University of Southern California Dissertations and Theses

Does procedural fairness matter in standardization? An examination of a drug standardization process in hospitals and its impact on hospital effectiveness

(USC Thesis Other)

Does procedural fairness matter in standardization? An examination of a drug standardization process in hospitals and its impact on hospital effectiveness

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DOES PROCEDURAL FAIRNESS MATTER IN STANDARDIZATION? AN EXAMINATION OF A DRUG STANDARDIZATION PROCESS IN HOSPITALS AND ITS IMPACT ON HOSPITAL EFFECTIVENESS Copyright 2003 by Seok-Woo Kwon A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (BUSINESS ADMINISTRATION) August 2003 Seok-Woo Kwon Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3116735 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3116735 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 900894695 This dissertation, written by Seok-W oo Kwon under the direction o f h - * - s dissertation committee, and approved by all its members, has been presented to and accepted by the Director o f Graduate and Professional Programs, in partial fulfillment o f the requirements fo r the degree of DOCTOR OF PHILOSOPHY Director Date A u gu st 1 2 , 2003 Dissertation Committee Chair \ J Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS LIST OF TABLES.. ............... v LIST OF FIGURES.....................................................................................................vii ABSTRACT. ....... ...viii 1. INTRODUCTION ........................ ........................................................ ..................1 2. LITERATURE REVIEW..... ...... 4 2.1. Effectiveness of Standardization......................................................... 4 2.1.1. Arguments for the Effectiveness of Standardization ..... 4 2.1.2. Arguments Against the Effectiveness of Standardization ..... 7 2.1.2.1. Criticisms from Cognitive Perspective .................. 8 2.1.2.2. Criticism from Affective Perspective ............ 9 2.2. Infusing Standardization with Fairness...................................................... . 11 2.2.1. Organizational Theory Perspective................... 12 2.3. Organizational Fairness................... 14 2.3.1. Outcome Fairness.. ...... 15 2.3.2. Procedural fairness................................... 18 2.3.3. Interactional & informational fairness. ..... 19 2.3.4. Overcoming Cognitive Downsides of Standardization.................. 23 2.3.5. Overcoming Affective Downside of Standardization ...... 25 2.4. Formularies and Formulary Systems.............. 27 2.4.1. Rapid Increases in Drug Costs..................................................... 29 2.4.2. Inappropriate Use of Drugs......................... ....30 2.4.3. Adverse Drug Events. ..... .31 2.5. Medicaid Formularies ..... 33 2.6. Hospital andHMO Formularies ............. .36 3. THEORETICAL MODEL........... .38 3.1. Arguments for the Effectiveness of Standardization......... ..... 38 3.2. Arguments Against the Effectiveness of Standardization ..... 39 3.3. Organizational Justice Perspective.. ........ ...................42 4. RESEARCH DESIGN ..... ..49 4.1. RESEARCH CONTEXT................ ......49 4.1.1. Choice of Samples and Sample Size... ..... ...49 4.1.2. Data sources ........ .............52 ii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.1.3. Choice of Diagnostic Categories and Drags.... ....... ...53 4.1.4. Pilot Test. ............................................................................................. 54 4.1.5. Survey Data Collection.................. .....................................................55 4.2. Analysis Methodology: Testing moderating effect ..... 57 4.3. Measures.............. ............... ...............................................................................58 4.3.1 Reliability and validity issues. .................... ......58 4.3.2. Hospital Effectiveness .... 58 4.3.3. Formulary characteristics..........................................................................60 4.3.3.1. Standardization...................................................................................60 4.3.3.2. Procedural Fairness............................................................................. 65 4.3.33. Formulary Incentive System........................ ....................................... 70 4.33.4. Nonphysician votes in P&T committee...............................................70 43.4. Patient Characteristics...............................................................................71 4.3.4.1. Patients’ health status ...... 71 43.4.2. Insurance. ..... 72 43.4.3. Demographic information. .............. 73 4.3.5. Hospital Characteristics...................................................................... ..........74 4.3.5.1. Ownership.......................................... 74 43.5.2. Urban/Rural.............................. .75 4.3.53. Geographic region.... ....... 75 43.5.4. Teaching institution.. ....................76 43.5.5. Multi-hospital system membership.... .................. ....76 43.5.6. Hospital size ..... .77 43.5.7. Hospital case mix index........................................... ......................77 43.5.8. Area Wage Index ......................... 78 43.5.9. Insurance mix................ 78 4.4. Preliminary data analysis and descriptive statistics....................... 78 4.4.1. Descriptive Statistics....................... .........78 4.4.1.1. The number of drugs.......................... .......79 4.4.1.2. Organizational context of drag standardization............... ......81 4.4.1.3. Organizational context of formulary practices.............. ....83 4.4.1.4. Standardization and fairness ..... ...........83 4.4.2. Correlations ........... ..............86 4.4.3. Assessment of the Representativeness of the Sample ....... 98 4.4.3.1. Comparison of respondents and mailing sample ........... 99 4.43.2. Comparison of early and late respondents 103 4.43.3. Comparison of variable means across hospital characteristics........ 105 5. RESULTS ......... 108 5.1. Diagnostics ...... 108 5.2. Hierarchical regression....... .................................................................... . 110 5.2.1. The base models...... ........ 110 5.3.3. Interaction models ................ .....115 iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6. CONCLUSIONS........ .......... .124 6.1. Limitations of the study..... ..... 125 6.2. Contributions....... ..... 126 REFERENCES.... ..... 128 APPENDIX.................................................................. .146 iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Table 1 Potential Advantages and Disadvantages of a Formulary System (Adapted From Blumenthal et al., 2000)..... 33 Table 2 Number of Surveys Mailed and Responded by State...................... 57 Table 3 Factor analysis of standardization measures ........ 65 Table 4 Factor analysis of procedural fairness measures... ...... 68 Table 5 Descriptive statistics of the number of drugs................... 79 Table 6 Organizational context of standardization ............ 82 Table 7 formulary practice and organizational context ............ 83 Table 8 Correlations........................................ 87 Table 9 Descriptive Statistics ....... .96 Table 10 Number of hospital beds of respondents to mailing sample ........... 99 Table 11 One-Sample Test.......................................................... .................................99 Table 12 Ownership of respondents and mailing sample ..... 100 Table 13 Chi-square test by ownership groups ............................................101 Table 14 System affiliation of respondents and mailing sample .............. 102 Table 15 Geographical location of respondents and mailing sample ..... 103 Table 16 Variable means and standard deviations for early and late respondents... 104 Table 17 system affiliation for early and late respondents ..... 105 Table 18 comparisons of means by system-affiliation groups ..... 106 Table 19 Comparison of means by ownership groups .......... 107 Table 20 The base regression models on pharmacy charge for aggregated dieseases _ .....112 Table 21 Main effect regression models for aggregated diseases ..... ....114 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 22 Interaction models with individual standardization and procedural fairness items ............. 117 Table 23 Interaction models with individual standardization items and overall fairness index .................................... ...119 Table 24 interaction models with overall standardization index and individual fairness items.............. 120 Table 25 Interaction models with overall indexes ............... 121 Table 26 List of the variables...................... 146 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Figure 1 Mechanisms underlying effects of procedural fairness....................... 23 Figure 2 Theoretical Framework................................................. 38 Figure 3 Standardization and procedural fairness........................... .48 Figure 4 Multiple research methods...................................................... ......52 Figure 5 Histogram of the number of cardiovascular drugs on formulary............ ....80 Figure 6 Histogram of the number of pneumonia drugs on formulary.......................80 Figure 7 Histogram of the number of ulcer drugs on formulary...................... 81 Figure 8 Scatter plot for cardiovascular diseases................................ 84 Figure 9 Scatter plot for pneumonia...................................... 85 Figure 10 Scatter plot for gastrointestinal diseases................................. ..85 Figure 11 Interaction................................................................................................... 122 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT In this study, I explore the degree to which procedural fairness determines responses to bureaucratic standardization. I examine this question in the context of drug formularies in hospitals. Drug formularies are a list of drugs available to physicians in a healthcare organization, and represent a standardization of the drug choices available to physicians. While some proponents advocate the use of drug formularies as cost-effective management, critics see the use of formularies as potentially harmful to patients and as an infringement on the clinical autonomy of physicians, thereby ultimately contributing to poorer quality and/or higher costs of patient care. Drawing on organizational justice theory, I examine the role of procedural fairness in shaping the effectiveness of bureaucratic standardization. In conclusion, I argue that the effect of drug formularies on organizational effectiveness depends on the degree to which the process of setting and implementing formulary policies is deemed fair by those individuals who use these formularies. Theoretically, this study contributes to the extension of procedural justice from its more traditional micro level formulation to the organization level. Rather than focusing on social psychological measurements of fairness and its outcomes, this study studies the structural features of procedural justice and their organizational consequences. From a practical point of view, understanding the conditions under which formulary standardization leads to organizational effectiveness is potentially an important means of improving the cost-effectiveness of the healthcare delivery system. The knowledge gained will help healthcare organizations and policy makers viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to design formulary standards that enhance, rather than undermine, organizational effectiveness. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1. Introduction Organizational research presents two conflicting views of performance benefits of standardization. According to the negative view, standardization creates a rigid and demoralized work environment which in turn reduces organizational effectiveness. According to the positive view, standardization contributes to overall organizational effectiveness as it enhances feelings of self-efficacy and collective- efficacy and provides efficiency benefits to organizations. This study attempts to partially reconcile these two conflicting views by considering the procedural fairness of standardization. Research on procedural justice suggests that an individual’s perception of the fairness of a procedure is critical to their satisfaction with a final outcome, even when the outcome is not the desired one. Elaborating on this perspective, I argue that people are willing to accept even a high degree of standardization without feeling stifled and demoralized when the process of standardization is fair. The literature suggests that procedural fairness can remove some - and perhaps all - of the potential psychological cost of standardization, without diminishing the features of standardization that enhance effectiveness. I will examine this proposition in the context of standardizing the pharmaceuticals used in hospitals; i.e. drug formularies. A drug formulary is a hospital-approved listing of prescription medications that are available to physicians for patient care. A hospital that has adopted an "open” or “voluntary" formulary allows physicians to prescribe drugs whether or not they are listed in the formulary. 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Conversely, a hospital that has adopted a "closed” (equivalently, “restrictive,” or “mandatory") formulary necessarily limits the physicians’ use of drugs to those listed in the formulary. The drug formulary thus represents a standardization of drugs available to physicians and is commonly used by hospitals to ensure that drug choices are cost-effective. Critics see the formulary as potentially harmful to patients and an infringement of the clinical autonomy of physicians, thereby ultimately contributing to poorer quality and/or higher costs of providing care. I propose to study the role of procedural fairness in creating and managing drug formularies, and how that affects the cost-effectiveness of patients. This study advances organization theory in three ways. First, although many studies have described and explained the process of standardization (see reviews in Hallstrom, 2000), the link between procedural fairness and the effectiveness of standards has not yet been explored. There have been very few explicit efforts to cross-pollenize standardization literature with fairness literature. This may in part due to the disciplinary division of labor: Standardization is often studied by economists and structural sociologists, while fairness is mainly researched by social psychologists. In this study, however, I argue that when the processes of standardization are fair, a high degree of standardization will be accompanied by positive attidudinal outcomes. This question is fundamental to organization theory, as standardization is a key element of bureaucratic organizations (March & Simon, 1958; Mintzberg, 1979; Thompson, 1967). 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Second, this study attempts to increase the theoretical relevance of procedural justice for organization theory. Procedural justice has been studied primarily at the individual level, and researchers have focused on social psychological measurements of fairness and its outcomes (e.g., citizenship behavior, commitment, or satisfaction). Hence, this literature has been relatively silent on the structural features of procedural justice at the organizational level and their organizational consequences. In this study, I extend procedural justice from its more traditional micro-level formulation to the organization level, and examine its impact on organizational performance. Third, standardization in professional settings is increasingly common as knowledge-intensive organizations come under competitive pressure to reduce costs, increase timeliness, and improve quality (Brennan & Berwick, 1996; Freidson, 1984; Stevens, 1989). Yet, among organizational researchers, considerable skepticism remains as to whether or not professional work can be effectively standardized to benefit organizational goals (Baer, 1986; Boreham, 1983). Hence, managers of professional organizations are often pulled in contradictory directions by conflicting recommendations. In this study, I contend that professional work can be standardized for organizational goals without alienating professionals if the procedures for standardization are deemed fair. 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2. Literature Review In the following sections, I first review general literature on standardization and its conflicting predictions on organizational effectiveness. Then I introduce fairness literature and suggest that it can be fruitfully combined with standardization to produce organizational effectiveness. Finally, I discuss briefly the empirical works done on drug formularies by health services researchers, to provide a context for our theoretical model. According to David & Greenstein (1990), there are two types of standardization: de facto and de jure. De facto standardization emerges spontaneously, often from market-mediated processes. On the other hand, de jure standardization is usually issued from administrative procedures and enforced by authorities that have some regulatory power. In this study, I am mainly interested in dejure standardization and use the term standards as specific kinds of rules created by authority, and standardization as the production of such rules (Brunsson & Jacobsson, 2000). In this sense, I exclude informal de facto standards from the focus of the study. 2.1. Effectiveness of Standardization 2.1.1. Arguments for the Effectiveness of Standardization A more task-oriented stream of organizational theory highlights the effectiveness benefits of standardization. Here the assumption is that standardization can yield cost saving, rather than wasteful outcomes. The argument goes that 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. standardization leads to greater predictability, the perfection of performance through repetition and routinization, and the economies of simplification by reducing variety (Brunsson et al., 2000). Invoking or implying assumptions such as those, for example, many writers in the operations management field (e.g., Deming, 1986; Schonberger, 1986) have endorsed standardized work inventory and work processes as a key feature of statistical quality control and total quality management. A number of theoretical approaches provide possible underlying mechanisms for a positive relationship between standardization and effectiveness. First, according to the economies of scale concept, standardization increases the number of similarly specified goods or services, and through the associated simplification decreases the average unit cost of a good or service. Thus when a good or service is standardized, it sppeds up task learning of employees through repetition and reduces the costs of purchased materials and services by giving the purchaser a better bargaining position through quantity discounts, and reduces the number of changeovers. Second, according to information processing perspective, standardization can improve effectiveness by economizing on scarce information processing capacity (Galbraith, 1977; March et al., 1958; Nelson & Winter, 1982). For example, if I know that another person or organizational unit is complying with a particular standard, it is easier for us to anticipate their behavior and adjust our actions accordingly. Thus, Schelling (1978) used the examples of traffic signals and clocks to illustrate how certain agreed-upon standards allow individuals to coordinate people’s activities without extensive exchange of information between them. 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Argyres (1999) also argued that the standardization achieved by the information system facilitated the coordination and governance of the project more efficient. Specifically, he attributed the success of developing B-2 bomber to the standardization created by B-2 product definition system, because it served as a system of communications codes useful for transmitting informal knowledge from one engineer to another. Third, researchers from the organizational learning perspective have argued that standardization contributes to effectiveness through individual and organizational learning. Individual employees can easily re-use the skills they learned previously if tasks are standardized. Also they can perform these tasks more efficiently and quickly when they could acquire more experience doing the standardized tasks. Organizations can also get learning benefits of standardization, by translating individual knowledge to organizational rules and polices, and turning these rules into a key depository of past learning and best practices (Craig, 1995; Deming, 1986; Levitt & March, 1988; Nelson et al., 1982; Simon, 1976). A good example of this is David and Rothwell’s (1996) comparison of the nuclear industry in the US and France. They suggested that, because of the lack of standardization in the U.S., the nuclear industry has lost competitiveness relative to the French nuclear industry. Fourth, according to advocates of modular design, standardization contributes to effectiveness by fostering innovation. Arguing against the view that standards bring about only uniformity, proponents of modularity design emphasize that 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. standards can actually favor variation and innovation. Baldwin & Clark (2000) argued that standardization allowed the computer industry to adopt a widespread use of modular designs and this in turn has dramatically increased its rate of innovation. Their argument is that by breaking up a product into subsystems or modules, computer designers have gained enormous flexibility because different companies can take responsibility for separate modules and be confident that a reliable product will arise from their collective efforts (see Langlois & Robertson, 1992). Finally, numerous studies in role stress theory found that formalization (a concept closely related to standardization) reduces role conflict and ambiguity, thereby increasing work satisfaction and reducing work stress (Jackson & Schuler, 1985). Other social psychological studies also reported results that support positive benefits from standardization (Michaels, Cron, Dubinsky, & Joachimstahler, 1988; Snizek & Bullard, 1983; Stevens, Diedriks, & Philipsen, 1992) 2.1.2. Arguments Against the Effectiveness of Standardization In response to the arguments by proponents of standardization, critics characterized such pro-standardization positions as “naive formalist” (Star, 1995, p. 16), and provided a number of counter arguments. Their criticisms can be organized in terms of three perspectives: cognitive, political, and affective. 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.1.2.1. Criticisms from Cognitive Perspective A key criticism in this view is that standards are an impoverished representation of reality. The emerging literature on the importance of communities of practice points to the importance of social networks and communities of knowledge as critical to gaining acceptance for ideas and knowledge. According to this view, standardization is a seriously flawed approach because of its limited vision of how work is accomplished and its ignorance of actual work practices. This criticism focuses on the fundamental poverty of standardization in comparison with the richness of the empirical world: Suchman (1987) described part of the problem as the “relation between normative accounts of how work gets done and actual practice.” Because standardization often involves deleting the details of that which it represents, thereby simplifying the reality represented, it is seen as a flawed approximation of the reality. According to the literature on communities of practice, tacit and working knowledge, essential to working efficiently, is bound to be lost with standardization (Orr, 1996).1 Looking at the worst-case scenario, standards can impose false uniformity on a large scale, and misguide work efforts. Contingency theory, however, points out that, depending on the complexity of task, standardization may not hinder performance. If reality is not all that “rich,” even an impoverished representation can be a serviceable approximation. 1 According to Orr (1998), the technicians he studied do use the standardized documentation, but only selectively in their own discretion: “they choose how and when to use a given test or procedure independently o f what the documentation requires . . . they do not dismiss the directive documentation, nor do they blame the designers when the solution for a problem is not found there. They believe that the behavior o f the machine is so complex that it cannot be predicted beforehand” 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Critics of standardization also argue that standardization can homogenize and drive out diversity. When an extreme degree of standardization is imposed, the problem is that it can severly circumscribe what is permitted and generates a ty ranny of systematization and uniformity, thereby stifling creativity and the scope for learning and progress. In evolutionary theory terms, it delimits the domain over which variation is able to operate and reduces the opportunity for learning. Consequently, standardization curtails the potential for the formation of new combinations and the regeneration of variety from which further selection will be possible. Hence, the critics argue that standardization and uniformity may bring short-term effectiveness gains, but only at the cost of foregoing some of the long term gains through innovation or experimentation. However, advocates of modular design have argued, to the contrary, that in fact standardization promotes innovation and creativity (Baldwin et al., 2000) 2.1.2.2. Criticism from Affective Perspective Standardization can be an unwelcome, unnecessary, and harmful intrusion into individual autonomy and freedom. For example, Robert Merton (1940) shows how goal displacement in bureaucratic organizations generates rigidity. Merton argues that the organization’s demand for reliability pressures its members to exhibit “an unusual degree of conformity with prescribed patterns of action” (1940, p. 562). This creates inflexibility. When standardization works in a rigid, inflexible way, it can deskill and dehumanize workers. This can lead to goal displacement or the 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. process of a rule becoming an end itself, rather than a means to an end. Consequently, “the very elements which conduce towards effectiveness in general produce ineffectiveness in special instances” (Merton, 1940). In social psychology literature, according to the job characteristic model, if jobs are designed in a way that increases the presence of core characteristics (i.e., skill variety, task identity, task significance, autonomy, task feedback), employees will find the work meaningful, feel responsible for work outcomes, and understand the consequences of their work, thereby increasing their work motivation and job satisfaction (Hackman & Oldham, 1980). Because work standardization is often believed to have a negative impact on those core characteristics (especially limiting variety and autonomy), researchers from the job characteristics approach argue that standardization will undermine commitment by employees and foster dissatisfaction. The negative impacts of standardization on employees’ attitudes are seen as even more destructive among professionals. Traditionally, much of what has been written about professionals working in bureaucratic organizations have typically assumed the professional-bureaucratic conflict model, according to which there is an inherent conflict between professional and bureaucratic ways of organizing work. Thus, when professionals’ work becomes bureaucratized by the implementation of formalized rules and standards, professional employees are expected to be demoralized and unproductive (Raelin, 1986). Some theorists (Derber, Schwartz, & Magrass, 1990) even advance a proletarianization interpretation that highlights professionals’ progressive subordination to bureaucratic standardization. 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, because standardization often spans across a variety of distinct work communities and multiple meaning systems using different representational frameworks, achieving standards that would work in all settings are not feasible. Rather there may be multiple, possibly inconsistent, competing standards, none of which has a unique claim to validity. Under such circumstances, problems can arise if a standard generated at a distance from a work site is transposed onto remote sites without careful regard for local work practices. Examples include administrative standards which go against doctors’ professional norms but which are nevertheless introduced into health care by administrators. In such conditions, questions arise, as Brunsson and Jacobsson put it, whether employees can “trust the expertise and goodwill of those who set the rules.” Also “(whether) the standardizes know best what is right” for employees is another touchy issue (Brunsson et al., 2000, p. 171). If such incongruent standards are forcibly imposed on certain groups, the critics argue that standardization privileges certain interests over others (Bowers, 1992; Suchman, 1987; Wood, 1992, p. 233) and may encounter strong resistance. 2.2. Infusing Standardization with Fairness Combining the two previous sections’ arguments, I submit that standardization produces mixed effects, both positive and negative consequences. While both effects merit serious considerations, we also need a theory that explains why the effects are sometimes positive, and sometimes negative. I will argue that the procedural fairness of standardization plays a key role in explaining the alternative 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. consequences. Research on standardization has traditionally focused on the degree of standardization, and has not looked at the types of standardization. Also, while some prior research (cited below) has suggested the general value of studying standardization processes, my contribution lies in bringing in procedural justice literature to the debate. Consistent with my proposition, previous researchers (Adler & Borys, 1996; Gouldner, 1954; Sitkin & Roth, 1993) have suggested that researchers should pay a more careful attention to the way that fairness can mitigate the negative outcomes of standardization or even make standardization a positive factor for employee motivation as well as organizational innovation. They believe that standardization is sometimes combined with fairness to produce effective and efficient outcomes. I review several studies in organization theory and legal theory that suggest such possibility. 2.2.1. Organizational Theory Perspective. Alvin Gouldner (1954) suggested that not all bureaucratic structures are alienating, and invoked several criteria— which closely resemble fairness— in distinguishing two different types of standardization in bureaucratic organizations: who initiates rules and whose interests do the rules serve. Based on those criteria, he distinguishes rules that are imposed (which he calls ‘punishment-centered bureaucracy’) and that are agreed upon (which he calls ‘representative bureaucracy’). Building on Gouldner’s work, Adler and Borys (1996) highlighted the importance of 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. social psychological variables in examining the effectiveness of formalization, a concept closely related to standardization. They argued that the differential outcomes of formalization are a consequence of whether those formal rules are seen by employees as enabling them better to master their tasks or as coercing them. They suggested two main types of formalization— enabling and coercive— and identified several key attributes of rules that characterize each type. Extending this work, Adler (2001; 1999) further identified trust as one of the key contextual factors in promoting enabling bureaucracy. Sim Sitkin (1995) also suggested the potential complementary relationship between legal rules (another concept closely related to standardization) and trust. Those works all suggest a possibility that fair processes and outcomes can mitigate or remove the negative consequences attributed to standardization. My contribution is that these types differ primarily in the process underlying standardization, in particular, by the fairness of the associated processes. According to Thibaut & Walker (1975), when the conflict of interest is high, procedural fairness matters a lot. Since standardization is often a political process and thus generates conflict of interest, I argue that procedural fairness must be a key consideration in standardization. The review of literature in organizational theory suggests that at least certain version of standardization can be implemented with little downsides and yet still can yield significant effectiveness benefits. I believe that fairness is one of a key 13 permission of the copyright owner. Further reproduction prohibited without permission. consideration for standardization, because it helps to preserve effectiveness features of standardization while at the same time mitigates some of the downsides of standardization. I maintain that, when standards are formulated and implemented in a fair manner, the commonly encountered negative outcomes of standardization would be replaced by positive outcomes generated by fairness. 2.3. Organizational Fairness Organizational justice research suggests that the fairness assessment involves both fairness in outcome and process (Greenberg, 1987). A process approach focuses on how various outcomes are determined. In contrast, outcome approach is concerned with the fairness of the resulting outcomes. This categorization is consistent with earlier research on legal justice by distinguishing between the way verdicts are derived and what those verdicts are (Walker, Lind, & Thibaut, 1979). However, I should also point out that there are some researchers who argue that this distinction between procedural and distributive justice can be questioned. Cropanzano & Ambrose (2001) argue that, because the same event can be seen as a process in one context and an outcome in another, people may not always perceive the distinction between procedural and outcome fairness. In support of their conceptual discussion, other researchers also found high correlations between them (Martocchio & Judge, 1995; Sweeney & McFarlin, 1997; Welboume, Balkin, & Gomez-Mejia, 1995). However, I leave this as an empirical issue to be examined with the data, since the unique nature of this study setting, formulary restriction in 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. healthcare settings, makes it difficult to predetermine whether fairness has one or multiple dimensions. The following sections expand on these two types of fairness judgments. 2.3.1. Outcome Fairness. Much of the research on outcome fairness has been concerned with the issues of distributive justice in exchange or allocation relationships. The focus has been on the distribution of economic goods, as well as the distribution of “conditions and goods that affect well-being, which includes psychological, physiological, economic, and social aspects” (Kabanoff, 1991). According to Greenberg (1987), in research on outcome fairness, a distinction can be made between studies of people’s decisions about the allocation norms that should be followed (i.e., research focusing on allocators’ behavior), and studies of how people judge and respond to unfair outcomes (i.e., research focusing on resource recipients’ justice perception). He calls the former a “proactive theory of justice,” and the latter a “reactive theory of justice” (Greenberg, 1987) According to studies that examined the rules by which allocators distribute resources, people use a variety of principles or values as the basis for distributing outcomes, and identified a number of them, which they call as “distributive rules.” Perhaps the most well known is that ofDeutsch (1975). He identifies three distinct distributive rules that are applied during allocation and exchange: equity (i.e., equality of outcome/input ratios, equality relative to individual contributions, or ratio 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. between actual outcome vs. “ just share”), equality (equal amounts to each recipient), and need (equality of outcomes taking into account need).2 Much of the research over the past decades has been attempts to discover which rules are appropriate in various social situations (for a review, see Cook & Hegtvedt, 1983; Deutsch, 1985). The main consensus in organizational studies seems to be that, in cooperative relations within which economic productivity is a primary goal, equity rather than equality tends to be the preferred rules, because equity rule allows differential and competitive distribution of awards. On the other hand, in relations in which the fostering or maintenance of cooperative social relations is the common goal, equality tends to be the dominant rule (Kabanoff, 1991; Meindl, 1989). Finally, when the concern for dignity and humaneness is paramount, individuals are rewarded based on what they need or require to survive as humans and to maintain their sense of individual dignity (Sheppard, Lewicki, & Minton, 1992). Consistent with these suggestions, Meindl (1989) found that equity will be preferred when the goal is to enhance productivity, and that equality will be preferred when the goal is to enhance enjoyable social relations. In a cross-cultural study, Leung & Lind (1986) supported Deutsch’s notions that concern for productivity is related to preference for equity, whereas concern for interpersonal relationships is related to preference for equality, with Korean as well as American subjects. 2 Others have identified a similar list o f distributive rules. For example, Eckhoff (1974) suggested a more elaborated list o f rules: equal amounts to each (objective equality), subjective equality, relative equality (equity), rank order equality, and equal opportunity. Lerner (1982) identified four principles: competition (allocations based on the outcome o f performance); parity (equality allocations); equity (allocations based on relative contributions); and Marxian justice (allocations based on needs). Fiske (1993) also suggests that humans use just four relational models (Communal Sharing, Authority Ranking, Equality Matching, and Market Pricing) to generate, interpret, coordinate, contest, evaluate, and sanction most social interaction. 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Other researchers have looked at people’s perceptions of the fairness of the outcomes they receive. While the research on reward allocation rules examined fairness from the allocator’s perspective, and thus was concerned with the beginning of the allocative process, the justice perception research took on the perspective of the allocatees and examined their evaluations of the end state of the allocation process. In other words, this stream of research has looked at how people derive fairness judgment, given that a certain distribution rule has been implemented. In general, the researchers found that people tend to be less satisfied with outcomes they perceive to be unfair than those they perceive to be fair. The core question in this stream of work has been to understand the process by which individuals form judgments of outcome fairness (e.g., Cropanzano & Greenberg, 1997). The findings suggest that fairness judgments are determined by the magnitude of the outcomes received, as well as by how these outcomes compare to a certain standard (e.g., Sweeney, McFarlin, & Inderrieden, 1990). The subsequent research has examined how people come to decide on the choice of comparison standard. For example, Sweeney, McFarlin, and Inderrieden (1990) found that although salary level was related to pay satisfaction, other various considerations (e.g., wants, past and future expectations, and entitlement) add to pay satisfaction beyond the amount of one’s income. 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3.2. Procedural fairness. In contrast to the focus of outcome justice on outcomes, process justice focuses on the methods by which decisions are made (for reviews see Cropanzano et al., 1997; Folger & Cropanzano, 1998; Konovsky, 2000; Tyler, 2000). Research in organizational justice suggests that an individual’s perception of the fairness of a procedure is critical to their satisfaction with the final outcome. For example, researchers have argued that procedurally fair treatment makes individuals more accepting of smoking bans (Greenberg, 1994), pay systems (Miceli, Jung, Near, & Greenberger, 1991), rationing decisions (Tyler & Degoey, 1995). On the other hand, when decision-making processes are perceived as unfair, it could lead to noncompliance of decision outcomes and even elicit retaliatory behavior from employees. Studies have found that, while procedurally unfair treatment would result in organizational retaliatory behaviors (Skarlicki, 1997), people are willing to obey personally unfavorable decisions if their decision making processes are perceived as fair (Tyler, 1990). For example, Kim & Mauborgne (1993) found that subsidiary top managers’ perception that their head offices exercised procedural justice motivated them to comply with head office directives. Other studies found that, when fairness perception is high, people even voluntarily monitor others’ behavior to achieve collective goals (Welboume et al., 1995). 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3.3. Interactional & informational fairness. Currently, the literature is not clear whether interpersonal interaction should be conceptualized merely as facets of existing concepts of fairness—namely process or outcome fairness—or as a fundamentally different fairness concept. On the one hand, Bies’ research on interactional justice suggested that people can distinguish three aspects of the allocation sequence—namely, procedures, interaction, and outcomes—and that it was conceptually meaningful to distinguish the concept of interactional justice from those of procedural justice and distributive justice. He further argued that interactional justice can be even further subdivided into two components: (a) clear and adequate explanations, or justifications, and (b) treatment with dignity and respect (Bies & Moag, 1986). Similarly, other theorists (e.g., Greenberg, 1994) have expanded the field of process justice to include additional criteria such as interactional justice (i.e., the fairness of interpersonal treatment people receive when procedures are enacted) and informational justice (i.e., the adequacy of the information provided about the procedure and its implementation). Hence, the three recent meta-analyses of the organizational justice literature, as cited in Cropanzano et al. all are unanimous in arguing for the separation of procedural from interactional justice. On the other hand, Cropanzano & Greenberg (1997) argue that these two types of fairness are difficult to distinguish from procedural justice, and that they should be treated as aspects of process justice as opposed to separate form of fairness. For example, Greenberg (1993) suggested that the respect and sensitivity aspects of 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. interactional justice might be viewed as interpersonal facets of distributive justice, because they alter reactions to decision outcomes. He further argued that the explanation aspect of interactional justice might best be viewed as an interpersonal facet of procedural justice, because explanations often provide the information needed to evaluate structural aspects of the procedure. On a similar vein, Lind and Tyler (1988) argued that procedural justice inherently involves considerations of both the structure of procedures and the quality of treatment. The idea that procedural and interactional justice were part of a single dimension is echoed by many conceptual pieces on organizational justice (Brockner, 1996; Cropanzano et al., 1997; Greenberg, 1990). So far the research on justice of formulary management is very sparse. Though a number of researchers have suggested the importance of justice in formulary decisions, they have not been able to empirically operationalize or test it. Given that justice research in this clinical setting is still in its infancy, it is appropriate that this study will examine procedural justice as well as informational and interactional justice criteria suggested in the literature, and see how they shape the process justice assessment of restricted formularies. Why is fairness important in formulary management? Greenberg (2001) suggested four key conditions under which justice becomes a salient issue. First, concerns about justice are triggered when people receive negative outcomes personally or collectively. After all, when people get what they want, they are not likely to consider fairness seriously. Second, justice becomes an issue when things 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. are in flux and uncertain. Third, concerns about justice are highlighted when resources are scarce. When there is more than enough resource to go around, questions about fairness do not emerge as often. Finally, concerns about justice are salient among people having different levels of power. With the power differential removed, justice often becomes a moot issue. The situation with formulary standardization meets many of these conditions. First, people, especially healthcare professionals and physicians, would often prefer autonomy over control and restrictive policies. For many physicians, formulary restrictions represent an illegitimate intrusion on their clinical authority, and thus are regarded as having negative outcomes. Second, though the formulary restriction is not a new issue, because the history of formulary is long in the US, it is an area that has attracted a lot of attention these days, as pharmacy costs and healthcare costs soar. With the tightening of the federal government support for healthcare costs and increasing role of the third-party payers in the delivery of care, the level of resources once taken for granted to physicians are dwindling very rapidly. Consequently, physicians and healthcare organizations are left with the less and less resources (Cassel, 1985). In addition, more and more physicians are experiencing their level of power over patients and vis-a-vis third-party payers declined. Hence, many of Greenberg’s conditions for making the fairness concerns salient are present in the case of formulary restrictions to a greater degree than before. Therefore, the focus on drug formularies in this study offers a unique opportunity to examine the influence of procedural fairness on the effectiveness of standardization. 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Furthermore, the formulary context is suitable because I can find variations in all three variables of our theoretical interests: the degree of standardization, the degree of fairness, and resulting organizational effectiveness. Hospitals vary in the number of drugs carried in the formularies as well as the degree to which hospital administrators enforce formularies rules. In a survey of 187 hospitals, Mannebach et al., (1999) found that most of the respondents (87.7%) reported that their institutions used closed formularies, but their degree of restrictions varied. Pedersen, Schneider, and Santell (2001) also found that more than 90% of hospitals use various formulary-system management techniques, but some techniques are used more often than others. Hence, the degree of standardization in each formulary varies from hospital to hospital, thereby offering a rich setting to examine the consequences of standardization. Moreover, hospitals vary in the procedures for formulating and implementing formularies: some use more fair procedures to physicians while others use less fair approach. Finally hospitals vary in their effectiveness in managing pharmacy costs. In the sections below, I explore how certain features of fairness compensate for the negative outcomes of standardization. I organize the discussion in terms of two downsides of standardization: Cognitive and affective downsides. Figure 1 summarizes the relationship in a diagram. 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1 Mechanisms underlying effects of procedural fairness Procedural fairness Reduces cognitive downsides of standardization Reduces affective downsides of standardization Increases org. cost effectiveness 2.3.4. Overcoming Cognitive Downsides of Standardization. I believe that, even though standardization simplifies and thus may even impoverish work reality sometimes, so long as it is done in a highly fair manner, it may still have effective outcomes. I believe that many features of fairness work jointly in addressing the cognitive downsides of standardization. Rather than elaborating on all the underlying processes, my discussion focuses selectively on two mechanisms: participation (i.e. voice) and due process. Participation enhances the flow and use of important information. Since workers typically have more complete knowledge of their work than standard setters, if they participate in standardization, resulting standards will be made with better pools of information. In addition, if workers participate in standardization, they will know more about implementing standards and much more committed to accepting them (for the productivity benefits of worker participation, Miller & Monge, 1986). 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. While organizational justice research has been relatively silent on effectiveness effects of justice, participation in decision making literature, a related literature to procedural justice, shows that participation is associated with positive satisfaction and, to a lesser extent, productivity outcomes (Miller et al., 1986; Wagner, 1994).3 For example, Ebert and Mitchell (1975) offer some rationale as to why participation has productivity effects. They suggest that participation in decision-making helps workers understand more clearly what is expected of them and increases the likelihood that they will work toward rewards that they value. Empirically, Cammann and Lawler (1973)showed that a group incentive system to improve productivity approved by a workgroup with a long history of participative decision-making was more effective than a similar system that had been imposed on a workgroup with no history of participation. Using an experimentally controlled study, Cooper, Dyck, and Frohlich (1992) also find that when people work under an impartially chosen fair rule, they realize significant productivity gains. However, working under a rule that has been imposed without participation does not lead to higher productivity. Another feature of fairness, due process helps to enhance cognitive downsides of standardization. Since standardization involves creating consensus knowledge among multiple interests and stakeholders, a key task is understanding and incorporating many different viewpoints into standardization process (Gerson & 3 However, this literature is also not without debate. For example, some researchers have raised questions about participation's ability to affect performance and satisfaction in the workplace (Ferris & Wagner, 1985; Locke & Schweiger, 1979). However, when the research is reviewed carefully, it appears that participation typically has a modest influence on employee productivity, motivation, and job satisfaction (Wagner, 1994). 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Star, 1986; Star, 1995). In that case, the key question becomes: In combining evidence from different viewpoints, how do you decide that sufficient, reliable, and fair amounts of evidence have been collected? Who, or what, does the reconciling, according to what set of rules? How can I assure that standardization make adequate provision for recognizing, weighing, and evaluating alternatives from conflicting sources? Gerson and Star (1986) call the work of ensuring due process “articulation.” Provision of an appeal process and flexible implementation facilitate some of the articulation work involved. They give employees chances to rectify problems and limitations of standards. At first glance, providing a way to question a standard’s specification through these due process mechanisms seems at odds with what the standardizer are trying to achieve. However, according to Gasser (1986), “far from acting irrationally, the informal practical actions of participants actually make systems more usable locally.” In this study, I argue that, while fairness in and by itself may not yield effectiveness outcomes, when it is combined with standardization, it would produce effectiveness outcomes. 2.3.5. Overcoming Affective Downside of Standardization. The organizational justice theory has yielded a number of studies that provide affective impacts on workers. For example, Moorman (1991) found that justice 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. positively influenced job satisfaction. Similarly, McFarlin and Sweeney (1992) found that both distributive and procedural justices was significant predictors of pay level satisfaction and job satisfaction. A number of other studies also supports this view (e.g., Aquino, 1997; Konovsky & Pugh, 1994). Based on the literature, I believe that procedural fairness can reduce the affective downsides of standardization proposed. While the link between procedural fairness and reduction of affective downsides is supported consistently in the literature, the link between positive affective outcomes and organizational effectiveness is much more controversial. Researchers often phrase this question as whether happy workers are more productive workers. Despite years of effort, organizational researchers have failed to find convincing evidence that satisfied workers are more productive. Over 40 years ago, Brayfield and Crockett (1955) reviewed a large body of literature and concluded that there was no "appreciable" evidence of a relationship between the two variables. A more recent meta-analytic review of the empirical evidence came to the same conclusion, finding that the two variables form only an "illusory" correlation (laffaldano & Muchinsky, 1985). Staw and Barsade (1993) note that most researchers now view a positive effect of satisfaction on performance as an unsubstantiated claim of practitioners and the popular press. However, recently, organizational researchers have discovered more links between happiness and positive behaviors at work. For example, Judge, Thoresen, Bono, and Patton (2001) reviewed over 300 organizational studies and found a 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stronger relationship between job satisfaction and job performance (r = +.3). Since satisfied workers, compared to dissatisfied workers, are more committed to the organization, and more likely to help coworkers, this could plausibly lead to higher job performance. Although these behaviors may not necessarily result in increases in individual productivity, they may affect overall organizational outcomes or productivity. Several studies (Ostroff, 1992; Ryan, Schmit, & Johnson, 1996) suggest that, when satisfaction and productivity data are gathered for the organization as a whole, rather than at the individual level, organizations with fewer dissatisfied employees tend to be more effective than organizations with fewer satisfied employees. As Robbins puts it, “while we might not be able to say that a happy worker is more productive, it might be true that happy organizations are more productive”(Robbins, 2003, p. 80). After examining the evidence, I propose that, on the whole, the behaviors of satisfied workers might possibly result in increased organizational productivity. 2.4. Formularies and Formulary Systems So far I have suggested that fairness can compensate or even remove the negative outcomes of standardization, thereby contributing to attitudinal and effectiveness outcomes. Empirically studying the argument outlined above requires a 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. setting that contains observable variations in the degree of standardization as well as procedural fair in formulating and implementing standards. The research context I chose for this study is drug standardization in hospitals, known as formularies. Drug formularies are a list of drugs available to physicians in a hospital, and represent a standardization of the drug choices to physicians. A hospital formulary is formulated and managed by a Pharmacy and Therapeutic committee, often abbreviated as P & T Committee. The committee is comprised of physicians and pharmacists who are experienced in evaluating the effectiveness and safety of prescription drugs. The committee meets on a regular basis to evaluate, appraise, and select from among numerous available drug entities and drug products, those that are considered most useful in patient care (Blumenthal & Herdman, 2000). The large number of drugs available to treat various medical conditions and the high cost of some of these drugs has led many P & T committees to restrict the medications that physicians can prescribe to patients However, the cost effectiveness of restricted drug formularies is still a matter of debate, and empirical studies of the issue have yielded mixed results. Some studies have documented cost savings through formulary restrictions (see Glassman et al., 2001 for a review), while others have found little or no evidence of greater cost effectiveness (e.g., Horn et al., 1996; Kozma, Reeder, & Lingle, 1990; Moore & Newman, 1993; Sloan, Gordon, & Cocks, 1993). The few studies that examined the impact of formulary restrictions on quality of care have also reported mixed results (See Horn et al., 1996; Kozma et al., 1990 for negative impact); (see Walser, Ross- 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Degnan, & Soumerai, 1996 for neutral impact). One possible explanation for these inconclusive findings is that (a) the effectiveness of restricted formularies may depend on physicians’ acceptance of restricted formularies,4 and (b) this acceptance may depend on the procedural fairness of creating and implementing formularies.5 In the subsequent sections, I discuss briefly the growth of restricting drug usage, known as formulary, in terms of three main reasons: rapid increases in drug costs; overuse of medications; adverse drug events. 2.4.1. Rapid Increases in Drug Costs. The accelerated increase in pharmaceutical prices has promoted many healthcare organizations to establish cost-containment policies. Factors contributing to this growth in drug spending include higher pharmaceutical prices and the introduction of new more expensive drugs. Tufts University researchers found that during the 1990s prescription drug costs increased from a 5.5 percent share of total health care spending to 8.5 percent. This is despite the fact hospital expenses 4 Numerous studies have shown that physicians’ acceptance o f formulary restrictions varies among individuals and among healthcare organizations. For example, Schectman, Elinsky, Kanwal, & Ott (1995) found that 58% of physicians in a network HMO and 30% o f physicians in a group HMO felt that restricted formularies were an inappropriate way to control costs. Donelan et al. (1997) found that 31 % o f approximately 2000 physicians surveyed in 1995 reportedly disapproved o f limitations on prescribing drugs. Finally, in their survey of VA hospital physicians, Glassman et al. (Glassman et al., 2001) found that 29% o f physicians’ survey stated that the formulary at their hospital limited their ability to provide the best quality o f care. 5 There is some evidence that physicians can, and often do, find ways to circumvent formulary restrictions if they do not agree with them. For example, according to (Green, Chawla, & Fong, 1985) analysis o f physicians’ orders under a restrictive formulary at an university hospital, physicians continued to prescribe non-formulary drugs, of which 65% were for drugs previously evaluated by the Pharmacy and Therapeutics (P&T) committee and denied admission to the formulary. In a different context (Wynia, Cummins, VanGeest, & Wilson, 2000) found that a sizable minority (39 %) o f physicians admitted manipulating reimbursement rules so that patients could receive care that the physicians perceived as necessary (we can plausibly imagine a similar scenario with regard to formulary restrictions, when physicians do not accept formulary restrictions). I conclude that physicians’ acceptance o f formulary restrictions may be a critical factor in the success of formulary standards. 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. decreased from 37 to 33 percent, and physician costs held constant over the same time period. This figure is also confirmed by Donald Moran (2000) in his survey of managed care executives: "While hospital and physician care spending trends remain in the 2-5 percent range," he reported, "pharmacy costs are rising rapidly." (A 2-5 percent increase represents a decrease or constancy in real terms.) The health insurance lobby claims an overall drug spending increase of 19 percent in 2000 (Lueck, 2001). Similarly, other studies report that prescription drug costs outpaced the growth in overall health care costs in the 1990s (Levit et al., 1998). In 1997, total health care costs increased 4.8%, but prescription drug costs rose 14% (Iglehart, 1999). All these surveys corroborate other reports by health insurers and self-insured employers of 15-20 percent annual increases in pharmacy costs. 2.4.2. Inappropriate Use of Drugs. Monitoring drug utilization has also been motivated by the perception that medications are often over-prescribed by physicians. Studies have found that some drugs are prescribed for inappropriate indications (Beers et al., 1988; Kunin, Johansen, Woming, & Daschner, 1990), while other drugs are substantially underutilized despite their life-saving consequences (e.g., cardiovascular agents) (Horwitz, Viscoli, Clemens, & Sadock, 1990) Furthermore the research documents marketing information from pharmaceutical sales representatives influences physician prescription behavior. Avom and his colleagues (Avom, Chen, & Hartley, 1982; Avom & Soumerai, 1983; 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chren & Landefield, 1994; Schwartz, Soumerai, & Avom, 1989) found that staff physician interactions with a drug manufacturer significantly influenced physician interest in adding products of that manufacturer to their hospital’s formulary. On the patient end, direct-to-consumer advertising, independent of drug efficacy, generates patient demand for advertised drugs. Formularies are in part created to respond to such advertisement-driven demand as well as to address over- and under-utilization of drugs. 2.4.3. Adverse Drug Events. Reducing over-utilization of drags is also seen as a way to decrease the incidence of adverse drag events and associated expenditures for physician and hospital services (Ray, Griffin, Schaffher, Baugh, & Melton, 1987). Prescribing drags with no clear clinical indication, or with less toxic therapeutic alternatives, unnecessarily increases the risk of adverse drag events. Numerous studies have documented the extent of inappropriate drag use in hospitals and its morbid and economic consequences. One recent study conducted at two prestigious teaching hospitals found that almost two percent of admissions experienced a preventable adverse drag event, resulting in an average increased length of stay of 4.6 days for those two percent of admissions and increased hospital cost of nearly $4,700 per admission (Bates et al., 1997). Similarly, in a matched case-control study of all patients admitted to a large teaching hospital from January 1990 through December 1993, it was found that the occurrence of an adverse drag event was associated with 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. an increased length of stay of 1.91 days and an increased cost of $2,262. (Classen, Pestotnik, Evans, Lloyd, & Burke, 1997) Reacting to these trends, healthcare organizations have been standardizing and restricting the medications that physicians can prescribe to their patients and are performing economic analyses (Glennie, Woloschuk, & Hall, 1993; Mannebach et al., 1999). However, while some physicians see these formularies as a legitimate tool for managing drug safety, others and as potentially harmful to patients. Advocates of formularies have claimed that a properly developed and maintained formulary will foster safe, effective, and appropriate drug use (e.g., Covington & Thornton, 1995; Dillon, 1999; Sheperd & Hejna, 1995). They argue that formularies can serve to limit the use of both ineffective and suboptimal drugs and drugs with undesirable adverse effects. Examples include the removal of DESI drugs, amphetamines, and many combination products from formularies (Abramowitz & Fletcher, 1986). Limiting the number of drugs prescribed also reduces adverse drug events. They also argue that it can reduce drug costs— due to utilization of more established, cost-effective products— without compromising the quality of care provided. On the other hand, opponents argue that highly standardized drug formularies can cause unintended reductions in quality of care and increased costs, including reduced use of cost-effective drug therapies; resulting declines in health status; substitution of less effective, more toxic, or more expensive medications for 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. restricted agents; and increased utilization of costly physician or hospital care. The pros and cons of having standardized formulary system are listed in the table 1 below. Table 1 Potential A dvantages and Disadvantages o f a Formulary System (A dapted From Blum enthal et al., 2000) Advantages Disadvantages • Educates physicians and patients about drugs • Can reduce adverse drug events • Can enhance cost-effective prescribing • Can increase quality of care through evidence-based management of disease • Can assure use of quality drug products • Administrative burden and inconvenience to participants • May not be an effective drug list for 100% of the population served • Can decrease quality of care by denying access to needed medications • May demoralize employees by depriving the of autonomy and flexibility • May cause unwanted or unexpected outcomes due to discontinuation of drug therapy Whether formularies actually promote quality of care and cost-effectiveness is still under debate, because empirical studies have yielded mixed results. I thus review the debates in the context of Medicaid formularies as well as hospital and HMO formularies. 2.5. Medicaid Formularies The Medicaid program grants federal financial assistance to states for providing medical care to low-income populations. In 1980s and 1990s, many state governments experimented with the drug coverage of their Medicaid formularies: 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. some decreases the number of drugs, while others did the opposite. The following section reviews the studies examining their experience. In reviewing 11 studies on the impact of restricted Medicaid formularies, Jang (1988) found the assumption that restriction of specific drugs results in savings in the drug costs was questionable. In some cases, physicians prescribed restricted drugs anyhow.6 When physicians prescribed alternate drugs, numerous studies found alternate formulary drugs tended to be more expensive than the restricted drugs.7 Whether the substitutes were therapeutically appropriate is not conclusive in these studies, because they generally did not give patient-specific details on diagnosis and alternative drugs employed. (An exception is a study of relaxing restrictive state Medicaid formularies as required under the 1990 Omnibus Budget Reconciliation Act (OBRA90). This study found substantial use of medications deemed by 2 physician panels as having ‘no additional therapeutic benefit.’) Likewise whether formulary restrictions actually lead to cost-effectiveness in relation to the total cost of treating illness needs further study. Some studies suggested that major shifts in costs occurred due to restrictive formularies because 6 (Smith & McKercher, 1984) found that 30.7% o f the patients were prescribed drugs eliminated from the formulary and they were willing to pay for these drugs themselves. 7 (Smith & MacLayton, 1977) found that analgesics and minor tranquilizers were substituted in substantial amounts in Mississippi. They projected that substitution o f other analgesic drugs would increase statewide analgesic costs in excess o f $100,000 per year. The restrictions on minor tranquilizers saved $136,000, less than 1% o f the drug budget. Smith and Mckercher in Michigan found the alternative prescriptions to have significantly higher acquisition costs for average daily doses ($0.38 versus $0.18). The Hefner study in Louisiana established a 3.5 fold increase in nonprescription expenses over decreases in drug program expenditure. Hence substituted drugs or services tend to be more expensive (Moore et al., 1993). 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. more expensive hospitalization was substituted in lieu of pharmaceuticals.8 For example, Soumerai et al., (1991) found that a drug payment cap in Medicaid was linked to increased institutionalization among frail elderly persons. In Kozma et al., (1990) study of the S. Carolina Medicaid drug formulary, when drug coverage was expanded, the number of prescriptions, physician visits, and outpatient visits all increased, but the number of inpatient hospital admissions declined. Hefner (1979) examined the effects of withdrawing reimbursement for certain drugs in the Louisiana Medicaid program, and found that, while prescription drug expenditures declined by 14 percent, there was a 248 percent increase in hospital admissions related to heart disease.9 On the other hand, there are some who argued that the restrictions might lead to some increases in physician reimbursements and hospital payments, but not to the extent that the restrictions were cost-ineffective (Bloom & Jacobs, 1985) Because, as Soumerai et al. (1993) argued, many studies of Medicaid formularies are plagued by inadequate research design, it is hard to conclude definitely what the cost implications of standardized formularies are. Only three of these twelve studies, reviewed by Soumerai, even partially controlled for possible exogenous influences on study results; four had no comparison group at all; and four 8 Soumerai distinguishes the tradeoffs between restricted drugs and (1) substituted medications (first-order effects), and (2) substitution o f other health services (second-order effects) (Soumerai, Ross-Degnan, Avom, McLaughlin, & Choodnovskiy, 1991). 9 (Soumerai, Ross-Degnan, Fortess, & Abelson, 1993) argued this finding implausible, given that the hospitalization effects were observed for the entire population, even those never affected by the drag restriction. They suspected that unmeasured, nonformulary factors may be responsible for the changes in hospitalization rates. 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. additional studies with comparison groups only measured outcomes after formulary changes had already been introduced. Hence, Schweitzer and Shiota (1992) review of Medicaid formularies concluded, “there is not a strong case that formularies either raise or reduce Medicaid expenditures” (405). 2.6. Hospital and HMO Formularies There are only two studies I can find on the cost effectiveness of hospital and HMO formularies. Horn and her colleagues collected extensive patient-level information to identify drug and medical care use associated with HMO cost containment. In this prospective study, the statistical analysis suggested that restrictive formularies may result in increased costs and lower quality care, though some contented that the study was flawed methodologically (e.g.,Barnett, 1996; Curtiss, 1996; Kravitz & Romano, 1996). As for hospital formularies, Sloan and his colleagues (1993) studied the influence of hospital formulary restrictions on pharmacy charges, non-drug-related hospital charges, and on length of stay, using a survey of hospital drug policies and hospital discharge data from Washington State. They found that restricting availability of drugs reduced pharmacy charges, but these savings were offset by increases in other charges. None of the studies I reviewed so far— whether they dealt with Medicaid formularies or with hospital or HMO formularies—have not looked at the processes and outcomes of formulary restrictions. They simply looked at the impact of restricting the number of available drugs in the formulary on costs. However, I 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. propose that the fairness of the outcomes and processes in formulary restrictions do make a difference in determining the cost of patient care. I will explore this in the next chapter. 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3. THEORETICAL MODEL The theoretical model draws on literature in the field of bureaucratic organization and organizational justice. As diagramed below, the degree of drag standardization has direct effects on effectiveness moderated by procedural fairness. I will elaborate on them in the sections below. Figure 2 Theoretical Framework Controls Procedural fairness of standardization Physicians' attitudinai outcomes Organizational Cost Effectiveness (pharmacy cost effectiveness) D egree of Standardization (# of drugs available) 3.1. Arguments for the Effectiveness of Standardization A number of rationales cited in the previous chapter suggest that standardization leads to organizational effectiveness. Applying those rationales to the formulary, one might argue that increased standardization of drug availability is associated with lower pharmacy cost. A number of researchers have suggested that the number of drugs covered influences drug costs (Schweitzer, 2000). For example, Sloan and his colleagues found that hospital drug costs decreased as the number of 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. available drugs become smaller. A number of Medicaid studies also found that drug costs went down after more stringent formularies were imposed (Kozma et al., 1990). There may be a number of possible explanations for the cost effectiveness effect. First, it may be that, by excluding certain drugs from the formulary or making them difficult to order them, drug standardization can drive cost down. Second, hospitals with a highly standardized drug formulary can concentrate on purchasing a smaller number of drugs in large volume from pharmaceutical companies, thereby gaining bargaining power vis-a-vis pharmaceutical companies. Finally, the reduced variety of drugs— as a result of standardization—may give physicians and healthcare professionals an opportunity to learn more about particular drugs on formulary list. This increased experience with particular drugs decreases the number of adverse drug events and increases more effective use of the drugs by physicians and healthcare professionals, leading to decreased overall cost of care. Based on these reasoning, one might hypothesize: HI. All else being equal, an increase in standardization will positively influence organizational effectiveness. (See Figure 3 for a graphic representation). 3.2. Arguments Against the Effectiveness of Standardization As discussed earlier in the study, Merton (1958) suggested how strict conformance to rules generated rigidity. The consequence of this rigidity is that it promotes local effectiveness at the expense of systemic effectiveness (i.e., goal 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. displacement). However, in the case of drug standardization, I argue that the excessive standardization can lead to local as well as systemic ineffectiveness. There are two scenarios under which this could happen. First, when a hospital imposes excessive drug standardization on its affiliated physicians, given professional autonomy, many physicians may simply ignore it and work around it, or actively resist it. There is some evidence that physicians can, and often do, find ways to circumvent formulary restrictions if they do not agree with them. For example, according to Green et al. (1985) analysis of physicians’ orders under a restrictive formulary at a university hospital, physicians continued to prescribe non-formulary drugs, of which 65% were for drugs previously evaluated by the Pharmacy and Therapeutics (P&T) committee and denied admission to the formulary. In a different context, Wynia et al., (2000) found that a sizable minority (39 %) of physicians admitted manipulating reimbursement rules so that patients could receive care that the physicians perceived as necessary: we can plausibly imagine a similar scenario with regard to formulary restrictions. Furthermore, one study suggested that, when physicians felt that they were unfairly constrained by formularies policies, they were more likely to be unsatisfied with their work and prescribe drugs which are prone to cause adverse reactions, or which are deemed inappropriate by medical consensus (Melville, 1980). Other study also found that, the more a physician experiences feelings of frustration, the more medicine (s)he will prescribe, thereby increasing the chance of adverse drug events and providing inappropriate care (Grol et al., 1985). Under this scenario, either 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. because of physicians’ resistance to the standardization or because of their loss of morale, drug standardization will not lead to hospital effectiveness. Similarly, under an alternative scenario in which physicians comply with excessively restricted formularies, hospitals may not achieve cost effectiveness. Many of the formulary studies reviewed consistently shows that pharmacy cost savings which is achieved through drug formularies, would be offset by increase in other medical charges (e.g., Horn et al., 1996; Kozma et al., 1990; Sloan et al., 1993). When the degree of standardization is excessive, drug formularies will have little effect upon the total hospital cost or even increase hospital cost to the extent that the unavailability of requested drug products increases utilization of other (more costly) medical services such as hospital days, surgical procedures or physician visits. Strict formularies may raise non-drug-related costs of taking care of patients if excluded drugs tend to be cost-effective, even though they are expensive. Reliance on cheaper, but less effective drugs would actually increase drug treatment costs (Hefner, 1979). Similarly one might predict a similar result for therapeutic substitution. Although drugs in a therapeutic class share similar pharmacological and therapeutic properties, they are not identical, and their differences may cause problems when they are interchanged across a large group of patients, some of whom will inevitably have differing physiological status. Caves and Hurwicz (1991) and Lu and Comanor (1998) noted, therapeutic categories tend to be over-inclusive in that all drugs in the therapeutic class would not actually be used for the same ailment. Hence, exchange 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without the approval of the prescriber is opposed by organized medicine (AMA, 1994; American College of Physicians, 1990). Finally one may also examine the implication for price composition of drug formularies. When a formulary restricts more expensive drugs-which tend to be more therapeutically efficacious (Lu et al., 1998), strict standardization may lead to higher non-drug-related cost and longer length of stay. Therefore, whereas the bureaucratic perspective would predict that reduced drug charges will be directly translated into lowering non-drug-related costs and shortening the length of stay, the bureaucratic dysfunctional perspective suggests an alternative hypothesis: Hlb. All else being equal, an increase in standardization will negatively influence organizational effectiveness. (See Figure 3 for a graphic representation). 3.3. Organizational Justice Perspective Eisenberg (1986) draws a parallel between physicians and player-manager of an athletic team, in that both call the plays as well as working with others to carry them out. This means that the physician influences the cost and quality of medical care in two ways: “first, by organizing and directing the production process; and second, by providing some of the productive input. Although physicians’ fees represent only about 20 percent of health care costs, as much as 80 percent of expenditures for medical care are for services prescribed by physicians” (Eisenberg, 1986). 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A critical feature of the demand for prescription pharmaceuticals is that the end consumer, the patient, does not select the drug he or she will consume. Instead, physician picks the drug therapy and also chooses either the brand or generic form. Hellerstein (1998) in an examination of physician prescribing behavior, finds that “all of the evidence indicates that physicians are indeed an important agent in determining whether patients receive either trade-name or generic drugs.” Physicians are often uninformed as to prices and availability of generics. Hence I believe that the cooperation from physicians is a critical factor in achieving cost effectiveness goal of drug standardization. However, numerous studies have shown that physicians’ acceptance of formulary restrictions varies between individuals and healthcare organizations. For example, Donelan et al., (1997)found that 31 % of approximately 2000 physicians surveyed in 1995 reportedly disapproved of limitations on prescribing drugs. In other survey of VA hospital physicians, Glassman et al., (2001) found that 29% of physicians surveyed stated that the formulary at their hospital limited on their ability to provide the best quality of care. These studies hint that physicians’ acceptance of formularies might be critical for achieving medication safety and cost-effectiveness through formulary. Research on organizational justice has shown that (1) people react positively to an unfavorable decision when the decision is perceived as fair, and (2) such fairness assessments typically involve both process and outcome considerations (respectively, process justice and outcome justice). Our theoretical model therefore 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. postulates that organizational effectiveness (as measured by pharmacy cost, non drug-related cost, and length of stay) are moderated by the process and outcome fairness of drug standardization. In applying the procedural justice perspective to formulary standardization, the first step involves identifying relevant decision-making processes associated with formulary standardization. Dillon (1999) suggests the three key decision processes are: (1) the selection process whereby certain drugs are included in, and others excluded from, the formulary, (2) the interchange process, which determines when and how the pharmacy will replace a drug prescribed a physician with a pharmaceutically equivalent drug, and (3) the exception process, which governs the provision of drugs not included in the formulary Second, each of these procedures needs to satisfy certain criteria suggested by the procedures. According to Leventhal (1980), a procedure’s fairness is evaluated using six criteria: consistency, bias-suppression, accuracy, correctability, representativeness, and ethicality. In reviewing many procedural justice measures, Colquitt suggested five criteria: voice, accuracy, appeal, consistency, and ethicality. Several healthcare researchers have already alluded to the importance of some of these criteria in rationing medical resources; but to date, these criteria have not been the object of systematic empirical study. For example, Fins (1998) made several recommendations to ensure the effectiveness of restricted formularies: the process should be fair, open to outside scrutiny, and locally credible. He wrote, “The process must be explicit (i.e. information) with an open and accessible appeals 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. process (i.e., correctability). Guidelines should be written with the scientific input of medical professionals as well as the oversight of managed care members who will be affected by these policies (i.e., representation).” Similarly, Daniels and Sabin (2001) suggested that a process for determining limits to care in general (specifically, the inclusion of a particular drug in a formulary) should meet three conditions: first, the rationale behind a specific decision must be publicly available (i.e. information); second, the rationale should be based on relevant criteria (i.e. accurate, bias- suppression); and third, decisions must be open to revision and appeal (i.e. correctability). In this study, I will empirically examine some of those criteria in drug standardization process. I also hypothesize that physicians are more likely to react positively to a restricted formulary when they believe that the key procedures and processes of the formulary system-which include the selection of which drugs to include, which drugs to substitute, and when to make exceptions to the list— are perceived as fair. Hence, Procedural fairness assessments moderate the relationship between the degree of formulary restrictiveness and physicians’ reaction to formulary restrictions. That is, physicians are more likely to react positively to a restricted formulary when they believe that its process is perceived as fair. As diagrammed in Figure 3 ,1 contend that the relationship between two variables (procedural fairness and the degree of standardization) predicts the outcome of a third variable (effectiveness). Focusing on the process of how formalization (a concept closely related to standardization) is initiated, Gouldner 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (1957) theoretically distinguishes between formal rules that are imposed (which he calls ‘punishment-centered bureaucracy’), and those that are agreed upon (which he calls ‘representative bureaucracy’). While the imposed rules are associated with negative outcomes, rules agreed upon in representative bureaucracies are not. Similarly, Adler et al. (1996) argued that rules made and implemented by processes that are open to suggestion, transparent, and flexible, are enabling. And conversely, when those features are absent from the process of formalization, the outcome are perceived as bad rules. These authors suggest that the level of effectiveness achieved by standardization will vary, depending on the degree to which procedural fairness is employed in setting standards. In other words, standardization has a negative effect on some organizations and a positive effect on others depending on the fairness of the standardization process. Stated formally, these hypotheses are as follows: H2a: When procedural fairness is low, increased standardization will negatively influence organizational effectiveness. (See Figure 3 for a graphic representation). H2b: When procedural fairness is high, increased standardization will positively influence organizational effectiveness. (See Figure 3 for a graphic representation). 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It is possible that hospital authority can also explicitly encourage the use of formulary drugs by offering financial incentives to physicians. However, the legality of offering such arrangement has been controversial. Until recently, federal officials have outlawed most forms of gain-sharing based on concerns that some physicians would reduce the quality of care to reap extra payments, or steer sicker, more costly patients away from hospitals that do not offer the cost saving incentives. For example, in a July 1999 ruling, the office of inspector general, established at the Department of Health and Human Services by Congress in 1976, specifically prohibited gain-sharing, a practice that rewards physicians for reducing costs by giving them a percentage of the total savings achieved. But the inspector general’s position has softened considerably recently, and even approved a limited gain- sharing arrangement at an Atlanta hospital, which made a deal with a surgical group linking physician bonuses to cost savings associated with specific medical equipment and drugs used in the operating room (Romano, 2001). Hospitals that financially reward physicians for savings on drug-related costs would elicit more cooperation from physicians in prescribing cost-effective drugs. So we will control for this. 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3 Standardization and procedural fairness High Pro. J. Low Pro. J. I Standardization C D g ? Standardization © O Standardization Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4. RESEARCH DESIGN In this section, I, first, describe the sample, discuss the rationale for selecting the sample, and explain the data collection process. Second, I discuss the analytical methodology. Third, I describe the construction of the measures. I conclude the chapter with a description of preliminary data analysis, which includes an assessment of the representativeness of the sample. 4.1. RESEARCH CONTEXT 4.1.1. Choice of Samples and Sample Size The data on hospital effectiveness (pharmacy charge) was from inpatient discharge dataset maintained by State Health Agencies. The uniform data in it made possible comparative studies of health care services and the use and cost of these services across hospitals and geographical regions. Because each state in the US collects its own data, there are multiple datasets. However, I selected three datasets based on the following criteria.1 0 First, because I am interested in pharmacy cost, state datasets that do not collect pharmacy charges for hospital inpatients were excluded. Second, to increase the sample size of participating hospitals from each state, I excluded those datasets containing less than 200 hospitals. The datasets from 1 0 Initially, I decided to include four sets o f data from four states: Florida, Illinois, New York and Texas. However, during the research process, the Illinois health agency responsible for the data collection in that state was reorganized and the newly organized agency decided not to distribute the dataset to researchers until they obtain further state funding. As o f the time that this dissertation is written, the data set from Illinois is not available. 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Florida, New York, and Texas met these two conditions. Though these states may not be representative of the entire states in the U.S., I believe they represent a spectrum in the restrictive degree of formularies: While I do not have direct information on the relative degree of formulary restrictiveness in these states, given the prevalence and power of managed care organizations in Florida (26.3% of total population enrolled in HMOs) and New York (30.9% in HMOs), it seems reasonable to assume that hospital formularies at least in these two states are at least as restrictive as those in other states. In Sloan et al.’s (1993) study of formulary restrictions in hospitals, they limited the sample population to hospitals having 100 beds or more. They reasoned that, in small hospitals, the drugs on the formulary may reflect the prescribing preferences of a handful of physicians at the hospital, that the formulary therefore will be less restrictive, and that the drugs included may constrain prescribing little if at all. Consistent with this view, one study found that large hospitals tend to have more restrictive formularies than small ones according to a recent survey on the subject (Pedersen et al., 2001). Hence, I limited the sample population to hospitals having 100 beds or more in the states of Florida, New York, and Texas. In each of the states identified above, state health agencies have collected, from hospitals in its state, detailed data on the amount charged each patient, the primary and secondary sources of payment, the primary and secondary diagnoses, length of stay, and information on some demographic characteristics of the patient. The inpatient discharge datasets are issued to public for a fee annually and quarterly 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. each year. Because using an annual dataset would yield unmanageable amount of data for the analysis, I decided to use the data from quarterly report.1 1 Specifically, I decided to use the first quarter data from the year 2002 because the first quarter corresponds to the winter season known for a large number of the hospitalization of people with pneumonia.1 2 I have no reason to suspect the other two diseases- coronary artery disease and gastrointestinal diseases— are seasonal. Since patient information identifies the hospital from which (s)he was discharged, survey responses from the hospitals could be matched with the discharge data. The observational unit in our analysis is thus the individual patient. Original hospital surveys were conducted for the hospitals 2002 fiscal year, to combine with hospital discharge data. These surveys will supply the data on drug standardization, organizational justice, and other relevant variables. In addition, the American Hospital Association 2001 Annual Survey data was be used to measure other baseline hospital characteristics not included in the original hospital surveys: staffed hospital beds, teaching activity (interns and residents per bed), and others variables. The most recent HMO penetration rate and regional salary data from the Centers for Medicare and Medicaid Services were also used. 1 1 For example, Sloan et al. (1993) used the annual data from the Washington State Hospital Commission. Even after they selected a 20% sample o f patients in their chosen diagnostic categories, they were left with almost 14,000 for coronary artery disease and 4,000 for pulmonary infections. 1 2 The exception is the dataset from Texas. Because the 2002 dataset was unavailable during the time the study was conducted, I used the first quarter dataset from 2001 for the Texas data. 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.1.2. Data sources Figure 4 summarizes the research processes. Figure 4 Multiple research methods T h eo ry gen eration i i Step 1: Participant observation in a hospital S tep 2: 20 sem i- structured interview s o f hospital pharmacy directors & physicians •Developm ent o f model & hypotheses T h eo ry Testing Step 3: Survey o f 645 hospital pharm acy directors (N = 2 7 2 s 42% R esponse rate) •Formulary policies •Degree o f standardization •Procedural fairness JL S tep 4: A rchival data: •Inpatient D ischarge Data (FL , IL, N Y , TX ) •2001 A H A Survey •C M M S data •Pharmacy cost •Patient length o f stay •Patient-level control vars. •Hospital-level control vars. •Environment-level control vars. Archival data for this study came from multiple sources (See Appendix I for a list of data sources). Data on drug formulary practices and policies were derived from surveys sent to the pharmacy directors in hospitals with more than 100 beds in Florida, New York, and Texas. The measures for hospital effectiveness were pharmacy costs for patients having one of the following diagnoses: cardiovascular diseases, pneumonia, and gastrointestinal diseases. These patient-level outcome data came from the Hospital Inpatient Discharge Dataset collected and maintained by the 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. State Government Health Agency in each respective state.131 used data from the first quarter of 2002 because this quarter of the year corresponded to the winter season traditionally known for a large number of hospitalizations of patients with pneumonia. Survey responses from the hospitals could be matched with the discharge data because information on each patient in the data set identifies the hospital from which they were discharged. 4.1.3. Choice of Diagnostic Categories and Drugs I chose a sample of patients in each of three diagnostic categories using the ICD-9-CM code: coronary artery diseases (acute myocardial infarction, angina pectoris, and coronary atherosclerosis: 410-414); pneumonia (excluding pneumonias caused by tuberculosis or sexually transmitted disease: 480-486); and gastrointestinal diseases (gastroduodenal ulcer, gastritis, and duodenitis: 531-535). Because the ICD- 9-CM comprises approximately 12,000 diagnostic codes, it is often difficult to use to identify cases for disease-specific study. Instead, I used Clinical Classification Software (http://www.ahrq.gov/data/hcup/ccs.htm) to aggregate individual ICD-9- CM codes into clinically meaningful categories that grouped similar diagnoses and selected relevant codes for the purpose of this study. I had three main criteria in mind when choosing the diagnostic categories of coronary artery diseases, pneumonia, and gastro-intestinal diseases. First I included 1 3 All hospitals in each State, except those owned by the federal government and those exempted as rural providers, are required to submit claims on all discharged inpatients attended or treated by physicians to the appropriate State Health Data Collection Agencies. 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. diagnostic categories with high volumes, so that patient care for these diseases could be expected to significantly influence overall costs at each hospital. Second, according to several pharmacology references (Goodman, Gilman, Hardman, & Limbird, 2001), each of these disease categories has a relatively well-defined set of treatment drugs associated with them. By linking drugs with diseases as closely as possible, I wanted to assure that any changes in drug availability would be translated into cost outcomes. Finally, I chose therapeutic categories associated with classes of drugs that include expensive items, so that some hospitals could be expected to exclude many of them from their formularies.1 4 Many of the drugs used in the therapeutic categories chosen for this study are relatively expensive and/or are listed as controlled access drugs.1 5 4.1.4. Pilot Test To ensure that the survey instrument is one that respondents can consistently understand and answer, I formally tested the instrument by using a method commonly referred to as the cognitive laboratory interview (Forysyth & Lessler, 1991; Lessler & Tourangeau, 1989). The goal of such interviews was to identify questions that are not consistently understood or that lead to ambiguous responses that cannot be meaningfully compared. Previous studies have shown that hospital 141 want to thank Carol Taketomo, Pharmacy Director at Children’s Hospital o f Los Angeles, for her guidance in selecting appropriate drugs for the study. 1 5 According to Scott-Levin survey, the 10 therapeutic classes with the highest number o f HMO plans reporting a restriction include antifungals, migraine treatments, select pain medications, antidepressants, cholesterol reducers, alpha-blockers, antiulcers/ulcer combinations, calcium channel blockers, sympathomimetic antiasthmatics and macrolides. Many o f the drugs in this study belong to the cited categories. 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. pharmacists were generally open to publicly sharing their formulary restriction information for research purposes (e.g., Green et al., 1985; Sloan et al., 1993). I also found many of them willing to help. I conducted a round of pilot interviews with twenty hospital pharmacists and faculty in pharmacy schools to ensure that the survey instrument was one that respondents could consistently understand and answer. Recruited pharmacists were asked to fill out the questionnaire as (s)he would if (s)he is part of a survey; then I led a discussion about the questionnaire. A number of topics was raised: (1) whether the instructions and the questions are clear; and (2) whether there are any problems in understanding what kind of responses are being elicited, or in responding to the questions as posed. Sometimes respondents were asked to “think aloud” while they were preparing their answers. In other cases, respondents were asked a set of questions about the way they understand each question and about issues related to their responses. The insights gathered from these interviews were incorporated as revisions to the questionnaire. 4.1.5. Survey Data Collection I found 469 hospitals fitting the inclusion criteria from the Directory of A m erican Hospital Association, and sent eight page surveys (attached in appendix) to the Director of Pharmacy in each hospital.. A potential difficulty of the proposed research methods is that the director of pharmacy might not respond to our 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. questionnaire (i.e., non-response error). This risk is known to be particularly high for mail surveys of healthcare professionals. I employed a number of measures to obtain a good response rate. The cover letter and survey were highly “professional” in appearance. The cover letter explained the purpose of the study and its procedures and our need for hospitals to participate. Respondents were assured complete confidentiality. Respondents were provided a pre-stamped return envelope. They were also offered the option of receiving a report at the conclusion of the study, which would help them to access their formulary practices. To achieve a high return rate, I conducted four mailings (i.e. the initial mailing and three reminders) as recommended by survey experts (Dillman, 1978; Mangione, 1995). Each of these mailings was spaced two weeks apart so that the total mailing period would be extended over approximately nine weeks. I put a complete package (respondent letter, questionnaire, and return envelope) in the first and third mailings. The second and fourth mailings included a postcard reminder. For those who did not respond after the final mailing, I appealed one last time by telephone. I kept track of the questionnaires being returned by placing a code number on the survey form.1 6 As shown in table 2, the response rate was 42.33%. 1 6 We will make it clear to respondents in the cover letter that we promise confidentiality, but not anonymity, and that their responses will be identified with a code number on the survey form. 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2 Number of Surveys Mailed and Responded by State State Florida New York Texas Total Mailed 120 184 165 469 Responded 57 77 65 199 Response rate 47.5% 41.84% 39.39% 42.43% 4.2. Analysis Methodology: Testing moderating effect To assess support for the moderation hypothesis (H2a and H2b), all variables (with the exception of dummy variables) are centered prior to calculating the interactions, as recommended by Aiken and West (1991) and by Jaccard, Turrisi, and Wan (1990). Centering consists of subtracting the sample mean from each independent variable. The adjusted variables each have a mean of zero, but their sample distribution remains unchanged. This procedure serves two purposes: (a) it alleviates problems associated with multicollinearity between the interaction terms and their constituent variables, and (b) it makes for meaningful interpretation of the main effects in the presence of interaction terms. I compute two interaction terms by multiplying the formulary standardization score with the procedural fairness score. The interaction term is entered in a separate step after the main terms (standardization and procedural fairness) have already been entered. If the addition of the interaction terms results in a statistically significant improvement over the regression model containing the main terms, then this would indicate support for the interaction model. 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3. Measures 4.3.1 Reliability and validity issues. Because there are no pre-existing scales specifically designed to measure the various aspects of formulary standardization, I either adapted several existing scales for our purpose or designed new ones. To enhance the reliability of these scales, I assessed how well items from a given scale indexed a common factor using an exploratory factor analysis. Items having the highest factor loadings will be combined into scales. 4.3.2. Hospital Effectiveness The relative cost position per specific treatment is important for the success of a hospital in an increasing competitive environment. The extended use of fixed payments for hospital care through DRGs, per diems, and capitation has increasingly elevated cost competition for hospitals as the primary means for financial success. Hospitals now must actively compete on price for patient volumes: those with lower relative costs can price more aggressively and more easily gain managed care contracts. Hence, the dependent variable used in this study is pharmacy cost. Pharmacy charge is a variable designed to capture the level of hospital resource use. From the State Inpatient Database, I use pharmacy charge data for patients in the three previously mentioned diagnostic categories: coronary artery diseases, gastrointestinal diseases, and pneumonia. In this study, I used pharmacy charge data as a proxy for pharmacy cost. Charges are the total dollars billed for patient services during the period and do not 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reflect reimbursement. Costs are the accumulated operational expenses for the period. The latter data are confidential and not made public. Hence, the state inpatient discharge data, from which I draw the cost measures, report charge, but not cost. Some researchers have used the cost-to-charge ratio (CCR) to calculate the underlying costs indirectly. Hospital wide cost-to-charge ratios (CCR) is reported on the Medicare Worksheet, and is determined by taking the total charges billed to Medicare and dividing it by the non-drug-related cost of providing care to Medicare patients. Dranove’s (1995) work indicates that the costs obtained in this way are generally within 10 percent of costs obtained from calculating costs for individual units of service and then aggregating them to obtain total hospital costs. However, I should also note that using cost data, instead of charge data, may pose a new set of problems. Sloan et al. objected to using cost data for two reasons: (1) some part of profit within charge amount is a legitimate part of economic cost, and (2) applying a hospital wide cost-to-charge ratio to specific cases does not recognize case-specific cost variation. At the same time, they also recognize that using charge data is not without problems: Because an increasingly smaller percentage of patients actually pay full billed charges, and hospital charges reflects different discount rates by some payers, charge data are only a rough proxy for real cost. To address this problem, they included a number of explanatory variables to account for systematic differences among hospitals in the cost-to-charge ratio, such as the hospital’s payer mix, ownership, teaching status, and metropolitan location. 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In this study, I used the both methods and did not find much difference. In fact, the pharmacy charge data were highly correlated with cost data calculated by using CCR. Thus I reported only the results using pharmacy charge data. Consistent with previous research, a logarithmic transformation was used to normalize the dependent variables (Bradbury, Golec, & Steams, 1991; Bums & Wholey, 1991; Johnson, Dowd, Morris, & Lurie, 1989). This log transformation of dependent variables improves the results of a test of normality to a near normal distribution. Separate equations were estimated for each of the three major disease categories, thus controlling for types of diseases. By analyzing disease categories separately, I can examine how patient and hospital characteristics differ by major disease categories in their associations with the dependent variables. 4.3.3. Formulary characteristics 4.3.3.1. Standardization In reviewing the previous measures of drug standardization, I found that they suffered from two major shortcomings. First, they were often measured in binary terms: either standardized or nonstandardized. For example, in Hanson and Shepherd’s study (1994), a formulary was judged as restrictive if prior approval for non-formulary dmgs was needed. However, as argued by Schweitzer (2000), formulary restrictions should be viewed along a continuum ranging from nonrestrictive to highly restrictive. Second, much of the previous research suffered from the exclusive use of subjective estimates of standardization. For example, some 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. researchers operationalized formulary restrictions by asking physicians to rate the restrictiveness of the drug policies in their hospitals (see Glassman et al., 2001; Nash, Shulkin, Owerbach, & Owerbach, 1992). However, according to a study done by Hasty, Schrager, & Wrenn (1999), emergency medicine physicians significantly overestimated the restrictions on their antibiotic prescribing practices, when their perceptions about formulary restriction was compared with objective measures. Hence, we need both objective and subjective measures of standardization, since they may differ. However, the previous measures do not distinguish multiple dimensions of standardization. To tease apart the various impacts of dimensions of standardization, I propose to measure each dimension separately. To measure objective dimensions of drug standardization, I examined the number of drugs in formularies and the number of drug categories therapeutically substitutable in each hospital. For subjective dimension, I also asked the director of pharmacy about their perception of the strength of enforcement. 4.3.3.1.1. Size of drug formulary To measure the formulary standardization, I use a new measure developed by Horn et al.(1996). I created an “availability ” variable to be the number of drugs available on the hospital formulary for specified therapeutic categories. “Formulary standardization” would be then defined as: formulary standardization = 1 minus availability. Thus, high formulary standardization is equivalent to low availability. In determining the total number of drugs for specified therapeutic categories, I examined the following sets of drug categories corresponding to the diagnoses 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. being analyzed: ace inhibitors, beta-blockers, nitrates, and calcium channel blockers for coronary artery diseases; miscellaneous beta-lactam antibiotics, macrolides, tetracyclines, quinolones, cephalosporins, aminoglycosides, and penicillins for pneumonia; and H2 -antagonists and proton pump inhibitors for gastrointestinal diseases. The availability of these drugs in hospital formularies is used to calculate drug standardization. Since I am measuring formulary standardization in three disease categories, each disease category will have its own degree of standardization. 43.3.1.2. Therapeutic Interchange. Therapeutic interchange involves substitution of a drug by a pharmacist that is therapeutically-equivalent but not chemically-identical. That is, it involves substitution of drug products that have different chemical structure but belong to the same pharmacological or therapeutic class. It is allowed or practiced in almost all US hospitals: Nash et al. (1992) found about the two thirds (76%) of the hospital surveyed practice it; Robinson & Casalino (1996) found the rate to be about 74.5%; Sloan, Whetten-Goldstein, & Wilson (1997) found the rate to be 79%; and Mannebach et al. (1999) found that 69% of hospitals had formal therapeutic interchange policies. In many hospitals, physicians agree to abide by hospital policies and procedures-including therapeutic substitution-when joining the hospital staff. Studies have found that the level of drug substitution as well as the number of categories involved in substitution reduced the number of drug items carried (Daniels & Wertheimer, 1982). 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To measure the extent to which therapeutic interchange is done in a hospital, I counted the number of drug classes that have a formal therapeutic interchange policy. 4.3.3.1.3. Strength of Enforcement. A standard can exist ceremonially, but may not command any compliance. Gouldner called this “mock bureaucracy”: rules come from outside organization and are generally ignored by organizational members. Under such circumstances, it’s hard to expect standards to have a real financial impact on organizations. On the other hand, if a standard is vigorously enforced, it is more likely to influence effectiveness outcomes. For example, according to Conrad et al. (1996), hospitals that provide written resource consumption reports to their physicians in general as well as for specific condition have lower cost than those that do not. To measure how vigorous hospital implements drug standardization, I asked the degree to which the director of pharmacy agrees with the following statements (All items are assessed with five-point Likert scales with anchors of 1 = “to a small extent” and 5 = “to a large extent.”): The hospital evaluates prescriber adherence to medication-use policies The hospital monitors prescribers for use of non-formulary drugs The hospital notifies prescribers when they don't comply with formulary policies The hospital uses the trend data of individual physicians on drug utilization for medical staff credentialing 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43.3.1.4. Overall standardization index (OSI) This index is constructed in both an additive form, as a simple average of the three component measures of standardization, and a multiplicative form, as the product of the three measures. For both forms, a low OSI score indicates a low degree of standardization. Before adding or multiplying component measures, I applied a linear transformation for each, such that each of them becomes a positive number. In that way, I could meaningfully interpret a sign of multiplicative terms. 4.3.3.1.5. Factor analysis of standardization measures The factor analysis of all three component measures can be seen in Table 3. All three factors (strength of enforcement, # of drugs in formulary, and the number of drug categories substitutable) turn out to be all separate dimensions of standardization in this factor analysis. They loaded on separate factors which imply that hospitals vary in their relative emphases on each dimension. 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3 Factor analysis of standardization measures Strength of enforcement # of drugs in formulary (reversed) # of drug categories substitutable The hospital monitors prescribers for use of non formulary drugs .836 .105 .093 The hospital evaluates prescriber adherence to medication-use policies .802 -.005 -.022 The hospital notifies prescribers when they don't comply with formulary policies .773 .053 .042 The hospital uses the trend data of individual physicians on drug utilization for medical staff credentialing .585 -.021 .034 # of cardiovascular drugs (reversed) .134 .896 -.002 #of pneumonia drugs (reversed) .026 .878 .109 # of gastrointestinal drugs (reversed) -.083 .581 .400 #of pneumonia drug categories therapeutically substitutable .095 .036 .782 #of cardiovascular drug categories therapeutically substitutable .206 .192 .739 #of gastrointestinal drug categories therapeutically substitutable -.108 .072 .723 Alpha 0.76 n.a. n.a. Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization, a Rotation converged in 4 iterations. 4.33.2. Procedural Fairness In organizational research, many researchers have echoed the call for a reliable standardized instrument with which to measure perceptions of various dimensions of fairness for some time (e.g., Greenberg, 1990; Lind et al., 1988); (see Colquitt, 2001 for a review). Despite efforts by several researchers (e.g., Beugre, 1996; Colquitt, 2001) to create standardized instrument, the measures of fairness seem to be context-specific. According to Greenberg (1996), “what makes a set of questions appropriate in one context may not make them equally appropriate in another. Questions about justice should be carefully matched to the context of interest. . . ” (402). Consistent with Greenberg’s recommendation, I tailored the 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. existing measures of fairness to accommodate the unique features of formulary decisions. Fairness has been examined at different levels of analysis: the individual level (e.g., most of organizational justice studies), the subunit level (Mossholder, Bennett, & Martin, 1998; Naumann & Bennett, 2000), and the organizational level (Sheppard et al., 1992)). The appropriate level of analysis is determined by the purpose of a study. In this study, I am mainly interested in fairness at organizational level. In organizational behavior research, many researchers have used a variety of standardized instruments to measure perceptions of various dimensions of fairness (Colquitt, 2001; Greenberg, 1990; Lind et al., 1988). While these tools are very useful in measuring individuals’ subjective perception of fairness, I had to tailor the existing measures in order to measure the structural, and thus more objective, features of procedural fairness which are the focus of this study. A key variable of interest in this study is the procedural fairness of drug standardization. Procedural fairness was thought to be assessed by comparing the process one experiences to several generalizable procedural “rules.” If the rules were upheld, the procedure was just. In the context of drug standardization, the rules included trust (e.g., decision makers are respected and trusted for their judgments), transparency (e.g., decision procedure is clear and public), input (e.g., all subgroups in the population affected by the decision are heard from), and due process (e.g., appeal procedures exist for correcting bad outcomes). Again, because there is no pre- 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. existing scale specifically designed to measure the procedural fairness of drug standardization, I created new measures. To measure this construct, I developed four component indices—formal review, flexibility, responsiveness, and due process-and an overall procedural fairness index.1 7 Each of the four component indices is composed of multiple variables, described below. All individual items are listed in Table 4. The Cronbach Alpha coefficients are above 0.6 and, thus, indicate acceptable reliability of the measures. 171 do not measure interactional justice here, though it is theoretically an integral part o f process fairness. The main reason is that we can not tap into the subjective perception o f individuals subject to drug standardization, since our study is at the organizational level. 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4 Factor analysis of procedural fairness measures Formal review Responsiv eness Due process Flexibility The P&T Committee uses a formal economic analysis or cost impact evaluation as part of the formulary review .874 .147 -.014 .039 The P&T Committee uses quantitative analytic techniques as part of the formulary review .C3S .019 .010 .053 The P&T Committee uses peer-reviewed journal articles as part of the formulary review ,621 .267 .225 -.242 The P&T Committee receives comments and recommendations from prescribers as part of the formulary review -.063 -.036 .199 The P&T Committee considers local standards of medical practice as part of the formulary review .076 .66? -.102 .183 The P&T Committee solicits inputs from all specialties and subspecialties as part of the formulary review .102 .648 -.090 -.155 The P&T Committee provides to practitioners a full account of procedures and criteria for formulary decisions .338 .56? .281 -.098 The P&T Committee and its members are respected by their colleagues for their knowledge and judgments .319 502 .127 .053 Does your hospital have an internal appeals process if a request for a formulary addition is denied? .032 -.039 y .s:i -.026 Does your hospital have an internal appeals process if a request for a non-formulary drug is denied? .092 ■ 015 .003 The inpatient pharmacy makes off-contract emergency purchases of drugs -.076 .069 .050 .835 The inpatient pharmacy stocks some non formulary drugs .052 .076 -.072 .810 Alpha 0.74 0.66 0.89 0.61 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization, a Rotation converged in 6 iterations. 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3.3.2.1. Formal review. This measures the extent to which formulary decisions are made according to formal standards, as opposed to personal preferences of P&T committee members. All items are assessed with five-point Likert scales with anchors of 1 = “to a small extent” and 5 = “to a large extent.” It consists of three items. 4.3.3.2.2. Flexibility. This measures the extent to which exceptions to standardized rules are made. All items are assessed with five-point Likert scales with anchors of 1 = “to a small extent” and 5 = “to a large extent.” It consists of two items. 4.3.3.2.3. Responsiveness. This index measures the extent to which formulary decision makers are responsive to physicians. All items are assessed with five-point Likert scales with anchors of 1 = “to a small extent” and 5 = “to a large extent.” It consists of five items. 4.3.3.2.4. Due process. If certain formulary decisions are not acceptable to physicians, some hospitals allow physicians to appeal the decisions to a higher authority. To measure whether the presence of appeal mechanism by which decisions on nonformulary addition and request can be reconsidered, I use two items (0 = absence of due process, 1= presence of due process). 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3.3.2.5. Procedural Fairness Index (PFI). Similar to Overall Standardization Index, Procedural Fairness Index was constructed in both an additive form and a multiplicative form. Before being additively or multiplicatively combined to form overall indices, component measures of fairness were standardized by conversion to z-scores. Each component measure in an index receives equal weight, because I feel that there is no clear conceptual basis for assigning differential weights. For ease of interpretation, I will apply a linear transformation to the summed z-scores for each component measure, such that 1 is the hospital with the lowest score in the sample and 2 is the hospital with the highest score. 4.3.3.3. Formulary Incentive System To measure the presence of financial incentive, I asked whether the hospital provide financial incentives (e.g., buying new equipment) to clinical programs or departments for hospital cost savings. This is measured in Yes or No. 4.3.3.4. Nonphysician votes in P&T committee Hospitals vary in terms of which professional groups can serve in the P&T committee, and more importantly which professional groups can vote in formulary decisions. Traditionally, P&T committees consist exclusively of physicians and, while nonphysician groups, such as pharmacists and nurses, attend the meeting, they were not given the voting privilege. However, increasingly this trend is changing and 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. many hospitals these days do give voting right to those nonphysician groups. I counted the number of nonphysicians members with voting right in a P&T committee. 4.3.4. Patient Characteristics. The main patient factor influencing costs and length of stay is related to patients’ health status, insurance, and demographic characteristics. 4.3.4.I. Patients’ health status. Because the relationship between the medical intervention and the outcome may differ depending on the patient’s severity of illness, I included severity of illness measures. For example, certain medication may be useful for improving the well being of stroke patients, but it may not be equally useful for all stroke patients. Patients who are completely comatose may benefit less from a medication than patients who are paralyzed but cognitively conscious. According to Kane (1997), there are three different concepts used to describe severity: severity of disease, comorbidity, and severity of illness(for a review of measures, (see Iezzoni, 1997). Severity of disease usually refers to the severity and importance of a particular diagnosis (often the principal diagnosis) to the patient’s risk, regardless of a patient’s other health conditions. Comorbidity, on the other hand, refers to additional diagnoses, not the principal diagnosis, for a patient. Finally, severity of illness is a combined score for a patient’s overall level of illness. Hence, it is a function of both the severity of disease and comorbidity. In this study, rather 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. than measuring severity of diseases and comorbidity separately and then combining them to create a measure for severity of illness, I use an alternative measure developed by Sloan and his colleagues. Sloan et al (1993) developed a number of proxy measures to measure a severity of disease—patient died,1 8 patient admitted to intensive (or coronary care) unit during the stay, number of listed diagnoses, and DRG case weight.1 9 Length of stay. Longer hospital stays are likely to result in higher charges. The State Inpatient Database also reports patients' hospital length of stay. It is based on the number of patient days accumulated from admission to discharge. A stay of less than one day (patient admission and discharge on the same day) is counted as one day. For patients admitted and discharged on different days, length of stay or the number of days of care is computed by counting all days from (and including) the date of admission to (but not including) the date of discharge. 4.3.4.2. Insurance. Patients’ insurance coverage has been found to be significantly associated with patients’ hospital resource use (Ellis & McGuire, 1986; Lave & Frank, 1990) because the nature of the incentives associated with different types of insurance 1 8 It is possible that, in contrast to other severity measures, this variable can have negative influence on length of stay, though deaths can incur higher costs. 1 9 The Health Care Financing Agency (CENTERS FOR MEDICARE & MEDICAID SERVICES) maintains a list o f relative value weights (RVUs) for inpatient hospitalization. These RVUs, known as the Diagnostic Related Group (DRG) weight, are based on the primary diagnosis when the patient enters the hospital. There were approximately 511 DRGs as o f 1999. Some o f the DRG groups have been discontinued over time, while others have been added. This reflects changes in morbidity and technology. Therefore, it is important to make sure that the DRG weight file is the same year as the utilization data being analyzed. The DRG relative value weight can be found on THE CENTERS FOR MEDICARE & MEDICAID SERVICES' web site. 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. affects providers’ decisions regarding patients’ hospital resource use. For example, if patients pay full charges or fee-for-service as in the pre-DRG era, there is little financial incentive for hospitals to reduce patients’ hospital resource use. However, under a fixed-case-based prospective payment system (or payment per case as in DRG-based Medicare PPS), a hospital's net revenue is reduced for each day of care provided by the marginal cost of a day of care. Hospitals have a strong financial incentive (i.e., to earn "profits") to reduce patients’ hospital resource use. Also, when payment is based on capitation or number of enrollees (as in the case of health maintenance organizations HMOs), the incentive is to reduce patients’ hospital resource use as much as possible because each reduced inpatient day represents savings on variable cost incurred during that day. Payment source was coded as five dummies: Workers' Compensation, Medicare, Medicaid, other government, and private/commercial insurance. 4.3.4.3. Demographic information. In addition to health status and insurance, other patient characteristics expected to influence patients’ hospital resource use are gender, age, race, and marital status (Lave et al., 1990). Men and women have certain physiological differences as well as different pain tolerance levels, which may influence patients’ hospital resource use. Age consistently has been associated with severity of illness and is included in the analysis as one of the proxies for patient case mix” (Bums et 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. al., 1991; Goldfarb, Hombrook, & Higgins, 1983). Race or ethnicity often is used as a proxy for socioeconomic status, which is significantly related to health status (Williams, Lavizzo-Mourey, & Warren, 1994). Racial variations in health status result primarily from variations among races in exposure and vulnerability to behavioral, psychosocial, material, and environmental risk factors and resources. Marital status reflects the presence of social support. Married people are more likely to have access to social support compared with nonmarried people, and hence have less need for extended hospital care. These patient characteristics will be included in our analysis. 4.3.5. Hospital Characteristics A variety of hospital factors influence patient pharmacy charge. Listed are some common reasons why hospital charge varies. 4.3.5.1. Ownership. In terms of ownership, private hospitals, generally more sensitive financially, are expected to be more responsive to the incentives embedded in insurance systems (Anderson & Pulcins, 1992). I reasoned that for profit hospitals would be more likely to be cost efficient than not-for-profit voluntary hospitals because of the formers' greater strategic flexibility, need to capture attractive market positions, and sensitivity to the competitive forces that drive innovation. 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To control for the influence of ownership of the hospital on effectiveness outcomes, I set up a dummy ownership variable: private for-profit, public (state or local government), and private not-for-profit ownership. Private nonprofit hospitals were controlled by nongovernmental, not-for-profit organizations, including religious organizations, community cooperative hospitals, and hospitals operated by fraternal societies. Public hospitals were controlled by a nonfederal government agency or the department of a state, county, city, city-county, or local hospital district or authority. Investor-owned hospitals were controlled on a for-profit basis by an individual, partnership, or profit-making corporation. 4.3.5.2. Urban/Rural Urban hospitals are located inside a Metropolitan Statistical Area, which is a geographically defined, integrated social and economic unit with a large population base. Rural hospitals are located outside a Metropolitan Statistical Area. 4.3.5.3. Geographic region Regional variations in length of stay and costs may be due to differences in the severity of disease for those admitted to hospitals, underlying regional differences in the health status of the population, as well as differences in physician practice patterns (Manheim, Feinglass, Shortell, & Hughes, 1992). To control for them, I set dummy state variables. 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Within each state, I also group hospitals in terms of MSA (Metropolitan Statistical Area). According to US census bureau, the general concept of a metropolitan area is that of a large population nucleus, together with adjacent communities having a high degree of social and economic integration with that core. Metropolitan areas comprise one or more entire counties, except in New England, where cities and towns are the basic geographic units. Healthcare research typically defines hospital market competition in terms of MSA. 4.3.5.4. Teaching institution A hospital is considered to be a teaching institution if it had either an American Medical Association-approved residency program or membership in the Council of Teaching Hospitals. Teaching hospitals are larger and located in large, urban areas. They offer more specialized services and provide more uncompensated care than nonteaching hospitals. Rural hospitals are not split according to teaching status because rural teaching hospitals are rare. The data came from the AHA survey. 4.3.5.5. Multi-hospital system membership Membership in hospital systems reflects network and information ties that may affect organizational cost positions. Network ties may generate institutional expectation and normative pressures (Galaskiewicz, 1989) and increase flows of information about innovative changes (Bums & Wholey, 1993). Such organizational characteristics may directly or indirectly affect the cost position of a hospital through 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. increasing institutional pressures and/or facilitating the change process. It is measured by a dummy variable with a value of one if the focal hospital was a member of a multi-hospital system and zero otherwise. 4.3.5.6. Hospital size Smaller hospitals generally are under greater financial pressure than larger hospitals, and are expected to reduce patients’ hospital resource use more drastically. They also have more flexibility to do so. Hospital size is also correlated significantly with patients' case mix with larger hospitals, including proportionally more sicker patients even for the same diagnosis. Hence, according to one study (Olsen & Goodman, 1977), hospital size is correlated with drug costs per patient day. To control for the effect, I measure hospital size as the average number of patient beds. 4.3.5.7. Hospital case mix index The average DRG weight for all of a hospital’s Medicare volume is called the case mix index (CMI). This index is useful in analysis since it indicates the relative severity of a patient population. When making comparisons among various hospitals, the case mix index can be used to adjust indicators such as average charges. The data came from The Medicare Prospective Payment data set maintained by the Centers for Medicare and Medicaid Services. In addition to measuring the average DRG weight for all of a hospital’s Medicare volume, I also measured the average DRG weight for specific diseases 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. studied. This measure would give a better indication of the specific case mix that a hospital has on the specific diseases. 4.3.5.8. Area Wage Index The wage index measures relative differences in the average hourly wage for the hospitals in each labor market area compared to the national average hourly wage. The data come from Centers of Medicare and Medicaid Services. 4.3.5.9. Insurance mix To reflect the insurance impact at the hospital level, I created several measures that show the percent of patients with primary source of payment at the time of the discharge (%Medicare, %Medicaid, %goverment plans, %commerical plans, and %worker compensation). These variables are calculated by dividing the number of patients with respective insurances by the total number of patients hospitalized. 4.4. Preliminary data analysis and descriptive statistics 4.4.1. Descriptive Statistics Table 10,11, and 12 show descriptive statistics at the patient-level. The level of pharmacy charges across three diseases was not very different. Death rate for pneumonia was a bit higher than other two diseases. However, the use oflCU was significantly higher for cardiovascular diseases. 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.1.1. The number of drugs Table 16 shows the wide variation among hospitals in terms of the total number of drugs carried for each disease. The variations of cardiovascular and pneumonia drugs across hospitals are much larger than that of ulcer drugs. This may due to a small initial number of drugs available to treat ulcer-related diseases. Table 5 Descriptive statistics of the number of drugs # of heart drugs it of pneumonia drugs # of ulcer drugs N Valid 186 185 185 Missing 4 5 5 Mean 30.23 38.13 3.12 Std. Deviation 8.480 7.438 1.422 Minimum 15 21 1 Maximum 53 63 8 The following three figures (5, 6, and 7) show the distribution of the number of drugs in hospitals. 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 5 Histogram of the number of cardiovascular drugs on formulary sum of all heart drugs Std. Dev = 8.48 Mean = 30.2 N = 186.00 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 17.5 22.5 27.5 32.5 37.5 42.5 47.5 52.5 sum of all heart drugs Figure 6 Histogram of the number of pneumonia drugs on formulary SUM PNMN Std. Dev = 7.44 Mean = 38.1 N = 185.00 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 SUM PNMN Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 7 Histogram of the number of ulcer drugs on formulary SUMJJLCE 1 0 0 - ,--------------- 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 SUMJJLCE 4.4.1.2. Organizational context of drug standardization The wide variation shown by hospitals in the number of drugs carried raises a question of what drives such diversity. While this question can be studied from many perspective, in this section, I examine the organizational characteristics that are associated with drug standardization: in particular, I look at ownership, size, urban location, system membership, teaching status, and case mix. I find some systematic differences across those categories. For example, non profit, teaching, urban hospitals with lower case-mix are more likely to carry a smaller number of cardiovascular drugs relative to other categories of hospitals. The literature is not clear why this is the case, and we need further research on this topic. These categories of ospitals do not differ significantly in terms of the number of therapeutic exchanges practiced or the degree of enforcing drug restrictions. 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 6 Organizational context of standardization Group # o f Heart drugs # of Pneumonia drugs # o f Ulcer drugs Enforcem ent # of therapeutic exchange among heart drugs # of therapeutic exchange among pneumonia drugs #of therapeutic exchange in ulcer drugs Ownership 1 .Public (24) 31.71 39.25 3.13 10.75 1.38 1.79 1.83 2.Nonprofit (123) 29.24 (8.08) 37.96 2.99 9.64 1.42 1.72 1.81 3.Profit (39) 32.44 37.97 3.55 10.12 1.26 2.05 1.74 F-statistic (186) + Size 1.<218 (93) 31 38.98 3.2 9.66 1.3 1.78 1.77 2.>=218 (93) 29.35 37.27 3.04 10.11 1.46 1.81 1.83 F-statistic (N) Urban l.Metropl(168) 28.54 37.19 2.92 1.49 1.81 1.81 9.93 2.Urban (66) 33.12 39.84 3.43 1.24 1.82 1.84 10.07 3 .Rural (30) 33.24 39.47 3.59 1.06 1.65 1.65 9 F-statistic (N) + System Membership O.No (95) 29.37 37.26 3.09 10.05 1.37 1.77 1.77 l.Yes (91) 31.12 39.04 3.16 9.72 1.40 1.82 1.84 F-statistic (186) Case Mix l.<Median (93) 28.76 36.75 3.02 9.87 1.4 1.81 1.83 2. ^Median (93) 31.69 39.49 3.23 9.91 1.37 1.78 1.77 F-statistic (N) * Teaching Hospital O.No (159) 31.07 38.17 3.18 9.84 1.35 1.82 1.81 l.Yes (27) 25.22 37.89 2.81 10.14 1.56 1.63 1.74 F-statistic (186) Post Hoc test: LSD +p<0.1, * p<0.05, **p<0.01, ***p<0.001 00 to 4.4.1.3. Organizational context of formulary practices. In addition to standardization, I also looked at whether formulary management practices vary across hospitals. Large teaching hospitals with high case- mix tend to have formal formulary review process. Also teaching hospitals are more responsive to physicians’ needs in formulary review and have due process when physicians do not agree with formulary decisions from P&T Committee. Table 7 formulary practice and organizational context Group Formal review Flexibility Responsiveness Due process Ownership l.Public(24) 9.92 5.63 18.41 .58 2.Nonprofit(123) 10.11 5.43 17.94 .72 3,Profit(39) 9.97 5.34 17.13 .46 F-statistic(186) Size 1.<218(93) 9.68 5.33 17.46 .63 2.>=218(93) 10.45 5.54 18.21 .66 F-statistic(N) * Urban l.Metropl(168) 10.14 5.47 18.02 .63 2.Urban(66) 10.02 5.45 17.65 .65 3.Rural(30) 9.59 5.12 17.06 .76 F-statistic(N) System Membership O.No(95) 9.87 5.32 17.55 .6 l.Yes(91) 10.26 5.56 18.13 .7 F-statistic(186) Case Mix l.<Median(93) 9.72 5.38 17.78 .53 2. SMedian(93) 10.4 5.49 17.88 .76 F-statistic(N) + + Teaching Hospital O.No(159) 9.84 5.42 17.59 .6 l.Yes(27) 11.37 5.52 19.32 .93 F-statistic(186) ** ■ u * + Post Hoc test: LSD +p<0.1, * p<0.05, **p<0.01, ***p<0.001 4.4.1.4. Standardization and fairness The scatter plots for cardiovascular diseases and pneumonia show that the relationship between standardization and fairness is largely independent. However, 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with gastrointestinal disease, many hospitals are located either in high standardization and low fairness quadrant or in high standardization and high fairness quadrant. The distribution of hospitals in these scatter plots seems to contradict the popular belief that high standardization is associated with unfairness. On the contrary, they seem to suggest that hospitals vary in both standardization and fairness dimensions when they manage drug formularies. Figure 8 Scatter plot for cardiovascular diseases 6.0 5.5 5.0 4.5 ■ g c o ? 1 3.5 a □ , a @ aa rat a a a o a . ° a n j | J& ana a a a oa a % & o f t , “ “ n °< % P “ o P tJ w ’Cl O n o „ t - m „aa tP a a 4.5 5.0 5.5 6.0 6.5 Overall fairness index (additive) 7.0 7.5 8.0 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 9 Scatter plot for pneumonia 5.5 5.0 T O 4.5 4.0 3.5 3.0 ° b a a □ aa a n ® * V 1 \ 0 . O g ° a 8 a £ Q tf1 a a a a H a “ e B ° °o S a □ » « ft °*n 4 ° ° | a Q % ° ° ° ° o a o a r f * 4.5 5.0 5.5 6.0 6.5 overall fairness index (additive) 7.0 7.5 8.0 Figure 10 Scatter plot for gastrointestinal diseases B T P 5.5 5.0 4.5 * E c a 3.5 * 3.0 a cd a c a n o a a a a a a a a aa _ _ _ a Sen3 a do a ' C Q I8S80 o aa m arfVD a D , 8 aS a .-B ° » cd? r a s Q a a u u rn a aa a a gp a a □ 3 a a 4.5 5.0 5.5 6.0 6.5 Overall fairness index (additive) 7.0 7.5 8.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.2. Correlations The presence of multicollinearity in the data can lead to large standard errors of the estimated regression coefficients resulting in insignificant t-ratios. To evaluate multicollinearity, pair-wise correlation data were inspected. Correlation table 17 contains multiplicative interaction terms that are centered. Centering reduces, but not completely eliminates, multicollinearity without changing the structural relationship among variables (Jaccard et al., 1990). Thus centering increases the likelihood that a statistically significant interaction effect can be identified. Though many of correlations between the overall indices and component measures are greater than 0.5, they are not entered into the model simultaneously. So the majority of correlations among variables in the models are less than 0.5 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 8 Correlations 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 . pharmacy charge 1 . 0 0 0 2 . # o f hospital beds -0.005 1 . 0 0 0 3. for-profit hospital .452** -.188** 1 . 0 0 0 4. teaching hospital -0.129 .570** -.2 1 2 ** 1 . 0 0 0 5. system-affiliated .340** 0.045 .272** -0.023 1 . 0 0 0 6 . Florida .251** 0.051 .156* -.161* 0.135 1 . 0 0 0 7. New York -.671** 0.074 -.352** .2 0 2 ** -.263** -.489** 1 . 0 0 0 8 . urban 0.072 -0.048 -0.005 -0.093 -0.014 0.025 -.180* 1 . 0 0 0 9. Area Wage Index -.263** .254** -.145* .288** -0.070 -.171* .540** -.368** 1 . 0 0 0 10. HMO penetration rate 0.098 .168* 0.006 0.059 .225** .310** -0.034 -.405** 0.106 1 . 0 0 0 11.% W orker Comp -0.066 -0.070 -0.065 -0 . 0 2 2 -0.099 0.058 -0.015 0.031 -0.019 0.043 1 . 0 0 0 12. % Medicare -.177* -.164* 0.066 -.2 1 0 ** 0.076 .204** 0 . 1 1 0 0.036 -0.069 -0.008 -0.023 1 . 0 0 0 13. % Medicaid -.156* 0.093 0.015 0.136 -0.132 -0 . 1 2 0 0 . 1 2 2 0.044 .219** -0.034 0.018 -.477** 14. % other Govt, plans 0.043 0.009 -0.039 0.013 0.118 0.127 -.244** .146* -.268** 0.018 0.055 -0.105 15. % commercial plans .247** .160* -0.069 0.136 0.067 -.156* -0.052 -0 . 1 2 1 0 . 1 2 0 0.093 -0.027 -.431** 16. average age -.147* -0.065 0.008 -0.108 0.090 .191** .227** -.152* 0.093 .181* -0.079 7 3 7 ** 17. % male .271** .313** -0.024 .249** 0.074 .204** -.2 0 2 ** 0.072 -0.063 -0 . 0 0 1 0.036 -0.014 18. avrg # o f diag. per pt. 0.024 0.023 0 . 0 1 2 0.018 0.067 .370** -0.017 0.077 -.207** 0.034 0.039 .384** 19. %died -.249** -0.042 -0.066 -0 . 1 2 0 -0.046 -.192** .405** -0 . 1 1 2 .263** 0.045 -0.135 .251** 20. % using ICU .214** 0.115 199** 0.008 0.015 -.273** -0.091 -0.069 0 . 1 2 1 -0.029 0.057 21. avrg DRG weight .244** .526** -0.045 .392** 0.118 -0.053 -0.063 0.087 0.024 0.009 -0.103 -0.129 22. avrg LOS -.191** .314** -0.134 .258** -0.133 -.356** .588** -0.130 .518** 0.015 -0.131 0.006 23. %Indian -0.031 0.126 -0.117 0.091 -0.028 -0.038 .186* -.153* .295** .177* 0.005 -.180* 24. %Asian -0.037 0.114 -0.043 .195** 0.031 -.152* 0.133 -0.126 .346** 0.036 -0.068 -.2 0 1 ** 25. %Black 0.064 .234** -0.013 .198** -0.071 0.044 -0.053 -0.141 .191** .159* 0.029 - 379** 26. % White -0.082 -.239** -.161* -.2 2 1 ** 0.033 .263** 0.065 -0.036 -0 . 1 0 0 -0.013 0 . 0 2 2 .367** 27. %Hispanic .2 0 2 ** 0.032 .207** -0.083 0.129 -0.074 -.257** -0.038 -0.076 .233** -0.043 -.278** 28. %pts on guidelines -.237** 0.066 -0.072 0.104 -0 . 0 2 0 0.026 0.142 -0.004 0.075 0.063 -0.107 0.089 29. communication channels -0.106 .237** -0 . 1 1 1 .403** 0 . 0 0 0 -.155* .162* -0.092 0.131 0.031 0.047 -.156* 30. # o f nonphysician votes in P&T -0.097 -0 . 0 0 1 0.014 0.009 -0.043 0.044 0.117 -0.004 0 . 1 2 0 0 . 1 0 0 -0.129 0.129 31. financial incentive 0.066 0.075 0.026 0.074 0.048 0.031 0.063 -0 . 0 1 1 0.098 .150* 0.132 -0.026 32. enforcement (z) 0.030 0 . 0 0 2 0 . 0 1 1 0.018 -0.053 -0.034 0.018 -0.016 .151* 0.064 0.083 -0 . 1 1 0 33. # o f drug classes exchangeable (z) 0.104 0.069 0.007 0.005 0.015 .262** _j9 j* » -0.031 -.234** .289** 0.084 0.024 34. # o f unavailable drugs on formulary (z) -.314** 0.013 -0.065 .147* -0.098 0 . 0 2 1 .429** -.219** .321** .257** .182* 0.049 35. Overall Standardization Index (add) -0.098 0.064 -0.034 0.092 -0.067 0.136 0.108 -0.114 0 . 1 1 0 .292** .173* -0.013 o o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 9 Correlations (Continued) 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 36. Overall Standardization Index (mult) -0.068 0.044 -0.015 0.078 -0.042 .144* 0.055 -0.097 0.068 .270** .184* -0.036 37. formal review (z) -0.082 .193** -0.031 .216** 0.049 0.038 0.105 -0.015 0.107 0.090 0.052 0.013 38. responsiveness (z) -0 . 1 2 2 0.044 -.148* .216** 0.068 -0.029 0.106 -0.044 .160* 0 . 1 1 1 -0.076 -0 . 0 0 1 39. due process (z) 0.008 0.062 -0.059 0.106 0.083 0.089 -0.086 -0.019 0.048 -0.073 0.008 0.023 40. flexibility (z) -0.029 0.006 -0.042 0.016 0.045 -0.069 0.042 0.038 0.105 -0.050 -.179* -0.025 41. Overall Fairness Index (add) -0.088 0.130 -0.118 .239** 0 . 1 1 2 0 . 0 2 0 0.064 -0 . 0 1 0 .172* 0.031 -0.082 0 . 0 0 2 42. Overall Fairness Index (mult.) -.177* -0.016 -0.061 0.009 0.060 0.049 0.033 0.025 0.043 -0.028 -0.042 0.091 43. #34 *#40 0.061 0.108 0.083 0.036 0.018 0.070 0.027 -.170* 0 . 1 2 2 0.033 -.213** 0.123 44.#34 * #39 -0.037 0.003 0.030 -0.031 0.119 0.014 -0.058 0.042 -0.037 0.029 -0.044 -0.018 45. #34 * #37 0.043 -0.056 0.105 0.045 0.092 -0.041 -0.055 -0.056 -0.007 -0.086 0.006 -0.076 46. #34 * #38 0.138 .155* 0.048 0.114 .182* -0.034 -0.006 -0.092 0.085 0.038 -0.114 -0.065 47. #34 * #41 0.085 0.092 0.114 0.070 .173* 0.003 -0.037 -0 . 1 2 1 0.074 0.008 -.154* -0.014 48. #34 * #42 -0.025 0.046 0.006 -0 . 0 0 2 0.071 0.037 0 . 0 0 1 0 . 0 2 1 0.062 -0 . 0 2 1 -0.013 -0.008 49. #33 * #40 -0 . 0 2 1 -0.060 -0.004 -0.093 0 . 1 0 0 0.066 0.065 0.141 -0.030 -0.086 -0.007 0.123 50. #33 * #39 -0.018 -0.134 0.046 -0.085 -0.004 0.139 -0.056 -0 . 0 0 2 -0.093 0.038 0 . 0 2 1 0.047 51. #33 *#37 0.040 -0.099 0.137 -0 .0 6 ^ 0.053 0.006 -.171* 0.017 -.153* -0.054 0.006 -0 . 0 1 1 52. #33 * #38 0.055 -.208** 0.132 -.204** 0.125 0.040 -0.106 0.074 -.146* -0.091 -0.018 0.004 53. #33 * #41 0.015 -.2 0 2 ** 0.124 -.181* 0.105 0.099 -0 . 1 0 2 0.088 -.164* -0.073 0 . 0 0 1 0.069 54. #33 * #42 -0.042 0.026 -0 . 0 1 2 -0.016 0.034 0.086 -0.061 0.095 -0 . 0 0 1 -0.047 0.065 0.043 55. #32 * #40 -0.054 -0.008 -0.088 -0.013 -0.052 0.046 0.079 0.033 0.029 -0.097 -0.039 0.087 56. #32 * #39 -0.052 -0.029 -0.088 -0 . 0 1 1 .181* -0.040 0.017 0 . 1 2 2 -0.041 -0 . 1 0 1 0 . 0 1 1 0.064 57. #32 *#37 -0.036 -0.028 -0.016 -0.135 0.066 0.072 -0.051 -0.016 0 . 0 1 0 0 . 0 2 2 -0 . 0 2 2 0.025 58. #32 * #38 -0 . 0 0 1 -.144* 0.081 -0.109 0.049 .167* -0.089 -0.056 0.078 -0.034 -0.028 0 . 0 0 0 59. #32 *#41 -0.049 -0.079 -0.043 -0 . 1 0 0 0.091 0.098 -0.018 0.033 0.025 -0.077 -0.029 0.063 60. #32 * #42 -0.039 -0 . 0 2 1 -0.037 -0.042 0.094 0.032 -0.036 0 . 1 0 1 0.058 -0.114 0.025 0.047 61. #35 *#40 0 . 0 1 0 -0.086 0.107 -0.079 0.109 0.013 -0.143 -0.007 -0.085 -0.059 -0.004 -0 . 0 2 1 62. #35 *#39 0.068 -0.096 0.128 -0.094 .185* 0.081 -0.126 -0.019 -0 . 0 1 0 -0.066 -0.073 -0.017 63. #35 *#37 -0.043 -0.085 -0.024 -0.062 0.140 0.058 -0 . 0 1 2 0.073 -0.063 -0.005 -0.007 0.042 64. #35 *#38 -0.019 0.025 -0 . 0 0 2 -0.031 0.037 0.094 0.059 0.030 0.053 -0.085 -0 . 1 2 0 .163* 65. #35 *#41 0.006 -0.094 0.085 -0.104 .184* 0.094 -0.089 0.027 -0.041 -0.084 -0.080 0.065 6 6 . #36 * #40 -0.092 0.139 0 . 0 2 2 .151* 0.115 0.055 0.031 -0.007 0.074 0.038 0.059 0.005 67. #36 * #39 -0.043 0.018 -0.060 -0.003 0.063 -0.031 0.064 0.067 0 . 1 1 2 -0.088 -.204** 0.059 6 8 . #36 * #37 -0.009 0.044 -0.109 0.088 0.116 0.097 -0.043 0.026 0.037 -0.050 0.013 0.035 69. #36 * #38 -0.113 0 . 0 0 2 -0.076 0.142 .152* 0.026 0 . 0 2 0 -0.039 0.131 0.043 -0.083 -0 . 0 0 1 70. #36 *#41 -0.103 0.006 -0.037 -0.018 0.088 0.083 -0.013 0.077 0.050 -0.043 0 . 0 1 0 0.069 p<01; *: p<.05 00 0 0 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 8 Correlations (continued) 13 14 15 16 17 18 19 2 0 2 1 2 2 23 24 13. % Medicaid 1 . 0 0 0 14. % other Govt, plans 0.023 1 . 0 0 0 15. % commercial plans -.170* -0.078 1 . 0 0 0 16. average age -.655** -0.108 -.217** 1 . 0 0 0 17. % male -.261** -0.018 .242** -0.031 1 . 0 0 0 18. avrg # of diag. per pt. -.386** 0.049 -.158* .465** .152* 1 . 0 0 0 19. %died -0 . 0 1 2 -0.098 -.203** .364** -.424** 0.078 1 . 0 0 0 20. % using ICU -.166* -0.047 0.082 0.141 0.085 -0.076 1 . 0 0 0 21. avrg DRG weight -.2 1 0 ** 0.015 3 9 9 ** -0.009 .674** 0.137 -.170* 0.132 1 . 0 0 0 22. avrg LOS 0 . 1 1 2 -0.137 -0.016 .208** -.156* 0.036 .550** -0.143 .2 0 2 ** 1 . 0 0 0 23. %Indian .216** -0.078 0.017 -0 . 1 1 0 0.090 -0.039 0.072 -0.080 0.056 .176* 1 . 0 0 0 24. %Asian .182* 0.048 0.076 -0 . 1 1 1 -0.057 -0.131 0.118 -0.096 0.024 .250** .434** 1 . 0 0 0 25. %Black .457** .217** -0 . 1 0 0 -.336** -.178* -.204** 0.059 0.028 -0.017 .173* .172* .162* 26. % White -.420** -0.089 -0.029 .385** 0.084 .366** 0.074 0.028 -.152* -0.117 -0.034 -0.126 27. %Hispanic .353** -0.067 0.046 -.233** -.2 1 2 ** -.305** -0 . 0 2 0 -0.050 -0.070 0.024 0.003 0.049 28. %pts on guidelines 0.026 -0.088 -0.156 0.097 0 . 0 2 1 .160* 0.097 -0.072 0.049 0.095 .207** 0.080 29. communication channels 0.087 0 . 0 0 1 0.091 -0.094 0.104 0.075 -0.083 -0.046 .251** .171* 0.083 0.057 30. # o f nonphysician votes in P&T -.147* -0.024 -0.004 .255** -0.077 0.076 0.075 0.059 -0.006 0.070 0.088 0.104 31. financial incentive -0.004 -0 . 0 1 0 0.034 0.050 -0.080 0.104 0 . 0 2 2 -0.045 0.016 0.089 0.113 -0.059 32. enforcement (z) .170* -0.065 0 . 0 1 1 -.145* -0.113 0 . 0 1 1 -0.034 -0.108 -0.080 0.008 0.040 -0.009 33. # o f drug classes exchangeable (z) -0.026 0 . 1 0 0 0.008 -0.037 .167* .176* -0.144 0.037 0.066 -0.090 0.057 -0.105 34. # of unavailable drugs on formulary (z) 0.096 -0.034 -.147* .146* -0.104 0.033 .2 0 1 ** -0.037 -.186* .236** .169* 0.088 35. Overall Standardization Index (add) 0.116 0.003 -0.071 -0.023 -0.014 0 . 1 1 1 0.007 -0.062 -0.094 0.070 0.133 -0 . 0 1 2 36. Overall Standardization Index (mult.) 0.130 -0.008 -0.052 -0.071 0.005 0 . 1 1 1 -0.017 -0.033 -0.075 0.027 0.124 -0.032 37. formal review (z) -0 . 0 1 0 -0.099 0 . 0 0 1 0.051 0 . 1 1 1 0.118 -0.040 0.006 .167* 0.088 0.007 -0.059 38. responsiveness (z) 0.003 -0.024 0.042 0.069 -0.059 0.038 -0.044 -0.009 0.035 0.067 -0.045 0.026 39. due process (z) -0.016 0.082 0.050 0.005 0.092 -0.045 -0.075 0.031 0.117 0.003 -0.017 0.015 40. flexibility (z) 0.052 -0 . 0 1 0 0 . 0 1 2 0.026 -.150* 0 . 0 1 1 .162* -0.097 -0.069 0.129 -0 . 0 2 2 0.106 41. Overall Fairness Index (add) 0.016 -0.018 0.044 0.061 0 . 0 0 2 0.062 -0.004 -0.027 0.116 0.119 -0.03! 0.038 42. Overall Fairness Index (mult.) 0.044 -0.019 -0.118 0.049 -0.107 0.078 .191** -0.019 -0.045 0.081 -0.123 -0.044 43. #34 * #40 -0.018 -0.069 -0.075 0.117 -0.066 0.038 .151* 0.093 0.008 .175* -0.058 0 . 0 1 2 44. #34 * #39 -0.079 0.071 .182* 0 . 0 2 0 0.017 0.004 0.030 -0.023 0.050 0.014 0.047 0.009 45. #34 * #37 0.039 0.028 0.088 -0.065 -0.044 0.004 -0.016 -0.033 -0 . 0 1 1 -0.030 -0.041 -0 . 0 1 0 46. #34 * #38 0.059 0.039 0.080 -0.099 -0.009 -0.029 -0.062 -0.038 0.042 0.023 -0.027 0.041 47. #34 * #41 0 . 0 0 2 0.027 0.114 -0 . 0 1 0 -0.044 0.004 0.045 0 . 0 0 0 0.036 0.079 -0.033 0 . 0 2 2 48. #34 *#42 -0 . 0 0 1 0.004 0.057 -0.004 -0.039 0.016 0.055 0.055 -0.037 0.029 -0.078 0 . 0 0 0 49. #33 * #40 -0.016 0.072 -.204** 0.057 -0.083 0.128 0.035 -0.028 -0.088 -0.006 -0.049 -0.130 O Q 'O T able 8 Correlations (continued) C N ■"t C O o in © C N © C O © © S O s 00 00 © in S os 8 vs Os © C O © C N © © r~- © © fN 8 © r~ ~ © © 00 © C O 8 © © © © © © © Os C O © o © © 9 © 1 9 © 1 © © © 9 © © © © i 9 9 © © © 9 2 3 r - «n o I © C N C O © 00 © in © © t> © C N •Nt o in 8 G © © © 00 C N © © © r-~ © © © © 8 C O C N © 00 © © © C O © co © © © (O © C N On © o © i © © © 1 © © © © © 9 9 © © 9 9 © © 1 © © © 2 2 ro O 00 C O © © On © S O O S © C O c n © © 00 © © c n S O © so © © © 8 C O © © C N © 00 C O © Os © © © © C O 8 oo C N © © 8 3 © © © i © 9 9 © o © i 9 © I © i © 9 9 © 9 © © © t © i 9 C N O" o in 00 © « ■ S O in n * C O in © © in © § # "<t © 00 © © © © O s 00 © © < N 3 Nf C O © © © © © 00 © co © 8 © C O © © 9 © < ■ ' 9 9 © © © i r 9 9 © © < © 9 © 1 © 9 © © © 1 2 0 j 00 © in <N C O so © O n © © S O 00 © in © * © in © © © c n © © 00 © © b - © I r - co © "It © © © © co © r - © © C N © co © © o © © © © © i © 1 © © © © © © © 1 © © © 1 © © © O S in in oc © in so © t> C N © C N * t 00 © © in r - © 00 © Os © © © © © © © © 8 © © © 00 00 © © © © r - © * r-~ - © © s © vs © f " © 9 © © © 9 © © © © © © © © 9 © © © i 9 © © 00 o C N C N C N © fN S O © C O C N S O © © r - C N O S S O © s Ov © C N O S © © © f - © © fN 00 © - © © © © C N © in © O S 00 © © © © © © © © © 9 © © © © © © © © © © © © 3 © * * © 00 C O c n C O © so C N © so 8 8 © © C N 00 © © t - "St © 9 C N C O © 00 - St © © C N © -*t O n © C O 00 © C O © © © © © i © © 9 © 9 9 9 9 9 © © 9 © 9 © © 9 V O © O S © S O 8 Os © © r - S O © Os © © in § © 8 Os © © © C O © 00 © co C N © © C O © C N r - © co 00 © © © © © C O © © © © © t - © 00 © t"** © © fN "3 * © © 9 © © © © © © 9 © © © i 9 © © © © © © © © © S O s «n in © © co © s in OS © in ■"t © » C O © Os © ^ t 8 © © © © b - © © 00 © © C N * 00 C N C O © C N C N © c-» © © co © b - © © m © © © 9 9 © 1 9 © © 9 9 © © 9 © »' 9 © © 1 © 9 9 Tj* in Os © «n in © so s in C O © so s C O s nt m © r - C O © © C O 8 C O © © t- * - © © *3 - co © © © © 00 © © fN © £ © © C N © a 00 © © © "5t C O © © 9 9 © © t 9 9 © © 9 9 © © © 9 © © o © 9 © 1 co 00 C N C N C O © fN © 00 C O © so 8 C O © t " © s © r-* m © O S © C O © © C N co © © © © * rf © C O C O © 00 C N © © C N ©© N " 00 © t - co ©1 ©©© 9 9 ©©©©©©©© T © 9 © © © © © #39 #37 00 C O #41 I #42 | #40 | #39 j #37 i 00 co % i C N #40 OS co % #3 7 1 #38 | © 3 #39 #37 #38 i n # » * * # # # » # * # * # * ■ $ # co C O % C O C O % C O C O % co co co C O % fN C O % fN C O sfc fN C O % C N C O % C N C O % C N C O % © co % © C O % © C O % © C O % © C O © C O = fc © C O % © C O % © C O % © C O © <n in C N in co in O " in in in sd »n r-~ tn 00 m O n in © © © C N © C O © © © © © © r*- © 00 © os © © t> Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 8 Correlations (continued) 25 26 27 28 29 30 31 32 33 34 35 36 25. %Black 1 . 0 0 0 26. % White -.423** 1 . 0 0 0 27. %Hispanic -0.032 -.343** 1 . 0 0 0 28. %pts on guidelines 0.083 0.036 -0.054 1 . 0 0 0 29. communication channels 0.052 -0.076 -0.045 .324** 1 . 0 0 0 30. # o f nonphysician votes in P&T -0.078 -0 . 0 0 2 -0.050 0.003 0.095 1 . 0 0 0 31. financial incentive -0.051 -0.060 0.116 -0.055 .154* 0.032 1 . 0 0 0 32. enforcement (z) .156* -0.141 0.139 .272** .2 2 1 ** 0.047 .171* 1 . 0 0 0 33. # of drug classes exchangeable (z) 0.005 -0 . 0 2 1 0.066 .184* .159* -0.025 0.084 0.143 1 . 0 0 0 34. # o f unavailable drugs on formulary (z) 0 . 1 1 0 -0.007 -0.125 .234** .143* 0 . 1 0 2 -0.007 .144* .264** 1 . 0 0 0 35. Overall Standardization Index (add) 0.134 -0.087 0.024 .365** .260** 0.043 0 . 1 1 1 .632** .699** .691** 1 . 0 0 0 36. Overall Standardization Index (m ult) 0 . 1 2 2 -0 . 1 0 1 0.027 .331** .236** 0.006 0.089 .606** .698** .641** .962** 1 . 0 0 0 37. formal review (z) 0.027 -0.023 -0 . 0 0 2 .191* .272** 0.047 199** .323** .189** .199** .359** .333** 38. responsiveness (z) 0.024 -0.078 0.063 .177* .199** 0.092 0.127 .388** 0.130 .141* .316** .297** 39. due process (z) 0.007 -0.060 -0.094 .170* .158* -0.059 0.032 .214** 0.091 0.060 . 2 0 2 ** .197** 40. flexibility (z) 0.031 -0.027 -0 . 0 2 2 0 . 0 2 1 -0.059 -0.060 0.099 0.029 -0.090 -0.088 -0.095 -0.073 41. Overall Fairness Index (add) 0.042 -0.083 -0 . 0 2 2 .237** .244** -0 . 0 0 1 197** .404** 0.140 0.132 .331** .320** 42. Overall Fairness Index (mult.) -0.038 0.043 -0.006 .174* 0.063 -0.054 -0.116 .175* 0.097 0.091 .2 0 0 ** .251** 43. #34 *#40 0.037 -0.013 0.055 -0.018 0.084 -0.077 -0.007 0.087 0 . 0 2 1 0.055 0.074 0.077 44. #34 * #39 -0.107 0.116 -0.044 -0.052 -0.051 -0 . 0 2 0 -0.081 -0.091 0.043 -0.057 -0.035 0.050 45. #34 * #37 -0.093 0.039 0.113 -0 . 1 0 0 -0.115 0.030 -0.047 -0.059 -0.078 -0 . 1 1 2 -.152* -0.051 46. #34 * #38 0 . 0 2 2 -0.115 0.060 -.177* 0.008 -0.009 -0.069 -0 . 0 1 1 -0.015 0 . 0 1 0 -0.051 0.061 47. #34 * #41 -0.058 0 . 0 1 1 0.077 -0.145 -0.028 -0.034 -0.086 -0.029 -0 . 0 1 1 -0.042 -0.066 0.059 48. #34 * #42 0 . 0 0 2 0 . 0 2 0 -0.073 -0.031 -0.070 -0.032 -.153* 0.091 0.129 0.070 .166* .275** 49. #33 * #40 0.113 0.096 -0.126 0.037 -.2 1 0 ** -0 . 1 0 2 0 . 0 1 2 0.068 -.170* 0.016 -0.047 -0.042 50. #33 * #39 -0 . 1 1 0 .154* -0.065 -0.047 -0.063 .166* -0.082 0.044 0 . 1 1 2 0.046 0.103 .183* 51. #33 *#37 0.037 -0.057 0.109 -0.154 -0.032 0.056 0 . 0 1 0 0.051 0.129 -0.066 0.054 .194** 52. #33 * #38 0.008 -0.033 0.050 -0.136 -0.068 0.031 0.088 -0.008 0.086 -0.008 0.027 .147* 53.#33 * #41 0.019 0.070 -0.016 -0.119 -.150* 0.069 0.008 0.070 0.063 -0 . 0 0 2 0.060 .199** 54. #33 * #42 -0.064 0 . 0 1 1 -0.036^ -0.005 -0 . 0 0 1 0.081 0.045 0.137 .185* 0.089 .206** .333** 55.#32 * #40 0.007 0.068 -0 . 1 1 0 0.125 -0.051 -0.091 0.008 .142* 0.061 0.076 0.085 0.055 56. #32 * #39 -0.115 .247** -0.099 0.123 0.045 -0.070 0 . 0 1 0 0.056 0.041 -0.090 0.042 0.075 57. #32 * #37 -0.051 0.013 0.107 0.045 -0 . 0 0 1 -0 . 0 2 2 .157* 0.074 0.045 -0.055 0.004 0.098 58. #32 * #38 0.045 -0.025 0.064 0.093 -0 . 0 1 2 0.083 .173* .224** -0.009 -0 . 0 1 2 0.064 .148* 59. #32 *#41 -0.036 0.105 -0 . 0 1 0 0.142 -0.008 -0.050 0.132 .188** 0.054 -0.026 0.071 0.139 60.#32 * #42 -0.083 0.036 -0.052 0.091 0 . 0 2 0 0 . 0 0 1 -0 . 0 1 1 .233** 0.115 0.049 .219** .317** 61. #35 *#40 -0.047 -0.005 0.137 -0.084 -0.087 0.017 0.052 -0.003 0.043 -0 . 1 2 2 -0.039 0 . 1 1 2 VO- H — i Table 8 Correlations (continued) © cn .157* 5 O d # O n O O ■ * -0.033 * * m 00 C N ■ * © • S t 00 r- cn «n cn § © 0 0 in © d vo »n © d vo cn © d £ # O N r- cn O N in © d 4 * * 00 cn C N s 3 « n a vo s d «n 8 d 00 vo © d 9 * C N r- 9 C N 00 © © * ■ n • N - fN © 1 3 3 | fN C N O d cn © d ri- © d O N s d * # C N 9 * s » cn vo * r- r- fN cn 8 O d 00 cn © d s © d © r-- © d o s d * n- « * * r- 00 cn ■ J t # 00 cn fN cn V O v > o d «n © d C N C N © d m © d # in C O © cn © d t-~ 8 © 8 © d i 1 3 0 1 in cn O d oo > n © © * C N in fN © d i O N «n © d § © 9 © 1 9 O N fN 8 © 9 v O C N © d < 9 C N © d i in <N © © i * ■ cn in 3 n (N © d 1 2 8 I © d © © 9 § © d d i fN in d r - n © d in cn d oo fN d © t"> (N in s © O n § 9 9 C N 8 9 cn cn © © 9 cn 9 n cn © d 00 © © 1 2 6 1 O v oo © d i # O N in C N © £ © § © © f- s 9 00 © © «n vo © © »n 9 in 8 © < N d * «n V O © d © 9 C N © © © d t-* 8 d cn s d O N in © 9 O n cn % * in cn % C N V O r - * - cn % # «n $ cn vo 00 cn % # in $ 2 3 # in cn % •n V O © 3 # V O cn = t f c 3 O N cn % * v O cn V O r- cn % # v o 8 s 00 cn % * © cn % O S © S # © cn % © r- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 8 Correlations (continued) 37 38 39 40 41 42 43 44 45 46 47 48 37. formal review (z) 1 . 0 0 0 38. responsiveness (z) .412** 1 . 0 0 0 39. due process (z) .164* 0.053 1 . 0 0 0 40. flexibility (z) -0 . 0 2 1 .150* -0.024 1 . 0 0 0 41. Overall Fairness Index (add) .664** .691** .509** .471** 1 . 0 0 0 42. Overall Fairness Index (mult.) 0.105 .189** .173* 0.085 .236** 1 . 0 0 0 43. #34 * #40 0.109 -0.009 0.030 0.024 0.066 .175* 1 . 0 0 0 44. #34 * #39 -0.106 -0.053 0 . 0 1 0 0.031 -0.050 0.008 0.090 1 .0 0 0 45. #34 * #37 -0.133 0.030 -0.111 0.115 -0.042 0.132 -0.083 .238** 1 .0 0 0 46. #34 * #38 0.028 -0.028 -0.054 -0.008 -0.026 0.099 0.116 0.029 .384** 1 .0 0 0 47. #34 *#41 -0.041 -0.026 -0.051 0.069 -0 . 0 2 1 .177* .490** .575** .643** .647** 1 .0 0 0 48. #34 * #42 0.138 0.107 0 . 0 1 1 .171* .183* .483** .216** .280** .344** .391** .522** 1 .0 0 0 49. #33 *#40 0.064 0.053 -0.037 0.094 0.077 0 . 1 0 1 .215** -0 . 0 0 2 0.061 0.038 0.131 .243** 50. #33 * #39 -0 . 1 0 2 -0 . 0 1 0 0.065 -0.035 -0.034 0.078 0 .0 0 0 .297** .178* 0.009 . 2 0 1 ** .204** 51. #33 *#37 -0 . 1 1 2 0.004 -0 . 1 1 2 0.068 -0.062 .154* 0.056 .193** .409** .194** .354** .351** 52. #33 * #38 0 . 0 0 1 -0.042 -0.015 0.060 0.003 .278** 0.037 0.015 .204** .245** .209** 4 4 4 ** 53. #33 * #41 -0.060 0.004 -0.037 0.077 -0.007 .245** 0.125 .209** .346** .193** .363** .500** 54. #33 * #42 0.105 .175* 0.054 0.063 .170* .281** .158* 0.144 .234** .280** .341** .615** 55. #32 * #40 0.131 0.087 0.081 0 . 1 0 0 .167* .423** .192** -.209** -0.077 0 . 0 2 2 -0.027 .160* 56. #32 * #39 0.043 0.070 .171* 0.097 .162* .246** -.219** 0.036 0 . 1 0 2 0.038 -0 . 0 2 2 .2 1 1 ** 57. #32 * #37 -0.092 -0 . 0 1 0 0.038 .144* 0.031 3 3 7 ** -0.050 0.117 .277** .180* .219** .235** 58. #32 * #38 -0 . 0 1 0 0.098 0.065 0.098 0.105 .307** 0.059 0.045 .160* 0.140 .171* .280** 59. #32 * #41 0.029 0.091 0.134 .165* .179* .509** 0.005 -0.014 .170* .144* 0.128 .336** 60. #32 *#42 .209** .188** .142* .276** .349** .642** 0.113 0.125 .142* .170* .234** .689** 61. #35 *#40 -.175* -0.009 -0.069 .146* -0.047 .296** -0.067 .290** .739** .316** .530** .429** 62. #35 * #39 -0.018 -0.029 0.025 0.052 0.013 .353** 0.082 0.091 .328** .644** .478** .553** 63. #35 *#37 -0.065 0.040 0.098 0.077 0.065 .166* -0.044 .658** .306** 0.086 .418** 64. #35 * #38 .144* 0.042 0.068 0.042 0.125 .385** .660** -0.053 -0.083 0.065 .251** .316** 65. #35 * #41 -0.048 0.016 0.046 0.126 0.059 .476** .250** .382** .516** .438** .661** .650** 6 6 . #36 * #40 .843** .378** .160* 0.023 .599** .320** 0.099 0.068 .2 1 1 ** .183* .234** .421** 67. #36 * #39 0.019 0.134 0.007 .873** .434** .292** .254** 0.029 0.095 0.068 .188* .367** 6 8 . #36 * #37 .155* 0.114 .884** 0.008 4 9 7 ** .255** 0.033 .248** 0.050 0.049 .159* .229** 69. #36 * #38 .354** .850** 0.108 0.138 .617** .419** 0.070 0.058 .173* .243** .227** .442** 70. #36 *#41 .177* .243** .147* .165* .312** .829** .216** 0.137 .225** .252** 3 4 7 ** .773** **: p<.01; *: p<.05 V O u » T able 8 Correlations (continued) o © © © © ft ft cn V O cn # V O 0 0 rf ft ft in © cn ft ft rj- 0 0 C N ft « vo V O in * t- in -li * ft C N cn ri .305** f t ft © 0 0 rf ft * in rf 0 0 O n in § © ft ft e ^ - rn in * ft 0 0 rf ft ft C N n in # ft C N 9 f t ft n C N cn si ft 0 0 O N so ft ft 0 0 O N C N ft f t cn cn ft ft © rf cn ft # cn ft ft cn 0 0 in 0 0 in 8 © ft * t"- cn p ft * C N p ft ft C N O N p » ft in © p * ft T — 1 p « © V © ft ft r- t- p ft ft cn © p ft © rf ft t- * # © © rf ft « - n n rf © ft ft in in in ft ft C N C N p ft ft C N vo p ft ft in p ft ft cn cn rf ft ft cn © p © e- © d i ft ft w n in « « r* p C N cn d ft e- r- ft ft © p ft ft in O N p © m o o p * ft O s cn p ft rf V O ft © © V O ft ft in cn s i ft V O e- C N ft ft C N # © vo C N cn © © « ft 'Sf cn rf ft ft © © C N * o 0 0 * © rf ft ft in O s ft ft n © cn 8 O C N C N d O N 8 d f t ft C N C N p ft ft vo in in ft ft in © p On O O © © 1 c- O N d C N © d s i ft C N C N p ft rf © p C N C N d ft ft e- p ft © n C N d « ft rf O n p > < } ■ W ) 8 p Os c- o o * ft © c n ■ » * C N t" C N ft ft r~ - S O cn ft ft cn rf cn ft « rf C N n ft ft cn O N cn # ft V O in • » ft cn ft ft s ft ft cn O N n ft ft in © ri * ft in in C N ft ft in in C N ft ft cn cn n ft ft C N e- m in p * * o r- in cn o o ft O N rf ft ft On 0 0 ft ft in O N fN * ft v O c- C N * « r- r t cn * « in z * ft in 0 0 in * « > 0 0 li ft ft vo cn ft ft O N in r- ft # 0 0 r- C N ft ft n <n C N « « cn C N ft ft © cn cn # * O s © p C N © o © p * ft in in p # ft © o in r- 0 0 O o On o ft ft cn © fN ft ft cn cn cn ft ft C N 0 0 C N * ft ?; cn ft # t - © rf « © ft cn r- * © in # 0 0 p ft C N n p ft © in in © © « ft cn rf p ft ft e- 0 0 rf »n 8 O S i ft o o p « # r- 0 0 © * * rf n cn in O N o o ■ 0 0 n o o ft ft V O V O C N # S f t cn »n # ft rf in p ft ft in p * ft 0 0 © rf # C N in p V O 8 d ft * r- P « ft C N © p © © © d in © © ft * 0 0 C N p ft ft C N cn p o m o o o S i ft V O rf fN ft On 0 0 * * O N f~ - in ft * 0 0 r- C N o o d ft in vo C- V O © © 8 d C N C N d ft O N V O ft ft C N p ft © in ft « 5 0 0 © © » ft * C N rf O s n © d © 9 C N © rf © d ft ft © p O N rf 8 O in r t o © 1 C N rf O 9 ft ft fN o fN # ft V O rf ft ft rf t" p s i ft V O in C N o C N o o On vo © 9 C N fN © 0 0 C N © cn 0 0 © © O N C N © 9 ft C" vo C N © d ft ft vo O N © ft ft V O C N p rfr C- © d # « C N rf C- © © © i # e- rf ft « © cn p O 3 O N cn % r- $ O O cn % i C N 3 o 3 O n cn % cn % 0 0 cn % C N 3 © % O s m t^- cn % 0 0 cn % s © 3 O N cn % e- cn % O O cn % 3 ft * * ft ■ » * ft » * « ■ k - ft « ft * ft ft ft ft ft ft ft m cn % cn cn % cn cn % cn cn cn cn % cn cn S fc C N cn % C N cn % C N §s C N cn % C N $ 2 C N s n cn % in cn % <n cn % n cn in cn % © cn © m % © cn % © cn I t © cn % O N rf d in «n C N in cn <n rf in *n in V O in C- n C O in O N in © V O vo C N V O cn vo vo in © © © C" © 0 0 © O s © © e- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T able 8 Correlations (continued) 8 8 8 8 l O o V a o V Q . 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10 Descriptive Statistics N Mean Std. Deviation Minimum Maximum 1. pharm acy charge 186 3033.813 2119.106 1.260 9599.690 2. # o f hospital beds 198 271.520 190.040 33.000 1120 3. for-profit hospital 198 0.210 0.406 0 1 4. teaching hospital 198 0.150 0.354 0 1 5. system -affiliated 195 0.490 0.501 0 1 6. Florida 198 0.280 0.449 0 1 7. N ew Y ork 198 0.380 0.488 0 1 8. urban 198 0.270 0.446 0 1 9. A rea W age Index 198 1.018 0.204 0.771 1.443 10. H M O penetration rate 187 14.839 13.283 0.040 46.140 11. % W orker Com p 186 0.121 0.434 0 4.830 12. % M edicare 186 59.842 13.785 0 88.490 13. % M edicaid 186 8.300 7.964 0 53.260 14. % other Govt, plans 186 0.981 2.236 0 16.520 15. % com m ercial plans 186 24.752 10.290 0 63.110 16. average age 186 64.596 6.921 29.890 76.310 17. % m ale 186 51.844 6.454 36.040 73.150 18. avrg # o f diag. per pt. 186 6.153 0.905 3.360 7.960 19. % died 186 5.227 2.485 0.910 16.440 20. % using ICU 186 39.681 21.163 0 94.190 21. avrg D R G w eight 186 1.629 0.540 0.900 3.400 22. avrg LOS 186 5.797 1.367 3.320 11.980 23. %Indian 186 0.136 0.376 0 2.750 24. % A sian 186 0.604 1.402 0 12.910 25. % Black 186 9.905 10.558 0 56.790 26. % W hite 186 77.503 22.927 0 99.280 27. % Hispanic 186 11.908 20.458 0 93.890 28. %pts on guidelines 161 41.037 34.921 0 100 29. com m unication channels 198 3.303 1.286 0 6.000 30. # o f nonphysician votes in P& T 193 2.031 1.698 0 7.000 31. financial incentive 197 0.060 0.240 0 1 32. enforcem ent 198 0 1 -1.624 2.474 33. # o f drug classes exchangeable 188 0 1 -1.377 2.960 34. # o f unavailable drugs on formulary 197 0 1 -3.167 2.137 35. Overall Standardization Index (add) 187 -0.067 1.997 -4.900 6.780 36. O verall Standardization Index (mult.) 187 65.059 33.874 11.820 241.840 37. formal review (z) 197 0 1 -2.759 1.917 38. responsiveness (z) 197 0 1 -2.998 2.402 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 11 Descriptive Statistics (Continued) 39. due process (z) 196 0 1 -0.719 1.536 40. flexibility (z) 198 0 1 -1.981 2.567 41. Overall Fairness Index (add) 195 -0.008 2.343 -5.010 7.410 42. Overall Fairness Index (mult.) 195 0.075 1.456 -5.180 10.500 43. #34 * #40 197 -0.087 1.018 -4.400 2.670 44. #34 * #39 195 0.060 0.994 -3.990 2.410 45. #34 *#37 196 0.198 0.971 -4.010 3.630 46. #34 * #38 196 0.139 0.989 -3.330 4.180 47. #34 * #41 194 0.309 2.347 -11.500 10.510 48. #34 * #42 194 0.133 1.485 -7.270 14.590 49. #33 * #40 188 -0.088 0.997 -3.420 4.090 50. #33 * #39 186 0.091 1.049 -2.130 4.450 51. #33 *#37 187 0.189 1.001 -2.640 5.490 52. #33 * #38 187 0.130 0.962 -2.310 6.880 53. #33 *#41 185 0.327 2.477 -6.920 20.860 54. #33 * #42 185 0.146 2.357 -5.050 28.960 55. #32 *#40 198 0.029 1.181 -3.280 6.350 56, #32 * #39 196 0.212 1.010 -2.500 3.800 57. #32 * #37 197 0.323 1.078 -2.630 4.740 58. #32 * #38 197 0.387 1.050 -2.790 5.940 59. #32 * #41 195 0.940 2.825 -3.880 18.030 60. #32 * #42 195 0.252 2.645 -6.720 25.030 61. #35 *#40 186 0.718 2.242 -8.000 13.000 62. #35 * #39 186 0.626 2.063 -4.210 16.280 63. #35 * #37 185 0.403 2.019 -5.750 10.420 64. #35 * #38 187 -0.184 2.138 -7.440 9.700 65. #35 *#41 184 1.556 5.398 -11.880 49.390 66. #36 * #40 186 10.611 74.277 -318.200 463.550 67. #36 * #39 187 -3.542 77.168 -255.770 345.820 68. #36 * #37 185 7.884 78.986 -133.800 371.580 69. #36 * #38 186 7.788 77.628 -168.670 580.810 70. #36 * #41 184 19.362 206.957 -282.730 2446.040 Valid N (listwise) 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.3. Assessment of the Representativeness of the Sample In order for me to derive general conclusions about the relationships between hospital practices and cost advantage from the data collected, it was important that the responding hospitals be representative of the mailing sample. I used two types of tests to assess the representativeness of the respondents. First, I compared respondents to the mailing sample along known dimensions. These dimensions include organization size, ownership, system affiliation, and geographical region. Second, I used wave analysis to investigate whether a self-selection bias existed such that firms following certain hospital formulary practices were more likely to respond to the survey. This procedure is based on the observation that in mail surveys, nonrespondents tend to be more similar to late respondents than to early respondents (Fowler, 1993). Wave analysis gauges nonresponse bias by comparing respondents who respond readily to a survey to those who respond after follow-up steps are taken. Comparisons of means and correlations for respondents to the first mailing and respondents to the third mailing revealed that the two groups did not differ significantly in either the levels of the variables or in the relationships between the variables. These results provide evidence that the respondents were representative of the mailing sample and that a self-selection bias was unlikely to exist. 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.3.1. Comparison of respondents and mailing sample. Four variables that are available for the whole mailing sample as well as the respondents are organization size, ownership, system affiliation, and geographical location. Comparisons of the respondents to the mailing sample along these four dimensions can be seen in the following section. As for organization size, the mean and the median for the respondents and the mailing sample are very similar (Table 5). T-test revealed no statistically significant difference for the mean size for the two groups (Table 6). Thus we can conclude that the respondents are representative of the mailing sample in terms of size. Table 12 Number of hospital beds of respondents to mailing sample Mailing sample Respondents Mean # of hospital beds 325.85 327.73 (standard deviation) (258.36) (227.97) Median # of hospital beds 250.5 252 Table 13 One-Sample Test Test Value = 325.8546 t df Sig. (2-tailed) Mean Difference 95% Confidence Interval of the Difference Lower Upper # of Beds .116 196 .908 1.8764 -30.1558 33.9086 Table 12 compares the respondents to the mailing sample by ownership. The response rates for government-owned hospitals and nonprofit hospitals are higher 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. than expected, whereas the response rate for for-profit hospitals are lower than expected. Table 14 Ownership of respondents and mailing sample COMPLETE Total 0 1 OWNERSHI govt Count Expected Count % within OWNERSHI % within COMPLETE % of Total 27 32.5 49.1% 9.5% 5.6% 28 22.5 50.9% 14.1% 5.8% 55 55.0 100.0% 11.4% 11.4% nonprofit Count Expected Count % within OWNERSHI % within COMPLETE % of Total 149 164.0 53.6% 52.3% 30.8% 129 114.0 46.4% 65.2% 26.7% 278 278.0 100.0% 57.6% 57.6% for profit Count Expected Count % within OWNERSHI % within COMPLETE % of Total 109 88.5 72.7% 38.2% 22.6% 41 61.5 27.3% 20.7% 8.5% 150 150.0 100.0% 31.1% 31.1% Total Count Expected Count % within OWNERSHI % within COMPLETE % of Total 285 285.0 59.0% 100.0% 59.0% 198 198.0 41.0% 100.0% 41.0% 483 483.0 100.0% 100.0% 100.0% To investigate these differences further, I performed a Chi-square test for the differences in observed and expected responses in the different ownership groups. This test rejects the null hypothesis that the response rates across ownership groups are equal (Table 13). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 15 Chi-square test by ownership groups Value df Asymp. Sig. (2-sided) Pearson Chi-Square 17.170a 2 .000 Likelihood Ratio 17.682 2 .000 Linear-by-Linear Association 14.904 1 .000 N of Valid Cases 483 a- 0 cells (.0%) have expected count less than 5. The minimum expected count is 22.55. Similarly, a cross tab table (Table 23) showed that system-affiliated hospitals were less likely to participate in this study than the independent hospitals were. Since the majority of system-affiliated hospitals are for-profit hospitals, this further showed the lack of participation from for-profit hospitals. A Chi-square test rejected the null hypothesis of equal response rate (chi-square (1 DF) = 5.476, p = 0.023). 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 16 System affiliation of respondents and mailing sample Responded? Total No Yes System? no Count Expected Count % within SYSTEM % within Respondents % of Total 100 112.4 54.1% 32.5% 19.7% 85 72.6 45.9% 42.7% 16.8% 185 185.0 100.0% 36.5% 36.5% yes Count Expected Count % within SYSTEM % within Respondents % of Total 208 195.6 64.6% 67.5% 41.0% 114 126.4 35.4% 57.3% 22.5% 322 322.0 100.0% 63.5% 63.5% Total Count Expected Count % within SYSTEM % within Respondents % of Total 308 308.0 60.7% 100.0% 60.7% 199 199.0 39.3% 100.0% 39.3% 507 507.0 100.0% 100.0% 100.0% The uneven response rate across ownership and system-affiliation give rise to two questions. First, what might be reasons for these differences, and second what are the implications of these differences for the generalizability of results. Reasons for the differences in response rate might be differences in the organizational policy to participate in outside surveys. Several system-affiliated hospitals returned the survey to me uncompleted with a note stating that their corporate headquarters did not allow participating in surveys. This uneven response across groups in terms of ownership and system-affiliation affects the generalizability of the results only if significant cross-group differences exist in the levels of the variables included in this study. I will explore potential differences across groups in the preliminary data analysis and the results sections. 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, I examined whether the response rate varied across three states. I found no statistically significant differences in observed and expected responses (Ch- square (2 DF) = 1.293, p = 0.524). Table 17 Geographical location of respondents and mailing sample Responded? Total No Yes STATE TX Count Expected Count % within STATE % within responded % of Total 100 100.4 60.6% 32.4% 19.7% 65 64.6 39.4% 32.7% 12.8% 165 165.0 100.0% 32.5% 32.5% FL Count Expected Count % within STATE % within responded % of Total 102 96.7 64.2% 33.0% 20.1% 57 62.3 35.8% 28.6% 11.2% 159 159.0 100.0% 31.3% 31.3% NY Count Expected Count % within STATE % within responded % of Total 107 111.9 58.2% 34.6% 21.1% 77 72.1 41.8% 38.7% 15.2% 184 184.0 100.0% 36.2% 36.2% Total Count Expected Count % within STATE % within responded % of Total 309 309.0 60.8% 100.0% 60.8% 199 199.0 39.2% 100.0% 39.2% 508 508.0 100.0% 100.0% 100.0% 4,4.3.2. Comparison of early and late respondents To test for differences between early and late respondents, I divided the completed surveys into two groups, responses to the first two mailings (n=135) and responses to the third mailing (n=60). Table 25 shows the means and standard deviations of the variables for each of the two groups, and the results ofT-tests for 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the difference in the means. These tests revealed that significant differences exist only for two variables at the 10% level. Differences between early and late respondents could indicate that non-respondents differ from respondents, i.e., that a response bias exists. Thus, these results strengthen our confidence in the representativeness of the sample. Table 18 Variable means and standard deviations for early and late respondents ______________ ______________ ______ Early respondents late respondents T-test responsiveness 17.6260 (3.15579) 18.3898 (2.53946) # of heart drugs 29.8370 (8.59035) 30.0500 (9.21389) # of pneumonia drugs 37.4370 (7.87743) 38.7833 (8.48946) # of ulcer drugs 2.9704 (1.28679) 3.3833 (1.64772) + Enforcement 9.7143 (3.39467) 10.2667 (4.09988) Formal review 9.9104 (2.6342) 10.5333 (2.3754) Due process .5672 (.86231) .8305 (.93131) + Hospital bed 287.30 (231.988) 257.12 (156.498) case mix 1.440644 (.2433119) 1.439762 (.2107068) %HMO penetration 14.2831 (13.41261) 16.5690 (13.07901) wage index 1.004633 (.1934860) 1.046035 (.2214737) Standard deviations in parentheses ***significant at 0.1% level ** significant at 1% level * significant at 5% level + significant at 10% level Differences between early and late respondents could also exist in terms of organizational characteristics. I used chi-square test to evaluate the association between type of respondents (early or late) and organizational characteristics. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. test shows that early and late respondents do not differ significantly in terms of ownership, teaching status, or urban/rural location. However, the chi-square test reveals that system hospitals are more likely to respond late at 10% significance level (Table 12). Table 19 system affiliation for early and late respondents late response No Yes Total system no Count 63 19 82 Expected Count 56.8 25.2 82.0 yes Count 72 41 113 Expected Count 78.2 34.8 113.0 Total Count 135 60 195 Expected Count 135.0 60.0 195.0 4.4.3.3. Comparison of variable means across hospital characteristics The comparison of the respondents to the mailing sample showed that the two groups differed in their system membership and ownership. In addition, a comparison of the early and the late respondents revealed that these groups differ in system membership. These differences might affect the generalizability of the results for the sample if the levels of the variables included in the model differ across system-membership and/or ownership. To test for differences in the levels of the variables, I performed F-tests for the quality of means across system-membership groups and ownership groups for each variable included in the models. I also performed pair-wise comparisons of the specific means for all of the ownership groups using Scheff s Test for Contrasts. The group means and the results of the F-tests for system-affiliation and ownership 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. groups can be seen in Table 27 and Table 28 respectively. The results of the Scheffe tests are discussed below. The comparison of the variable means across all groups reveals that the null hypothesis that all the group means are equal can not be rejected at the five percent level. The null hypothesis can be rejected at the 10% level for the independent variable responsiveness. However, the Scheffe Tests for the variable responsiveness did not reveal any significant differences across ownership groups. Thus, the level of all variables does not different significantly across groups. These results indicate that even though respondents might not be representative for the whole mailing sample in terms of the ownership and system-affiliation groups, results of this study might not be affected by this difference. Table 20 comparisons of means by system-affiliation groups Sum of Squares df Mean Square F Sig. Enforcement Between Groups 14.565 1 14.565 1.108 .294 Within Groups 2510.927 191 13.146 Total 2525.492 192 Formal Review Between Groups .006 1 .006 .001 .975 Within Groups 1271.932 192 6.625 Total 1271.938 193 Due Process Between Groups .423 1 .423 .533 .466 Within Groups 151.618 191 .794 Total 152.041 192 Responsiveness Between Groups 6.873 1 6.873 .767 .382 Within Groups 1685.569 188 8.966 Total 1692.442 189 # of ulcer drugs Between Groups .000 1 .000 .000 .999 Within Groups 389.149 193 2.016 Total 389.149 194 # of Pneumonia drugs Between Groups 58.681 1 58.681 .900 .344 Within Groups 12584.007 193 65.202 Total 12642.687 194 # of heart drugs Between Groups 112.179 1 112.179 1.464 .228 Within Groups 14786.970 193 76.616 Total 14899.149 194 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 21 Comparison of means by ownership groups Sum of Squares df Mean Square F Sig. Enforcement Between Groups 36.522 2 18.261 1.394 .251 Within Groups 2488.970 190 13.100 Total 2525.492 192 Formal Review Between Groups 1.806 2 .903 .136 .873 Within Groups 1270.132 191 6.650 Total 1271.938 193 Due Process Between Groups .880 2 .440 .553 .576 Within Groups 151.162 190 .796 Total 152.041 192 Responsiveness Between Groups 42.910 2 21.455 2.432 .091 Within Groups 1649.532 187 8.821 Total 1692.442 189 # of ulcer drugs Between Groups 5.350 2 2.675 1.338 .265 Within Groups 383.799 192 1.999 Total 389.149 194 # of Pneumonia drugs Between Groups 124.369 2 62.185 .954 .387 Within Groups 12518.318 192 65.200 Total 12642.687 194 # of heart drugs Between Groups 193.596 2 96.798 1.264 .285 Within Groups 14705.552 192 76.591 Total 14899.149 194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5. RESULTS 5.1. Diagnostics Since regression analysis requires certain conditions, regression diagnostics were conducted first. I examined the univariate frequency distribution of each variable for outlying observations and normality assumptions. Visual inspection of the plots of independent variables against the dependent variable revealed no significant outliers for all independent variables. I also tested for curvilinear relationships by putting in squared terms of standardization variables. There were no significant relationships between squared independent terms and the dependent variable. The presence of multicollinearity in the data can lead to large standard errors of the estimated regression coefficients resulting in insignificant t-ratios. To evaluate multicollinearity, pair-wise correlation data were inspected first. The correlations among the independent variables shown in Table 17 were reviewed for multicollinearity. These correlations are generally low (below 0.3), with a few exceptions. Because I centered continuous independent variables first (i.e., subtract the mean from each case) and then computed the interaction terms, most of the interactions terms were not correlated with the terms used to compute them. Even after centering, occasionally a high correlation (above 0.8) was found among some interaction terms. However, they were not entered into a same regression model. 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Indices such as Tolerance Statistics, the Variance Inflation Factors (VIF) and Condition Indices recommended by (Belsley, Kuh et al. 1980) were also used. Multicollinearity is a problem if Tolerance is less than 0.1, VIF is greater than 10, or condition index is greater than 30. Tolerance statistics is the reverse of the extent to which each individual regressor is linearly dependent on the other regressors. Thus, a value of “tolerance” near 1 means that the respective variable is largely “indedenpdent” from the other regressors. VIF measure the inflation in the variances of the parameter estimates due to the collinearities that exist among independent variables. A high VIF indicates that the multiple correlation coefficient of the explanatory variable X * regressed on the remaining explanatory variables is nearly unity, and thus points to collinearity. Condition indices refer to the square root of the ratio of the largest to the smallest characteristic root of X’X. An index of 5 or 10 points has weak dependencies, while moderate to strong relationships have indices over 30. Examinations of condition indices and of variance inflation factors indicated that multicollinearity was not a major problem. A small set of the data can have a disproportionate influence on the estimated parameters in a regression equation. Two basic diagnostic tools are the diagonal elements of the least-square project matrix hi, and the studentized residuals. It is suggested that when hi is more than twice its average value, p/n, the ith observation is a leveraging point and may be an influential observation (Belsley, Kuh et al. 1980). The studentized residuals are used to identify data points that have an extreme effect 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. on the regression results. Observations in which the studentized residuals exceed a value of two are potential outliers (Belsley, Kuh et al. 1980). The analysis revealed that four observations are potentially influential in the regression. In order to examine the influence of these variables, a series of regression analyses were conducted by eliminating these observations, but R2 was not significantly increased in all the Models when observations are eliminated. Therefore, the cases were kept in the data set and the subsequent results for regression equations were based on 198 data points without eliminating those potential outliers. 5.2. Hierarchical regression I used hierarchial regression to test the hypotheses at the organizational level of analysis. Control variables were entered in the first step, the variables of theoretical interests, standardization and procedural fairness variables, in the second step, and interaction terms in the third step. 5.2.1. The base models To control for other influences on hospital pharmacy charge, I have included several control variables in Table 29. Model 1 shows the “base case” variables regressed on pharmacy charge. Some of the control variables are statistically significant and have expected signs. A for-profit ownership of the hospitals, affiliation with system, and average lenth of hospital stay have positive signs, 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. meaning that they contribute to higher pharmacy costs. Being located in New York state has a negative sign Issues of model specification and omitted variable bias are always a concern in selecting appropriate control variables. To address this, I incorporated a variety of other hospital-level variables in the models, including teaching status, urban/rural location, hospital size, system affiliation, regional HMO penetration rate and regional wage index. None of these variables signivcantly affected the findings. Model 2 tested whether giving financial incentive to clinical departments influences average pharmacy charge. The result was insignificantly. Model 3 examined whether increasing the number of nonphysician votes in the P&T committee impacts pharmacy charge. Again, it was not significant. Model 4 tests whether effective implementation of clinical guidelines reduces pharmacy charge. The result shows a significant reduction in pharmacy charge when hospitals enrolled larger number of patients on the guidelines. Thus I included this variable in the subsequent analyses that follows. Ill Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 22 The base regression models on pharmacy charge for aggregated dieseases Model 1 Model 2 Model 3 Model 4 Control variables Stndrzd Coeff. (Constant) # of hospital beds -.139 -.138 -.139 -.160* For-profit hospital .239*** 237*** .240*** .238*** Teaching hospital -.011 -.014 -.012 .014 System-affiliated .138* ,140* .136* .139* Florida -.097 -.102 -.092 -.061 New York -.699*** -.699*** -.696*** -.627*** Urban .040 .040 .045 .046 Area Wage Index .070 .069 .073 .084 HMO penetration rate .073 .071 .075 .063 %Worker Comp -.048 -.051 -.052 -.068 % Medicare -.055 -.055 -.058 -.039 % Medicaid -.100 -.095 -.103 -.151 % other Govt, plans -.077 -.078 -.076 -.083 % commercial plans .105 .105 .105 .079 average age -.043 -.037 -.033 -.088 % male .164 .170 .156 .157 avrg # of diag. per pt. .031 .029 .028 .054 %died -.091 -.091 -.095 -.076 % using ICU .000 .003 .000 -.016 avrg DRG weight .059 .056 .062 .094 avrg LOS .364*** .361*** .363*** .319*** %Indian .062 .056 .065 .084 %Asian -.081 -.076 -.077 -.078 %Black .067 .072 .061 .100 %White .079 .082 .074 .083 %Hispanic .026 .026 .025 .062 Incentive .021 voting # of nonphysicians -.031 % of patients on guidelines -.143** Adj. R2 .625 .624 .613 .674 AR2 .002 .001 .018 F change .335 .308 8.243** d.f. 155 154 151 126 + p < .10 * p < .05, **p< .01, ***p < .001 5.2.2. Main effect models The next set of tables adds variables of substantive interest, one at a time, to the “base case” variables. I individually entered standardization variables first, then procedural fairness variables, and finally both of them together into the equations. There were no significant results for the effects of either standardization or procedural fairness variables when they are entered item by item or by overall 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. indexes. Hence those variables did not add much to the R squared explained. The changes in F statistics were insignificant and minimal. The control variables maintained similar levels of significance for the main effect models. The percentage of patients enrolled in guidelines, the number of hospital beds, and being located in New York state reduced pharmacy charges significantly. On the other hand, a member of system, for-profit hospital ownership, and average length of hospital stay all contributed to higher pharmacy charge. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 23 Main effect regression models for aggregated diseases coefficients Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Control variables # of hospital beds -.165* -.161* -.164* -.162* -.169* -.165* For-profit hospital .226*** .237*** .239*** .235*** 228*** .232*** Teaching hospital .009 .013 .009 .018 .007 .019 System-affiliated .136* .141* .141* .142* .142* .146* Florida -.066 -.072 -.097 -.057 -.103 -.067 New York -.663*** -.635*** -.616*** -.627*** -.643*** -.637*** Urban .045 .046 .059 .046 .058 .047 Area Wage Index .076 .089 .119 .086 .114 .094 HMO penetration rate .055 .054 .076 .061 .062 .047 %Worker Comp -.084 -.074 -.085 -.069 -.098 -.077 % Medicare -.019 -.041 -.045 -.038 -.030 -.040 % Medicaid -.149 -.148 -.164 -.151 -.162 -.148 % other Govt, plans -.086 -.081 -.085 -.084 -.087 -.081 % commercial plans .085 .075 .066 .080 .069 .076 average age -.104 -.081 -.093 -.087 -.102 -.077 % male .145 .153 .163 .154 .151 .148 avrg # of diag. per pt. .065 .051 .084 .053 .090 .049 %died -.084 -.077 -.080 -.076 -.089 -.077 % using ICU -.023 -.011 -.003 -.018 -.005 -.012 avrg DRG weight .125 .102 .085 .095 .115 .106 avrg LOS .323** .315** 294** .320** .295** .315** %lndian .091 .087 .072 .083 .078 .085 %Asian -.087 -.078 -.068 -.079 -.075 -.079 %Black .107 .099 .122 .101 .126 .100 %White .089 .090 .091 .079 .097 .086 %Hispanic .073 .063 .081 .061 .089 .061 % of patients on guidelines -.160** -.154** -.151*** -.140* -.167** -.152** Standardization variables # of unaviable drugs on formulary # of drug classes exchangeable Enforcement OS! (additive) -.003 -.033 .091 .034 .006 -.014 .076 .044 Procedural fairness variables Formal review Flexibility Responsiveness Due process OFI (additive) -.005 -.070 .092 -.024 -.016 -.011 -.073 .084 -.019 -.028 Adj. R2 AR2 F change d.f. .615 .001 .127 152 .619 0 .141 154 .617 .004 .428 150 .622 .001 .548 153 .609 .007 .475 146 .616 .002 .518 151 * p < .05, * * p < .01, ***p < .001 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3.3. Interaction models Interaction effects between standardization and procedural fairness are shown in next four tables (31, 32, 33, & 34). Table 31 contains the first hierarchical regression analysis. Model 1 provides the base case, with all control variables and all main effect variables. Model 2 to model 13 add the two-way interaction terms one at a time, and model 14 adds all the interaction terms at once. Next, as shown in Table 32, individual interaction terms of the component items of the overall standardization index with overall procedural fairness index were entered one at a time and then altogether in one equation. Table 33 examines the effects of the interaction terms between the overall standardization index and individual items of the overall procedural fairness index. Finally in table 34 the interaction beteen the overall standardization and procedural fairness indexes was entered. Multiplicative interaction terms are often criticized because they are so highly correlated with the component variables, creating problems of multicollinearity that can inflate standard errors. To deal with this problem, many recommends a linear transformation known as “centering,” in which the mean value for a variable is subtracted from each other. Centering reduces multicollinearity among predictor variables, and makes meaningless regression coefficients meaningful. Yet, centering doesn’t really affect anything of interest. Simple slopes will be the same in centered as in uncentered equations, their standard errors and t- tests will be the same, and interaction plots will look exactly the same, only with different values on the x-axis (Aiken, 1991 #5840). 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In assessing the goodness-of-fit of the regression models, I looked at an “adjusted” R square and the stardard error of the estimate statistics (Lewis-Beck, 1980 #5867). The regression model presented in Table 33 has the most favorable goodness-of-fit statistics. Its adjusted R squre (0.703) is the highest among the regression models examined, and the standard error of the estimate Y (1137) is the smallest. The results in Table 33 shows that, of the four interaction terms, two interaction terms (Overall standardization index*formal review and Overall standardization*due process) were statistically significant while the Overall standardization index*responsiveness interaction term and the overall standardization index*flexibility term are not. This indicates that, when all other main effects and interaction terms are held constant, hospitals with high interaction scores for the overall standardization index*formal review term and the overall standardization index*due process has lower pharmacy charge. Though the significance level for each interaction term varies across the regression models, the directions of the terms were more consistent. Across all the models, I find the majority of interaction terms having the minus sign. Also in all models, the adjusted R square is higher than the “main effect” case and in many models the increase is statistically significant. Hence the two-way interaction terms do explain variance in pharmacy charge beyond that captured by the individual main effect terms. 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 24 Interaction models with individual standardization and procedural fairness items coefficients Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 1 2 Model 13 Model 14 Control variables # o f hospital beds -.163* -.166* -.164* -.172* -.156* -.161 -.167* -.166* -.176* -.145* -.164* -.173* -.166* -.159 For-profit hospital 2 4 4 *** .241*** .244*** .252*** .246 .243*** 242*** .244*** .241*** .243*** .241*** .241*** .246*** Teaching hospital . 0 1 2 .024 .008 .014 .030 .027 . 0 2 0 .009 .013 . 0 0 1 .013 .015 .014 .048 System-affiliated .133* .141* .136* .124* .138* .141* .129* .138* .136* .137* .133* .141* .152* .125* Florida -.096 -.088 - . 1 0 1 -.094 -.085 -.108 -.093 -.099 -.091 -.123 -.094 -.084 - . 1 0 2 - . 1 0 2 New York -.627*** -.624*** -.627*** -.623*** .628*** -.676 -.636*** -.628*** -.619*** -.629*** -.627*** .626*** .612*** .669*** Urban .058 .052 .067 .060 .072 .058 .054 .059 .060 .058 .058 .055 .067 .088 Area Wage Index .114 .109 . 1 1 0 .108 .103 .086 . 1 1 2 .115 .117 .119 .117 .125 .125 .073 HMO penetration rate .065 .049 .070 .065 .069 .046 .063 .061 .060 .076 .062 .062 .049 .067 % Worker Comp -.099 -.092 -.085 -.094 - . 1 0 1 -.092 -.097 - . 1 0 0 - . 1 0 1 -.089 - . 1 0 0 -.099 -.091 -.060 % Medicare -.028 -.032 -.039 -.029 -.028 -.030 -.033 -.032 -.037 -.015 -.027 -.028 -.031 -.026 % Medicaid -.182 -.177 -.175 -.183 -.183 -.140 -.175 -.182 -.190 -.183 -.184 -.192 -.195 -.150 % other G ovt plans -.086 -.083 -.088 -.086 -.073 -.106 -.089 -.088 -.076 -.072 -.086 -.081 -.090 -.084 % commercial plans .081 .091 .083 .078 .106 .113 .089 .077 .074 . 1 0 0 .078 .072 .076 .144 average age -.128 -.129 - . 1 2 1 -.124 -.130 -.082 - . 1 2 0 -.126 -.126 -.158 -.132 -.148 -.147 -.106 % male .151 .148 .156 .153 .140 .119 .150 .152 .145 .154 .150 .145 .147 .108 avrg # o f diag. per pt. .096 .096 .095 .093 .079 .085 .089 .098 . 1 0 0 . 1 2 1 .097 .094 . 1 0 2 .073 %died -.074 -.067 -.078 -.070 -.061 -.056 -.071 -.072 -.095 -.057 -.074 -.072 -.068 -.032 % using ICU -,0 H -.015 -.014 - . 0 1 0 -.018 .005 -.014 -.008 - . 0 0 2 .018 -.013 - . 0 0 2 -.004 . 0 1 0 avrg DRG weight . 1 0 1 .096 .093 .111 .096 .088 . 1 0 2 .099 . 1 1 0 .075 .104 .099 . 1 1 0 .093 avrg LOS .282** .282** .280** .277** .291** .287 .287** .278** .288** .270** 2 7 9 ** .282** .276** .283** %Indian .074 .076 .076 .078 .088 .082 .074 .075 .085 .091 .076 .082 .073 .119 %Asian -.066 -.072 -.065 -.067 -.077 -.068 -.061 -.067 -.062 -.079 -.067 -.076 -.069 -.088 %Black .124 .116 .116 .125 . 1 1 0 .111 .113 .123 .115 .105 .124 . 1 1 2 .124 .081 %White .094 .098 .092 . 1 0 0 . 1 1 0 .092 .089 .092 .108 .092 .095 .088 .124 .136 %Hispanic .092 . 1 0 2 .084 .094 .085 .078 .092 .093 .097 .119 .091 .090 . 1 0 1 .077 % o f patients on guidelines -.169** -.167** -.163** -.165** -.170** -.186 -.167** -.173** -.180** -.152** -.169** -.160** -.167** -.150** Standardization variables Enforcement # o f drug exchangeable # o f drugs .005 -.015 .073 . 0 0 2 -.013 .066 - . 0 0 1 -.013 .065 .006 -.016 .073 . 0 0 1 -.008 .073 .027 .023 .070 . 0 0 1 -.008 .073 .006 -.011 .073 .018 -.007 .080 . 0 0 2 -.003 .060 .006 -.014 .073 .014 - . 0 1 2 .064 .004 .005 .063 . 0 2 1 .041 .046 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 25 Interaction models with individual standardization and procedural fairness items (Continued) coefficients Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12 Model 13 Model 14 Formal review Responsiveness Due process -.009 -.076 .085 -.017 - . 0 2 0 -.072 .079 -.009 -.015 -.072 .088 - . 0 1 1 - . 0 1 1 -.068 .085 -.017 -.018 -.068 .076 -.024 - . 0 2 2 -.069 .069 - . 0 1 1 - . 0 1 0 -.078 .084 - . 0 2 2 -.007 -.078 .086 -.015 -.017 -.078 .090 -.013 -.006 -.083 .092 .023 -.007 -.076 .087 -.017 -.014 -.071 .085 - . 0 1 1 -.013 -.074 .104 . 0 0 1 -.042 -.034 .080 .018 # o f unaviable drugs * formal review # o f unaviable drugs * flexibility # o f unaviable drugs * responsiveness # o f unaviable drugs * -.066 .050 .035 -.107* .059 - . 1 0 1 .024 .076 # o f drugs exchangeable # formal review # o f drugs exchangeable # flexibility # o f drugs exchangeable # responsiveness # o f drugs exchangeable -.140** .042 - . 0 2 1 -.076 -.004* - . 0 0 2 -.162 .067 standardization enforcement * formal review gfandardt7ariftn enforcement * flexibility standardization enforcement * responsiveness standardization enforcement * due process -.138** -.015 -.070 -.084 -.035 -.042 -.080 -.023 Adj. R2 .67 .67 .674 .672 .684 .69 .673 .672 .677 .691 .672 .677 .677 .695 AR2 .74 .0 0 . 0 0 2 .0 0 1 .0 1 .014 . 0 0 1 0 .005 .015 0 .004 .005 .04 F change 10.24 *** 1.64 .867 .412 4.703* 7.002** .632 .148 2.176 7.258** .086 1.894 2.247 1 . 6 8 d.f. 118 117 117 117 117 117 117 117 117 117 117 117 117 106+ 0 0 Table 26 Interaction models with individual standardization items and overall fairness index coefficients Model 1 Model 2 Model 3 Model 4 Model 5 Control variables # of hospital beds -.160* -.155* -.166* -.162* -.166* For-profit hospital .242*** .247*** .240*** .235*** .236*** Teaching hospital .017 .024 .0 1 1 .019 .015 System-affiliated .132* .139* .144* .143* .151* Florida -.060 -.056 -.068 -.057 -.062 New York -.647*** -.648*** -.652*** -.649*** -.653*** Urban .044 .041 .050 .046 .049 Area Wage Index .077 .079 .077 .096 .094 HMO penetration rate .052 .046 .043 .043 .037 %Worker Comp -.087 -.093 -.090 -.086 -.089 % Medicare -.018 -.014 -.026 -.015 -.021 % Medicaid -.166 -.167 -.165 -.180 -.177 % other Govt, plans -.085 -.081 -.084 -.077 -.077 % commercial plans .099 .106 .092 .091 .088 average age -.127 -.132 -.117 -.157 -.147 % male .146 .140 .139 .139 .134 avrg # of diag. per pt. .069 .069 .076 .078 .082 %died -.070 -.066 -.071 -.058 -.060 % using ICU -.026 -.028 -.014 -.021 -.013 avrg DRG weight .109 .105 .104 .108 .104 avrg LOS .311** .316** .307** .304** .302** %Indian .085 .085 .092 .095 .099 %Asian -.077 -.080 -.079 -.088 -.089 %Black .104 .102 .103 .091 .092 “ /.White .085 .087 .090 .093 .095 "/.Hispanic .071 .075 .070 .076 .075 % of patients on guidelines -.158** -.162** -.173** -.149** -.162** Enforcement .006 .005 .018 .007 .016 # of drug exchangeable -.030 -.028 -.017 -.017 -.010 # of drugs .085 .086 .086 .073 .076 OFI -.021 -.027 -.017 .000 .000 Interaction effects # of unaviable drugs * OFI -.044 -.006 # of drugs exchangeable * OFI -.075 -.055 standardization enforcement * OFI -.094+ -.082 Adj. R2 .668 .668 .671 .675 .672 AR2 .736 .002 .004 .007 .01 F change 10.882*** .743 2.074 3.377+ 1.505 d.f. 121 120 120 120 118 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 27 interaction models with overall standardization index and individual fairness items coefficients Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Control variables # of hospital beds -.162* -.154* -.162* -.163* -.171* -.158* For-profit hospital .249*** .258*** .249*** .247*** .250*** .262*** Teaching hospital .017 .030 .018 .015 .030 .044 System-affiliated .137* .150* .136* .146* .155* .143* Florida -.096 -.105 -.097 -.096 -.085 -.098 New York -.605*** -.636*** -.608*** -.608*** -.592*** -.622*** Urban .060 .051 .060 .059 .075 .064 Area Wage Index .124 .100 .116 .129 .122 .084 HMO penetration rate .063 .045 .067 .060 .051 .046 %Worker Comp -.093 -.080 -.086 -.096 -.092 -.070 % Medicare -.043 -.034 -.048 -.044 -.050 -.038 % Medicaid -.184 -.156 -.176 -.186 -.200 -.160 % other Govt, plans -.083 -.082 -.085 -.083 -.068 -.072 % commercial plans .073 .115 .081 .070 .081 .130 average age -.113 -.112 -.104 -.118 -.129 -.107 % male .154 .136 .157 .152 .140 .132 avrg # of diag. per pt. .086 .097 .082 .089 .085 .089 %died -.068 -.044 -.069 -.068 -.075 -.050 % using ICU -.005 .011 -.004 -.002 .003 .010 avrg DRG weight .087 .066 .083 .082 .101 .084 avrg LOS .273** .276** .278** .273** .286** .287** %lndian .071 .086 .070 .072 .089 .097 %Asian -.060 -.076 -.056 -.062 -.065 -.072 %Black .119 .095 .112 .116 .103 .088 %White .093 .094 .089 .087 .136 .134 %Hispanic .086 .105 .084 .086 .094 .110 % of patients on guidelines -.166** -.166** -.164** -.169** -.177** -.165** Main effects OS! .042 .057 .039 .046 .067 .063 Formal review -.011 -.027 -.015 -.010 -.026 -.042 Responsiveness -.083 -.078 -.083 -.086 -.078 -.067 Due process .085 .073 .082 .085 .097 .079 Flexibility -.019 .015 -.019 -.016 -.004 .020 Interaction effects OSI * formal review -.158** -.158* OSI * flexibility .034 .015 OSI * responsiveness -.031 .068 OSI * due process -.134** -.090 Summary statistics Adi. R2 .677 .7 .676 .676 .693 .703 0^ < 3 .745 .02 .001 .001 .014 .028 F change 10.976*** 9.872** .378 .434 7.014** 3.538** d.f. 120 119 119 119 119 116 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 28 Interaction models with overall indexes coefficients Model 1 Model 2 Control variables # of hospital beds -.158* -.158* For-profit hospital .249*** .250*** Teaching hospital .024 .027 System-affiliated .136** .156** Florida -.060 -.058 New York -.620*** -.630*** Urban .047 .048 Area Wage Index .093 .102 HMO penetration rate .049 .033 %Worker Comp -.078 -.088 % Medicare -.037 -.033 % Medicaid -.170 -.177 % other Govt, plans -.081 -.073 % commercial plans .089 .090 average age -.108 -.128 % male .149 .135 avrg # of diag. per pt. .056 .067 %died -.061 -.052 % using ICU -.019 -.012 avrg DRG weight .092 .086 avrg LOS .300** .299** %Indian .081 .092 %Asian -.069 -.082 %Black .098 .089 %White .083 .093 %Hispanic .064 .071 % of patients on guidelines -.154** -.164** OSI .039 .054 OFI -.028 -.018 Interaction terms OSI * OFI -.102* Adj. R2 .67 .679 AR2 .733 .009 F change 11.665*** 4.246* d.f. 123 122 + p < .10 * p < .05, **p< .01, ***p < .001 5.3. Summary results The results suggest that the relationships of the standardization and procedural fairness to organizational effectiveness are complex. While the significance levels are not consistent for each effect across regression models, we find some emerging patterns of interaction effects from the standardization and procedural fairness variables. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The main effects of standardization and procedural fairness on pharmacy charge are inconclusive and statistically insignificant. Consistent with previous literature on formulary effectiveness, standardization did not contribute to pharmacy charge in either direction. Similarly, procedural fairness did not influence pharmacy charge at a statistically significant level. The interaction models, however, suggest more complex interpretations. To better explain the form of interactions reported in the above hierarchical regression analysis, I plotted the interaction effects in the graphs shown in Figure 11, using one standard deviation above and below the mean to capture high and low procedural fairness (Cohen & Cohen, 1983). Figure 11 Interaction Low Pro. J. High Pro. J. Standardization The figure above shows that, when a standard is fair, increasing standardization leads to high organization effectivenss. But when a standard is unfair, increasing standardization leads to lower organizational effectiveness, which 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. confirms hypothesis 3. It also helps to explain the standardization variables’ lack of statistical significance in the main effect models. If, depending on the level of procedural fairness associated with a standard, a standard’s impact on organizational effectiveness varies, as shown in Figure 11, the positive influence of a procedurally fair standard may be cancelled out by the negative influence of a procedurally unfair standard in the main effect models. 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6. CONCLUSIONS Past research on the relationship between standardization and organizational effectiveness has often failed to recognize the process dimension of standardization. In particular, procedural fairness issues of formulating and implementing standards have been much neglected. In this study I have investigated whether they matter. I developed several measures of standardization and procedural fairness in the context of drug formularies in hospitals. I found that the impact of standardization on organization effectiveness depends on procedural fairness. I also found that when the moderating role of procedural fairness is not considered in the model, standardization has no impact on organizational effectiveness. This may explain the conflicting results reported in the formulary literature on the effectiveness of drug standardization. Because the literature has failed to consider the fairness dimensions in standardization, I believe, the literature failed to find any consistent effect of formulary standardization. In contrast to the popular belief that formulary restrictions lead to lower pharmacy cost, my results show that restricting the number of drugs in formularies, therapeutically exchanging drugs, and enforcing those policies had no statistically significant impact on pharmacy cost. The main effect of procedural fairness features on organizational cost effectiveness was also largely insignificant. However, the interaction effects between standardization and procedural fairness showed that standardization does lead to organizational cost effectiveness when it is done in a procedurally fair manner. On the other hand, if a 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. standard is unfairly formulated and implemented, it contributes to organizational ineffectiveness. 6.1. Limitations of the study There are several limitations to this study. One key limitation of the study is that I do not measure quality of care. It is possible that some hospitals may standardize their formularies to the extent that they may achieve financial efficiency at the expense of providing quality care. Potentially certain variables such as re admittance rate, may serve as a proxy for quality of care. Unfortunately, however, such measures were not available for this study. Though a patient’s length of study can be used by some health care researchers as a proxy for quality, I decided not to use it because it is strongly determined by hospital’s administrative policy. Also the data are cross-sectional, so causality cannot be definitively determined. It was not possible to determine whether standardization leads to hospital effectiveness, or the other way around. It may be possible that hospitals experiencing operational inefficiency may try to further standardize their internal operations. In that case, the positive correlations found in the study between standardization and organizational ineffectiveness may not be attributed to the failure of standardization effect. Findings from hospitals, which have unique institutional environments, may not generalize well to other settings. Hospitals are under strong institutional pressure as well as market pressure. If institutional pressure determines hospitals’ practice of formulary management to a greater extent, the latter may not always translate to 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. operational efficiency outcomes. In fact, we found hospitals vary in their formulary management practices in terms of hospital institutional characteristics. Hence, by measuring organizational effectiveness only through pharmacy charge, we may not capture the full spectrum of effectiveness outcomes of formulary standardization. This study did not measure the fairness perception of physicians directly. Instead it looked at the organizational practices of formulary procedures, and assume that such practices would lead to fairness perceptions of physicians. Future studies should examine this assumption. How individual physicians’ acceptance of and willingness to cooperate with formulary standardization work would help to better understand the social psychological mechanism and impact of standardization. 6.2. Contributions Organizational literature has implicitly assumed that bureaucratic work structure is inappropriate for professional workers as well as knowledge-intensive work settings. However, this is rather one-sided view that does not take into bureaucratiziation processes, in particular procedural fairness in it. Rather than banishing bureaucratic work structure from professional work settings, this study suggests focusing on the features of bureaucratization pertaining to fairness. On a more macro level, it also suggests the possible feasibility of organizing effective professional work environment in a bureaucratic manner. Unlike some scholars that suggest professional organization form as a distinct from bureaucracy (Raelin, 1986 #2911), this study argues that the hybrid form can be also effective. 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. From a practical point of view, understanding the conditions under which formulary standardization leads to organizational effectiveness is potentially an important means of improving the cost-effectiveness of the healthcare delivery system. The knowledge gained will help healthcare organizations and policy makers to design formulary standards that enhance, rather than undermine, organizational effectiveness. Our results will also be of interest to people concerned with other forms of standardization (e.g. Information Systems) or with the role of fairness in the process of creating standards. Theoretically, this study contributes to building an organization theory of procedural justice. Using micro-research as building blocks for a macro-characterization of procedural justice, I highlight the strategic and performance implications of procedural fairness. 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES Abramowitz, P. W., & Fletcher, C. V. 1986. Counterpoint: let's expand the formulary system and renew its vigor. Am JHosp Pharm, 43(11): 2834-2838. Adler, P. S. 2001. Market, hierarchy, and trust: The knowledge economy and the future of capitalism. Organization Science. Adler, P. S., & Borys, B. 1996. Two types of bureaucracy: Enabling and coercive. Administrative Science Quarterly, 41: 61-89. Adler, P. S., Goldoftas, B., & Levine, D. I. 1999. Flexibility versus efficiency? A case study of model changeovers in the Toyota production system. Organization Science, 10(1): 43-68. Aiken, L. S., & West, S. G. 1991. Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. AMA. 1994. AMA policy on drug formularies and therapeutic interchange in inpatient and ambulatory patient care settings. American Medical Association. Am JHosp Pharm, 51(14): 1808-1810. American College of Physicians. 1990. Therapeutic substitution and formulary systems. American College of Physicians. Ann Intern Med, 113(2): 160-163. Anderson, G. M., & Pulcins, I. R. 1992. High variation medical conditions as an explanation of regional and temporal differences in hospital utilization. Med Care, 30(5): 461-465. Aquino, K. 1997. Integrating justice constructs into the turnover process: A test of a referent cognitions model. Academy of Management Journal, 40(5): 1208. Argyres, N. S. 1999. The impact of information technology on coordination: Evidence from the B-2 "Stealth" bomber. Organization Science, 10(2): 162- 180. 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Avom, J., Chen, M., & Hartley, R. 1982. Scientific versus commercial sources of influence on the prescribing behavior of physicians. Am J Med, 73(1): 4-8. Avom, J., & Soumerai, S. B. 1983. Improving drug-therapy decisions through educational outreach. A randomized controlled trial of academically based "detailing". N Engl J Med, 308(24): 1457-1463. Baer, W. C. 1986. Expertise and professional standards. Work and Occupations, 13(4): 532-552. Baldwin, C. Y., & Clark, K. B. 2000. Design rules. Cambridge, Mass.: MIT Press. Bamett, A. A. 1996. HMO formularies lead to higher care costs. Lancet, 347(9005): 894. Bates, D. W., Spell, N., Cullen, D. J., Burdick, E., Laird, N., Petersen, L. A., Small, S. D., Sweitzer, B. J., & Leape, L. L. 1997. The costs of adverse drag events in hospitalized patients. Adverse Drug Events Prevention Study Group. Jama, 277(4): 307-311. Beers, M., Avom, J., Soumerai, S., Everitt, D., Sherman, D., & Salem, S. 1988. Psychoactive medication use in intermediate-care facility residents. Journal of the American Medical Association, 260(20): 3016-3020. Beugre. 1996. Managing fairness in organizations. Westport, CT: Quorum. Bies, R. J., & Moag, J. S. 1986. Interactional justice: Communication criteria of fairness. In R. J. Lewicki, B. H. Sheppard, & M. H. Bazerman (Eds.), Research on negotiation in organizations : a biannual research series'. 43-55. Greenwich, Conn.: JAI Press. Bloom, B. S., & Jacobs, J. 1985. Cost effects of restricting cost-effective therapy. Medical Care, 23(7): 872-880. Blumenthal, D., & Herdman, R. (Eds.). 2000. Description and analysis of the V A national formulary. Washington, D.C.: National Academy Press. 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Boreham, P. 1983. Indetermination: Professional knowledge, organization and control. Sociological Review, 31: 693-718. Bowers, J. 1992. The politics of formalism, Contexts o f Computer-Mediated Communication. Bradbury, R. C., Golec, J. H., & Steams, F. E. 1991. Comparing hospital length of stay in independent practice association HMOs and traditional insurance programs. Inquiry, 28(1): 87-93. Brayfield, A. H., & Crockett, W. H. 1955. Employee attitudes and employee performance. Psychological Bulletin, 52: 396-424. Brennan, T. A., & Berwick, D. M. 1996. New rules : regulation, markets, and the quality o f American health care (1st ed.). San Francisco: Jossey-Bass Publishers. Brockner, J. 1996. Understanding the interaction between procedural and distributive justice: The role of tmst. In R. M. Kramer (Ed.), Trust in organizations: Frontiers o f theory and research: 390-413. Thousand Oaks, CA: Sage Publications Inc. Brunsson, N., & Jacobsson, B. 2000. A world o f standards. Oxford; New York: Oxford University Press. Bryk, A. S., & Raudenbush, S. W. 1992. Hierarchical linear models : applications and data analysis methods. Newbury Park: Sage Publications. Bums, L. R., & Wholey, D. R. 1991. The effects of patient, hospital, and physician characteristics on length of stay and mortality. Med Care, 29(3): 251-271. Bums, L. R., & Wholey, D. R. 1993. Adoption and Abandonment of Matrix Management Programs: Effects of Organizational Characteristics and Interorganizational Networks. Academy o f Management Journal, 36(1): 106- 138. 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cammann, C., & Lawler, E. E. 1973. Employee reactions to a pay incentive plan. Journal o f Applied Psychology, 58(2): 163-172. Caves, R., M, W., & Hurwicz, M. 1991. Patent expiration, entry, and competition in the U. S. Pharmaceutical Industry. Brookings Papers on Economic Activity: Microeconomics: 1-66. Chren, M.-M., & Landefield, S. 1994. In reply letter. J. Am. Med. Assoc., 272(5): 355. Classen, D. C., Pestotnik, S. L., Evans, R. S., Lloyd, J. F., & Burke, J. P. 1997. Adverse drug events in hospitalized patients. Excess length of stay, extra costs, and attributable mortality. Jama, 277(4): 301-306. Cohen, J., & Cohen, P. 1983. Applied multiple regression correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, N.J.: L. Erlbaum Associates. Colquitt, J. A. 2001. On the dimensionality of organizational justice: A construct validation of a measure. Journal o f Applied Psychology, 86(3). Conrad, D., Wickizer, T., Maynard, C., Klastorin, T., Lessler, D., Ross, A., Soderstrom, N., Sullivan, S., Alexander, J., & Travis, K. 1996. Managing care, incentives, and information: an exploratory look inside the "black box" of hospital efficiency. Health Serv Res, 31(3): 235-259. Cook, K. S., & Hegtvedt, K. 1983. Distributive justice, equity, and equality. Annual Review o f Sociology, 9: 217-241. Cooper, C. L., Dyck, B., & Frohlich, N. 1992. Improving the Effectiveness of Gainsharing: The Role of Fairness and Participation. Administrative Science Quarterly, 37(3): 471. Craig, T. 1995. Achieving innovation through bureaucracy: Lessons from the Japanese brewing industry. California Management Review, 38(1): 8-36. 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cropanzano, R., & Ambrose, M. L. 2001. Procedural and distributive justice are more similar than you think: A monistic perspective and a research agenda. In J. Greenberg, & R. Cropanzano (Eds.), Advances in organizational justice'. 119-151. Stanford, CA: Stanford University Press. Cropanzano, R., & Greenberg, J. 1997. Progress in organizational justice: Tunneling through the maze. International Review o f Industrial and Organizational Psychology, 12: 317-372. Curtiss, F. R. 1996. Drug formularies provide a path to best care. Amercian Journal o f Health-System Pharmacy, 53(18): 2201-2203. Daniels, C. E., & Wertheimer, A. I. 1982. Analysis of hospital formulary effects on cost control. Topics in hospital pharmacy management, 2(2): 32-49. Daniels, N., & Sabin, J. E. 2001. What are fairness and consistency in a national pharmacy benefit? Med Care, 39(4): 312-314. David, P. A., & Greenstein, S. 1990. The Economics of Compatibility Standards: An Introduction to Recent Research. Economics o f Innovation and New Technology, 1(1-2): 3-41. Deming, W. E. 1986. Out o f the crisis. Cambridge, MA: MIT Press. Derber, C., Schwartz, W., & Magrass, Y. 1990. Power in the highest degree: Professionals and the rise o f a new mandarin order. N.Y.: Oxford University Press. Deutsch, M. 1975. Equity, equality, and need: What determines which value will be used as the basis for distributive justice? Journal of Social Issues, 31: 137- 150. Deutsch, M. 1985. Distributive ju stice: a social-psychological perspective. New Haven: Yale University Press. Dillman, D. A. 1978. Mail and telephone surveys : the total design method. New York: Wiley. 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dillon, M. J. 1999. Drug formulary management. In R. Navarro (Ed.), Managed care pharmacy practice1 . 145-184. Gaithersburg, Maryland: Aspen Publishers. Donelan, K., Blendon, R. J., Lundberg, G. D., Calkins, D. R., Newhouse, J. P., Leape, L. L., Remler, D. K., & Taylor, H. 1997. The new medical marketplace: physicians' views. Health Aff (Millwood), 16(5): 139-148. Dranove, D. 1995. A measuring costs. In F. A. Sloan (Ed.), Valuing health care: Costs, benefits, and effectiveness o f pharmaceuticals and other medical technologies1 . 61-76: Cambridge University Press. Ebert, R. J., & Mitchell, T. R. 1975. Organizational decision processes : concepts & analysis. New York: Crane Russak. Eckhoff, T. E. 1974. Justice: its determinants in social interaction. Rotterdam: Rotterdam University Press. Eisenberg, J. M. 1986. Doctors' decisions and the cost o f medical care : the reasons for doctors' practice patterns and ways to change them. Ann Arbor, Mich.: Health Administration Press Perspectives. Ellis, R. P., & McGuire, T. G. 1986. Provider behavior under prospective reimbursement. Cost sharing and supply. J Health Econ, 5(2): 129-151. Ferris, G. R., & Wagner, J. A. 1 .1985. Quality Circles in the United States: A Conceptual Reevaluation. The Journal o f Applied Behavioral Science, 21(2): 155. Fins, J. J. 1998. Drug benefits in managed care: seeking ethical guidance from the formulary? J Am Geriatr Soc, 46(3): 346-350. Fiske, A. P. 1993. Structures o f social life : the four elementary forms o f human relations : communal sharing, authority ranking, equality matching, market pricing (1st Free Press pbk. ed.). New York Toronto: Free Press ; Collier Macmillan Canada; Maxwell Macmillan International. 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Folger, R., & Cropanzano, R. 1998. Organizational justice and human resource management. Thousand Oaks: Sage Publications. Forysyth, B. H., & Lessler, J. 1991. Cognitive laboratory methods: A taxonomy. In P. Biemer, R. Groves, L. Lyberg, N. Mathiowetz, & S. Sudman (Eds.), Measurement errors in surveys: 393-418. NY: John Wiley. Freidson, E. 1984. The changing nature of professional control. Annual Review of Sociology, 10: 1-20. Galaskiewicz, J. 1989. Mimetic Processes Within an Interorganizational Field: An Empirical Test. Administrative Science Quarterly, 34(3): 454-479. Galbraith, J. 1977. Organizational design. Reading, MA: Addison-Wesley. Gasser, L. 1986. The Integration of Computing and Routine Work. ACM Transactionson Office Information Systems, 4(3): 205-225. Gerson, E. M., & Star, S. L. 1986. Analyzing due process in the workplace. ACM Transactions on Information Systems, 4: 257-270. Glassman, P. A., Good, C. B., Kelley, M. E., Bradley, M., Valentino, M., Ogden, J., & Kizer, K. W. 2001. Physician perceptions of a national formulary. Am J Manag Care, 7(3): 241-251. Glennie, J. L., Woloschuk, D. M., & Hall, K. W. 1993. High technology drugs for cancer: the decision process for adding to a formulary. Pharmacoeconomics, 4(6): 405-413. Goldfarb, M. G., Hombrook, M. C., & Higgins, C. S. 1983. Determinants of hospital use: a cross-diagnostic analysis. Med Care, 21(1): 48-66. 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Goodman, L. S., Gilman, A. G., Hardman, J. G., & Limbird, L. E. 2001. Goodman & Gilman's the pharmacological basis o f therapeutics (10th / ed.). New York: McGraw-Hill Medical Pub. Division. Gouldner, A. W. 1954. Patterns o f industrial bureaucracy. Glencoe, 1 1 1 .,: Free Press. Gouldner, A. W. 1957. Cosmopolitans and locals: Toward an analysis of latent social roles. Administrative Science Quarterly, 2(3): 281-306. Green, J. A., Chawla, A. K., & Fong, P. A. 1985. Evaluating a restrictive formulary system by assessing nonformulary- drug requests. Am JH osp Pharm, 42(7): 1537-1541. Greenberg, J. 1987. A taxonomy of organizational justice theories. Academy of Management Review, 12(1): 9-22. Greenberg, J. 1990. Organizaitonal justice: Yesterday, today, and tomorrow. Journal o f Management, 16: 399-432. Greenberg, J. 1993. The social side of fairness: Interpersonal and informational classes of organizational justice. In R. Cropanzano (Ed.), Justice in the workplace: Approaching fairness in human resource management: 79-103. Hillsdale, NJ: Erlbaum. Greenberg, J. 1994. Using socially fair treatment to promote acceptance of a work site smoking ban. Journal o f Applied Psychology, 79(2): 288. Greenberg, J. 1996. The quest for justice on the job: Essays and experiements. Thousand Oaks, CA: Sage. Greenberg, J. 2001. Setting the justice agenda: Seven unanswered questions about "what, why, and how". Journal o f Vocational Behavior, 58: 210-219. Grol, R., Mokkink, H., Smits, A., van Eijk, J., Beek, M., Mesker, P., & Mesker- Niesten, J. 1985. Work satisfaction of general practitioners and the quality of patient care. Fam Pract, 2(3): 128-135. 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hackman, J. R., & Oldham, G. R. 1980. Work redesign. Reading, Mass.: Addison- Wesley. Hallstrom, K. T. 2000. Organizing the process of standardization. In B. Jacobsson (Ed.), A world o f standards'. 85-99. Oxford: Oxford University Press. Hanson, E., & Shepherd, M. 1994. Formulary restrictiveness in health maintenance organizations. Journal o f Social and Administrative Pharmacy, 11(1): 54-56. Hasty, M., Schrager, J., & Wrenn, K. 1999. Physicians' perceptions about managed care restrictions on antibiotic prescribing. J Gen Intern Med, 14(12): 756-758. Hefner, D. L. 1979. A study to determine the cost-effectiveness o f a restrictive formulary: The Louisiana experience. Washington, D. C: National Pharmaceutical Council. Hellerstein, J. K. 1998. The importance of the physician in the generic versus trade name prescription decision. Rand JEcon, 29(1): 108-136. Horn, S. D., Sharkey, P. D., Tracy, D. M., Horn, C. E., James, B., & Goodwin, F. 1996. Intended and Unintended Consequences of HMO Cost-Containment Strategies: Results from the Managed Care Outcomes Project. The American Journal o f Managed Care, 2(3): 253-264. Horwitz, R. I., Viscoli, C. M., Clemens, J. D., & Sadock, R. T. 1990. Developing improved observational methods for evaluating therapeutic effectiveness. Am J Med, 89: 630-638. Iezzoni, L. I. 1997. The risks of risk adjustment. Jama-Journal o f the American Medical Association, 278(19): 1600-1607. Iglehart, J. K. 1999. The American health care system-expenditures. N Engl J Med, 340(1): 70-76. Jaccard, J., Turrisi, R., & Wan, C. K. 1990. Interaction effects in multiple regression. Newbury Park, CA: Sage. 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Jackson, S. E., & Schuler, R. S. 1985. A meta-analysis and conceptual critique of research on role ambiguity and role conflict in work setting. Organizational Behavior and Human Decision Processes, 36(1): 16-78. Jang, R. 1988. Medicaid formularies: a critical review of the literature. J Pharm Mark Manage, 2(3): 39-61. Johnson, A. N., Dowd, B., Morris, N. E., & Lurie, N. 1989. Differences in inpatient resource use by type of health plan. Inquiry, 26(3): 388-398. Judge, T. A., Thoresen, C. J., Bono, J. E., & Patton, G. K. 2001. The job satisfaction- job performance relationship: A qualitative and quantitative review. Psychological Bulletin, 127: 376-407. Kabanoff, B. 1991. Equity, equality, power and conflict. Academy o f Management Review, 16: 416-441. Kane, R. L. 1997. Understanding health care outcomes research. Gaithersburg, Md.: Aspen Publishers. Kim, W. C., & Mauborgne, R. A. 1993. Procedural justice, attitudes, and subsidiary top management compliance with multinationals' corporate strategic decisions. Academy o f Management Journal, 36(3): 502-526. Konovsky, M. 2000. Understanding procedural justice and its impact on business organizations. Journal o f Management, 26(3): 489-512. Konovsky, M. A., & Pugh, S. D. 1994. Citizenship behavior and social exchange. Academy O f Management Journal, 37(3): 656. Kozma, C. M., Reeder, C. E., & Lingle, E. W. 1990. Expanding Medicaid drug formulary coverage. Effects on utilization of related services. Med Care, 28(10): 963-977. Kravitz, R. L., & Romano, P. S. 1996. Managed care cost containment and the law of unintended consequences. American Journal o f Managed Care, 2(3): 323- 324. 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kreft, I. G. G., & Leeuw, J. d. 1998. Introducing multilevel modeling. London ; Thousand Oaks, Calif.: Sage. Kunin, C. M., Johansen, K. S., Woming, A. M., & Daschner, F. D. 1990. Report on a symposium of use and abuse of antibiotics worldwide. Review o f Infectious Disease, 12: 12-19. laffaldano, M. T., & Muchinsky, P. M. 1985. Job satisfaction and job performance: A meta-analysis. Psychological Bulletin, 97: 251-273. Langlois, R. N., & Robertson, P. L. 1992. Networks and innovation in a modular system: Lessons from the microcomputer and stereo component industries. Research Policy. Lave, J. R., & Frank, R. G. 1990. Effect of the structure of hospital payment on length of stay. Health Serv Res, 25(2): 327-347. Lessler, J., & Tourangeau, R. 1989. Questionnaire design in the cognitive research laboratory. Washington, DC: Government Printing Office. Leung, K., & Lind, E. A. 1986. Procedural justice and culture: Effects of culture, gender, and investigator status on procedural preferences. Journal of Personality & Social Psychology, 50(6): 1134-1140. Leventhal, G. S. 1980. What should be done with equity theory?: New approaches to the study of fairness in social relationships. In K. J. Gergen, M. S. Greenberg, & R. H. Willis (Eds.), Social exchange: Advances in theory and research: 27- 55. Plenum Press: N.Y. Levit, K., Cowan, C., Braden, B., Stiller, J., Sensenig, A., & Lazenby, H. 1998. National health expenditures in 1997: more slow growth. Health Aff (Millwood), 17(6): 99-110. Levitt, B., & March, J. G. 1988. Organizational Learning. Annual Review o f Sociology, 14: 319-340. 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lind, E. A., & Tyler, T. R. 1988. The social psychology ofprocedural justice. New York, NY: Plenum Press. Locke, E., & Schweiger, D. 1979. Participation in decision making: One more look. New Directions in Organizational Behavior, 1: 265-339. Lu, Z., & Comanor, W. 1998. Strategic pricing of new pharmaceuticals. Review of Economics and Statistics: 108-118. Lueck, S. 2001. Spending on Prescription Drugs Rose 19 Percent, Wall Street Journal. Mangione, T. W. 1995. Mail surveys : improving the quality. Thousand Oaks, Calif.: Sage Publications. Manheim, L. M., Feinglass, J., Shortell, S. M., & Hughes, E. F. 1992. Regional variation in Medicare hospital mortality. Inquiry, 29(1): 55-66. Mannebach, M. A., Ascione, F. J., Gaither, C. A., Bagozzi, R. P., Cohen, I. A., & Ryan, M. L. 1999. Activities, functions, and structure of pharmacy and therapeutics committees in large teaching hospitals. Am J Health Syst Pharm, 56(7): 622-628. March, J. G., & Simon, H. A. 1958. Organizations (Second ed.). Oxford: Basil Blackwell Inc. Martocchio, J. J., & Judge, T. A. 1995. When we don't see eye to eye: Discrepanceis between supervisors and subordinates in absence disciplinary decisions. Jounal o f Management, 21: 251-278. McFarlin, D. B., & Sweeney, P. D. 1992. Distributive and Procedural Justice as Predictors of Satisfaction with Personal and Organizational Outcomes. Academy o f Management Journal, 35(3): 626. Meindl, J. R. 1989. Managing To Be Fair: An Exploration Of Values,Motives,And L. Administrative Science Quarterly, 34(2): 252. 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Merton, R. K. 1940. Bureaucratic structure and personality. Social Forces, 18: 560- 568. Miceli, M. P., Jung, I., Near, J. P., & Greenberger, D. B. 1991. Predictors and Outcomes of Reactions to Pay-for-Performance. Journal o f Applied Psychology, 76(4): 508. Michaels, R. E., Cron, W., L., Dubinsky, A. J., & Joachimstahler, E. A. 1988. Influence of formalization on the organizational commitment and work alienation of salespeople and industrial buyers. Journal o f Marketing Research, 25(4): 376-383. Miller, K. I., & Monge, P. R. 1986. Participation, satisfaction, and productivity: A meta-analytic review. Academy o f Management Journal, 29(4): 727-753. Mintzberg, H. 1979. The structuring o f organizations : a synthesis o f the research. Englewood Cliffs, N.J.: Prentice-Hall. Moore, W. M., & Newman, R. J. 1993. Drug formulary restrictions as a cost- containment policy in medicaid programs. Journal o f Law & Economics, XXXVI: 71-97. Moorman, R. H. 1991. Relationship Between Organizational Justice and Organizational Citizenship Behaviors: Do Fairness Perceptions Influence Employee Citizenship? Journal o f Applied Psychology, 76(6): 845. Moran, D. W. 2000. Prescription drugs and managed care: can 'free-market detente' hold? Health Aff (Millwood), 19(2): 63-77. Mossholder, K. W., Bennett, N., & Martin, C. L. 1998. A multilevel analysis of procedural justice context. Journal o f Organizational Behavior, 19(2): 131- 141. Nash, D., Shulkin, D., Owerbach, J., & Owerbach, J. 1992. Physician attitudes toward managed care formularies: An initial survey. Journal o f Research in Pharmaceutical Economics, 4(2): 31-41. 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Naumann, S. E., & Bennett, N. 2000. A case for procedural justice climate: Development and test of a multilevel model. Academy o f Management Journal, 43(5): 881-889. Nelson, R. R., & Winter, S. G. 1982. An evolutionary theory o f economic change. Cambridge, MA: Belknap Press. Olsen, G. G., & Goodman, M. E. 1977. The cost-benefit of clinical pharmacy in a general hospital. Australian Journal o f Hospital Pharmacy, 7(3): 102-107. Qrr, J. E. 1996. Talking about machines : an ethnography o f a modem job. Ithaca, N.Y.:ILR Press. Ostroff, C. 1992. The relationship between satisfaction, attitudes, and performance: An organizational level analysis. Journal o f Applied Psychology. 963-974. Pedersen, C., Schneider, P., & Santell, J. 2001. ASHP national survey of pharmacy practice in hospital settings: Prescribing and transcribing-2001. Am J Helaht-Syst Pharm, 58: 2251-2266. Raelin, J. A. 1986. The clash o f cultures : managers and professionals. Boston, Mass.: Harvard Business School Press. Raudenbush, S. W., & Bryk, A. S. 2002. Hierarchical linear models : applications and data analysis methods (2nd ed.). Thousand Oaks: Sage Publications. Ray, W., Griffin, M., Schaffner, W., Baugh, D., & Melton, L. 1987. Psychotropic drug use and the risk ofhip fracture. N Engl J Med, 3(16): 363-369. Robbins, S. P. 2003. Organizational Behavior (10 ed.). Upper Saddle River: Prentice Hall. Robinson, J. C., & Casalino, L. P. 1996. Vertical integration and organizational networks in health care. Health Aff (Millwood), 15(1): 7-22. Romano, M. 2001. New ruling more than a bonus. Modern Healthcare, 31(4): 16. 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ryan, A. M., Schmit, M. J., & Johnson, R. 1996. Attitudes and effectiveness: Examining relations at an organizational level. Personnel Psychology: 853- 882. Schectman, J. M., Elinsky, E. G., Kanwal, N. K., & Ott, J. E. 1995. HMO physician attitudes toward drug cost containment strategies. HMO Pract, 9(3): 116-119. Schelling, T. C. 1978. Micromotives and macrobehavior (1st ed.). New York: Norton. Schonberger, R. 1986. World class manufacturing: the lessons o f simplicity applied. New York: Free Press. Schwartz, R. K., Soumerai, S. B., & Avom, J. 1989. Physician motivations for nonscientific drug prescribing. Soc Sci Med, 28(6): 577-582. Schweitzer, S. O. 2000. Differences in managed care drug formularies: what can consumers learn? Med Care Res Rev, 57(3): 326-339. Schweitzer, S. O., & Shiota, S. R. 1992. Access and cost implications of state limitations on Medicaid reimbursement for pharmaceuticals. Annual Review o f Public Health, 13: 399-410. Sheppard, B. H., Lewicki, R. J., & Minton, J. W. 1992. Organizational justice : the search fo r fairness in the workplace. New York: Lexington Books. Simon, H. A. 1976. Administrative behavior: a study o f decision-making processes in administrative organization (3d ed. / ed.). New York: Free Press. Sitkin, S. B. 1995. On the positive effect of legalization on trust. In R. J. Bies, R. J. Lewicki, & B. H. Sheppard (Eds.), Research on negotiation in organizations, Vol. 5: 185-217. Greenwich, CT: JAI Press. Sitkin, S. B., & Roth, N. L. 1993. Explaining the limited effectiveness of legalistic "remedies" for trust/distruct. Organization Science, 4(3): 367-392. 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Skarlicki, D. P. 1997. Retaliation in the workplace: The roles of distributive, procedural, and interactional justice. Journal o f Applied Psychology, 82(3): 434. Sloan, F. A., Gordon, G. S., & Cocks, D. L. 1993. Hospital drug formularies and use of hospital services. Med Care, 31(10): 851-867. Sloan, F. A., Whetten-Goldstein, K., & Wilson, A. 1997. Hospital pharmacy decisions, cost containment, and the use of cost- effectiveness analysis. Soc Sci Med, 45(4): 523-533. Smith, D. M., & McKercher, P. L. 1984. The elimination of selected drug products from the Michigan Medicaid formulary: a case study. Hospital Formulary, 19(5): 366-372. Smith, M. C., & MacLayton, D. W. 1977. The effect of closing a Medicaid formulary on the prescription of analgesic drugs. Hospital Formulary, 12(1): 36-38,41. Snizek, W., & Bullard, J. 1983. Perception ofbureaucracy and changing job satisfaction: A longitudinal analysis. Organizational Behavior and Human Perform. Soumerai, S. B., Ross-Degnan, D., Avom, J., McLaughlin, T., & Choodnovskiy, I. 1991. Effects of Medicaid drug-payment limits on admission to hospitals and nursing homes. N Engl J Med, 325(15): 1072-1077. Soumerai, S. B., Ross-Degnan, D., Fortess, E. E., & Abelson, J. 1993. A critical analysis of studies of state drug reimbursement policies: research in need of discipline. Milbank Q, 71(2): 217-252. Star, S. L. 1995. Ecologies o f knowledge: work and politics in science and technology. Albany: State University ofNew York Press. Staw, B. M., & Barsade, S. G. 1993. Affect and Managerial Performance: A test of the sadder-but-wiser vs. happier-and-smarter hypotheses. Administrative Science Quarterly, 38: 304-331. 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Stevens, F., Diedriks, J., & Philipsen, H. 1992. Physician satisfaction, professional characteristics, and behavior formalization in hospitals. Social Science and Medicine, 35(3): 295-303. Stevens, R. 1989. In sickness and in wealth : American hospitals in the twentieth century. New York: Basic Books. Suchman, L. A. 1987. Plans and situated actions : the problem o f human-machine communication. Cambridge Cambridgeshire ; New York: Cambridge University Press. Sweeney, P. D., & McFarlin, D. B. 1997. Process and outcome: Gender differences in the assessment of justice. Journal o f Organizational Behavior, 18: 83-98. Sweeney, P. D., McFarlin, D. B., & Inderrieden, E. J. 1990. Using relative deprivation theory to explain satisfaction with income and pay level: A multistudy examination. Academy o f Management Journal, 33(2): 423-436. Tate, R. L., & Wongbundhit, Y. 1983. Random versus nonrandom coefficient models for multilevel analysis. Journal o f Educational Statistics, 8: 103-120. Thibaut, J. W., & Walker, L. 1975. Procedural justice : a psychological analysis. Hillsdale, N.J.: L. Erlbaum Associates. Thompson, J. D. 1967. Organizations in action; social science bases o f administrative theory. New York,: McGraw-Hill. Tyler, T. 2000. Social justice: Outcome and procedure. International Journal o f Psychology, 35(2): 117-125. Tyler, T. R. 1990. Why people obey the law. New Haven: Yale University Press. Tyler, T. R., & Degoey, P. 1995. Collective restraint in social dilemmas: Procedural justice and social identification effects on support for authorities. Journal of Personality & Social Psychology, 69(3): 482-497. 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wagner, J. A. I. 1994. Participation's effects on performance and satisfaction: A reconsideration of research evidence. Academy o f Management. The Academy o f Management Review, 19(2): 312. Walker, L., Lind, E. A., & Thibaut, J. 1979. The relation between procedural justice and distributive justice. Virginia Law Review, 65: 1401-1420. Walser, B. L., Ross-Degnan, D., & Soumerai, S. B. 1996. Do Open Formularies Increase Access To Clinically Useful Drugs? Health Affairs, 15(3): 95 — 109. Welboume, T. M., Balkin, D. B., & Gomez-Mejia, L. R. 1995. Gainsharing and mutual monitoring: A combined agency-organizational justice interpretation. Academy o f Management Journal, 38(3): 881. Williams, D. R., Lavizzo-Mourey, R., & Warren, R. C. 1994. The concept of race and health status in America. Public Health Rep, 109(1): 26-41. Williamson, O. E. 1991. Comparative Economic Organization: The Analysis of Discrete Structural Alternatives. Administrative Science Quarterly, 36: 269- 296. Wood, D. 1992. The power o f maps: Guilford Press. Wynia, M. K., Cummins, D. S., VanGeest, J. B., & Wilson, I. B. 2000. Physician manipulation of reimbursement rules for patients: between a rock and a hard place [see comments]. Jama, 283(14): 1858-1865. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Appendix Table 29 List o f the variables Items Data sources Hospital effectiveness Pharmacy cost Length of stay Inpatient Discharge Data Drug standardization Size of drug formulary Original Hospital Surveys Strength of Enforcement Original Hospital Surveys Therapeutic interchange Original Hospital Surveys Use of formalized tools Original Hospital Surveys Formulary Incentive System Original Hospital Surveys Organizational justice Procedural fairness Original Hospital Surveys Patient characteristics Severity of illness Inpatient Discharge Data Severity of disease comorbidity Inpatient Discharge Data insurance Inpatient Discharge Data -Age category: 0-17 years 18-44 years 45-64 years Inpatient Discharge Data -sex: Male or female Inpatient Discharge Data -race/ethnicity Inpatient Discharge Data -insurance coverage: private insurance Medicare Medicaid HMO/Managed care Self-paid/uninsured Inpatient Discharge Data Hospital Characteristics -Teaching hospital: An American Medical Association-approved residency program or membership in the Council of Teaching Hospitals AHA Annual Survey of Hospitals -Ownership: Government-owned Investor-owned AHA Annual Survey of Hospitals -Hospital size: Bed size AHA Annual Survey of Hospitals Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. -Location: Urban or Rural AHA Annual Survey of Hospitals -Multihospital membership AHA Annual Survey of Hospitals -Centralized vs. decentralized pharmacy Original Hospital Surveys Wage: The HCFA Area Wage Index The Federal Register Geographical region: Florida, Illinois, & New York HMO penetration Centers for Medicare and Medicaid Services D esigning Effective Formulary Systems! A H ospital Pharm acist Survey IN S T R U C T IO N S 1. The focus o f this survey is the inpatient formulary systems in the hospital. The questionnaire should be completed by someone familiar with the inpatient formulary. 2. You will receive a summary o f results if you enclose .a business card in the return envelope. 3. Please complete and return the survey within 7 days of receipt to the address below, using the enclosed, preaddressed, postage-paid envelope: Seok-Woo Kwon Management and Organization Bridge Hall 306 University o f Southern California Los Angeles, CA 90089-0808 1. If a prescribe? wants to order a drug jggg listed oil the formulary, which of the following Ss the standard process? {C heck a il th a t apply.) a. □ The presenter needs to do no more than simply writing the order only b. □ The prescribec needs to communicate an oral request to authorized physician, pharmacy dept, or P&T committee c. DThe prescribec needs to submit a written request to authorized physician, pharmacy dept, or P&T committee d. □ Other (s pe cif) )' -________________________________________________________________ _______ 2. Suppose a patient has been admitted to your hospital and Is talcing a medication that had been prescribed before hospitalization. If the drug the patient is taking is not listed on your formulary, which is the most frequent course of action? (Checkotie.) a. O Convert the patient to a drug available in the formulary b. □ Procure and supply the non-formulary drug to continue the prescription c. 0 Ask the patient to bring in his/her medicine to the hospital d. 0 Other ____________________ . ................. ............... 1 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3. In the p u t 12 months, BOTmiwtttly. what percentage of non-formulary medicatio ns In each of the following categories were procured by the pharmacy for inpatient use? {C ircle o n e fo r each category.') 1 2 3 4 5 Under 10% 10-20% 21-30% 31-40% Above 40% a. Cardiovascular drugs...................................... 1 % (e.g., ACE inhibitors, Beta-blockers, CCBs) b. Anti-infective agents........................................... 1 2 (e.g., Macrolides, Quinolones, Aminoglycosides, Cephalosporins) c. Gastrointestinal d ru g s................................. (e.g., Hj-antagonists, Proton pump inhibitors) 4* Which o f the following drug classes, if any, have a f {C h e c k ify e sJ a. □ HMG-CoA reductase inhibitors b. □ ACE inhibitors c. □ Calcium channel blockers d. OThrombolytics e. □ Heparin and LMW heparins f. □ Macrolides g. □ .Penicillins h. 0 Quinolones i. □ Aminoglycosides j. □ Cephalosporins k. nH^antagonists 1. □ Proton pump inhibitors 5. If a practitioner w ants to add a new drug to the formulary, which of the following is the standard process? {C heck a ll th a t apply.) a. D The practitioner is .asked to be present during formulary review at the P&T committee meeting b. Q The practitioner is asked to, supply supporting materials to the P&T committee c. Q The practitioner is asked to have the request countersigned by a division or dept, head d. 0 The practitioner is asked to disclose any ties with the drug company e. Q Other (spedfii: ________ ____________________________________________________ 6. O n average, how long does it take from the time that a practitioner requests anew drug addition (to the formulary) to the time that it is available as a formulary drug?. ...................................... 7 a Please Indicate whether the following are members of the P&T committee* and whether they can vote. {C H E C K A L L T H A T A P P LY .) . hold b . hem notin g " T * tf+ a. □ Physician(s)............... .........I b. □ Medical directors)... ..... a c. □ Pharmacists)............ ...... D d. D N urse(s).................. .......a e. 0 Risk manager(s)....... ...... 0 f. D Case manager(s)...... ...... a a , hold b. hem voting mmbmbij) proihff & □ U R/Q A officer®........... h. □ Financial officers)......... ...a i. □ Purchasing manager®.,.... a j. other: ...a k. other: ...a 2 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 . To what extent does each o f the following statements tmieorNo S o m e yktd c r a te G r e a t V ery G r e a t apply to the P&T committee in your hospital? E xten t E xten t E xteat E xten t E xten t (C ircle one fa r each ite m .) a. The P&T committee and its members are respected by their colleagues for their Imowledge and judgments................... 1 2 3 4 5 h. T *he P S a T * corrun.tttr'? prov'dc* to brar.tihoners a full PBJ * c. The P&T committee uses peer-reviewed journal articles as part o f the formulary review.................................................. 1 2 3 4 5 d. The ?&T a terms: fcouormr analysis or cost impact evs‘uafeor< as pact of a w % formulary rrv.ew ., * > . S i . . e. The P&T committee uses quantitative analytic techniques as part of the formulary review.......................................................1 2 3 4 5 .£ The P&T commute* cosssdcES local standards o f medicfd p ra ctsa c pact d f iV formulary review ............... i l l l l l i l 3 4 5 g. The P&T committee receives comments and recommendations from prescri b e s as part o f the formulary review........................ 1 2 3 4 5 •tL : The P&T committee ccr.wcteas the overa.!jate asnacmteo wttn usmg a < 5 r< £ as oart oFttw form uj& rv- mv.ew ...................I . . .r - i. The P&T committee relies on prescribers> personal testimonials as part o f the formulary review................................1 s 3 4 S The P&T ucrnswWce sohots inputs all aoRaathft? arvo. subsnecudties: as part o f the fcrnv.osKy Eevww. ... . . . . 1 . li i i i l f c i ;I|? 3 4:. . . . k. Formulary decisions o f the P&T committee ace influenced by the hospital's membership in a buyinggroup..............................., 1 2 3 4 5 %. ■ The P&T committee wosks cibsely with fhe oudjiy assurance isswmnltteft sm a u fttatatt m& management committee t . . * 2 ... . . ■ 5 m. When a new drug is added to the formulary, the P&T committee seeks to delete another in the same therapeutic class................ 1 2 3 4 5 9, During the p ast6 m onths, how many P&T meetings (Including subcommittee m eetings) were held? (C h eck one.) a. □ None b. OOne c. □ Two d. □ Three e, 0 More than three 10. W hat percentage of the P&T members attend the P&T meetings on average? (C heck one.) a. 0 Under 25% b. 0 26-50% c. □ 51-75% d, 0 Aboue 75% 1L Please check "yes" or “no” for each item. the F& 7 committee ser; aee.^ . " :s b. Does the P&T committee report through the medical staff (vs. the hospital) structure for approval o f its policies?..................................................................................... □ Yea □ No x. *.,oe? the "WT c sm m ita « o e ct its members tc disclose m s P e> 'b r < < -R r ties with a drug compare?......... * , * x l l l l i l l l : a m * 3 S ® d. Is a practitioner who requested formulary review allowed to vote if (s)he is a voting member of the P&T committee?............................................................................ □ Yes □ No 3 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12. To what extent does each of the following statements uttieorMa Soma Moderate Great Very Great apply to your hospital? ( C i r c l e o n e fo r e a c h ite m .) E x ten t Extent Extent Extent Extent lormuiairy its pniic??; 2 * • b. Our practitioners prescribe the most cost-effective medication when there are several therapeutically equivale it drugs.............. ] 2 - c The icpA vie.tr, pharmacy s*cefc? only a singfc generic epyivaient 2 : 'il& S im * d. The inpatient phcrrr.acv stocks some non formulary drugs 1 2 . ■ e. The sR D atiefi* pharmac? i t » < ik c s ofzVcontract *mrri?ercv ; ' 4 t i f. The hospital accepts physician preference as a legitimate reason for approving non formulary drugs.................. .......................... 1 2 . ■ % . The* ha&s&ti svaU.aks ores.~r.bc; asfccrfT.ce sc . mftdicatic;: use do-icies 2 % h. The hospital monitors presctibers for use o f non-formulary drugs.......................................... 2 3 4 5 The nostonal notifies p/escr.^it?; when the? do rt comply with &u.snulsuypolicies ... , S . . j. The hospital uses the trend dat&of individual physicians on drug utilisation for medical staff credenbaling ............................1 2 3 4 5 tu The hospital, legplsEiy updates the mp&tkn? formulary . 1 l l * * * 5 1 . The hospital uses pharmacist interventions designed to help monitor prescriber compliance with established medication-use policies........... .......................... ............................1 2 3 4 5 m. Pharmacists call a ocsscr.b-s: each ame Anon-fbcraulary order :■ ...............1 ..............• : iS. Is the hospital involved in the activities for each category? (C h e c k n il t h a t a p p l y .) Heart attack Ulcer Pneumonia b. Comparing quality improvement results against those of other health cate organizations,,,, ......... 0 a 0 ( 1 Cse djgnnthms. oaetu-e-nrotoccis* ' T iib ii cf, i f checked, approximated wnat percent of patent* X 4 I are on applicable jWnMmes. ....................... 14. Which processes are used to communicate information on appropriate drug use to prescribers? { C h e c k a ll th a t a p p l y ,) a. □ In-service education meetings. f. 0 Individual consultation by pharmacists/ b. D Grand rounds presentations. physicians c. Q Newsletters/memos g. 0 Retreats d. □ Emails h. Q Other (specify):. e. □ Web pages/electronic access 4 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15. Please check "yes” or "no” for each item. Ttem the M o E m ste tsm te/pm xtsw c? «Scpk amoam a la t e a . dmsr amiastcQfi o f ......................... E m * b. Can the information system generate drug utilization data by prescribed................ □ Yes □ N o Are ihpfihf’ .nt phaz.nssy'serwcss prowded by decerttrafe&erf-chsic&i. pharmacists?. ,... Q ytm 3**:... d. Does your hospital have a designated utilization review pharmacist who deals full-time with quality and cost issues?............................................................. D Yes □ No process#'a request far a mtmukrg Addition is denied?.............. .............. ...................... ....... ........ T ,..... O lfe s O N # ■ f. Does vour hospital have an internal appeals process if a request for a non-Fnrmularv drug is denied?......................................................................................................................... 0 Yes □ N o £- Doer your hosniral ahow ohstrmacisfc tc Ibewseiihc.sJly substitute for certain medication getting approval from prescribe^ •••,,•„:• .v • 3 y « . □ N o h. Does your hospital provide the "relative cost" o f formulary drugs to prescribes?..., Doe?; yuui' hospi;«i restrict t;ie activities of pharmaceutical naies reps on its. □ Yes □ N o premises?. . .......... ... ...., .. , .. ... .. .. . HYm □ No i- Does your hospital provide financial incentives (e.g., bonuses) to staff pharmacists for pharmacy cost savings?................................................................................... , ............... □ Y es □ N o P oes your hospital pccvtde fm&nasl maa&X9& s btrfmgimweqr^pment) so d^icai jungEtt^-d? depa&m&sSs-for i& sspit& l soat , . , . . , . — S ¥ « s D N » j I Has the formulary or formulary policies undergone a major overhaul during the past two □ Yes □ No 16. Is the hospital part of a health system in w hich a parent organization owns the hospital? a. 0 Yes b. □ N o -* GO TO QUESTION 18 17. To w hat extent does each o f the following statements Little c* No Sem e Moderate Great VeryGrest apply to your H ealth System? Extent Extent Extent Extent Extent a. The Health System orovrdas financial incentives to its a'Prate hosrsiials foe coat savincs............................................. .................. 1 7, 2 4 K f b. Formularies and formulary policies are standardized across the Health System..................................................................1 2 3 4 5 ■ The Health System influences drug oroduet dertsisns and.policies in its affiliate hospitals ..................... .... • -.v t Z 3 .a r > d. The Health System seeks input from its affiliate hospitals when making drug product decisions and policies...................... 1 2 3 4 5 •it. when sm%m$dsug:js£Gdue?tits& a m and-pohdes,. . . . . . . . 2 3 4 5 f. The Health System considers the needs ofyouc hospital and prescribers when making drug product decisions and policies... 1 2 3 4 5 & The losal. pm stizz ftsnsleMs or:it» .................................. ............. « .........— * 2 3 4 5 h. Hospitals are free to reject or modify drug product decisions and policies of the Health System without penalty...... ............... 1 2 3 4 5 5 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18. For each item below, please check the box if it is included in the inpatient formulary and write “R” next to the box if It is restricted. The accuracy o f your answers here is critical to our study. 1 f you are not sure about some o f the items below, we encourage you to refer to the inpatient formulary. ______________ Aee Inhibitors ; Be«BRpr3 (eg , Latfiaai}.. ....... 0 Captopril (e.g., Capoten)........................... □ 0 Fosinopril (ag, Monopril)........................ □ .liantspa!feg,Ze«*dl).., ......... 0 Lisinopril-Hydrochlorothiazide (e.g., Piinside)..................................................... □ ;Mo«3tt|s£fl (pgt Dswaec} ........ 0 Perinetopril (e.g, Ateon)......................... 0 (fc g , A c c t^ S .},............__ Q . Ramipril (e.g., Altace).............................. □ Tr.tc',dciurcj (e g , IViaviki......................... Q Trandolapril/Verapamil (e.g., Tarka) 0 Beta-Blockcrs Aeebuaolnt Sm xsl). _______ 0 Atenolol (e.g., Tenormin)......................... □ Aceriola-ChiorthaiuloriS <e.a. Tfenoretic; .. .. 0 Betaxolol (e.g., Kedone)......................... □ .ijB.g.^Zisc}........., ; . ..... L ............... 0 Carteolol (e.g,, Cartxol) ......................... □ ;Cawedilol(»gf € « e g |...... ....... 0 Esmold (e.g., Brevibloc)....................... 0 Labetaioi (e.g, 1 iormodyns; ....... 0 Metoprolol (e.g, Lopressor).................. 0 Penbutolol (e.g, Levatol)........................ □ i & t d o l d ( e ^ , T R s f c s r s ) Q Propranolol (e.g, federal)........................ Q i S o f a f e ! f a g , , B t t t s p a t * } , .............................. jQ ■nmolol (e.g, Blocadren)........................ 0 Heparin and LMW heparins (M se g a sm ( e - g , 0 Enoxapann (e.g., Lovenox).................... □ p t jw iis .. . . . . . ........, , . , . . . . . , . . . . 3 Tinzaparin (e.g., Innohep)....................... □ Calcium channel blockers i A m l o s S f s k f f i N t s s a s e } 0 Amlodipine-Benazepril (e.g., Lotrel) □ (B qxafi |Kg» Vaaaa!) B Diltiazem (e.g, Cardizem)....................... 0 i F y o f f l j ^ ( ^ g . J P t e * s d 3 ) . . . . . ....................... 0 Isradipme (e.g., DynaCirc)...................... □ M i o w r f i g i i ® ( e g * C a r t e s e } . 0 Nifedipine (e.g, Adalat).......................... □ N a s o d i g a s ( 8 , g A . N « a o £ B g | ...................... Q Nisoldipine (e.g, Sular)........................... 0 jTemgunS 0 6 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Continued) Thrombolytics iAfeepiiwe (e.5 , AcS«*e)................... ... 0 ; Anistreplase (e.g., Eminase)............. .... □ - __ • , ■ ■ ■ Keteplase (e.g, Retavase;. ... □ . ■StesBKsbaa®# (e-gf, Sseplase).......... ... 0 Tenecteplase (e.g, TNKase)........... □ Bacanspicillin (e.g., Spectrobid)............. iCtoaariffin (e.gi, C3t»p®t) □ 0 : C M m ase (s^ A W b k I .,. ..... 0 Dieloxacillin (e.g., Pathodl)................. .. □ HMG-CoA reductase inhibitors iCa&nwdls. (e,g, Geodlsa}.......... 0 - Atewsstaiin (a.S> .Iiss»s),............... Fluvastatin (e.g., Lescol)................... 0 . . . . . □ Mezlocillin (e.g, Mezlin)......................... 'MrfdlHa (B.g.f fStSpesj,......................... □ 0 iLowstaia. (erg, Mwaoat)................ Pravastatin (e.g., Pravachol)............. Q - .... □ OxadEin (e.g, BactodE)......................... Brakt&m G Benzattete fa-g, Sicafe),.. □ a .. ; StamBtatin (e.g., Zoeor). ............ ..... 0 ' : Penicillin G Potassium (e.g, Pfderpen).. □ Macrolides iPeiwaEm; G Rrocaira $sg, WyeBia)...... Penicillin V (e.g, Yeetids)........................ . 0 a * - Piperadii®. and Tsecbwtam ' » » * h u t ;is.g, «&sspi>,................. ............... ...... u aaritbrom ydn (e.g., Biaxin)............. □ TicardEin (e.g, Ticar)............................. □ Diafitroasjeifi (e.g, DjBitbse)........ Erythromycin (e.g., E-Mycin).......... . ... Q .... □ Quinolones ■lasfhroiayca Bthfimmdmts Gij*oSe®sdft (e,g, CSpm)..................... 0 ..... Erythromycin Lactobionate ... i3 : Enoxadn (e.g, Penetrex),,...................... □ (e.g, Erythrodn)............................... .... □ ^Qtffitfitawott. (e®, Tcqdji)............... 0 . S r f6 « r a » .f c s n .@ te a ja s e ........ ...... 0 Levofloxadn (e.g., Levaquin)................. □ Troleandomydn (e.g., TAO)........... .... D ■LsaneSiaada (e g , f t a a s q w B } . .. a Penicillins Moxtfloxadn (e.g, Avelox)...................... (Kg* Maeostn)...................... □ a A m sasrflk fc-g.. MasaaSj.......... .... a OfLoxadn (e.g, Floxin).......................... □ AmoxidHin and Clavulanate (StffltaariB fe.g, Z^paa) — ............... 0 (e.g., Augmentin).............................. 0 Trovafioxadn (e.g., Trovan)................... □ 7 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Continued) A m inoglycosides A mtonti|e.|g,Assfcn). ......... B Gentamicin (e.g, Garamycin). ............... □ JfeKnydn ................................ Q Streptomycin.................................... 0 ■IfctaKBjrata feg,, Nebdsi) ...... 0 C ephalosporins iCfe&dor(«g,Cedar).. ....... Q Cefadroxil (e.g., Duricef)....................... 0 ! C e f t m « f a k ...... £J Cefazolin (e.g., Ancef)............................. 0 Cefifcff {e.g, QmnsceCi, ......... ., 3 Cefepime (e.g, Maapime)....................... □ Cefiaote (•# » S q p at).............. 0 Cefopetazone (e.g., Cefobid),................. □ iCefiataaJs* (e,g» Cteferaa) ......... 0 Cefotetan (e.g, Cefotan)......................... 0 Ce&atia fe-g, Mefeba}.....,. ............ 0 Cefpodosdme (e.g., Vantan)..................... □ , € f i f j s r « I ( & g , O f i a l ) . .;.. . . . . . . . . . . . . , . . . . . 0 Ceftazidime (e.g, Ceptaz).................... □ CeffiSwteft («tg, C ede),.— ........ 0 Ceftizoxime (e.g., CeftZox)................... 0 !CeftQ»»s»(e.g}Ito«f4t«)... ........... 0 Ceforoxime (e.g., Ceftin) ...................... □ :<kpfestos» (s.g, Kssfiex),.................. Q Cephradine (e.g., Yelosef)....................... 0 M iscellaneous Antibiotics Chloramphenicol (e.g, Chloromycetin).,,. 0 Smpews®, andC Issm * Sodium . . . . . . . . . . 0 Loracarbef(e.g, Lorabid)............................ 0 (e.g,. Meawn}... — .......... Q Miscellaneous Anti-infectives Ateysaqpsae fs g , M e j s - o n ) ..............0 Clindamycin (e.g, Cleocin)..........................□ iCo“ttiB»»»ole.(a,g., Saettitti)................ 0 Iinezolid (e.g., Zyrox)..................................□ •MeawBAiaote (e.gs HsgfJ) , ............ 0 Pentamidine (e,g, Nebupent)................. 0 I ’ SKnegBanfte (eg , KfadBwwn) ......... 0 Vancomycin (e.g., Vancodn)................... □ U,-antagonists ;CisKSsSt»(e'.g.,T^sae!:)... 3 Famotidine (e,g, Pepcid)........................ □ (b & a tf c f a e (f e g ,A a d d h . 0 Ranitidine (e.g., Zantac)........................... 0 Proton pum p inhibitors i i t a s o f M S r f O Omeprazole (e.g, Prilosec)..........................0 ;Par>ts§®»ie Prtssoiss) ...0 Rabeprazole (e.g., Adphex).........................0 THANK YOU. 8 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The effects of case mix, hospital competition, and managed care penetration on hospital costs and revenues

PDF

The role of cognition and emotion regulation in conflict: A study of the impact on organizational and virtual teams

PDF

No laughing matter: The impact of supervisor humor on leader -member exchange (LMX) quality

PDF

The effects of Medi -Cal managed care on access to mammography screening for women residing in Los Angeles County

PDF

Performance management in health care in Iceland

PDF

Toward a theory of networks and network effectiveness: An analytical framework for the study of multilevel networks in health care service delivery

PDF

The effects of explicitly vs. implicitly valued premiums on consumer evaluation and choice

PDF

Physician profiling and clinical pathways: Combining the tools to change physician resource utilization

PDF

Nurse executive leadership and the relationship to organizational commitment among nurses

PDF

Prescription drug profiles as health risk adjusters in capitated payment systems: An applied econometric analysis

PDF

A quantitative and qualitative evaluation of the least burdensome provisions of the Food and Drug Administration Modernization Act of 1997

PDF

Task formalization and organizational performance: A learning-oriented synthesis of perspectives

PDF

Health care finance reform and market failure at King/Drew Medical Center, Los Angeles: An analysis of pediatric social outcomes in an urban public safety -net hospital

PDF

Six -month -one -day employee stock option repricing: An examination of accounting considerations and incentive implications

PDF

A study of employee health plan choice and medical cost: Panel data probit regression and sample selection model

PDF

Out -of -pocket health expenditures by older adults in relation to age, race, and insurance

PDF

Effects of a formulary expansion of the use of SSRIs and health care services by depressed patients in the California Medicaid program

PDF

Using network perspective to examine the organization of community -based elder care systems across four communities

PDF

The extent and effectiveness of nursing home regulation in the 50 states

PDF

Optimism and accuracy in forecasts of financial analysts: The effects of earnings predictability and experience

Asset Metadata

Creator Kwon, Seok-Woo (author)

Core Title Does procedural fairness matter in standardization? An examination of a drug standardization process in hospitals and its impact on hospital effectiveness

Degree Doctor of Philosophy

Degree Program Business Administration

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag business administration, management,health sciences, health care management,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-640757

Unique identifier UC11340038

Identifier 3116735.pdf (filename),usctheses-c16-640757 (legacy record id)

Legacy Identifier 3116735.pdf

Dmrecord 640757

Document Type Dissertation

Rights Kwon, Seok-Woo

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA