University of Southern California Dissertations and Theses

Characterization of health outcomes in patients with hemophilia A and B: Findings from psychometric and health economic analyses

(USC Thesis Other)

Characterization of health outcomes in patients with hemophilia A and B: Findings from psychometric and health economic analyses

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1 Characterization of Health Outcomes in Patients with Hemophilia A and B: Findings from Psychometric and Health Economic Analyses By Yuchen Ding 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 (HEALTH ECONOMICS) August 2018 2 Table of Contents DEDICATION ....................................................................................................................................... i ACKNOWLEDGEMENTS ................................................................................................................... ii LIST OF TABLES ................................................................................................................................ iv LIST OF FIGURES ............................................................................................................................... v ABSTRACT ......................................................................................................................................... vi CHAPTER 1. Introduction ............................................................................................................... 1 1.1 Overview of hemophilia ...................................................................................................................................................... 1 1.2 Characterization of health outcomes ........................................................................................................................... 3 1.3 Hemophilia Utilization Group Studies (HUGS)......................................................................................................... 5 1.4 Overview of three essays..................................................................................................................................................... 6 1.5 Chapter references ................................................................................................................................................................ 8 CHAPTER 2: Psychometric Analysis of VERITAS-Pro, an Adherence Scale in Hemophilia ...................................................................................................................................... 11 ABSTRACT ....................................................................................................................................................................................... 11 2.1 Introduction ........................................................................................................................................................................... 12 2.2 Materials and methods...................................................................................................................................................... 14 2.2.1Data source ................................................................................................................................................................... 14 2.2.2 VERITAS-Pro ............................................................................................................................................................... 15 2.2.3 Statistical analyses ................................................................................................................................................... 16 2.3 Results ....................................................................................................................................................................................... 18 2.3.1 Descriptive analyses ............................................................................................................................................... 18 2.3.2 Psychometric properties of VERITAS-Pro ................................................................................................... 19 2.3.3 Modified VERITAS-Pro ........................................................................................................................................... 20 2.3.4 Interval-scaled scoring of modified VERITAS-Pro................................................................................... 21 2.4 Discussion and conclusion ............................................................................................................................................... 21 2.5 Chapter references .............................................................................................................................................................. 24 CHAPTER 3: Concurrent and Predictive validities for VERITAS-Pro and mVERITAS- Pro: Validation Using Prescription and Dispensing Records ............................................. 33 ABSTRACT ....................................................................................................................................................................................... 33 3.1 Introduction ........................................................................................................................................................................... 35 3.2 Materials and methods...................................................................................................................................................... 37 3.2.1 Study design and population .............................................................................................................................. 37 3.2.2 VERITAS-Pro and mVERITAS-Pro .................................................................................................................... 38 3 3.2.3 Criterion measures .................................................................................................................................................. 38 3.2.4 Statistical analysis .................................................................................................................................................... 39 3.3 Results ....................................................................................................................................................................................... 41 3.4 Discussion and conclusion ............................................................................................................................................... 43 3.5 Chapter references .............................................................................................................................................................. 46 CHAPTER 4: Transition from childhood to adulthood and health outcomes in persons with hemophilia A: evidence from longitudinal analyses in the USA .............................. 54 ABSTRACT ....................................................................................................................................................................................... 54 4.1 Introduction ........................................................................................................................................................................... 55 4.2 Materials and methods...................................................................................................................................................... 57 4.2.1 Data source .................................................................................................................................................................. 57 4.2.2 Health outcomes ....................................................................................................................................................... 58 4.2.3 Study groups ............................................................................................................................................................... 60 4.2.4 Statistical analyses ................................................................................................................................................... 60 4.3 Results ....................................................................................................................................................................................... 62 4.3.1 Sample characteristics ........................................................................................................................................... 62 4.3.2 Temporal trends in health outcomes ............................................................................................................. 62 4.3.3 Multivariable regression analyses ................................................................................................................... 63 4.4 Discussion and conclusion ............................................................................................................................................... 65 4.5 Chapter references .............................................................................................................................................................. 68 CHAPTER 5. Conclusion ................................................................................................................ 79 APPENDIX ......................................................................................................................................... 81 Appendix 1. Modified VERITAS-Pro ..................................................................................................................................... 81 Appendix 2. Supplemental materials for Chapter 4 .................................................................................................... 83 i DEDICATION To my dearest parents and grandma, for your unconditional love and support, and motivating me to go the extra mile; To Peng, for your endless love, trust, and encouragement. ii ACKNOWLEDGEMENTS First and foremost, I would like to express my sincere gratitude and appreciation to Dr. Michael B. Nichol, my PhD advisor and committee chair, for giving me tremendous guidance on research and professional development over the past two years. His mentorship has inspired me to be perseverant and self-driven, without which my dissertation would not be possible. I would also like to thank my dissertation committee, Dr. Jason Doctor, Dr. Rebecca Myerson, Dr. Shinyi Wu, and Dr. John Romley for their invaluable and generous inputs and time spent to make my dissertation a better piece. I am sincerely grateful to the Hemophilia Utilization Group Studies (HUGS) team for offering an amazing opportunity to hone my technical and interpersonal skills which would solve real-world problems in the research of hemophilia. Thank you my “HUGS family” – Mimi, Joanne, Randy, Megan, Brenda, and Judith! I am also deeply grateful to Bioverativ Therapeutics, who funded my two-year pre- doctoral fellowship and offered me an unique opportunity work onsite with cross- functional experts in hemophilia. Thank you Sangeeta, Jun, Nick, Jing, Desilu, Elisa, Kun, Nga, Mathura, and Donna, I would not be able to have such a great journey without you. iii I am deeply indebted to the faculty at the Department of Pharmaceutical and Health Economics at USC for offering me the invaluable opportunity to pursue a PhD in health economics. My dissertation benefited a lot from the strong technical foundation laid in the graduate program. Special thanks to Dr. Jeff McCombs, the “captain” of our program, who has always been encouraging and inspiring. Finally, my PhD journey would have been less enjoyable without the accompany of my friends and fellow classmates. Thank you Xiayu, Bo, Cynthia, Tianyi, Xue, Jianhui, Yingying, Chia-Wei, Christina, and Emmanuel, our memories at USC will always be cherished. iv LIST OF TABLES Table 2.1 Sample Characteristics ....................................................................................................................... 27 Table 2.2 Distribution of Responses to VERITAS-Pro .............................................................................. 28 Table 2.3 Summary of Modifications for VERITAS-Pro ........................................................................... 29 Table 2.4 Category Structure for modified VERITAS-Pro ....................................................................... 30 Table 3.1 Sample characteristics at baseline ................................................................................................ 49 Table 3.2 Rank-biserial correlations with Criterion Measure 1 .......................................................... 50 Table 3.3 Rank-biserial correlations with Criterion Measure 2 .......................................................... 51 Table 3.4 Phi coefficients of correlation between VERITAS-Pro and Criterion Measure 1 .... 52 Table 3.5 Phi coefficients of correlation between VERITAS-Pro and Criterion Measure 2 .... 53 Table 4.1 Sample characteristics by age group at HUGS Va initial survey ...................................... 70 Table 4.2 Sample characteristics by age group at HUGS LTS initial survey ................................... 71 Table 4.3 Random effect negative binomial models for association between six-month bleeding rate and age .............................................................................................................................................. 72 v LIST OF FIGURES Figure 2.1 Item-person Map for VERITAS-Pro............................................................................................. 31 Figure 2.2 Item-person Map for modified VERITAS-Pro ........................................................................ 32 Figure 4.1b. Temporal trend of six-month bleeding rate (6moBR) in on-demand patients .. 74 Figure 4.2a. Temporal trend of six-month factor utilization in prophylactic patients ............. 75 Figure 4.2b. Temporal trend of six-month factor utilization in on-demand patients ............... 76 Figure 4.3 Temporal trend of the proportion of patients adhering to prophylaxis ................... 77 Figure 4.4 Temporal trend of the proportion of patients on prophylaxis ...................................... 78 vi ABSTRACT Hemophilia is a hereditary blood disorder characterized by deficiencies in coagulation factors VIII (hemophilia A) and IX (hemophilia B), which imposed significant disease burden on patients, caregivers, and the society. Bleeding is the most important clinical manifestation of hemophilia. Repeated bleeding into the joints leads to irreversible joint damage, impaired health outcomes, and increased health resource utilization. Over the past decades, hemophilia has been treated with factor replacement therapies which replenish the deficient coagulation factors through intravenous infusions on either prophylactic (or prophylaxis) or episodic (or on-demand) regimen. Prophylaxis administers factors regularly to prevent anticipated bleeding; on-demand regimen administers factors “as-needed”, either after bleeding occurs or as extra doses to prevent breakthrough bleeding occurs at special events (e.g. sports). Prophylaxis has been recommend by the World Federation of Hemophilia as the optimal regimen to preserve joint function and increasingly practiced worldwide especially in children who would benefit most from early initiation of prophylaxis. Despite tremendous benefits associated with prophylaxis, the high costs and time-consuming nature of it may significantly reduce adherence, which is critical for successful management of hemophilia. Prior studies measure adherence using highly variable methods, underscoring the lack of gold standard measures of adherence in hemophilia research. One of the most widely used subjective measures is the Validated Hemophilia Regimen Treatment Adherence Scales – Prophylaxis (VERITAS-Pro), which is also the only standardized measure of adherence to prophylaxis. VERITAS-Pro is a patient-/parent-reported questionnaire which was developed in 2010 by researchers in the Indiana Hemophilia Treatment Center (IHTC) in the USA. The measure consists of 24 questions under 6 domains, and has been widely used and translated into more than 30 languages. However, VERITAS-Pro have not been fully examined for psychometric properties vii (i.e. construct validity, predictive validity, rating scale performance, and discriminative ability), which hindered its further applications and raised concerns about the credibility of studies using this measure. Another emerging issue in hemophilia research is the lack of studies examining health outcomes in adolescents and young adults with hemophilia, who are in transition from childhood to adulthood and experiencing composite clinical and psychosocial challenges not experienced by patients in other age groups. Adherence to prophylaxis was found to be the lowest among adolescents and young adults, however it is still unclear how much of an impact it will have on other important health outcomes such as bleeding and factor utilization. Although national policy makers have identified transition as a priority issue for patients with blood disorders, no studies have ever quantified the impact of transition from childhood to adulthood on health outcomes in hemophilia patients. My dissertation consists of three studies characterizing health outcomes in persons with hemophilia A and B, using data collected in the Hemophilia Utilization Group Studies (HUGS) in the USA. The first study (Chapter 2) assessed construct validity, rating scale performance, and discriminative ability for VERITAS-Pro and proposed a 18-item modified version (mVERITAS- Pro) with improved psychometric properties based on Rasch analysis. The second study (Chapter 3) found weak concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro with criterion measures calculated using factor prescription and dispensing records. The third study (Chapter 4) showed that transition from childhood to adulthood was associated with increased bleeding compared to childhood, and increased factor utilization compared to childhood and adulthood using a longitudinal analysis of hemophilia A patients. viii Altogether, findings from these studies provided important psychometric and health economic evidence for the characterization of health outcomes in patients with hemophilia A and B. More importantly, they shed light on areas for improvement regarding the measurement of adherence and unmet need in vulnerable hemophilia populations who are in transition from childhood to adulthood. Furthermore, methods developed and used in these studies can be applied in future research of hemophilia as well as other chronic conditions. 1 CHAPTER 1. Introduction 1.1 Overview of hemophilia Hemophilia is a congenital blood disorder inherited in an X-linked recessive pattern, thus primarily affecting males. There are two major types of hemophilia, characterized by a deficiency of coagulation factor VIII (hemophilia A) or factor IX (hemophilia B). Hemophilia A is five times more common than hemophilia B (also known as “Christmas disease”), with an incidence of 1 in 5,000 live male births versus 1 in 30,000 live male births.[1,2] There are approximately 20,000 hemophilia A and B patients in the USA and 400,000 worldwide.[3,4] Severity of disease is defined by the amount of coagulation factors in blood, measured as the number of international units (IU) of factor per milliliter (mL), or as the percentage of normal factor activity, which is 0.05-0.40 IU mL -1 (or 5 to <40% of normal) for mild patients, 0.01-0.05 IU mL -1 (or 1-5% of normal) for moderate patients, and <0.01 IU mL -1 (or <1% of normal) for severe patients.[4] As the most important clinical symptom, bleeding more often occurs in target joints (knees, elbows, and ankles) and muscles than in digestive organs, neck/throat, and brain. Bleeding is categorized into spontaneous or traumatic, with spontaneous bleeding occurs without any obvious physical trauma. Hemophilia patients experience frequent spontaneous bleeding which lead to joint and muscle pain, joint deformity, and ultimately may lead to joint arthropathy and muscle atrophy, all of which will adversely impact patients’ well-being and incur substantial health care resource utilization and costs. Hemophilia is treated with intravenous infusion of the deficient coagulation factor concentrates regularly (prophylactic regimen or prophylaxis) to prevent anticipated bleeding, or as-need after bleeding occurs or as extra doses to prevent anticipated bleeding for special events such as sports 2 (episodic or “on-demand” regimen). Prophylaxis has been recommended as the optimal regimen by the World Federation of Hemophilia, based on mounting evidence of benefits associated with bleeding and joint function.[4] Although cost-effective in the long run, high costs associated with prophylaxis have so far limited its applications to countries with ample health care resources. In developed countries, approximately 80% of children and 42% of adults are treated with prophylaxis.[5,6] Although a rare disease, hemophilia is associated with disproportionately high costs which have posed significant financial burden on patients, payers, and the society.[7] Implications for hemophilia are not limited to financial burden, but also profound psychosocial impact throughout a patient’s life. Hemophilia patients face varying psychosocial challenges at different stages of life from childhood, adolescence, to adulthood. Studies have shown that children with hemophilia had higher rates of absenteeism resulting in lower test scores and academic achievement compared to healthy children.[8] Parents for hemophilia children also suffered from substantial emotional stress, missed work time, reduced productivity, and financial burden.[9] Adults with hemophilia experience difficulties with employment and interpersonal relationships and comorbidities with aging. However, probably the most challenging life stage for hemophilia patients is adolescence. The transition from childhood to adulthood is a time when people experience tremendous developmental and psychological challenges, such as teenage rebellion, risk-seeking behaviors, independence and separation, and adapting to adulthood.[10] Compared to healthy adolescents, these challenges aggravate in adolescents with hemophilia who start to manage hemophilia on their own with diminishing influence from parents.[11] Several challenges have been identified for patients in transition from childhood to adulthood in prior studies. First, the need to switch from pediatric care providers with whom they have long and trusted relationships to adult care providers they are unfamiliar with, and from a pediatric 3 hemophilia treatment center (HTC) setting with more parental involvement to an adult HTC setting requesting proactive patient engagement.[12] Second, unawareness of the benefits associated with prophylaxis a result of extensive prophylaxis during childhood during which bleeding rarely occurs.[12] Lastly, the lack of psychological maturity in taking over the responsibilities of care, financial and employment barriers associated with hemophilia, and the pressing need to secure health insurance coverage by the age of 26, an issue specific to patients in the USA.[12,13] In summary, hemophilia is chronic debilitating disorder posing significant clinical, psychosocial, and financial burden on patients, caregivers, and the society. Next section highlights issues around characterization of health outcomes in patients with hemophilia. 1.2 Characterization of health outcomes Hemophilia treatment requires high level of motivation and commitment by patients and their caregivers. This is particularly true for patients receiving prophylaxis which requires frequent intravenous infusions of factor concentrates. However, adherence may have been compromised due to the high costs and time-consuming nature of prophylaxis, which may result in increased bleeding leads to irreversible joint damage. Geraghty et al. found that 41% of children and 83-94% of adults only achieved low-to-moderate adherence in a global survey of hemophilia A patients.[14] Du Treil et al. identified 74% of children and 61% of adults were low-to-moderate adherent in a retrospective analysis of patients with hemophilia A and B.[15] In addition to high costs and the time-consuming nature of treatment, other barriers for adherence to prophylaxis were identified as pain and discomfort associated with intravenous injections and emotional stress.[16,17] Indeed, studies have shown that nonadherence and discontinuation to prophylaxis resulted in a higher 4 number of bleeding episodes, development and worsening of target joints, and reduced health- related quality of life (HRQoL) in both children and adult patients.[5,6] Although there is an urgent need to measure adherence using valid and reliable measures, prior adherence studies have employed highly variable methods resulting in tremendous variations across studies. Measures included standardized instruments such as the Validated Hemophilia Regimen Treatment Adherence Scales and modified Morisky adherence scale, and non- standardized instruments such as patient- or provider-reported surveys, infusion logs, dispensing records, and insurance claims.[14–16,18–26] As a result, the percentage of patients adherent to treatment ranged from 6% to 93% depending on the method used. Each method has its own limitations. Non-standardized surveys cannot be assessed for psychometric properties; infusion logs are subject to selection bias since patients keeping the infusion logs may be more adherent in nature; dispensing records reflect the amount of factor dispensed but not the amount utilized since patients may save factors for future use; insurance claims are developed for administrative billing purposes, thus may not capture all variables necessary for defining adherence; standardized surveys, as all subjective measures, are subject to recall and reporting bias. However, standardized surveys are easier to administer and can be assessed for psychometric properties compared to the other methods. VERITAS measures were developed by Duncan et al. from IHTC to assess adherence to prophylaxis (VERITAS-Pro) and on-demand (VERITAS-PRN) factor replacement therapies in persons with hemophilia A or B.[26,27] VERITAS-Pro has been one of the most widely used instruments in clinical trials and observational studies and translated into more than 30 languages since its introduction.[28] However, psychometric properties for VERITAS-Pro have not been fully examined in prior 5 validation study.[26] Specifically, important psychometric properties such as construct validity, predictive validity, rating scale performance, and discriminative ability have never been examined. Health outcome studies in hemophilia have largely focused on children and adults, but not those in transition from childhood to adulthood, although transition has been identified as a priority focus for persons with bleeding disorders by the US Health Resource and Services Administration (HRSA).[29] Adolescents with hemophilia have to deal with unique clinical and psychosocial challenges leading to higher risk of nonadherence and subsequently reduced physical and mental well-being. Prior studies found that adherence to prophylaxis was much lower in adolescents (aged>12 years) compared to young children (aged≤12 years).[14,30–32] In the USA, approximately 55% of hemophilia patients are young people and many of whom are in transition from childhood to adulthood.[33] Although there is an urgent need to study this population, very few studies exist and none of them has quantified the impact of transition from childhood to adulthood on important health outcomes such as bleeding, factor utilization, and adherence. 1.3 Hemophilia Utilization Group Studies (HUGS) Decades ago, the Hemophilia Treatment Centers (HTCs) were founded to provide comprehensive and coordinated hemophilia care encompassing diagnosis, treatment, prevention, education, outreach, and surveillance services to patients in geographical diverse regions.[34] This integrated care model enabled hemophilia patients to build a long and trusted relationship with their local HTC providers at an early age, and also the HTCs to serve as a research and data collection channel to track patients longitudinally. 6 The Hemophilia Utilization Group Studies (HUGS) were created in California in the 1990s to evaluate the health services utilization and costs of hemophilia, and gradually expanded to 12 HTCs spanning across eleven geographically diverse states in the USA.[7] HUGS provided rich datasets collecting key clinical, treatment, and socio-demographic data for health economics and outcomes research. More importantly, HUGS enrolled hemophilia patients from geographically diverse HTCs which ensured a relatively large sample size for this rare disease and more importantly good external validity of study findings. My dissertation will utilize data from three HUGS projects: HUGS Va, Long-term Follow-up Study (LTS), and HUGS VI. HUGS Va was one of the largest multi-center study evaluating cost and burden of illness in hemophilia A from July 2005 and July 2007. In 2014, LTS was initiated to gather long-term data on health outcomes, utilization, and costs from a subset of persons originally participated in HUGS Va. HUGS VI is an ongoing observational study evaluating adherence to factor replacement therapies in real-world setting. 1.4 Overview of three essays The first essay (Chapter 2) will assess construct validity, rating scale performance, and discriminative ability for VERITAS-Pro using interim baseline data collected in HUGS VI as of July 2017. The second essay (Chapter 3) will evaluate concurrent and predictive validities for VERITAS-Pro against objective criterion measure computed using prescription and dispensing records based on an interim analysis of HUGS VI data collected as of Apr 2018. The third essay (Chapter 4) will characterize temporal trends in health outcomes and the impact of transition on bleeding in hemophilia A patients who were in transition from childhood to adulthood. Taken together, three essays will provide additional evidence as well as shed light on areas for 7 improvement regarding characterization of health outcomes in patients with hemophilia A and B using psychometric and health economics analysis. 8 1.5 Chapter references 1 Centers for Disease Control and Prevention. Hemophilia Data & Statistics. 2 Konkle BA. Hemophilia B. GeneReviews(®). Seattle (WA): University of Washington, Seattle; 1993. 3 Soucie JM, Evatt B, Jackson D. Occurrence of hemophilia in the United States. Am J Hematol 1998; 59: 288–94. 4 Srivastava A, Brewer AK, Mauser-Bunschoten EP, Key NS, Kitchen S, Llinas A, et al. Guidelines for the management of hemophilia. Haemophilia 2013; 19: e1–47. 5 Manco-Johnson MJ, Sanders J, Ewing N, Rodriguez N, Tarantino M, Humphries T, et al. Consequences of switching from prophylactic treatment to on-demand treatment in late teens and early adults with severe haemophilia A: the TEEN/TWEN study. Haemophilia 2013; 19: 727–35. 6 Walsh CE, Valentino LA. Factor VIII prophylaxis for adult patients with severe haemophilia A: results of a US survey of attitudes and practices. Haemophilia 2009; 15: 1014–21. 7 Globe DR, Curtis RG, Koerper MA, For the HUGS Steering Committee. Utilization of care in haemophilia: a resource-based method for cost analysis from the Haemophilia Utilization Group Study (HUGS). Haemophilia 2004; 10: 63–70. 8 Shapiro AD, Donfield SM, Lynn HS, Cool VA, Stehbens JA, Hunsberger SL, et al. Defining the Impact of Hemophilia: The Academic Achievement in Children With Hemophilia Study. Pediatrics 2001; 108: e105–e105. 9 DeKoven M, Karkare S, Kelley LA, Cooper DL, Pham H, Powers J, et al. Understanding the experience of caring for children with haemophilia: cross-sectional study of caregivers in the United States. Haemophilia 2014; 20: 541–9. 10 Sanders RA. Adolescent Psychosocial, Social, and Cognitive Development. Pediatr Rev 2013; 34: 354–9. 11 Lindvall K, Colstrup L, Wollter I-M, Klemenz G, Loogna K, Grönhaug S, et al. Compliance with treatment and understanding of own disease in patients with severe and moderate haemophilia. Haemophilia 2006; 12: 47–51. 12 Quon D, Reding M, Guelcher C, Peltier S, Witkop M, Cutter S, et al. Unmet needs in the transition to adulthood: 18- to 30-year-old people with hemophilia. Am J Hematol 2015; 90: S17–22. 9 13 Patient Protection and Affordable Care Act, 42 U.S.C. § 18001 (2010). 14 Geraghty S, Dunkley T, Harrington C, Lindvall K, Maahs J, Sek J. Practice patterns in haemophilia A therapy – global progress towards optimal care. Haemophilia 2006; 12: 75–81. 15 Du Treil S, Rice J, Leissinger CA. Quantifying adherence to treatment and its relationship to quality of life in a well-characterized haemophilia population. Haemophilia 2007; 13: 493–501. 16 Hacker MR, Geraghty S, Manco-Johnson M. Barriers to compliance with prophylaxis therapy in haemophilia. Haemophilia 2001; 7: 392–6. 17 Carcao MD, Aledort L. Prophylactic factor replacement in hemophilia. Blood Rev 2004; 18: 101–13. 18 De Moerloose P, Urbancik W, Van Den Berg HM, Richards M. A survey of adherence to haemophilia therapy in six European countries: results and recommendations. Haemophilia 2008; 14: 931–8. 19 Llewellyn CD, Miners AH, Lee CA, Harrington C, Weinman J. The Illness Perceptions and Treatment Beliefs of Individuals with Severe Haemophilia and their Role in Adherence to Home Treatment. Psychol Health 2003; 18: 185–200. 20 Thornburg CD, Pipe SW. Adherence to prophylactic infusions of factor VIII or factor IX for haemophilia. 2006; . 21 Thornburg CD. Physicians’ perceptions of adherence to prophylactic clotting factor infusions. Haemophilia 2008; 14: 25–9. 22 Ho S, Gue D, McIntosh K, Bucevska M, Yang M, Jackson S. An objective method for assessing adherence to prophylaxis in adults with severe haemophilia. Haemophilia 2014; 20: 39–43. 23 Chen CX. The impact of treatment decisions and adherence on outcomes in small hereditary disease populations (Doctoral dissertation). Retrieved from the University of Southern California Digital Library Database. 2016. 24 Armstrong EP, Malone DC, Krishnan S, Wessler MJ. Adherence to clotting factors among persons with hemophilia A or B. Hematology 2015; 20: 148–53. 25 Tencer T, Roberson C, Duncan N, Johnson K, Shapiro A. A haemophilia treatment centre-administered disease management programme in patients with bleeding disorders. Haemophilia 2007; 13: 480–8. 10 26 Duncan N, Kronenberger W, Roberson C, Shapiro A. VERITAS-Pro: a new measure of adherence to prophylactic regimens in haemophilia. Haemophilia 2010; 16: 247–55. 27 Duncan NA, Kronenberger WG, Roberson CP, Shapiro AD. VERITAS-PRN: a new measure of adherence to episodic treatment regimens in haemophilia. Haemophilia 2010; 16: 47–53. 28 Thornburg CD, Duncan NA. Treatment adherence in hemophilia. Patient Prefer Adherence 2017; 11: 1677–86. 29 The American Thrombosis and Hemostasis Network (ATHN). Action Guide for Improving Care for People with Bleeding Disorders. 2016. 30 van Dijk K, Fischer K, van der Bom JG, Scheibel E, Ingerslev J, van den Berg HM. Can long-term prophylaxis for severe haemophilia be stopped in adulthood? Results from Denmark and the Netherlands. Br J Haematol 2005; 130: 107–12. 31 Duncan N, Shapiro A, Ye X, Epstein J, Luo MP. Treatment patterns, health-related quality of life and adherence to prophylaxis among haemophilia A patients in the United States. Haemophilia 2012; 18: 760–5. 32 Fischer K, Van Der Bom JG, Prejs R, Mauser-Bunschoten EP, Roosendaal G, Grobbee DE, et al. Discontinuation of prophylactic therapy in severe haemophilia: incidence and effects on outcome. Haemophilia 2001; 7: 544–50. 33 Nazzaro A-M, Owens S, Hoots WK, Larson KL. Knowledge, Attitudes, and Behaviors of Youths in the US Hemophilia Population: Results of a National Survey. Am J Public Health 2006; 96: 1618–22. 34 Baker JR, Riske B, Drake JH, Forsberg AD, Atwood R, Voutsis M, et al. US Hemophilia Treatment Center population trends 1990–2010: patient diagnoses, demographics, health services utilization. Haemophilia 2013; 19: 21–6. 11 CHAPTER 2: Psychometric Analysis of VERITAS-Pro, an Adherence Scale in Hemophilia ABSTRACT Introduction The Validated Hemophilia Regimen Treatment Adherence Scale - Prophylaxis (VERITAS- Pro) was developed to measure adherence to prophylactic factor replacement therapies in hemophilia. Despite wide applications, psychometric properties of VERITAS-Pro have not been fully assessed. Aim To examine construct validity, rating scale performance, and discriminative ability of VERITAS-Pro and propose a modified version using Rasch analysis. Methods Rasch polytomous rating scale model was employed to analyze VERITAS-Pro collected in the initial surveys of the Hemophilia Utilization Group Studies Part VI as of July 2017. Results Using VERITAS-Pro data completed by 118 prophylactic persons with hemophilia A or B from eight hemophilia treatment centers in the USA, several improvements have been identified for construct validity, rating scale performance, and discriminative ability. Six items misfit the Rasch model and had high factor loadings on the principal component of 12 residuals, indicating impaired construct validity. Disordered step thresholds for middle response categories ("Rarely" and "Sometimes") indicated insufficient rating scale performance. A large difference (1.31 logits) in mean person and item measures indicated a lack of discriminative ability. The modified version (mVERITAS-Pro) removed six items impairing construct validity and collapsed responses into a 3-point scale, and demonstrated improved psychometric properties. Conclusion Areas for improvement regarding psychometric properties have been identified for VERITAS-Pro. Future research should assess concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro using criterion measures such as prescription and dispensing records. 2.1 Introduction Hemophilia is a rare congenital blood disorder characterized by deficiencies in blood coagulation factors VIII (hemophilia A) and IX (hemophilia B). Over the past decades, hemophilia is treated with intravenous infusions of the deficient factor concentrates. Persons with hemophilia (PWH) receive infusions “as-needed” (episodic or on-demand regimen) or multiple times a week to prevent anticipated bleeding (prophylactic regimen or prophylaxis).[1] Prophylaxis has been recommended as the optimal regimen to prevent joint damage and preserve joint function by the World Federation of Hemophilia and increasingly practiced worldwide in recent years.[1] 13 Adherence to prophylaxis is critical to successful management of hemophilia, since non- adherent patients had increased bleeding, development and worsening of target joints, and reduced health-related quality of life (HRQoL).[2,3] However, the high costs and time- consuming nature of prophylaxis may lead to nonadherence.[4,5] Adherence in PWH has been measured using highly variable methods, including patient- or provider-reported surveys, the modified Morisky adherence scale, patient-reported infusion logs, dispensing records, and insurance claims.[4,6–15] As a result, tremendous variations of adherence were reported in prior studies, with the percentage of patients adherent to treatment ranging from 6% to 93% depending on the method used.[4,6–12,15] In 2010, Duncan et al. developed and validated the first standardized measure of adherence to prophylaxis, i.e. the Validated Hemophilia Regimen Treatment Adherence Scale – Prophylaxis (VERITAS-Pro). VERITAS-Pro has been widely used and translated into more than 30 languages ever since its introduction.[16,17] Although VERITAS-Pro plays an important role in the measurement of adherence in hemophilia, its psychometric properties have not been fully examined. Prior validation study of VERITAS-Pro did not assess construct validity and predictive validity, which justify the validity of a psychometric instrument such as VERITAS-Pro along with content validity and concurrent validity.[18] Specifically, construct validity determines the extent to which VERITAS-Pro measures adherence as it intends to measure, and predictive validity determines the ability of VERITAS-Pro to predict real-world adherence in the future.[16] In addition, rating scale performance and discriminative ability for VERITAS-Pro were found to be unsatisfactory in prior adherence studies.[16,19] 14 Rasch analysis provides the best approach to assessing these important psychometric properties. Rasch model has deep roots in measurement theory, and assumes that the probability of a respondent (or “person”) affirming a question (or “item”) is determined by the distance between difficulty of the item (e.g. level of adherence required) and ability of the person (e.g. level of adherence presented) measured on a shared linear scale of the construct (e.g. adherence).[20] The probability is described using logistic functions and the unit of measurement is log odds (or “logit”).[21] Compared to Classical Test Theory methods, Rasch analysis produces sample-independent results and has the ability to transform raw scores from ordinal responses into interval-scaled scores which conform to measurement theory. [22–25] This study was conducted with two objectives. First, to assess the construct validity, rating scale performance, and discriminative ability for VERITAS-Pro using Rasch analysis. Second, to propose a modified version (mVERITAS-Pro) with improved psychometric properties and an algorithm to convert raw scores into interval-scaled scores. 2.2 Materials and methods 2.2.1Data source This study utilized VERITAS-Pro data collected at initial surveys in the Hemophilia Utilization Group Studies Part VI (HUGS VI) as of July 2017. HUGS VI is an ongoing multi- center observational study assessing adherence to factor replacement therapies in persons with hemophilia A and B. HUGS VI enrolled patients aged≥6 years and receiving the 15 majority of hemophilia care at one of the eight HTCs in different regions of the USA. VERITAS-Pro was completed four time, once at initial survey and three time at follow-up surveys administered once every 3 months during a period of 9 months. A total of 127 prophylactic patients completed VERITAS-Pro at initial surveys as of July 2017. We included 118 prophylactic patients who had complete VERITAS-Pro data, since missing imputation is not recommended for Rasch analysis.[26] 2.2.2 VERITAS-Pro VERITAS-Pro was jointly developed by a panel of hemophilia patients and providers in the Indiana HTC in 2010. Details of the development procedures were described elsewhere.[16] VERITAS-Pro has 24 items and 6 domains (Time, Dose, Plan, Remember, Skip, and Communicate). Eleven items are negatively worded (Q6, 7, 11, 13, 14, 16-20, 23) to describe non-adherent activities and thirteen items are positively worded to describe adherent activities.[16] Responses are rated on a 5-point rating scale (1-Always, 2-Often, 3- Sometimes, 4-Rarely, and 5-Never), with higher scores indicating lower adherence. In line with Rasch recommendations and for the ease of interpretation, individual scores were reversed for negatively worded items then all items were reversely scored so that higher scores indicated better adherence. 16 2.2.3 Statistical analyses Descriptive analyses were conducted to compare sample characteristics between our study and the prior validation study.[16] Data distributions were analyzed and non-normality (skewness or kurtosis >2.0 or <-2.0) as well as ceiling and floor effects (>20% of the response clustered at the highest and lowest end of the rating scale) were identified.[27–29] We chose the Rasch-Andrich polytomous rating scale model since all items were rated on the same polytomous response scale.[30] First, global model fit was assessed using separation and reliability statistics. Separation is the number of distinct strata of persons and items discernable by corresponding items and persons, with good separation indicated by a person separation statistic of ≥2 (3 strata) and an item separation statistic of ≥3 (4 strata).[31] Reliability is indicated by separation indexes, and can be interpreted in a way similar to Cronbach's α, with good reliability indicated by a person separation index of ≥0.8 and an item separation index of ≥0.9.[31] Second, construct validity was evaluated based on mean- square outfit and infit statistics. Outfit statistic is the sum of squared standardized residuals; infit statistic is a weighted sum of squared residuals, with weight being the variance of squared residuals.[32] Compared to infit statistic, outfit statistic is more sensitive to outliers. Item misfit was identified by infit or outfit statistics falling outside the range of 0.6-1.4.[33] In addition, principal component analysis (PCA) of the residuals was performed to detect violations of construct validity. Unidimensionality is violated if items have high factor loadings (>0.4 or <-0.4) on the principal component in residuals, the principal component has an eigenvalue ≥3.0, and there are person measures outside the 95% confidence interval (CI) on the cluster measure plot.[34] 17 Moreover, rating scale performance was evaluated using Rasch step thresholds which represent the point at which the probabilities of choosing two adjacent categories are equal. Step thresholds are disordered if they do not advance monotonically with increasing response categories, which often occurs due to redundancies in rating scale or a lack of clarity in the description of response categories.[35] Discriminative ability was assessed using the Rasch item-person map which plots items and persons along a shared linear scale representing adherence. Good discriminative ability is indicated by normally distributed and evenly spread-out items and persons and equal item difficulty and person ability measures on the item-person map.[36] Modifications on VERITAS-Pro were made as follows. First, items impairing construct validity as detected by infit and outfit statistics and the PCA of residuals were removed. Second, categories with disordered step thresholds were collapsed with adjacent categories to have a new rating scale which satisfies the following: step difficulties within 1.4-5.0, outfit statistic <2.0, ≥10 observations in each response category, and improved separation and reliability indexes and response category structure.[35] Specifically, response category structure was assessed by infit and outfit statistics, step thresholds, category probability curves depicting the probability of observing each response category, and coherence representing the relationship between ratings and measures characterized by the percentage of ratings actually observed in the category as expected according to the measures (MàC) and the percentage of occurrences of a category placed by the measures in that category (CàM). As a rule of thumb, MàC or CàM ≥40% indicates good coherence.[35] Next, discriminative ability was compared between the original and modified VERITAS-Pro in 18 terms of person-item distributions. Lastly, nonlinear regression analyses were performed to generate an algorithm to convert raw ordinal scores to interval-scaled scores which conform to measurement theory, assuming a double-asymptotic non-linear relationship between average ordinal rating and person measure.[37] Descriptive analyses and nonlinear regression analyses were performed using SAS® software Version 9.4. Rasch analyses were performed using WINSTEPS® software Version 3.92.[38] 2.3 Results 2.3.1 Descriptive analyses Compared to prior validation study which included 67 patients from a single HTC, our sample was larger (N=118) and geographically diverse. Specifically, our sample has an older average age (25.0 vs. 15.2 years), higher percentages of persons with hemophilia A (89.0% vs. 83.6%), African Americans (26.3% vs. 14.9%), and Asians (8.5% vs. 1.5%), and lower percentage of adult patients and parents for pediatric patients educated at college level or above (70.3% vs. 76.1%) (Table 2.1).[16] Non-normal distribution were identified for six items (Q5, Q6, Q8, Q9, Q10, Q11) (Table 2.2). Significant ceiling effects were present for all items, which may be due to that this sample was very adherent or VERITAS-Pro was too "easy" for them. 19 2.3.2 Psychometric properties of VERITAS-Pro Data fit well to the Rasch-Andrich polytomous rating scale model. Item (2.06) and item (4.08) separation were above the thresholds, as were person (0.81) and item (0.94) reliability. Insufficient construct validity was identified by infit and outfit statistics and PCA of residuals. Ten items had out-of-range outfit or infit statistic, six (Q3, 6, 8, 11, 12, 24) had fit statistics >1.4, and four (Q13, 16, 17, 20) had fit statistics <0.6. Five of the ten misfitting items were reversely worded (Q6, 11, 13, 16, 17, 20). PCA of residuals found violations of unidimensionality, as indicated by six items with high factor loadings (Q3, 6, 8, 11, 12, 24), a large eigenvalue (3.3) for the principal component of residuals, and many person measures lying outside the 95% CI on the cluster measure plot. Rating scale performance was also unsatisfactory, as indicated by disordered step thresholds (-.50 and -.56, respectively) between “Rarely” (Category 2 after reverse scoring) and “Sometimes” (Category 3 after reverse scoring), and problematic response category structure. Specifically, distribution of responses was highly skewed, with highest counts in “Always” (Category 5 after reverse scoring), outfit statistic >2.0 for “Never” (Category 1 after reverse scoring), coherence (MàC) was borderline (43%) sufficient for “Sometimes” and “Often” (Categories 3 and 4 after reverse scoring). Furthermore, discriminative power was insufficient, as indicated by several issues in the item-person map (Figure 2.1). First, persons and items were not normally distributed along the linear scale. Second, the mean person ability was much higher (1.31 logits) than the mean item difficulty. Lastly, many highly adherent respondents did not correspond to any item, and eight items of low difficulties did not correspond to any person on the map. 20 2.3.3 Modified VERITAS-Pro Based on results above, several modified versions were generated based on item reduction and response category re-calibration, and the final modified version was selected based on global model fit (Error! Reference source not found.). The first version increased i tem separation (4.08 to 4.22) and reliability (0.94 to 0.95) by removing six items (Q3, 6, 8, 11, 12, 24), with some loss in person separation (2.06 to 1.95) and reliability (0.81 to 0.79). The second version removed four more items (Q13, 16, 17, 20), with further increase in item separation (4.22 to 4.42) at the expense of greater loss in person separation (1.95 to 1.64) and reliability (0.79 to 0.73). As a result, the first version was chosen for response re-calibration since item separation and reliability were improved without significant loss in person separation and reliability, and also based on conservative considerations that the six misfitting items removed were confirmed by fit statistics and PCA of residuals altogether. Next, the 5-point rating scale was collapsed into 4, 3, and 2 categories to see which would yield the best response category structure and global model fit (Table 2.3 and 2.4). We found that Version 4 demonstrated significant improvement in response category structure and global model fit, and was chosen as the final mVERITAS-Pro (Appendix 1). This version included 18 items and a 3-point rating scale, with the most difficult item being Q23 (Make treatment decision myself rather than calling the hemophilia center), and the easiest item being Q5 (Use doctor-recommended dose for infusion). Discriminative power was significantly improved, since persons and items were normally distributed, and items were evenly spread out along the linear scale (Figure 2.2). Difference between mean person ability and item difficulty decreased significantly from 1.31 to 0.66 logits, however the mean 21 person ability was still higher. Although the number of persons without corresponding items decreased, there were still few persons not corresponding to any item on the item-person map. 2.3.4 Interval-scaled scoring of modified VERITAS-Pro An interval-scaled scoring algorithm was developed to convert raw ordinal scores into interval scores.[37] The scoring procedures are described as follows: first, reversely score positively worded items (Q1-7, Q10, Q16, Q17) so that higher scores indicate better adherence; second, calculate average ordinal rating as the mean item score across all items and convert the calculated average ordinal rating into Rasch person measure using the formula as follows: Person Measure = 1.22 * log(Average Ordinal Rating)/(3- Average Ordinal Rating)) - 0.92 It should be noted that the algorithm should be applied for complete data only since it was developed using complete VERITAS-Pro data. 2.4 Discussion and conclusion VERITAS-Pro is the only standardized measure of adherence to prophylaxis which has played an important role in the measure of adherence in hemophilia. Our study was the first to examine construct validity, rating scale performance, and discriminative ability for VERITAS-Pro, identified several improvements, and proposed a modified version using Rasch analysis. VERITAS-Pro has six items impairing construct validity, redundancies in the 5-point rating scale, and insufficient discriminative ability. The modified version 22 (mVERITAS-Pro) removed six items impairing construct validity and collapsed five response categories into three (1-Always, 2-Sometimes, 3-Never). mVERITAS-Pro demonstrated significant improvement in psychometric properties and has the potential to reduce respondent and administrative burden owing to fewer numbers of items and response categories. Moreover, an interval-scaled scoring algorithm was provided as a tool for users unfamiliar with Rasch analysis to transform raw ordinal scores into interval scores which enables a more accurate interpretation of changes and differences in adherence. using a geographically diverse sample of patients with hemophilia A and B which ensures good external validity of the results.[39] With a larger sample size (118 vs. 67 in prior validation study), our study employed Rasch analysis as a psychometric method with strong theoretical underpinnings in measurement theory which has enabled rigorous psychometric assessments. However, several limitations must be noted. Although the sample size was reasonably large for hemophilia studies, it was relatively small for psychometric analyses. If possible, a larger sample could provide greater statistical power.[35] Second, mean person and item measures was different by 0.66 logits for mVERITAS-Pro, indicating slight misfit between persons and items. Lastly, mVERITAS-Pro was generated from a psychometric perspective, which may have removed items with clinical significance. In conclusion, our study identified areas for improvements for VERITAS-Pro regarding psychometric properties, and proposed a 18-item mVERITAS-Pro with a 3-point rating scale which demonstrated improved construct validity, rating scale performance, and discriminative ability. Future research should examine concurrent and predictive validities 23 for VERITAS-Pro and mVERITAS-Pro against criterion measure computed using prescription and dispensing records, which would lend additional credibility for these subjective measures of adherence. Moreover, how much of an impact mVERITAS-Pro will have on respondent and administrative burden needs to be carefully evaluated in clinical practice. Most importantly, collaborative efforts should be made by clinicians, patients, caregivers, and psychometricians to generate additional items with the ability to discern prophylactic patients with high adherence which could be added to mVERITAS-Pro. 24 2.5 Chapter references 1 Srivastava A, Brewer AK, Mauser-Bunschoten EP, Key NS, Kitchen S, Llinas A, et al. Guidelines for the management of hemophilia. Haemophilia 2013; 19: e1–47. 2 Manco-Johnson MJ, Sanders J, Ewing N, Rodriguez N, Tarantino M, Humphries T, et al. Consequences of switching from prophylactic treatment to on-demand treatment in late teens and early adults with severe haemophilia A: the TEEN/TWEN study. Haemophilia 2013; 19: 727–35. 3 Walsh CE, Valentino LA. Factor VIII prophylaxis for adult patients with severe haemophilia A: results of a US survey of attitudes and practices. Haemophilia 2009; 15: 1014–21. 4 Hacker MR, Geraghty S, Manco-Johnson M. Barriers to compliance with prophylaxis therapy in haemophilia. Haemophilia 2001; 7: 392–6. 5 Carcao MD, Aledort L. Prophylactic factor replacement in hemophilia. Blood Rev 2004; 18: 101–13. 6 Geraghty S, Dunkley T, Harrington C, Lindvall K, Maahs J, Sek J. Practice patterns in haemophilia A therapy – global progress towards optimal care. Haemophilia 2006; 12: 75–81. 7 Du Treil S, Rice J, Leissinger CA. Quantifying adherence to treatment and its relationship to quality of life in a well-characterized haemophilia population. Haemophilia 2007; 13: 493– 501. 8 De Moerloose P, Urbancik W, Van Den Berg HM, Richards M. A survey of adherence to haemophilia therapy in six European countries: results and recommendations. Haemophilia 2008; 14: 931–8. 9 Llewellyn CD, Miners AH, Lee CA, Harrington C, Weinman J. The Illness Perceptions and Treatment Beliefs of Individuals with Severe Haemophilia and their Role in Adherence to Home Treatment. Psychol Health 2003; 18: 185–200. 10 Thornburg CD, Pipe SW. Adherence to prophylactic infusions of factor VIII or factor IX for haemophilia. 2006; . 11 Thornburg CD. Physicians’ perceptions of adherence to prophylactic clotting factor infusions. Haemophilia 2008; 14: 25–9. 12 Ho S, Gue D, McIntosh K, Bucevska M, Yang M, Jackson S. An objective method for assessing adherence to prophylaxis in adults with severe haemophilia. Haemophilia 2014; 20: 39–43. 13 Chen CX. The impact of treatment decisions and adherence on outcomes in small hereditary disease populations (Doctoral dissertation). Retrieved from the University of Southern California Digital Library Database. 2016. 25 14 Armstrong EP, Malone DC, Krishnan S, Wessler MJ. Adherence to clotting factors among persons with hemophilia A or B. Hematology 2015; 20: 148–53. 15 Tencer T, Roberson C, Duncan N, Johnson K, Shapiro A. A haemophilia treatment centre- administered disease management programme in patients with bleeding disorders. Haemophilia 2007; 13: 480–8. 16 Duncan N, Kronenberger W, Roberson C, Shapiro A. VERITAS-Pro: a new measure of adherence to prophylactic regimens in haemophilia. Haemophilia 2010; 16: 247–55. 17 Thornburg CD, Duncan NA. Treatment adherence in hemophilia. Patient Prefer Adherence 2017; 11: 1677–86. 18 Cronbach LJ, Meehl PE. Construct validity in psychological tests. Psychol Bull 1955; 52: 281– 302. 19 Os SB van, Troop NA, Sullivan KR, Hart DP. Adherence to Prophylaxis in Adolescents and Young Adults with Severe Haemophilia: A Quantitative Study with Patients. PLOS ONE 2017; 12: e0169880. 20 Massof RW. An Interval-Scaled Scoring Algorithm for Visual Function Questionnaires. Optom Vis Sci 2007; 84: E689. 21 Tennant A, Conaghan PG. The Rasch measurement model in rheumatology: What is it and why use it? When should it be applied, and what should one look for in a Rasch paper? Arthritis Care Res 2007; 57: 1358–62. 22 Wright BD, Linacre JM. Observations are always ordinal; measurements, however, must be interval. Arch Phys Med Rehabil 1989; 70: 857–60. 23 Massof R. The Measurement of Vision Disability. Optometry & Vision Science 2002; 79: 516– 52. 24 Tennant A, McKenna SP, Hagell P. Application of Rasch Analysis in the Development and Application of Quality of Life Instruments. Value Health 2004; 7: S22–6. 25 Pallant JF, Tennant A. An introduction to the Rasch measurement model: An example using the Hospital Anxiety and Depression Scale (HADS). Br J Clin Psychol 2007; 46: 1–18. 26 Hardouin J-B, Conroy R, Sébille V. Imputation by the mean score should be avoided when validating a Patient Reported Outcomes questionnaire by a Rasch model in presence of informative missing data. BMC Med Res Methodol 2011; 11: 105. 27 Pesudovs K, Garamendi E, Keeves JP, Elliott DB. The Activities of Daily Vision Scale for Cataract Surgery Outcomes: Re-evaluating Validity with Rasch Analysis. Invest Ophthalmol Vis Sci 2003; 44: 2892–9. 26 28 Putten JJMF van der, Hobart JC, Freeman JA, Thompson AJ. Measuring change in disability after inpatient rehabilitation: comparison of the responsiveness of the Barthel Index and the Functional Independence Measure. J Neurol Neurosurg Psychiatry 1999; 66: 480–4. 29 Scientific advisory committee of the Medical Outcomes Trust. Instrument review criteria. Medical Outcomes Trust Bulletin; 1995. 30 Andrich D. A rating formulation for ordered response categories. Psychometrika 1978; 43: 561–73. 31 Duncan PW, Bode RK, Min Lai S, Perera S. Rasch analysis of a new stroke-specific outcome scale: the stroke impact scale. Arch Phys Med Rehabil 2003; 84: 950–63. 32 Prieto L, Alonso J, Lamarca R. Classical test theory versus Rasch analysis for quality of life questionnaire reduction. Health Qual Life Outcomes 2003; 1: 27. 33 Wright B. Reasonable mean-square fit values. Rasch Meas Trans 1994; 8: 370. 34 Linacre JM. Winsteps and Facets User Manuals. 35 Linacre JM. Optimizing rating scale category effectiveness. J Appl Meas 2002; 3: 85–106. 36 Institute for Objective Measurement, Inc. The Rasch Model as a Construct Validation Tool. 37 Massof RW. Application of Stochastic Measurement Models to Visual Function Rating Scale Questionnaires. Ophthalmic Epidemiol 2005; 12: 103–24. 38 Linacre JM, Wright B. WINSTEPS: Multiple-choice, rating scale, and partial credit Rasch analysis [Computer software]. Chicago: MESA; 2000. 39 Clark LA, Watson D. Constructing validity: Basic issues in objective scale development. Psychol Assess 1995; 7: 309–19. 27 Table 2.1 Sample Characteristics Variables/Statistics, n (%) Total sample (N=118) Age - mean years (SD) 25.0 (11.6) Gender: Male 116 (98.3) Hemophilia type Hemophilia A 105 (89.0) Hemophilia B 13 (11.0) Severity Severe 111 (94.1) Mild/Moderate 7 (5.9) Race/ethnicity [1] Non-Hispanic White 67 (56.8) African American 31 (26.3) Hispanic 2 (1.7) Asian 10 (8.5) Other 7 (5.9) Education: College and above [2] 83 (70.3) HTC, State Center For Inherited Blood Disorders, CA 7 (6.0) Children’s Hospital Los Angeles, CA 19 (16.1) University of Colorado Denver, CO 23 (19.5) Emory University, GA 20 (17.0) Indiana Hemophilia & Thrombosis Center, IN 10 (8.5) University of Pittsburgh, Hemophilia Center of Western Pennsylvania, PA 7 (5.9) Gulf States Hemophilia & Thrombophilia Center, TX 27 (22.9) Bloodworks Northwest, WA 5 (4.2) [1] 1 patient did not specify race/ethnicity; [2] Education for adults aged ≥ 18 years and parents of children aged 6-17 years. 28 Table 2.2 Distribution of Responses to VERITAS-Pro Item No. Domain Description Skewness [1] Kurtosis [1] Ceiling effect (%) [1,2] Floor effect (%) [1,2] 1 Time Do prophylaxis infusions on scheduled days 0.96 1.04 34.75 0.85 2 Time Infuse recommended number of times per week 1.17 0.81 49.15 0.85 3 Time Do prophylaxis infusion in the morning 0.25 -0.78 23.73 5.93 4 Time Do infusions according to schedule 1.03 0.75 33.90 4.24 5 Dose Use doctor-recommended dose for infusion 2.79 9.21 82.20 0.85 6 Dose Infuse at lower dose than prescribed 2.93 9.30 79.66 0.85 7 Dose Increase or decrease dose without calling HTC 1.23 0.39 59.32 1.69 8 Dose Use correct number of factor boxes to total recommended dose 3.20 11.99 78.81 1.69 9 Plan Plan ahead to have enough factor at home 2.39 6.69 76.27 0.85 10 Plan Track factor and supplies at home 2.02 4.35 69.49 0.85 11 Plan Run out of factor before ordering more 1.97 3.83 63.56 3.39 12 Plan Have a factor tracking system at home 1.27 0.31 56.78 9.32 13 Remember Forget to do prophylaxis infusions 0.40 -0.48 26.27 5.93 14 Remember Difficult to remember to do prophylaxis 0.94 0.56 33.05 3.39 15 Remember Remember to infuse on schedule prescribed by HTC 1.11 1.17 34.75 3.39 16 Remember Miss infusions because of forgetting 0.28 -0.82 30.51 3.39 17 Skip Skip prophylaxis infusions 0.40 -0.88 37.29 2.54 18 Skip Infuse less often than prescribed 1.37 1.84 54.24 1.69 19 Skip Skip infusion if it is inconvenient 0.72 -0.65 46.61 6.78 20 Skip Miss infusions because of skipping 0.67 -0.74 47.46 3.39 21 Communicate Call HTC for questions 0.57 -0.93 38.14 8.47 22 Communicate Call HTC if there is health concerns or changes 0.91 -0.05 44.07 5.93 23 Communicate Make treatment decisions myself 0.39 -0.96 33.90 6.78 24 Communicate Call HTC before medical interventions 1.68 1.65 68.64 3.39 [1] Numbers are bolded if they fall outside the acceptable ranges for corresponding statistics, i.e. >2 or <-2 for skewness and kurtosis, and >20% for ceiling and floor effects.[27–29] [2] Rating scale is reversed so that higher score indicates higher adherence. Ceiling and floor effects correspond to Categories 5 and 1 (originally Categories 1 and 5 before reversing) respectively. 29 Table 2.3 Summary of Modifications for VERITAS-Pro Version Person separation Item Separation Person Reliability Item Reliability Modifications were made on Modified content Original version 2.06 4.08 0.81 0.94 N/A N/A Modified Version 1 1.95 4.22 0.79 0.95 Original version Deleted items Q3, 6, 8, 11, 12, 24 Modified Version 2 1.64 4.42 0.73 0.95 Modified version 1 Deleted items Q13, 16, 17, 20 Modified Version 3 2.02 4.27 0.80 0.95 Modified version 1 Collapsed categories 1 and 2 Modified Version 4* (mVERITAS-Pro) 2.14 4.29 0.82 0.95 Modified version 1 Collapsed categories 1, 2, and 3 Modified Version 5 2.15 4.12 0.82 0.94 Modified version 1 Collapsed categories 1, 2, 3, and 4 Abbreviation: N/A, not applicable. *indicates the final modified version which was chosen. 30 Table 2.4 Category Structure for mVERITAS-Pro Category label Category before reversing rating scale Obs Count Obs Count % Measure Coherence Infit Mnsq Outfit Mnsq Step Thrld Step SE Score to Measure Avg Exp M—> C C—> M At Cat Zone 1 3 522 25 -.63 -.67 72% 38% 1.07 1.51 None None -1.74 -INF -.97 2 2 621 29 .17 .24 43% 76% .85 .85 -.40 .06 .00 -.97 .97 3 1 981 46 1.36 1.33 82% 61% .97 .97 .40 .05 1.74 .97 +INF Abbreviations: Obs, observed; Avg, average; Exp, expected; M, measure; C, category; Mnsq, mean square; Thrld, threshold; SE, standard error; At Cat, measures corresponding to an expected score equal to each category value of the rating scale, or a reasonable proxy for the extreme categories. [1] Step difficulties (average measure in this table) should be within 1.4-5.0, and each response category should have an outfit statistic <2.0 and ≥10 observations. MàC or CàM ≥40% indicates useful coherence.[35] 31 Figure 2.1 Item-person Map for VERITAS-Pro 32 Figure 2.2 Item-person Map for mVERITAS-Pro 33 CHAPTER 3: Concurrent and Predictive validities for VERITAS-Pro and mVERITAS-Pro: Validation Using Prescription and Dispensing Records ABSTRACT Introduction The Validated Hemophilia Regimen Treatment Adherence Scale - Prophylaxis (VERITAS- Pro) and modified version (mVERITAS-Pro) are the only standardized measures of adherence to prophylaxis in hemophilia. However, their concurrent and predictive validities were understudied. Aim To assess concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro against criterion measures calculated using prescription and dispensing records. Methods We utilized data collected in the Hemophilia Utilization Group Studies – Part VI (HUGS VI) as of Apr 2018. mVERITAS-Pro scores were derived from VERITAS-Pro scores, and two criterion measures were computed using prescription and dispensing records. Concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro were evaluated based on rank- biserial and Phi correlation coefficients and P-values. Results 34 The sample consisted of 144 prophylactic patients (109 adults aged³18 years and 35 children aged 6-17 years) with predominantly hemophilia A (88.9%) and severe hemophilia (93.1%). Correlations between VERITAS-Pro and mVERITAS-Pro and two criterion measures were weak and not statistically significant (rank-biserial and Phi coefficients>-0.5, P-value>0.05). Adults had higher rank-biserial correlation coefficients compared to the overall sample. Conclusions Our study identified weak concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro against criterion measures calculated using prescription and dispensing records, based on an interim analysis of HUGS VI data. Future study should analyze complete HUGS VI data and more efforts should be made to develop and validate objective measures which can be used to validate subjective measures such as VERITAS-Pro and mVERITAS-Pro. 35 3.1 Introduction Hemophilia is a congenital blood disorder characterized by deficiencies in blood coagulation factors. For decades, hemophilia patients have been treated with factor replacement therapies which intravenously infuse the deficient factors using prophylactic (or prophylaxis) or episodic (or on-demand) regimen. Prophylaxis administers factors regularly to prevent anticipated bleeding, and on-demand regimen administers factors as-needed after bleeding occurs or as extra doses for special events (e.g. sports).[1] Prophylaxis has been recommended as the optimal regimen for preventing joint bleeds and preserving joint function, and increasingly practiced worldwide.[1] In developed countries, approximately 42% of adults and 80% of children are receiving prophylaxis.[1–3] Adherence is the extent to which patients utilize the treatment as prescribed. Although adherence to prophylaxis is important for successful management of hemophilia, nonadherence is still common. A global survey of hemophilia treatment centers (HTCs) found that nonadherence to prophylaxis was prevalent across all age groups, ranging from 10% to 65%.[4] Reasons for nonadherence were identified as infrequent bleeds, high costs of factor concentrates, and the time-consuming nature of prophylaxis.[4–6] Prior studies showed that nonadherence to prophylaxis led to worsened health outcomes such as increased bleeding, decreased target joint function, and reduced health-related quality of life (HRQoL) in hemophilia patients.[2,3] Methods utilized for measuring adherence have been highly variable. Some studies employed subjective measures such as patient- or provider-reported surveys, infusion logs , general adherence rating, and the modified Morisky adherence scale.[4,5,7–14] The other 36 studies employed objective measures such as dispensing records and insurance claims.[15– 17] Few studies employed both subjective and objective measures.[18–20] Although subjective measures are subject to recall and reporting bias, the majority of adherence studies in hemophilia relied on subjective measures as opposed to objective measures, which may be due to that subjective measures are able to elicit adherence directly from patients, caregivers, and providers without the interpretation by clinicians, and less cumbersome to administer and analyze compared to objective measures.[21,22] The Validated Hemophilia Regimen Treatment Adherence Scale – Prophylaxis (VERITAS-Pro) is a widely used measure of adherence to prophylaxis in hemophilia.[10,13,23,24] Recently, we proposed a modified version (mVERITAS-Pro) using Rasch analysis, which demonstrated improved construct validity, rating scale performance, and discriminative ability, as well as the potential to reduce administrative and response burden. Despite the importance of these two measures for measurement of adherence, concurrent and predictive validities were understudied.[10] Collectively, concurrent and predictive validities are “criterion validity” determing the extent to which a measure correlates with a criterion measure of the same construct which is usually the “gold standard”.[25] Specifically, concurrent validity determines the association between VERITAS-Pro or mVERITAS-Pro and the criterion measure administered contemporarily, and predictive validity represents the utility of VERITAS-Pro or mVERITAS-Pro to predict real-world adherence in the future.[26] Prior validation study of VERITAS-Pro assessed concurrent validity without tapping into predictive validity, and both concurrent and predictive validities for mVERITAS-Pro were yet to be established.[10] 37 The objective of this study was to assess concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro against objective criterion measures calculated using factor prescription and dispensing records. 3.2 Materials and methods 3.2.1 Study design and population The Hemophilia Utilization Group Studies Part VI (HUGS VI) is a multi-center observational study assessing adherence to factor replacement therapies in persons with hemophilia A and B in the USA. Patients aged ≥6 years, had been treated with conventional factor products, or with extended half-life (EHL) products (<6 weeks prior to enrolment) were enrolled from eight hemophilia treatment centers (HTCs) in different regions of the USA. In HUGS VI, patient surveys were administered at enrolment and every three months to collect data on socio-demographics, insurance status, bleeding, pain, motion limitation, comorbidities, HRQoL, and adherence which was measured using VERITAS-Pro. Medical charts as well as factor prescription and dispensing records were collected six months pre- enrolment to nine months post-enrolment, and reviewed by study coordinators at each participating HTC to extract useful variables. Prescription records documented the date of prescription, brand and total units of factors prescribed, frequency of administration, number of refills, treatment regimen (prophylaxis or episodic), and patient weight. Dispensing records documented the date of dispensing as well as the type (recombinant or plasma- derived), brand (conventional or EHL factor products), and total units of factors dispensed. 38 3.2.2 VERITAS-Pro and mVERITAS-Pro VERITAS-Pro is a widely used patient- or parent-reported questionnaire consisting of 24 items and 6 domains (Time, Dose, Plan, Remember, Skip, and Communicate). Responses are rated on a 5-point rating scale (1-Always, 2-Often, 3-Sometime, 4-Rarely, 5-Never), with higher scores indicating lower adherence. We calculated adherence score as a continuous and as a binary variable, with the continuous score being the sum of scores for all items and the binary score being the continuous score dichotomized into adherence (<57) and non- adherence (³57) based on the cut-off score proposed in prior study.[10] mVERITAS-Pro was derived from VERITAS-Pro based on Rasch analysis (described in Chapter 2), consisting of 18 items and a 3-point rating scale (1-Always, 2-Sometime, 3-Never). As for VERITAS-Pro, higher scores on mVERITAS-Pro indicate lower adherence. Since VERITAS-Pro and mVERITAS-Pro were collected four times once every three months, we calculated the adherence scores corresponding to four study periods, i.e. Period 1 (3 months pre-enrolment) and Period 2, 3, 4 (Months 1-3, 4-6, and 7-9 post-enrolment respectively). 3.2.3 Criterion measures Although various methods have been used to measure adherence to prophylaxis, there was a lack of the gold standard. Therefore we undertook an exploratory approach to calculate an objective measures of adherence as the proportion of factors dispensed as prescribed for prophylaxis using prescription and dispensing data (Criterion Measure 1) (Eq. 1): 39 !"ℎ$%$&'$ ( = +,-../01 234.-5 6/7890796 2-5 85-8:;,3</7 +,-../01 234.-5 859745/=96 2-5 85-8:;,3</7 (1) where clotting factor dispensed/prescribed was the total units of factors dispensed/prescribed (IU/kg) for prophylaxis in each study period regardless of the type and brand of factors. We also utilized another objective measure developed by Chen et al. calculating adherence to prophylaxis as follows (Criterion Measure 2) (Eq. 2)[15]: !"ℎ$%$&'$ > = (+,-../01 234.-5 6/7890796)A(B8/7-6/4 6-79) ´ (C,996/01 98/7-697) (D5-8:;,3</7 6-79) ´ (D5-8:;,3</7 /02E7/-07 895 F99G)´ (H ´ I.KL (2) where clotting factor dispensed was the total units of factors dispensed, episodic dose was assumed to be 57.5 IU/kg and 113 IU/kg for hemophilia A and B respectively, bleeding episodes were the total number of bleeds reported in each study period, prophylaxis dose was assumed to be 27.5 IU/kg and 35 IU/kg for hemophilia A and B respectively, prophylaxis infusions per week was assumed to be 3 and 2 times for hemophilia A and B respectively, 13 weeks represented the length of each study period, 0.85 represented the 85% threshold required for continuous prophylaxis according to the World Federation of Hemophilia. Both criterion adherence measures were calculated for four study periods, and dichotomized into adherent (³ 0.8) and non-adherent (< 0.8) using a cut-off of 0.8, which is a commonly chosen threshold for adherence vs. nonadherence.[27] 3.2.4 Statistical analysis Concurrent and predictive validities were evaluated by correlations between VERITAS- Pro or mVERITAS-Pro and two criterion measures. Concurrent validity was evaluated based on correlations between VERITAS-Pro or mVERITAS-Pro and two criterion measures 40 administered in the same study period. For example, VERITAS-Pro in Period 1 and mVERITAS-Pro in Period 1. Predictive validity was assessed based on correlations between VERITAS-Pro or mVERITAS-Pro administered in an earlier study period and the criterion measures calculated for later study periods. For example, the 3-month predictive validity of VERITAS-Pro was characterized by correlation between VERITAS-Pro scores Period 1 and criterion measures in Period 2. We hypothesized that that the strength of correlation decreases as the interval between measures increases due to a longer recall period. We employed two types of correlation coefficients which did not require normality of data, i.e. rank-biserial correlation coefficient and Phi coefficient, since data were not normally distributed for VERITAS-Pro, mVERITAS-Pro, or criterion measures. Rank- biserial correlation characterizes the association between a binary variable and the ranks of a continuous variable, and was calculated for VERITAS-Pro and mVERITAS-Pro.[28] Phi coefficient is a special case of the Pearson correlation coefficient when two variables are coded as binary, and was calculated for VERITAS-Pro only since the cut-off value was not established for mVERITAS-Pro at the time of analysis.[29] Since correlations between VERITAS-Pro or mVERITAS-Pro and the criterion measures should be negative, the strength of correlations were interpreted as follows: ³-0.30, negligible correlation; -0.30 to - 0.50, low correlation; -0.50 to -0.70, moderate correlation; -0.70 to -0.90, high correlation; - 0.90 to -1.00, very high correlation.[30] Ad-hoc sample size calculations were conducted using G*Power version 3.1.[31] All other statistical analyses were performed using SAS® software version 9.4. 41 3.3 Results A total of 144 prophylactic patients who completed at least one survey and had prescription and/or dispensing records collected as of Apr 2018 were included in the analyses (Table 3.1). Among them, 136 patients received prophylaxis since initial survey and 8 patients switched from episodic regimen to prophylactic regimen during the 9-month follow- up. The sample consisted of predominantly persons with hemophilia A (88.9%) and severe hemophilia (93.1%), with the majority experiencing joint bleeds six months prior to enrollment (54.5%), having moderate-to-extreme joint pain (59.0%) and none-to-a-little-bit motion limitation (52.1%). Compared to prior validation study of VERITAS-Pro, our sample had larger size (144 vs. 67) and an older mean age (25.5 vs. 15.2 years), consisted of a slightly higher percentage of hemophilia A (88.9% vs. 83.6%), African American (25.9% vs. 14.9%) and Asian (8.4% vs. 1.5%), and lower percentage of severe hemophilia (93.1% vs. 95.5%) and non-Hispanic white (57.3% vs. 76.1%). Sample sizes available for correlation calculations differed across study periods due to missing data, which were lowest for VERITAS-Pro, mVERITAS-Pro, and Criterion Measure 2 in Period 2 and for Criterion Measure 1 in Period 4. Ad hoc sample size calculations indicated that a sample size of 26 was able to detect a two-tail large correlation (effect size=0.5) at a significance level of 0.05 for rank-biserial correlations, and that of 29 for Phi coefficients. Based on correlations with sample sizes satisfying ad hoc calculations, VERITAS-Pro and mVERITAS-Pro had weak and not statistically significant correlations (coefficients>- 0.50, P-value>0.05) with two criterion measures (Tables 3.2-3.5). In the overall sample, 42 difference in rank-biserial correlations between VERITAS-Pro and mVERITAS-Pro as well as between two criterion measures were minimal, and difference in Phi coefficients between two criterion measures were not pronounced as well. However, it is worth nothing that the strengths of rank-biserial correlations between VERITAS-Pro or mVERITAS-Pro and two criterion measures were higher for adults compared to the overall sample when the sample sizes were sufficient based on ad hoc calculations (Tables 3.2 and 3.3). For Criterion Measure 1, correlations with VERITAS-Pro were -0.48 vs. -0.15, and those with mVERITAS-Pro were -0.50 vs. -0.14 in Period 1. Correlations between VERITAS-Pro in Period 1 and Criterion Measure 1 in Periods 2 and 3 were -0.06 vs. 0.08 and -0.16 vs. -0.08 respectively. Correlations between mVERITAS-Pro in Period 1 and Criterion Measure 1 in Periods 2 and 3 were -0.03 vs. 0.09 and -0.24 vs. -0.12 respectively. For Criterion Measure 2, correlations with VERITAS-Pro in Periods 1, 3, and 4 were -0.27 vs. -0.12, -0.42 vs. -0.30, and -0.34 vs. -0.18 respectively. Correlations between mVERITAS-Pro and Criterion Measure 2 in Periods 1, 3, and 4 were -0.24 vs. -0.10, -0.36 vs. -0.24, and -0.27 vs. -0.12 respectively. Correlations between VERITAS-Pro in Period 1 and Criterion Measure 2 in Periods 3 and 4 were -0.50 vs. -0.36 and -0.26 vs. -0.11 respectively, and those between VERITAS-Pro in Period 3 and Criterion Measure 2 in Period 4 was -0.34 vs. -0.14. For adults versus the overall sample, rank-biserial correlations between mVERITAS-Pro in Period 1 and Criterion Measure 2 in Periods 3 and 4 were -0.48 vs. -0.33 and -0.21 vs. -0.08 respectively, and those between mVERITAS-Pro in Period 3 and Criterion Measure 4 in Period 4 was -0.24 vs. -0.07. However, such differences between adults and the overall sample was observed for rank- biserial correlations but not for Phi coefficients (Tables 3.4 and 3.5). 43 In contrary to our hypothesis, the strength of rank-biserial association between VERITAS-Pro or mVERITAS-Pro and two criterion measures did not decrease with longer intervals between measurements, but significantly increased from a 3- to 6-month interval, then decreased from a 6- to 9-month interval (Tables 3.2 and 3.3). However, the Phi correlations slightly decreased for VERITAS-Pro and Criterion Measure 1, and but not with Criterion Measure 2. 3.4 Discussion and conclusion VERITAS-Pro and mVERITAS-Pro are the only standardized patient- or parent-reported measures of adherence to prophylaxis in hemophilia. VERITAS-Pro has been widely used and translated into more than 30 languages, and mVERITAS-Pro was recently proposed using Rasch analysis which demonstrated improved psychometric properties and the potential to reduce administrative and respondent burden.[10,13,23,24] Our study identified insufficient concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro against objective measures calculated using prescription and dispensing records, based on an interim data analysis of 144 prophylactic patients in HUGS VI. Our study was the first to examine predictive validity for VERITAS-Pro and concurrent and predictive validities for mVERITAS-Pro, and the first to evaluate these validities against objective criterion measures calculated using prescription and dispensing records. Findings from our analyses contradicted the prior study which found moderate-to-strong concurrent validity for VERITAS-Pro, and may be explained by several reasons.[10] First, we utilized a different type of criterion measures compared to prior study. Two subjective measures were 44 employed in prior study, the provider-reported general adherence rating scale and the percentage of recommended infusions administered computed using patient-reported infusion logs. However, this study employed two objective measures calculated using prescription and dispensing records as the criterion. Second, the prior study assessed correlations utilizing Pearson correlation coefficients assuming normality of data without specifying the data distributions which raised questions about validity of the results. In contrary, our study employed rank-biserial correlation and Phi coefficients which relaxed the assumption of normality and were better suited for analyzing non-normally distributed data. Third, the prior sample was less heterogeneous compared to our sample which included patients from geographically diverse HTCs in the USA and equipped with better external validity. However, findings from this study should be interpreted with a few limitations. First, the results were based on interim analysis of HUGS VI data, since data collection of follow-up surveys as well as prescription and dispensing records were still on-going at the time of analysis. As a result, sample sizes were insufficient at some data points and interpretation of those results was not possible. Therefore, a re-analysis of the complete HUGS VI data may show different results compared to present analysis. Second, data reporting issues associated with prescription and dispensing records were present, since some patients did not have matching prescription and dispensing records. Third, we had to develop and utilize objective criterion measures that were yet to be validated due to a lack of gold standard criterion measures. Although we believed that an evaluation of concurrent and predictive validities using objective criterion measures would significantly complement prior studies using 45 subjective measures, uncertainty around the validity of criterion measures may have biased the results. Furthermore, there was a possibility that adherence to prophylaxis was not accurately captured since hemophilia treatments are administered intravenously which are drastically different from oral medications. In conclusion, this interim analysis of patients with hemophilia A and B from different region in the USA showed that VERITAS-Pro and mVERITAS-Pro had weak concurrent and predictive validities against criterion measures computed using prescription and dispensing records. In addition, a paucity of gold standard measures of adherence to prophylaxis in hemophilia was observed. Future research should aim at analyzing complete HUGS VI data to address the uncertainty around this interim data analysis, and developing and validating objective measures which can be used as criterion to assess concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro. 46 3.5 Chapter references 1 Srivastava A, Brewer AK, Mauser-Bunschoten EP, Key NS, Kitchen S, Llinas A, et al. Guidelines for the management of hemophilia. Haemophilia 2013; 19: e1–47. 2 Manco-Johnson MJ, Sanders J, Ewing N, Rodriguez N, Tarantino M, Humphries T, et al. Consequences of switching from prophylactic treatment to on-demand treatment in late teens and early adults with severe haemophilia A: the TEEN/TWEN study. Haemophilia 2013; 19: 727–35. 3 Walsh CE, Valentino LA. Factor VIII prophylaxis for adult patients with severe haemophilia A: results of a US survey of attitudes and practices. Haemophilia 2009; 15: 1014–21. 4 Geraghty S, Dunkley T, Harrington C, Lindvall K, Maahs J, Sek J. Practice patterns in haemophilia A therapy – global progress towards optimal care. Haemophilia 2006; 12: 75–81. 5 Hacker MR, Geraghty S, Manco-Johnson M. Barriers to compliance with prophylaxis therapy in haemophilia. Haemophilia 2001; 7: 392–6. 6 Carcao MD, Aledort L. Prophylactic factor replacement in hemophilia. Blood Rev 2004; 18: 101–13. 7 De Moerloose P, Urbancik W, Van Den Berg HM, Richards M. A survey of adherence to haemophilia therapy in six European countries: results and recommendations. Haemophilia 2008; 14: 931–8. 8 Thornburg CD, Pipe SW. Adherence to prophylactic infusions of factor VIII or factor IX for haemophilia. 2006; . 9 Thornburg CD. Physicians’ perceptions of adherence to prophylactic clotting factor infusions. Haemophilia 2008; 14: 25–9. 10 Duncan N, Kronenberger W, Roberson C, Shapiro A. VERITAS-Pro: a new measure of adherence to prophylactic regimens in haemophilia. Haemophilia 2010; 16: 247–55. 11 Lock J, Raat H, Duncan N, Shapiro A, Beijlevelt M, Peters M, et al. Adherence to treatment in a Western European paediatric population with haemophilia: reliability and validity of the VERITAS-Pro scale. Haemophilia 2014; 20: 616–23. 12 Duncan NA, Kronenberger WG, Krishnan S, Shapiro AD. Adherence To Prophylactic Treatment In Hemophilia As Measured Using The Veritas-Pro And Annual Bleed Rate (Abr). Value Health 2014; 17: A230. 13 McLaughlin JM, Witkop ML, Lambing A, Anderson TL, Munn J, Tortella B. Better adherence to prescribed treatment regimen is related to less chronic pain among adolescents and young adults with moderate or severe haemophilia. Haemophilia 2014; 20: 506–12. 47 14 Tencer T, Roberson C, Duncan N, Johnson K, Shapiro A. A haemophilia treatment centre- administered disease management programme in patients with bleeding disorders. Haemophilia 2007; 13: 480–8. 15 Chen CX. The impact of treatment decisions and adherence on outcomes in small hereditary disease populations (Doctoral dissertation). Retrieved from the University of Southern California Digital Library Database. 2016. 16 Chen C, Eldar-lissai A, Zhou J, Buckley B. Changes in adherence after initiating treatment with prolonged half-life clotting factors for hemophilia. Haemophilia 2016; 22. 17 Buckley B, Hagberg B, Zhou J, Eldar-lisaai A. Adherence to Treatment in Hemophilia: A Comparison of Conventional and Prolonged Half-life Therapies. Haemophilia 2016; 22: 91– 2. 18 Du Treil S, Rice J, Leissinger CA. Quantifying adherence to treatment and its relationship to quality of life in a well-characterized haemophilia population. Haemophilia 2007; 13: 493– 501. 19 Shapiro A, Gruppo R, Pabinger I, Collins PW, Hay CR, Schroth P, et al. Integrated analysis of safety and efficacy of a plasma- and albumin-free recombinant factor VIII (rAHF-PFM) from six clinical studies in patients with hemophilia A. Expert Opin Biol Ther 2009; 9: 273– 83. 20 Ho S, Gue D, McIntosh K, Bucevska M, Yang M, Jackson S. An objective method for assessing adherence to prophylaxis in adults with severe haemophilia. Haemophilia 2014; 20: 39–43. 21 Stull DE, Leidy NK, Parasuraman B, Chassany O. Optimal recall periods for patient-reported outcomes: challenges and potential solutions. Curr Med Res Opin 2009; 25: 929–42. 22 Murri R, Ammassari A, Trotta MP, Luca AD, Melzi S, Minardi C, et al. Patient-reported and Physician-estimated Adherence to HAART. J Gen Intern Med 2004; 19: 1104–10. 23 Lock J, Raat H, Duncan N, Shapiro A, Beijlevelt M, Peters M, et al. Adherence to treatment in a Western European paediatric population with haemophilia: reliability and validity of the VERITAS-Pro scale. Haemophilia 2014; 20: 616–23. 24 Thornburg CD, Duncan NA. Treatment adherence in hemophilia. Patient Prefer Adherence 2017; 11: 1677–86. 25 Terwee CB, Bouwmeester W, Elsland SL van, Vet HCW de, Dekker J. Instruments to assess physical activity in patients with osteoarthritis of the hip or knee: a systematic review of measurement properties. Osteoarthritis Cartilage 2011; 19: 620–33. 26 Cronbach LJ, Meehl PE. Construct validity in psychological tests. Psychol Bull 1955; 52: 281– 302. 48 27 Dunbar-Jacob J, Mortimer-Stephens MK. Treatment adherence in chronic disease. J Clin Epidemiol 2001; 54: S57–60. 28 Kraemer HC. Biserial Correlation. Encyclopedia of Statistical Sciences. American Cancer Society; 2006. 29 Kraemer HC. Correlation coefficients in medical research: from product moment correlation to the odds ratio. Stat Methods Med Res 2006; 15: 525–45. 30 Hinkle DE. Applied Statistics for the Behavioral Sciences. Cengage Learning; 2002. 31 Faul F, Erdfelder E, Buchner A, Lang A-G. Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behav Res Methods 2009; 41: 1149–60. 49 Table 3.1 Sample Characteristics at Baseline Variables/Statistics, n (%) Total sample (N=144) Age - mean years (SD) 25.5 (11.6) Age group Adults 109 (75.7) Children 35 (24.3) Race/ethnicity [2] Non-Hispanic White 82 (57.3) African American 37 (25.9) Hispanic 2 (1.4) Asian 12 (8.4) Other 10 (7.0) Hemophilia type Hemophilia A 128 (88.9) Hemophilia B 16 (11.1) Severe hemophilia 134 (93.1) Treatment regimen at initial survey Prophylaxis 136 (94.4) On-demand [1] 8 (5.6) Experienced joint bleeds six month pre-enrolment [2] 78 (54.5) Joint pain None to a little bit 59 (41.0) Moderate to extreme 85 (59.0) Motion limitation None to a little bit 75 (52.1) Moderate to extreme 69 (47.9) Education: College and above [2,3] 105 (95.5) Employed [2] 97 (69.3) Annual household income>$50,000 [2] 52 (46.8) Notes: [1] Eight patients who switched from on-demand to prophylaxis during follow-up were included in the analyses. [2] Total did not add up to N=144 due to missing data. [3] Education for adults aged ≥ 18 years and parents of children aged 6-17 years. 50 Table 3.2 Rank-biserial Correlations with Criterion Measure 1 Rank-biserial correlation (Sample size) Criterion Measure 1 Period 1 Period 2 Period 3 Period 4 Adults Children Overall Adults Children Overall Adults Children Overall Adults Children Overall VERITAS-Pro Period 1 -0.48 (n=43) 0.71 (n=9) -0.15 (n=52) -0.06 (n=43) 0.52 (n=10) 0.08 (n=53) -0.16 (n=27) 0.07 (n=12) -0.08 (n=39) -0.05 (n=30) -1.00 (n=4) -0.10 (n=34) Period 2 - - - 0.54 (n=16) 0.25 (n=6) 0.48 (n=22) 1.00 (n=7) 1.00 (n=4) 1.00 (n=11) 1.00 (n=10) N/A (n=0) 1.00 (n=10) Period 3 - - - - - - -0.24 (n=19) -0.38 (n=10) -0.27 (n=29) -0.17 (n=22) -1.00 (n=3) -0.34 (n=25) Period 4 - - - - - - - - - 0.00 (n=23) N/A (n=1) -0.02 (n=24) mVERITAS-Pro Period 1 -0.50 (n=42) 0.71 (n=9) -0.14 (n=51) -0.03 (n=43) 0.57 (n=10) 0.09 (n=53) -0.24 (n=27) 0.11 (n=12) -0.12 (n=39) -0.06 (n=30) -1.00 (n=4) -0.11 (n=34) Period 2 - - - 0.54 (n=16) 0.75 (n=6) 0.58 (n=22) 1.00 (n=7) 1.00 (n=4) 1.00 (n=11) 1.00 (n=10) N/A (n=0) 1.00 (n=10) Period 3 - - - - - - -0.24 (n=19) -0.13 (n=10) -0.21 (n=29) -0.06 (n=22) -1.00 (n=3) -0.25 (n=25) Period 4 - - - - - - - - - -0.02 (n=23) N/A (n=1) -0.03 (n=24) Note: Interpretation of correlation coefficient: ³-0.30, negligible correlation; -0.30 to -0.50, low correlation; -0.50 to -0.70, moderate correlation; -0.70 to -0.90, high correlation; -0.90 to -1.00, very high correlation. 51 Table 3.3 Rank-biserial Correlations with Criterion Measure 2 Rank-biserial correlation (Sample size) Criterion Measure 2 Period 1 Period 2 Period 3 Period 4 Adults Children Overall Adults Children Overall Adults Children Overall Adults Children Overall VERITAS-Pro Period 1 -0.27 (n=68) 0.41 (n=18) -0.12 (n=86) 0.05 (n=18) 1.00 (n=7) 0.08 (n=25) -0.50 (n=41) 0.08 (n=14) -0.36 (n=55) -0.26 (n=44) 0.44 (n=10) -0.11 (n=54) Period 2 - - - 0.03 (n=19) 0.33 (n=8) 0.07 (n=27) -0.14 (n=15) 0.25 (n=6) -0.09 (n=21) 0.35 (n=13) 0.50 (n=6) 0.44 (n=19) Period 3 - - - - - - -0.42 (n=45) 0.05 (n=15) -0.30 (n=60) -0.34 (n=35) 0.40 (n=10) -0.14 (n=45) Period 4 - - - - - - - - - -0.34 (n=46) 0.40 (n=10) -0.18 (n=56) mVERITAS-Pro Period 1 -0.24 (n=67) 0.35 (n=18) -0.10 (n=85) 0.05 (n=18) 1.00 (n=7) 0.15 (n=25) -0.48 (n=41) 0.04 (n=14) -0.33 (n=55) -0.21 (n=44) 0.40 (n=10) -0.08 (n=54) Period 2 - - - 0.14 (n=19) 0.60 (n=8) 0.19 (n=27) -0.14 (n=15) 0.75 (n=6) -0.02 (n=21) 0.43 (n=13) 0.50 (n=6) 0.51 (n=19) Period 3 - - - - - - -0.36 (n=45) 0.05 (n=15) -0.24 (n=60) -0.24 (n=35) 0.28 (n=10) -0.07 (n=45) Period 4 - - - - - - - - - -0.27 (n=46) 0.44 (n=10) -0.12 (n=56) Note: Interpretation of correlation coefficient: ³-0.30, negligible correlation; -0.30 to -0.50, low correlation; -0.50 to -0.70, moderate correlation; -0.70 to -0.90, high correlation; -0.90 to -1.00, very high correlation. 52 Table 3.4 Phi Coefficients of Correlation between VERITAS-Pro and Criterion Measure 1 Phi coefficient (Sample size) Criterion Measure 1 Period 1 Period 2 Period 3 Period 4 Adults Children Overall Adults Children Overall Adults Children Overall Adults Children Overall VERITAS-Pro Period 1 0.23 (n=43) -0.38 (n=9) 0.09 (n=52) -0.04 (n=43) -0.43 (n=10) -0.12 (n=53) 0.03 (n=27) -0.17 (n=12) -0.03 (n=39) 0.00 (n=30) N/A (n=4) -0.03 (n=34) Period 2 - - - -0.28 (n=16) -0.32 (n=6) -0.29 (n=22) -0.17 (n=7) N/A (n=4) -0.15 (n=11) -0.17 (n=10) N/A (n=0) -0.17 (n=10) Period 3 - - - - - - 0.07 (n=19) -0.17 (n=10) 0.01 (n=29) 0.08 (n=22) N/A (n=3) 0.05 (n=25) Period 4 - - - - - - - - - -0.08 (n=23) N/A (n=1) -0.08 (n=24) Note: Interpretation of correlation coefficient: ³-0.30, negligible correlation; -0.30 to -0.50, low correlation; -0.50 to -0.70, moderate correlation; -0.70 to -0.90, high correlation; -0.90 to -1.00, very high correlation. 53 Table 3.5 Phi Coefficients of Correlation between VERITAS-Pro and Criterion Measure 2 Phi coefficient (Sample size) Criterion Measure 2 Period 1 Period 2 Period 3 Period 4 Adults Children Overall Adults Children Overall Adults Children Overall Adults Children Overall VERITAS-Pro Period 1 0.00 (n=68) -0.27 (n=18) -0.05 (n=86) 0.12 (n=18) -0.26 (n=7) 0.07 (n=25) 0.10 (n=41) 0.09 (n=14) 0.09 (n=55) 0.06 (n=44) -0.22 (n=10) 0.00 (n=54) Period 2 - - - -0.09 (n=19) -0.29 (n=8) -0.12 (n=27) 0.04 (n=15) -0.32 (n=6) -0.03 (n=21) 0.23 (n=13) -0.63 (n=6) -0.09 (n=19) Period 3 - - - - - - 0.13 (n=45) -0.13 (n=15) 0.06 (n=60) 0.21 (n=35) -0.22 (n=10) 0.09 (n=45) Period 4 - - - - - - - - - 0.33 (n=46) 0.00 (n=10) 0.26 (n=56) Note: Interpretation of correlation coefficient: ³-0.30, negligible correlation; -0.30 to -0.50, low correlation; -0.50 to -0.70, moderate correlation; -0.70 to -0.90, high correlation; -0.90 to -1.00, very high correlation. 54 CHAPTER 4: Transition from childhood to adulthood and health outcomes in persons with hemophilia A: evidence from longitudinal analyses in the USA ABSTRACT Introduction Persons with hemophilia (PWH) in transition from childhood to adulthood are at a higher risk of adverse health outcomes compared to children and adults, however health outcomes in this population remained understudied. Aim To compare trends in health outcomes between hemophilia patients in transition from childhood to adulthood and those not in transition, and to examine the impact of transition on bleeding. Methods We utilized longitudinal data collected from two Hemophilia Utilization Group Studies (HUGS) spanning across 7 to 10 years. Descriptive trend analyses were performed to compare health outcomes between adolescents (the transition group) and children and adults (the non-transition groups). Random effects negative binomial regression analyses was employed to examine the association between transition and bleeding. Results Trend analyses showed adolescents had higher bleeding and factor utilization compared to children regardless of treatment regimen. Compared to adults, adolescents had lower bleeding in on- 55 demand population, similar bleeding in prophylactic population, and higher factor utilization regardless of treatment regimen. Adolescents had highest rate of bleeding despite highest factor utilization among prophylactic patients. Multivariable regression analyses found the six-month bleeding rates (6moBRs) for prophylactic and on-demand patients were 16% and 12% lower in children compared to adolescents. The 6moBR was 203% higher in adults than adolescents after controlling for interactions between age and adherence. Conclusion In a multi-center sample of persons with hemophilia A in the USA, transition from childhood to adulthood was associated with increased bleeding compared to childhood, and increased factor utilization compared to childhood and adulthood. Actions should be taken to improve health outcomes in PWH in transition from childhood to adulthood. 4.1 Introduction Hemophilia A is a rare hereditary blood disorder characterized by deficiencies of coagulation factor VIII (FVIII), with an incidence of 1 in 5,000 live male births per year.[1] Persons with hemophilia (PWH) are treated with regular infusions of the deficient coagulation factor to prevent anticipated bleeding (prophylactic regimen or prophylaxis) or as-needed after bleeding occurs (episodic or on-demand regimen). Prophylaxis has been recommended as the optimal regimen to control bleeds and preserve joint function, and approximately 80% of children and 42% of adults are receiving prophylaxis in developed countries.[2,3] Hemophilia has differential impacts on patients at different ages. Pediatric patients are usually taken care of by their parents, and adult patients manage hemophilia on their own. Adolescent 56 patients may begin to manage hemophilia as early as 14 years old with diminishing influence from parents, however it is worth noting that the drastic social developmental, cognitive, and physical challenges they experience during the phase of transition from childhood to adulthood may significantly impact hemophilia treatment.[4] Several challenges for patients in transition from childhood to adulthood have been identified.[5] First, switching from pediatric to adult providers and hemophilia treatment centers (HTCs) settings, which requires a great amount of efforts to build new relationships with new providers in a new environment. Second, a high risk of quitting prophylaxis due to unawareness of the benefits, since bleeding rarely occurs in the childhood of this generation of patients who were extensively covered by prophylaxis. Third, challenges for all adolescents such as teenage rebellion, disclosure of disease in interpersonal relationships, financial and employment-related barriers, and health insurance coverage which is an issue just for PWH in the USA.[5,6] Approximately 55% of hemophilia patients are young people and many of whom are in transition from childhood to adulthood in the USA.[7] Transition was identified as a priority focus for patients with bleeding disordered by the Health Resource and Services Administration (HRSA), and transitional care programs have been established in some HTCs in the USA.[8] However, non- adherence to prophylaxis was lowest in adolescents across all age groups, and more than 30% of adolescents and young adults had once discontinued prophylaxis in their lives.[9–12] Nonadherence to prophylaxis has been linked to adverse health outcomes such as increased bleeding, worsened joint function, and reduced health-related quality of life.[2,3] It is likely that health outcomes in hemophilia patients in transition from childhood to adulthood may be compromised due to nonadherence to prophylaxis, however this particularly vulnerable population remained understudied. 57 The study was initiated with two objectives. First, to compare trends in health outcomes such as bleeding, factor utilization, adherence, and prevalence of prophylaxis in PWH in transition from childhood to adulthood versus those not in transition. Second, to examine the association between transition from childhood to adulthood and bleeding using regression analysis. 4.2 Materials and methods 4.2.1 Data source Our analyses utilized data collected in the Hemophilia Utilization Group Studies (HUGS) Part Va (HUGS Va) and the Long Term Study (HUGS LTS). HUGS Va evaluated cost and burden of illness in 329 persons with hemophilia A (aged 2-64 years) treated at six geographically diverse HTCs from July 2005 to July 2007. In 2014, HUGS LTS gathered additional data on health outcomes, burden of illness, and costs in 109 patients originally enrolled in HUGS Va. Patient- reported data on socio-demographics, hemophilia treatment regimen, health insurance, and clinical outcomes were collected through surveys. HUGS Va surveys were administered at initial interview, monthly in the first year of follow-up and semi-annually in the second year. HUGS LTS patient survey was administered one-time with a 6-month recall period. In addition, data coordinators at participating HTCs reviewed medical charts and factor dispensing records and gathered data on patient clinical characteristics such as weight and inhibitor development, health resources utilization such as outpatient, hospital, and emergency room (ER) visits, and factor utilization. Medical charts was collected during the first year of follow-up in HUGS Va, retrospectively for six months and prospectively for five months in HUGS LTS. 58 Dispensing records were collected during two years of follow-up in HUGS Va, and retrospectively for six months and prospectively for five months in HUGS LTS. 4.2.2 Health outcomes We examined temporal trends in health outcomes including bleeding, factor utilization, treatment regimen (prophylaxis or on-demand), and adherence to treatment from HUGS Va to HUGS LTS. Six-month bleeding Rate (6moBR) and treatment regimen were directly obtained from patient surveys administered every six months (Months 13-18 and 19-24 in HUGS Va, and within 6 months prior to HUGS LTS) except the first year of HUGS Va which administered patient surveys monthly. Therefore, the first year of HUGS Va was divided into two 6-month study periods (Months 1-6 and 7-12). 6moBR was calculated as the total number of bleeds occurred and treatment regimen was identified as the most frequently reported (>3 out of 6 months) treatment regimen during those periods. As for bleeding and treatment regimen, factor utilization and adherence to treatment were also computed for five study periods (Months 1-6, 7-12, 13-18, and 19-24 in HUGS Va, and within 6 months prior to HUGS LTS). Factor utilization (international units per kg, or IU/kg) was the sum of factors shipped to the patient and those dispensed during hospitalization and ER visits (if there were any) divided by patient weight (kilogram, or kg). Both plasma and recombinant FVIII products were included in the calculation. Since prescription records were not collected in HUGS Va or HUGS LTS, adherence to treatment was calculated using a proxy measure developed by Chen et al. using factor dispensing records (Eq. 1): !"ℎ$%$&'$ ()*( +,-./01 = (5-*66078 9.:6*) ;01(<71<;)>(?(01*;0: ;*1<) ´ (@-<<;078 <(01*;<1) (A)*( +,-./01 ;*1<) ´ (A)*( +,-./01 079B10*71 (<) C<<D)´ EF ´ G.IJ (1) 59 where clotting factor dispensed was the total units of factors dispensed, episodic dose was assumed to be 57.5 IU/kg, bleeding episodes were the total number of bleeds reported in each study period, prophylaxis dose was assumed to be 27.5 IU/kg, prophylaxis infusions per week was assumed to be 3 times, 26 weeks indicated the length of each study period, 0.85 represented the 85% threshold required for continuous prophylaxis according to the World Federation of Hemophilia.[13] Adherence to prophylaxis was further dichotomized into adherent (≥80%) and non-adherent (<80%) using the cut-off of 80% which is commonly used to define adherence vs. nonadherence for chronic conditions.[14] Since adherence to on-demand regimen was rarely defined in the literature, and even more difficult when prescription records were not available as in our study, we defined an exploratory measure of adherence to on-demand regimen as follows (Eq. 2): !"ℎ$%$&'$ *7>;<K.7; = 5-*66078 9.:6*) ;01(<71<; ?(01*;0: ;*1< × @-<<;078 <(01*;<1 × 100% (2) where clotting factor dispensed was factor utilization (IU/kg) during the 6-month study period, episodic dose was assumed to be 57.5 IU/kg based on clinical trials of hemophilia A factor products,[15,16] and bleeding episodes were the 6moBR reported for each study period. Adherence to on-demand regimen was further dichotomized into adherent (80-150% in children aged <18 years, 80-120% in adults aged ≥18 years) and non-adherent (<80% or >150% in children aged <18 years, <80% or >120% in adults aged ≥18 years), with the cut-off values chosen based on clinical considerations employed by Du Treil et al.[17] 60 4.2.3 Study groups We excluded patients with inhibitors since they may have consumed excessive amount of factor products for immune tolerance induction (ITI) treatment. In addition, we excluded patients without dispensing records since it was impossible to calculate their adherence. Our sample was divided into three age groups – adolescents (aged 12-20 years) who were in transition from childhood to adulthood, children (aged 0-11 years), and adults (aged 21+ years) following the characterization of developmental stages in hemophilia patients by the National Hemophilia Foundation.[18] 4.2.4 Statistical analyses Sociodemographic, clinical, and insurance-related characteristics were compared between adolescents, children, and adults at the baseline of HUGS Va and HUGS LTS. Temporal trends in median 6moBR, median factor utilization (IU/kg), treatment regimen, and adherence to treatment were also compared between three age groups. Median values were chosen over mean values to minimize the impact of outlier data points. We employed random-effect multivariable negative binomial (NB) regression analyses to examine the association between transition from childhood to adulthood and bleeding since prior study found NB model was the best approach to analyzing highly skewed bleeding data compared to other models.[19] In our sample, distributions of the number of bleeds were highly skewed regardless of treatment regimen in all study periods (Tables 1a-5c in Appendix 2), therefore NB model was most suitable for our analyses. The unit of analysis was an individual patient. Panel data specifications were added to account for serial correlations between observations collected in the same individual at different points in 61 time. The unobserved time-invariant characteristics were specified as random effects (RE) rather than fixed effects (FE) for two reasons. First, FE models do not estimate the effect of time-invariant variables which may be the interest of our analyses.[20] Second, from a statistical perspective, no systematic difference were found between RE and FE estimators using the Hausman specification test.[21] The associations between transition from childhood to adulthood and bleeding were estimated in prophylaxis and on-demand patients separately, using models specified as follows: P%QR ℎSTUVWX: Zit=]i+_1 !`$1it +_2 !`$3it+_3 !"ℎ$%$&c it+ _nenit+ fit (2) g&−"$iU&": Zit= ]i+_1 !`$1it+_2 !`$3it+_nenit+ fit (3) P%QR ℎSTUVWX: Zit=]i+_1 !`$1it+_2 !`$3it+_3 !`$1it∗!"ℎ$%$&c it+ _4 !`$2it∗ !"ℎ$%$&c it+_5 !`$3it∗!"ℎ$%$&c it+_nenit+ fit (4) where Zit was the 6moBR for person W at time c, ]i was the RE accounting for all unobserved time-invariant characteristics for person W at time c, !`$1it , !`$2it , and !`$3it were dummy variables indicating whether person W at time c belong to age group 1 (children), 2 (adolescents), or 3 (adults), !"ℎ$%$&c it was a dummy variable indicating whether person W at time c was adherent to treatment, fit was the residual term, and enit indicated the &th covariate for person W at time c selected using univariate analyses (P-value=0.10) Specifically, Model 1 analyzed the association in prophylactic patients, with adolescents non- adherent to prophylaxis being the reference group. Model 2 estimated the association in on- demand patients, with adolescents being the reference group. Adherence was not included since the proxy measure of adherence to on-demand regimen was only exploratory and not fully validated. Model 3 assessed the association in prophylactic patients including interaction between age and adherence, with adolescents non-adherent to prophylaxis being the reference group. The 62 interaction effects were included to assess whether the impact of transition from childhood to adulthood on bleeding differed by adherence to prophylaxis. Data processing and descriptive analyses were conducted using SAS software version 9.4. Univariate and regression analyses were performed using Stata software version 14.0 (StataCorp LP, College Station, TX, USA). 4.3 Results 4.3.1 Sample characteristics Adolescents, children, and adults had significantly different clinical, insurance-related, and sociodemographic characteristics at HUGS Va baseline (Table 4.1). The percentage of patients having moderate-to-severe joint pain was highest in adults (73.1%), followed by adolescents (43.8%), and children (31.3%). The proportion of patients having private insurance was lowest (31.3%) in adolescents compared to children and adults. Most of the differences across age groups were muted at HUGS LTS survey (Table 4.2) except the proportion of patients with moderate-to- severe joint pain, household income above $75,000, and having private insurance. 4.3.2 Temporal trends in health outcomes Temporal trends in 6moBR across age groups differed between prophylactic and on-demand populations. In HUGS Va, median 6moBR was lowest in children and similar between adolescents and adults receiving prophylaxis (Figure 4.1a), but was highest in adults, followed by adolescents and children receiving on-demand treatment (Figure 4.1b). Regardless of treatment regimen, 63 adolescents generally had the highest factor utilization in HUGS Va; the lowest factor utilization in prophylactic population was observed in children (Figure 4.2a) and that in on-demand population was observed in in adults (Figure 4.2b). The proportion of patients adhering to prophylaxis was highest in adolescents and lowest in children in HUGS Va (Figure 4.3). The proportion of patients receiving prophylaxis was consistently lowest in adults and similar between adolescents and children in HUGS Va and HUGS LTS (Figure 4.4). Interestingly, differences between adolescents and children in bleeding, factor utilization, and adherence to treatment were in the opposite directions in HUGS LTS compared to HUGS Va. Temporal trend of adherence to on-demand regimen was not plotted, since the exploratory measure resulted in excessive amount of out-of-range values (>200%) and a majority of patients identified as non-adherent. 4.3.3 Multivariable regression analyses Variables of primary interest were age and adherence (Models 1 and 3, Eq. 2 and 4 respectively), and the interaction between the two (Model 3, Eq. 4). Additionally, having moderate-to-severe joint pain and having private insurance were selected into the multivariable regression models based on univariate analyses. Across three models, 6moBR in adolescents was higher compared to that in children, but lower compared to that in adults (Table 4.3). Adhering to prophylaxis and having private insurance were associated with reduced 6moBR, having moderate- to-severe joint pain was associated with increased 6moBR. Without accounting for interactions between age and adherence, 6moBR in children was 16% lower (P-value>0.05), and that in adults was 43% higher (P-value>0.05) compared to adolescents in prophylactic population; 6moBR in prophylactic patients with moderate-to-severe joint pain was 63% higher (P-value=0.01) compared to those with little-to-none joint pain; 6moBR in patients having private insurance was 40% lower 64 (P-value=0.00) compared to those having public insurance. Accounting for interactions between age and adherence, the differences in 6moBR between adolescents and children were muted and not statistically significant (P-value>0.05), however the differences between adolescents and adults were magnified, with adults having 203% higher 6moBR (P-value=0.00) compared to adolescents. The effects of having moderate-to-severe joint pain and private insurance were similar with and without accounting for interactions between age and adherence. Adhering to prophylaxis was associated with 14% reduction (P-value>0.05) in 6moBR in prophylactic patients. Moreover, the interaction effects illustrated the differential impact of adherence on bleeding between adolescents, children, and adults. Adhering to prophylaxis was associated with 51% reduction (P-value=0.01) in 6moBR among adults, and 29% increase (P- value>0.05) in 6moBR among adolescents, and minimal impact on bleeding in children. However, results should not be considered conclusive for adolescents and children, since the P-values did not reach statistical significance (P-value>0.05) and 95% confidence intervals (CIs) were wide (0.61-1.84 in children and 0.69-2.39 in adolescents). In on-demand population, differences in 6moBR between adolescents, children, and adults were less pronounced as in prophylactic population. Children had 12% lower (P-value>0.05), and adults had 10% higher (P-value>0.05) 6moBR compared to adolescents. As opposed to prophylactic population, the associations between bleeding and having severe joint pain and private insurance were muted in on-demand population and did not reach statistical significance (P-value>0.05). 65 4.4 Discussion and conclusion Hemophilia in transition from childhood to adulthood face numerous clinical and psychosocial challenges, but how and how and to what extent it will have an impact on health outcomes were unknown. Our study was the first to analyze trends in important health outcomes such as bleeding, factor utilization, treatment regimen, and adherence to treatment in patients who were in transition from childhood to adulthood, and to examine the impact of transition on bleeding using multivariable regression analyses. Our analyses made use of longitudinal data spanning across seven up to ten years in a sample of hemophilia A patients from geographically diverse HTCs in the USA. As hypothesized, adolescents in transition from childhood to adulthood had higher bleeding and factor utilization compared to children regardless of treatment regimen. Adolescents receiving prophylaxis had highest bleeding despite highest factor utilization among all age groups, possibly due to the fact that adolescents did not infuse the factors as dispensed either because they forgot or were unwilling to infuse. This finding is alarming, since bleeding was not sufficiently controlled for by using a large amount of factors dispensed, resulting in increased burden of illness on patients and also payers. Therefore, policy makers, payers, and providers should design and utilize alternative approaches to closely monitoring factor consumption and to improving adherence in patients in transition from childhood to adulthood. In contrary to our hypothesis, adolescents did not have higher rates of bleeding compared to adults regardless of treatment regimen. In prophylactic patients, adolescents and adults had similar rate of bleeding; in on-demand patients, adolescents had lower bleeding compared to adults. Our speculation is that adults are at higher risks of bleeding in nature since damages caused by bleeding are accumulative over time. Another finding contradicting our hypothesis was that the proportion 66 of patients adhering to prophylaxis was highest in adolescents among all age groups which may be due to the fact that adherence was calculated using a proxy measure not accurately reflect the extent to which factors were utilized as prescribed. Interestingly, differences between adolescents and children in bleeding, factor utilization, and adherence to prophylaxis were in opposite directions between HUGS Va and HUGS LTS, possibly due to anomalies associated with the one- time collection of self-reported bleeding data, and data loss due to retrospective collection of medical charts and dispensing records in HUGS LTS. Therefore, we deemed that the temporal trends observed in HUGS Va should be a more reliable representation of the differences across age groups compared to HUGS LTS. Multivariable regression analyses confirmed that regardless of treatment regimen, transition from childhood to adulthood was associated with higher bleeding compared to childhood but not compared to adulthood. In prophylactic population, we observed differential impact of adhering to treatment on bleeding by age group. Adhering to prophylaxis resulted in significant reduction in bleeding in adults, but not in adolescents and children, which may be due to small sample sizes of the adolescent and child samples compared to the adult sample, or issues with the proxy measure of adherence, or that adhering to prophylaxis is less important to adolescents and children as to adults who are more likely to bleed. However, our findings should be interpreted with several limitations. First, although the sample size was reasonable for hemophilia studies, it was small for multivariable regression analyses, especially the adolescent and child samples. Second, bleeding and treatment regimen was based on patient-reported data which were subject to recall and reporting bias. Third, HUGS LTS collected patient-reported data only one-time, and retrospectively collected medical charts and dispensing records within six months prior to enrollment, which may lead to biased results 67 compared to HUGS Va which collected data multiple times prospectively. Lastly, our study was not designed to trace whether patients infused factors as dispensed, which should be a more accurate reflection of adherence compared to the proxy measures based on dispensing records. In summary, our study was the first to demonstrate that in a USA multi-center sample of persons with hemophilia A, transition from childhood to adulthood was associated with increased bleeding compared to childhood, but not compared to adulthood. Bleeding was not sufficiently controlled for in adolescent hemophilia patients although a substantial amount of factors were consumed. Actions should be taken by policy makers, payers, and providers to design and implement interventions tailored to monitor and improve health outcomes in hemophilia patients in transition from childhood to adulthood. 68 4.5 Chapter references 1 Centers for Disease Control and Prevention. Hemophilia Data & Statistics. 2 Manco-Johnson MJ et al. Consequences of switching from prophylactic treatment to on- demand treatment in late teens and early adults with severe haemophilia A: the TEEN/TWEN study. Haemophilia 2013; 19: 727–35. 3 Walsh CE et al. Factor VIII prophylaxis for adult patients with severe haemophilia A: results of a US survey of attitudes and practices. Haemophilia 2009; 15: 1014–21. 4 Lindvall K et al. Compliance with treatment and understanding of own disease in patients with severe and moderate haemophilia. Haemophilia 2006; 12: 47–51. 5 Quon D et al. Unmet needs in the transition to adulthood: 18- to 30-year-old people with hemophilia. Am J Hematol 2015; 90: S17–22. 6 Patient Protection and Affordable Care Act, 42 U.S.C. § 18001 (2010). 7 Nazzaro A-M et al. Knowledge, Attitudes, and Behaviors of Youths in the US Hemophilia Population: Results of a National Survey. Am J Public Health 2006; 96: 1618–22. 8 The American Thrombosis and Hemostasis Network (ATHN). Action Guide for Improving Care for People with Bleeding Disorders. 2016. 9 Duncan N et al. Treatment patterns, health-related quality of life and adherence to prophylaxis among haemophilia A patients in the United States. Haemophilia 2012; 18: 760–5. 10 Geraghty S et al. Practice patterns in haemophilia A therapy – global progress towards optimal care. Haemophilia 2006; 12: 75–81. 11 van Dijk K et al. Can long-term prophylaxis for severe haemophilia be stopped in adulthood? Results from Denmark and the Netherlands. Br J Haematol 2005; 130: 107–12. 12 Fischer K et al. Discontinuation of prophylactic therapy in severe haemophilia: incidence and effects on outcome. Haemophilia 2001; 7: 544–50. 13 Chen CX. The impact of treatment decisions and adherence on outcomes in small hereditary disease populations (Doctoral dissertation). Retrieved from the University of Southern California Digital Library Database. 2016. 14 Dunbar-Jacob J et al. Treatment adherence in chronic disease. J Clin Epidemiol 2001; 54: S57– 60. 15 Ozelo M et al. Long-term patterns of safety and efficacy of bleeding prophylaxis with turoctocog alfa (NovoEight®) in previously treated patients with severe haemophilia A: interim results of the guardian TM 2 extension trial. Haemophilia 2015; 21: e436–9. 69 16 Shapiro AD et al. The safety and efficacy of recombinant human blood coagulation factor IX in previously untreated patients with severe or moderately severe hemophilia B. Blood 2005; 105: 518–25. 17 Du Treil S et al. Quantifying adherence to treatment and its relationship to quality of life in a well-characterized haemophilia population. Haemophilia 2007; 13: 493–501. 18 Johson MJ et al. Child Development with a Bleeding Disorder and Transition. National Hemophilia Foundation; 2013. 19 Den Uijl IEM et al. Analysis of low frequency bleeding data: the association of joint bleeds according to baseline FVIII activity levels. Haemophilia 2011; 17: 41–4. 20 Frees EW. Longitudinal and Panel Data. Cambridge University Press; 2004. 21 Hausman JA. Specification Tests in Econometrics. Econometrica 1978; 46: 1251–71. 70 Table 4.1 Sample characteristics by age group at HUGS Va baseline Study group Variables, N (%) Total Children (aged 0-11 years) Adolescents (aged 12-20 years) Adults (aged 21+ years) P-value [4] Number of patients [1] 74 32 16 26 Prophylaxis [2] 41 (56.2) 23 (74.2) 10 (62.5) 8 (30.8) 0.004 Moderate to severe joint pain 36 (48.6) 10 (31.3) 7 (43.8) 19 (73.1) 0.006 Insured for the entire year 66 (89.2) 32 (100.0) 15 (93.8) 19 (73.1) 0.004 Private insurance [2] 42 (60.9) 24 (75.0) 5 (31.3) 13 (61.9) 0.01 Married [3] 45 (60.8) 27 (84.4) 9 (56.3) 9 (34.6) 0.001 Education (college or above) [2,3] 53 (72.6) 27 (84.4) 8 (50.0) 18 (72.0) 0.04 Employed [3] 53 (71.6) 26 (81.3) 10 (62.5) 17 (65.4) 0.27 Annual household income above $75,000 [2] 21 (28.8) 16 (50.0) 3 (18.8) 2 (8.0) 0.001 [1] Patients were included if they did not have inhibitors at any point in HUGS Va and HUGS LTS, and their dispensing records were available at HUGS Va initial survey. [2] Numbers did not sum to total number of patients in each age category due to missing data. Column percentages were computed using non-missing data. [3] For adults≥18 years or parents of children<18 years. [4] P-values were calculated using Pearson’s chi-square tests. 71 Table 4.2 Sample characteristics by age group at HUGS LTS baseline Study group Variables, N (%) Total Children (aged 0-11 years) Adolescents (aged 12-20 years) Adults (aged 21+ years) P-value [4] Number of patients [1] 79 3 29 47 Prophylaxis [2] 60 (78.9) 3 (100.0) 23 (85.2) 34 (73.9) 0.34 Moderate to severe joint pain 47 (59.5) 1 (33.3) 12 (41.4) 34 (72.3) 0.02 Insured for the entire year 73 (92.4) 3 (100.0) 27 (93.1) 43 (91.5) 0.85 Private insurance [2] 47 (61.8) 0 (0.0) 23 (82.1) 24 (53.3) 0.004 Married [2,3] 39 (50.0) 2 (66.7) 11 (39.3) 26 (55.3) 0.34 Education (college or above) [2,3] 75 (97.4) 3 (100.0) 28 (100.0) 44 (95.7) 0.50 Employed [2,3] 49 (63.6) 3 (100.0) 16 (57.1) 30 (65.2) 0.32 Annual household income above $75,000 [2] 21 (29.6) 1 (33.3) 13 (46.4) 7 (17.5) 0.04 [1] Patients were included if they did not have inhibitors at any point in HUGS Va and HUGS LTS, and their dispensing records were available at HUGS LTS survey. [2] Numbers did not sum to total number of patients in each age category due to missing data. Column percentages were computed using non-missing data. [3] For adults≥18 years or parents of children<18 years. [4] P-values were calculated using Pearson’s chi-square tests. 72 Table 4.3 Random-effect negative binomial models for the association between transition and bleeding Variables Six-month bleeding rate Model 1: Prophylaxis, without interactions Model 2: On-demand, without interactions Model 3: Prophylaxis, with interactions IRR (95% CI) P-value IRR (95% CI) P-value IRR (95% CI) P-value Age 0-11 years 0.84 (0.57, 1.24) 0.38 0.88 (0.47, 1.63) 0.69 0.97 (0.48, 1.97) 0.94 Age 21+ years 1.43 (0.98, 2.10) 0.07 1.10 (0.62, 1.93) 0.75 3.03 (1.42, 6.45) 0.00 Adherent 0.86 (0.62, 1.20) 0.38 - - - - Severe joint pain 1.63 (1.12, 2.38) 0.01 1.27 (0.80, 2.01) 0.30 1.60 (1.09, 2.34) 0.02 Private insurance 0.60 (0.43, 0.84) 0.00 0.94 (0.60, 1.47) 0.79 0.59 (0.41, 0.83) 0.00 Interactions (age & adherence) Age 0-11 years & adherent - - - - 1.06 (0.61, 1.84) 0.84 Age 12-20 years & adherent - - - - 1.29 (0.69, 2.39) 0.42 Age 21+ years & adherent - - - - 0.49 (0.28, 0.84) 0.01 Constant 1.85 (0.97, 3.50) 0.06 2.07 (0.93, 4.60) 0.07 1.58 (0.75, 3.33) 0.23 Number of observations 207 135 207 Number of patients 65 52 65 Abbreviation: IRR, incident rate ratio; CI, confidence interval. 73 Figure 4.1a. Temporal trend in six-month bleeding rate (6moBR) in prophylactic patients Study period Age group HUGS Va Mo 1-6 HUGS Va Mo 7-12 HUGS Va Mo 13-18 HUGS Va Mo 19-24 HUGS LTS Mo -6 to 0 Children 1.0 1.0 0.0 1.5 4.0 Adolescents 2.0 4.0 2.5 3.0 2.0 Adults 1.0 4.0 2.5 3.0 3.0 74 Figure 4.1b. Temporal trend in six-month bleeding rate (6moBR) in on-demand patients Study period Age group HUGS Va Mo 1-6 HUGS Va Mo 7-12 HUGS Va Mo 13-18 HUGS Va Mo 19-24 HUGS LTS Mo -6 to 0 Children 2.0 3.5 1.0 1.5 0.0 Adolescents 2.5 3.5 1.0 3.0 2.0 Adults 6.5 9.0 7.0 5.0 2.0 75 Figure 4.2a. Temporal trend in six-month factor utilization in prophylactic patients Study period Age group HUGS Va Mo 1-6 HUGS Va Mo 7-12 HUGS Va Mo 13-18 HUGS Va Mo 19-24 HUGS LTS Mo -6 to 0 Children 3194 2607 1030 829 4771 Adolescents 1686 2612 3043 3747 3000 Adults 2357 2438 2380 2261 2545 76 Figure 4.2b. Temporal trend in six-month factor utilization in on-demand patients Abbreviation: N/A, not applicable. Note: No children (aged 0-11 years) reported as receiving on-demand treatment at HUGS LTS survey. Study period Age group HUGS Va Mo 1-6 HUGS Va Mo 7-12 HUGS Va Mo 13-18 HUGS Va Mo 19-24 HUGS LTS Mo -6 to 0 Children 1841 2391 1840 937 N/A Adolescents 526 1830 2727 1868 833 Adults 594 857 789 413 992 77 Figure 4.3 Temporal trend in the proportion of patients adhering to prophylaxis Study period Age group HUGS Va Mo 1-6 HUGS Va Mo 7-12 HUGS Va Mo 13-18 HUGS Va Mo 19-24 HUGS LTS Mo -6 to 0 Children 78 71 47 50 100 Adolescents 50 86 92 100 70 Adults 75 67 75 70 76 78 Figure 4.4 Temporal trend in the proportion of patients on prophylaxis Study period Age group HUGS Va Mo 1-6 HUGS Va Mo 7-12 HUGS Va Mo 13-18 HUGS Va Mo 19-24 HUGS LTS Mo -6 to 0 Children 63 62 70 71 80 Adolescents 58 54 82 77 76 Adults 23 19 28 33 61 79 CHAPTER 5. Conclusion Hemophilia is a rare chronic condition posing substantial disease burden on patients and caregivers and disproportionately high financial burden on payers and the society. Characterization of health outcomes is critical for evaluating the cost-effectiveness of various existing therapies and programs and designing new treatment modalities to address the unmet need in the hemophilia population. This dissertation aimed to shed light on the characterization and measurement of health outcomes in hemophilia using psychometric and health economic analyses. Specifically, two studies assessed important psychometric properties for the Validated Hemophilia Regimen Treatment Adherence Scale – Prophylaxis (VERITAS-Pro), the only standardized measure of adherence to prophylaxis in hemophilia to date. The third study was performed to evaluate the impact of transition from childhood to adulthood on health outcomes in persons with hemophilia A. Chapter 2 identified several areas for improvement for VERITAS-Pro, including construct validity, rating scale performance, and discriminative ability, and proposed a modified version (mVERITAS-Pro) using psychometric analyses. Chapter 3 identified a lack of concurrent and predictive validities for VERITAS-Pro and mVERITAS-Pro against criterion measures calculated using factor prescription and dispensing records. Chapter 4 found that transition from childhood to adulthood led to increased bleeding compared to childhood, and increased factor utilization compared to childhood and adulthood using a longitudinal analysis of hemophilia A patients. In conclusion, findings from this dissertation work shed light on the psychometric properties for VERITAS-Pro and suboptimal health outcomes in hemophilia patients in transition from childhood to adulthood. Future research is needed to further assess the concurrent and predictive validities and determine the cut-off score for mVERITAS-Pro using validated criterion measures, 80 and to test the utility of mVERITAS-Pro in different patient subpopulations by age, race/ethnicity, , and disease severity. More importantly, collaborative efforts need to be made by patients, clinicians, and psychometricians to generate and validate additional items with higher ability to discern patients at varying levels of adherence. Finally, actions should be taken to design and implement interventions which could improve adherence and reduce disease burden in hemophilia patients, and particularly those who are in transition from childhood to adulthood. 81 APPENDIX Appendix 1. Modified VERITAS-Pro Modified VERITAS-Pro Managing hemophilia is a challenging task. The questions below ask about how you manage hemophilia and prophylaxis. We’d like to get an idea of how often you have done each of these things in the past three months. There are no right or wrong answers. The most important thing is for you to answer each question as honestly as possible. Please answer each question using the following scale: Always – all of the time, 100% of the time Sometimes – occasionally, at least 50% of the time Never – not at all, 0% of the time 1. I do prophylaxis infusions on the scheduled days. Always Sometimes Never 2. I infuse the recommended number of times per week. Always Sometimes Never 3. I do infusions according to the schedule provided by the treatment center. Always Sometimes Never 4. I use the doctor-recommended dose for infusions. Always Sometimes Never 5. I increase or decrease the dose without calling the treatment center. Always Sometimes Never 6. I plan ahead so I have enough factor at home. Always Sometimes Never 7. I keep close track of how much factor and how many supplies I have at home. Always Sometimes Never 82 8. I forget to do prophylaxis infusions. Always Sometimes Never 9. Remembering to do prophylaxis is difficult. Always Sometimes Never 10. I remember to infuse on the schedule prescribed by the treatment center. Always Sometimes Never 11. I miss recommended infusions because I forget about them. Always Sometimes Never 12. I skip prophylaxis infusions. Always Sometimes Never 13. I choose to infuse less often than prescribed. Always Sometimes Never 14. If it is inconvenient to infuse, I skip the infusion that day. Always Sometimes Never 15. I miss recommended infusions because I skip them. Always Sometimes Never 16. I call the treatment center when I have questions about hemophilia or treatment. Always Sometimes Never 17. I call the treatment center when I have hemophilia-related health concerns or when changes occur. Always Sometimes Never 18. I make treatment decisions myself rather than calling the hemophilia center. Always Sometimes Never 83 Appendix 2. Supplemental materials for Chapter 4 Table 1a. Distribution of bleeds in HUGS Va Month 1-6, overall sample (N=103) Table 1b. Distribution of bleeds in HUGS Va Month 1-6, prophylaxis sample (N=47) Table 1c. Distribution of bleeds in HUGS Va Month 1-6, on-demand sample (N=54) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 1 2 3 4 5 6 7 8 9 10 11 12 15 16 17 18 19 23 24 25 Proportion of patients (%) No. bleeds in HUGS Va Month 1-6 0.00 0.10 0.20 0.30 0.40 0 1 2 3 4 5 8 9 12 23 24 25 Proportion of patients (%) No. bleeds in HUGS Va Month 0-6 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 1 2 3 4 5 6 7 8 9 10 11 12 15 16 17 18 19 20 21 Proportion of patients (%) No. bleeds in HUGS Va Month 0-6 84 Table 2a. Distribution of bleeds in HUGS Va Month 7-12, overall sample (N=87) Table 2b. Distribution of bleeds in HUGS Va Month 7-12, prophylaxis sample (N=37) Table 2c. Distribution of bleeds in HUGS Va Month 7-12, on-demand sample (N=38) 0.00 0.05 0.10 0.15 0.20 0.25 0 1 2 3 4 5 7 8 9 10 11 13 14 15 16 17 18 19 20 29 31 32 33 34 Proportion of patients (%) No. bleeds in HUGS Va Month 7-12 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 1 2 3 4 5 7 9 10 11 19 29 32 33 34 Proportion of patients (%) No. bleeds in HUGS Va Month 7-12 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0 1 2 3 4 5 7 8 9 10 13 14 15 16 17 18 20 31 32 33 Proportion of patients (%) No. bleeds in HUGS Va Month 7-12 85 Table 3a. Distribution of bleeds in HUGS Va Month 13-18, overall sample (N=80) Table 3b. Distribution of bleeds in HUGS Va Month 13-18, prophylaxis sample (N=43) Table 3c. Distribution of bleeds in HUGS Va Month 13-18, on-demand sample (N=25) 0.00 0.10 0.20 0.30 0.40 0.50 0 1 2 3 4 5 6 7 8 9 10 11 13 15 19 27 28 29 Proportion of patients (%) No. bleeds in HUGS Va Month 13-18 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0 1 2 3 4 5 6 8 9 19 20 21 Proportion of patients (%) No. bleeds in HUGS Va Month 13-18 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0 1 2 3 4 7 8 9 10 11 13 15 27 28 29 Proportion of patients (%) No. bleeds in HUGS Va Month 13-18 86 Table 4a. Distribution of bleeds in HUGS Va Month 19-24, overall sample (N=88) Table 4b. Distribution of bleeds in HUGS Va Month 19-24, prophylaxis sample (N=50) Table 4c. Distribution of bleeds in HUGS Va Month 19-24, on-demand sample (N=23) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 1 2 3 4 6 7 8 10 11 12 14 17 18 20 24 28 29 30 Proportion of patients (%) No. bleeds in HUGS Va Month 19-24 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 1 2 3 4 6 7 8 10 11 12 14 18 20 24 25 26 Proportion of patients (%) No. bleeds in HUGS Va Month 19-24 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0 1 2 3 4 6 7 8 10 12 14 17 28 29 30 Proportion of patients (%) No. bleeds in HUGS Va Month 19-24 87 Table 5a. Distribution of bleeds in HUGS LTS Month -6 to 0, overall sample (N=107) Table 5b. Distribution of bleeds in HUGS LTS Month -6 to 0, prophylaxis sample (N=72) Table 5c. Distribution of bleeds in HUGS LTS Month -6 to 0, on-demand sample (N=29) 0.00 0.05 0.10 0.15 0.20 0.25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 15 20 35 52 54 75 76 77 Proportion of patients (%) No. bleeds in LTS Month -6 to 0 0.00 0.05 0.10 0.15 0.20 0.25 0 1 2 3 4 5 6 7 8 9 10 11 12 15 20 35 52 75 76 77 Proportion of patients (%) No. bleeds in LTS Month -6 to 0 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 1 2 3 5 6 8 9 15 54 55 56 Proportion of patients (%) No. bleeds in LTS Month -6 to 0

Abstract (if available)

Abstract Hemophilia is a hereditary blood disorder characterized by deficiencies in coagulation factors VIII (hemophilia A) and IX (hemophilia B), which imposed significant disease burden on patients, caregivers, and the society. Bleeding is the most important clinical manifestation of hemophilia. Repeated bleeding into the joints leads to irreversible joint damage, impaired health outcomes, and increased health resource utilization. Over the past decades, hemophilia has been treated with factor replacement therapies which replenish the deficient coagulation factors through intravenous infusions on either prophylactic (or prophylaxis) or episodic (or on-demand) regimen. Prophylaxis administers factors regularly to prevent anticipated bleeding

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Three essays on emerging issues in hemophilia care

PDF

Discriminating changes in health using patient-reported outcomes

PDF

Burden of illness in hemophilia A: taking the patient’s perspective

PDF

Essays on opioid prescribing and abatement

PDF

Economic, clinical, and behavioral outcomes from medical and pharmaceutical treatments

PDF

The value of novel antihyperlipidemic treatments in the U.S. healthcare system: Reducing the burden of cardiovascular diseases and filling the gap of low adherence in statins

PDF

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

PDF

Value in health in the era of vertical integration

PDF

Three essays on estimating the effects of government programs and policies on health care among disadvantaged population

PDF

Treatment of hepatitis C in era of direct acting antivirals (DAAs): selected essays in health economics

PDF

Essays in health economics

PDF

Essays in pharmaceutical and health economics

PDF

Application of concepts from prospect theory and response shift when estimating the minimally important difference for health utility measures

PDF

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

PDF

Impact of medication persistence on clinical and health care cost outcomes in patients with major depressive disorder using retrospective claims data

PDF

A series of longitudinal analyses of patient reported outcomes to further the understanding of care-management of comorbid diabetes and depression in a safety-net healthcare system

PDF

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

PDF

Essays in health economics and provider behavior

PDF

Outcomes of antibiotic use among children with acute respiratory tract infections

Asset Metadata

Creator Ding, Yuchen (author)

Core Title Characterization of health outcomes in patients with hemophilia A and B: Findings from psychometric and health economic analyses

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Health Economics

Publication Date 07/26/2018

Defense Date 06/18/2018

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

Tag adherence,Health Economics,hemophilia,hemophilia utilization group studies,OAI-PMH Harvest,psychometrics,rare disease,Rasch analysis

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Nichol, Michael B. (committee chair), Doctor, Jason (committee member), Myerson, Rebecca (committee member), Romley, John A. (committee member), Wu, Shinyi (committee member)

Creator Email erinding27@gmail.com,yuchendi@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-35549

Unique identifier UC11671745

Identifier etd-DingYuchen-6527.pdf (filename),usctheses-c89-35549 (legacy record id)

Legacy Identifier etd-DingYuchen-6527.pdf

Dmrecord 35549

Document Type Dissertation

Format application/pdf (imt)

Rights Ding, Yuchen

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