University of Southern California Dissertations and Theses

Evaluating approaches to reduce inappropriate antibiotic use in the United States

(USC Thesis Other)

Evaluating approaches to reduce inappropriate antibiotic use in the United States

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

EVALUATING APPROACHES TO REDUCE INAPPROPRIATE ANTIBIOTIC USE IN THE
UNITED STATES

by
Cynthia Lee Gong, Pharm.D.

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PHARMACEUTICAL ECONOMICS AND POLICY)

December 2017

Copyright 2017 Cynthia Lee Gong
ii

Contents
Acknowledgements ...................................................................................................................................... iv
Chapter 1: Introduction ................................................................................................................................. 1
References ........................................................................................................................................ 5
Chapter 2: Prescriber Preferences for Behavioral Economics Interventions to Improve Treatment of Acute
Respiratory Infections: A Discrete Choice Experiment ................................................................................ 6
Abstract ............................................................................................................................................ 6
Introduction ...................................................................................................................................... 8
Methods ........................................................................................................................................... 9
Results ............................................................................................................................................ 12
Discussion ...................................................................................................................................... 17
Conclusion ..................................................................................................................................... 20
Figure 1: Discrete choice experiment treatment scenarios ............................................................ 21
Appendix ........................................................................................................................................ 22
References ...................................................................................................................................... 25
Chapter 3: The use of behavioral economics and social psychology to improve treatment of acute
respiratory infections (BEARI): a cost-effectiveness analysis .................................................................... 27
Abstract .......................................................................................................................................... 27
Introduction .................................................................................................................................... 28
Methods ......................................................................................................................................... 29
Results ............................................................................................................................................ 33
Discussion ...................................................................................................................................... 34
Conclusion ..................................................................................................................................... 36
Table 1: Model Inputs .................................................................................................................... 37
FIGURE 1: Markov Model Structure ............................................................................................ 39
FIGURE 2: Cost-Effectiveness Plane ............................................................................................ 40
References ...................................................................................................................................... 43
Chapter 4: Management of Asymptomatic Term & Late Preterm Newborns Exposed to Maternal
Intrapartum Fever: A Societal Cost Benefit Analysis of the Proposed “Triple I” Algorithm ..................... 46
Abstract .......................................................................................................................................... 46
Introduction .................................................................................................................................... 47
Methods ......................................................................................................................................... 48
Results ............................................................................................................................................ 51
Discussion ...................................................................................................................................... 52
iii

Conclusion ..................................................................................................................................... 55
Figure 1: Decision Model – Management of Asymptomatic Term and Late Preterm Newborns
Exposed to Maternal Intrapartum Fever ........................................................................................ 56
Figure 2: One-Way Sensitivity Analysis ....................................................................................... 57
Figure 3: Probabilistic Sensitivity Analysis (Likelihood of Net Benefit) ...................................... 58
Table 1: Intraamniotic Infection Classification ............................................................................. 59
Table 2: Decision Model Inputs ..................................................................................................... 60
Table 3: Base Case Clinical Outcomes
§
......................................................................................... 61
References ...................................................................................................................................... 62
Chapter 5: Summary and Future Research Directions ................................................................................ 66

iv

Acknowledgements
Special thanks to my advisor Dr. Joel Hay for his support, guidance, and the motivation required
to complete this dissertation, and for his advice on anything and everything whenever I needed it. Thank
you for believing in me even when I stopped believing in myself.
Thank you to my committee members: Dr. Kenneth Zangwill for his invaluable clinical guidance
and input, and Dr. Jason Doctor for providing support and guidance for the BEARI project, without which
none of this work would have been possible.
I would also like to acknowledge the individuals who have helped develop and guide me
throughout my graduate school years. Thank you to Drs. John Romley, Steven Fox, and Daniella
Meeker, without whose collaboration, expertise, and support this research could not have happened, as
well as Dr. Shom Dasgupta, whose clinical expertise was invaluable to my work. Special thanks to Dr.
Jeffrey McCombs for encouraging me to pursue this course of study.
To my classmates: thank you for the endless support, chats, and advice as we struggled through
coursework and projects along the way. I am grateful to have gotten to know each and every one of you,
and am lucky to call you my friends. I know I can always reach out to you if I ever have any questions or
need advice.
To Ngoc, my husband: I am grateful for your infinite patience and support, and for providing
realistic advice even at the most trying of times throughout this process.
Finally, to my parents: thank you for everything that you do, because I certainly would not have
achieved all that I have without your discipline, support, and encouragement, which have made me who I
am today.

1

Chapter 1: Introduction
Antibiotics represent one of the greatest achievements in modern medicine to date, with the
discovery of penicillin in 1928.[1] As antibiotics evolved to fight ever more infections however, so too
did the bacteria, necessitating the implementation of antibiotic stewardship programs in hospitals
nationwide, and a Presidential Executive Order in 2014 to address antimicrobial resistance.[2] In fact, the
Centers for Disease Control and Prevention (CDC) estimate that antibiotic resistance alone causes 2
million infections and 23,000 deaths each year in the United States.[3] The CDC classifies drug-
resistance threats according to the increasing hazard levels “concerning”, “serious”, and “urgent”. The
most common cause of bacterial upper respiratory infections, Streptococcus pneumoniae, is categorized
as a serious threat, while Group B Streptococcus (GBS), which is the most common causative pathogen
of neonatal sepsis and meningitis, is classified as “concerning”.[4-6] We emphasize these two organisms
because among adults in the United States, most antibiotics are prescribed for acute upper respiratory
infections (ARIs), about half of which are issued for non-bacterial diagnoses, while among the neonatal
population, nearly all infants born to mothers with intrapartum fever are administered empiric antibiotics
even in the absence of clinical signs/symptoms of infection, despite the risk of true infection being quite
low.[7]
The most effective way to reduce inappropriate antibiotic use is by reducing inappropriate
antibiotic prescriptions, which requires changing prescribing behavior. Evidence-based clinical
guidelines are one means of achieving this, by helping guide physician behavior towards more judicious
prescribing habits to improve quality, outcomes, and cost-effectiveness of care. However, physician
guideline adoption and adherence is estimated to be just 47% among primary care providers due to four
main barriers: 1) Weak financial incentives to change behavior due to volume-based vs. outcomes-based
payments; 2) Insufficient access to point-of-care guidelines due to limited health information technology
resources; 3) Lack of feedback on physician performance, leading to the development of specific beliefs
and habits that may not be evidence-based; 4) Lack of transparency in the guideline development process,
2

and the inability of guidelines to account for the complexities associated with real clinical practice.[8] To
overcome these barriers, the New England Health Institute offers several possible solutions, including
clinician engagement in the guideline development and review process, and improved feedback on
clinician behavior within his/her own practice relative to peers.[8] These concepts are consistent with the
idea that engaging clinicians is essential to the adoption and maintenance of interventions to improve
quality, and that behavioral incentives may be more impactful than financial ones in changing
behavior.[9]
Unfortunately, while improving prescribing behavior can successfully control antimicrobial
resistance in the hospital setting, the impact on reversing resistance in the community setting is unlikely
or at best, likely to occur at a very slow rate.[10-12] Nonetheless, there are additional benefits, from
minimizing common antibiotic adverse drug reactions that can require emergency department attention
(and thus unnecessary healthcare resource utilization), to reducing hospital lengths-of-stay and the
associated risks that come with a hospitalization, such as exposure to multi-drug resistant organisms,
Clostridium difficile, medication errors, and adverse drug reactions. However, it is unclear whether these
theoretical benefits of reduced antibiotic prescribing lead to decreases in unnecessary healthcare resource
utilization without increasing adverse clinical outcomes, and if so, at what cost. The economic impact
and costs associated with changing antibiotic prescribing behavior have previously not been quantified in
a transparent manner that incorporates the impact of antimicrobial resistance.[13-15] In addition, cost-
effectiveness evaluations of health information technology interventions designed to improve patient care
often fail to use comprehensive methods of economic evaluation, and fail to report results using a
standard metric, such as costs per quality-adjusted life years.[16, 17] Further, even if such interventions
are successful, previous studies have not evaluated the preferences of prescribing clinicians for
interventions that change their prescribing behavior, focusing instead on patient preferences for various
treatments or procedures.[18, 19] This is of particular interest as clinician engagement is crucial in the
development and longevity of clinical quality improvement initiatives.
3

Thus, we aim to describe clinician preferences for behavioral economics interventions and their
economic impact in the next two chapters of this dissertation. Our research draws data from the “Effect
of Behavioral Interventions on Inappropriate Antibiotic Prescribing Among Primary Care Practices”
(BEARI) study, which was a randomized controlled trial comparing three behavioral interventions to a
control group consisting of an educational module with a review of clinical guidelines for the treatment of
acute respiratory infections.[20] The three interventions included: 1) Suggested Alternatives (SA), which
presented electronic order sets suggesting nonantibiotic treatments; 2) Accountable Justification (JA),
which prompted clinicians to enter free-text justifications for prescribing antibiotics into patients’
electronic health records; and 3) Peer Comparison (PC), which sent periodic emails to clinicians that
compared their antibiotic prescribing rates with those of “top performers” (those with the lowest
inappropriate prescribing rates). The results of this trial indicated that only JA and PC had a significant
impact on reducing the rates of inappropriate antibiotic prescriptions, while SA was not significantly
different from the control group, nor significantly different from JA and PC.
Using a discrete choice experiment, Chapter 2 analyzes which of the three interventions clinicians
feel are most likely to improve their inappropriate antibiotic prescribing. The clinicians’ preferences are
then compared to their actual prescribing behavior from the randomized controlled clinical trial. The
analysis represents one of very few in the health economics literature that compares physicians’ revealed
preferences (i.e. true behavior) to their stated preferences (i.e. those reported via questionnaire), and
provides insight into how management-driven interventions might compare to clinician-driven ones.
Given the efficacy of the behavioral economic interventions in changing prescribing behavior, we then
analyze the cost-effectiveness of these interventions in Chapter 3, conducting a comprehensive economic
evaluation that includes the impact of antimicrobial resistance and provides outcomes in a standardized
format of costs per quality-adjusted life years.
In Chapter 4, we focus on the neonatal population, and analyze the societal cost-benefit of
managing neonates exposed to maternal intrapartum fever according to a newly proposed clinical
4

algorithm that minimizes empiric antibiotic use for all well-appearing infants >34 weeks gestational age,
instead recommending that only ill-appearing infants be treated with empiric antibiotics for presumed
GBS infection.[21] These newly proposed guidelines contrast with the currently established clinical
guidelines, which recommend empiric antibiotic use regardless of the clinical status of the infant due to
the severe consequences of neonatal early onset sepsis and/or meningitis.[22] As the American Academy
of Pediatrics Committee on Infectious Diseases is on the verge of releasing a statement on the use of
antibiotics in this specific population, our analysis is timely and may help guide the development of
clinical practice in this area.
Finally, we summarize the overall findings of each analysis in Chapter 5, and provide some
direction for future research in these areas.
5

References
1. American Chemical Society International Historic Chemical Landmarks. Discovery and
Development of Penicillin. 1999 [cited 2017 October 10]; Available from:
https://www.acs.org/content/acs/en/education/whatischemistry/landmarks/flemingpenicillin.html.
2. The White House, Executive Order - Combating Antibiotic-Resistant Bacteria, Office of the Press
Secretary, Editor. 2014.
3. Centers for Disease Control and Prevention. About Antimicrobial Resistance: Four Core Actions
to Fight Resistance. [Online] 2015 September 8, 2015 [cited 2017 January 23]; Available from:
https://www.cdc.gov/drugresistance/about.html.
4. Fokkens, W.J., R. Hoffmans, and M. Thomas, Avoid prescribing antibiotics in acute
rhinosinusitis. BMJ, 2014. 349: p. g5703.
5. Schwartz, L.E. and R.B. Brown, Purulent otitis media in adults. Arch Intern Med, 1992. 152(11):
p. 2301-4.
6. Sokol, W., Epidemiology of sinusitis in the primary care setting: results from the 1999-2000
respiratory surveillance program. Am J Med, 2001. 111 Suppl 9A: p. 19S-24S.
7. Escobar, G.J., et al., Stratification of risk of early-onset sepsis in newborns >/= 34 weeks'
gestation. Pediatrics, 2014. 133(1): p. 30-6.
8. Kenefick, H., J. Lee, and V. Fleishman, Improving Physician Adherence to Clinical Practice
Guidelines: Barriers and Strategies for Change. 2008, New England Healthcare Institute. p. 63.
9. Frolich, A., et al., A behavioral model of clinician responses to incentives to improve quality.
Health Policy, 2007. 80(1): p. 179-93.
10. Andersson, D.I. and D. Hughes, Antibiotic resistance and its cost: is it possible to reverse
resistance? Nat Rev Microbiol, 2010. 8(4): p. 260-71.
11. Andersson, D.I. and D. Hughes, Persistence of antibiotic resistance in bacterial populations.
FEMS Microbiol Rev, 2011. 35(5): p. 901-11.
12. Austin, D.J. and R.M. Anderson, Studies of antibiotic resistance within the patient, hospitals and
the community using simple mathematical models. Philos Trans R Soc Lond B Biol Sci, 1999.
354(1384): p. 721-38.
13. Hunter, R., Cost-effectiveness of point-of-care C-reactive protein tests for respiratory tract
infection in primary care in England. Adv Ther, 2015. 32(1): p. 69-85.
14. Michaelidis, C.I., et al., Cost-effectiveness of procalcitonin-guided antibiotic therapy for
outpatient management of acute respiratory tract infections in adults. J Gen Intern Med, 2014.
29(4): p. 579-86.
15. Oppong, R., et al., Cost-effectiveness of point-of-care C-reactive protein testing to inform
antibiotic prescribing decisions. Br J Gen Pract, 2013. 63(612): p. e465-71.
16. Bassi, J. and F. Lau, Measuring value for money: a scoping review on economic evaluation of
health information systems. J Am Med Inform Assoc, 2013. 20(4): p. 792-801.
17. O'Reilly, D., et al., The economics of health information technology in medication management:
a systematic review of economic evaluations. J Am Med Inform Assoc, 2012. 19(3): p. 423-38.
18. de Bekker-Grob, E.W., M. Ryan, and K. Gerard, Discrete choice experiments in health
economics: a review of the literature. Health Econ, 2012. 21(2): p. 145-72.
19. Clark, M.D., et al., Discrete choice experiments in health economics: a review of the literature.
Pharmacoeconomics, 2014. 32(9): p. 883-902.
20. Meeker, D., et al., Effect of Behavioral Interventions on Inappropriate Antibiotic Prescribing
Among Primary Care Practices: A Randomized Clinical Trial. JAMA, 2016. 315(6): p. 562-70.
21. Higgins, R.D., et al., Evaluation and Management of Women and Newborns With a Maternal
Diagnosis of Chorioamnionitis: Summary of a Workshop. Obstet Gynecol, 2016. 127(3): p. 426-
36.
22. Schrag, S., et al, Prevention of Perinatal Group B Streptococcal Disease – Revised Guidelines
from CDC. MMWR, 2002. 51(RR-11): p. 1*24.
6

Chapter 2: Prescriber Preferences for Behavioral Economics Interventions to Improve Treatment
of Acute Respiratory Infections: A Discrete Choice Experiment
Gong CL, Hay JW, Meeker D, Doctor JN. Prescriber preferences for behavioural economics interventions
to improve treatment of acute respiratory infections: a discrete choice experiment. BMJ Open
2016;6:e012739.
Abstract

Objective: To elicit prescribers’ preferences for behavioral economics interventions designed to reduce
inappropriate antibiotic prescribing, and compare these to actual behavior.
Design: Discrete choice experiment.
Setting: 47 primary care centers in Boston and Los Angeles.
Participants: 234 primary care providers, with an average 20 years of practice.
Main Outcomes and Measures: Results of a behavioral economic intervention trial were compared to
prescribers’ stated preferences for the same interventions relative to monetary and time rewards for
improved prescribing outcomes. In the randomized controlled trial (RCT) component, the three
Computerized Prescription Order Entry -triggered interventions studied included: Suggested Alternatives
(SA), an alert that populated non-antibiotic treatment options if an inappropriate antibiotic was
prescribed; Accountable Justifications (JA), which prompted the prescriber to enter a justification for an
inappropriately prescribed antibiotic that would then be documented in the patient’s chart; and Peer
Comparison (PC), an email periodically sent to each prescriber comparing his/her antibiotic prescribing
rate with those who had the lowest rates of inappropriate antibiotic prescribing. A discrete choice
experiment (DCE) study component was administered to determine whether prescribers felt SA, JA, PC,
pay-for-performance, or additional clinic time would most effectively reduce their inappropriate antibiotic
prescribing. Willingness-to-pay (WTP) was calculated for each intervention.
Results: In the RCT, Peer Comparison and Accountable Justifications were found to be the most effective
interventions to reduce inappropriate antibiotic prescribing, whereas Suggested Alternatives was not
significantly different from controls. In the DCE however, regardless of treatment intervention received
during the RCT, prescribers overwhelmingly preferred Suggested Alternatives, followed by Peer
Comparison, then Accountable Justifications. WTP estimates indicated that each intervention would be
significantly cheaper to implement than pay-for-performance incentives of $200/month.
Conclusion: Prescribing behavior and stated preferences are not concordant, suggesting that relying on
stated preferences alone to inform intervention design may eliminate effective interventions.
7

Article Summary: Strengths and Limitations of this Study
• In this discrete choice experiment, prescribers were asked about which interventions they
felt would be most effective in reducing their rates of inappropriate antibiotic
prescriptions for acute respiratory infections; such information can be valuable when
developing clinical quality improvement interventions.
• This is one of few studies in the healthcare literature that not only elicits prescriber
preferences using a discrete choice experiment, but also compares these stated
preferences to actual prescribing behavior as observed in a randomized controlled trial.
• Results indicate that stated and revealed preferences are not concordant, suggesting that
clinical quality improvement programs should not rely solely on clinician input, but
instead combine both expert-driven and clinician-driven approaches.
• The discrete choice experiment may not necessarily capture true preference, but instead
be a reflection of convenience and ease of use of a specific clinical quality improvement
intervention.
• Stated preferences for this group of healthcare providers may not necessarily reflect those
of a national sample of providers.

8

Introduction
According to the Centers for Disease Control and Prevention, up to 50% of antibiotics are not
optimally prescribed, leading to an estimated 2 million illnesses and 23,000 deaths due to antibiotic
resistance alone[1]. Most antibiotics in the United States are prescribed for acute respiratory tract
infections (ARIs), and about half of these prescriptions are issued to patients with nonbacterial
diagnoses.[2, 3] Despite attempts to curb inappropriate antibiotic prescribing through interventions such
as physician and patient education, electronic clinical decision support, and financial incentives, these
have only resulted in modest reductions in antibiotic prescribing rates for nonbacterial ARIs.[4]
Clinical Quality Improvement interventions frequently rely upon changing clinicians’ practice
behaviors, such as reducing orders for inappropriate treatments or diagnostic tests. The consensus-
recommended best practices in design and implementation of quality improvement interventions include
some component of “local participatory” approaches, i.e. working with frontline staff, including the target
population.[5-7] This engagement may directly or indirectly influence the intervention design, but it is
unclear if this practice yields the optimal design. For example, both theory and survey results suggest that
physicians are likely to indicate preferences for direct financial incentives (bonuses).[8-10] However,
studies of effectiveness have shown mixed results of direct incentives in practice.[10, 11] This raises
questions regarding whether stated preferences for intervention features elicited in a participatory process
should be incorporated into design.
One approach to changing prescribing behavior applies ideas from the behavioral sciences, using
social cues and subtle changes in the clinic environment to influence clinical decision making.[12, 13] In
fact, the UK government implements behavioral economics “nudges” into its health services, and a recent
study in the UK found that social norm feedback from England’s Chief Medical Officer was highly
effective in reducing inappropriate antibiotic prescribing at a low cost.[14-16] In the US, the “Use of
behavioral economics and social psychology to improve treatment of acute respiratory infections
(BEARI)” study, a multi-site cluster randomized-controlled trial, applied behavioral techniques and
9

assessed the impact of various behavioral interventions on the rates of inappropriate antibiotic prescribing
in various practices in Illinois, Massachusetts and Southern California.[17, 18] As part of the BEARI
study, a discrete choice experiment (DCE) was conducted to elicit prescribers’ stated preferences to
evaluate prescriber preferences for one intervention compared to another.
The use of discrete methods to elicit preferences for various programs and interventions in
healthcare has significantly increased in recent years.[19-22] These methods have been utilized in the
context of health system reforms or quality improvement programs, and health care policies, programs,
services, incentives, and interventions.[9, 20] While they provide valuable feedback regarding factors
that should be considered for a given program or intervention, few studies in healthcare have evaluated
how stated preferences compare to real-life behavior. Those that have assessed external validity have
generally found that stated preferences are consistent with actual decision-making behavior on an
aggregate level, while individual level concordance is limited.[23, 24] Previous studies evaluating an
individual’s decisions for vaccination or disease screening indicated a positive predictive value for DCEs
of 85%, but that the negative predictive value was only 26%; however, the authors noted that the majority
of people opt for preventive healthcare in these situations, yielding the overall predictive values of the
DCE.[21, 25, 26] Furthermore, only one study has specifically considered physician decision-making –
the ultimate driver of quality and cost of care.[27]
Based on these studies, DCE-elicited responses may generally reflect real-life behaviors.
However, few studies have tested behavior in a randomized controlled trial and compared the results to
DCE-elicited preferences for the same group of individuals (particularly physicians) to validate the DCE
responses.[28] Thus, the objective of this study was to elicit prescriber preferences for different
behavioral economics interventions to reduce antibiotic prescribing, and compare these to actual
prescribing behavior as revealed in the BEARI study.
Methods
Design
10

Details for the BEARI study have been described elsewhere.[17] Briefly, this was a multicenter
trial conducted in Illinois, Massachusetts and Southern California to determine whether behavioral
economic interventions could influence prescriber behavior. The interventions implemented in the
BEARI study included: 1) Suggested Alternatives (SA), which utilized computerized clinical decision
support to suggest both over-the-counter (OTC) and prescription non-antibiotic treatment choices to
clinicians prescribing an antibiotic for an ARI; 2) Accountable Justifications (JA), which prompted
clinicians to enter an explicit justification when prescribing an antibiotic for an ARI that would then
appear in the patient’s electronic health record; and 3) Peer Comparison (PC), which was an email sent
periodically to each prescriber that compared his/her rate of inappropriate antibiotic prescribing relative to
top-performing peers[17]. All clinicians received an education module reviewing ARI diagnosis and
treatment guidelines, and each was assigned to 0, 1, 2, or all 3 interventions. Interventions were in place
for 18 months at each practice site.
Upon completion of the study, all prescribers were asked to complete a computerized exit survey,
which assessed prescribers’ level of satisfaction with each of the interventions, and included the discrete
choice experiment (DCE). A DCE is used to determine an individual’s preferences for specific
alternatives given a specific scenario, rather than observing an individual’s behavior in real markets to
determine his/her revealed preferences.[29] The individual is presented with a pair of choices with
varying levels of attributes, and is asked to choose the preferred alternative based on the attributes
presented [refer to Supplement for details].
In the DCE, subjects were presented with the following scenario, followed by a set of ten choice
pairs from which prescribers had to choose a preferred alternative that they felt would most likely reduce
their ARI antibiotic prescribing:
Suppose that your Health Care Organization implements various measures
to reduce overprescribing of antibiotics for acute respiratory infections
(e.g. Acute bronchitis, Acute Pharyngitis, Sinusitis, etc.). Assume
implementation with an electronic health record system with electronic
prescribing options and physician monitoring that can be automatically
11

enabled or disabled. You will be provided with a short patient handout
that discusses antibiotic overprescribing for acute respiratory infections.
The five attributes included for each alternative were Accountable Justifications (JA), Suggested
Alternatives (SA), Peer Comparison (PC), Pay for Performance ($100, $200, or $0), and Additional Time
spent with the patient (5 minutes or 0 minutes) (Table 1).
Table 1: Attributes and Descriptions
Attribute Description
Suggested Alternatives (SA)
If you enter a target ARI diagnosis into the patient electronic health record the
EHR will prominently display a list of appropriate non-antibiotic prescription
and non-prescription treatment alternatives to antibiotics.
Accountable Justification (JA)
If you order an antibiotic for a patient with an ARI, a new EHR screen will pop
up that asks you to enter a short written justification for why the antibiotic
prescription was necessary. What you write in this screen becomes part of the
patient’s permanent medical record, and therefore visible to other providers.
Peer Comparison (PC)
Each other week, you and your peers will receive in your work mailboxes (or
email) an updated ranking comparing your antibiotic prescribing rate to the top
performing (i.e., “best”) decile of your clinical peers.
Pay for Performance (P4P)
If you are able to reduce your rate of antibiotic prescribing for all of your ARI
patients to the lowest 10th percentile of your peers you will receive a
predetermined monthly reward payment of either $100 or $200.
Additional Time (AT)
Your productivity management record system will change the RVU/visit time
allowed for each ARI patient to increase by five minutes per visit as you
respond to the patient’s concerns regarding the alternative treatments for their
ARI diagnosis. If you achieve the antibiotic prescription reduction goal of
reducing your antibiotic prescriptions by 50% this five minute per visit time
will be preserved.

These attributes and their levels were chosen based on interviews with prescriber focus groups in Los
Angeles. Each attribute was toggled on or off in the choice pairs presented to respondents (Figure 1).
Respondents were asked to evaluate a total of 10 choice pairs which were assigned using a fractional
factorial design with orthogonality of main effects.[30]
A pilot survey was administered to a small number of prescribers in Los Angeles and Boston who
were not involved in the BEARI study and thus not exposed to any of the interventions, and to a subset of
prescribers in Chicago who did receive exposure to the BEARI interventions. The DCE was then
administered as part of the exit survey for all prescribers involved in the BEARI study.
12

Data Analysis
Baseline provider characteristics were collected at the start of the study, and information about
the clinic environment and patient population mix were collected during the exit survey. Descriptive
statistics were performed on these data.
Preference data from the DCE were analyzed using multinomial logit and mixed logit models
drawn 500 times to obtain a robust output. In addition, the models were run on the subset of responders
who completed 6 or more of the 10 choice tasks, and the results compared to those of the entire sample.
The impact of intervention assignment on stated preference was evaluated by including intervention
assignment as an explanatory variable in the mixed logit model, and interacting intervention assignment
on each program. Willingness-to-pay (WTP) for each intervention type was calculated by taking the ratio
of the estimated coefficient for a given attribute to the cost coefficient, for example:
𝑊𝑇𝑃 𝑃𝑒𝑒𝑟 𝐶𝑜𝑚𝑝𝑎𝑟𝑖𝑠𝑜𝑛 =
𝛽 𝑃𝑒𝑒𝑟 𝐶𝑜𝑚𝑝𝑎𝑟𝑖𝑠𝑜𝑛 𝛽 𝑃𝑎𝑦 𝑓𝑜𝑟 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐 𝑒 . This calculation was done for each intervention to quantify the
monetary value of each attribute included in the DCE.[31]
All data were analyzed in Stata 12.0.[32]
Results
Details for the outcomes of the BEARI study have been described elsewhere.[18] Briefly, the
BEARI study found that the most significant reduction in the rates of inappropriate antibiotic prescribing
was achieved with Accountable Justifications and Peer Comparison, as compared to the education
control. Suggested Alternatives (SA) yielded a non-significant reduction in inappropriate antibiotic
prescribing rates. [18] These results suggest that Peer Comparison and Accountable Justifications may be
the most effective interventions to influence prescribing.
Of the 253 prescribers recruited in the BEARI RCT study, n = 234 responses were received for
the DCE survey, although only 157 provided responses to all ten treatment scenarios. Of the responders,
13

26 were exposed to all three BEARI interventions, 35 to Suggested Alternatives and Accountable
Justifications, 32 to Suggested Alternatives and Peer Comparison, and 23 to Accountable Justifications
and Peer Comparison in the BEARI RCT. Prescribers were primarily medical doctors with an average of
20 years of clinical practice, with more males in the control group (Table 2).
Table 2: Baseline Demographics
Total (n=234) Control (n=24) Treatment (n=210)
Characteristics Number (%) Number (%) Number (%)
Age (years, mean) 48.5 47.8 48.6

Gender
Female 137 (58.5%) 12 (50.0%) 125 (59.5%)
Male 81 (34.6%) 11 (45.8%) 70 (33.3%)
No Response 16 (6.8%) 1 (4.2%) 15 (7.1%)

Years of Practice (mean) 19.9 19.0 21.2

Professional Credentials
Medical Doctor 174 (74.4%) 18 (75.0%) 156 (74.3%)
Nurse Practitioner 24 (10.3%) 2 (8.3%) 22 (10.5%)
Physician Assistant 13 (5.6%) 2 (8.3%) 11 (5.2%)
Doctor of Osteopathy 6 (2.6% 1 (4.2%) 5 (2.4%)
No Response 17 (7.3%) 1 (4.2%) 16 (7.6%)

Specialty
Internal Medicine 126 (53.5%) 14 (58.3%) 112 (53.3%)
Pediatrics 23 (9.8%) 2 (8.3%) 21 (10%)
Family / General Practice 28 (12.0%) 3 (12.5%) 25 (11.9%)
Geriatrics 3 (1.3%) 0 3 (1.3%)
Rheumatology 1 (0.43%) 0 1 (0.43%)
Preventive Medicine 1 (0.43%) 0 1 (0.43%)
Infectious Disease 1 (0.43%) 0 1 (0.43%)
Midwife 1 (0.43%) 0 1 (0.43%)
No Response / Other 50 (21.4%) 5 (20.8%) 45 (21.4%)

Results of the mixed logit model are presented in Table 3.
Table 3: BEARI Study Mixed Logit Regression Results
14

Exposure Group OR [95% CI] Coefficient Std. Err. [95% CI] p-value p-value*
Full Sample (n=239) <0.001
Pay for Performance 1.007 [1.006, 1.008] 0.007 0.001 [0.006, 0.008] 0.000
Suggested Alternatives 2.108 [1.727, 2.573] 0.746 0.102 [0.546, 0.945] 0.000
Accountable Justifications 1.092 [0.85, 1.404] 0.088 0.128 [-0.162, 0.339] 0.490
Peer Comparison 1.446 [1.184, 1.766] 0.369 0.102 [0.169, 0.568] 0.000
Additional Time 1.159 [1.105, 1.216] 0.147 0.024 [0.1, 0.195] 0.000

Controls (n=24) 0.414
Pay for Performance 1.006 [1.003, 1.009] 0.006 0.002 [0.003, 0.009] 0.000
Suggested Alternatives 2.074 [1.302, 3.303] 0.729 0.237 [0.264, 1.195] 0.002
Accountable Justifications 0.924 [0.512, 1.67] -0.079 0.302 [-0.67, 0.513] 0.794
Peer Comparison 1.563 [1.01, 2.421] 0.447 0.223 [0.01, 0.884] 0.045
Additional Time 1.085 [0.99, 1.191] 0.082 0.047 [-0.01, 0.174] 0.082

Suggested Alternatives (SA) (n=135) <0.001
Pay for Performance 1.007 [1.006, 1.009] 0.007 0.001 [0.006, 0.009] 0.000
Suggested Alternatives 2.199 [1.626, 2.973] 0.788 0.154 [0.486, 1.09] 0.000
Accountable Justifications 1.009 [0.719, 1.418] 0.009 0.173 [-0.33, 0.349] 0.957
Peer Comparison 1.552 [1.173, 2.054] 0.440 0.143 [0.159, 0.72] 0.002
Additional Time 1.192 [1.117, 1.274] 0.176 0.034 [0.11, 0.242] 0.000

Accountable Justifications (JA) (n=121) <0.001
Pay for Performance 1.006 [1.005, 1.008] 0.006 0.001 [0.005, 0.008] 0.000
Suggested Alternatives 1.955 [1.504, 2.54] 0.670 0.134 [0.408, 0.932] 0.000
Accountable Justifications 1.451 [1.061, 1.984] 0.372 0.160 [0.059, 0.685] 0.020
Peer Comparison 1.429 [1.097, 1.861] 0.357 0.135 [0.093, 0.621] 0.008
Additional Time 1.174 [1.095, 1.259] 0.160 0.036 [0.09, 0.23] 0.000

Peer Comparison (PC) (n=101) <0.001
Pay for Performance 1.007 [1.005, 1.009] 0.007 0.001 [0.005, 0.009] 0.000
Suggested Alternatives 2.589 [1.819, 3.685] 0.951 0.180 [0.598, 1.304] 0.000
Accountable Justifications 1.079 [0.684, 1.703] 0.076 0.233 [-0.38, 0.532] 0.744
Peer Comparison 1.285 [0.901, 1.832] 0.250 0.181 [-0.105, 0.605] 0.167
Additional Time 1.200 [1.103, 1.305] 0.182 0.043 [0.098, 0.266] 0.000
Results indicate the odds ratio of each alternative relative to the control group, broken down by exposure group (full
sample, controls, SA, JA, and PC).
*p-value for significance across all co-efficients for that sample
Chi-square statistic indicates that coefficients were significantly different from one another (p < 0.001).
Overall, prescribers tended to prefer any of the presented alternatives to prescriber education and
guideline review, regardless of exposure group. Prescribers did not prefer Peer Comparison as much as
other interventions for reducing inappropriate antibiotic prescriptions. Instead, prescribers consistently
15

preferred Suggested Alternatives regardless of which intervention they were actually exposed to,
including those in the control group. These results match those found in the pilot study; those who were
not exposed to any of the BEARI interventions also strongly preferred Suggested Alternatives (OR =
2.75, p = 0.003). Results were similar in the pilot group exposed to the interventions, though slightly less
pronounced (OR = 2.07, p = 0.018) (eTable 1). These trends remained consistent whether prescribers
were exposed to just one, two, or all of the interventions, although the magnitude and significance of
preference varied compared to the results by individual intervention (eTable 2).
In contrast, Peer Comparison was not strongly preferred as an intervention most likely to reduce
ARI antibiotic prescribing, though still statistically significant among controls (OR = 1.56, p = 0.045),
and those exposed to Suggested Alternatives (OR = 1.55, p = 0.002) and JA (OR = 1.43, p = 0.008). In
fact, those who were exposed to Peer Comparison had a non-significant preference for this intervention
(OR = 1.285, p = 0.167), even though it was highly effective in reducing inappropriate antibiotic
prescribing rates in the BEARI experiment. Overall, prescribers were either indifferent to, or did not
prefer Accountable Justifications (although this was non-significant), unless they were exposed to that
intervention in the study, in which case there was a significant preference for the intervention (OR = 1.45,
p = 0.02). In contrast, the pilot study showed that JA was non-preferred among non-exposed prescribers
(OR = 0.68, p = 0.049), though preferred among those exposed (OR = 1.79, p = 0.247).
Pay-for-performance was marginally preferred among prescribers across all groups (control and
treatment) with an OR = 1.007, implying that each additional $100 in financial incentives would increase
the probability of preferring pay-for-performance by 70%. This suggests that clinicians feel that every
$100 increase in compensation would improve their inappropriate antibiotic prescribing rates by 70%
relative to the control group. These results also generally held true in the pilot study. Additional office
time was more strongly preferred than pay-for-performance, although this preference was non-significant
in the control group (OR = 1.085, p = 0.082). Among those exposed to SA, JA, and PC, additional office
time was highly significantly preferred, with OR = 1.19, 1.17, and 1.2, respectively (p < 0.05). These
16

results imply that overall, clinicians felt that additional time spent with a patient to form an appropriate
diagnosis and treatment plan would be more effective than financial incentives to reduce inappropriate
antibiotic prescribing.
When adjusted for intervention in the interacted model, trends in stated preferences remained
stable (eTable 3). Those exposed to Accountable Justifications did not favor Suggested Alternatives as
strongly as those exposed to Suggested Alternatives or Peer Comparison (OR = 1.889, p < 0.001, vs. OR
= 2.085, p < 0.001 and OR = 2.406, p < 0.001, respectively), as in the base case model. In addition,
Accountable Justifications remained non-preferred even when stratified by exposure group, except for
those exposed to JA in the BEARI trial.
Willingness-to-pay estimates indicate the value of each intervention to prescribers (Table 4).
Table 4: Willingness-To-Pay Estimates
Willingness to Pay Monthly Annually

Willingness to Pay Monthly Annually
Controls

Accountable Justifications (JA)
Suggested Alternatives -$120.52 -$1,446.28

Suggested Alternatives -$109.21 -$1,310.46
Accountable Justifications $12.98 $155.81

Accountable Justifications -$60.65 -$727.81
Peer Comparison -$73.84 -$886.09

Peer Comparison -$58.16 -$697.90
Additional Time -$13.55 -$162.61

Additional Time -$26.10 -$313.23

Suggested Alternatives (SA)

Peer Comparison (PC)
Suggested Alternatives -$108.21 -$1,298.56

Suggested Alternatives -$129.43 -$1,553.13
Accountable Justifications -$1.29 -$15.48

Accountable Justifications -$10.34 -$124.06
Peer Comparison -$60.38 -$724.56

Peer Comparison -$34.07 -$408.80
Additional Time -$24.17 -$290.04

Additional Time -$24.79 -$297.43
*Negative values indicate the value of that particular alternative to the prescriber, thus the absolute value is the
amount the prescriber is willing to accept; a positive number implies that the prescriber would be willing to give
up that dollar amount rather than use the alternative presented
Suggested Alternatives was equivalent to $1299-$1553 in financial incentives per year, whereas Peer
Comparison was only worth $408-$886 per year. Each minute of additional office time for patient
education was only worth an average $265 per year. These estimates imply that prescribers could be paid
an average of $1400, $650, or $265 per year to achieve the same results as Suggested Alternatives, Peer
Comparison, and additional time, respectively. These results also imply that Suggested Alternatives is
17

equivalent to an additional five to six minutes of office time, and Peer Comparison three minutes. In
contrast, prescribers in the control group would rather sacrifice $13 of pay per month (or $156 annually)
than be required to enter a justification for every inappropriate antibiotic prescription (Accountable
Justifications).
Compared to the results of the DCE, the exit survey yielded an overwhelmingly ambivalent
attitude towards the BEARI study as a whole, and towards each individual intervention. While 60% of
prescribers agreed that providing feedback on clinician performance and using electronic decision support
tools are effective ways to improve quality care, 42% of prescribers felt that the BEARI interventions
were neither useful nor useless in improving antibiotic prescribing practices. When asked about the
usefulness of the Peer Comparison emails, 37% of prescribers responded neutrally, while 14% felt the
information was not at all useful, and only 7% found it very useful. For Suggested Alternatives and
Accountable Justification, 30% of prescribers responded neutrally, ~20% felt the interventions were not at
all useful, and only 10% found the interventions to be very useful.
Discussion
Overall, the DCE-elicited preferences did not reflect actual behavior as revealed in the
randomized controlled trial of prescriber interventions to alter antibiotic prescribing. Although the DCE
overwhelmingly favored Suggested Alternatives as the most effective method for reducing inappropriate
antibiotic prescriptions across both treatment and controls, this intervention yielded similar antibiotic
prescribing rates to controls throughout the duration of the BEARI trial.[18] On the other hand, most
prescribers generally felt that Accountable Justifications was ineffective in reducing their ARI antibiotic
prescribing, despite its significant impact on inappropriate antibiotic prescription rates in the trial. Peer
Comparison was generally significantly preferred to provider education, although this preference was not
as strong as that for Suggested Alternatives. Regardless of the intervention group to which subjects were
assigned in the trial, the general trends in stated preference were similar across all groups in the DCE
survey, while the trial showed that the trajectory of antibiotic prescribing rates for Accountable
18

Justification and Peer Comparison relative to the control group consistently decreased compared to
Suggested Alternatives.
Differences between stated and revealed preferences may be due to a number of factors.
Prescribers did not get to choose their intervention in the randomized trial; therefore, prescribing behavior
may not represent the prescriber’s true “revealed preference”. Another consideration is that in the DCE
task, prescribers may not be choosing the intervention that best reduces inappropriate antibiotic
prescribing; instead, the stated preference might be for the intervention that is the least inconvenient for
the prescriber. The DCE results are also reflected in the exit survey, which indicated that the majority of
prescribers rated the usefulness of each intervention as either 1, 2, or 3 on a scale of 1-5 (1 being not at all
useful, 5 being very useful). Only 10% of prescribers found Suggested Alternatives to be very useful
(rating of 5 on a scale of 1-5), while just 8% and 6% found Accountable Justifications and Peer
Comparison to be very useful interventions, respectively. Given the lack of enthusiasm in the survey
responses, it is unsurprising that even Suggested Alternatives, the most preferred intervention according
to the DCE, was only worth up to $1500 annually to clinicians, and the other interventions even less.
Suggested Alternatives may be the most preferred intervention overall because it is a clinical
decision support tool, as opposed to a socially motivated intervention such as JA or PC. The stated
preference for SA is consistent with design recommendations that if EHR alerts are used, they should
include actionable information.[33, 34]. On the other hand, socially motivated interventions force the
prescriber to conform to “social norms” due to a concern about his/her social reputation and/or an
awareness of the social norm of cooperation.[35] Rather than applying social pressure, Suggested
Alternatives bears no social consequences if the prescriber chooses to ignore the alternatives presented.
In contrast, prescribers’ dislike of Accountable Justifications is consistent with findings describing the
phenomenon of “alert fatigue”, and relatively poor efficacy of alerts that do not include actionable
recommendations.[34, 36] However, JA in BEARI was actually highly effective in the randomized
controlled trial, and differs from similar studies that do not include peer accountability[34].
19

One limitation of the BEARI study is that there were many incomplete responses to the DCE
task. However results for the entire sample were compared to results for those who completed six or
more choice tasks, and there were no significant differences. In addition, geographic variation in
prescribing habits and patient attitudes may have affected the rate at which inappropriate antibiotics were
prescribed, regardless of the presence of a behavioral economic intervention. However, it is assumed that
such attitudes and behaviors would have been balanced among all treatment groups after randomization.
Responses may have been different had the DCE been administered before the RCT took place, since
prescribers would not have actually experienced the interventions prior to being asked about them. The
pilot DCE was administered to prescribers who did not receive any of the interventions, yet it showed
similar statistically significant stated preferences to those elicited from prescribers who participated in the
trial. This, along with the fact that actual intervention assignment did not affect the overall trend of stated
preferences suggests that prescribers appear to express the same preferences regardless of exposure to the
BEARI interventions. Finally, contrary to consensus regarding the inclusion of an opt-out option in DCE
to reflect a respondent’s choice in real life, no opt-out option was included.[37, 38] However, one study
empirically evaluated the effect of an opt-out option on attribute preference, and found that while attribute
value estimates differed, there were no notable differences in the relative order of the attributes (as
compared to each other).[37] Based on their findings, the authors recommended that an opt-out option
always be included in a DCE if non-participation is an option in real life as well. In BEARI, prescribers
did not have the option to opt out of participation, and it is likely that if the BEARI interventions were
implemented in practice, prescribers would also be unable to opt out.
Compared to previous studies attempting to evaluate the external validity of DCEs in healthcare,
this study provides stronger evidence that stated and revealed preferences can be non-concordant.
Specifically, revealed preferences were determined through a randomized controlled trial, and the same
individuals were then asked to complete the DCE. Participants were not given a choice as to which
intervention they wished to be implemented in clinic; instead, real-life behavior was captured in the trial.
20

Prescribers were also from various types of practice settings across the country, increasing the external
validity of the results, rather than eliciting responses from a single practice site. Previous studies did not
utilize an experimental design to determine revealed preferences, and thus the results are not as robust.
Ultimately, had the BEARI Trial relied upon the DCE preferences of the clinicians unexposed to
the interventions to inform intervention selection, the least effective intervention (SA) would have been
adopted and one of the effective interventions (JA or PC) would have been rejected. Relying on the exit
survey responses would have led to none of the interventions being adopted. Our findings are consistent
with recommendations that quality improvement interventions combining local participatory approaches
with expert driven approaches are likely to be most effective.[39] In particular, approaches that engage
targets in the implementation strategy rather than the design and development of interventions may
optimally ensure that interventions are deployed in a pragmatic way without being influenced by the
stated preferences of clinicians.
Conclusion
This study is part of a growing body of literature comparing stated and revealed preferences in a
healthcare setting. Consistent with similar studies evaluating the external validity of DCE-stated
preferences, this study showed that they are not concordant, and suggests that quality improvement
interventions should not rely on front-line staff input alone. Future work should continue to test the
external validity of DCE preferences in healthcare and the factors that contribute to differences between
stated and revealed preferences.
21

Figure 1: Discrete choice experiment treatment scenarios

Consider each of the following Choice pairs. For each separate Choice Pair indicate which
treatment option is most likely to impact your ARI antibiotic prescribing. You may only indicate
one choice per pair.
CHOICE A CHOICE B
EHR Alternative
Prescribing Screen
ON ON
Required Justification
Note
OFF OFF
Peer Performance
Feedback
ON OFF
Pay for Performance $100/month $200/month
Additional ARI Therapy
Explanation Time
5 minutes
per visit
0 minutes
per visit

MY CHOICE X

We will now give you 10 choice pairs and ask you to indicate your choice preference for
each pair sequentially. Please pick the one of the two alternatives that you think is better.

22

Appendix

DCEs have a theoretical basis rooted in random utility theory (RUT), which proposes that an individual’s
“utility”, U ni of a choice alternative can be broken down into two components: a systematic, explainable
component V ni that an individual n associates with alternative i, and a random error component ε in for
individual n associated with alternative i, i.e. 𝑈 𝑛𝑖
= 𝑉 𝑛𝑖
+ 𝜀 𝑛𝑖
.[23, 35] In this study, the utility estimation
for a given alternative is given by the following equation;
𝑈 𝑛𝑖
= 𝑉 𝑛𝑖
+ 𝜀 𝑛𝑖
= 𝛽 1
𝑝𝑎𝑦 𝑓𝑜𝑟 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 + 𝛽 2
𝑠𝑢𝑔𝑔𝑒𝑠𝑡𝑒𝑑 𝑎𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑣𝑒𝑠 + 𝛽 3
𝑎𝑐𝑐𝑜𝑢𝑛𝑡𝑎𝑏𝑙𝑒 𝑗𝑢𝑠𝑡𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛 + 𝛽 4
𝑝𝑒𝑒𝑟 𝑐𝑜𝑚𝑝𝑎𝑟𝑖𝑠𝑜𝑛 + 𝛽 5
𝑎𝑑𝑑𝑖𝑡𝑖𝑜𝑛𝑎𝑙 𝑡𝑖𝑚𝑒 + 𝜀 𝑛𝑖

where U is the utility derived from choosing a specific treatment, V is the observed utility derived from
the 𝛽 1−5
parameter estimates, and ε represents the unobservable error term.
The multinomial logit is traditionally used for analysis of DCE data but is limited by its assumption of
independence of irrelevant alternatives (IIA), which implies that the unobserved portion of utility for one
alternative has a fixed relationship to the unobserved portion of utility for another alternative.[36] While
IIA may hold true for some situations, it is a fairly restrictive assumption and thus the DCE data were
analyzed using mixed logit to verify the robustness of findings.
Mixed logit models allow for individual-specific attribute coefficients, accounting for preference
heterogeneity among different decision makers and allowing flexible substitution patterns, which more
realistically represents a choice situation the decision-maker faces.[36] In addition, the mixed logit model
adjusts the standard errors of the utility estimates for each individual to account for repeated choices made
by the same individual.

eTable 1: Pilot Study Mixed Logit Regression Results
Pilot Study OR [95% CI] Coefficient Std. Err. [95% CI] p-value z-score
Unexposed to BEARI
Pay for Performance 1.004 [1.002, 1.006] 0.004 0.001 [0.002, 0.006] 0.000 3.900
Suggested Alternatives 2.754 [1.424, 5.327] 1.013 0.337 [0.354, 1.673] 0.003 3.010
Accountable Justifications 0.684 [0.469, 0.998] -0.379 0.193 [-0.756, -0.002] 0.049 -1.970
Peer Comparison 0.995 [0.693, 1.428] -0.005 0.184 [-0.367, 0.356] 0.976 -0.030
Additional Time 1.175 [1.079, 1.279] 0.161 0.043 [0.076, 0.246] 0.000 3.720

Exposed to BEARI
Pay for Performance 1.008 [1.004, 1.011] 0.008 0.002 [0.004, 0.011] 0.000 4.000
Suggested Alternatives 2.072 [1.131, 3.796] 0.729 0.309 [0.123, 1.334] 0.018 2.360
Accountable Justifications 1.794 [0.667, 4.828] 0.585 0.505 [-0.405, 1.574] 0.247 1.160
Peer Comparison 1.641 [0.798, 3.372] 0.495 0.368 [-0.225, 1.216] 0.178 1.350
Additional Time 1.103 [0.963, 1.264] 0.098 0.693 [-0.037, 0.234] 0.156 1.420

23

eTable 2: Mixed Logit Regression Results by BEARI Exposure Group
OR [95% CI] Coefficient Std. Err. [95% CI] p-value p-value*
Exposure Group: SA, JA, PC (n=26) 0.004
Pay for Performance 1.008 [1.004, 1.012] 0.008 0.002 [0.004, 0.012] 0.000
Suggested Alternatives 1.760 [0.951, 3.258] 0.565 0.314 [-0.05, 1.181] 0.072
Accountable Justifications 1.713 [0.74, 3.968] 0.538 0.428 [-0.301, 1.378] 0.209
Peer Comparison 1.138 [0.618, 2.096] 0.129 0.312 [-0.481, 0.74] 0.678
Additional Time 1.219 [1.079, 1.378] 0.198 0.062 [0.076, 0.321] 0.001

Exposure Group: SA, JA (n=35) 0.004
Pay for Performance 1.006 [1.003, 1.009] 0.006 0.001 [0.003, 0.009] 0.000
Suggested Alternatives 2.015 [1.197, 3.392] 0.701 0.266 [0.18, 1.221] 0.008
Accountable Justifications 1.217 [0.781, 1.898] 0.197 0.227 [-0.247, 0.641] 0.385
Peer Comparison 1.921 [1.137, 3.246] 0.653 0.268 [0.128, 1.178] 0.015
Additional Time 1.223 [1.075, 1.393] 0.202 0.066 [0.072, 0.331] 0.002

Exposure Group: SA, PC (n=32) <0.001
Pay for Performance 1.008 [1.004, 1.011] 0.008 0.002 [0.004, 0.011] 0.000
Suggested Alternatives 5.520 [2.221, 13.716] 1.708 0.464 [0.798, 2.619] 0.000
Accountable Justifications 0.482 [0.198, 1.177] -0.729 0.455 [-1.622, 0.163] 0.109
Peer Comparison 1.706 [0.797, 3.655] 0.534 0.389 [-0.227, 1.296] 0.169
Additional Time 1.167 [0.989, 1.377] 0.154 0.084 [-0.011, 0.32] 0.068

Exposure Group: JA, PC (n=23) <0.001
Pay for Performance 1.006 [1.003, 1.009] 0.006 0.002 [0.003, 0.009] 0.000
Suggested Alternatives 2.153 [1.209, 3.834] 0.767 0.294 [0.19, 1.344] 0.009
Accountable Justifications 2.187 [1.041, 4.595] 0.782 0.379 [0.04, 1.525] 0.039
Peer Comparison 1.257 [0.753, 2.098] 0.229 0.261 [-0.284, 0.741] 0.382
Additional Time 1.221 [1.003, 1.488] 0.200 0.101 [0.003, 0.398] 0.047
SA: Suggested Alternatives, JA: Accountable Justification, PC: Peer Comparison, AT: Additional Time
Note: Prescribers were randomized to one of four main exposure groups consisting of more than one intervention
during the BEARI trial.
*p-value for significance across all coefficients for that sample

24

eTable 3: Mixed Logit Regression Results by Intervention and with Interaction Terms
Program Choice OR [95% CI] Coefficient Std. Err. [95% CI] p-value z-score
Suggested Alternatives
Pay for Performance 1.004 [1.004, 1.005] 0.004 0.000 [0.004, 0.005] 0.000 11.580
Suggested Alternatives 1.678 [1.462, 1.926] 0.518 0.070 [0.379, 0.656] 0.000 7.350

SA Interacted Model
Pay for Performance 1.005 [1.004, 1.006] 0.005 0.000 [0.004, 0.006] 0.000 11.900
Suggested Alternatives 2.085 [1.584, 2.745] 0.735 0.140 [0.46, 1.01] 0.000 5.240
Accountable Justifications 1.047 [1.303, 1.427] 0.046 0.158 [0.265, 0.356] -0.773 0.290
Peer Comparison 1.510 [1.168, 1.951] 0.412 0.131 [0.155, 0.668] 0.002 3.150
Additional Time 1.177 [1.109, 1.249] 0.163 0.030 [0.103, 0.222] 0.000 5.360

Accountable Justifications
Pay for Performance 1.005 [1.004, 1.006] 0.005 0.000 [0.004, 0.006] 0.000 11.820
Accountable Justifications 1.077 [0.895, 1.295] 0.074 0.094 [-0.111, 0.258] 0.434 0.780

JA Interacted Model
Pay for Performance 1.005 [1.004, 1.006] 0.005 0.000 [0.004, 0.006] 0.000 11.660
Suggested Alternatives 1.889 [1.474, 2.42] 0.636 0.126 [0.388, 0.884] 0.000 5.030
Accountable Justifications 1.428 [1.058, 1.928] 0.356 0.153 [0.057, 0.656] 0.020 2.330
Peer Comparison 1.420 [1.102, 1.83] 0.351 0.129 [0.097, 0.604] 0.007 2.710
Additional Time 1.164 [1.09, 1.244] 0.152 0.034 [0.086, 0.218] 0.000 4.500

Peer Comparison
Pay for Performance 1.004 [1.004, 1.005] 0.004 0.000 [0.004, 0.005] 0.000 11.460
Peer Comparison 1.294 [1.128, 1.485] 0.258 0.070 [0.121, 0.395] 0.000 3.680

PC Interacted Model
Pay for Performance 1.005 [1.004, 1.006] 0.005 0.000 [0.004, 0.006] 0.000 11.910
Suggested Alternatives 2.406 [1.754, 3.302] 0.878 0.161 [0.562, 1.195] 0.000 5.440
Accountable Justifications 1.061 [1.429, 1.609] 0.059 0.212 [0.357, 0.475] -0.780 0.280
Peer Comparison 1.268 [0.918, 1.75] 0.237 0.165 [-0.085, 0.56] 0.150 1.440
Additional Time 1.173 [1.09, 1.264] 0.160 0.038 [0.086, 0.234] 0.000 4.240

Additional Time
Pay for Performance 1.005 [1.004, 1.006] 0.005 0.000 [0.004, 0.006] 0.000 11.900
Additional Time 1.117 [1.079, 1.157] 0.111 0.018 [0.076, 0.146] 0.000 6.230
SA: Suggested Alternatives, JA: Accountable Justification, PC: Peer Comparison, AT: Additional Time
Note: Regression results are shown for each treatment group exposed to that same intervention, e.g. SA-exposed prescribers’
preference for Suggested Alternatives followed by model results for SA interacted with each intervention. There is not an
interacted model for additional time since this was not an intervention in the BEARI trial.
25

References
1. Shapiro, D.J., et al., Antibiotic prescribing for adults in ambulatory care in the USA, 2007-09. J
Antimicrob Chemother, 2014. 69(1): p. 234-40.
2. Steinman, M., et al., Changing use of antibiotics in community-based outpatient practice, 1991-
1999. Ann Intern Med, 2003. 138: p. 525-33.
3. Grijalva, C.G., J.P. Nuorti, and M.R. Griffin, Antibiotic prescription rates for acute respiratory
tract infections in US ambulatory settings. JAMA, 2009. 302(7): p. 758-66.
4. Ranji, S.S., et al., Antibiotic Prescribing Behavior Vol. 4: Closing the Quality Gap: A Critical
Analysis of Quality Improvement Strategies. Technical Review 9 (Prepared by Stanford
University-UCSF Evidence-based Practice Center under Contract No. 290-02-0017. 2006,
Agency for Healthcare Research and Quality: Rockville, MD.
5. van Bokhoven, M.A., G. Kok, and T. van der Weijden, Designing a quality improvement
intervention: a systematic approach. Qual Saf Health Care, 2003. 12(3): p. 215-20.
6. Demakis, J., et al., Quality Enhancement Research Initiative (QUERI): A Collaboration Between
Research and Clinical Practice. Medical Care, 2000. 38(6 Suppl 1): p. I17.
7. Rubenstein, L.V., et al., From understanding health care provider behavior to improving health
care: the QUERI framework for quality improvement. Quality Enhancement Research Initiative.
Med Care, 2000. 38(6 Suppl 1): p. I129-41.
8. Robinson, J.C., Theory and practice in the design of physician payment incentives. Milbank Q,
2001. 79(2): p. 149-77, III.
9. Scott, A., Eliciting GPs' preferences for pecuniary and non-pecuniary job characteristics. J
Health Econ, 2001. 20(3): p. 329-47.
10. Armour, B.S., et al., The effect of explicit financial incentives on physician behavior. Arch Intern
Med, 2001. 161(10): p. 1261-6.
11. Town, R., et al., Economic incentives and physicians' delivery of preventive care: a systematic
review. Am J Prev Med, 2005. 28(2): p. 234-40.
12. Loewenstein, G., T. Brennan, and K. Volpp, Asymmetric paternalism to improve health
behaviors. JAMA, 2007. 298: p. 2415-7.
13. Frolich, A., et al., A behavioral model of clinician responses to incentinves to improve quality.
Health Policy, 2007. 80: p. 179-93.
14. Thaler, R.H. and C.R. Sunstein, Nudge: Improving Decisions About Health, Wealth, and
Happiness. 2009, New York: Penguin Books.
15. Who We Are. [cited 2016 May 19]; Available from: http://www.behaviouralinsights.co.uk/about-
us/.
16. Hallsworth, M., et al., Provision of social norm feedback to high prescribers of antibiotics in
general practice: a pragmatic national randomised controlled trial. The Lancet, 2016.
387(10029): p. 1743-1752.
17. Persell, S., et al., Use of behavioral economics and social psychology to improve treatment of
acute respiratory infections (BEARI): rationale and design of a cluster randomized controlled
trial - study protcol and baseline practice and provider characteristics. BMC Infectious Diseases,
2013. 13: p. 290-300.
18. Meeker, D., et al., Effect of Behavioral Interventions on Inappropriate Antibiotic Prescribing
Among Primary Care Practices. Jama, 2016. 315(6): p. 562-70.
19. Ryan, M. and K. Gerard, Using discrete choice experiments to value healt hcare programmes:
current practice and future research reflections. App Health Econ and Health Pol, 2003. 2: p. 55-
64.
20. Whitty, J.A., et al., A systematic review of stated preference studies reporting public preferences
for healthcare priority setting. Patient, 2014. 7(4): p. 365-86.
21. De Bekker-Grob, E., M. Ryan, and K. Gerard, Discrete choice experiments in health economics:
a review of the literature. Health Econ, 2012. 21: p. 145-72.
26

22. Clark, M.D., et al., Discrete choice experiments in health economics: a review of the literature.
Pharmacoeconomics, 2014. 32(9): p. 883-902.
23. Krucien, N., A. Gafni, and N. Pelletier-Fleury, Empirical Testing of the External Validity of a
Discrete Choice Experiment to Determine Preferred Treatment Option: The Case of Sleep Apnea.
Health Econ, 2015. 24(8): p. 951-65.
24. Linley, W. and D. Hughes, Decision-Makers’ Preferences for Approving New Medicines in
Wales: A Discrete-Choice Experiment with Assessment of External Validity. Pharmacoeconomics,
2013. 31(4): p. 345-55.
25. Ryan, M. and V. Watson, Comparing welfare estimates from payment card contingent valuation
and discrete choice experiments. Health Econ, 2009. 18(4): p. 389-401.
26. Lambooij, M.H., IA; Veldwijk, J; de Melker, H; Mollema, L; van Weert, YWM; de Wit, GA,
Consistency between stated and revealed preferences: a discrete choice experiment and a
behavioural experiment on vaccination behaviour compared. BMC Med Res Methodol, 2015.
15(19).
27. Mark, T.L. and J. Swait, Using stated preference modeling to forecast the effect of medication
attributes on prescriptions of alcoholism medications. Value Health, 2003. 6(4): p. 474-82.
28. Arrow, K.S., R; Portney, PR; Leamer, EE; Radner, R; Schuman, H, Report of the NOAA Panel on
Contingent Valuation. 1993.
29. Louviere, J., T. Flynn, and R. Carson, Discrete Choice Experiments Are Not Conjoint Analysis.
Journal of Choice Modelling, 2010. 3: p. 57-72.
30. Louviere, J., D. Hensher, and J. Swait, Stated Choice Methods Analysis and Applications. 2000,
Cambridge, UK: Cambridge University Press.
31. Lancaster, K.J., A new approach to demand theory. J Polit Econ, 1966. 74: p. 132-57.
32. StataCorp, Stata Statistical Software: Release 12. 2011, StataCorp LP: College Station, TX.
33. Maviglia, S.M., et al., Automating complex guidelines for chronic disease: lessons learned. J Am
Med Inform Assoc, 2003. 10(2): p. 154-65.
34. Kawamoto, K., et al., Improving clinical practice using clinical decision support systems: a
systematic review of trials to identify features critical to success. BMJ, 2005. 330(7494): p. 765.
35. Kerr, N., Anonymity and Social Control in Social Dilemmas. Resolving Social Dilemmas:
Dynamics, Structural, and Intergroup Aspects, ed. M.S. Foddy, M; Schneider, S; Hogg, M. 1999,
Philadephia, PA: Psychology Press. 103-9.
36. Kesselheim, A.S., et al., Clinical decision support systems could be modified to reduce 'alert
fatigue' while still minimizing the risk of litigation. Health Aff (Millwood), 2011. 30(12): p. 2310-
7.
37. Veldwijk, J., et al., The effect of including an opt-out option in discrete choice experiments. PLoS
One, 2014. 9(11): p. e111805.
38. Lancsar, E. and J. Louviere, Conducting discrete choice experiments to inform healthcare
decision making: a user's guide. Pharmacoeconomics, 2008. 26(8): p. 661-77.
39. Parker, L.E., et al., Balancing health care evidence and art to meet clinical needs: policymakers'
perspectives. J Eval Clin Pract, 2009. 15(6): p. 970-5.

27

Chapter 3: The use of behavioral economics and social psychology to improve treatment of acute
respiratory infections (BEARI): a cost-effectiveness analysis

Abstract
Background: Behavioral economics interventions have been shown to effectively reduce the rates of
inappropriate antibiotic prescriptions for acute respiratory infections (ARIs).
Objective: To determine the cost-effectiveness of three behavioral economic interventions designed to
reduce inappropriate antibiotic prescriptions for ARIs.

Design: 30-year Markov model from the US societal perspective with inputs derived from the literature
and CDC surveillance data.

Subjects: 45-year-old adults with signs and symptoms of ARI presenting to a healthcare provider.

Interventions: (1) Provider education on guidelines for the appropriate treatment of ARIs; (2) Suggested
Alternatives, which utilizes computerized clinical decision support to suggest non-antibiotic treatment
choices in lieu of antibiotics; (3) Accountable Justification, which mandates free-text justification into the
patient’s electronic health record when antibiotics are prescribed; and (4) Peer Comparison, which sends a
periodic email to prescribers about his/her rate of inappropriate antibiotic prescribing relative to clinician
colleagues.

Main Measures: Discounted costs, quality-adjusted life years (QALYs), and incremental cost-
effectiveness ratios.

Key Results: Each intervention has lower costs but higher QALYs compared to provider education.
Total costs for each intervention were $1,004, $519, $487, and $447, and total QALYs were 14.32, 14.39,
14.40, and 14.41 for the control, Suggested Alternatives, Accountable Justification, and Peer Comparison
groups respectively. Results were most sensitive to the quality-of-life of the uninfected state, and the
likelihood and costs for antibiotic-associated adverse events.

Conclusions: Behavioral economics interventions can be cost-effective strategies for reducing
inappropriate antibiotic prescriptions by reducing healthcare resource utilization.

Acknowledgements: CLG, KMZ, and JWH made substantial contributions to the conception and design of this project. CLG, KMZ, JWH, DM
and JND assisted with data acquisition, analysis and interpretation of data for this manuscript. CLG drafted the manuscript, and KMZ, JWH, DM
and JND revised it critically for important intellectual content. All authors approved of the final version to be published. CLG is the guarantor.
The original BEARI Trial (NCT01454947) was supported by the American Recovery & Reinvestment Act of 2009 (RC4 AG039115) from the
National Institutes of Health/National Institute on Aging and Agency for Healthcare Research and Quality (Principal Investigator: Dr Doctor,
University of Southern California). The project also benefited from technology funded by the Agency for Healthcare Research and Quality
through the American Recovery & Reinvestment Act of 2009 (R01 HS19913-01) (Dr Ohno-Machado, University of California, San Diego). Data
for the project were collected by the University of Southern California's Medical Information Network for Experimental Research (Med-INFER)
which participates in the Patient Scalable National Network for Effectiveness Research (pSCANNER) supported by the Patient-Centered
Outcomes Research Institute (PCORI), Contract CDRN-1306-04819 (Dr Ohno-Machado). The authors are independent from these funding
organizations, which played no role in design and conduct of the study; collection, management, analysis, and interpretation of the data;
preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.
28

Introduction
In the United States, it is estimated that more than 50% of outpatient-prescribed antibiotics are
inappropriate, predominantly among patients seeking treatment for acute respiratory infections (ARIs)
caused by viruses. Such unnecessary antibiotic use leads to increased risk of adverse events and
emergency department (ED) visits for such events and additional financial costs to the healthcare
system.(1-3) Furthermore, excess antibiotic use contributes to the ever-increasing problem of antibiotic
resistance.(4-6) The Centers for Disease Control and Prevention (CDC) notes that the single most
important action needed to slow the spread of antibiotic-resistant infections is to reduce the amount of
inappropriate and unnecessary antibiotic use in humans and animals.(7) A large body of work describes
various attempts to curb inappropriate antibiotic prescribing through traditional interventions such as
physician and patient education, electronic clinical decision support, and financial incentives. These have
only resulted in modest reductions in antibiotic prescribing rates for nonbacterial ARIs.(8, 9)
An alternative approach to changing prescribing behavior applies ideas from the behavioral
sciences, using social cues and subtle changes in the clinic environment to influence clinical decision
making.(10, 11) Efforts to change antibiotic prescribing through the use of behavioral insights have been
implemented recently in the U.S. and also in the U.K.(12, 13) In the US, a multi-site cluster randomized
clinical trial, the BEARI study, evaluated the effectiveness of behavioral interventions on the rates of
inappropriate antibiotic prescribing in primary care practices with existing electronic health records
systems in Boston and Los Angeles.(12) The interventions implemented were: (1) Suggested
Alternatives, which utilizes computerized clinical decision support to suggest non-antibiotic treatment
choices in lieu of antibiotics; (2) Accountable Justification, which prompts entry of free-text justification
that become part of the patient’s electronic health record when antibiotics are prescribed; and (3) Peer
Comparison, which sends a periodic email to prescribers about his/her rate of inappropriate antibiotic
prescribing relative to colleagues. The study found that all three interventions led to absolute reductions
of 16-18% in inappropriate antibiotic use over an eighteen-month period.
29

Previous cost-effectiveness analyses on reducing inappropriate outpatient antibiotic prescriptions
among outpatients have focused on the cost impact of using biomarker point-of-care tests (C-reactive
protein (CRP), procalcitonin) to identify patients with possible bacterial lower respiratory tract
infection.(14-16) These models have shown that such tests do not significantly increase costs nor impact
patients, while having a significant effect on reducing inappropriate antibiotic prescriptions, and are thus
cost-effective. No other cost-effectiveness models have assessed other interventions to reduce outpatient
antibiotic prescribing. To assess the tradeoffs between costs and benefits and inform public policy, we
conducted a cost-utility analysis from the US societal perspective to determine the BEARI interventions’
value in reducing inappropriate antibiotic prescriptions.
Methods
Model Background
Because the BEARI interventions focused on reducing inappropriate antibiotic use for
nonspecific upper respiratory tract infections commonly caused by viruses, we included the following
primary ARIs that could result in justifiable antibiotic use in adults: acute otitis media, sinusitis, and
pharyngitis.(17) Statistics and model parameters for these infections and resistance rates were based on
data for Streptococcus pneumoniae, as this is the most common causative pathogen for community-
acquired respiratory tract infections.(18) We assumed that an individual could be infected with either
susceptible bacterial strains, i.e. likely to clinically resolve with just one course of antibiotics, or resistant
ones that may require multiple courses of antibiotic treatment.
Model Structure
We constructed a Markov model to simulate utilization of antibiotics, cost of care, and health
outcomes for a 45-year-old adult presenting to a healthcare provider with signs and symptoms of ARI
potentially with complications, as this was the approximate average age of the population in the BEARI
trial, and for which age-specific data on inappropriate antibiotic prescribing rates were available (Figure
30

1). The model framework was identical for each treatment arm (control, Suggested Alternatives,
Accountable Justification, and Peer Comparison), with treatment-specific model inputs. We used this
model to estimate the cumulative costs, QALYS and ICERs of three interventions relative to the control
of no intervention over a 30-year period and from the US societal perspective. This time horizon was
used as the estimated duration for amortization of any costs associated with the initial implementation of
the interventions. Model computation was done in R version 3.3.1 using the markovchain package.(19)
The model was split into two major groups: those vaccinated against pneumococcal disease and
those who are not. An individual began as unvaccinated, and transitioned to the vaccinated group at age
65 and older at a rate based on the overall change in pneumococcal vaccination coverage per year.
Within each group, the individual could contract either a viral ARI, susceptible bacterial ARI (sinusitis,
otitis media, pharyngitis), or resistant bacterial ARI, due to the most common pathogens associated with
these diseases in adults including S. pyogenes (pharyngitis), S. pneumoniae, H. influenzae, and M.
catarrhalis (otitis media, sinusitis) and S. aureus in some cases of sinusitis.(20, 21) In all three clinical
conditions, individuals who received antibiotics were at risk for experiencing drug-associated adverse
reactions that either self-resolved or resulted in an emergency department visit and very rarely, death.
Hospitalization was assumed to occur only for severe bacterial ARIs regardless of susceptibility, with
complications such as mastoiditis and brain abscess (acute otitis media), orbital infection (sinusitis), or
rheumatic heart disease, tonsillar / retropharyngeal abscess, and glomerulonephritis (pharyngitis).(22, 23)
Probabilities
Annual transition probabilities for each state were derived from available literature and
information regarding resistance patterns from CDC (Table 1). The baseline probabilities of receiving an
antibiotic for undifferentiated viral infection, sinusitis, acute otitis media, and pharyngitis were derived
from a national study evaluating antibiotic prescriptions dispensed in the ambulatory setting, as well as
the true prevalence of bacterial URI.(24) Reductions in antibiotic prescribing were as reported in the
BEARI study.(12) Rates of hospitalization for ARI and complications related to these hospitalizations
31

were derived from the Agency for Healthcare Research and Quality Inpatient Stay data, as well as expert
opinion from infectious diseases clinicians.(25) Rates of respiratory antibiotic-associated adverse drug
reactions (ADRs) and anaphylaxis were derived from various studies evaluating emergency department
visits for antibiotic-associated ADRs.(1, 2, 26) Baseline rates of antibiotic resistance, as well as rate of
susceptible to resistant strain conversion, were based on data from the CDC’s Active Bacterial Core
Surveillance Report.(27) In addition, the rate of pneumococcal vaccination was based on CDC
Behavioral Risk Factor Surveillance System (BRFSS) Reports.(28) Finally, baseline age-adjusted
mortality based on actuarial life tables was incorporated to account for death from other causes in addition
to deaths resulting from hospitalization and/or complications of S. pneumoniae infection.(29)
Costs
Costs were in 2016 US dollars (USD) and derived from the literature and the Centers for
Medicare and Medicaid Services (CMS) reimbursement for outpatient encounters based on Common
Procedural Terminology (CPT) codes. Costs included intervention implementation costs, provider office
visit for respiratory infection, average cost of over-the-counter and symptomatic treatment for acute
respiratory infections, and average antibiotic costs. Hospitalization and complication costs were also
included for susceptible vs. resistant infections. Costs for the BEARI interventions were calculated based
on the Bureau of Labor Statistics compensation rate for physicians, and the approximate amount of time a
clinician would spend on an encounter if a BEARI alert or email was generated. Cost of an outpatient
encounter was based on the CMS Physician Fee Schedule Healthcare Common Procedure Coding System
(HCPCS) code for a minor self-limiting problem. Average over-the-counter, symptomatic treatments and
antibiotic costs for acute respiratory infections were based on literature estimates, as well as
hospitalization costs for acute respiratory infections and related complications.(2, 25)
All costs were adjusted to 2016 USD using the Medical Consumer Price Index (CPI) as shown in
Table 1. An annual discount rate of 3% was applied to all costs.
32

Quality of Life
Quality of life (QOL) utility weights were based on literature related to acute respiratory
infections. For non-infected individuals, a baseline utility value of 0.87 was used across all groups.(30)
Acute respiratory infections were assigned a utility of 0.684 (range 0.671-0.696).(31, 32) Treatment for
acute respiratory infections was assigned a utility of 0.814 (range 0.801-0.825) for over-the-counter and
symptomatic treatment, and 0.806 (0.698-0.884) for antibiotic prescription treatment.(31) Emergency
department visits were assigned a utility of 0.622 (range 0.37-0.94). Hospitalization utility values were
0.53 (range 0.43-0.63) for severe infections, and 0.3 (range 0.237-0.363) for complications resulting from
hospitalization. (33) Each utility value was then used to adjust for the time spent in that health state per
year to calculate an overall utility value for that health state using the following formula:
QOL health_state = utility health_state * (
𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 ℎ𝑒𝑎𝑙𝑡 ℎ 𝑠𝑡𝑎𝑡𝑒 𝑂𝑛𝑒 𝑦𝑒𝑎𝑟 (365 𝑑𝑎𝑦𝑠 )
) + utility non-infected * (1 –
𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 ℎ𝑒𝑎𝑙𝑡 ℎ 𝑠𝑡𝑎𝑡𝑒 𝑂𝑛𝑒 𝑦𝑒𝑎𝑟 (365 𝑑𝑎𝑦𝑠 )
)
This calculated utility is shown for each health state in Table 1. Quality-adjusted life years were
discounted by an annual rate of 3%.
Analyses
Outcomes measures for this analysis included QALYs and total costs. Treatment arms were
compared to the control group in terms of cost per QALY using ICERs. The ICER is the ratio of the
difference in costs to the difference in effectiveness between two alternatives:(34)
𝐼𝐶𝐸𝑅 =
𝐶𝑜𝑠𝑡𝑠 𝐼𝑛𝑡𝑒𝑟𝑣𝑒𝑛𝑡𝑖𝑜𝑛 − 𝐶𝑜𝑠𝑡𝑠 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑄𝐴𝐿𝑌𝑠 𝐼𝑛𝑡𝑒𝑟𝑣𝑒𝑛𝑡𝑖𝑜𝑛 − 𝐶𝑜𝑠𝑡𝑠 𝐶𝑜𝑛𝑡𝑟𝑜𝑙
The ICER allows different interventions to be compared across a standard metric. In this analysis, all
treatment arms were compared to the control group. An annual discount rate of 3% was applied to all
costs and QALYs.
33

One-way sensitivity analyses were conducted to test the effect of individual parameters on the
results of the model. Probabilities, costs, and utility values were varied per reported ranges published in
the literature. Incremental cost-effectiveness ratios were recalculated accordingly.
A net monetary benefit (NMB) analysis was conducted to assess the cost-effectiveness of each
therapy at varying WTP thresholds.(35) NMB is calculated as:
𝑁𝑀𝐵 = (𝑄𝐴𝐿𝑌𝑠 × 𝑊𝑇𝑃 ) − 𝐶𝑜𝑠𝑡
The NMB is determined at each willingness-to-pay threshold. The treatment with the highest
NMB at a given WTP is considered the most cost-effective at that WTP threshold. An intervention is
considered “dominated” if its NMB is always lower than another intervention.
Results
In the base case scenario, all three BEARI interventions dominated the control group, each
yielding more QALYs at a lower cost compared to the control. The QALYs yielded were 14.32 for the
control group, compared to 14.39, 14.40, and 14.41 QALYs for Suggested Alternatives, Accountable
Justification, and Peer Comparison respectively, while costs were $1,004 for the control group compared
to $519, $487, and $447 for each intervention respectively. The distribution of each intervention is
shown in the cost-effectiveness plane in Figure 2.
One-way sensitivity analyses revealed that the results remained robust to changes in model
parameters; the BEARI interventions continued to yield more QALYs at a lower cost compared to the
control group despite parameter changes, and sensitivity analysis results are thus shown as net monetary
benefits (Figure 3). Results were most sensitive to the utility of the uninfected health state, as most of the
time in the model is spent uninfected. Notably, results were sensitive to the probability of experiencing
an adverse drug reaction to antibiotics, and going to the emergency department for antibiotic-induced
adverse drug reactions, highlighting the significant impact the interventions have on reducing exposure to
34

adverse drug effects. Nonetheless, the BEARI interventions remained dominant over the control group.
In addition, results were not affected by changes in bacterial resistance patterns.
Net monetary benefit analysis showed that the BEARI interventions all yielded marginally higher
net monetary benefits compared to the control group at very low willingness-to-pay thresholds (Figure 4).
This effect remained consistent even at higher willingness-to-pay thresholds (not displayed on the graph
for clarity).
Discussion
All the BEARI interventions are cost-effective, yielding lower costs for more QALYs compared
to no intervention. Model results were most sensitive to the likelihood of ED visits and antibiotic-
associated adverse events, yet each intervention remained cost-effective even when these probabilities
were varied per reported ranges. This reduction in costs highlights the significant impact that reducing
ED visits and antibiotic-associated adverse events can have on reducing overall healthcare costs. This
suggests that reducing inappropriate antibiotic prescriptions can substantially affect healthcare resource
utilization beyond simply improving clinical practice. Overall patient care would benefit from any
intervention that minimizes the likelihood of substandard clinical practice, patient-level adverse drug
events, and hospital and individual costs to treat them. Furthermore, the cost effectiveness of these
BEARI interventions is consistent with evidence showing that behavioral economics can and should be
used to design effective policies and programs to improve health, education, and the economy.(36)
Our data extend the existing few data available on the cost-effectiveness of active antimicrobial
stewardship in the adult primary care healthcare setting for URI. Previous studies utilized models that did
not capture the impact of changes in antibiotic resistance resulting from the effects of an antibiotic
stewardship intervention(14-16), except for one model, which attempted to quantify the cost of resistance
associated with each antibiotic prescription dispensed. However, these calculations were rough estimates
that could not be verified for accuracy, and thus were not considered for use in our analysis. [18] Our
35

model is also one of few cost-effectiveness analyses in health information technology that includes a full
accounting of costs and outcomes of the intervention implemented; many previous studies evaluating
health information technology have only provided cost data without a full economic evaluation that
includes outcomes, and particularly, standardized outcomes (such as quality-adjusted life years).(37, 38)
In contrast, our model provides a full economic evaluation of the technology used to implement the
interventions.
There are some limitations to our analysis. We were forced to make assumptions for which few or
only non-robust data are available. Nonetheless, the variables identified as the major cost drivers were
based on credible national datasets (ED visits and ADR information). We also did not include children in
this analysis as the referent BEARI trial included only those >18 years of age. Children, however,
represent a large proportion of inappropriate antibiotic utilization in the United States. Their inclusion in
our analysis may have revealed an even larger economic impact of the BEARI intervention. Although we
did not find that changes in drug resistance was an important driver of costs, our model was unable to
include the potential effects of antibiotic use in animals, which has been shown to contribute to resistance
patterns in humans.(39) Another limitation is the assumption that resistance rates stay constant over time.
Per CDC’s Active Bacterial Core (ABC) Surveillance Report, rates of bacterial resistance have fluctuated
widely over the past ten years, but with a net increase of 1.2% resistant isolates per year. One-way
sensitivity analyses allowed testing of resistance rates based on historical trends, and model results
remained robust. It is notable that improving prescribing behavior is highly successful in the control of
antimicrobial resistance in the hospital setting, but available data suggest that the impact on reversing
resistance in the community setting is unlikely or at best, likely to occur at a very slow rate.(40, 41) Only
two studies have evaluated this issue and neither showed an impact on pneumococcal antimicrobial
resistance in the community.(42, 43) Regardless, appropriate antibiotic prescribing can reduce healthcare
costs by up to 20-30%.(9)
36

A practical concern is that the BEARI interventions will not have persistent effects on decreasing
antibiotic prescribing rates, or that they will be less effective in a non-experimental setting. In our
analysis, the rates of antibiotic prescribing were varied based on confidence intervals reported in the
randomized controlled trial to verify if improved rates of antibiotic prescribing among the control group
would affect the cost-effectiveness of the other interventions. Even so, the interventions continue to
remain cost-effective primarily because of the significant impact that even a small reduction in antibiotic
prescribing would have on adverse drug events and associated emergency department visits.
A major benefit of the BEARI interventions is the ease of implementation and the lack of a need
for point-of-care blood testing (e.g. procalcitonin, CRP) in patients who usually have an uncomplicated
ARI at presentation. While an underlying assumption for this model is that an existing electronic health
record is already in place, thus making the implementation of the interventions inexpensive with little
overhead cost, 87% of all office-based physicians and 96% of non-federal hospitals in the United States
had an electronic health record system in 2015.(44) The high adoption rate of electronic health record
systems allows any electronic interventions to be implemented with due diligence and efficiency without
the need to install a completely new system simultaneously. The few remaining organizations lacking
electronic health records are likely to eventually convert as the benefits may outweigh the upfront costs.
(45, 46)
Conclusion
In this cost-effectiveness analysis, the BEARI interventions were all shown to be cost-effective
relative to the control group, assuming an existing electronic health record is in place. We believe our
data are robust and reveal the cost-effectiveness of each BEARI intervention; a complement to prior work
noting its potential to facilitate improved patient care for those with ARIs, minimize adverse outcomes
associated with inappropriate antibiotic use, and potentially mitigate against the development of drug
resistance.
37

Table 1: Model Inputs
Key Transition Probabilities Base Case (Range) Reference
Probability of Inappropriate Antibiotics, age 20-64
*
Control 0.430 (0.367-0.495) (24)
Suggested Alternatives 0.119 (0.101-0.137) (12)
Accountable Justification 0.096 (0.082-0.111) (12)
Peer Comparison 0.080 (0.068-0.092 (12)
Probability of Inappropriate Antibiotics, age >65
*
Control 0.394 (0.272-0.531) (24)
Suggested Alternatives 0.109 (0.075-0.147) (12)
Accountable Justification 0.088 (0.061-0.119) (12)
Peer Comparison 0.073 (0.051-0.099) (12)
Prevalence of true bacterial infections, age 20-64 0.045 (0.029-0.051) (24)
Prevalence of true bacterial infections, age >65 0.063 (0.051-0.75) (24)
Baseline population resistance 0.163 (0-0.313) (27)
Conversion of susceptible  resistant strain 0.013 (0-0.143) (27)
Likelihood of antibiotic adverse drug reaction (ADR) 0.451 (0.151-0.751) (1, 2)
Likelihood of ADR requiring ED visit 0.306 (0.103-0.510) (1, 2)
Likelihood of death due to anaphylaxis 0.003 (0-0.0084) (26)
Likelihood of hospitalization for URI 0.004 (0.002-0.005) (25)
Likelihood of complications 0.010 (0-0.020) expert opinion; (47-49)
Likelihood of pneumococcal vaccination
†
0.033 (0-0.065) (28)
Costs
Base Case Cost
(Range, 2016 US Dollars)
Reference
Implementation

Suggested Alternatives 1.91 (0-5.73)
Expert Opinion
Bureau of Labor Statistics
Accountable Justification 3.82 (0-9.55)
Peer Comparison 0.95 (0-3.82)
MD Visit (HCPCS 99212) 35.06 (28.84-44.45) CMS Physician Fee Schedule
Antibiotics (Susceptible Infection) 8.65 (0.17-46.20) VA Federal Supply Schedule
Antibiotics (Resistant Infection) 11.11 (4.19-53.00) VA Federal Supply Schedule
OTC / Symptomatic Treatment 4.98 (0-10.31) VA Federal Supply Schedule
38

Complications
‡
17,313 (16,102-18,523) (25), (50)
Emergency Department Visit 4,088 (3,553-4,632) (2)
Health States Base Case Utility (Range) Reference
Non-infected (“healthy”) 0.8700 (0.8600-0.8800) (30)
Upper respiratory infection 0.8649 (0.8645-0.8652) (31, 32)
Antibiotic treatment
§
0.8682 (0.8653-0.8704) (31), (51)
OTC / symptomatic treatment 0.8685 (0.8653-0.8704) (31)
ED visit for infection 0.8693 (0.8686-0.8702) (31, 32)
Hospitalization for severe infection 0.8635 (0.8616-0.8654) (33)
Inpatient complications 0.8591 (0.8544-0.8603) (33)

*
Probability of inappropriate antibiotics for BEARI interventions derived based on a reduction in antibiotic
prescriptions relative to the rate of antibiotic prescribing reported in the study by Fleming-Dutra et al.
†
Represents the probability of getting a vaccination from year to year. BRFSS data only report the total percentage
of individuals who are vaccinated (vaccine coverage), not the percentage of new vaccinations each year.
‡
Includes mastoiditis, intracranial abscess, orbital cellulitis, peritonsillar abscess, retropharyngeal abscess,
glomerulonephritis, and Clostridium difficile.
§
Incorporates quality-of-life decrements for ADRs related to antibiotic treatment, such as C. difficile associated
diarrhea.

39

FIGURE 1: Markov Model Structure

Figure 1 depicts the Markov framework. “U_” and “V_” designate unvaccinated (UNVACC) and vaccinated
(VACC) individuals respectively, while “s” and “r” subscripts represent carriers of susceptible or resistant bacterial
strains. As individuals get vaccinated over time, they move from the UNVACC to the VACC pool. An individual
may contract one of three types of infections: viral acute respiratory infection (VARI), susceptible bacterial acute
respiratory infection (BARIs), or resistant bacterial acute respiratory infection (BARIr). For VARI, treatment is
either over-the-counter and symptomatic treatment (VOTC), or inappropriate antibiotics (VABX), which may lead
to an adverse drug reaction (VADR) and possible emergency department visit (VED) and/or anaphylactic death
(VDEATH). Otherwise, the infection will resolve (VRESOLVED) and patients return to the pool of
unvaccinated/vaccinated individuals. For BARI, all individuals should receive antibiotics (BABX), which may also
lead to subsequent adverse drug reaction (ADR) and emergency visit (BED). In addition, the infection may become
severe requiring inpatient hospitalization (BHOSP) and possible infectious and hospitalization complications
(BCOMP). Not shown is background mortality, which assumes individuals may exit the model at any state due to
death from natural causes.

40

FIGURE 2: Cost-Effectiveness Plane

The cost-effectiveness plane depicts the incremental costs and quality-adjusted life years of each intervention
relative to the control group. The further down the X and Y-axes the intervention is, the more cost-effective it is
relative to the control.

$(600.00)
$(500.00)
$(400.00)
$(300.00)
$(200.00)
$(100.00)
$-
$100.00
0.000 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090
Incremental Cost
Incremental Quality-Adjusted Life Years
Cost-Effectiveness Plane
Control
Accountable Justification
Peer Comparison
Suggested Alternatives
41

Figure 3: One-Way Sensitivity Analysis
*

*
One-way sensitivity analyses yielded similar trends for each intervention, with results most sensitive to the utility
of the uninfected health state, followed by the likelihood and costs associated with adverse events due to antibiotics.
Results have been transformed into net monetary benefits as even in one-way sensitivity analyses, the interventions
remained dominant over the control group, therefore yielding negative ICERs.
$128 $130 $132 $134 $136 $138 $140 $142 $144 $146 $148
Probability of Antibiotic Rx - Control
Utility of Health State - Complications
Probability of Death due to Complications
Utility of Health State - Hospitalization
Cost of Complications
Utility of Health State - Emergency Department Visit
Probability of Complications
Probability of Baseline Resistance
Probability of Vaccine Efficacy
Probability of Resistance Conversion
Cost of Vaccination
Utility of Health State - Adverse Drug Reaction
Probability of Vaccination
Cost of Hospitalization
Utility of Health State - Antibiotic Treatment
Probability of Antibiotic Anaphylaxis
Probability of Hospitalization for URI
Utility of Health State - Non-Antibiotic Treatment
Cost of BEARI Interventions
Utility of Health State - Viral ARI
Cost of Antibiotics
Probability of Bacterial URI
Cost of Symptomatic Drugs
Probability of Antibiotic Rx - BEARI
Cost of Encounter
Cost of Emergency Department Visit
Probability of Emergency Dept Visit
Probability of Adverse Drug Reaction
Utility of Health State - Uninfected
Net Monetary Benefits, Hundreds of Dollars
42

FIGURE 4: Net Monetary Benefits

Net monetary benefits indicate that all BEARI interventions provide greater benefit than the control group. The
willingness-to-pay axis does not exceed $800 for graphical clarity.

$(5,000)
$-
$5,000
$10,000
$15,000
$20,000
$25,000
$- $100 $200 $300 $400 $500 $600 $700 $800
Net Monetary Benefits
Willingness-To-Pay per QALY
Net Monetary Benefits
Control Suggested Alternatives Accountable Justification Peer Comparison
43

References
1. Shehab N, Patel PR, Srinivasan A, Budnitz DS. Emergency department visits for antibiotic-
associated adverse events. Clin Infect Dis. 2008;47(6):735-43.
2. Llop CJ, Tuttle E, Tillotson GS, LaPlante K, File TM, Jr. Antibiotic treatment patterns, costs, and
resource utilization among patients with community acquired pneumonia: a US cohort study.
Hosp Pract (1995). 2017:1-8.
3. Jones SC, Budnitz DS, Sorbello A, Mehta H. US-based emergency department visits for
fluoroquinolone-associated hypersensitivity reactions. Pharmacoepidemiol Drug Saf.
2013;22(10):1099-106.
4. Falagas ME, Bliziotis IA. Pandrug-resistant Gram-negative bacteria: the dawn of the post-
antibiotic era? Int J Antimicrob Agents. 2007;29(6):630-6.
5. Brown ED, Wright GD. Antibacterial drug discovery in the resistance era. Nature.
2016;529(7586):336-43.
6. CDC. Antibiotic resistance threats. Atlanta, GA: US Department of Health and Human Services;
2013. Accessed May 2017. Available: https://www.cdc.gov/drugresistance/pdf/ar-threats-2013-
508.pdf
7. CDC. About Antimicrobial Resistance: Four Core Actions to Fight Resistance. Centers for
Disease Control and Prevention; 2015. Accessed May 2017. Available:
https://www.cdc.gov/drugresistance/about.html
8. Ranji SR, Steinman MAS, Kaveh G;, Sundaram V, Lewis R, Arnold S, Gonzales R. Antibiotic
Prescribing Behavior Vol. 4: Closing the Quality Gap: A Critical Analysis of Quality
Improvement Strategies. Technical Review 9 (Prepared by Stanford University-UCSF Evidence-
based Practice Center under Contract No. 290-02-0017. Rockville, MD: Agency for Healthcare
Research and Quality; 2006.
9. Drekonja D, Filice G, Greer N, Olson A, MacDonald R, Rutks I, et al. Antimicrobial Stewardship
Programs in Outpatient Settings: A Systematic Review. Washington (DC); 2014.
10. Loewenstein G, Brennan T, Volpp KG. Asymmetric paternalism to improve health behaviors.
JAMA. 2007;298(20):2415-7.
11. Frolich A, Talavera JA, Broadhead P, Dudley RA. A behavioral model of clinician responses to
incentives to improve quality. Health Policy. 2007;80(1):179-93.
12. Meeker D, Linder JA, Fox CR, Friedberg MW, Persell SD, Goldstein NJ, et al. Effect of
Behavioral Interventions on Inappropriate Antibiotic Prescribing Among Primary Care Practices:
A Randomized Clinical Trial. JAMA. 2016;315(6):562-70.
13. Hallsworth M, Chadborn T, Sallis A, Sanders M, Berry D, Greaves F, et al. Provision of social
norm feedback to high prescribers of antibiotics in general practice: a pragmatic national
randomised controlled trial. Lancet. 2016;387(10029):1743-52.
14. Hunter R. Cost-effectiveness of point-of-care C-reactive protein tests for respiratory tract
infection in primary care in England. Adv Ther. 2015;32(1):69-85.
15. Michaelidis CI, Zimmerman RK, Nowalk MP, Fine MJ, Smith KJ. Cost-effectiveness of
procalcitonin-guided antibiotic therapy for outpatient management of acute respiratory tract
infections in adults. J Gen Intern Med. 2014;29(4):579-86.
16. Oppong R, Jit M, Smith RD, Butler CC, Melbye H, Molstad S, et al. Cost-effectiveness of point-
of-care C-reactive protein testing to inform antibiotic prescribing decisions. Br J Gen Pract.
2013;63(612):e465-71.
17. Zoorob R, Sidani MA, Fremont RD, Kihlberg C. Antibiotic Use in Acute Upper Respiratory
Tract Infections. Am Fam Physician. . 2012 86(9):817-22.
18. Jenkins SG, Brown SD, Farrell DJ. Trends in antibacterial resistance among Streptococcus
pneumoniae isolated in the USA: update from PROTEKT US Years 1-4. Ann Clin Microbiol
Antimicrob. 2008;7:1.
44

19. Spedicato GA, Kang TS, Yalamanchi SB, Yadav D. markovchain. 0.6.6 ed: R; 2017.
20. Schwartz LE, Brown RB. Purulent otitis media in adults. Arch Intern Med. 1992;152(11):2301-4.
21. Sokol W. Epidemiology of sinusitis in the primary care setting: results from the 1999-2000
respiratory surveillance program. Am J Med. 2001;111 Suppl 9A:19S-24S.
22. Chow AW, Benninger MS, Brook I, Brozek JL, Goldstein EJ, Hicks LA, et al. IDSA clinical
practice guideline for acute bacterial rhinosinusitis in children and adults. Clin Infect Dis.
2012;54(8):e72-e112.
23. Shulman ST, Bisno AL, Clegg HW, Gerber MA, Kaplan EL, Lee G, et al. Clinical practice
guideline for the diagnosis and management of group A streptococcal pharyngitis: 2012 update
by the Infectious Diseases Society of America. Clin Infect Dis. 2012;55(10):e86-102.
24. Fleming-Dutra KE, Hersh AL, Shapiro DJ, Bartoces M, Enns EA, File TM, Jr., et al. Prevalence
of Inappropriate Antibiotic Prescriptions Among US Ambulatory Care Visits, 2010-2011. JAMA.
2016;315(17):1864-73.
25. AHRQ. National Statistics on All Stays. Healthcare Cost and Utilization Project (HCUP): Agency
for Healthcare Research and Quality; 2016. Accessed May 2017. Available:
https://hcupnet.ahrq.gov/#setup
26. Ma L, Danoff TM, Borish L. Case fatality and population mortality associated with anaphylaxis
in the United States. J Allergy Clin Immunol. 2014;133(4):1075-83.
27. CDC. Active Bacterial Core Surveillance Report, Emerging Infections Program Network,
Streptococcus pneumoniae, 2014. Centers for Disease Control and Prevention; 2014. Accessed
Jun 2017. Available: https://www.cdc.gov/abcs/reports-findings/surv-reports.html
28. Centers for Disease Control and Prevention. BRFSS Prevalence & Trends Data. National Center
for Chronic Disease Prevention and Health Promotion; 2015.
29. Social Security Administration. Period Life Table, 2013. In: Administration SS, ed. Actuarial
Life Table: Social Security Administration; 2016. Accessed Jun 2017. Available:
https://www.ssa.gov/oact/STATS/table4c6.html
30. Luo N, Johnson JA, Shaw JW, Feeny D, Coons SJ. Self-reported health status of the general adult
U.S. population as assessed by the EQ-5D and Health Utilities Index. Medical Care.
2005;43(11):1078-86.
31. Bergus GR, Weber CA, Ernst ME, Ernst EJ. Do antibiotics affect the quality of life of patients
with upper respiratory tract illnesses? It might depend on one's luck. Int J Clin Pract.
2008;62(6):855-9.
32. Johnson FR, Banzhaf MR, Desvousges WH. Willingness to pay for improved respiratory and
cardiovascular health: a multiple-format, stated-preference approach. Health Economics.
2000;9(4):295-317.
33. Egger ME, Myers JA, Arnold FW, Pass LA, Ramirez JA, Brock GN. Cost effectiveness of
adherence to IDSA/ATS guidelines in elderly patients hospitalized for Community-Aquired
Pneumonia. BMC Med Inform Decis Mak. 2016;16:34.
34. Neumann PJ, Sanders GD, Russell LB, Siegel JE, Ganiats TG. Cost-Effectiveness in Health and
Medicine: Second Edition. New York: Oxford University Press; 2016.
35. Stinnett AA, Mullahy J. Net health benefits: a new framework for the analysis of uncertainty in
cost-effectiveness analysis. Med Decis Making. 1998;18(2 Suppl):S68-80.
36. Using Behavioral Science Insights to Better Serve the American People. Online: The White
House Office of the Press Secretary; 2015.
37. Bassi J, Lau F. Measuring value for money: a scoping review on economic evaluation of health
information systems. J Am Med Inform Assoc. 2013;20(4):792-801.
38. O'Reilly D, Tarride JE, Goeree R, Lokker C, McKibbon KA. The economics of health
information technology in medication management: a systematic review of economic evaluations.
J Am Med Inform Assoc. 2012;19(3):423-38.
39. Antibiotic Use in Food-Producing Animals. National Antimicrobial Resistance Monitoring
System for Enteric Bacteria (NARMS).
45

40. Andersson DI, Hughes D. Antibiotic resistance and its cost: is it possible to reverse resistance?
Nat Rev Microbiol. 2010;8(4):260-71.
41. Andersson DI, Hughes D. Persistence of antibiotic resistance in bacterial populations. FEMS
Microbiol Rev. 2011;35(5):901-11.
42. Belongia EA, Sullivan BJ, Chyou PH, Madagame E, Reed KD, Schwartz B. A Community
Intervention Trial to Promote Judicious Antibiotic Use and Reduce Penicillin-Resistant
Streptococcus pneumoniae Carriage in Children. Pediatrics. 2001;108(3):575-83.
43. Hennessy TW, Petersen KM, Bruden D, Parkinson AJ, Hurlburt D, Getty M, et al. Changes in
antibiotic-prescribing practices and carriage of penicillin-resistant Streptococcus pneumoniae: A
controlled intervention trial in rural Alaska. Clin Infect Dis. 2002;34(12):1543-50.
44. Electronic Health Record (EHR) Adoption. In: Technology TOotNCfHI, ed. Health Information
Technology Data Summaries: The Office of the National Coordinator for Health Information
Technology; 2015.
45. Gilmer TP, O'Connor PJ, Sperl-Hillen JM, Rush WA, Johnson PE, Amundson GH, et al. Cost-
effectiveness of an electronic medical record based clinical decision support system. Health Serv
Res. 2012;47(6):2137-58.
46. Shekelle P, Morton SC, Keeler EB. Costs and Benefits of Health Information Technology.
Evidence Report/Technology Assessment No. 132. (Prepared by the Southern California
Evidence-based Practice Center under Contract No. 290-02-0003.) AHRQ Publication No. 06-
E006. Rockville, MD: Agency for Healthcare Research and Quality; 2006.
47. Bird J, Biggs TC, Thomas M, Salib RJ. Adult acute rhinosinusitis. BMJ. 2013;346:f2687.
48. Fokkens WJ, Hoffmans R, Thomas M. Avoid prescribing antibiotics in acute rhinosinusitis. BMJ.
2014;349:g5703.
49. Little P, Stuart B, Hobbs FD, Butler CC, Hay AD, Campbell J, et al. Predictors of suppurative
complications for acute sore throat in primary care: prospective clinical cohort study. BMJ.
2013;347:f6867.
50. Zhang S, Palazuelos-Munoz S, Balsells EM, Nair H, Chit A, Kyaw MH. Cost of hospital
management of Clostridium difficile infection in United States-a meta-analysis and modelling
study. BMC Infect Dis. 2016;16(1):447.
51. Shupo F, Dorey J, Aballea S, McGarry T, Odeyemi II, Toumi M. PIN75 Health-Related Quality
of Life of Clostridium Difficile Infection: A Methodological Contribution to Direct Utility
Elicitation by TTO. Value in Health. 2012;15(7):A399.

46

Chapter 4: Management of Asymptomatic Term & Late Preterm Newborns Exposed to Maternal
Intrapartum Fever: A Societal Cost Benefit Analysis of the Proposed “Triple I” Algorithm
Abstract
Importance: The diagnosis of maternal chorioamnionitis has direct implications for subsequent neonatal
management. A recently proposed algorithm (Triple I) reconceptualizes the disease and provides clinical
guidance to minimize diagnostic testing and antibiotic use in asymptomatic newborns, using an early-
onset sepsis (EOS) calculator.

Objective: To perform an economic evaluation of neonatal management according to the Triple I
algorithm, compared to existing guidelines (CDC/AAP).

Design: We performed a cost-benefit analysis using a decision tree model from the U.S. societal
perspective. Published literature was used to input the probability of septicemia, meningitis,
consequences of infection and antibiotic use, and indirect costs for long-term disability and mortality,
Direct medical costs were based on Healthcare Cost and Utilization Project data and published literature.

Setting: U.S. 2015 birth cohort.

Participants: Asymptomatic newborns >34 weeks gestational age exposed to maternal intrapartum fever.

Interventions: Management per Triple I vs. CDC/AAP.

Main Outcomes: The potential net monetary benefit (NMB) of neonatal care under the Triple I algorithm
compared to existing CDC/AAP clinical guidelines.

Results: In the base case, there was a NMB of US$4114 per Triple I-treated infant exposed to maternal
intrapartum fever, compared to the CDC/AAP approach; 49% was due to the direct costs for medical care.
In probabilistic sensitivity analysis (10,000 Monte Carlo simulations), there was a 66% likelihood of
NMB favoring Triple I. One-way sensitivity analyses revealed that results were most sensitive to the
presence of bacteremia/clinical sepsis, mortality, due to early-onset sepsis/meningitis, and
aminoglycoside-associated ototoxicity. Per 1,000 live births with exposure to maternal intrapartum fever,
incidences of outcomes for Triple I vs. CDC/AAP respectively, were: 134 vs. 808 admissions to a higher
level of care, 0.73 vs. 0.61 cases of meningitis; 36.7 vs. 30.7 cases of sepsis; 1.9 vs. 11.3 cases of
aminoglycoside-associated ototoxicity; and 1.2 vs. 1.0 deaths. Adherence to Triple I would provide $473
million NMB per year in the U.S.

Conclusions and Relevance: Incorporating the Triple I/calculator algorithm into routine clinical practice
results in a net societal benefit that is robust to a variety of clinical probabilities.

47

Introduction
Fetal exposure to chorioamnionitis may result in a variety of neonatal complications, including
early-onset infection.[1-7] In 2002, the Centers for Disease Control and Prevention (CDC) recommended
empiric broad-spectrum antibiotic therapy for all asymptomatic infants born to women who had received
intrapartum antibiotics for suspected chorioamnionitis. [2] In 2010, CDC expanded its recommendation to
include maternal intrapartum fever and/or suspected chorioamnionitis alone as sufficient cause for
empiric therapy.[4] It further recommended that at a minimum, a complete blood count and culture be
obtained, regardless of the presence (or absence) of other risk factors for or clinical signs of early-onset
sepsis (EOS). Subsequently, the American Academy of Pediatrics’ Committee on Fetus and Newborn
(AAP-COFN) issued a report in 2012 to clarify its recommendations for the newborn at risk for sepsis. [4,
8-10].
Contemporaneous data on the effectiveness of these neonatal management strategies, however,
are not robust[1, 11] and therefore substantial variability in clinical practice exists.[3, 4, 12-14] Moreover,
unnecessary diagnostic testing and/or therapy has important biosocial implications for the patient, his/her
family, and society, such as impacts on hospital length of stay; maternal-infant bonding; breastfeeding
success; and harm related to diagnostic procedures and therapies.[1, 4, 15-20]
Acknowledging many of these concerns, a National Institute of Child Health and Human
Development expert panel developed guidance on management of chorioamnionitis and neonates born to
mothers with intrapartum fever.[21] The panel reconceptualized chorioamnionitis as “intrauterine
inflammation or infection or both” (Triple I) towards minimizing obstetric and pediatric clinical practice
variation. Within this paradigm, neonatal management is dependent on the clinical classification of
maternal fever as “Isolated Maternal Fever,” “Suspected Triple I,” or “Confirmed Triple I,” (Table 1).
Neonates with clear signs of sepsis or born to mothers with “Confirmed” (a very small proportion of all
women) are excluded.[2, 4, 15, 22] For those with “Suspected”, the authors endorse a validated EOS
calculator for well-appearing newborns ≥34 weeks gestation.[3, 7, 12, 13, 21] The potential clinical and
48

economic impacts of widespread adoption of the Triple I/calculator paradigm are unknown. We thus
performed a cost-benefit analysis (CBA) from the U.S. societal perspective to compare the Triple-I-based
management algorithm to current practices.
Methods
We performed a CBA using a decision tree to model the clinical course of well-appearing infants
born at >34 weeks gestational age exposed to maternal intrapartum fever (the target population for the
Triple I approach, Figure 1).[7, 15] We chose a CBA over a cost-utility analysis to allow complete
consideration of the clinical and economic implications of each management strategy, particularly given
the difficulty in ascertaining quality-of-life for neonatal outcomes, and assignment of health state utilities
to neonates.[23-27]
We defined empiric antibiotic use as administration within the first 12-24 hours of life requiring
transfer to a Level II or greater level of nursery for diagnostic evaluation and treatment. Otherwise,
infants were assumed to remain in Level I unless s/he became ill-appearing requiring admission to level
II/III/IV.[9, 19, 20, 28, 29]. Both the CDC/AAP and Triple I approach recommend empiric antibiotic
treatment for a minimum of 48 hours, by which point modern incubation systems will have detected
bacteria in >98% of cultures, if present.[30-33] Lacking direct data on antibiotic use, the base case
treatment duration equaled the length-of-stay (LOS) derived from HCUP data: non-bacteremic
septicemia, median LOS 2 days, bacteremia 7 days (mean 10.9 days) and meningitis, 14 days (mean 15.7
days); durations consistent with the published literature.[9, 20, 34-36] We calculated the antibiotic
utilization rate as total antibiotic days per 1000 patient-days.[37]
We included adverse outcomes that have quantifiable clinical and economic implications over the
lifetime of the affected individual. The most clinically significant antibiotic-related adverse effects among
newborns are reversible nephrotoxicity and permanent sensorineural hearing loss due to exposure to
aminoglycosides.[38-41] Infants who survive meningitis may suffer permanent neurocognitive deficits
such as intellectual disability, cerebral palsy, vision impairment, and hearing loss, among others.[28, 29,
49

34, 42] We also included post-discharge readmission because a significant proportion is related to
infection, [43] and management and complications are similar to that of an otherwise at-risk admitted
infant. We did not incorporate antibiotic-associated hypersensitivity reactions, (permanent)
nephrotoxicity, neurotoxicity, and myelosuppression as these are rare and/or reversible.[40, 44]
We derived probabilities of EOS, meningitis and mortality from published literature, with the
latter verified using data from the Agency for Healthcare Research and Quality Healthcare Cost and
Utilization Project (HCUP) National Inpatient Sample (NIS). [15, 28, 29, 34, 42, 45, 46] Probabilities
correspond to symbols in the decision tree as follows (Figure 1): γ for ill-appearing or positive blood
cultures; θ for positive CSF (meningitis); and δ, σ, μ for death due to CSF-positive disease, death due to
CSF-negative disease, and death after readmission for new-onset clinical illness, respectively. The
probabilities for readmission (ρ) and long-term sequelae of meningitis or drug-induced ototoxicity (λ, π)
were derived from additional published data. [43, 47-53] Prime superscripts (e.g. γ’, θ’, ρ’) designate
corresponding probabilities among infants who do not receive initial empiric antibiotics.
Direct costs associated with hospitalization were based on HCUP NIS cost data by ICD9
diagnosis code which include national estimates of all direct costs associated with acute management for
each diagnosis, including adverse drug reactions, medication errors, procedure complications, and
hospital-acquired infections.[54] We used ICD9 codes for healthy newborn infants who remained in the
Level I nursery, and for those admitted to higher levels of care for therapy, we used corresponding codes
for infants diagnosed with septicemia or infants diagnosed with streptococcal or Gram-negative
meningitis (Table 2). We weighted costs for meningitis by organism type according to their relative
prevalence in HCUP and contemporary epidemiologic data, and etiology-specific recommendations for
minimum duration of therapy to calculate an average composite cost of hospitalization.[34, 42, 55-58]
We excluded Listeria monocytogenes as it has become very uncommon in the U.S. [36, 55, 56, 58, 59]
Moreover, queries in the 2000-2014 HCUP data yielded no cases of listeriosis (ICD9 027.0) or bacterial
meningitis of other etiologies (ICD9 320.7) among neonates. Indirect costs for aminoglycoside-associated
50

ototoxicity and neurocognitive disability included lifetime costs of healthcare and special education
services attributable to these sequelae, as well as losses in individual economic productivity into
adulthood.[60-62] We used HCUP costs and LOS to estimate overall costs for newborns admitted to
level II/III/IV units for therapy and subsequently discharged after 48-72 hours. We used median costs for
the base case, and the expected minimum and maximum LOS to estimate the upper and lower bounds of
the range of costs.
The value of a life lost due to neonatal disease was based on published estimates of $4-10 million,
controlling for observed and unobserved differences in the population. [40-42] This range is also used in
economic valuations by the U.S. Environmental Protection Agency to analyze the impact of policies on
health.[63] We used a base estimate of $7 million, inflated to $9.6 million in 2017 US Dollars.[24, 25,
64] We subjected this parameter to threshold analysis (while holding other model inputs constant), to
identify the value at which the net cost-benefit for Triple I vs. CDC/AAP reaches equipoise.
All probabilities and costs are displayed in Table 2. These parameters were used to calculate the
cost-benefit per infant for the Triple I algorithm vs. the CDC/AAP consensus guidelines. All costs were
converted to 2017 US Dollars using the Medical Consumer Price Index.[64] Sensitivity analyses were
conducted to assess robustness of the base case to changes in model parameters. One-way sensitivity
analysis was conducted to determine the relative influence of each model parameter.[63] We also
conducted probabilistic sensitivity analyses using 10,000 Monte Carlo simulations to further evaluate
uncertainty. In the absence of discrete data on the distribution of each parameter estimate, all parameters
for cost and probability were varied according to a triangular distribution, which only requires a
minimum, maximum, and mode of a parameter, as per ranges reported in the literature (Table 1). We also
estimated the cumulative incremental net benefit in the U.S. per year with increasing levels of adoption of
the Triple I algorithm in clinical practice. All calculations were performed using Microsoft Excel, 2016.
51

Results
In the base case, the total direct and indirect cost per infant ≥34 weeks gestational age exposed to
maternal intrapartum fever was $17,174 for the Triple I algorithm compared to $21,288 under the current
guidelines; an incremental net benefit per infant of $4,114 under Triple I. Total direct costs for acute
medical care were $1,659 and $3,684 per infant for each algorithm respectively, yielding a net per-patient
benefit of $2,025 in direct medical costs alone - favoring the Triple I approach.
One-way sensitivity analyses show that the probability of receiving antibiotic thus necessitating a
transfer to a higher level of care, followed by costs associated with mortality or ototoxicity, had the
greatest effect on the magnitude of net benefit in favor of the Triple I/calculator approach (Figure 2). In
most cases, there is a net benefit under Triple I, ranging from $58 to $12,544. In rare instances, there is a
net cost of up to $1,290. Despite the sensitivity of the results to the probabilities and costs associated with
poor neonatal outcomes, the Triple I approach was always associated with a net benefit in 90% of the
models. A net benefit was predicted with the Triple I approach even when the proportion of infants
receiving antibiotics under the CDC algorithm was as low as 6.7% (from a base case of 80%).
In the base case, only 104 Triple I-treated infants require higher level of nursery care (to receive
antibiotic therapy and monitoring), compared with 808 infants under CDC/AAP, per 1,000 live births
exposed to maternal fever. Further, the burden of total antibiotic utilization was 230 antibiotic days per
1000 livebirths (range 188-474) and 824 (818-955) for the Triple I and CDC approaches, respectively. In
the base case, the absolute change in the incidence for each outcome per 1000 live birth was <1%
following a change from the CDC to the Triple I approach. Our model predicted fewer cases of
ototoxicity and readmission yet more cases of meningitis, clinical sepsis/bacteremia, and all-cause
mortality in the Triple I algorithm (Table 3). Setting the likelihood of ototoxicity (the outcome with the
greatest change favoring Triple I) at 0% from a base case of 1.4% still resulted in a per-infant net benefit
of $227.
52

Our probabilistic sensitivity analysis revealed - per 1,000 live-births exposed to maternal
intrapartum fever - the following outcomes for Triple I vs. CDC/AAP, respectively: median 1.40 [IQR
0.91-1.86] vs. 1.15 [0.76-1.57] deaths due to EOS or meningitis; 1.47 [IQR 0.79-2.44] vs. 1.20 [0.65-
2.04] cases of meningitis; 35.9 [27.9-42.19] vs. 29.7 [23.4-35.46] cases of sepsis/bacteremia; 4.55 [3.73-
5.68] vs. 4.49 [3.69-5.63] cases of neurocognitive disability due to early-onset meningitis; 2.55 [1.59-
3.43] vs. 15.33 [11.24-18.82] cases of ototoxicity; 21.26 [18.69-24.67] vs. 21.28 [18.66-24.80]
readmissions; and 145.9 [114.12-170.32] vs. 868.69 [806.43-916.67] admissions to non-Level I nursery.
Management according to the Triple I algorithm yielded a net benefit in 66% of the simulations. The
median NMB per infant was $9667 [$82,116 NMB to $46,455 net cost] with a 95% range from $57,000
in net benefit to $33,000 in net cost.
Among an estimated 3,978,497 live births in 2015,[62] of which 97.2% are of gestational age >34
weeks, approximately 3.3% (range 1-4.5%) were born of mothers with intrapartum fever.[62, 65]
Therefore we estimate 127,614 infants exposed to maternal intrapartum each year, yielding $473 million
in aggregate net societal benefit per year if 90% of clinicians transitioned from the CDC/AAP guidelines
to the proposed Triple I algorithm (base case). At 10% adoption, $52 million per year in net monetary
benefits is predicted, with an increase of approximately $52-55 million increase per year, up to $525
million with 100% adoption.
Discussion
Our results suggest that use of the Triple I/calculator approach for management of newborns
exposed to intrapartum fever yields a NMB of >$4100 per infant compared to CDC/AAP guidelines.
Nearly half of this benefit is derived from avoidance of transfer to a higher level of care for empiric
antibiotic use and monitoring for presumed at-risk neonates. It has been frequently suggested that too
many neonates born of mothers with intrapartum fever are receiving diagnostic evaluation and antibiotics
- the primary motivation for convening the NIH expert panel that developed the Triple I concept.[30, 66]
53

Our data provide first evidence that adoption of the Triple I algorithm would provide substantial NMB
when applied to the U.S. birth cohort.
The base case and sensitivity analyses demonstrate that the location of care is the most important
driver of cost. Even if indirect costs are excluded from our analyses, the Triple I approach still yields a
per-patient NMB (>$2000 per infant) that is entirely attributable to reduction of excess direct medical
costs. Only ~10% of intrapartum fever-exposed neonates managed according to the Triple I/calculator
approach would require admission to NICU or a step-down unit, compared to ~81% of neonates under
CDC/AAP management according to our base case, representing tremendous direct medical cost savings.
Our base case NMB results were robust in all sensitivity analyses and consistent with clinical
expectations and the published literature. In the one-way sensitivity analysis, Triple I was associated with
higher costs only at extreme inputs for antibiotic use and likelihood of death due to sepsis or meningitis,
the two most important drivers of cost. Further, when the clinical outcome with the largest change in
incidence (ototoxicity) was set to zero, NMB persisted, albeit at a low value per infant. In the probabilistic
sensitivity analysis, per-infant NMB was found in 66% of the simulations. Finally, we also found that the
NMB threshold (the transition point to net cost) was reached only at unrealistically low levels of
antibiotic use in the CDC paradigm (<10%) and a very high value for a neonatal life lost of >$30 million,
a value which far exceeds the accepted values typically used in economics literature and government
policy practices.[63, 67-70]
Herein we present the first estimates of the burden of antibiotic use avoided under the EOS
calculator among the US cohort of neonates born of mothers with intrapartum fever. Use of the calculator
necessarily reduces the number infants that receive empiric antibiotics since it identifies only those with
an elevated prior probability of sepsis (beyond maternal fever alone) where the CDC/AAP guideline does
not. Existing studies note that ~3-5% of infants would be recommended to receive empiric therapy with
the calculator, most of whom were exposed to maternal fever. This, however, may underestimate the
actual number who would receive antibiotics given that these cohorts may underrepresent populations at
54

higher risk for neonatal sepsis. It is for this reason that we chose a base case of 10% antibiotic use and a
range up to 15% and a higher risk for sepsis therefrom in our models.
The absolute change in incidence for all clinical outcomes was <1% between Triple I and the
CDC approach. The largest change was observed for ototoxicity in the Triple I group, a consequence of
less aminoglycoside exposure. Overall, hearing loss is detected in 2-3% of neonates who receive
aminoglycosides, which is the most prevalent risk factor for hearing loss in this population.[47, 48, 71,
72] The 84% relative decline in incidence we observed highlights the impact of aminoglycoside use,
despite the possible existence of co-morbidities associated with hearing loss in some patients. Our data
were not sufficiently robust to enable evaluation of other known antibiotic-associated complications such
as renal damage, drug-drug interactions, and development of bacterial resistance. Our models also predict
slightly more cases per year of meningitis, sepsis, and death using Triple I. For clinical sepsis/bacteremia,
others have shown that only ~21% of cases of “sepsis” are definitively proven (blood culture positive),
equivalent to 159 cases in the U.S. in our base case. Nonetheless, the number of excess cases may fall to
zero given the limits of each outcome by clinical condition, as shown in our probabilistic analysis.
We believe our base case results underestimate the potential NMB associated with transition to
the Triple I/calculator approach. We were unable to include potential cost savings associated with less
aggressive rule out sepsis management with outcomes, for which incidence and/or cost data are not
available. These include reductions in unnecessary laboratory testing, clinical monitoring, and other
patient harms, and less well-defined adverse consequences related to delay in initiation of breast-feeding,
maternal peripartum depression, and potential consequence of changes in the gut microbiome from broad
spectrum antibiotic use [16, 22, 73]. We believe our base-case analysis represents a clinically justified
scenario, given the available data for each of the model parameters. Further, our probabilistic sensitivity
analysis provides contextual information on the statistical uncertainty of each of our parameters to
generate a prediction of model outcomes based on the assumed distribution of each model parameter.
55

We acknowledge certain limitations of our analysis. First, we included only newborns of
gestational age >34 weeks to mirror the Triple I population target. Second, we assumed that neonates who
receive empiric therapy are transferred from the lowest level of care. Third, limitations of HCUP data and
the absence of granular published data required us to combine neonates with documented bacteremia and
those with clinical sepsis into one subgroup. Fourth, our model could not include other strategies to lessen
antibiotic use such as stewardship strategies or emerging biomarkers.[66, 74-78] Finally, we could not
model infants born to mothers without intrapartum fever due to lack of available risk and outcome data, as
noted by the NIH Triple I expert panel.[21]
We appreciate that administrative data are variably robust [79], and obtaining precise costs and
probabilities for each health state was not possible. We also did not have robust clinical data on the
impact of intrapartum chemoprophylaxis on neonatal disease risk, as noted in prior studies of the EOS
calculator studies.[7, 8, 54] Further, there are few high-quality clinical data to stratify this risk reduction
according to the Triple I chorioamnionitis classification scheme (Table 1). Our model, however, used
maternal fever as this is the most common clinical scenario. [21, 80]
Conclusion
Our model provides compelling first evidence that the Triple I - and by extension the EOS
calculator – clinical management approach for neonates exposed to intrapartum fever provides a net
monetary benefit to society. Our data suggest that under various scenarios, this management strategy has
a >65% likelihood of resulting in net societal benefit. At the highest levels of adoption, the Triple
I/calculator approach could yield US$525 million in aggregate value, and likely more, had we been able
to include the long-term psychosocial benefits of reduced antibiotic treatment, as compared to the
CDC/AAP status quo. The robustness of the NMB, combined with a consideration of the general benefits
of de-escalating care for newborns deemed at lower risk of sepsis, presents useful data for consideration
by hospital stakeholders in their decision-making regarding neonatal sepsis management.
56

Figure 1: Decision Model – Management of Asymptomatic Term and Late Preterm Newborns
Exposed to Maternal Intrapartum Fever
1
Maternal intrapartum
fever-exposed,
asymptomatic neonate
>34 weeks GA
Empiric antibiotics
administered
No empiric antibiotics
administered
Becomes ill-appearing
and/or blood cultures (+)
Remains well-appearing
and blood cultures (-)
Discharged home
CSF (+)
CSF (-)
Death
Discharged home
Readmitted
Ototoxicity
Healthy
Death
Discharged home
Death
Discharged home
Healthy
Neurocognitive disability
Healthy
Neurocognitive disability
Healthy
Ototoxicity
Discharged home Readmitted
Remain healthy
Becomes ill-appearing
and/or blood cultures (+)
CSF (+)
CSF (-)
Death
Discharged home
Death
Discharged home
Healthy
Neurocognitive disability
Healthy
Ototoxicity Remains well-appearing
and blood cultures (-)
Death
Discharged home
Healthy
Neurocognitive disability
12-24 HOL 7 DOL 48-72 HOL
α
β
γ'
η
η'
π
ρ
ρ'
δ
1 - δ
1 - σ
λ
μ
1 - μ
σ
λ
π
λ
λ
θ
14-21+ DOL
θ'
1 - δ
1 - σ
1 - μ
δ
σ
μ
1
1 - λ
1 - λ
1 - π
1 - λ
1 - λ
1 - π
π
1 - γ
1 - γ’
γ
1 - θ
1 - θ’
Birth Longterm

Under the CDC guidelines, most (if not all) neonates will receive empiric antibiotics at birth. In the Triple I
algorithm, most neonates will not receive empiric antibiotics at birth, remaining under observation with or without
limited evaluation (e.g. blood culture), following the branch on the bottom half of the decision tree. HOL: Hours of
Life. DOL: Days of Life.

57

Figure 2: One-Way Sensitivity Analysis

The range of net benefits is shown in thousands of dollars. One-way sensitivity analysis reveals that the
probabilities of death due to CSF(+) under Triple I or CDC/AAP, becoming ill-appearing under CDC/AAP, and the
cost of death under Triple I have the most significant impact on model results, resulting in net costs under Triple I at
the extreme end of each parameter’s range. Ranges for each parameter can be found in Table 1.
-$12.54
-$9.35
-$9.07
-$8.94
-$9.03
-$8.50
-$8.17
-$6.97
-$5.00
-$5.92
-$4.84
-$5.24
-$4.79
-$4.71
-$4.50
-$4.99
-$4.73
-$4.60
-$4.19
-$4.55
-$4.47
-$4.64
-$4.19
-$4.65
-$4.37
-$4.24
-$4.48
-$4.31
-$4.30
-$4.48
-$4.19
-$4.27
-$4.29
-$4.29
-$4.18
-$4.18
-$4.12
-$4.11
-$4.12
-$4.11
-$0.91
$1.12
$1.29
$0.71
-$0.16
$0.27
-$0.06
-$1.26
-$2.30
-$3.23
-$2.21
-$3.27
-$3.49
-$3.43
-$3.24
-$3.72
-$3.58
-$3.51
-$3.30
-$3.82
-$3.76
-$4.03
-$3.59
-$4.07
-$3.80
-$3.68
-$4.01
-$3.92
-$3.92
-$4.11
-$3.90
-$3.98
-$4.05
-$4.10
-$4.04
-$4.06
-$4.05
-$4.07
-$4.11
-$4.11
-$14 -$12 -$10 -$8 -$6 -$4 -$2 $0 $2
Probability - ill-appearing, γ', Triple I
Probability - Death, CSF (+), Triple I
Probability - ill-appearing, γ, CDC
Cost - Death, Triple I
Cost - Ototoxicity, CDC
Probability - Death, CSF (+), CDC
Cost - Death, CDC
Probability - Ototoxicity, CDC
Cost - Neurocognitive disability, Triple I
Cost - Neurocognitive disability, CDC
Probability - ill-appearing, γ', CDC
Probability - antibiotics given, CDC
Probability - ill-appearing, γ, Triple I
Probability - Readmission, ρ', Triple I
Probability - Neurocognitive disability, Triple I
Probability - Neurocognitive disability, CDC
Probability - Readmission, ρ, CDC
Cost - Ototoxicity, Triple I
Probability -CSF cultures (+), θ', CDC
Probability - antibiotics given, Triple I
Probability - Ototoxicity, Triple I
Cost - Readmission, CDC
Cost - Readmission, Triple I
Probability -CSF cultures (+), θ, CDC
Cost - No antibiotics, Triple I
Cost - Blood cultures (+), Triple I
Cost - Blood cultures (+), CDC
Probability - Death, CSF (-), CDC
Probability - Death, CSF (-), Triple I
Cost - Antibiotics, CDC
Probability - Death, readmission, Triple I
Probability - Readmission, ρ', CDC
Probability - Death, readmission, CDC
Probability -CSF cultures (+), θ', CDC
Probability - Readmission, ρ, Triple I
Cost - No antibiotics, CDC
Probability -CSF cultures (+), θ, Triple I
Cost - Antibiotics, Triple I
Cost - CSF cultures (+), CDC
Cost - CSF cultures (+), Triple I
Net Benefit ($, Thousands)
One-Way Sensitivity Analysis
58

Figure 3: Probabilistic Sensitivity Analysis (Likelihood of Net Benefit)

There is a net benefit in 66% of cases. We expect a net benefit of up to approximately $10,000 in about 50% of
cases. 34% of cases yield either no benefit, or increased costs under Triple I management vs. CDC/AAP.

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
$(100) $(80) $(60) $(40) $(20) $- $20 $40 $60
Probability of Net Benefit
Net Benefit ($, Thousands)
Thousands
Net Benefit per Infant
NET BENEFIT NO NET BENEFIT
59

Table 1: Intraamniotic Infection Classification
Terminology Triple I Definition[21] ACOG Definition[80]
Isolated maternal
fever
(“documented”
fever)
Maternal oral temperature 39.0°C
(102.2°F) or greater on any one occasion is
documented fever. IF the oral temperature
is between 38.0°C (100.4°F) and 39.0°C
(102.2°F), repeat the measurement in 30
minutes; if the repeat value remains at least
38.0°C (100.4°F), it is documented fever.
Any temperature between 38°C and
38.9°C with no other clinical criteria
indicating intraamniotic infection, and with
or without persistent temperature
elevation.
Suspected
intraamniotic
infection
Fever without a clear source plus any of the
following:
1) Baseline fetal tachycardia (Greater
than 160 beats per minute for 10
minutes of longer, excluding
accelerations, decelerations, and
periods of marked variability)
2) Maternal white blood cell count
greater than 15,000 per mm
3
in the
absence of corticosteroids
3) Definite purulent fluid from the
cervical os
Fever of >39°C with no other clinical risk
factors present, absent an obvious
alternative source.
Confirmed
intraamniotic
infection
All of the above plus:
1) Amniocentesis-proven infection
through a positive Gram stain
2) Low glucose or positive amniotic
fluid culture
3) Placental pathology revealing
diagnostic features of infection
No practical distinction from “Suspected”;
confirmed infection will most commonly
be made postpartum, based on
histopathologic study of the placenta.

60

Table 2: Decision Model Inputs
Parameter Health State Cost Range ICD9 Code Reference
α Antibiotics given $3,933 $3,933-19,663 V29.0 [45]
β No antibiotics given $1,002 $705-1,363 V30.0 [45]
γ, γ’ Blood culture (+) $11,917 $8,512-23,833 771.81 [45]
Blood culture (-) $3,933 $3,933-19,663 [45]
θ, θ' CSF (+) $24,117 $15,961-33,518 320.2, 320.82 [45]
CSF (-) $11,917 $8,512-23,833 [45]
ρ, ρ' Readmission $5,093 $428-35,593 [43]
λ Neurocognitive disability $1,087,960 $828,668-1,619,117 [60]
π Ototoxicity $393,151 $42,366-828,668 [60]
δ, σ, μ Mortality $9,653,532 $5,516,304-13,790,761 [67-70]

Parameter Health State Probability Range ICD9 Code Reference
α CDC Antibiotics given – CDC 0.80 0.65-1
[15, 20, 29,
65, 81]
α Triple I Antibiotics given - Triple I 0.10 0.025-0.15 [29, 82]
β No antibiotics given =1 - α
γ
Becomes ill-appearing OR blood
culture (+)
0.029 0.007-0.049
[15, 28, 29,
83]

Remains well-appearing AND blood
culture (-)
=1 - γ
γ'
Becomes ill-appearing OR blood
culture (+)
0.0375 0.007-0.049 [15, 28, 29]

Remains well-appearing AND blood
culture (-)
=1 – γ’
θ CSF (+) 0.02 0.013-0.098 [2, 81, 82]
CSF (-) – antibiotics given =1 - θ
θ' CSF (+) – no antibiotics given 0.02 0.013-0.098 [2, 81, 82]
CSF (-) – no antibiotics given =1 - θ'
ρ Readmitted – antibiotics given 0.0179 0.0148-0.215 [43]
ρ' Readmitted – no antibiotics given 0.0179 0.0148-0.215 [43]
61

λ Neurocognitive disability 0.19 0.17-0.235 [50-53]
π Ototoxicity 0.014 0.005-0.023 [47, 48, 84]
δ Mortality - CSF (+) 0.040000 0.0293-0.071 320.2, 320.82 [42, 45]
σ Mortality - CSF (-) 0. 0311 0.016-0.0462 320.2, 320.82 [45, 56]
μ Mortality - readmission 0.00117 0-0.0023 V30.0 [45]
*Cost of hospitalization (α) for asymptomatic newborns admitted to high levels of care (Level II) for “rule-out EOS”
(LOS 48-72h) derived from blood culture (+) (γ), as described in Methods.

Table 3: Base Case Clinical Outcomes
§

Outcome* CDC
(per 1000 live
births)^
Triple I
(per 1000 live
births)^
Absolute
difference
(%)
Excess
cases/year
in U.S.
Admission to higher level of care
Ototoxicity
Clinical sepsis/bacteremia
All cause mortality
Readmission
Meningitis
Neurocognitive changes
808
11.28
30.17
0.98
17.35
0.61
3.41
104
1.85
36.65
1.17
17.24
0.73
3.40
-7.0
-0.94
0.65
0.02
-0.01
0.01
-0.001
-85908
-1204
759**
24
-14
15
<1

* With base case 90% and 80% adherence to Triple I and CDC algorithm, respectively
** Per published literature, an average of 21% of babies with this ICD9 code is blood culture positive (159 neonates
in our base case/year in the U.S.)
^ Median per 1000 neonates born to a mother with maternal fever
§ Because results are generated from a model, standard population-based estimates of dispersion are not appropriate
to report. Interquartile ranges are reported in the probabilistic sensitivity analysis results.

62

References

1. Cotten, C.M., Antibiotic stewardship: reassessment of guidelines for management of neonatal
sepsis. Clin Perinatol, 2015. 42(1): p. 195-206, x.
2. Schrag, S., et al, Prevention of Perinatal Group B Streptococcal Disease – Revised Guidelines
from CDC. MMWR, 2002. 51(RR-11): p. 1*24.
3. Malloy, M.H., Chorioamnionitis: epidemiology of newborn management and outcome United
States 2008. J Perinatol, 2014. 34(8): p. 611-5.
4. Randis, T.M., R.A. Polin, and G. Saade, Chorioamnionitis: time for a new approach. Curr Opin
Pediatr, 2017. 29(2): p. 159-164.
5. Fishman, S.G. and S.E. Gelber, Evidence for the clinical management of chorioamnionitis. Semin
Fetal Neonatal Med, 2012. 17(1): p. 46-50.
6. Johnson, C.T., A. Farzin, and I. Burd, Current management and long-term outcomes following
chorioamnionitis. Obstet Gynecol Clin North Am, 2014. 41(4): p. 649-69.
7. Tita, A.T. and W.W. Andrews, Diagnosis and management of clinical chorioamnionitis. Clin
Perinatol, 2010. 37(2): p. 339-54.
8. Brady, M.T. and R.A. Polin, Prevention and management of infants with suspected or proven
neonatal sepsis. Pediatrics, 2013. 132(1): p. 166-8.
9. Polin, R.A., F. Committee on, and Newborn, Management of neonates with suspected or proven
early-onset bacterial sepsis. Pediatrics, 2012. 129(5): p. 1006-15.
10. Taylor, J.A. and D.J. Opel, Choriophobia: a 1-act play. Pediatrics, 2012. 130(2): p. 342-6.
11. Polin, R.A., et al., The conundrum of early-onset sepsis. Pediatrics, 2014. 133(6): p. 1122-3.
12. Escobar, G.J., et al., Stratification of risk of early-onset sepsis in newborns >/= 34 weeks'
gestation. Pediatrics, 2014. 133(1): p. 30-6.
13. Puopolo, K.M., et al., Estimating the probability of neonatal early-onset infection on the basis of
maternal risk factors. Pediatrics, 2011. 128(5): p. e1155-63.
14. Benitz, W.E., J.L. Wynn, and R.A. Polin, Reappraisal of guidelines for management of neonates
with suspected early-onset sepsis. J Pediatr, 2015. 166(4): p. 1070-4.
15. Kiser, C., et al., Role of guidelines on length of therapy in chorioamnionitis and neonatal sepsis.
Pediatrics, 2014. 133(6): p. 992-8.
16. Saari, A., et al., Antibiotic exposure in infancy and risk of being overweight in the first 24 months
of life. Pediatrics, 2015. 135(4): p. 617-26.
17. Zeissig, S. and R.S. Blumberg, Life at the beginning: perturbation of the microbiota by
antibiotics in early life and its role in health and disease. Nat Immunol, 2014. 15(4): p. 307-10.
18. Gibson, M.K., T.S. Crofts, and G. Dantas, Antibiotics and the developing infant gut microbiota
and resistome. Curr Opin Microbiol, 2015. 27: p. 51-6.
19. Vangay, P., et al., Antibiotics, pediatric dysbiosis, and disease. Cell Host Microbe, 2015. 17(5):
p. 553-64.
20. Mukhopadhyay, S., et al., Effect of early-onset sepsis evaluations on in-hospital breastfeeding
practices among asymptomatic term neonates. Hosp Pediatr, 2015. 5(4): p. 203-10.
21. Higgins, R.D., et al., Evaluation and Management of Women and Newborns With a Maternal
Diagnosis of Chorioamnionitis: Summary of a Workshop. Obstet Gynecol, 2016. 127(3): p. 426-
36.
22. Li, M., M. Wang, and S.M. Donovan, Early development of the gut microbiome and immune-
mediated childhood disorders. Semin Reprod Med, 2014. 32(1): p. 74-86.
23. Akker-van Marle, M.E., et al., Cost-effectiveness of different treatment strategies with
intrapartum antibiotic prophylaxis to prevent early-onset group B streptococcal disease. BJOG,
2005. 112(6): p. 820-6.
24. Grosse, S.D., The use of economic evaluation to inform newborn screening policy decisions: the
Washington State experience. Milbank Q, 2016. 4(2): p. 366-91.
63

25. Ding, Y., et al., Cost-Effectiveness/Cost-Benefit Analysis of Newborn Screening for Severe
Combined Immune Deficiency in Washington State. J Pediatr, 2016. 172: p. 127-35.
26. Ding, Y., et al., Cost-benefit analysis of in-hospital influenza vaccination of postpartum women.
Obstet Gynecol, 2012. 119(2 Pt 1): p. 306-14.
27. Drummond, M.F., et al., Cost-Benefit Analysis, in Methods for the Economic Evaluation of
Health Care Programmes. 2005, Oxford University Press: New York. p. 211-245.
28. Jackson, G.L., et al., Hospital course and short-term outcomes of term and late preterm neonates
following exposure to prolonged rupture of membranes and/or chorioamnionitis. Pediatr Infect
Dis J, 2012. 31(1): p. 89-90.
29. Shakib, J., et al., Management of newborns born to mothers with chorioamnionitis: is it time for a
kinder, gentler approach? Acad Pediatr, 2015. 15(3): p. 340-4.
30. Kumar, Y., et al., Time to positivity of neonatal blood cultures. Archives of Disease in Childhood
- Fetal and Neonatal Edition, 2001. 85(3): p. 182F-186.
31. Kurlat, I., B.J. Stoll, and J.E. McGowan, Jr., Time to positivity for detection of bacteremia in
neonates. J Clin Microbiol, 1989. 27(5): p. 1068-71.
32. Sarkar, S.S., et al., Does maternal intrapartum antibiotic treatment prolong the incubation time
required for blood cultures to become positive for infants with early-onset sepsis? Am J
Perinatol, 2015. 32(4): p. 357-62.
33. Leazer, R., et al., Epidemiology of Cerebrospinal Fluid Cultures and Time to Detection in Term
Infants. Pediatrics, 2017. 139(5).
34. Ku, L.C., K.A. Boggess, and M. Cohen-Wolkowiez, Bacterial meningitis in infants. Clin
Perinatol, 2015. 42(1): p. 29-45, vii-viii.
35. Oliver, E.A., et al., Patterns of Empiric Antibiotic Administration for Presumed Early-Onset
Neonatal Sepsis in Neonatal Intensive Care Units in the United States. Am J Perinatol, 2017.
34(7): p. 640-647.
36. Simonsen, K.A., et al., Early-onset neonatal sepsis. Clin Microbiol Rev, 2014. 27(1): p. 21-47.
37. National Center for Emerging and Zoonotic Infectious Diseases. NHSN Antimicrobial Use and
Resistance (AUR) Module. 2017 [cited 2017 October 31]; Available from:
https://www.cdc.gov/nhsn/pdfs/training/aur/aur-training.pdf.
38. Matz, G.J., Aminoglycoside cochlear ototoxicity. Otolaryngol Clin North Am, 1993. 26(5): p.
705-12.
39. McCracken, G.H., Jr., Aminoglycoside toxicity in infants and children. Am J Med, 1986. 80(6B):
p. 172-8.
40. APP Pharmaceuticals, Gentamicin Package Insert. 2013: Lake Zurich, IL.
41. American Academy of Pediatrics Committee on, Fetus, and Newborn, Levels of neonatal care.
Pediatrics, 2012. 130(3): p. 587-97.
42. Ouchenir, L., et al., The Epidemiology, Management, and Outcomes of Bacterial Meningitis in
Infants. Pediatrics, 2017. 140(1).
43. Young, P.C., K. Korgenski, and K.F. Buchi, Early readmission of newborns in a large health
care system. Pediatrics, 2013. 131(5): p. e1538-44.
44. Shaffer, C.L., et al., Ampicillin-induced neurotoxicity in very-low-birth-weight neonates. Ann
Pharmacother, 1998. 32(4): p. 482-4.
45. Healthcare Utilization Project (HCUP) National Inpatient Sample (NIS). 2014, Agency for
Healthcare Research and Quality (AHRQ).
46. Towers, C.V., et al., Incidence of fever in labor and risk of neonatal sepsis. Am J Obstet Gynecol,
2017. 216(6): p. 596 e1-596 e5.
47. Contopoulos-Ioannidis, D.G., Extended-Interval Aminoglycoside Administration for Children: A
Meta-analysis. Pediatrics, 2004. 114(1): p. e111-e118.
48. Nestaas, E., et al., Aminoglycoside extended interval dosing in neonates is safe and effective: a
meta-analysis. Arch Dis Child Fetal Neonatal Ed, 2005. 90(4): p. F294-300.
64

49. Nizet, V. and J.O. Klein, Bacterial Sepsis and Meningitis, in Infectious Diseases of the Fetus and
Newborn Infant. 2016, Elsevier: Philadelphia, PA.
50. de Louvois, J., S. Halket, and D. Harvey, Neonatal meningitis in England and Wales: sequelae at
5 years of age. Eur J Pediatr, 2005. 164(12): p. 730-4.
51. Klinger, G., et al., Predicting the outcome of neonatal bacterial meningitis. Pediatrics, 2000.
106(3): p. 477-82.
52. Levent, F., et al., Early outcomes of group B streptococcal meningitis in the 21st century. Pediatr
Infect Dis J, 2010. 29(11): p. 1009-12.
53. Libster, R., et al., Long-term outcomes of group B streptococcal meningitis. Pediatrics, 2012.
130(1): p. e8-15.
54. Agency for Healthcare Research and Quality. National Inpatient Sample (NIS) Overview. 2017
March 22 [cited 2017 July 11]; Available from: https://www.hcup-us.ahrq.gov/nisoverview.jsp.
55. Stoll, B.J., et al., Early onset neonatal sepsis: the burden of group B Streptococcal and E. coli
disease continues. Pediatrics, 2011. 127(5): p. 817-26.
56. Weston, E.J., et al., The burden of invasive early-onset neonatal sepsis in the United States, 2005-
2008. Pediatr Infect Dis J, 2011. 30(11): p. 937-41.
57. Greenhow, T.L., et al., The changing epidemiology of serious bacterial infections in young
infants. Pediatr Infect Dis J, 2014. 33(6): p. 595-9.
58. Thigpen, M.C., et al., Bacterial meningitis in the United States, 1998-2007. N Engl J Med, 2011.
364(21): p. 2016-25.
59. Lee, B., J.G. Newland, and R. Jhaveri, Reductions in neonatal listeriosis: "Collateral benefit" of
Group B streptococcal prophylaxis? J Infect, 2016. 72(3): p. 317-23.
60. Institute of Medicine (US) Committee on Understanding Premature Birth and Assuring Healthy
Outcomes, Societal Costs of Preterm Birth, in Preterm Birth: Causes, Consequences, and
Prevention, R.E. Behrman and A.S. Butler, Editors. 2007, National Academies Press:
Washington (DC)
61. Centers for Disease Control and Prevention, Economic costs associated with mental retardation,
cerebral palsy, hearing loss, and vision impairment—United States, 2003. Morbidity and
Mortality Weekly Report, 2004. 53(3): p. 57-59.
62. Martin, J.A., et al., Births: final Data for 2015. National Vital Statistics Reports, 2017. 66(1): p.
1-70.
63. US Environmental Protection Agency. Guidelines for Preparing Economic Analyses. 2014 [cited
2017 August 29]; Available from: https://www.epa.gov/environmental-economics/guidelines-
preparing-economic-analyses#whatare.
64. Neumann, P.J., et al., Cost-Effectiveness in Health and Medicine: Second Edition. 2016, New
York: Oxford University Press.
65. Verani, J.R., et al, Prevention of Perinatal Group B Streptococcal Disease - Revised Guidelines
from CDC, 2010. MMWR, 2010. 59(RR-10): p. 1-31.
66. Cantey, J.B. and S.D. Baird, Ending the Culture of Culture-Negative Sepsis in the Neonatal ICU.
Pediatrics, 2017. 140(4).
67. Hoffer, S., et al., Economic Values for Evaluation of Federal Aviation Administration Investment
and Regulatory Program. Report FAA-APO-98-8. 1998, U.S. Federal Aviation Administration:
Washington, D.C.
68. Kniesner, T.J., et al., The Value of a Statistical Life: Evidence from Panel Data. The Review of
Economics and Statistics, 2012. 94(1): p. 74-87.
69. Viscusi, W.K., The Value of Life in Legal Contexts: Survey and Critique. American Law and
Economics Review, 2000. 2(1): p. 195-222.
70. Viscusi, W.K. and J.E. Aldy, The Value of a Statistical Life: A Critical Review of Market
Estimates Throughout the World. The Journal of Risk and Uncertainty, 2003. 27: p. 5-76.
71. Bielecki, I., A. Horbulewicz, and T. Wolan, Risk factors associated with hearing loss in infants:
an analysis of 5282 referred neonates. Int J Pediatr Otorhinolaryngol, 2011. 75(7): p. 925-30.
65

72. Hess, M., et al., Hearing screening in at-risk neonate cohort. Int J Pediatr Otorhinolaryngol,
1998. 46(1-2): p. 81-9.
73. Yassour, M., et al., Natural history of the infant gut microbiome and impact of antibiotic
treatment on bacterial strain diversity and stability. Sci Transl Med, 2016. 8(343): p. 343ra81.
74. Arora, H.S., et al., Enhanced Identification of Group B Streptococcus and Escherichia Coli in
Young Infants with Meningitis Using the Biofire Filmarray Meningitis/Encephalitis Panel.
Pediatr Infect Dis J, 2017. 36(7): p. 685-687.
75. Brozanski, B.S., et al., Use of polymerase chain reaction as a diagnostic tool for neonatal sepsis
can result in a decrease in use of antibiotics and total neonatal intensive care unit length of stay.
J Perinatol, 2006. 26(11): p. 688-92.
76. Patel, T.S., et al., Cost Analysis of Implementing Matrix-Assisted Laser Desorption Ionization-
Time of Flight Mass Spectrometry Plus Real-Time Antimicrobial Stewardship Intervention for
Bloodstream Infections. J Clin Microbiol, 2017. 55(1): p. 60-67.
77. Stocker, M., et al., Procalcitonin-guided decision making for duration of antibiotic therapy in
neonates with suspected early-onset sepsis: a multicentre, randomised controlled trial (NeoPIns).
The Lancet, 2017. 390(10097): p. 871-881.
78. Stranieri, I., et al., Usefulness of a 16S rDNA real-time PCR to monitor neonatal sepsis and to
assist in medical decision to discontinue antibiotics. J Matern Fetal Neonatal Med, 2016. 29(13):
p. 2141-4.
79. Ladha, K.S. and M. Eikermann, Codifying healthcare--big data and the issue of misclassification.
BMC Anesthesiol, 2015. 15: p. 179.
80. Heine, R.P., et al., ACOG Committee Opinion No. 712: Intrapartum Management of
Intraamniotic Infection. Obstet Gynecol, 2017. 130(2): p. e95-101.
81. Wortham, J.M., et al., Chorioamnionitis and Culture-Confirmed, Early-Onset Neonatal
Infections. Pediatrics, 2016. 137(1).
82. Kuzniewicz, M.W., et al., A Quantitative, Risk-Based Approach to the Management of Neonatal
Early-Onset Sepsis. JAMA Pediatr, 2017. 171(4): p. 365-371.
83. Benitz, W.E., J.B. Gould, and M.L. Druzin, Risk factors for early-onset group B streptococcal
sepsis: estimation of odds ratios by critical literature review. Pediatrics, 1999. 103(6): p. e77.
84. Fuchs, A., et al., Gentamicin Exposure and Sensorineural Hearing Loss in Preterm Infants. PLoS
One, 2016. 11(7): p. e0158806.

66

Chapter 5: Summary and Future Research Directions
It is evident that unnecessary and inappropriate antibiotic prescribing represents a significant
economic and clinical burden. Unfortunately, influencing behavior is challenging, and the benefits of
doing so are difficult to quantify. This research offers some insight into the economic benefits of
changing behavior and the long-term effects of reducing inappropriate antibiotic use.
Understanding prescriber preferences and how they respond to behavioral economic interventions
can help design future mechanisms that influence prescribing behavior to achieve desired outcomes.
Based on our cost-effectiveness analysis, we believe that any such interventions will represent significant
cost-savings due to large decreases in unnecessary healthcare resource utilization, even if we are unable to
definitively affect antimicrobial resistance. Future work in this area should integrate complex
epidemiological models of antimicrobial evolution and resistance with economic models to
comprehensively quantify the true costs and effects of antimicrobial resistance. Such a model would
serve as an important framework for further research on the economic burden associated with antibiotic
resistance.
Finally, widespread adoption of a clinical guideline that has been recommended and validated by
peer clinicians can be used as an additional tool to influence prescribing behavior that not only achieves
beneficial outcomes for the target neonatal population by reducing the acute risks of extended
hospitalization associated with antibiotic administration, but also represents significant societal cost-
savings by minimizing life-long consequences of hospitalization and maternal separation. While any
adoption of new treatment guidelines takes time, future studies that evaluate the economic impact of
clinical guidelines will provide further evidence and rationale for practice changes.

Abstract (if available)

Abstract BACKGROUND: Inappropriate antibiotic use is a serious public health problem, leading to problems such as antibiotic resistance and unnecessary healthcare utilization. Guideline adherence and behavioral economic interventions can be used to effectively reduce inappropriate antibiotic use. Prescriber preferences and the economic impact of such interventions is unknown. ❧ OBJECTIVES: to describe clinician preferences for behavioral economics interventions and their cost-effectiveness, and to analyze the societal cost-benefit of managing neonates exposed to maternal intrapartum fever according to a newly proposed clinical algorithm that minimizes empiric antibiotic use for all well-appearing infants >34 weeks gestational age. ❧ METHODS: We perform a discrete choice experiment to identify clinician preferences for behavioral economic interventions designed to reduce inappropriate antibiotic prescriptions. We estimate the economic impact of these interventions by performing a cost-effectiveness evaluation of these interventions. Finally, we evaluate the economic impact of changing currently existing guidelines in the management of neonates born to mothers with intrapartum fever to reduce inappropriate antibiotic use. ❧ RESULTS: We find that contrary to the results of a randomized controlled trial, clinicians highly prefer interventions that were not effective in the trial. Further, we find that the interventions are significantly cost-effective

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Advances and applications for economic evaluation methods in health technology assessment (HTA)

PDF

Essays on the economics of infectious diseases

PDF

Selected papers on the evaluation of healthcare costs of prematurity and necrotizing enterocolitis using large retrospective databases

PDF

Economic aspects of obesity

PDF

The impact of treatment decisions and adherence on outcomes in small hereditary disease populations

PDF

Three essays on comparative economic and clinical outcomes of antipsychotics using retrospective claims data

PDF

New approaches using probabilistic graphical models in health economics and outcomes research

PDF

Evaluating cancer treatments with health economics methods

PDF

Physician behavior and low-value care

PDF

Impact of pharmacy-based transitional care on healthcare utilization and costs

PDF

Evaluating treatment options for metastatic, castration-resistant prostate cancer: a comprehensive value assessment

PDF

Discriminating changes in health using patient-reported outcomes

PDF

Methicillin-resistant Staphylococcus aureus: molecular epidemiology, virulence in disease, and antibiotic modulation of virulence

PDF

Household carbon footprints: how to encourage adoption of emissions‐reducing behaviors and technologies

PDF

Delivering better care for children with special health care needs: analyses of patient-centered medical home and types of insurance

PDF

Simulation modeling to evaluate cost-benefit of multi-level screening strategies involving behavioral components to improve compliance: the example of diabetic retinopathy

PDF

An evaluation of risk predictors for disease progression in Veteran Affairs patients with chronic hepatitis C

PDF

Assessing value defects in limb preservation care

PDF

The causal-effect of childhood obesity on asthma in young and adolescent children

PDF

The preclinical evaluation of moxidectin as a platform for drug development for alcohol use disorder

Asset Metadata

Creator Gong, Cynthia Lee (author)

Core Title Evaluating approaches to reduce inappropriate antibiotic use in the United States

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Pharmaceutical Economics and Policy

Publication Date 11/14/2017

Defense Date 10/16/2017

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

Tag antibiotics,cost-benefit,cost-effectiveness,discrete choice,OAI-PMH Harvest,resistance,stewardship

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Hay, Joel W. (committee chair), Doctor, Jason N. (committee member), Zangwill, Kenneth M. (committee member)

Creator Email gongc@usc.edu,thia911@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-455090

Unique identifier UC11264144

Identifier etd-GongCynthi-5901.pdf (filename),usctheses-c40-455090 (legacy record id)

Legacy Identifier etd-GongCynthi-5901.pdf

Dmrecord 455090

Document Type Dissertation

Rights Gong, Cynthia Lee

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 a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA