University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

THE IMPACT OF TREATMENT DECISIONS AND ADHERENCE ON OUTCOMES IN
SMALL HEREDITARY DISEASE POPULATIONS

By Christina Xiaoyue Chen

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

August 2016
2
Dedication

To my family:

Brian,
For being my rock, for constantly inspiring me, for loving me more than I could have ever
imagined, and for never failing to make me laugh;

My parents,
For making all of this possible;

My little buddy, Kaylen,
For always reminding me to appreciate the small things;

And Marge and Gary,
For loving me like one of their own.

3
Acknowledgements
First and foremost, I would like to thank my advisor, Professor Michael Nichol, for
encouraging my growth as an independent researcher while imparting invaluable knowledge and
guidance throughout the last few years. I am truly grateful for his mentorship and all he has
taught me about health economics, hemophilia, the academic writing process, and external
collaboration. So much of what I have accomplished would not have been possible without his
wisdom, support, and encouragement.
This work was also shaped by the incredible knowledge and guidance from the professors
who have supported my research and served on my qualifying exam and dissertation committees.
I am grateful to Professor Joel Hay for his contributions and input on the cost-effectiveness
model included in this dissertation. He has taught me a tremendous amount about
pharmacoeconomics, as well as challenged me to conduct and publish high quality research.
Thank you to Professor Jason Doctor for serving as co-chair of my dissertation committee and
providing thoughtful comments on my research. Additionally, thank you to Professor Geoffrey
Joyce for always pushing me to consider the “big picture.” Finally, thank you to Professors
Jeffrey McCombs and Shinyi Wu for astute input on my dissertation proposal.
More than words, I am also grateful to Professor McCombs, the beloved Dr. J, for always
having an open door, an open ear, and an open heart for his students. This graduate program is
one of the most respected health economics programs in the country in part because of his
dedication to the department and students. I will cherish and miss all our laughs and our
discussions on health economics, baseball, travel, life, and our feline friends.
Much of this dissertation focuses on research in hemophilia. For this, I must acknowledge
the entire HUGS group and hemophilia community. The members of the Steering Committee
4
fully accepted me into their close circle and provided insightful and candid guidance on my
presentations and manuscripts throughout the years. The study coordinators and principal
investigators have incredible passion for their work and made the HUGS project possible. Mimi
Lou was there, day in and day out, to help me with questions and requests no matter how big or
small. Additionally, Joanne Wu, Dr. Zheng-Yi Zhou, and Xiaoli Niu were always willing to
provide their analytical and subject matter expertise, time, and advice. Most importantly, thank
you to all the HUGS participants and other individuals with hemophilia who have touched my
life and inspired me with their devotion, resilience, and courage. Thank you for sharing your
personal journeys and motivating me to conduct research with real impact to improve lives.
Working as a pre-doctoral fellow at Biogen while defending and completing my
dissertation has been one of the most challenging, unique, and rewarding experiences. Thank you
to everyone at Biogen, especially the Health Economics and Outcomes Research team, for the
opportunity to learn, lead, and take on responsibilities beyond what is normally expected of a
graduate student. Thank you to Dr. Adi Eldar-Lissai and Dr. Sangeeta Krishnan for mentoring
me and filling the last two years with tremendous learnings outside of academic research. In
particular, I am grateful to Adi for teaching me the ins and outs of a professional career, for
guiding my decision in post-USC endeavors, for being more than generous with her time and
advice, and for introducing me to all the best food around Cambridge.
Finally, thank you also to the friends and teachers who have made my last 8 years as an
undergraduate and graduate student at USC a worthwhile and unforgettable experience. Dr.
Christin Thompson has been an incredibly helpful and supportive friend, without whom I would
never have entered into this graduate program nor made it through with most of my hair intact.
Also, all of the students in my program, including Justin McGinnis, Cynthia Gong, Darshan
5
Mehta, Yuchen Ding, and Dr. Emmanuel Drabo, were the most intelligent, encouraging, and fun
comrades a graduate student could hope for.
It is amazing to look back on all the people and accomplishments that have led me to
where I am today both personally and professionally. To all the individuals who have helped me
close this chapter of my life and develop the skills to succeed in the new chapters open before
me, thank you and Fight On!
6
Table of Contents
Dedication ..................................................................................................................................................... 2
Acknowledgements ....................................................................................................................................... 3
Table of Contents .......................................................................................................................................... 6
List of Tables ................................................................................................................................................ 9
List of Figures ............................................................................................................................................. 10
ABSTRACT ................................................................................................................................................ 11
CHAPTER 1: Introduction ......................................................................................................................... 13
1.1. Background ...................................................................................................................................................... 13
1.2. Overview of papers .......................................................................................................................................... 19
1.3. Chapter references ........................................................................................................................................... 20
CHAPTER 2: Burden of Illness among Persons with Hemophilia B: Examining the Associations of
Severity and Treatment Regimens with Costs and Annualized Bleed Rates .............................................. 23
ABSTRACT ........................................................................................................................................................... 23
2.1. Introduction ...................................................................................................................................................... 25
2.1.1. Background .............................................................................................................................................. 25
2.1.2. Objectives................................................................................................................................................. 27
2.2. Methods ........................................................................................................................................................... 27
2.2.1. Hemophilia Utilization Group Studies Part Vb ........................................................................................ 27
2.2.2. Determination of direct costs ................................................................................................................... 28
2.2.3. Determination of indirect costs ................................................................................................................ 29
2.2.4. Statistical analyses ................................................................................................................................... 30
2.3. Results .............................................................................................................................................................. 30
2.3.1. Baseline characteristics ............................................................................................................................ 30
2.3.2. Healthcare services utilization, dispensing, and work productivity losses ............................................... 34
7
2.3.3. Bleeding episodes..................................................................................................................................... 36
2.3.4. Direct and indirect costs ........................................................................................................................... 36
2.3.5. Costs by participant subgroups ................................................................................................................ 36
2.4. Discussion and limitations ............................................................................................................................... 39
2.4.1. Discussion ................................................................................................................................................ 39
2.4.2. Limitations ............................................................................................................................................... 41
2.5. Conclusions ...................................................................................................................................................... 41
2.6. Chapter references ........................................................................................................................................... 43
CHAPTER 3: The Impact of Adherence to Prophylactic Clotting Factor Replacement Therapy on
Bleeding Episodes among Persons with Hemophilia in the United States ................................................. 46
ABSTRACT ........................................................................................................................................................... 46
3.1. Introduction ...................................................................................................................................................... 48
3.1.1. Background .............................................................................................................................................. 48
3.1.2. Objectives................................................................................................................................................. 50
3.2. Data and methods ............................................................................................................................................. 51
3.2.1. Data sources ............................................................................................................................................. 51
3.2.2. Methodology ............................................................................................................................................ 53
3.2.3. Statistical analyses ................................................................................................................................... 57
3.3. Results .............................................................................................................................................................. 62
3.3.1. Baseline participant characteristics .......................................................................................................... 62
3.3.2. Descriptive results .................................................................................................................................... 64
3.3.3. Negative binomial regression results ....................................................................................................... 67
3.3.4. Sensitivity analyses .................................................................................................................................. 74
3.3.5. Matched cohort sensitivity analysis ......................................................................................................... 75
3.4. Discussion and limitations ............................................................................................................................... 78
3.4.1. Discussion ................................................................................................................................................ 78
3.4.2. Limitations ............................................................................................................................................... 81
3.5. Conclusions ...................................................................................................................................................... 83
8
3.6. Chapter references ........................................................................................................................................... 84
Appendix A. Distribution of annualized adherence ................................................................................................ 87
Appendix B. Distribution of annualized bleed rate ................................................................................................. 88
Appendix C. Distribution of independent variables after matching ........................................................................ 89
CHATPER 4: Cost-effectiveness Analysis of Alternative Screening and Treatment Strategies for
Heterozygous Familial Hypercholesterolemia in the United States............................................................ 90
ABSTRACT ........................................................................................................................................................... 90
4.1. Introduction ...................................................................................................................................................... 92
4.1.1. Background .............................................................................................................................................. 92
4.2.1. Objectives................................................................................................................................................. 94
4.2. Methods ........................................................................................................................................................... 94
4.2.1. The model ................................................................................................................................................ 94
4.2.2. Transition probabilities ............................................................................................................................ 98
4.2.3. Health-state utilities ............................................................................................................................... 100
4.2.4. Costs ....................................................................................................................................................... 101
4.2.5. Sensitivity analysis ................................................................................................................................. 104
4.3. Results ............................................................................................................................................................ 105
4.3.1. Cost-effectiveness ratios ........................................................................................................................ 105
4.3.2. Sensitivity analysis results ..................................................................................................................... 106
4.4. Discussion and limitations ............................................................................................................................. 109
4.4.1. Discussion .............................................................................................................................................. 109
4.4.2. Study limitations .................................................................................................................................... 111
4.5. Conclusions .................................................................................................................................................... 112
4.6. Chapter references ......................................................................................................................................... 113
CHAPTER 5: Conclusion ......................................................................................................................... 117

9
List of Tables
Table 2.1. Baseline characteristics among excluded and included participants with hemophilia B ............................ 32
Table 2.2. Baseline characteristics among included participants by hemophilia B severity ........................................ 33
Table 2.3. Annual healthcare resource utilization, work productivity loss, and bleeding episodes among
individuals with hemophilia B ..................................................................................................................................... 35
Table 2.4. Annual per-person hemophilia B-related costs in United States dollars ..................................................... 37
Table 3.1. Episodic dose and prophylaxis regimens .................................................................................................... 56
Table 3.2. Description of independent variables for regression analyses .................................................................... 59
Table 3.3. Baseline participant characteristics ............................................................................................................. 63
Table 3.4. Unadjusted comparison of independent variables among high and low prophylaxis adherers ................... 65
Table 3.5. Calculated independent variables and outcomes for participant subgroups................................................ 66
Table 3.6. Negative binomial regression for ABR: IRR results with continuous baseline period bleeding
episode variable ........................................................................................................................................................... 71
Table 3.7. Negative binomial regression for ABR: IRR results with binary baseline period bleeding episode
variable ........................................................................................................................................................................ 72
Table 3.C1. Distribution of independent variables among matched high and low prophylaxis adherers .................... 89
Table 4.1. Transition probability adjustment base case parameters ............................................................................. 99
Table 4.2. Input health-state utilities ......................................................................................................................... 101
Table 4.3. Input costs ................................................................................................................................................. 104
Table 4.4. Markov model results ............................................................................................................................... 106
10
List of Figures
Figure 2.1. Mean annual non-factor healthcare services utilization costs by hemophilia B severity and
treatment regimen ........................................................................................................................................................ 38
Figure 2.2. Mean annual indirect costs by hemophilia B severity and treatment regimen .......................................... 38
Figure 3.1. Selection of participants for statistical analyses ........................................................................................ 53
Figure 3.2. Distribution of annualized study period bleeding episodes among individuals using prophylaxis
(N=100) ....................................................................................................................................................................... 58
Figure 3.3. Unadjusted study period annualized bleed rate among high and low prophylaxis adherers ..................... 67
Figure 3.4. Adjusted study period annualized bleed rate among high and low prophylaxis adherers
(continuous baseline bleed variable)............................................................................................................................ 73
Figure 3.5. Adjusted study period annualized bleed rate among high and low prophylaxis adherers (binary
baseline bleed variable) ............................................................................................................................................... 73
Figure 3.6. Sensitivity analyses of negative binomial regression models among participant subgroups using
varied data inclusion criteria ........................................................................................................................................ 76
Figure 3.7. Sensitivity analyses of negative binomial regression models among participant subgroups using
varied prophylaxis dose ............................................................................................................................................... 77
Figure 3.8. Study period annualized bleeding episodes among 1:1 matched high and low adherers .......................... 77
Figure 3.A1. Distribution of annualized adherence by participant age ........................................................................ 87
Figure 3.A2. Distribution of annualized adherence by participant hemophilia type ................................................... 87
Figure 3.B1. Distribution of annualized bleed rate by participant age ........................................................................ 88
Figure 3.B2. Distribution of annualized bleed rate by participant hemophilia type .................................................... 88
Figure 4.1. Decision tree for familial hypercholesterolemia diagnosis........................................................................ 97
Figure 4.2. Markov model of disease progression ....................................................................................................... 98
Figure 4.3a. One-way sensitivity analysis tornado plot of Lipid Screening + AD vs. Lipid Screening ICER .......... 108
Figure 4.3b. One-way sensitivity analysis tornado plot of Genetic Screening vs. Lipid Screening ICER ................ 108
Figure 4.4. Cost-effectiveness acceptability curve of Lipid Screening + AD vs. Lipid Screening ICER .................. 109

11
ABSTRACT
Hereditary diseases are generally chronic and impose a lifetime of burden of illness. The
majority of these diseases are rare conditions that individually affect only a small proportion of
the population. Many do not have effective treatments, and even among those with disease-
specific pharmaceutical interventions, treatment regimens may not be optimal. It has been
established that approximately 50% of patients with chronic conditions do not adhere to their
prescribed treatment. For individuals with hereditary or chronic diseases, this eventually leads to
poor clinical and quality of life (QoL) outcomes, increased morbidity, and unnecessary
healthcare expenditures over the course of a lifetime. Further, while individually uncommon,
together there are thousands of rare diseases that create significant burden of illness for over 350
million people globally, but thorough understanding of real-world treatment patterns and patient
experiences is limited by the geographic dispersion and low prevalence of individuals with these
conditions. In order to highlight unmet treatment needs among individuals with hereditary
diseases and advance efforts to optimize treatments for patient experiences, this dissertation aims
to better understand variations in clinical and economic outcomes, examine the impact of
treatment decisions and adherence on clinical and economic outcomes, and assess the potential
for cost-effective treatment optimization among patients with two different examples of
hereditary diseases. The studies here report findings from hemophilia and familial
hypercholesterolemia (FH), which represent two conditions with varying prevalence, diagnosis
rate, severity and onset of clinical symptoms, and ease and cost of pharmaceutical treatment.
The first study used prospective longitudinal data from the Hemophilia Utilization Group
Studies (HUGS) to determine societal burden of illness, including direct and indirect costs and
annualized bleed rate (ABR), for persons with hemophilia B in the United States (US) and to
12
conduct subgroup analyses, which found that costs and ABR differed significantly among
subgroups by hemophilia severity and treatment regimen. The second study employed the
relatively large hemophilia sample sizes and rich datasets from HUGS in regression analyses to
evaluate the covariate-adjusted impact of adherence to prophylaxis on ABR among persons with
hemophilia A or B in the US, and to identify other socio-demographic and clinical variables
significantly associated with bleeding episodes. The results showed a statistically significant
impact of adherence on bleeding episodes even after controlling for many patient and disease
characteristics. While this relationship between adherence and outcomes was found in all
subgroups by age and hemophilia type, the impact of adherence was not significant among
children. The third study used decision tree and Markov modeling to show that implementing a
statin adherence program in addition to currently recommended lipid-based cascade screening
for FH diagnosis and management is cost-effective compared with genetic cascade screening or
lipid-based screening alone for individuals with high cholesterol and a family history of FH or
heart disease in the US.
Together, the studies demonstrated that hereditary diseases can be costly conditions, in
which healthcare expenditures, treatment patterns, and clinical outcomes differ significantly
between clinically meaningful patient subgroups. Further, adherence was a common issue across
different types of hereditary diseases regardless of the severity of outcomes and cost or ease of
treatment. The results also suggested that management of these conditions could be tailored to
meet individual needs and clinical goals and that improving treatment adherence could be cost-
effective to society. The types of data and methods employed here may by appropriate for
broader research regarding treatment optimization in other chronic and hereditary conditions.

13
CHAPTER 1: Introduction
1.1. Background
Hereditary diseases are caused by inherited abnormalities in an individual’s genome and
can impose a lifetime of treatment, quality of life (QoL), and socioeconomic burdens.
1,2

Individuals affected by these diseases are born with a genetic condition inherited either
dominantly or recessively from parents. No cures exist for the majority of these oftentimes
debilitating or life-threatening hereditary diseases, but appropriate treatment and management
can significantly improve QoL and extend life expectancy in many individuals for which
disease-specific treatment is available.
3-6
Currently, there are approximately 7,000 recognized
hereditary diseases, the majority of which are rare conditions that individually only affect a small
portion of the population.
7
Within the United States (US), rare diseases are defined as conditions
affecting less than 200,000 individuals, and within the European Union (EU), as conditions
affecting less than 1 in 2,000 individuals.
8,9
While individually uncommon, together there are
thousands of rare diseases that create significant burden of illness for approximately 350 million
people globally.
8,10

Without appropriate treatment, many individuals with hereditary conditions face
decreased QoL due to severe clinical events requiring hospitalizations and ongoing medical care.
Disease-specific treatment is currently only available for about 400 rare diseases and is usually
very expensive.
7
Further, personalized treatment guidelines do not exist for many rare and/or
hereditary conditions and it remains unclear if affected individuals are treated optimally. As
more studies examining the epidemiology, natural history, outcomes, and budget impact of
hereditary diseases reveal significant unmet needs for these patient populations and high costs
14
for healthcare systems, there is an increasing effort to refine hereditary disease management for
improved individual experiences.
Most hereditary diseases pose considerable burden on affected individuals, their families,
and society, including burden of treatment, decreased QoL, and socio-economic burden. One
systematic literature review has been conducted on QoL in rare hereditary diseases and found
that QoL in affected individuals was consistently lower than in those unaffected.
2
Because
hereditary diseases often have chronic and debilitating clinical outcomes and either scant
treatment options or difficult and time-consuming treatment regimens, family members and other
caregivers also report significant burden in caring for these individuals.
11,12
In addition to QoL
detriments, socio-economic burdens are also common. A review for cost of illness studies in ten
rare diseases revealed that cystic fibrosis and hemophilia were relatively well studied, whereas
data was extremely limited in conditions without pharmaceutical treatment.
1
For many of these
conditions, per person direct medical costs totaled hundreds of thousands of dollars annually.
Additionally, indirect costs were a substantial portion of total costs in most of the diseases
reporting these costs.
Hemophilia is a congenital bleeding disorder and one example of a well-studied rare
hereditary disease with serious outcomes and costly treatment options. Studies have shown that
for individuals with this condition, repeated bleeding into joints and soft tissue can lead to
painful arthropathy and significant QoL consequences.
13-17
While prophylactic treatment using
clotting factor replacement therapy can effectively prevent spontaneous bleeding and thus delay
the onset of arthropathy, many individuals with hemophilia still face frequent bleeding in routine
clinical practice, presumably due to suboptimal treatment.
18-21
Poor outcomes despite high
clotting factor usage is concerning as previous literature has estimated average annual per-person
15
clotting factor therapy costs upwards of $200,000.
1,18,22
This translates into around $3.4 billion to
patients and society to treat approximately 20,000 individuals in the US each year, and does not
account for other hemophilia-related costs, such as bleed-related hospitalizations, orthopedic
surgeries, and indirect costs.
23
Further, decisions regarding prophylaxis regimens and dosing
schedules are highly individualized, but the impact of variations in treatment patterns on
hemophilia-related outcomes has not been fully studies.
24,25

For individuals with familial hypercholesterolemia (FH), a chronic hereditary disorder
affecting 1 in 1 million to 1 in 500 individuals depending on the disease form, cardiovascular and
cerebrovascular events are major concerns.
26
Without appropriate cholesterol-lowering
treatment, an event such as an acute myocardial infarction (AMI) or cerebral vascular attack can
lead to a lifelong 30-50% reduction in health-state utility.
27
In contrast with costly clotting factor
injections used to treat hemophilia, inexpensive generic statins with well-defined prescribing
guidelines and easy oral administration are readily available to treat high cholesterol. However,
only an estimated 1-20% of actual FH cases have been diagnosed in most countries and
adherence to statins is a well-recognized issue among individuals at risk for cardiovascular
disease (CVD) or coronary heart disease (CHD).
28
Although statins are inexpensive, high costs
for individuals with FH manifest through a lifetime of chronic CVD and CHD. The American
Heart Association (AHA) estimates that in the US, CHD and stroke from all causes cost $108.9
billion and $53.9 billion each year, respectively, including direct and indirect costs.
29
At an
individual level, first-year treatment and hospitalization costs for major events can be over
$23,000 for an AMI and $16,000 for a stroke, with annual post-event costs of thousands of
dollars covering other pharmacy and medical services.
30

16
Taken together, the examples from these two conditions highlight unmet treatment needs
among individuals affected by different types of hereditary diseases with varying ease and cost
of treatment. Namely, approximately 50% of patients with chronic conditions do not adhere to
their prescribed treatment.
31
Adherence is a particularly crucial issue in chronic conditions, as
affected individuals face lifelong burdens of illness, and poor adherence can have long-term
consequences. For individuals with hereditary or chronic diseases, suboptimal adherence
eventually leads to poor clinical and QoL outcomes, increased morbidity, and unnecessary
healthcare expenditures.
31-34
According to a 2003 report from the World Health Organizations
(WHO), “increasing the effectiveness of adherence interventions may have a far greater impact
on the health of the population than any improvement in specific medical treatments.”
31

A few observational studies in hemophilia have found that up to 70% of individuals do
not adhere to treatment and discontinuation of prophylaxis with clotting factors is common.
35-37

The benefits of reduced hemophilic arthropathy and improved QoL are mitigated among
individuals who do not fully adhere to prophylaxis because of inconvenient and frequent
intravenous infusions and the higher costs of this approach.
25,38
It has been suggested that
prophylactic treatment should be personalized based on individual needs and bleeding
experiences.
24
Although some studies suggest that adherence is likely to be lowest among young
adults who transition from being infused by a caregiver to self-infusion, more long-term studies
of real-world experiences are needed to inform personalized treatment guidelines.
38-40
Further,
more frequent bleeding episodes observed in routine clinical practice compared with clinical trial
settings are likely due to poor adherence.
18-21
However, the precise impact of adherence on
bleeding episodes and how this impact varies among patient subgroups is not fully understood
17
because high-quality data quantifying both adherence and outcomes within one study is often
difficult to collect for rare diseases.
Adherence is also a problem among individuals with FH. Despite the fact that statin
treatment is inexpensive, has easy oral administration, and significantly reduces cholesterol
levels, less than 50% of adults with high cholesterol in the US are receiving appropriate
treatment according to the Centers for Disease Control and Prevention (CDC). Of those treated,
discontinuation rates are high and adherence to prescribed statin therapy generally declines to
50-60% or lower after multiple years of treatment, which is associated with higher incidence of
CVD events and mortality.
41-45
Because the costs associated with treating lifetime CVD and
CHD dwarf the cost of generic statin treatment, programs to increase FH diagnosis through
familial cascade screening and to encourage treatment adherence could be cost-effective in
improving outcomes for individuals with FH.
46-48

Many of the relationships between poor adherence, suboptimal outcomes, and high costs
have been established in well-known diseases affecting a large proportion of the population.
31

However, the low prevalence of most hereditary or rare diseases hinders obtaining sufficient
sample sizes and comprehensive datasets to investigate the impact of treatment decisions and
adherence on outcomes and to identify opportunities for adherence improvement.
49,50
Further,
randomization is not always ethical in these patient populations, and subgroup analyses needed
to study the individualized impact of hereditary diseases and their treatment options are difficult
to conduct in limited study samples. As such, investigators must adopt innovative and alternative
study designs to address unmet needs for optimized treatment in these patient populations. In
particular, non-interventional or observational studies, including retrospective studies, multi-
center registries, observational cohort studies, and meta-analyses, employ alternative designs to
18
provide valuable real-world information that supplements randomized clinical trial (RCT)
findings and fills current evidence gaps for these disease populations. Unlike most RCTs, which
are designed to minimize selection bias and powered to detect specific treatment effects,
observational studies can be designed to maximize data collection and analysis. Secondary data
collection through meta-analyses also expands the body of evidence regarding hereditary
diseases. Compiling data from multiple sources can fill in important evidence gaps where local
studies, direct treatment comparisons, large study samples, and other specific outcomes are not
readily available.
The work presented here uses two types of non-interventional study data from two
different hereditary diseases to evaluate patient experiences in light of varying treatment
decisions and adherence and to identify potential opportunities for cost-effective treatment
optimization. As with any analysis conducted on non-randomized, observational data, findings
must be interpreted in light of limitations regarding potential biases and narrow external validity.
By capitalizing on relatively large hereditary disease datasets, employing novel and rigorous
statistical methods, controlling for endogeneity with a comprehensive set of covariates,
conducting subgroup analyses, and testing the robustness of results in sensitivity analyses, this
work will provide a better understanding of the true impact of treatment patterns on key
outcomes of care and could help healthcare system influencers allocate resources and treatment
efforts to better address the needs of individuals with hereditary diseases. According to the
WHO, “interventions that target adherence must be tailored to the particular illness-related
demands experienced by the patient.”
31
The current analyses aim to provide new insights and
evidence for such tailored treatments designed to maximize individual adherence and optimize
patient experiences in small hereditary disease populations.
19
1.2. Overview of papers
Non-interventional studies were used to examine the impact of real-word treatment
decisions and adherence on outcomes and to assess the potential for cost-effective treatment
optimization among patients with hereditary diseases. Observational data from two prospective,
longitudinal cohort studies were used to determine burden of illness for persons with hemophilia
and evaluate the impact of adherence and other socio-demographic and clinical characteristics on
bleeding episodes. Chapter 2 used this data to determine the total societal burden of illness,
including direct and indirect costs and bleeding episodes, for individuals with hemophilia B, and
further examined the associations of hemophilia B severity and treatment regimen with multi-
dimensional components of burden. Chapter 3 expanded on these analyses to evaluate the impact
of adherence to prophylaxis on bleeding episodes among individuals with hemophilia A or B and
conducts subgroup analyses by age and hemophilia type. This chapter capitalized on the
relatively large cohort datasets for studying a rare disease, which included a rich set of
individual-level variables and longitudinal data on treatment patterns and outcomes, in regression
models that controlled for potential biases between the variable of interest and bleeding episodes.
Chapter 4 used a meta-analysis of published data from Europe and the US and combined
decision tree and Markov modeling techniques to assess the cost-effectiveness of alternative FH
screening and treatment adherence approaches in the US in terms of incremental cost-
effectiveness ratios (ICERs) in costs per quality-adjusted life-years (QALYs). Finally, Chapter 5
concludes and provides suggestions for future research.

20
1.3. Chapter references
1. Angelis A, Tordrup D, Kanavos P. Socio-economic burden of rare diseases: A systematic
review of cost of illness evidence. Health Policy. 2014;119(7):964-979.
2. Cohen JS, Biesecker BB. Quality of life in rare genetic conditions: a systematic review of
the literature. American Journal of Medical Genetics. Part A. 2010;152a(5):1136-1156.
3. O'Connor TP, Crystal RG. Genetic medicines: treatment strategies for hereditary
disorders. Nature Reviews. Genetics. 2006;7(4):261-276.
4. Konkle BA, Josephson NC, Nakaya Fletcher S. Hemophilia B. In: Pagon RA, Adam MP,
Ardinger HH, et al., eds. GeneReviews(R). Seattle (WA): University of Washington,
Seattle; 1993-2015.
5. Konkle BA, Josephson NC, Nakaya Fletcher S. Hemophilia A. In: Pagon RA, Adam MP,
Ardinger HH, et al., eds. GeneReviews(R). Seattle (WA): University of Washington,
Seattle; 1993-2015.
6. Soutar AK. Rare genetic causes of autosomal dominant or recessive
hypercholesterolaemia. IUBMB Life. 2010;62(2):125-131.
7. Sun HY, Hou TJ, Zhang HY. Finding chemical drugs for genetic diseases. Drug
Discovery Today. 2014;19(12):1836-1840.
8. Communication on Rare Diseases: Europe's Challenges. Brussels: Commission of the
European Communities; Nov 11, 2008 2008.
9. Schieppati A, Henter JI, Daina E, Aperia A. Why rare diseases are an important medical
and social issue. Lancet. 2008;371(9629):2039-2041.
10. Azie N, Vincent J. Rare diseases: the bane of modern society and the quest for cures.
Clinical Pharmacology and Therapeutics. 2012;92(2):135-139.
11. Anderson M, Elliott E, Zurynski Y. Australian families living with rare disease:
experiences of diagnosis, health services use and needs for psychosocial support.
Orphanet Journal of Rare Diseases. 2013;8(1):22.
12. Dellve L, Samuelsson L, Tallborn A, Fasth A, Hallberg LR. Stress and well-being among
parents of children with rare diseases: a prospective intervention study. Journal of
Advanced Nursing. 2006;53(4):392-402.
13. Choiniere M, Melzack R. Acute and chronic pain in hemophilia. Pain. 1987;31(3):317-
331.
14. Humphries TJ, Kessler CM. Managing chronic pain in adults with haemophilia: current
status and call to action. Haemophilia. 2015;21(1):41-51.
15. Luck JV, Jr., Silva M, Rodriguez-Merchan EC, Ghalambor N, Zahiri CA, Finn RS.
Hemophilic arthropathy. The Journal of the American Academy of Orthopaedic
Surgeons. 2004;12(4):234-245.
16. Witkop M, Lambing A, Divine G, Kachalsky E, Rushlow D, Dinnen J. A national study
of pain in the bleeding disorders community: a description of haemophilia pain.
Haemophilia. 2012;18(3):e115-119.
17. Olsson A, Hellgren M, Berntorp E, Baghaei F. Association between bleeding tendency
and health-related quality of life in carriers of moderate and severe haemophilia.
Haemophilia. 2015.
18. Zhou ZY, Koerper MA, Johnson KA, et al. Burden of illness: direct and indirect costs
among persons with hemophilia A in the United States. Journal of Medical Economics.
2015:1-9.
21
19. Alexander M, Barnes C, Barnett P. Prospective audit of patients with haemophilia:
bleeding episodes and management. Journal of Paediatrics and Child Health.
2012;48(2):177-179.
20. Flood E, Pocoski J, Michaels LA, McCoy A, Beusterien K, Sasane R. Patient-reported
experience of bleeding events in haemophilia. European Journal of Haematology.
2014;93 Suppl 75:19-28.
21. Schramm W, Royal S, Kroner B, et al. Clinical outcomes and resource utilization
associated with haemophilia care in Europe. Haemophilia. 2002;8(1):33-43.
22. Carlsson KS, Hojgard S, Lindgren A, et al. Costs of on-demand and prophylactic
treatment for severe haemophilia in Norway and Sweden. Haemophilia. 2004;10(5):515-
526.
23. Soucie JM, Evatt B, Jackson D. Occurrence of hemophilia in the United States. The
Hemophilia Surveillance System Project Investigators. American Journal of Hematology.
1998;59(4):288-294.
24. Astermark J, Petrini P, Tengborn L, Schulman S, Ljung R, Berntorp E. Primary
prophylaxis in severe haemophilia should be started at an early age but can be
individualized. British Journal of Haematology. 1999;105(4):1109-1113.
25. Fischer K. Prophylaxis for adults with haemophilia: one size does not fit all. Blood
Transfusion. 2012;10(2):169-173.
26. Youngblom E, Knowles JW. Familial Hypercholesterolemia. In: Pagon RA, Adam MP,
Ardinger HH, et al., eds. GeneReviews(R). Seattle (WA)1993-2015.
27. Sullivan PW, Ghushchyan V. Preference-Based EQ-5D index scores for chronic
conditions in the United States. Medical Decision Making 2006;26(4):410-420.
28. Nordestgaard BG, Chapman MJ, Humphries SE, et al. Familial hypercholesterolaemia is
underdiagnosed and undertreated in the general population: guidance for clinicians to
prevent coronary heart disease: consensus statement of the European Atherosclerosis
Society. European Heart Journal. 2013;34(45):3478-3490a.
29. Heidenreich PA, Trogdon JG, Khavjou OA, et al. Forecasting the future of cardiovascular
disease in the United States: a policy statement from the American Heart Association.
Circulation. 2011;123(8):933-944.
30. Pignone M, Earnshaw S, Tice JA, Pletcher MJ. Aspirin, statins, or both drugs for the
primary prevention of coronary heart disease events in men: a cost-utility analysis.
Annals of Internal Medicine. 2006;144(5):326-336.
31. Sabaté E. Adherence to Long-Term Therapies: Evidence for Action. Geneva, Switzerland:
World Health Organization;2003.
32. Brown MT, Bussell JK. Medication adherence: WHO cares? Mayo Clinic Proceedings.
2011;86(4):304-314.
33. Garcia-Dasi M, Aznar JA, Jimenez-Yuste V, et al. Adherence to prophylaxis and quality
of life in children and adolescents with severe haemophilia A. Haemophilia.
2015;21(4):458-464.
34. Krishnan S, Vietri J, Furlan R, Duncan N. Adherence to prophylaxis is associated with
better outcomes in moderate and severe haemophilia: results of a patient survey.
Haemophilia. 2015;21(1):64-70.
35. Armstrong EP, Malone DC, Krishnan S, Wessler MJ. Adherence to clotting factors
among persons with hemophilia A or B. Hematology. 2015;20(3):148-153.
22
36. 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(5):493-501.
37. Fischer K, Van Der Bom JG, Prejs R, et al. Discontinuation of prophylactic therapy in
severe haemophilia: incidence and effects on outcome. Haemophilia. 2001;7(6):544-550.
38. Khair K, Gibson F, Meerabeau L. The benefits of prophylaxis: views of adolescents with
severe haemophilia. Haemophilia. 2012;18(3):e286-289.
39. 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(5):760-765.
40. Pipe SW, Valentino LA. Optimizing outcomes for patients with severe haemophilia A.
Haemophilia. 2007;13 Suppl 4:1-16; quiz 13 p following 16.
41. Bates TR, Connaughton VM, Watts GF. Non-adherence to statin therapy: a major
challenge for preventive cardiology. Expert Opinion on Pharmacotherapy.
2009;10(18):2973-2985.
42. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term
persistence in use of statin therapy in elderly patients. JAMA : The Journal of the
American Medical Association. 2002;288(4):455-461.
43. Caspard H, Chan AK, Walker AM. Compliance with a statin treatment in a usual-care
setting: retrospective database analysis over 3 years after treatment initiation in health
maintenance organization enrollees with dyslipidemia. Clinical Therapeutics.
2005;27(10):1639-1646.
44. Huser MA, Evans TS, Berger V. Medication adherence trends with statins. Advances in
Therapy. 2005;22(2):163-171.
45. Rannanheimo PK, Tiittanen P, Hartikainen J, et al. Impact of Statin Adherence on
Cardiovascular Morbidity and All-Cause Mortality in the Primary Prevention of
Cardiovascular Disease: A Population-Based Cohort Study in Finland. Value in Health.
2015;18(6):896-905.
46. Aarnio E, Korhonen MJ, Huupponen R, Martikainen J. Cost-effectiveness of statin
treatment for primary prevention in conditions of real-world adherence--estimates from
the Finnish prescription register. Atherosclerosis. 2015;239(1):240-247.
47. Chen CX, Hay JW. Cost-effectiveness analysis of alternative screening and treatment
strategies for heterozygous familial hypercholesterolemia in the United States.
International Journal of Cardiology. 2015;181:417-424.
48. Nherera L, Calvert NW, Demott K, et al. Cost-effectiveness analysis of the use of a high-
intensity statin compared to a low-intensity statin in the management of patients with
familial hypercholesterolaemia. Current Medical Research and Opinion. 2010;26(3):529-
536.
49. Gagne JJ, Thompson L, O'Keefe K, Kesselheim AS. Innovative research methods for
studying treatments for rare diseases: methodological review. British Medical Journal
(Clinical Research Ed.). 2014;349:g6802.
50. Shapiro AD, Soucie JM, Peyvandi F, Aschman DJ, DiMichele DM. Knowledge and
therapeutic gaps: a public health problem in the rare coagulation disorders population.
American Journal of Preventive Medicine. 2011;41(6 Suppl 4):S324-331.

23
CHAPTER 2: Burden of Illness among Persons with Hemophilia B:
Examining the Associations of Severity and Treatment Regimens with Costs
and Annualized Bleed Rates

ABSTRACT
BACKGROUND AND OBJECTIVES: To determine US societal burden of illness, including
direct and indirect costs and annual bleed rate (ABR), for persons with hemophilia B (HB), a rare
and debilitating genetic disorder, and to examine associations of hemophilia severity and
treatment regimens with costs and ABR.
DATA AND METHODS: From 2009-2014, the Hemophilia Utilization Group Studies (HUGS)
Part Vb collected prospective data from ten US Hemophilia Treatment Centers. Adult or parents
of pediatric participants with HB completed initial surveys on socio-demographics, clinical
characteristics, and treatment patterns. During the two-year follow-up, participants reported
bleeding episodes, work absenteeism, and caregiver time quarterly, which were used to calculate
ABR and indirect costs. Direct costs were calculated using one-year clinical chart and two-year
dispensing records.
RESULTS: Of 148 participants, 112 with complete medical records and ≥1 follow-up survey
were included. Total mean annual per-person costs were $85,852 (median: $20,160) for
mild/moderate HB, $198,733 ($147,891) for severe HB, and $140,240 ($63,617) for all
individuals without inhibitors (p<0.0001). Mean ABR was 5.2 bleeds/year (standard deviation:
6.7). Clotting factor and indirect costs accounted for 85% and 9% of total costs, respectively.
Compared with episodic treatment, prophylaxis use in those with severe HB was associated with
2.5-fold higher clotting factor costs (p<0.01), significantly more missed parent work days
24
(p<0.0001) and clinician (p<0.001) or nursing visits (p<0.0001) but lower hospitalizations costs
(p=0.17) and ABR (p<0.0001).
CONCLUSIONS: HB is associated with high economic burden and ABR in routine clinical
practice. Benefits associated with prophylaxis may be used to refine treatment for optimal patient
outcomes.
25
2.1. Introduction
2.1.1. Background
Hemophilia is a rare, congenital blood disorder, which primarily affects males and causes
potentially fatal internal bleeding into the head and gastrointestinal tract, as well as frequent
bleeding into joints and soft tissues.
1,2
This disorder affects approximately 20,000 individuals
total in the United States (US). Hemophilia B (factor IX deficiency or HB) is much rarer than
hemophilia A (factor VII deficiency or HA), occurring in about 4,400 of these US residents.
3
For
individuals with hemophilia, acute bleeding episodes can occur spontaneously and after trauma
or surgery. Repeated bleeding into joints may eventually lead to debilitating and painful chronic
hemophilic arthropathy.
2,4

Although there is no cure, hemophilia can be effectively managed by integrated teams
expert in diagnosis and management with clotting factor replacement therapy either administered
following a bleeding episode (episodic or on-demand treatment) or regularly to prevent bleeding
episodes (prophylaxis).
5-7
People with mild or moderate hemophilia commonly use episodic
treatment, which can control bleeding, relieve pain, and restore joint mobility, but cannot prevent
arthropathy.
8
Prophylaxis results in fewer joint bleeds, delays the onset of arthropathy, and
improves quality of life (QoL), and is currently considered optimal care for individuals with
severe hemophilia.
5,7,9-12

The low prevalence of HB limits obtaining cohorts of sufficient size to robustly examine
burden associated with HB distinct from that specific to HA.
13
Due to individual variations in
different hemophilia severity, treatment regimens, and underlying therapeutic response, costs and
outcomes can differ significantly in terms of bleeding rate, healthcare resource utilization, and
QoL.
12,14
Prophylaxis compared with episodic treatment has been associated with lower bleeding
26
rate across persons with HA and HB, but at the price of higher clotting factor costs.
14,15
The
degree to which prophylaxis can improve outcomes to mitigate other hemophilia-related costs
from work productivity losses and healthcare services utilization, and how the benefit of
prophylaxis varies across HB remains unclear.
13
Further, about 25-30% of individuals with HA
and 3-5% of those with HB develop inhibitors (anti-drug antibodies) to clotting factors.
16
These
individuals require higher doses of clotting factors or other bypassing agents and can accrue
annual costs over three times higher than costs for individuals without inhibitors.
17

Prior studies have estimated that 45-94% of total direct medical costs are due to clotting
factor usage.
15,18,19
One French study of 126 individuals with HB found that the extra direct
medical cost of prophylaxis versus episodic treatment was approximately $24,695 per bleeding
episode prevented.
15
Although it remains unclear whether clotting factor consumption differs
significantly between individuals with HA and HB, it is possible that persons with HA have
more severe outcomes and could bear total costs different than those with HB.
20,21
Thus, it is
useful to obtain more comprehensive estimates of burden of illness by hemophilia type and other
clinical subgroups.
Recent studies have estimated hemophilia-related burden of illness specifically in the
US.
14,22-25
Four studies calculated direct costs across both HA and HB from a payer’s perspective
using claims data.
22-25
However, claims data lack detailed clinical and socio-demographic
variables to identify hemophilia severity, treatment regimen, and inhibitor status, and generally
do not record information regarding work productivity or bleeding episodes. A 2015 study used
prospective patient-reported outcomes and medical record extraction to calculate both direct and
indirect costs and annualized bleed rate (ABR) among 222 individuals with HA in the US,
27
revealing total mean annual per-person costs of $195,332 (median: $139,571) in 2011 US
dollars.
14

This study will employ prospective, longitudinal cohort data from the Hemophilia
Utilization Group Studies Part Vb (HUGS Vb), a multicenter study designed to examine burden
of illness among persons with HB at federally-supported hemophilia treatment centers (HTCs) in
the US. HTC care is a multi-disciplinary, team-based care delivery model that aims to prevent
orthopedic complications and maximize physical and psychological functioning and socio-
economic benefits.
6

2.1.2. Objectives
The objective of this study is to determine societal burden of illness, including direct and
indirect costs and ABR, for persons with HB in the US, and to examine associations of
hemophilia severity and treatment regimens with costs and ABR.

2.2. Methods
2.2.1. Hemophilia Utilization Group Studies Part Vb
From 2009-2014, ten HTCs collected prospective data on individuals from fifteen states
*

using the HUGS Vb protocol. All participants provided informed consent or assent. The
inclusion criteria were: 1) age 2-64 years at initial interview; 2) physician diagnosis of factor IX
deficiency ≤ 30%, with or without history of inhibitors; 3) receiving at least 90% of hemophilia
care from the HTC; 4) English or Spanish speaking; 5) seen at the HTC within two years prior to

*
Participants in HUGS Vb originated from Arkansas, California, Colorado, Illinois, Indiana,
Kansas, Massachusetts, Michigan, Mississippi, Montana, Ohio, South Dakota, Texas,
Washington, and Wyoming.
28
the study’s initiation. The protocol was approved by the Institutional Review Board (IRB) of the
University of Southern California and of each participating HTC.
All adults ≥18 years of age or parents of pediatric participants <18 years of age
completed a baseline survey to collect information regarding socio-demographics, clinical
characteristics, and treatment patterns. Participants or parents completed a follow-up survey
quarterly over a two-year study period (8 follow-up surveys) to track work or school
absenteeism, unpaid hemophilia-related caregiver time, bleeding episodes, and health outcomes.
ABR was annualized from the sum of participant-reported bleeding episodes.
Baseline clinical chart information included weight, inhibitor status, treatment patterns,
and comorbidities. Follow-up clinical information regarding healthcare services utilization,
changes in treatment pattern, inhibitor development, and new medical problems was collected
monthly from clinical charts in the first year of the two-year study period. Prescription data was
collected monthly from dispensing records throughout the two-year period.
2.2.2. Determination of direct costs
Each recorded instance of healthcare services utilization or drug dispensation was
multiplied by the price associated with the service or product to estimate direct costs, which were
adjusted to 2014 US dollars using the Consumer Price Index (CPI) for Medical Care. Only
patients with complete chart and dispensing records were included in the analysis.
Direct costs from clinical charts included hospitalizations, emergency room (ER) visits
and outpatient services, and related units of clotting factor received. The length of stay (LOS)
and primary diagnoses were used to calculate hospitalization costs. The average daily inpatient
cost was obtained from the Agency for Healthcare Research and Quality’s Healthcare Cost and
Utilization Project National Inpatient Sample (HCUP-NIS) average LOS and costs, based on
29
hospital-specific cost-to-charge ratios, for each International Classification of Diseases, Ninth
Revision (ICD-9) code recorded.
26
The average cost of an ER visit was based on the Medical
Expenditure Panel Survey statistical briefs.
27

Outpatient services included various types of HTC visits (comprehensive, nursing,
clinician, physical therapist, and social work/psychology), laboratory tests, and outpatient
procedures. Comprehensive visits refer to annual multidisciplinary evaluations, which involve
the HTC team of specialists, nurses, and hematologists and include laboratory testing,
assessment of treatment, and various training and counseling. Costs were estimated from the
2014 Medicare fee schedule, based on Current Procedural Terminology (CPT) codes.
28
A list of
laboratory tests required during comprehensive visits, which varied by age, use of recombinant
or plasma-derived clotting factor, and virological status, was summarized and reviewed by a
hematologist previously.
14

Annual medication costs were also included in direct costs and calculated using the
average of dispensing records over two years. The unit cost for clotting factors and bypassing
agents was obtained from payment allowance limits for Medicare Part B.
29
All hospital-supplied
factors recorded in the clinical charts were priced and added to healthcare services utilization
costs. The Veterans Affairs Federal Supply Schedule was used to obtain the cost for
aminocaproic acid.
30

2.2.3. Determination of indirect costs
The human capital approach was used to calculate indirect costs.
31
In this method, work
productivity losses are estimated through lost earnings using wages as a proxy for work time
output. Indirect costs included lost wages due to missed days of work among those employed and
unpaid hemophilia-related caregiver time reported in participant- or parent-completed follow-up
30
surveys, as well as hemophilia-related part-time employment or unemployment reported in the
baseline survey. Average civilian worker compensation obtained from the US Bureau of Labor
Statistics was $31.96/hour in 2014.
32
Full-time work was assumed to be 40 hours per week, and
part-time work was assumed to be 20 hours per week. Participants with at least one follow-up
survey were included, and all data were annualized using total follow-up days for each
participant.
2.2.4. Statistical analyses
Summary statistics were reported for all participants. Bivariate analyses were conducted
to examine the associations of hemophilia severity and treatment regimen with outcomes
variables. The chi-square statistic for categorical variables and Kruskal-Wallis test for
continuous variables were used to test groups for statistically significant differences. The chi-
square statistic for Poisson distributions was used to test for statistical significance in count data.
All analyses were conducted using SAS statistical software, version 9.4 (SAS Institute, Cary,
NC).

2.3. Results
2.3.1. Baseline characteristics
Of 148 recruited participants, 112 (75.7%) with complete chart and dispensing records
and at least one follow-up survey were included. The 112 individuals completed an average of 6
[median (med): 7] follow-up surveys with 21.4 (23.8) months of follow-up. Compared to
excluded participants, included subjects were more likely to be children (55.4% vs. 33.3%,
p<0.03), have public or private health insurance only (48.2% vs. 2.8% or 39.3% vs. 22.2%,
31
respectively, overall p<0.04), and have severe disease (49.1% vs. 25%, p<0.02) (Table 2.1). In
addition, included subjects were more likely to have annual household income >$20,000
compared with excluded subjects (75.9% vs. 55.6%, p<0.01). No other baseline characteristics
differed significantly between excluded and included participants. At the initial interview, two
included participants had inhibitors to clotting factors.
The mean age of included participants was 22.1 [standard deviation (SD): 17.6] and
almost half of the participants or parents were employed full-time (47.3%) (Table 2.1). As
expected, only 3 of 57 mild/moderate participants (5.3%) used prophylaxis compared to 31 of 55
severe participants (56.4%) (Table 2.2). Further, severe participants were more likely to be
antibody-positive for the hepatitis C virus (HCV) compared to mild/moderate participants
(29.1% vs. 12.3%, p<0.03). Finally, severe compared to mild/moderate participants were more
likely to have annual household income ≤$20,000 (27.3% vs. 7%, p<0.01).
32
Table 2.1. Baseline characteristics among excluded and included participants with
hemophilia B
Variables Excluded (n=36) Included (n=111) P-value
Socio-demographics
Age, mean (SD) 27.6 (17.5) 22.1 (17.6) 0.0542
Child (2 to <18 years old) 12 (33.3%) 62 (55.4%) 0.0215
Male 35 (97.2%) 111 (99.1%) 0.3942
Race 0.6974
White, non-Hispanic 27 (75%) 70 (62.5%)
Black, non-Hispanic 1 (2.8%) 8 (7.1%)
Hispanic 4 (11.1%) 18 (16.1%)
Asian Pacific Islander 1 (2.8%) 5 (4.5%)
Other
a
3 (8.3%) 11 (9.8%)
Employment status
b,c
0.3698
Full-time 14 (38.9%) 53 (47.3%)
Part-time 7 (19.4%) 15 (13.4%)
Not employed 15 (41.7%) 38 (33.9%)
Retired 0 (0%) 5 (4.5%)
Married/with partner
b
20 (55.6%) 77 (68.8%) 0.1111
Education >12 years
b
25 (69.4%) 74 (66.1%) 0.7083
Income
b,c
0.0035
≤$20,000 14 (38.9%) 19 (17%)
$20,001 - $40,000 2 (5.6%) 29 (25.9%)
$40,001 - $75,000 4 (11.1%) 24 (21.4%)
$75,000 14 (38.9%) 32 (28.6%)
Insurance type
c
0.0336
Public 1 (2.8%) 54 (48.2%)
Private 8 (22.2%) 44 (39.3%)
Both public and private 22 (61.1%) 10 (8.9%)
No insurance 4 (11.1%) 3 (2.7%)
Clinical Characteristics
Using prophylaxis 6 (16.7%) 34 (30.4%) 0.1076
Comorbidities
≥1 comorbidity 13 (36.1%) 33 (29.5%) 0.888
HIV/AIDS 2 (5.6%) 4 (3.6%) 0.5995
HCV 8 (22.2%) 23 (20.5%) 0.8287
Hemophilia Severity 0.0111
Mild/Moderate 27 (75%) 57 (50.9%)
Severe 9 (25%) 55 (49.1%)
Abbreviations: SD - Standard deviation, HIV/AIDS - Human immunodeficiency virus/Acquired immunodeficiency
syndrome, HCV - Hepatitis C virus
All statistics reported in N(%), unless otherwise specified
a
Other races include American Indian or Alaskan Native and others
b
Applies to participants ≥18 years old, or parents of participants 2 to <18 years old
c
Does not add up to total sample because of missing data
33
Table 2.2. Baseline characteristics among included participants by hemophilia B severity
Variables
Mild/Moderate
(N=57)
Severe
(N=55)
P-value
Socio-demographics
Age, mean (SD) 21.9 (16.7) 22.4 (18.6) 0.7683
Child (2 to <18 years old) 31 (54.4%) 31 (56.4%) 0.8333
Male 57 (100%) 54 (98.2%) 0.3065
Race 0.0192
White, non-Hispanic 37 (64.9%) 33 (60%)
Black, non-Hispanic 2 (3.5%) 6 (10.9%)
Hispanic 6 (10.5%) 12 (21.8%)
Asian Pacific Islander 2 (3.5%) 3 (5.5%)
Other
a
10 (17.5%) 1 (1.8%)
Employment status
b,c
0.4312
Full-time 27 (47.4%) 26 (47.3%)
Part-time 7 (12.3%) 8 (14.5%)
Not employed 22 (38.6%) 16 (29.1%)
Retired 1 (1.8%) 4 (7.3%)
Married/with partner
b
42 (73.7%) 35 (63.6%) 0.2439
Education >12 years
b
33 (57.9%) 41 (74.5%) 0.0628
Income
b,c
0.0318
≤$20,000 4 (7%) 15 (27.3%)
$20,001 - $40,000 17 (29.8%) 12 (21.8%)
$40,001 - $75,000 15 (26.3%) 9 (16.4%)
$75,000 17 (29.8%) 15 (27.3%)
Insurance type
c
0.2947
Public 29 (50.9%) 25 (45.5%)
Private 21 (36.8%) 23 (41.8%)
Both public and private 4 (7%) 6 (10.9%)
No insurance 3 (5.3%) 0 (0%)
Clinical Characteristics
Using prophylaxis 3 (5.3%) 31 (56.4%) <0.0001
Comorbidities
≥1 comorbidity 14 (24.6%) 19 (34.5%) 0.0486
HIV/AIDS 1 (1.8%) 3 (5.5%) 0.2915
HCV 7 (12.3%) 16 (29.1%) 0.0277
Abbreviations: SD - Standard deviation, HIV/AIDS - Human immunodeficiency virus/Acquired immunodeficiency
syndrome, HCV - Hepatitis C virus
All statistics reported in N(%), unless otherwise specified
a
Other races include American Indian or Alaskan Native and others
b
Applies to participants ≥18 years old, or parents of participants 2 to <18 years old
c
Does not add up to total sample because of missing data
34
2.3.2. Healthcare services utilization, dispensing, and work productivity losses
During the one-year clinical chart follow-up, 18 participants (16%) had at least one ER
visit and 7 (6.3%) had at least one hospitalization. Overall, severe prophylaxis versus episodic
treatment users had more ER visits (12 vs. 1, p<0.04) but fewer hospitalizations (1 vs. 3, p=0.24)
with shorter mean LOS (1.5 vs. 6.3 days, p=0.09) (Table 2.3).
Severe compared to mild/moderate participants had significantly more nursing
(p<0.0001), clinician (doctor, nurse practitioner, or physician’s assistant) (p<0.01), and social
work/psychology visits (p<0.03) at the HTC (Table 2.3). The numbers of these types of HTC-
related visits were also significantly higher among severe participants treating prophylactically
versus episodically (p<0.0001, p<0.001, p<0.05, respectively).
Furthermore, severe participants had significantly higher mean annual clotting factor
dispensing measured by international units/kilogram of body weight (IU/kg) than mild/moderate
participants (p<0.0001) (Table 2.3). Prophylaxis users had significantly higher mean factor
dispensing than episodic treatment users among severe participants [4,945 (SD: 4,184) vs. 1,486
(SD: 1,613), p<0.01].
Hemophilia severity and treatment regimen were also associated with different work
productivity losses (Table 2.3). Severe versus mild/moderate hemophilia was associated with
significantly more days of parental work absenteeism (p<0.01) and hours of caregiver time
(p<0.0001). On average, severe adult participants lost 5.2 days (SD: 5.5) of work productivity
annually and parents of severe pediatric participants lost 1.6 (3.5) days, of which 3.5 (4.5) and
1.2 (2.8) days were due to hemophilia, respectively. Among mild/moderate participants, adults
lost 5.8 (9.5) days annually and parents of pediatric participants lost 0.9 (3.6) days, of which 3.2
(7.5) and 0.5 (1.5) days were due to hemophilia, respectively. Compared with those treating
35
prophylactically, more severe participants or parents of severe pediatric participants treating
episodically were unemployed (3 vs. 2, p=0.44) or employed part-time due to hemophilia (3 vs.
1, p=0.19), but prophylaxis was associated with more missed parental work days (p<0.0001).
Table 2.3. Annual healthcare resource utilization, work productivity loss, and bleeding
episodes among individuals with hemophilia B
Variables
Total
(N=112)
a

Hemophilia Severity Severe Hemophilia Treatment
Mild/
Moderate
(N=57)
Severe
(N=55)
P-value

Episodic
(N=24)
Prophylaxis
(N=31)
P-value

Annual Healthcare Service
Utilization, mean (SD)
(# visits/person/year)
b

Comprehensive visits 1 (0.8) 0.8 (0.8) 1.2 (0.8) 0.1083 1 (0.9) 1.3 (0.8) 0.5273
Nursing visits 0.8 (4.8) 0.4 (0.7) 1.3 (6.9) <0.0001 0.3 (0.7) 2 (9.1) <0.0001
Other clinician (MD/PA/NP) visits 0.5 (1.1) 0.3 (0.6) 0.6 (1.5) 0.0042 0.1 (0.4) 1.1 (1.8) 0.0005
Physical therapist visits 0.2 (0.7) 0.2 (0.6) 0.3 (0.8) 0.0742 0 (0) 0.6 (1) --
Social work/psychology visits 0.2 (0.6) 0.1 (0.4) 0.3 (0.8) 0.0233 0.1 (0.5) 0.5 (1) 0.0435
Emergency room visits 0.2 (0.6) 0.2 (0.5) 0.2 (0.7) 0.9274 0 (0.2) 0.4 (0.8) 0.032
Hospitalizations 0.1 (0.2) 0.1 (0.2) 0.1 (0.3) 0.672 0.1 (0.3) 0 (0.2) 0.2408
Length of stay (days/patient/year)
c
3.9 (3.5) 2.3 (0.6) 5.1 (4.4) 0.0723 6.3 (4.5) 1.5 (-) 0.0894
Outpatient procedures 0.1 (0.3) 0.1 (0.3) 0 (0.2) 0.193 0.1 (0.3) 0 (0) --
Annual Clotting Factor Dispensed,
mean (SD)
(IU/kg body weight/year)
d

2,372
(3,392)
1,299
(2,597)
3,548
(3,775)
<0.0001 1,486
(1,613)
4,945
(4,184)
0.0018
Employment status due to
hemophilia, N (%)

Employed part-time 4 (3.6%) 0 (0%) 4 (7.3%) 0.0381 3 (12.5%) 1 (3.2%) 0.189
Unemployed 8 (7.1%) 3 (5.3%) 5 (9.1%) 0.4317 3 (12.5%) 2 (6.5%) 0.439
Missed days of work, mean (SD)
Missed days due to all reasons,
parent
e

1.3 (3.5) 0.9 (3.6) 1.6 (3.5) 0.002 0.7 (1.8) 2.3 (4.3) <0.0001
Missed days due to HB,
parent
e

0.9 (2.2) 0.5 (1.5) 1.2 (2.8) 0.0002 0.5 (1.3) 1.7 (3.5) 0.0002
Missed days due to all reasons,
adult participant
f

5.5 (7.8) 5.8 (9.5) 5.2 (5.5) 0.1705 4.7 (4.6) 5.6 (6.1) 0.149
Missed days due to HB,
adult participant
f

3.6 (6) 3.8 (7.1) 3.5 (4.5) 0.3429 3.6 (4.8) 3.4 (4.4) 0.7583
Unpaid caregiver hours,
mean (SD)
6.5 (25.1) 3.2 (7.5) 9.8 (34.8) <0.0001 10 (42.6) 9.7 (28.1) 0.7101
ABR, mean (SD) 5.2 (6.7) 3.5 (5.1) 7.1 (7.7) <0.0001 9 (7) 5.5 (7.9) <0.0001
Abbreviations: SD - Standard deviation, MD - Doctor of medicine, PA - Physician’s assistant, NP - Nurse
practitioner, IU - International unit, kg - Kilogram, HB - Hemophilia B, ABR - Annual bleed rate
a
Includes all participants with (n=2) and without (n=110) inhibitors
b
Visits refer to in-person visits
c
Only applies to participants with hospital stays (N=7)
d
Three participants with missing weight data were excluded. Inhibitor-related bypassing agent dispensations were
excluded. Participants with inhibitors did not have clotting factor dispensation in addition to bypassing agents.
e
Parents of pediatric participants 2 to <18 years of age (mild/moderate N=31, severe N=30)
f
Adult participants ≥18 years of age (mild/moderate N=26, severe N=23)
36
2.3.3. Bleeding episodes
Mean ABR was 5.2 (SD: 6.7, med: 2.5) (Table 2.3). Severe compared with
mild/moderate participants had significantly higher ABR (p<0.0001). Episodic treatment users
had significantly higher ABR than prophylaxis users in severe [mean (SD): 9 (7) vs. 5.5 (7.9),
p<0.0001] and mild/moderate participants [3.6 (5.2) vs. 1 (1.3), p<0.03].
2.3.4. Direct and indirect costs
Mean annual total (direct plus indirect) costs per participant without inhibitors (N=110)
was $140,240 (med: $63,617) (Table 2.4). Clotting factor costs accounted for an average 85%
(med: 98%) of total costs and 92% (med: 99%) of direct costs. Indirect costs accounted for 9%
(med: 0%) of total costs, while lost wages from unemployment or part-time employment
accounted for 96% of indirect costs for those who were underemployed (n=11). Mean total costs
for mild/moderate and severe participants without inhibitors were $85,852 (med: $20,160) and
$198,733 (med: $142,891), respectively (p<0.0001). Among participants with inhibitors (N=2),
mean direct and indirect costs were $1,424,364 and $34,638, respectively.
2.3.5. Costs by participant subgroups
Severe versus mild/moderate participants had higher indirect [mean(med): $8,421($204)
vs. $4,416($0), p=0.11] and direct costs [$190,312($141.879) vs. $81,435($19,146), p<0.0001]
(Table 2.4). Prophylaxis versus episodic treatment was associated with lower indirect costs for
severe participants [$6,477($408) vs. $10,957($131), p=0.96] but significantly higher direct
costs [$256,775($205,575) vs. $103,630($63,765), p<0.01]. The costs of hospitalizations and ER
visits accounted for the majority of non-factor healthcare services utilization costs across
subgroups (Figure 2.1). Severe versus mild/moderate participants had significantly higher HTC
visit costs [mean(med): $243($187) vs. $139($111), p<0.04] and laboratory test costs [$225(160)
37
vs. $113($107), p<0.01]. In severe participants, prophylaxis compared with episodic treatment
was associated with lower hospitalization costs [$68($0) vs. $5,389($0), p=0.17) and
significantly higher HTC visit costs [$325($223) vs. $135($111), p<0.03]. Lost wages from part-
time employment or unemployment due to hemophilia accounted for the majority of indirect
costs across subgroups (Figure 2.2), and differences were not statistically significant.
Table 2.4. Annual per-person hemophilia B-related costs in United States dollars
Annual Costs
Total
(N=110)
Hemophilia Severity
Treatment Regimen,
Severe Hemophilia Only

Mild/
Moderate
(N=57)
Severe
(N=53)

Episodic
(N=23)
Prophylaxis
(N=30)
Total costs
(Direct+Indirect)***,
†††
140,240
(170,392)
[63,617]

85,852
(20,160)
[20,160]
198,733
(178,246)
[147,891]

114,577
(90,336)
[95,353]
263,253
(202,128)
[208,999]
Total direct costs***
,†††

133,894
(167,768)
[51,814]

51,435
(137,459)
[19,146]
190,312
(137,459)
[141,879]

103,620
(90,256)
[63,765]
256,775
(203,391)
[205,575]
Healthcare services
utilization costs*
,a
2,303
(8,508)
[438]

1,439
(3,548)
[250]
3,231
(11,682)
[500]

6,122
(17,453)
[500]
1,015
(1,381)
[500]
Clotting factor
costs***
,†††,b

131,574
(167,606)
[51,205]

79,974
(137,097)
[18,927]
187,070
(180,514)
[131,837]

97,490
(88,570)
[62,658]
255,747
(203,062)
[205,085]
Total indirect costs
c

6,346
(17,296)
[159]

4,416
(14,977)
[0]
8,421
(19,417)
[204]

10,957
(21,544)
[131]
6,477
(17,747)
[408]
Factor costs as
proportion of total costs
0.85
(0.27)
[0.98]

0.86 (0.25)
[0.96]
0.86 (0.29)
[0.99]

0.82 (0.3)
[0.98]
0.87 (0.28)
[0.99]
Costs reported as mean (standard deviation) [median] in 2014 United States dollars ($)
*** p<0.01, ** p<0.05, * p<0.1 between hemophilia severity groups
†††
p<0.01,
††
p<0.05,
†
p<0.1 between treatment regimen groups among those with severe hemophilia only
a
Includes hemophilia treatment center visits, laboratory tests, emergency room visits, hospitalizations, and outpatient
procedures
b
Cost of non-factor hemophilia-related medication (aminocaproic acid) was also included in Total direct sosts
c
Includes lost wages from part-time or unemployment due to hemophilia, lost wages from missed work due to all
reasons, and unpaid caregiver time

38
Figure 2.1. Mean annual non-factor healthcare services utilization costs by hemophilia B
severity and treatment regimen

Abbreviations: HB – Hemophilia B, ER - Emergency room, HTC - Hemophilia treatment center
Numeric labels represent total mean per-person healthcare services utilization costs; error bars represent 95%
confidence intervals
Figure 2.2. Mean annual indirect costs by hemophilia B severity and treatment regimen

Abbreviations: HB - Hemophilia B
Numeric labels represent total mean per-person indirect costs; error bars represent 95% confidence intervals
-$2,000
$0
$2,000
$4,000
$6,000
$8,000
$10,000
$12,000
$14,000
Mild/Moderate (N=57) Severe (N=53) Episodic (N=23) Prophylactic (N=30)
Per-Person Non-Factor Healthcare Services
Utilization Costs
Outpatient procedures
Hospitalizations
ER visits
Laboratory tests
HTC visits
$3,231
$6,122
$1,015
$1439
Hemophilia Severity Treatment Regimen, Severe HB Only
$0
$5,000
$10,000
$15,000
$20,000
Mild/Moderate (N=57) Severe (N=53) Episodic (N=23) Prophlaxis (N=30)
Per-Person Indirect Costs
Unpaid caregiver cost
Lost wages from missed work
due to HB
Lost wages from missed work
due to other reasons
Lost wages from part-time or
unemployment due to HB
Hemophilia Severity Treatment Regimen, Severe HB Only
$4,416
$8,421
$10,957
$6,477
39
2.4. Discussion and limitations
2.4.1. Discussion
Data from HUGS Vb was used to estimate and examine annual hemophilia-related burden of
illness for individuals with HB in the US. Although a similar analysis has been done for HA,
studies evaluating hemophilia-related costs from a societal perspective with detailed information
on patient characteristics and bleeding patterns are scarce, especially in HB.
14
Polack et al. used
French national health insurance data from 126 subjects with HB to calculate mean annual per-
person medical costs of approximately $104,459 (SD:$90,828) and mean ABR of 3.57 (SD:
6.55), but did not estimate indirect costs other than travel time to the clinic and only followed-up
for one year.
15
Other recent studies evaluating HA- and HB-related costs from US payers’
perspectives relied on claims databases that lack detailed patient information.
22-25

These results evaluate the economic and clinical impact of HB separate from HA on both
patients and society. HB is a costly disorder with lower ABR and annual per-person costs
compared to estimates for HA.
14
However, further comparisons of economic and clinical
outcomes specific to HB versus HA should be made cautiously as different enrollment criteria,
data collection, and analysis methods from multiple studies may limit the comparability of
separate study results. Subgroup analyses from HUGS Vb showed that severe HB is associated
with more work productivity losses, costs and bleeds compared with mild/moderate HB. Within
severity classes, treating prophylactically versus episodically is associated with lower ABR at the
price of significantly higher direct costs. Among severe individuals, although prophylaxis use is
associated with more HTC-related visits and missed parental work days, presumably due to the
need for more frequent factor infusions, there is preliminary evidence of cost-savings through
fewer hospitalizations with shorter LOS and less part-time employment or unemployment
40
compared to episodic treatment use. However, these savings do not fully offset higher factor
costs as these other costs are generally low compared to factor costs across all subgroups. While
cost and bleeding patterns among clinical subgroups in HUGS Vb are similar to those found in
HA, a direct comparison of economic and clinical outcomes is still needed to evaluate
differences between HA and HB.
ABR observed in HUGS Vb is higher than what has been reported in clinical trials of
individuals with severe or moderately severe HB and by Polack et al.
15,33,34
However, there is
other evidence of frequent patient- or clinician-reported bleeding episodes despite clotting factor
therapy from studies that captured outcomes in routine clinical practice.
14,35,36
Despite the
effectiveness of clotting factor treatment regimens observed in highly regulated clinical trial
environments, the results suggest that individuals using episodic or prophylactic treatments will
still experience bleeding episodes in routine clinical practice.
Combined, the total US societal cost to treat these 112 HB patients from HUGS Vb
would be $15.5 million annually. This sample represents 2-3% of the individuals with HB
receiving care at an HTC.
37
Given that clotting factor usage accounts for 85% of total costs and
prophylaxis may lead to fewer hospitalizations and lower indirect costs, additional studies may
enhance our understanding of the cost-effectiveness of individual treatment decisions. Further,
the degree to which reduced ABR and hospitalizations provide long-term non-monetary benefits
to patients, in terms of joint health, QoL, and caregiver burden, has yet to be fully assessed in
prospective longitudinal studies of HB. Combined with future studies that assess the long-term
impact of clotting factor treatment regimens, the current results can shed light on opportunities to
personalize treatment of HB for optimal outcomes beyond reduction in ABR.
41
2.4.2. Limitations
This study has a few limitations that emphasize the need for future analyses. First, the
results rely on patient-reported data, which may be subject to recall, social response, or other
biases. In addition, no adherence data is available to corroborate the clotting factor usage
suggested by dispensing records. Second, analyses by subgroups were not based on patients
randomized to prophylaxis or episodic treatment. Because multivariate analyses were limited by
the skewed nature of medical costs and the small sample size of subjects with complete covariate
information, potential selection bias should be factored into interpretations of the comparisons
between treatment regimens. Third, 36 participants without complete follow-ups, who tended to
be children with severe HB from poorer households, were excluded. As such, these individuals
may have different access to healthcare and social resources compared with the included subjects
and may face disparities in health outcomes that could bias results. Finally, this study examined
males receiving care at 10 of 141 HTCs in the US, potentially limiting the generalizability of the
results to the entire US HB population. In 2010, about 70% of the US HB population was treated
at HTCs, and it is unclear whether individuals treated at HTCs are different from those treated
elsewhere.
37
Additionally, any costs for hemophilia-related care received outside the HTCs or
HTC-affiliated hospitals were not fully captured by the HUGS Vb survey forms.

2.5. Conclusions
HB is associated with high societal costs and surprisingly high ABR in routine clinical
practice in the US. This is the first study to examine burden of illness for the US HB population
and the results demonstrated significant associations of hemophilia severity and treatment
regimens with costs and ABR. Overall, indirect and healthcare services utilization costs were
42
low compared with clotting factor costs. Severe versus mild/moderate HB and prophylaxis
versus episodic treatment use in severe HB were significantly associated with more HTC-related
visits and missed parental work days. Although frequent prophylactic infusions may necessitate
more HTC visits and work absenteeism, evidence of both lower hospitalization costs and lower
ABR compared with episodic treatment suggests potential long-term benefits to prophylaxis use
for HB. Future studies should evaluate how individualized treatment regimens and other patient
characteristics impact factor use and ABR and should assess the long-term impact of these
findings on joint health and overall patient well-being.
43
2.6. Chapter references
1. Mejia-Carvajal C, Czapek EE, Valentino LA. Life expectancy in hemophilia outcome.
Journal of Thrombosis and Haemostasis. 2006;4(3):507-509.
2. Luck JV, Jr., Silva M, Rodriguez-Merchan EC, Ghalambor N, Zahiri CA, Finn RS.
Hemophilic arthropathy. The Journal of the American Academy of Orthopaedic
Surgeons. 2004;12(4):234-245.
3. Soucie JM, Evatt B, Jackson D. Occurrence of hemophilia in the United States. The
Hemophilia Surveillance System Project Investigators. American Journal of Hematology.
1998;59(4):288-294.
4. Witkop M, Lambing A, Divine G, Kachalsky E, Rushlow D, Dinnen J. A national study
of pain in the bleeding disorders community: a description of haemophilia pain.
Haemophilia. 2012;18(3):e115-119.
5. Manco-Johnson MJ, Abshire TC, Shapiro AD, et al. Prophylaxis versus episodic
treatment to prevent joint disease in boys with severe hemophilia. The New England
Journal of Medicine. 2007;357(6):535-544.
6. Baker JR, Crudder SO, Riske B, Bias V, Forsberg A. A model for a regional system of
care to promote the health and well-being of people with rare chronic genetic disorders.
American Journal of Public Health and the Nation's Health. 2005;95(11):1910-1916.
7. Srivastava A, Brewer AK, Mauser-Bunschoten EP, et al. Guidelines for the management
of hemophilia. Haemophilia. 2013;19(1):e1-47.
8. Pipe SW, Valentino LA. Optimizing outcomes for patients with severe haemophilia A.
Haemophilia. 2007;13 Suppl 4:1-16; quiz 13 p following 16.
9. National Hemophilia Foundation MASAC [Medical and Scientific Advisory Council]
Recommendation 179: MASAC Recommendation Concerning Prophylaxis (Regular
Administration of Clotting Factor Concentrate to Prevent Bleeding). 2007;
http://www.hemophilia.org/NHFWeb/MainPgs/MainNHF.aspx?menuid=57&contentid=1
007. Accessed May 22, 2015, 2015.
10. Gringeri A, Lundin B, von Mackensen S, Mantovani L, Mannucci PM. A randomized
clinical trial of prophylaxis in children with hemophilia A (the ESPRIT Study). Journal
of Thrombosis and Haemostasis. 2011;9(4):700-710.
11. Manco-Johnson MJ, Kempton CL, Reding MT, et al. Randomized, controlled, parallel-
group trial of routine prophylaxis vs. on-demand treatment with sucrose-formulated
recombinant factor VIII in adults with severe hemophilia A (SPINART). Journal of
Thrombosis and Haemostasis. 2013;11(6):1119-1127.
12. Royal S, Schramm W, Berntorp E, et al. Quality-of-life differences between prophylactic
and on-demand factor replacement therapy in European haemophilia patients.
Haemophilia. 2002;8(1):44-50.
13. Gater A, Thomson TA, Strandberg-Larsen M. Haemophilia B: impact on patients and
economic burden of disease. Thrombosis and Haemostasis. 2011;106(9):398-404.
14. Zhou ZY, Koerper MA, Johnson KA, et al. Burden of illness: direct and indirect costs
among persons with hemophilia A in the United States. Journal of Medical Economics.
2015:1-9.
15. Polack B, Calvez T, Chambost H, et al. EQOFIX: a combined economic and quality-of-
life study of hemophilia B treatments in France. Transfusion. 2015.
44
16. Franchini M, Mannucci PM. Inhibitors of propagation of coagulation (factors VIII, IX
and XI): a review of current therapeutic practice. British Journal of Clinical
Pharmacology. 2011;72(4):553-562.
17. Armstrong EP, Malone DC, Krishnan S, Wessler MJ. Costs and utilization of hemophilia
A and B patients with and without inhibitors. Journal of Medical Economics.
2014;17(11):798-802.
18. Globe DR, Curtis RG, Koerper MA. Utilization of care in haemophilia: a resource-based
method for cost analysis from the Haemophilia Utilization Group Study (HUGS).
Haemophilia. 2004;10 Suppl 1:63-70.
19. Carlsson KS, Hojgard S, Lindgren A, et al. Costs of on-demand and prophylactic
treatment for severe haemophilia in Norway and Sweden. Haemophilia. 2004;10(5):515-
526.
20. Nagel K, Walker I, Decker K, Chan AK, Pai MK. Comparing bleed frequency and factor
concentrate use between haemophilia A and B patients. Haemophilia. 2011;17(6):872-
874.
21. Melchiorre D, Linari S, Manetti M, et al. Clinical, instrumental, serological and
histological findings suggest that hemophilia B may be less severe than hemophilia A.
Haematologica. 2016;101(2):219-225.
22. Guh S, Grosse SD, McAlister S, Kessler CM, Soucie JM. Healthcare expenditures for
males with haemophilia and employer-sponsored insurance in the United States, 2008.
Haemophilia. 2012;18(2):268-275.
23. Guh S, Grosse SD, McAlister S, Kessler CM, Soucie JM. Health care expenditures for
Medicaid-covered males with haemophilia in the United States, 2008. Haemophilia.
2012;18(2):276-283.
24. Tencer T, Friedman HS, Li-McLeod J, Johnson K. Medical costs and resource utilization
for hemophilia patients with and without HIV or HCV infection. Journal of Managed
Care Pharmacy 2007;13(9):790-798.
25. Valentino LA, Pipe SW, Tarantino MD, Ye X, Xiong Y, Luo MP. Healthcare resource
utilization among haemophilia A patients in the United States. Haemophilia.
2012;18(3):332-338.
26. Tingley K, Chakraborty P, Little J, et al. Candida infective endocarditis: an observational
cohort study with a focus on therapy. Genetics in Medicine. 2015;59(4):2365-2373.
27. Machlin S, Chowdhury S. Medical Expenditure Panel Survey Statistical Brief #318:
Expenses and Characteristics of Physician Visits in Different Ambulatory Care Settings,
2008. http://meps.ahrq.gov/mepsweb/data_files/publications/st318/stat318.shtml.
Accessed November 14, 2014.
28. Cole JA, Taylor JS, Hangartner TN, Weinreb NJ, Mistry PK, Khan A. Reducing selection
bias in case-control studies from rare disease registries. Orphanet Journal of Rare
Diseases. 2011;6:61.
29. Colina M, Zucchini W, Ciancio G, Orzincolo C, Trotta F, Govoni M. The evolution of
adult-onset Still disease: an observational and comparative study in a cohort of 76 Italian
patients. Seminars in Arthritis and Rheumatism. 2011;41(2):279-285.
30. National Acquisition Center (NAC) Contract Catalog Search Tool (CCST).
http://www.va.gov/nac/index.cfm?template=Search_Pharmaceutical_Catalog. Accessed
November 14, 2014.
45
31. Hodgson TA, Meiners MR. Cost-of-illness methodology: a guide to current practices and
procedures. The Milbank Memorial Fund Quarterly. Health and Society. 1982;60(3):429-
462.
32. Pariser AR, Gahl WA. Important role of translational science in rare disease innovation,
discovery, and drug development. Journal of General Internal Medicine. 2014;29 Suppl
3:S804-807.
33. Valentino LA, Rusen L, Elezovic I, Smith LM, Korth-Bradley JM, Rendo P. Multicentre,
randomized, open-label study of on-demand treatment with two prophylaxis regimens of
recombinant coagulation factor IX in haemophilia B subjects. Haemophilia.
2014;20(3):398-406.
34. Windyga J, Lissitchkov T, Stasyshyn O, et al. Pharmacokinetics, efficacy and safety of
BAX326, a novel recombinant factor IX: a prospective, controlled, multicentre phase I/III
study in previously treated patients with severe (FIX level <1%) or moderately severe
(FIX level ≤2%) haemophilia B. Haemophilia. 2014;20(1):15-24.
35. Alexander M, Barnes C, Barnett P. Prospective audit of patients with haemophilia:
bleeding episodes and management. Journal of Paediatrics and Child Health.
2012;48(2):177-179.
36. Schramm W, Royal S, Kroner B, et al. Clinical outcomes and resource utilization
associated with haemophilia care in Europe. Haemophilia. 2002;8(1):33-43.
37. Baker JR, Riske B, Drake JH, et al. US Hemophilia Treatment Center population trends
1990-2010: patient diagnoses, demographics, health services utilization. Haemophilia.
2013;19(1):21-26.

46
CHAPTER 3: The Impact of Adherence to Prophylactic Clotting Factor
Replacement Therapy on Bleeding Episodes among Persons with Hemophilia
in the United States

ABSTRACT
BACKGROUND AND OBJECTIVES: In chronic conditions, poor treatment adherence may
lead to poor outcomes and increased morbidity. However, there is limited evidence
demonstrating the relationship between adherence and outcomes among individuals with
hemophilia. The objective of this study is to evaluate the impact of adherence to prophylactic
clotting factor replacement therapy and of other covariates on bleeding episodes among persons
with hemophilia A (HA) or B (HB).
DATA AND METHODS: Between 2005-2007 and 2009-2012, the Hemophilia Utilization
Group Studies Parts Va and Vb (HUGS Va and Vb), two prospective cohort studies in the United
States (US), enrolled participants with HA or HB, respectively. Participants were followed-up
for two years to collect information on socio-demographics, clinical characteristics, treatment
patterns, healthcare resource utilization, and outcomes of care. Bivariate analyses were used to
examine associations between adherence and unadjusted annualized bleed rate (ABR). Negative
binomial regressions were used to evaluate the impact of adherence on ABR, while controlling
for covariates, and conducted separately among the total sample and subgroups by age and
hemophilia type. Adherence was calculated as dispensed prophylactic factor divided by amount
of factor required for “continuous prophylaxis” based on the World Federation of Hemophilia
47
guidelines and dichotomized to indicate high adherence 80%. Sensitivity analyses were
conducted to assess robustness of results.
RESULTS: Of 477 enrolled participants, there were 100 individuals who used prophylaxis and
met data inclusion criteria. Overall, 73% of individuals were considered high adherers. A smaller
proportion of high versus low adherers were adults (26.0% vs. 44.4%, p=0.0771). High versus
low adherers had regression-adjusted mean ABR of 6.3 [(standard error (SE): 2.4] versus 3.7
(1.4) (p<0.05). In the total sample, low versus high adherence was associated with a statistically
significant 69% higher ABR (p<0.05). The impact was large and highly significant among adults
(N=31) and individuals with HA (N=81). The regression did not converge in the limited HB
sample (N=19); however, bivariate analyses showed that low versus high adherence was
associated with higher ABR [mean (SD): 5.0 (4.3) vs. 4.2 (7.0), p=0.7890]. A number of
covariates were generally associated with ABR, including hours of paid nursing time and
insurance status. Findings were largely robust to reasonable variations in study parameters and in
cohort matching sensitivity analysis.
CONCLUSIONS: Poor adherence leads to poor clinical outcomes among individuals with
hemophilia, and robust differences in adherence and outcomes exist between subgroups. Efforts
to increase adherence and tailor prophylaxis regimens for individual treatment goals may lead to
clinical benefits with long-term quality of life (QoL) and economic implications.

48
3.1. Introduction
3.1.1. Background
Adherence to treatment is critical for effective patient care and achieving favorable
outcomes. Medication adherence is particularly crucial for patients with chronic conditions, as
these individuals face lifelong burdens of illness, and poor adherence can have significant long-
term consequences. According to the World Health Organization (WHO), adherence only
averages approximately 50% among individuals with chronic conditions in developed countries.
1

For individuals with chronic or inherited diseases, such as cardiovascular disease or hemophilia,
suboptimal adherence eventually leads to poor clinical and quality of life (QoL) outcomes,
increased morbidity, and increased healthcare expenditures.
1-4
As interest in understanding and
treating rare diseases grows, the negative impact of poor adherence among individuals with
hemophilia is an increasing concern.
Hemophilia is a rare inherited bleeding disorder, which occurs in approximately 20,000
individuals in the United States (US).
5
In these individuals, deficiency in factor VIII (hemophilia
A; HA) or factor IX (hemophilia B; HB) impairs the body's ability to form proper blood clots.
Bleeding episodes in the joints or soft tissues can occur spontaneously and after trauma or
surgery, and internal bleeding can be fatal.
6,7
Repeated bleeding into joints can eventually lead to
the development of debilitating and painful chronic hemophilic arthropathy and necessitate
major joint surgeries.
7,8
Hemophilia can be treated with clotting factor replacement therapy
administered following a bleeding episode (episodic or on-demand treatment), but such treatment
cannot prevent arthropathy. Alternatively, prophylaxis with clotting factor requires regular
infusions to maintain a factor activity level >1% for bleeding episode prevention. Prophylaxis
49
has been shown to result in fewer bleeds, delay the onset of arthropathy, and improve QoL, and
is currently considered optimal care for individuals with severe hemophilia.
9-14

Despite effective treatment with clotting factors, real world studies have demonstrated
high burden of illness associated with hemophilia. For individuals with hemophilia, frequency of
bleeding episodes is a key clinical outcome of interest. In contrast with pivotal clinical trials
reporting low mean and median annualized bleed rates (ABRs) of less than 3 or 4 for individuals
using recombinant clotting factor therapies, there is evidence of high ABR despite clotting factor
therapy in routine clinical practice, possibly due to poor adherence.
15-21
Over time, excessive
bleeding from suboptimal clotting factor treatment may result in joint disease and disability by
young adulthood.
5,7,22-24
Studies have also shown high cost burdens totaling over $200,000
annually per person with severe hemophilia, with costs for individuals on prophylaxis versus on-
demand treatment being significantly higher.
17,21,22,25
Further, about 25-30% of individuals with
HA and 3-5% of those with HB develop inhibitors (anti-drug antibodies) to clotting factors.
26

These individuals require higher doses of clotting factors or other bypassing agents to stop
bleeding and can accrue annual costs over three times higher than costs for individuals without
inhibitors.
27
Newer factors with extended half-lives have been shown to provide similar bleed
protection as conventional factors with decreased frequency of prophylactic infusions and could
potentially impact treatment adherence, long-term outcomes, and costs.
28,29
Thus, a better
understanding of the impact of real-world treatment adherence for this expensive and debilitating
disorder will be critical for efforts to improve and individualize care.
A small number of observational studies have studied adherence in hemophilia and found
that up to 70% of individuals with hemophilia do not adhere to prophylactic treatment and
discontinuation of prophylaxis is common.
30-33
There is also evidence suggesting adherence is
50
likely to be lowest among young adults who transition from being infused by a caregiver to self-
infusion and who consider a switch from prophylaxis to on-demand treatment.
34,35
While
national guidelines provide recommendations for prophylaxis protocols, treatment regimens are
often adjusted to satisfy individual needs and bleeding experiences, and frequent intravenous
infusions up to 3-4 times weekly for prophylaxis can hinder adherence. Recently, two studies
have demonstrated that higher adherence may be associated with fewer bleeding episodes based
on bivariate analyses and regression analyses with limited explanatory variables.
3,4
However, this
association has not been fully examined as studies evaluating the longitudinal relationships
between adherence or other patient characteristics and bleeds are rare and often limited by small
samples, narrowly representative patient subgroups, and/or limited covariates.
3,4,32,36
Given the
clinical significance of appropriate bleed management in individuals with hemophilia, the impact
of adherence and other important socio-demographic and clinical variables on bleeding episodes
should be further explored using larger, longitudinal datasets that can quantify and link
adherence, bleeding episodes and relevant individual characteristics within one study.
3.1.2. Objectives
The primary objective of this study is to evaluate the impact of adherence to prophylaxis on
bleeding episodes among persons with HA or HB treated at hemophilia treatment centers (HTCs)
in the US. Additionally, this study will identify other individual socio-demographic and clinical
variables significantly associated with bleeding episodes.

51
3.2. Data and methods
3.2.1. Data sources
3.2.1.1. The Hemophilia Utilization Groups Studies Va and Vb
The Hemophilia Utilization Group Studies Parts Va and Vb (HUGS Va and Vb), are two
prospective longitudinal cohort studies designed to examine the burden of illness among persons
with hemophilia at federally-supported HTCs in the US. By following participants longitudinally
and capturing a broad range of information on participant characteristics, hemophilia treatment,
outcomes of care, healthcare resource utilization, and work productivity losses, HUGS Va and
Vb provide rich datasets with which to explore the subtleties of hemophilia management and to
assess the impact of adherence on bleeding episodes through rigorous statistical methods.
Between 2005-2007 and 2009-2012, HUGS Va and Vb recruited individuals with HA or HB,
respectively, from ten HTCs treating patients from fifteen geographically diverse states
2
. Study
procedures and inclusion/exclusion criteria have been detailed previously.
37,38
Briefly, 329 and
148 individuals with HA or HB, respectively, were recruited and followed for two years. At the
initial interview, adult participants ≥18 years of age or parents of pediatric participants <18 years
of age completed an initial survey to collect information on socio-demographics, treatment
patterns, and QoL. Follow-up surveys were administered every month in the first year and semi-
annually in the second year in HUGS Va and every three months in HUGS Vb to track
information on work or school absenteeism, caregiver time, nursing time, bleeding episodes,
QoL, and other outcomes of care.

2
Participants in HUGS Va and Vb originated from Arkansas, California, Colorado, Illinois,
Indiana, Kansas, Massachusetts, Michigan, Mississippi, Montana, Ohio, South Dakota, Texas,
Washington, and Wyoming.
52
Clinical information and healthcare services resource utilization were abstracted from
medical charts and dispensing records. Baseline clinical information included clinical
characteristics, inhibitor status, hemophilia history, and treatment regimen. Follow-up clinical
information regarding HTC visits, emergency room (ER) and inpatient visits, outpatient
procedures, changes in treatment pattern, and inhibitor development was collected monthly from
clinical charts in the first year. Prescription data, including brand and amount of clotting factor as
well as other non-factor medications, was collected from dispensing records throughout the two-
year follow-up period.
3.2.1.2. Data cleaning and sample selection
The HUGS Va and Vb dataset was cleaned to identify prophylaxis users with minimal
missing data for statistical analyses. Participants without inhibitors and with complete baseline
covariates for analysis, complete dispensing records, ≥6 months of participant follow-up surveys,
and ≥6 months of follow-up clinical charts were included in this analysis (see Figure 3.1). The
final included sample had 290 individuals, of which 100 individuals with HA or HB who used
prophylaxis were included in the analyses.
Because clinical chart information was only collected in the first year of HUGS Va and
Vb and to address underlying disease complexity and management, the two-year follow-up
period was divided into two periods. The first 6 months were defined as the baseline period and
used to calculated variables describing baseline hemophilia severity, disease complexity, and
disease management (see Section 3.2.2). The next 18 months were defined as the study period of
interest and used to calculate the main independent and outcome variables of interest, adherence
to prophylaxis and number of bleeding episodes, respectively.
53
Figure 3.1. Selection of participants for statistical analyses

Abbreviations: HA - Hemophilia A, HB - Hemophilia B, HUGS Va - Hemophilia Utilization Group Study Part Va,
HUGS Vb - Hemophilia Utilization Group Study Part Vb
3.2.2. Methodology
3.2.2.1. Methodological considerations
The causal effect of adherence on clinical outcomes can be difficult to determine using
non-interventional data. Without randomization to different levels of adherence, which would be
unethical, there may be endogeneity and imbalance in observable and unobservable
characteristics between high and low adherence groups. In assessing the impact of adherence to
prophylaxis on number of bleeding episodes, potential unobserved heterogeneity in individual
hemophilia experiences and simultaneous causality between adherence and bleeding episodes
54
must be considered. Although the majority of prophylaxis users have severe hemophilia,
individual patterns of bleeding episodes and disease severity can vary greatly between
individuals. It is well documented that when used adherently and appropriately, prophylaxis is
effective in maintaining protection from bleeding episodes in clinical trial settings.
11,13,16,28

However, in routine clinical practice, ABR tends to be much higher than that reported in clinical
trials and likely due to suboptimal treatment.
17-21
Further, although improved adherence to
prophylaxis could lead to better clinical outcomes in terms of fewer bleeding episodes, it is also
possible that better outcomes could impact adherence, but the direction of this impact is unclear.
This study takes advantage of the panel datasets from HUGS Va and Vb to minimize
endogeneity when evaluating the impact of adherence. These are some of the largest prospective
and longitudinal datasets in hemophilia, encompassing a rich set of time-invariant individual
characteristics and two years of follow-up that link hemophilia treatment and outcomes. As such,
this analysis was able to use multivariable regression analyses that expand on previously
published studies examining the impact of adherence, which have been limited by smaller
sample sizes and/or incomplete explanatory covariates.
3,4
By including a relatively
comprehensive set of participant characteristics and a set of baseline variables calculated from
the first 6 months of the follow-up period to control for hemophilia complexity, level of disease
management, and number of bleeding episodes preceding the study period of interest, the
regression analyses in this study attempt to address both confounding unobserved hemophilia
characteristics and simultaneous causality between adherence and outcome. Thus, the baseline
bleeding episodes represent both a predictor of study period bleeding episodes and a lagged
dependent variable for study period bleeding, which, if excluded, could potentially bias estimates
of the impact of adherence in the current period.
55
3.2.2.2. Estimating adherence to prophylaxis
Because HUGS Va and Vb did not collect prescription data or days supply for clotting
factor dispensations, a proxy for adherence to prophylaxis was calculated based on the World
Federation of Hemophilia (WFH) definition for prophylactic factor replacement therapy
protocols.
9
The adherence proxy was a measure of total international units (IU) of clotting factor
per kilogram (kg) of body weight dispensed for prophylaxis in HUGS Va or Vb divided by the
total IU/kg of clotting factor required for “continuous prophylaxis” based on WFH definitions,
and multiplied by 100% to obtain percent adherence. Episodic clotting factor doses used to treat
bleeding episodes were excluded from the numerator to isolate total dispensing attributed to
prophylaxis use from that used to stop bleeds. Adherence for each included prophylaxis
participant in the study period of interest was calculated as:
Adherence =
(Clotting factor dispensed) - (Episodic dose) × (Bleeding episodes)
(Prophylaxis dose) × (Prophylaxis infusions/week) × (Weeks) × (0.85)
(1)
where Clotting factor dispensed was the sum of clotting factors reported in HUGS dispensing
records in IU/kg; Episodic dose was assumed to be 57.5 IU/kg for HA and 113 IU/kg for HB,
based on clinical trial data;
15,39
Bleeding episodes was the sum of participant-reported bleeding
episodes in HUGS follow-up surveys; Prophylaxis dose for HA was assumed to be 27.5 IU/kg
per infusion based on the average of the Malmo and Utrecht protocols
9
and 35 IU/kg per infusion
for HB based on a review of treatment with factor IX;
40
Prophylaxis infusions/week was assumed
to be 3 for HA and 2 for HB;
9
Weeks represents the length of follow-up time analyzed in
multivariable analyses (18 months = 78 weeks); and 0.85 represents the 85% threshold of weeks
of infusions required for “continuous prophylaxis” based on WFH definitions. Table 3.1 details
the assumed episodic doses and prophylaxis regimens by hemophilia type.
56
Table 3.1. Episodic dose and prophylaxis regimens
Episodic
dose
Prophylaxis
dose
Prophylaxis
frequency
Hemophilia A 57.5 IU/kg
15
Malmo protocol: 25-40 IU/kg
9

Utrecht protocol: 15-30 IU/kg
9

3 infusions/week
Hemophilia B 113 IU/kg
39
30-40 IU/kg
40
2 infusions/week
Abbreviations: IU - International unit, kg - Kilograms of body weight
In order to adjust for differences in study period length of follow-up, an annualized
adherence variable was calculated using Bleeding episodes equal to ABR:
ABR =
(Bleeding episodes) × 365 days
(Days of study period follow up)
(2)
and Weeks equal to 52. Additionally, because the adherence proxy could take negative values or
values greater than 1, adherence was dichotomized to indicate high versus low adherence for
statistical analyses. High adherence was defined as recorded factor consumption for prophylaxis
80% of the amount of factor required for “continuous prophylaxis”; factor consumption <80%
was considered low adherence. This is a commonly used cutoff to evaluate differences in
outcomes due to adherence.
2,33,41

3.2.2.3. Outcome variable of interest
The outcome variable of interest was number of bleeding episodes in the study period
calculated as the sum of bleeding episodes reported in study period participant follow-up
surveys. For non-annualized regression analyses, the days of study period follow-up was
included as an additional independent variable to adjust for differences in length of follow-up.
For annualized analyses, ABR was the outcome variable of interest and modeled without days of
study period follow-up as a covariate.
57
3.2.2.4. Independent variables
The independent variables used in the statistical analyses can be categorized in two
groups: baseline participant characteristics collected at the initial interview and baseline period
variables describing underlying hemophilia complexity and management. Participant
characteristics included socio-demographics such as age, marital status, education level, work
status, insurance type, and race, as well as variables related to hemophilia status and care, such as
years on prophylaxis, hours of travel to get to the HTC, hemophilia type, and hemophilia severity
based on clinical definitions. Because hemophilia manifests differently between patients and
treatment is highly individualized, underlying hemophilia complexity and management was
assessed through variables calculated from the baseline period. These included 6-month baseline
healthcare resource utilization (ER, inpatient, or HTC visits and outpatient procedures), work
productivity losses (hemophilia-related work absenteeism, unpaid caregiver hours, and paid
nurse hours), and number of bleeding episodes, which signal and control for the underlying
individual variations in hemophilia severity and appropriateness of current treatment regimens.
3.2.3. Statistical analyses
3.2.3.1. Descriptive statistics
Baseline participant characteristics were reported for all participants. Bivariate analyses
were conducted to examine differences in baseline period and study period variables between
participant subgroups. Descriptive statistics were reported in aggregate and also by clinically
meaningful subgroups in order to understand the distribution of important variables among high
versus low adherers, adults versus children, and participants with HA versus HB. The chi-square
statistic for categorical variables and Kruskal-Wallis test for continuous variables was used to
58
test groups for statistically significant differences. The chi-square statistic for negative binomial
distributions was used to test for statistical significance in count data. All statistical analyses
were conducted using SAS
®
statistical software, version 9.4 (SAS Institute, Cary, NC).
3.2.3.2. Negative binomial regression
Number of bleeding episodes is an example of count data, which is bounded at zero and
positively skewed (see Figure 3.2). As such, generalized linear models incorporating a negative
binomial distribution and a log-link function were used to evaluate the impact of adherence to
prophylaxis on bleeding episodes while adjusting for participant characteristics and baseline
period variables in SAS
®
.
Figure 3.2. Distribution of annualized study period bleeding episodes among individuals
using prophylaxis (N=100)

The theoretical model used to evaluate the impact of adherence is given by:
Y
i
= α
i
+ βP
i
+ δ
'
X
i
+ ε
i
(3)
59
where Y
i
is number of bleeding episodes for each participant i, 
i
is the individual intercept,  is
the coefficient of the variable of interest, P
i
is a binary indicator for high adherence to
prophylaxis,  = [ 
1
, 
2
, …, 
k
]’ is the coefficient vector of all other independent variables, X
i
=
[X
i1
, X
i2
, … X
ik
]’ are the other independent variables, and 
i
is the individual error term. Table 3.2
summarizes the observable independent variables X
ij
for j = 1, 2, …, k.
Table 3.2. Description of independent variables for regression analyses
Independent variables Variable type
Study period variables
Variable of interest
High adherence or Binary
High adherence, annualized Binary
Days of follow-up Continuous
Baseline period variables
Number of bleeding episodes or Count
 5 bleeding episodes Binary
ER visit(s) Binary
Inpatient visit(s) Binary
Outpatient procedure(s) Binary
Number of HTC visits Count
Days of hemophilia-related work absenteeism Count
Hours of unpaid caregiver time Count
Hours of paid nurse time Count
Patient characteristic variables
Adult or Binary
Age Continuous
Race Binary
Marital status Binary
Education level Binary
Insurance type Binary
Employment status Categorical
Hemophilia type Binary
Hemophilia severity Binary
Years using prophylaxis Continuous
Hours of distance to HTC Continuous
Abbreviations: HTC - Hemophilia treatment center
The base case regression model examined the impact of annualized adherence to
prophylaxis on ABR among the total included sample of prophylaxis users, adjusting for
60
continuous baseline period bleeding episodes and all other baseline period and participant
characteristic variables. The regression-estimated impact of each independent variable on ABR is
reported in incidence rate ratios (IRR), which equals the coefficient exponentiated, for more
straightforward interpretability and 95% confidence intervals. In order to maximize degrees of
freedom in this relatively small sample, annualized adherence was used as the base case variable
of interest rather than non-annualized adherence, which requires also including days of follow-up
as a covariate. Additionally, a binary indicator for adults versus children was included rather than
a continuous age variable. The negative binomial regression was evaluated in aggregate, as well
as separately in clinically meaningful subgroups, including adults, children, HA, and HB, to
identify differences in the relationships between adherence or other independent variables on
bleeding episodes between groups. Regression-adjusted mean ABR between high and low
adherers was also reported.
The log likelihood, Akaike information criterion (AIC) and Bayesian information
criterion (BIC) were examined to assess model goodness of fit. Log likelihood values closer to 0
indicate better model fit, as do smaller values of AIC and BIC. In addition, model fit compared
to the null model was assessed through likelihood ratio testing, which uses a chi-squared test on
the difference of log likelihoods between two models. In this testing, a p-value of <0.05 indicates
that the model fits the data significantly better than the null model.
3.2.3.3. Sensitivity analyses
Sensitivity analyses were performed to assess the robustness of base case negative
binomial regression estimates for the IRR of the variable of interest. First, base case model
specifications and data cleaning criteria were varied in additional regressions. The base case
regression analyses were conducted in a number of scenarios: a binary variable indicating
61
number of baseline period bleeding episodes 5 replacing the continuous baseline period
bleeding episode variable; non-annualized study period adherence and bleeding episodes and a
covariate for length of follow-up replacing annualized adherence and ABR; and both a binary
variable for baseline period bleeding episodes and non-annualized study period variables. All
regression analyses were also conducted using a dataset in which the length of study period
exclusion criteria was extended to <12 months instead of <6 months in the base case. Finally, all
regression analyses were conducted using a dataset in which the baseline period was the first 12
months of follow-up and the study period was the next 12 months of follow-up compared with a
6-month baseline and 18-month study period in the base case. While these two data cleaning
sensitivity analyses limit the final sample sizes, results from these analyses can help demonstrate
the robustness of the impact of adherence on outcomes in study samples with minimal missing
data and/or stronger baseline period variables.
IRR sensitivity analyses were also conducted on reasonable variations in the assumed
prophylaxis doses for HA and HB. The lower bound tested was 80% of the assumed prophylaxis
doses (22 IU/kg in HA and 28 IU/kg in HB). Because the denominator of the adherence proxy,
which included prophylaxis dose, is entirely multiplicative, this equates to multiplying the
adherence proxy by a factor 1.25. Using the minimum of doses reported in Table 3.1 was not
clinically meaningful, as most individuals with hemophilia are rarely treated prophylactically at
such low doses, especially in the US. The upper bounds tested were the maximum doses reported
in Table 3.1 (40 IU/kg for HA and HB).
A matched cohort analysis was also conducted on the base case dataset as an exploratory
sensitivity analyses. Using this method, high adherers were matched to a cohort of low adherers
based on a propensity score calculated using all independent variables. Matching was carried out
62
using the SAS® macro %gmatch, which uses a greedy matching algorithm, also referred to as
nearest neighbor matching, to select a control (low adherer) for each case (high adherer) that is
the one with the smallest difference between propensity scores.
42
Participant characteristics and
baseline period variables were compared between high and low adherers before and after
matching. The differences in study period bleeding episodes between matched groups were
compared in bivariate analyses to assess robustness of regression analysis results.

3.3. Results
3.3.1. Baseline participant characteristics
A total of 290 participants were included in this study after data cleaning. Of these
participants, 100 individuals who used prophylaxis were included in statistical analyses. Baseline
participant characteristics for excluded (N=187) and included participants (N=290) are presented
in Table 3.3. Compared with excluded participants, included participants were more likely to be
white versus other races (73.1% and 52.4% white, respectively; p<0.0001) and use episodic
treatment versus prophylaxis (65.5% and 55.1% episodic treatment, respectively; p<0.05). No
other differences in characteristics were statistically significant between the two groups.
The mean age of prophylaxis users included in the analyses was 15.9 [(standard
deviation) SD: 12.3], and the mean years of prophylaxis use at the initial interview was 5.9 (SD:
5.1). The sample included 81 participants with HA and 19 with HB, which is highly
representative of actual population-level differences in HA and HB prevalence.
5
As expected,
among included participants, episodic treatment versus prophylaxis users were more likely to be
adults (62.1% vs. 31.0%, p<0.0001), have mild or moderate hemophilia (62.6% vs. 5%,
p<0.0001), and have HB (34.7% vs. 19%, p<0.001). Race, education level, insurance type, and
63
employment status also differed between included participants using episodic treatment versus
prophylaxis.
Table 3.3. Baseline participant characteristics
Variables

All participants
Included participant treatment
regimen
Excluded
(N=187)
Included
(N=290)
P-value

Episodic
(N=190)
Prophylaxis
(N=100)
P-value

Adult
a
89 (47.6%) 149 (51.4%) 0.4195 118 (62.1%) 31 (31.0%) <0.0001
Years of age,
mean (SD)
21.1 (15.1) 22.9 (16.5) 0.353 26.5 (17.3) 15.9 (12.3) <0.0001
Male 184 (98.4%) 287 (99.0%) 0.5857 188 (98.9%) 99 (99.0%) 0.9664
Race <0.0001 0.0112
Non-White 89 (47.6%) 78 (26.9%) 42 (22.1%) 36 (36.0%)
White 98 (52.4%) 212 (73.1%) 148 (77.9%) 64 (64.0%)
Married or with partner 108 (59.3%) 182 (62.8%) 0.4577 119 (62.6%) 63 (63.0%) 0.9508
Education >12 years 117 (64.6%) 206 (71.0%) 0.1459 127 (66.8%) 79 (79.0%) 0.0300
Household income <$20,000 41 (24.4%) 46 (17.1%) 0.0629 30 (17.1%) 16 (17.0%) 0.9799
Insurance type 0.0620 0.0243
Private only 94 (50.3%) 171 (59.0%) 121 (63.7%) 50 (50.0%)
Other 93 (49.7%) 119 (41.0%) 69 (36.3%) 50 (50.0%)
Employment status
b
0.5033 0.0209
Not employed 60 (32.3%) 103 (35.5%) 76 (40.0%) 27 (27.0%)
Part-time 35 (18.8%) 61 (21.0%) 32 (16.8%) 29 (29.0%)
Full-time 91 (48.9%) 126 (43.4%) 82 (43.2%) 44 (44.0%)
Hours of distance to HTC, mean (SD) 1.2 (1.5) 1.5 (2.3) 0.1621 1.5 (2.4) 1.4 (2.1) 0.6761
Hemophilia severity 0.8212 <0.0001
Mild/moderate 78 (41.7%) 124 (42.8%) 119 (62.6%) 5 (5.0%)
Severe 109 (58.3%) 166 (57.2%) 71 (37.4%) 95 (95.0%)
Hemophilia treatment regimen 0.0222
Episodic 103 (55.1%) 190 (65.5%) 190 (100%) n/a
Prophylaxis 84 (44.9%) 100 (34.5%) n/a 100 (100%)
Years on prophylaxis,
mean (SD)
c

7.1 (6.8) 5.9 (5.1) 0.4985 n/a 5.9 (5.1) n/a
Hemophilia type 0.3128 0.0051
Hemophilia B 63 (33.7%) 85 (29.3%) 66 (34.7%) 19 (19.0%)
Hemophilia A 124 (66.3%) 205 (70.7%) 124 (65.3%) 81 (81.0%)
Abbreviations: SD - Standard deviation, HTC - Hemophilia treatment center
All statistics reported in N (%), unless otherwise specified
a
Adults defined as participants ≥18 years old
b
May not add up to 100 percent due to missing data
c
Only applies to participants using prophylaxis

64
3.3.2. Descriptive results
The annualized adherence proxy was censored to the range of 0.0 to 1.0, where higher
values represent higher adherence to prophylaxis, and dichotomized. Among all included
prophylaxis users, 73% of individuals were considered high prophylaxis adherers with adherence
80%. At the 10% significance level, a significantly smaller proportion of high adherers
compared with low adherers were adults (26.0% vs. 44.4%, p<0.1), had baseline period bleeding
episodes 5 (17.8% vs. 33.3%, p<0.1), or any baseline period ER visits (8.2% vs. 22.2%, p<0.1)
(Table 3.4). High versus low adherers also had more mean years on prophylaxis [mean (SD): 6.3
(5.2) vs. 4.6 (4.6) years, p<0.05] and hours of travel distance to the HTC [1.6 (2.2) vs. 0.8 (6.2)
hours, p<0.1].
Differences also existed between participant subgroups by age and hemophilia type.
More children (N=69) versus adults (N=31) and more participants with HA (N=81) versus HB
(N=19) tended to have high adherence to prophylaxis (Appendix A). Study period ABR was
positively skewed in all subgroups (Appendix B). Table 3.5 shows that differences in study
period variables were statistically significant at the 10% level between adults and children but
not between HA and HB, possibly due to the small sample size of participants with HB. High
adherers comprised 61.3% of adults and 78.3% of children (p<0.1) (Table 3.5). Mean ABR was
10.5 (SD: 12.3) among adults and 4.3 (5.8) among children, (p<0.01). No calculated baseline
period variables were significantly different between adults and children. Across all subgroups,
length of follow-up was longest among adults [mean (SD) days: 500 (79)] and significantly
longer than length of follow-up for children [460 (106) days, p<0.05].
Mean unadjusted study period ABR was 6.2 (SD: 8.8) in the total sample. The ABRs for
low and high prophylaxis adherers in participant subgroups are compared in Figure 3.3. Being a
65
low versus high adherer was significantly associated with higher ABR among the total sample of
participants, adult participants, and participants with HA. Low adhering adults had the highest
mean ABR of 16.8 (SD: 17.1), while low adhering children and individuals with HB had the
lowest ABR of 4.2 (3.4) and 4.2 (7.0), respectively. Although not statistically significant, high
versus low adherers had similar ABR [4.3 (6.4) vs. 4.2 (3.4), p=0.95] among children and higher
ABR among participants with HB [5.0 (4.3) vs. 4.2 (7.0), p= 0.79], suggesting a need for
covariate adjustment.
Table 3.4. Unadjusted comparison of independent variables among high and low
prophylaxis adherers
Variables
High adherer
(N=73)
Low adherer
(N=27)
P-value
Baseline period variables:
Number of bleeds, mean (SD) 2.6 (4) 4.8 (6.2) 0.0501
Number of bleeds ≥5 13 (17.8%) 9 (33.3%) 0.0961
≥1 ER visit 6 (8.2%) 6 (22.2%) 0.0557
≥1 hospitalization 6 (8.2%) 1 (3.7%) 0.4320
≥1 outpatient procedure 11 (15.1%) 4 (14.8%) 0.9748
Number of HTC visits, mean (SD) 1.6 (2.2) 2.6 (6.2) 0.7841
Days of work absenteeism,
a
mean (SD) 1.1 (3.5) 1.4 (4.4) 0.7408
Hours of caregiver time,
b
mean (SD) 1.7 (5.8) 3.2 (9.7) 0.5057
Hours of nurse time,
c
mean (SD) 0.1 (0.6) 0.9 (4.2) 0.1291
Baseline participant characteristics
Adult 19 (26.0%) 12 (44.4%) 0.0771
White 46 (63.0%) 18 (66.7%) 0.7355
Married/with partner
a
46 (63.0%) 17 (63.0%) 0.9963
Education >12 years
a
57 (78.1%) 22 (81.5%) 0.7110
Private insurance only 34 (46.6%) 16 (59.3%) 0.2601
Employment status
a
0.7581
Not employed 21 (28.8%) 6 (22.2%)
Part-time 20 (27.4%) 9 (33.3%)
Full-time 32 (43.8%) 12 (44.4%)
Hemophilia A 61 (83.6%) 20 (74.1%) 0.2830
Severe hemophilia 70 (95.9%) 25 (92.6%) 0.5017
Years on prophylaxis, mean (SD) 6.3 (5.2) 4.6 (4.6) 0.0499
Hours of distance to HTC, mean (SD) 1.6 (2.2) 0.8 (6.2) 0.0593
Abbreviations: SD - standard deviation, HTC - Hemophilia treatment center
All statistics reported in N (%), unless otherwise specified
a
Applies to participants ≥18 years old, or parents of participants <18 years old
b
Hours of unpaid caregiver help with household chores such as cooking, cleaning, transportation, or shopping
c
Hours of paid skilled nursing, medication administration (such as clotting factor infusion), wound care and/or
physical therapy
66
Table 3.5. Calculated independent variables and outcomes for participant subgroups
Calculated variables
Total
sample
(N=100)
Participant age Participant hemophilia type

Adults
(N=31)
Children
(N=69)
P-
value

HA
(N=81)
HB
(N=19)
P-
value
Study period variables
High adherence,
a
N (%) 76
(76.0%)
19
(61.3%)
57
(82.6%)
0.0210 63
(77.8%)
13
(68.4%)
0.3901
High annualized
adherence,
a,b
N (%)
73
(73.0%)
19
(61.3%)
54
(78.3%)
0.0771 61
(75.3%)
12
(63.2%)
0.2830
Days of follow-up,
mean (SD)
472
(100)
500
(79)
460
(106)
0.0101 477
(99)
451
(103)
0.6603
Number of bleeds,
mean (SD)
8.04
(11.1)
13.6
(14.6)
5.5
(8.1)
0.0022 8.6
(11.8)
5.7
(7.4)
0.2896
Annualized bleeds (ABR),
b

mean (SD)
6.2
(8.8)
10.5
(12.3)
4.3
(5.8)
0.0032 6.5
(9.4)
4.7
(5.3)
0.506
Baseline period variables
Number of bleeds,
mean (SD)
3.2
(4.8)
4.3
(6.3)
2.7
(3.9)
0.7087 3.5
(5.1)
1.99
(3.2)
0.0804
≥5 bleeds, N (%) 22
(22.0%)
8
(25.8%)
14
(20.3%)
0.5380 19
(23.5%)
3
(15.8%)
0.4678
≥1 ER visit, N (%) 12
(12.0%)
2
(6.5%)
10
(14.5%)
0.2524 8
(9.9%)
4
(21.1%)
0.1773
≥1 hospitalization, N (%) 7
(7.0%)
2
(6.5%)
5
(7.2%)
0.8855 7
(8.6%)
0
(0.0%)
0.1839
≥1 outpatient procedure,
N (%)
15
(15.0%)
5
(16.1%)
10
(14.5%)
0.8322 15
(18.5%)
0
(0.0%)
0.0419
Number of HTC visits,
mean (SD)
1.9
(3.7)
1.4
(2.01)
2.1
(4.2)
0.4194 1.5
(2.2)
3.7
(7.02)
0.4316
Days of work absenteeism,
c

mean (SD)
1.2
(3.7)
2.2
(6.1)
0.8
(1.9)
0.5695 1.4
(4)
0.6
(2.1)
0.2771
Hours of caregiver time,
d
mean (SD)
2.1
(7.02)
1.3
(4.4)
2.5
(7.9)
0.3244 1.99
(7.4)
2.6
(5.4)
0.6859
Hours of nurse time,
e

mean (SD)
0.3
(2.3)
0.9
(3.96)
0.1
(0.5)
0.4433 0.1
(0.5)
1.4
(5.04)
0.0932
Abbreviations: SD - standard deviation, ABR - Annualized bleed rate, HTC - Hemophilia treatment center, HA -
Hemophilia A, HB - Hemophilia B
a
High adherence defined as adherence proxy ≥80%
b
Annualized variables = (Variable)*(365/Days of follow-up)
c
Applies to participants ≥18 years old, or parents of participants <18 years old
d
Hours of unpaid caregiver help with household chores such as cooking, cleaning, transportation, or shopping
e
Hours of paid skilled nursing, medication administration (such as clotting factor infusion), wound care and/or
physical therapy
67
Figure 3.3. Unadjusted study period annualized bleed rate among high and low
prophylaxis adherers
Abbreviations: ABR - Annualized bleed rate, HA - Hemophilia A, HB - Hemophilia B
Numeric labels represent mean (standard deviation) of ABR for each participant subgroup
*** p<0.01, ** p<0.05, * p<0.1
3.3.3. Negative binomial regression results
The base case negative binomial regression model was conducted separately using a
continuous baseline period bleeding episode variable and also using a binary baseline period
bleeding episode variable indicating ≥5 bleeds in the total sample and among participant
subgroups by age and hemophilia type. The regression for participants with HB did not converge
due to limited sample size after data cleaning (N=19). Based on the AIC and BIC, all subgroup
models had improved fit compared to the total sample model, with the adult model having best
fit (Tables 3.6 and 3.7). In adults and children, a continuous variable for age was included as an
9.8 (13.1)
16.8 (17.1)
4.2 (3.4)
11.7 (14.3)
4.2 (7.0)
4.9 (6.2)
6.5 (5.4)
4.3 (6.4)
4.8 (6.5)
5.0 (4.3)
0
2
4
6
8
10
12
14
16
18
Total (N=100)*** Adults (N=31)** Children (N=69) HA (N=81)*** HB (N=19)
Mean study period ABR
Low Adherers
High Adherers
68
independent variable. This variable was not included in the total sample and HA models as it
worsened model fit and was not significantly associated with ABR.
The IRRs for ABR and associated 95% confidence intervals are presented in Tables 3.6
and 3.7. In the base case model with a continuous baseline bleed variable, low compared to high
adherence was significantly associated with 69% higher ABR in the total sample (p<0.05),
almost 6 times higher ABR among adults (p<0.01), and 2.37 times higher ABR among
participants with HA (p<0.01) (Table 3.6). Among children, low compared with high adherence
is associated with 11% higher ABR, but the IRR was not statistically significant. In all four
groups, the likelihood ratio test suggested a significantly better fit compared with the null model
for study period ABR (all p<0.01).
As expected, higher baseline bleeding episodes, included as a predictor of study period
outcomes and underlying disease severity, was directly associated with higher study period ABR
and significant in the total sample and adults (p<0.05 and p<0.1, respectively). Further, HB
compared with HA was generally associated with lower ABR. Hemophilia severity was not
significant due to the small number of participants with mild/moderate hemophilia who used
prophylaxis. Being a child versus adult was significantly associated with 43% lower ABR among
the total sample (p<0.05) and 41% lower ABR among participants with HA (p<0.1). Having no
baseline ER visits or outpatient procedures was associated with higher ABR, but the IRRs were
not significant in most groups. Baseline hours of paid skilled nursing time were associated with
lower ABR across subgroups, and the effect was significant in the total sample (p<0.05),
children (p<0.1), and participants with HA (p<0.05). Having public or no insurance (no
insurance N=1, public N=39, both public and private N=10) compared with only private
insurance was significantly associated with higher ABR across subgroups. Finally, more hours of
69
travel distance to the HTC was generally associated with higher ABR, and the IRR was
significant in adults (p<0.01). In adults, additional variables significantly associated with ABR
included hours of unpaid caregiver time, race, and education level (all p<0.1). In children, days
of work absenteeism was also significant (p<0.1).
The model specifications using a binary baseline bleed variable showed similar trends in
the impact of adherence on ABR, but had slightly worse fit compared with the model using a
continuous baseline bleed variable (Table 3.7). Although the magnitude of the IRRs varied
somewhat between model specifications, most of the significant variables in the models with a
continuous baseline bleed variable were also significant here. Low compared with high
adherence was significantly associated with 99% higher ABR in the total sample (p<0.01),
almost 6 times higher ABR among adults (p<0.01), and 2.62 times higher ABR in participants
with HA (p<0.01). In children, low adherence is associated with 21% higher ABR compared
with low adherence, but, again, the IRR was not statistically significant. Having baseline bleeds
<5 compared with 5 and being a child versus adult was generally associated with lower study
period ABR, but most results were not significant.
Mean regression-adjusted study period ABRs are shown in Figure 3.4, which adjusted for
a continuous baseline bleed variable, and Figure 3.5, which adjusted for a binary variable.
Compared with unadjusted ABRs in Figure 3.3, low versus high adherence remained
significantly associated with higher ABR in the total sample, adults, and participants with HA in
both model specifications (all p<0.05). After adjustment, mean ABR was highest among low
adhering participants with HA. In addition, mean adjusted ABRs in high and low adhering adults
were noticeably lower than unadjusted ABRs among the same groups. In children, the adjusted
ABR was lower for high versus low adherers, but remained not statistically significant. In the
70
model including a continuous baseline bleed variable, mean ABR for low versus high adherers
was 6.3 [standard error (SE): 2.4] versus 3.7 (1.4) among the total sample (p<0.05), 4.0 (2.3)
versus 0.7 (0.5) among adults (p<0.01), 2.6 (1.4) versus 2.3 (1.1) among children (p=0.75), and
10.8 (4.4) versus 4.5 (1.9) among participants with HA (p<0.01). Similarly, in the model
including a binary baseline bleed variable, mean ABR for low versus high adherers was 6.7 (SE:
2.7) versus 3.3 (1.3) among the total sample (p<0.01), 4.8 (3.0) versus 0.8 (0.6) among adults
(p<0.01), 2.6 (1.4) versus 2.1 (1.1) among children (p=0.58), and 11.6 (4.9) versus 4.4 (2.0)
among participants with HA (p<0.01).
71
Table 3.6. Negative binomial regression for ABR: IRR results with continuous baseline
period bleeding episode variable
Variables
Total sample
(N=100)
Participant age Hemophilia A
(N=81) Adult (N=31) Child (N=69)
Low adherence
(Ref: High adherence)
a

1.69**
(1.01-2.85)
5.996***
(2.74-13.15)
1.11
(0.57-2.16)
2.37***
(1.35-4.15)
Baseline period variables:
No ER visits
(Ref: ≥1 ER visits)
1.59
(0.64-3.96)
10.52***
(1.87-59.09)
1.55
(0.52-4.63)
1.69
(0.52-5.44)
No hospitalizations
(Ref: ≥1 hospitalizations)
0.82
(0.24-2.82)
0.24
(0.02-2.42)
0.86
(0.15-5.1)
0.69
(0.19-2.42)
No outpatient procedures
(Ref: ≥1 outpatient procedures)
1.63
(0.57-4.66)
1.25
(0.28-5.56)
1.95
(0.44-8.6)
1.29
(0.42-3.97)
Number of HTC visits 0.996
(0.94-1.06)
0.97
(0.84-1.11)
1.005
(0.93-1.08)
0.92
(0.81-1.04)
Days of work absenteeism
b
1.01
(0.95-1.08)
1.01
(0.95-1.08)
1.16*
(0.99-1.37)
1.02
(0.96-1.09)
Hours of caregiver time
c
1.01
(0.98-1.04)
0.93*
(0.85-1.01)
1.02
(0.98-1.05)
1.01
(0.98-1.04)
Hours of nurse time
d
0.66 **
(0.43-0.99)
0.89
(0.73-1.09)
0.24*
(0.06-1.03)
0.46**
(0.23-0.93)
Number of bleeds 1.08**
(1.02-1.14)
1.06*
(0.99-1.14)
1.04
(0.96-1.14)
1.04
(0.98-1.11)
Baseline participant characteristics
Child
(Ref: Adult)
0.57**
(0.33-0.99)
-- -- 0.59*
(0.33-1.05)
Non-white
(Ref: White)
e

1.17
(0.69-1.98)
4.59*
(2.36-8.93)
0.97
(0.46-2.04)
1.07
(0.61-1.88)
Not married/not with partner
(Ref: Married/with partner)
b

1.19
(0.69-2.05)
2.03
(0.85-4.87)
0.8
(0.4-1.6)
1.24
(0.7-2.23)
Education ≤12 years
(Ref: >12 years)
b

1.29
(0.78-2.14)
2.06*
(1.1-3.87)
0.93
(0.45-1.92)
1.18
(0.64-2.17)
Public or no insurance
(Ref: Private insurance only)
1.57**
(1.02-2.44)
1.76*
(0.998-3.09)
2.62***
(1.3-5.27)
1.91***
(1.17-3.11)
Not employed
(Ref: Full-time)
b

1.09
(0.65-1.8)
1.09
(0.51-2.35)
1.38
(0.67-2.84)
1.004
(0.57-1.77)
Part-time
(Ref: Full-time)
b

1.01
(0.61-1.66)
0.68
(0.3-1.58)
1.11
(0.59-2.07)
0.95
(0.58-1.55)
Hemophilia B
(Ref: Hemophilia A)
0.87
(0.47-1.61)
0.35**
(0.14-0.85)
0.78
(0.35-1.72)
--
Mild/moderate hemophilia
(Ref: Severe hemophilia)
1.25
(0.49-3.23)
0.14***
(0.04-0.54)
1.44
(0.39-5.26)
1.38
(0.47-4.09)
Years on prophylaxis 1.001
(0.96-1.05)
0.98
(0.93-1.04)
0.96
(0.89-1.05)
0.99
(0.94-1.04)
Hours of distance to HTC 1.05
(0.95-1.17)
1.59***
(1.18-2.14)
1.08
(0.95-1.22)
1.07
(0.97-1.17)
Years of age -- 0.99
(0.97-1.02)
0.99
(0.9-1.08)
--
Goodness of fit measures
Deviance/degrees of freedom 1.4568 3.9209 1.6172 1.5448
Log likelihood -257.1788 -81.4558 -158.0626 -208.0315
AIC 558.3575 206.9115 360.1252 458.063
BIC 615.6713 238.4592 409.2755 508.3464
Abbreviations: ABR - Annualized bleed rate, IRR - Incidence rate ratio, Ref - Reference group, ER - Emergency room, HTC -
Hemophilia treatment center, AIC - Akaike information criterion, BIC Bayesian information criterion
Results presented in IRR (95% confidence interval); *** p<0.01, ** p<0.05, * p<0.1
a
Low adherer defined as participants with adherence proxy <80%
b
Applies to participants ≥18 years old, or parents of participants <18 years old
c
Hours of unpaid caregiver help with household chores such as cooking, cleaning, transportation, or shopping
d
Hours of paid skilled nursing, medication administration, wound care and/or physical therapy
e
Non-white races include black, Hispanic, Asian Pacific Island, and other
72
Table 3.7. Negative binomial regression for ABR: IRR results with binary baseline period
bleeding episode variable
Variables
Total sample
(N=100)
Participant age Hemophilia A
(N=81) Adult (N=31) Child (N=69)
Low adherence
(Ref: High adherence)
a

1.99***
(1.2-3.32)
5.92***
(2.74-12.77)
1.21
(0.62-2.33)
2.62***
(1.55-4.45)
Baseline period variables:
Number of bleeds <5
(Ref: ≥5 bleeds)
0.65
(0.38-1.11)
0.53
(0.24-1.2)
1.06
(0.5-2.24)
0.81
(0.46-1.4)
No ER visits
(Ref: ≥1 ER visits)
1.65
(0.65-4.22)
8.27**
(1.43-47.79)
1.65
(0.55-4.99)
1.68
(0.51-5.46)
No hospitalizations
(Ref: ≥1 hospitalizations)
1.36
(0.42-4.4)
0.26
(0.02-2.72)
0.94
(0.16-5.57)
0.82
(0.24-2.79)
No outpatient procedures
(Ref: ≥1 outpatient procedures)
1.19
(0.43-3.3)
1.27
(0.27-6)
1.78
(0.4-7.85)
1.11
(0.38-3.29)
Number of HTC visits 0.99
(0.93-1.06)
0.96
(0.83-1.12)
1.002
(0.93-1.08)
0.9*
(0.8-1.02)
Days of work absenteeism
b
1.04
(0.987-1.1)
1.03
(0.97-1.08)
1.23***
(1.05-1.43)
1.04
(0.98-1.1)
Hours of caregiver time
c
1.002
(0.97-1.03)
0.92**
(0.84-0.995)
1.02
(0.98-1.055)
1.003
(0.98-1.03)
Hours of nurse time
d
0.68*
(0.44-1.03)
0.899
(0.73-1.1)
0.28*
(0.07-1.13)
0.46**
(0.22-0.95)
Baseline participant characteristics
Child
(Ref: Adult)
0.62*
(0.35-1.09)
-- -- 0.62
(0.35-1.1)
Non-white
(Ref: White)
e

1.27
(0.74-2.19)
4.54***
(2.33-8.88)
1.02
(0.48-2.16)
1.14
(0.65-2)
Not married/not with partner
(Ref: Married/with partner)
b

1.43
(0.84-2.45)
2.7***
(1.32-5.55)
0.83
(0.41-1.66)
1.41
(0.82-2.42)
Education ≤12 years
(Ref: >12 years)
b

1.14
(0.69-1.9)
2.02**
(1.05-3.89)
0.93
(0.45-1.93)
1.09
(0.6-1.97)
Public or no insurance
(Ref: Private insurance only)
1.74**
(1.12-2.72)
2.002**
(1.11-3.6)
2.77*
(1.37-5.61)
2.03***
(1.26-3.29)
Not employed
(Ref: Full-time)
b

1.02
(0.6-1.71)
0.92
(0.44-1.95)
1.52
(0.73-3.15)
0.95
(0.54-1.66)
Part-time
(Ref: Full-time)
b

0.8 (0.5-1.29) 0.51*
(0.26-1.02)
1.08
(0.57-2.02)
0.828
(0.52-1.31)
Hemophilia B
(Ref: Hemophilia A)
0.84
(0.44-1.61)
0.32**
(0.12-0.83)
0.78
(0.35-1.74)
--
Mild/moderate hemophilia
(Ref: Severe hemophilia)
1.13
(0.43-2.97)
0.16***
(0.04-0.64)
1.29
(0.35-4.77)
1.31
(0.43-3.98)
Years on prophylaxis 0.99
(0.95-1.04)
0.97
(0.93-1.02)
0.95
(0.87-1.04)
0.99
(0.94-1.04)
Hours of distance to HTC 1.1**
(1.002-1.21)
1.53***
(1.11-2.12)
1.13**
(1.03-1.24)
1.09**
(1.01-1.19)
Years of age -- 1.001
(0.97-1.03)
0.998
(0.91-1.09)
--
Goodness of fit measures
Deviance/Degrees of freedom 1.449 3.8149 1.6087 1.5452
Log likelihood -259.0852 -81.7955 -158.4897 -208.556
AIC 562.1705 207.5909 360.9794 459.1121
BIC 619.4842 239.1386 410.1298 509.3955
Abbreviations: ABR - Annualized bleed rate, IRR - Incidence rate ratio, Ref - Reference group, ER - Emergency room, HTC -
Hemophilia treatment center, AIC - Akaike information criterion, BIC Bayesian information criterion
Results presented in IRR (95% confidence interval); *** p<0.01, ** p<0.05, * p<0.1
a
Low adherer defined as participants with adherence proxy <80%
b
Applies to participants ≥18 years old, or parents of participants <18 years old
c
Hours of unpaid caregiver help with household chores such as cooking, cleaning, transportation, or shopping
d
Hours of paid skilled nursing, medication administration, wound care and/or physical therapy
e
Non-white races include black, Hispanic, Asian Pacific Island, and other
73
Figure 3.4. Adjusted study period annualized bleed rate among high and low prophylaxis
adherers (continuous baseline bleed variable)

Abbreviations: HA - Hemophilia A, ABR - Annualized bleed rate
Numeric labels represent mean (standard error) of ABR for each participant subgroup
*** p<0.01, ** p<0.05, * p<0.1
Figure 3.5. Adjusted study period annualized bleed rate among high and low prophylaxis
adherers (binary baseline bleed variable)

Abbreviations: HA - Hemophilia A, ABR - Annualized bleed rate
Numeric labels represent mean (standard error) of ABR for each participant subgroup
*** p<0.01, ** p<0.05, * p<0.1
6.3 (2.4)
4.0 (2.3)
2.6 (1.4)
10.8 (4.4)
3.7 (1.4)
0.7 (0.5)
2.3 (1.1)
4.5 (1.9)
0
2
4
6
8
10
12
14
Total (N=100)** Adults (N=31)*** Children (N=69) HA (N=81)***
Mean study period ABR
Low Adherers High Adherers
6.7 (2.7)
4.8 (3.0)
2.6 (1.4)
11.6 (4.9)
3.3 (1.3)
0.8 (0.6)
2.1 (1.1)
4.4 (2.0)
0
2
4
6
8
10
12
14
Total (N=100)*** Adults (N=31)*** Children (N=69) HA (N=81)***
Mean study period ABR
Low Adherers High Adherers
74
3.3.4. Sensitivity analyses
The base case dataset was varied in three scenarios (see Section 3.2.3.3) to assess the
robustness of the impact of adherence to prophylaxis on bleeding episodes, in terms of IRRs.
When using a <12 month study period exclusion criteria, 85 participants were included (N=30
adults, N=55 children, N=69 HA and N=16 HB). When using the first 12 months as the baseline
and the next 12 months as the study period, 87 participants were included (N=31 adults, N=56
children, N=69 HA and N=18 HB). Compared with the base case, additional excluded
participants in these two scenarios tended to be children and have HA.
A forest plot of IRRs for low versus high adherence on study period bleeding episodes in
the base case and sensitivity analysis scenarios are presented in Figure 3.6. IRR estimates were
>1.0 in all scenarios and subgroups, indicating higher study period bleeding episodes for low
versus high adherers, and significant results were generally consistent within subgroups. Among
the total sample, low adherence was significantly associated with higher bleeding episodes (i.e.
95% confidence intervals did not cross 1.0) for all sensitivity analysis models except using a 12-
month baseline period, and all IRRs ranged 1.37-2.44. Low adherence was also significantly
associated with higher bleeding episodes among adults in all but one model, with IRR ranging
3.53-6.30. Adherence was not significant in any of the pediatric models, but IRRs ranged 1.04-
1.39. Finally, adherence was significant in all HA models, with IRR ranging 2.09-3.18.
A similar forest plot in Figure 3.7 shows the IRR sensitivity analysis results of varying
the assumed prophylaxis dose to 80% of the base case and to maximum doses from Table 3.1.
Again, IRRs were >1.0 in all models and significant results were generally consistent within
subgroups. The IRR of study period bleeding episodes for low versus high adherers ranged 1.40-
75
1.75 among total sample models, 3.52-6.00 among adult models, 1.03-1.29 among pediatric
models, and 1.70-2.39 among HA models.
Overall, although the estimated IRR of study period bleeding episodes for low versus
high adherers did vary somewhat between regression models within participant subgroups, the
sensitivity analysis results demonstrated that the impact of adherence on bleeding episodes was
significant and generally robust in all scenarios tested.
3.3.5. Matched cohort sensitivity analysis
The distribution of independent variables among high and low adherers after propensity
score cohort matching is shown in Appendix C. High and low adherers were matched one-to-one
using the base case dataset and either a continuous (N=16 cases and controls) or binary (N=17
cases and controls) baseline bleed variable. After matching, there were no significant differences
between high and low adhering groups when matching was conducted using the continuous
bleed variable. When using the binary variable, only days of work absenteeism differed
significantly at the 5% level between the two groups.
After matching, low versus high adherers had higher mean ABR when using the
continuous bleed variable [mean (SD): 8.5 (9.3) vs. 4.6 (4.9), p=0.14], and, significantly, when
using the binary bleed variable [8.1 (9.2) vs. 3.9 (4.4), p<0.1] (Figure 3.8). The negative impact
of low adherence on clinical outcomes is consistent with regression-adjusted mean ABRs
reported in Section 3.3.3. Due to the small sample size after matching, high statistical
significance was not found and subgroup analyses were not possible.

76

Figure 3.6. Sensitivity analyses of negative binomial regression models among participant subgroups using varied data
inclusion criteria
77
Figure 3.7. Sensitivity analyses of negative binomial regression models among participant
subgroups using varied prophylaxis dose

Abbreviations: BL - baseline, IRR - incidence rate ratios
All estimates of IRR >1.0
Error bars represent IRR 95% confidence intervals
Figure 3.8. Study period annualized bleeding episodes among 1:1 matched high and low
adherers

Abbreviations: ABR - Annualized bleed rate
Numeric labels represent mean (standard deviation) of ABR in each group
*** p<0.001, ** p<0.05, * p<0.1
8.5 (9.3)
8.1 (9.2)
4.6 (4.9)
3.9 (4.4)
0
1
2
3
4
5
6
7
8
9
10
Matched on continuous baseline period
bleeding episode variable (N=32)
Matched on binary baseline period
bleeding episode variable (N=34)*
Mean study period ABR
Low Adherer
High Adherer
78
3.4. Discussion and limitations
3.4.1. Discussion
Using prospective, longitudinal data from HUGS Va and Vb, this study found that
adherence to prophylaxis had a significant impact on bleeding episodes among individuals with
hemophilia. Overall, regression-adjusted mean ABR was significantly higher for low versus high
adherers. In negative binomial regressions including a broad set of independent variables, low
compared with high adherence was associated with a statistically significant 69% higher number
of bleeding episodes, the main clinical outcome of interest in evaluating hemophilia care. In
subgroups, the magnitude of the impact of adherence on bleeds varied, but the direction was
consistent and the impact was highly significant among adults and individuals with HA. A
significant impact was not observed among children as over 78% of children had high adherence,
resulting in limited variation in the variable of interest. It is likely the impact of adherence
among children could be significant in larger samples with more power to detect an effect. All
findings were robust in the majority of sensitivity analyses conducted.
This work significantly expands upon many of the conclusions from previously published
studies, and also provides new evidence on the negative consequences of poor adherence in
subgroups within the hemophilia population. Thus far, the rarity and geographic dispersion of
these individuals has hindered obtaining datasets with sufficient sample size and richness of
information to link the effect of treatment decisions and adherence to outcomes over time. There
is some preliminary evidence demonstrating relationships between adherence and hemophilia-
related outcomes.
3,4,30,31,34,36
In a sample of 78 children and adolescents with HA, García-Dasí et
al. found a statistically significant difference in number of bleeding episodes between the
adherent (mean 1.4) and infra-adherent (mean 4.5) groups.
3
Overall, 52.6% of the sample was
79
adherent, 33.3% was infra-adherent and 14.1% was over adherent. Because this study only
included individuals aged 6-20 and did not collect detailed patient, disease, and treatment
information, these results do not encompass differences among older hemophilia populations and
are based on cross-sectional bivariate analyses. Another recent study used the Validated
Hemophilia Regimen Treatment Adherence Scale for Prophylaxis (VERITAS-Pro) to quantify
adherence to prophylaxis and negative binomial regression modeling to assess the relationship
between adherence and bleeding events among a sample of 110 adults or children with HA or
HB.
4
Based on VERITAS-Pro score <57, 74.5% of adults and 92.7% of children were adherent.
The study found that poorer adherence was associated with a greater number of bleeding
episodes in adults, but no significant association was found in children. Similar to patterns of
adherence found in HUGS, the range of adherence scores in this study was narrower in children
versus adults. However, the pediatric model only incorporated adherence as a predictor, while
age was also included as a covariate in the adult model. Thus, limited conclusions can be drawn
from previously published studies, which have not evaluated the true impact of adherence on
clinical outcomes while controlling for patient heterogeneity and potential endogeneity.
The current analysis controlled for a rich set of variables describing patient and
hemophilia characteristics, underlying disease severity, and baseline level of management to
evaluate the impact of adherence on bleeding episodes in a number of regression models.
Including these independent variables is critical to disentangle the impact of adherence on
outcomes from potential confounding due to other individual and disease characteristics and
behaviors. Moreover, HUGS’ relatively large sample size compared with previous datasets used
for hemophilia research and extensive set of variables enabled subgroup analyses identifying
important variations within the hemophilia population. Overall, this study found a larger
80
proportion of high adherers compared with some other studies, which have reported that less
than 40% of individuals with hemophilia have high adherence.
30,31
However, measures of
adherence can vary greatly between studies, which limits direct comparisons between findings
and conceptualization of results.
43,44
In order to test the robustness of the assumptions used to
calculate the adherence proxy and final dataset used in this study, numerous sensitivity analyses
were conducted around the methods, and found that results largely remained consistent.
This study also showed that other independent variables were associated with bleeding
episodes. For example, being a child compared with adult was associated with significantly
lower ABR in both bivariate and regression analyses. Negative binomial regression suggested
that having no baseline ER visits or outpatient procedures was associated with higher ABR.
Although the effect was not always significant, it is possibly explained by risk-adverse behavior
and higher treatment adherence for improved bleed protection among individuals who have
recently experienced serious events requiring such healthcare resource utilization. Having private
insurance only was also significantly associated with higher ABR, suggesting that patients
experiencing more severe outcomes tended to be on Medicaid or some other form of public
insurance. In addition, more hours of nursing time and less hours of travel time required to reach
the HTC were significantly associated with lower ABR. This suggests that level of management
and barriers to care may have important implications on clinical outcomes. These new and
preliminary findings imply that multiple aspects beyond treatment choices could impact bleeding
episodes and that hemophilia care could potentially be systematically individualized for
improved outcomes.
The results from subgroup analyses by participant age and hemophilia type confirmed
results from previous studies demonstrating differences between these groups. Adherence to
81
prophylaxis tends to be highest in children, and decreases significantly with age.
34,35
This study
found that adults had significantly higher ABR and more widely distributed estimates of
adherence than children. There is also growing evidence suggesting HA is more severe than
HB.
45
Although regression analysis was not possible in the small HB sample, unadjusted mean
ABRs were higher among participants with HA than HB. Further, overall and subgroup mean
ABR in HUGS are higher than those reported in highly regulated clinical trial settings.
15,16,28,29

As such, these suboptimal clinical outcomes observed in routine clinical practice and differences
between subgroups emphasize a need for clotting factor treatment regimens optimized for
individual characteristics and clinical goals.
In all, this study presents valuable new evidence and perspectives linking low adherence
to a higher number of bleeding episodes among individuals with hemophilia. Because bleeding
episodes can result in debilitating arthropathy, decreased QoL, and increased healthcare
expenditures, more research on the impact, variations, and drivers of adherence is critical for
efforts to cost-effectively improve hemophilia care and outcomes.
33
Currently, prophylactic
hemophilia treatment is expensive and highly individualized, but guidelines for treatment
optimization have not been established due to lack of concrete evidence linking outcomes to
specific treatment regimens or patterns and other individual characteristics or
behaviors.
17,22,25,27,46-48
Future research is needed to better understand the impact of treatment
patterns on other hemophilia-related outcomes, the long-term consequences of poor adherence,
and the socio-economic implications of poor adherence on patients and healthcare systems.
3.4.2. Limitations
As with any study, the results of these analyses should be interpreted in light of the
limitations. Selection bias is possible due to the observational nature of HUGS Va and Vb.
82
Participants were not randomly selected or randomized to treatment regimens. In addition, data
cleaning was conducted to select a dataset of participants with minimal incomplete records and
sufficient length of follow-up. However, while missing data was minimized and participant
characteristics did not differ significantly between excluded and included groups, there were still
differences in length of follow-up that required calculation of annualized study period variables
or adjustment for follow-up time. The assumption that annualized variables are representative of
actual patterns is supported by the fact that mean length of follow-up exceed one year in all
groups. This study also minimized sources of potential selection biases by using longitudinal
data and including a rich set of explanatory variables in the regression analyses. Further, this
study is relatively large considering the rarity of hemophilia, but small sample size limited the
subgroups analyses possible and the ability to detect a significant impact of adherence in some
groups. The study also included individuals receiving care at 10 of 141 HTCs across the US.
Because approximately 30% of the US hemophilia population is not treated at HTCs and it is
unclear whether individuals treated at HTCs are different from those treated elsewhere, the
current findings may have limited generalizability to a broader population.
49

Although HUGS includes a rich dataset of variables, there are limitations to the data. The
results from this study may be subject to recall or social response bias as information on socio-
demographics, bleeding episodes, and work productivity losses were participant-reported.
Additionally, prescribing information or clotting factor infusion logs were not collected. Without
such data, calculations of adherence are limited by assumptions regarding actual utilization since
observed dispensing does not always indicate appropriate and timely infusions. Given the lack of
information, the adherence proxy used in this study was calculated based on assumptions for
“continuous prophylaxis.” Nevertheless, sensitivity analyses suggested that the impact of
83
adherence found here was robust to the majority of reasonable variations in the regression
models and assumptions.
Lastly, the new findings from this study do not paint the entire picture regarding
adherence. While this study found a significant and robust impact of adherence on bleeding
episodes, it did not identify the socio-demographic and clinical characteristics that influence
adherence. Future studies that link the drivers of adherence to more long-term outcomes will be
able to build on the results from this study in efforts to optimize care for meaningful
improvements to individuals with hemophilia.

3.5. Conclusions
This study is the first to demonstrate the impact of adherence to prophylaxis on bleeding
episodes after adjusting for a rich set of individual and clinical characteristics in regression
analyses using longitudinal data. Different approaches were used to test assumptions and control
for potential endogeneity and simultaneous causality, and showed consistent results. Overall, low
compared with high adherence was significantly associated with higher bleeding episodes.
Robust differences in adherence and outcomes were found between participant subgroups,
highlighting a need for more tailored approaches to individualize prophylaxis regimens for
optimal clinical goals. The findings indicate that efforts to increase adherence to prophylaxis
may lead to clinical benefits with longer-term QoL and economic implications. Further research
is needed to understand the drivers of adherence and impacts of adherence beyond bleeding
episodes in order to shape decisions for improved hemophilia care among patients, healthcare
practitioners, and policy makers.
84
3.6. Chapter references
1. Sabaté E. Adherence to Long-Term Therapies: Evidence for Action. Geneva,
Switzerland: World Health Organization;2003.
2. Brown MT, Bussell JK. Medication adherence: WHO cares? Mayo Clinic Proceedings.
2011;86(4):304-314.
3. Garcia-Dasi M, Aznar JA, Jimenez-Yuste V, et al. Adherence to prophylaxis and quality
of life in children and adolescents with severe haemophilia A. Haemophilia.
2015;21(4):458-464.
4. Krishnan S, Vietri J, Furlan R, Duncan N. Adherence to prophylaxis is associated with
better outcomes in moderate and severe haemophilia: results of a patient survey.
Haemophilia. 2015;21(1):64-70.
5. Soucie JM, Evatt B, Jackson D. Occurrence of hemophilia in the United States. The
Hemophilia Surveillance System Project Investigators. American Journal of Hematology.
1998;59(4):288-294.
6. Mejia-Carvajal C, Czapek EE, Valentino LA. Life expectancy in hemophilia outcome.
Journal of Thrombosis and Haemostasis. 2006;4(3):507-509.
7. Luck JV, Jr., Silva M, Rodriguez-Merchan EC, Ghalambor N, Zahiri CA, Finn RS.
Hemophilic arthropathy. The Journal of the American Academy of Orthopaedic
Surgeons. 2004;12(4):234-245.
8. Witkop M, Lambing A, Divine G, Kachalsky E, Rushlow D, Dinnen J. A national study
of pain in the bleeding disorders community: a description of haemophilia pain.
Haemophilia. 2012;18(3):e115-119.
9. Srivastava A, Brewer AK, Mauser-Bunschoten EP, et al. Guidelines for the management
of hemophilia. Haemophilia. 2013;19(1):e1-47.
10. National Hemophilia Foundation MASAC [Medical and Scientific Advisory Council]
Recommendation 179: MASAC Recommendation Concerning Prophylaxis (Regular
Administration of Clotting Factor Concentrate to Prevent Bleeding). 2007;
http://www.hemophilia.org/NHFWeb/MainPgs/MainNHF.aspx?menuid=57&contentid=1
007. Accessed May 22, 2015, 2015.
11. Gringeri A, Lundin B, von Mackensen S, Mantovani L, Mannucci PM. A randomized
clinical trial of prophylaxis in children with hemophilia A (the ESPRIT Study). Journal
of Thrombosis and Haemostasis. 2011;9(4):700-710.
12. Manco-Johnson MJ, Abshire TC, Shapiro AD, et al. Prophylaxis versus episodic
treatment to prevent joint disease in boys with severe hemophilia. The New England
Journal of Medicine. 2007;357(6):535-544.
13. Manco-Johnson MJ, Kempton CL, Reding MT, et al. Randomized, controlled, parallel-
group trial of routine prophylaxis vs. on-demand treatment with sucrose-formulated
recombinant factor VIII in adults with severe hemophilia A (SPINART). Journal of
Thrombosis and Haemostasis. 2013;11(6):1119-1127.
14. Royal S, Schramm W, Berntorp E, et al. Quality-of-life differences between prophylactic
and on-demand factor replacement therapy in European haemophilia patients.
Haemophilia. 2002;8(1):44-50.
15. Ozelo M, Misgav M, Abdul Karim F, et al. Long-term patterns of safety and efficacy of
bleeding prophylaxis with turoctocog alfa (NovoEight®) in previously treated patients
85
with severe haemophilia A: interim results of the guardian™2 extension trial.
Haemophilia. 2015;21(5):e436-e439.
16. Windyga J, Lissitchkov T, Stasyshyn O, et al. Pharmacokinetics, efficacy and safety of
BAX326, a novel recombinant factor IX: a prospective, controlled, multicentre phase I/III
study in previously treated patients with severe (FIX level <1%) or moderately severe
(FIX level ≤2%) haemophilia B. Haemophilia. 2014;20(1):15-24.
17. Zhou ZY, Koerper MA, Johnson KA, et al. Burden of illness: direct and indirect costs
among persons with hemophilia A in the United States. Journal of Medical Economics.
2015:1-9.
18. Schramm W, Royal S, Kroner B, et al. Clinical outcomes and resource utilization
associated with haemophilia care in Europe. Haemophilia. 2002;8(1):33-43.
19. Alexander M, Barnes C, Barnett P. Prospective audit of patients with haemophilia:
bleeding episodes and management. Journal of Paediatrics and Child Health.
2012;48(2):177-179.
20. Flood E, Pocoski J, Michaels LA, McCoy A, Beusterien K, Sasane R. Patient-reported
experience of bleeding events in haemophilia. European Journal of Haematology.
2014;93 Suppl 75:19-28.
21. Chen CX, Baker JR, Nichol MB. Burden of Illness Among Persons with Hemophilia B:
Examining the Association of Severity and Treatment Regimens with Costs and Annual
Bleed Rate. 2016.
22. Polack B, Calvez T, Chambost H, et al. EQOFIX: a combined economic and quality-of-
life study of hemophilia B treatments in France. Transfusion. 2015.
23. Choiniere M, Melzack R. Acute and chronic pain in hemophilia. Pain. 1987;31(3):317-
331.
24. Olsson A, Hellgren M, Berntorp E, Baghaei F. Association between bleeding tendency
and health-related quality of life in carriers of moderate and severe haemophilia.
Haemophilia. 2015.
25. Carlsson KS, Hojgard S, Lindgren A, et al. Costs of on-demand and prophylactic
treatment for severe haemophilia in Norway and Sweden. Haemophilia. 2004;10(5):515-
526.
26. Franchini M, Mannucci PM. Inhibitors of propagation of coagulation (factors VIII, IX
and XI): a review of current therapeutic practice. British Journal of Clinical
Pharmacology. 2011;72(4):553-562.
27. Armstrong EP, Malone DC, Krishnan S, Wessler MJ. Costs and utilization of hemophilia
A and B patients with and without inhibitors. Journal of Medical Economics.
2014;17(11):798-802.
28. Powell JS, Pasi KJ, Ragni MV, et al. Phase 3 study of recombinant factor IX Fc fusion
protein in hemophilia B. The New England Journal of Medicine. 2013;369(24):2313-
2323.
29. Mahlangu J, Powell JS, Ragni MV, et al. Phase 3 study of recombinant factor VIII Fc
fusion protein in severe hemophilia A. Blood. 2014;123(3):317-325.
30. Armstrong EP, Malone DC, Krishnan S, Wessler MJ. Adherence to clotting factors
among persons with hemophilia A or B. Hematology. 2015;20(3):148-153.
31. 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(5):493-501.
86
32. Fischer K, Van Der Bom JG, Prejs R, et al. Discontinuation of prophylactic therapy in
severe haemophilia: incidence and effects on outcome. Haemophilia. 2001;7(6):544-550.
33. Schrijvers LH, Kars MC, Beijlevelt-van der Zande M, Peters M, Schuurmans MJ, Fischer
K. Unravelling adherence to prophylaxis in haemophilia: a patients' perspective.
Haemophilia. 2015;21(5):612-621.
34. 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(5):760-765.
35. Brand B, Dunn S, Kulkarni R. Challenges in the management of haemophilia on
transition from adolescence to adulthood. European Journal of Haematology. 2015;95
Suppl 81:30-35.
36. 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(1):39-43.
37. Niu X, Poon JL, Riske B, et al. Physical activity and health outcomes in persons with
haemophilia B. Haemophilia. 2014;20(6):814-821.
38. Zhou ZY, Wu J, Baker J, et al. Haemophilia utilization group study - Part Va (HUGS
Va): design, methods and baseline data. Haemophilia. 2011;17(5):729-736.
39. Shapiro AD, Di Paola J, Cohen A, 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(2):518-525.
40. Franchini M, Frattini F, Crestani S, Sissa C, Bonfanti C. Treatment of hemophilia B:
focus on recombinant factor IX. Biologics : Targets & Therapy. 2013;7:33-38.
41. Carls GS, Roebuck MC, Brennan TA, Slezak JA, Matlin OS, Gibson TB. Impact of
medication adherence on absenteeism and short-term disability for five chronic diseases.
Journal of Occupational and Environmental Medicine. 2012;54(7):792-805.
42. Bergstralh E, Kosanke J. GMATCH SAS Macro. Mayo Clinic. 2004.
http://mayoresearch.mayo.edu/mayo/research/biostat/sasmacros.cfm Accessed Mar 2016.
43. Duncan N, Kronenberger W, Roberson C, Shapiro A. VERITAS-Pro: a new measure of
adherence to prophylactic regimens in haemophilia. Haemophilia. 2010;16(2):247-255.
44. Lam WY, Fresco P. Medication Adherence Measures: An Overview. BioMed Research
International. 2015;2015:12.
45. Melchiorre D, Linari S, Manetti M, et al. Clinical, instrumental, serological and
histological findings suggest that hemophilia B may be less severe than hemophilia A.
Haematologica. 2016;101(2):219-225.
46. Fischer K. Prophylaxis for adults with haemophilia: one size does not fit all. Blood
Transfusion. 2012;10(2):169-173.
47. Franchini M, Mannucci PM. Prophylaxis for adults with haemophilia: towards a
personalised approach? Blood Transfusion. 2012;10(2):123-124.
48. Astermark J, Petrini P, Tengborn L, Schulman S, Ljung R, Berntorp E. Primary
prophylaxis in severe haemophilia should be started at an early age but can be
individualized. British Journal of Haematology. 1999;105(4):1109-1113.
49. Baker JR, Riske B, Drake JH, et al. US Hemophilia Treatment Center population trends
1990-2010: patient diagnoses, demographics, health services utilization. Haemophilia.
2013;19(1):21-26.

87
Appendix A. Distribution of annualized adherence
Figure 3.A1. Distribution of annualized adherence by participant age

Figure 3.A2. Distribution of annualized adherence by participant hemophilia type

Abbreviations: HA - Hemophilia A, HB - Hemophilia B
88
Appendix B. Distribution of annualized bleed rate
Figure 3.B1. Distribution annualized bleed rate by participant age

Figure 3.B2. Distribution of annualized bleed rate by participant hemophilia type

Abbreviations: HA - Hemophilia A, HB - Hemophilia B
89
Appendix C. Distribution of independent variables after matching
Table 3.C1. Distribution of independent variables among matched high and low
prophylaxis adherers
Variables
Matched
(Continuous baseline bleeds)

Matched
(Binary baseline bleeds)
High
adherer
(N=16)
Low
adherer
(N=16)
P-value
b

High
adherer
(N=17)
Low
adherer
(N=17)
P-value
b

Baseline period variables:
Number of bleeds, mean (SD) 3.3 (4.1) 3.4 (4.4) 0.9432 2.3 (2.7) 4 (4.9) 0.2235
Number of bleeds ≥5 3 (18.8%) 4 (25.0%) 0.6689 2 (11.8%) 5 (29.4%) 0.492
≥1 ER visit 1 (6.3%) 2 (12.5%) 0.5442 1 (5.9%) 2 (11.8%) 0.5454
≥1 hospitalization 1 (6.3%) 1 (6.3%) 1.0000 1 (5.9%) 1 (5.9%) 1.0000
≥1 outpatient procedure 2 (12.5%) 2 (12.5%) 1.0000 1 (5.9%) 2 (11.8%) 0.5454
Number of HTC visits, mean (SD) 2.4 (3.4) 2.1 (3.2) 0.7841 2.5 (3.3) 2 (3.2) 0.6634
Days of work absenteeism,
a
mean (SD) 0.4 (1.5) 0.7 (1.8) 0.6773 0.1 (0.2) 0.7 (1.7) 0.0441
Hours of caregiver time,
b
mean (SD) 0.6 (2.2) 0.2 (0.5) 0.5365 0.5 (2.1) 0.2 (0.5) 0.5375
Hours of nurse time,
c
mean (SD) 0.3 (1) 0 (0) 0.99996 0.2 (1) 0 (0) 0.99996
Baseline participant characteristics
Adult
b
6 (37.5%) 6 (37.5%) 1.0000 6 (35.3%) 6 (35.3%) 1.0000
White 12 (75.0%) 13 (81.3%) 0.6689 13 (76.5%) 14 (82.4%) 0.6715
Married or with partner
a
7 (43.8%) 11 (68.8%) 0.1540 8 (47.1%) 12 (70.6%) 0.1634
Education >12 years
a
12 (75.0%) 12 (75.0%) 1.0000 12 (70.6%) 13 (76.5%) 0.6975
Private insurance only 7 (43.8%) 11 (68.8%) 1.0000 7 (41.2%) 12 (70.6%) 0.0842
Employment status
a
1.0000 0.9195
Not employed 5 (31.3%) 5 (31.3%) 5 (29.4%) 5 (29.4%)
Part-time 6 (37.5%) 6 (37.5%) 5 (29.4%) 6 (35.3%)
Full-time 5 (31.3%) 5 (31.3%) 7 (41.2%) 6 (35.3%)
Hemophilia A 14 (87.5%) 14 (87.5%) 1.0000 15 (88.2%) 15 (88.2%) 1.0000
Severe hemophilia 16 (100%) 16 (100%) 0.5442 17 (100%) 17 (100%) 0.5017
Years on prophylaxis, mean (SD) 5.2 (3.6) 4.7 (3.5) 0.7053 5.1 (3.5) 4.6 (3.5) 0.6037
Hours of distance to HTC, mean (SD) 1.1 (3.4) 0.9 (3.2) 0.518 1.2 (3.3) 0.9 (3.2) 0.3656
Abbreviations: SD - standard deviation, HTC - Hemophilia treatment center
All statistics reported in N(%), unless otherwise specified
a
Applies to participants ≥18 years old, or parents of participants <18 years old
b
Hours of unpaid caregiver help with household chores such as cooking, cleaning, transportation, or shopping
c
Hours of paid skilled nursing, medication administration (such as infusing their medication), wound care and/or
physical therapy

90
CHATPER 4: Cost-effectiveness Analysis of Alternative Screening and
Treatment Strategies for Heterozygous Familial Hypercholesterolemia in the
United States
*

ABSTRACT
BACKGROUND AND OBJECTIVES: Familial hypercholesterolemia (FH) is a genetic disorder
that leads to premature heart disease or stroke if untreated. Statins are effective for individuals
with FH but less than 20% of actual cases are diagnosed in the United States (US) and many
people are not adherent to treatment. Using new knowledge regarding mutations responsible for
FH, some European countries have developed genetic FH screening strategies, many of which
have been shown to be cost-effective. This study evaluates the cost-effectiveness of genetic
screening and lipid-based screening with statin adherence measures compared to lipid-based
screening alone in the US.
METHODS: A decision tree was used to estimate disease detection with the three screening
strategies, while a Markov model was used to model disease progression until death, quality-
adjusted life years (QALYs), and costs from a US societal perspective.
RESULTS: The results showed that genetic screening cost $15,594 for 18.29 QALYs per person
and lipid screening with adherence measures cost $16,385 for 18.77 QALYs compared with
$10,396 for 18.28 QALYs for lipid screening alone. The incremental cost-effectiveness ratio
(ICER) of genetic screening versus lipid screening was $519,813/QALY and that of lipid

*
Reprinted from The International Journal of Cardiology, 181, Christina X. Chen and Joel W.
Hay, Cost-effectiveness analysis of alternative screening and treatment strategies for
heterozygous familial hypercholesterolemia in the United States, 417-424, Copyright (2015),
with permission from Elsevier.

91
screening with adherence measures versus lipid screening alone was $12,223/QALY. At a US
willingness-to-pay threshold of $150,000/QALY genetic screening is not cost-effective
compared with lipid screening. Sensitivity analyses showed that results were robust to reasonable
variations in model parameters.
CONCLUSIONS: Although genetic screening is currently not a cost-effective option in the US,
health outcomes for FH individuals could benefit from adherence measures encouraging statin
use.

92
4.1. Introduction
4.1.1. Background
Heterozygous familial hypercholesterolemia (FH) is a genetic disorder that affects about
1 in 500 Caucasians in the United States (US).
1
While the exact pathways of this disease are still
unknown, recent research has focused on mutations of the low-density lipoprotein (LDL)
receptor gene (LDLR) and the gene for apolipoprotein B (APOB) as indicators for genetic FH
diagnosis.
2
In individuals with FH, mutations in the genes responsible for plasma low-density
lipoprotein cholesterol (LDL-C) clearance cause abnormal accumulation of cholesterol in the
blood and premature coronary heart disease (CHD) and stroke. According to estimates of CHD
risk from the Health Technology Assessment program in the UK, untreated Caucasian
heterozygous FH individuals are up to four times more likely to develop CHD by the age of 60
than the non-FH population.
3
Treatment and outcomes for homozygous FH are different than
those for the heterozygous form, and not the topic of this study.
1

Together, the economic and quality-of-life consequences of premature CHD present a
huge burden in the US. The American Heart Association (AHA) estimates that CHD and stroke
cost roughly $108.9 billion and $53.9 billion in the US each year, respectively, including direct
and indirect costs.
4
In addition, for up to one-third of CHD patients, their first disease symptom
is sudden cardiac death (SCD), while health-related quality-of-life for survivors decreases 30-
50% following a major event, such as an acute myocardial infarction (AMI), angina, or stroke.
5,6

Currently, the American College of Cardiology (ACC) and AHA cholesterol guidelines
encourage more widespread use of statins in potential cardiovascular disease (CVD) patients,
such as people with LDL-C levels ≥190 mg/dL, which includes FH individuals.
7
However, it is
estimated that less than 20% of actual FH cases are diagnosed in the US and, according to the

93
Centers for Disease Control and Prevention (CDC), less than 50% of adults with high cholesterol
are getting treatment.
8,9
To address low levels of diagnosis and treatment, many European
countries have established guidelines encouraging genetic testing for FH. For instance, the UK’s
National Institute for Health and Clinical Excellence (NICE) recommends using cascade
screening that incorporates genetic testing to detect new FH cases for immediate lipid-lowering
treatment. Cascade screening involves identifying index cases with a previous diagnosis of FH
and screening first, second, and, possibly, third degree relatives for FH as soon as possible.
Recently, European analyses have used both economic modeling and data from country-specific
cohorts to demonstrate the cost-effectiveness of these newer genetic cascade screening strategies
in improving treatment rates and health outcomes for the FH population.
1,10-13

However, no recent cost-effectiveness analyses (CEAs) have been conducted from a US
perspective. Further, the current cost of genetic sequencing and mutation detection tests can be
thousands of US dollars, yet the sensitivity of these tests is variable because hundreds of
biomarkers could potentially be linked to FH.
1
FH genetic testing kits validated in European
countries have not been validated in the US where the common population mutations may vary
from those identified in European patient populations. In addition to genetic screening, another
possible option to improve outcomes for FH individuals in the US is the use of statin adherence
programs, many of which have been shown to improve adherence and heart disease outcomes in
randomized trials.
14
Because the treatment and disease consequences of FH are clinically
identical to those due to high cholesterol from other reasons, such a program would not only
benefit FH individuals but also high cholesterol individuals with no FH gene mutations. By using
the information available between Europe and the US, this study is the first to attempt a complete

94
cost-effectiveness model for the US FH population, which considers the effect of statin
adherence and includes modeling of both disease diagnosis and progression.
4.2.1. Objectives
The objective of this study is to evaluate the cost-effectiveness of two FH screening and
treatment strategies not currently used in the US, compared with the lipid cascade screening
strategy currently recommended for individuals with high cholesterol and a family history of FH
or heart disease (Lipid Screening).
7,15
The two new strategies will include genetic cascade
screening of at-risk relatives from an index case (Genetic Screening) and an enhanced lipid
cascade screening strategy that includes a statin adherence program (Lipid Screening + AD).
This study will evaluate the incremental cost-effectiveness ratios (ICERs) between Lipid
Screening, Genetic Screening and Lipid Screening + AD, in 2013 US dollars per quality-adjusted
life year (QALY). The results of this study will add to the limited cost-effectiveness literature
regarding FH in the US and provide insight into where current screening and treatment pathways
can be best improved.

4.2. Methods
4.2.1. The model
The analysis was conducted with a US societal perspective and lifetime time horizon. An
initial cohort of 1000 Caucasian male adults with a family history of FH and high-risk baseline
cholesterol levels of 46 mg/dL high-density lipoprotein cholesterol (HDL-C), 224 mg/dL LDL-
C, and 305 mg/dL total cholesterol were followed in a Markov model simulation using Microsoft
Excel. Because females have different baseline health-state utilities and risk profiles for heart
disease, the model focuses on male patients. The baseline levels were adapted from the study

95
population for the UK’s Simon Broome Register of Familial Hyperlipidaemia, where genetic
testing and FH data is most available.
16
Average systolic blood pressure by age group was
obtained from a report based on the Framingham Heart Study, as the Simon Broome Register
data did not include blood pressure by age.
17
Parameters for transition probabilities, health-state
utilities, and costs were derived from peer-reviewed literature and publically available databases.
All costs and QALYs were discounted using an annual discount rate of 3%.
18

The model uses a decision tree to estimate first year screening costs and diagnosis
probabilities, and a Markov model to simulate heart disease progression and cost outcomes for
the initial cohort in each of the three screening arms. The decision tree for FH screening
differentiates between the two lipid cascade screening strategies and the genetic cascade
screening strategy. Figure 4.1 outlines the different procedures. In Genetic Screening, index
cases are individuals with a previous clinical diagnosis of FH based on cholesterol levels.
Because common US FH gene mutations have not been identified and are possibly different from
European mutations, DNA sequencing is conducted in index cases to identify familial mutations
in the LDLR or APOB genes and improve efficiency of genetic mutation detection.
3
Because of
the large number of mutations linked to FH and possible high cholesterol due to non-genetic
factors, approximately 3.4 index cases must be sequenced to reach one genetic diagnosis of FH
and identify one familial FH mutation.
1
Index cases with a genetic diagnosis are assumed to
provide an average of 2.5 relatives who are male adults, alive, at risk and agree to be screened.
1

These relatives will make up the initial cohort for the Markov model. Using the DNA mutation
identified from a respective index case, relatives will be tested for FH using a site-specific DNA
mutation detection test with an assumed sensitivity based on genetic testing strategies in the UK
of 78.5%.
1,2
Given the accessibility, affordability, and effectiveness of statins, all positive FH

96
cases from genetic screening will be prescribed statin therapy. Negative cases will be given an
LDL-C test with a sensitivity of 91% to diagnose FH cases missed in genetic testing and to
ensure that all high LDL-C cases are identified.
1
Because cholesterol levels are highly variable in
an individual at any given time and LDL-C tests do not directly measure the amount of LDL
particles, LDL-C tests cannot provide a 100% guarantee of FH diagnosis, but will help identify
high cholesterol individuals who require statin therapy. Those who test positive will be
prescribed statin therapy, while the remaining negative cases will continue to be tested with
LDL-C tests every two years, as long as they adhere to the intervention.
The approach for lipid cascade screening, with or without a statin adherence program,
does not involve extra testing of the index cases. At-risk relatives are identified from index cases
and diagnosed for FH based on LDL-C levels. Again, all positive FH cases are treated with
statins, plus an adherence program in the Lipid Screening + AD arm, while negative cases will
continue to be tested for high LDL-C until everyone from the initial cohort with dyslipidemia is
on statin therapy.
All individuals from these screening programs will enter the Markov model to simulate
their health outcomes at 1-year intervals until death. The Markov model is shown in Figure 4.2
and includes three health states: Pre-CVD, CVD Event/Stroke, and Death. All individuals start in
the Pre-CVD state and enter the CVD Event/Stroke state following a first AMI, angina, or stroke
event. While FH individuals are mainly at risk for AMI or angina, high cholesterol also increases
the risk for cardiovascular events and stroke.
1
After an event individuals will transition to the
Death state in the case of a CVD-related or non-CVD-related fatality. Individuals can also
transition directly from the Pre-CVD state to Death due to either SCD, a CVD-related fatality or
a non-CVD-related fatality. Approximately 22.7% of individuals who experience a first CVD or

97
Figure 4.1. Decision tree for familial hypercholesterolemia diagnosis

Abbreviations: FH - Familial hypercholesterolemia, LDL-C - Low-density lipoprotein cholesterol
Initial cohort is comprised of relatives from the boxes outlined in bold. Both genetic cascade screening and lipid
cascade screening options for the first year of screening are shown. Only the Lipid Screening + AD option includes
a statin adherence program. Continued treatment and diagnosis cycles are the same for remaining true (+), false (+),
true (-), and false (-) cases past the first year.
stroke event will die from SCD.
5,19
Similar models with pre-event, post-event, and death states
have been used previously to simulate cardiovascular outcomes, although not specifically for an
FH population.
20-23
Event incidences from the US were used to estimate the proportion of CHD
individuals who had an AMI, angina, or stroke event in the CVD Event/Stroke state for health-
state utility and cost calculations.
19
Different events were not modeled separately due to
insufficient evidence regarding statin efficacy for specific events in a US FH setting. The final
outcome of the Markov model is a calculation of life expectancy and discounted QALYs for
individuals from the initial cohort of each screening arm. These values are used in cost and ICER
calculations.

98
Figure 4.2. Markov model of disease progression

Abbreviations: CVD - Cardiovascular disease
Individuals begin in the Pre-CVD state, and transition to the CVD Event/Stroke state following acute myocardial
infarction-, angina-, or stroke-related hospitalization, and finally to Death. Alternatively, individuals can transition
directly to Death from Pre-CVD, or remain in a state for multiple cycles.

4.2.2. Transition probabilities
Transition probabilities between the health states in the Markov model depend on the risk
of CVD and CVD-related death calculated with the Framingham Heart Study risk equations.
24,25

The main parameters are summarized in Table 4.1 with 95% confidence intervals, when
available. Statin efficacy in all arms is based on a moderate daily dose of 10 mg atorvastatin to
minimize possibility of medication-related side effects, but adherence decreases with time on
statin treatment across arms and is increased with the statin adherence program in the Lipid
Screening + AD arm only.
26-31
Individuals diagnosed with FH or high LDL-C in the Pre-CVD
and CVD Event/Stroke state in this arm receive an annual lipid test and physician follow-up,
monthly mailed educational pamphlets regarding CVD risk and statin therapy, monthly mailed
refill reminders, and 10-minute monthly pharmacist counseling calls to discuss adherence. Here,
a fairly comprehensive adherence program is described to ensure that all reasonable cost
elements will be included in the model. The percent increase in statin adherence with a treatment
adherence program was derived from a randomized controlled trial of a comprehensive
pharmacy care program with similar components.
31
While not a long-term experiment, the study

99
used prospective, rather than retrospective, methods to provide a conservative but realistic
estimate for the effect of a medication adherence program on a similar high cholesterol patient
population taking chronic medications.
The statin efficacy and adherence parameters as well as the proportion diagnosed based
on the decision tree are used to adjust cholesterol levels from baseline and calculate risk of major
events and death.
22
The resulting cholesterol levels and Framingham Heart Study risk equations
were used to generate the 10-year probability of CVD and the 10-year probability of CVD-
related death by male age groups, in 10-year intervals.
25
These risks were converted into 1-year
health state transition probabilities for Pre-CVD to CVD Event/Stroke and Pre-CVD to Death,
respectively, using the DEALE method.
25,32

Table 4.1. Transition probability adjustment base case parameters
Parameters Effect
Hazard ratio of death after CVD event or stroke 5.0
33-36

Increase in adherence with adherence program 38%
Adherence without program 69.1% (65.4 - 72.8)
31

Adherence with program 95.5% (94.3 - 96.7)
31

Effect of Statin Therapy on Cholesterol Levels
Decrease in total Cholesterol 28.00% (25.9 - 30.1)
26,27,30

Decrease in LDL Cholesterol 38.00% (35.7 - 40.3)
26,27,30

Increase in HDL Cholesterol 5.50% (2.75 - 8.25)
26,27,30

Adherence to Statin Therapy (No adherence program)
Baseline percent adherent (1-9 years of therapy) 56% (12 - 94)
28

Baseline percent adherent (10+ years of therapy) 42% (0 - 90)
28

HR non-adherence in elderly (≥ 60 years) compared to non-
elderly (< 60 years)
0.86 (0.76 - 0.98)
29

Abbreviations: AMI - Acute myocardial infarction, LDL - Low-density lipoprotein, HDL - High-density lipoprotein,
HR - Hazard ratio
Effects presented in point estimate (95% confidence interval)
Because long-term FH cohort data from the US is not available to directly calculate the
rate of death after a CVD event or stroke, the transition probability for CVD Event/Stroke to

100
Death is estimated by multiplying the probability of CVD-related death with the hazard ratio of
death after an event, following previously published work.
20,21
The hazard ratio of death after a
CVD event or stroke is assumed to be 5.0 based on US studies that provide long-term survival
prognosis 5 to 10 years after AMI or stroke in the Framingham Heart Study population.
33-36

Since 95% confidence intervals were not available from the literature, this parameter was varied
between 4.0 and 6.0 in sensitivity analyses. Non-CVD-related death rates for different age
groups are derived from U.S. vital statistics data to modify all transition probabilities to the
Death state. Transition probabilities for each screening arm are separated by 10-year age groups
to account for changing heart disease progression and death rates.
4.2.3. Health-state utilities
Utilities for each state in the Markov model were estimated from studies using a US
societal perspective, with reference to a perfect health utility of 1.0 and death state utility of
0.0.
37,38
The societal perspective was used over patient assessment to avoid over or under-
estimating the burdens of disease. Because moderate-dose statins are well tolerated and there is
no disutility from hyperlipidemia alone, the disutility from statin therapy for FH patients is
assumed to be minimal.
30,39,40
In the Pre-CVD state, there is a 0.004 disutility to reflect the act of
taking a daily statin prescription with mild side effects, which is varied between 0.95 and 1.0 in
sensitivity analyses.
41
The disutility in the CVD Event/Stroke state is estimated from a study of
nationally representative EQ-5D index scores for chronic ICD-9 codes.
6
The health-state utility
is based on preference scores reported for AMI, angina, and stroke multiplied with the
proportions of these events occurring in the US population. Other studies have likewise assumed
a constant utility during each cycle within a post-event state, with similar utility parameters

101
following AMI or stroke.
20,23
All health-state utilities in Table 4.2 were adjusted by age group
according to US male age utility weights for use in the Markov model.
42

Table 4.2. Input health-state utilities
State Utility
Pre-CVD 0.996
41

CVD Event/Stroke 0.68 (0.65 - 0.71)
6

Death 0
Abbreviations: CVD - Cardiovascular disease
Utility presented in point estimate (95% confidence interval)
4.2.4. Costs
Costs were evaluated and reported from a US societal perspective and adjusted to 2013
US dollars using the CPI for Medical Care, when needed. The costs of genetic and lipid
screening, statin treatment, CVD and stroke management, and statin adherence measures were
included in this study (Table 4.3).
Screening costs in each treatment arm were applied in the first year to all individuals as
they enter the Markov models. The direct screening costs of Lipid Screening and Lipid
Screening + AD include two LDL-C tests, to confirm FH or high cholesterol diagnosis, plus the
administrative costs for testing and the indirect cost of work-hours lost due to physician visits.
Those who are diagnosed will also incur costs for disease counseling. In the Genetic Screening
arm, screening costs per person include the cost of FH sequencing multiplied by 3.4 index cases
required to identify one familial mutation and divided by 2.5 relatives referred from each index
case to be included in the initial cohort. In addition, the cost of a site-specific DNA mutation
detection test is included for each person in the initial cohort. Individuals incur similar
administrative, indirect, and counseling costs as those individuals from the lipid screening arms,
while FH-negative individuals would also be screened with LDL-C tests to identify FH or high

102
cholesterol cases missed by genetic screening alone. The costs for DNA sequencing and
mutation testing were obtained from the Ambry Genetics company website, and CPT codes were
used to price all other screening costs using the Medicare Physician Fee Schedule.
43,44
After the
first year, remaining FH-negative cases from all screening arms are tested for high cholesterol
every two years and receive disease counseling if diagnosed until all individuals are on statin
treatment.
The statin treatment costs for those diagnosed with FH or high cholesterol in the Pre-
CVD state of all screening strategies include the cost of medication and statin-related adverse
events. The annual cost of 10 mg generic atorvastatin per day was obtained from the 2013
Veterans Affairs Federal Supply Schedule. The main adverse events associated with statins are
generally mild, affect approximately 46% of individuals, and include localized pain, pharyngitis,
myalgia, and headaches.
30
The costs of these side effects are much less than costs for FH
screening and CVD or stroke treatment, so they were estimated as a continual annual cost for
statin users, based on prices for common over-the-counter treatments from an online pharmacy
website, drugstore.com. Because serious events are rare, costs for these events were not
included.
39,45
Further, 46% is a high estimate for the proportion of statin-users who will report
adverse events, and both the cost of statin medication and adverse events are varied in sensitivity
analyses to address the possibility of higher costs. In the Lipid Screening + AD arm, costs for the
adherence program are also included.
14

Costs in the CVD Event/Stroke state for all screening arms include first year treatment
costs for AMI, angina, and stroke hospitalizations, as well as ongoing annual costs for each event
in subsequent years. Estimates were based on a cost-utility analysis of statins and aspirin for
CHD events, which calculated first year costs for these events directly from US hospital

103
utilization data and estimated the ongoing costs in following years from published literature.
46

While estimates for CHD event costs can vary, these figures were compared with other cost
estimates in published literature and each figure was varied in sensitivity analyses to see if there
was any major effect on ICER results.
19,47,48
First year and ongoing costs for an AMI, angina, or
stroke event were multiplied by the proportions of these events occurring in the population to
obtain total costs per person. Ongoing annual costs for maintenance care include pharmacy,
inpatient, outpatient, emergency, and home health services.
47

In the Lipid Screening + AD arm, the annual cost for a statin adherence program is also
included in the CVD Event/Stroke state. Figures from Chapman et al.’s CEA of adherence-
improving interventions for cardiovascular drugs were the most appropriate estimates for statin
adherence in a high cholesterol patient population and used for costs of monthly mailed
educational pamphlets and refill reminders.
14
Using methods from Chapman et al., the cost for
each monthly 10-minute pharmacist counseling call was estimated using average compensation
rates from the Bureau of Labor Statistics.
14
Finally, CPT codes were used to price the cost for an
annual lipid test and 10-minute physician follow-up using the Medicare Physician Fee
Schedule.
44

104
Table 4.3. Input costs

Lipid
Screening
Lipid
Screening
+ AD
Genetic
Screening
Selected input costs
Comprehensive FH sequencing
43
-- -- $3,480
Site-specific DNA test for FH
43
-- -- $400
LDL-C test
44
$17 $17 $17
Annual statin medication $42 $42 $42
Annual statin adverse events
30
$186 $186 $186
First year cost of AMI
46
$23,123 $23,123 $23,123
Annual cost of AMI (ongoing)
46
$3,703 $3,703 $3,703
First year cost of angina
46
$8,139 $8,139 $8,139
Annual cost of angina (ongoing)
46
$3,536 $3,536 $3,536
First year cost of stroke
46
$16,044 $16,044 $16,044
Annual cost of stroke (ongoing)
46
$2,392 $2,392 $2,392
Annual statin adherence program
14,44
-- $238 --
Patient-time lost per physician visit
a
$124 $124 $124

Total costs (per person)
First year screening costs
b
$334 $334 $5,528
Cholesterol testing for FH (-) cases (every 2 years)
c
$233 $233 $233
Annual statin treatment costs
d
$106 $352 $106
First year CVD event/stroke costs
e
$16,551 $16,551 $16,551
Ongoing annual CVD event/stroke costs
f
$3,382 $3,382 $3,382
Death $0 $0 $0
Abbreviations: AD - Adherence program, FH - Familial hypercholesterolemia, LDL-C - Low-density lipoprotein
cholesterol, AMI - Acute myocardial infarction, CVD - Cardiovascular disease
a
Assume 4 hours total required per doctor’s visit at $31.09 average hourly compensation from the Bureau of Labor
Statistics
b
Includes costs of genetic or LDL-C testing, physician visits, and disease counseling for index cases and relatives in
the initial cohort
c
Includes costs of LDL-C testing and disease counseling for remaining FH (-) cases.
d
Includes cost of statin medication, which is adjusted for adherence in the Markov model, and cost of statin-related
adverse events, which is also adjusted for adherence as well as proportion of individuals who report symptoms; cost
of statin adherence program is included in the Lipid Screening + AD column
e
Cost calculated by multiplying 1
st
-year cost of AMI, angina, and stroke by respective proportions reported in the US
and summing
f
Costs after the first year of an event: cost calculated by multiplying annual ongoing cost of AMI, angina, and stroke
by respective proportions reported in the US and summing
4.2.5. Sensitivity analysis
Sensitivity analyses were necessary to address the lack of complete FH screening and
treatment data from a US population. One-way sensitivity analyses of the ICER for Genetic

105
Screening versus Lipid Screening and Lipid Screening + AD versus Lipid Screening were
conducted to determine model robustness. When available, parameters were varied using upper
and lower limits from the literature or 95% confidence intervals. All other parameters were
varied +/- 20%.
Probabilistic sensitivity analyses (PSA) were conducted using Monte Carlo simulations
(10,000 iterations) to obtain cost-effectiveness acceptability curves. Again, the parameters used
in one-way sensitivity analyses were varied using upper and lower limits or distributional
information provided in the literature, or +/- 20% when no literature was available. Parameters
for transition probability and health-state utility calculations were drawn from normal
distributions, with the exception of the hazard ratio of death after a CVD event or stroke and the
disutility from statin medication in the Pre-CVD state. These two parameters with varied using a
triangular distribution with ranges specified in Sections 4.2.2 and 4.2.3, respectively. Medication
and adherence program costs were assumed to have a triangular distribution with maximum and
minimum +/- 20% of the baseline value, respectively. First year and ongoing costs of AMI,
angina, and stroke were assumed to have a gamma distribution to model the skewed nature of
healthcare costs. Threshold sensitivity analyses were also used to evaluate the effect of Genetic
Screening costs and adherence to statins on results.

4.3. Results
4.3.1. Cost-effectiveness ratios
Results are presented in total costs and QALYs per individual in each screening arm, as
well as incremental costs and QALYs and ICERs between arms (Table 4.4). Genetic Screening
cost $5,198 more per person than Lipid Screening due to the high price for FH sequencing and

106
DNA mutation detection tests, but only produced 0.01 more QALYs, resulting in an ICER of
$519,813/QALY, which is not cost-effective at a US willingness-to-pay threshold of
$150,000/QALY.
49
Lipid Screening + AD compared to Lipid Screening and Genetic Screening
produced ICERs of $12,223/QALY and $1,648/QALY, respectively. Lipid Screening + AD
versus Lipid Screening has more incremental costs than Genetic Screening versus Lipid
Screening ($5,989 compared with $5,198) because of the lifetime costs for a statin adherence
program, but Lipid Screening + AD versus Lipid Screening also produces more incremental
QALYs due to the prolonged beneficial effect of increased statin adherence on heart disease- or
stroke-related events and death (0.49 QALYs compared with 0.01 QALYs). Because Genetic
Screening mainly affects the probability of diagnosis in the first few years, it does not lead to a
sustained improvement in health outcomes and QALY gain. Based on the ranking algorithm for
comparing cost-effectiveness among multiple treatments, Lipid Screening and Lipid Screening +
AD have extended dominance over Genetic Screening.
50

Table 4.4. Markov model results

Total
Costs
Total
QALYs
Incremental
Costs
Incremental
QALYs
ICER
($/QALY)
Lipid Screening (reference) $10,396 18.28 -- -- --
Lipid Screening + AD $16,385 18.77 $5,989 0.49 $12,223
Genetic Screening $15,594 18.29 $5,198 0.01 $519,813
Abbreviations: QALYs - Quality-adjusted life years, ICER - Incremental cost-effectiveness ratio, AD - Adherence
program
Lipid Screening + AD vs. Genetic Screening ICER: $1,648/QALY
4.3.2. Sensitivity analysis results
One-way sensitivity analyses showed that the main drivers for the ICER results between
Lipid Screening + AD and Lipid Screening included the discount factor, utility in the CVD
Event/Stroke state, adherence program costs, and odds ratio of adherence with the adherence

107
program, but results did not exceed $15,037/QALY (Figure 4.3a). However, only genetic
sequencing and testing costs had a large effect on the ICER results between Genetic Screening
and Lipid Screening, which remained consistently above the $150,000/QALY willingness-to-pay
threshold (Figure 4.3b). All other parameters only produced changes in this ICER of less than
1%. Overall, all model results were robust to the parameters varied.
A cost-effectiveness acceptability curve was generated for the Lipid Screening + AD
versus Lipid Screening ICER using PSA (Figure 4.4). At a willingness-to-pay threshold of
$150,000/QALY, 99% of the simulated ICERs for Lipid Screening + AD versus Lipid Screening
and only 55% of those for Genetic Screening versus Lipid Screening were cost-effective.
Using a threshold sensitivity analysis, Genetic Screening will only be cost-effective
compared with Lipid Screening at first year screening costs of less than $1,830 per person,
assuming a willingness-to-pay cost-effectiveness threshold of $150,000/QALY. This suggests
that the current first year per person screening costs of $5,528 must decrease by 67% in order for
Genetic Screening to be cost-effective compared with Lipid Screening. Threshold sensitivity
analyses for adherence parameters showed that the ICER results between Lipid Screening + AD
and Lipid Screening are robust. The ICER consistently decreased as the odds ratio of improved
statin adherence with an adherence program was varied between 1.0, meaning no improvement
in adherence, and 3.0 from a base case odds ratio of 1.38. Even with a small odds ratio of 1.05,
the ICER remains below $14,000/QALY. When the percent adherence to statins during the first
9 years of therapy and10 years of therapy or greater was varied between 0% and 100%, the ICER
increased with adherence, but never exceeded $14,476/QALY and $18,233/QALY, respectively.

108
Figure 4.3a. One-way sensitivity analysis tornado plot of Lipid Screening + AD vs. Lipid
Screening ICER

Abbreviations: ICER - Incremental cost-effectiveness ratio, QALY - Quality-adjusted life year, CVD -
Cardiovascular disease, OR - Odds ratio, HR - hazard ratio, AMI - Acute myocardial infarction, HDL-C - High-
density lipoprotein cholesterol
Figure 4.3b. One-way sensitivity analysis tornado plot of Genetic Screening vs. Lipid
Screening ICER

Abbreviations: ICER - Incremental cost-effectiveness ratio, QALY - Quality-adjusted life year, FH - Familial
hypercholesterolemia, AMI - Acute myocardial infarction, HDL-C - High-density lipoprotein cholesterol

109
Figure 4.4. Cost-effectiveness acceptability curve of Lipid Screening + AD vs. Lipid
Screening ICER

Abbreviations: ICER - Incremental cost-effectiveness ratio, QALY - Quality-adjusted life year
At a willingness-to-pay of $150,000/QALY, 99% of simulated ICERs were cost-effective
4.4. Discussion and limitations
4.4.1. Discussion
Our CEA model compared the cost-effectiveness of three different screening strategies
for diagnosis and treatment of FH in the US. Based on model parameters derived from peer-
reviewed literature and publically available reports, the results demonstrated the cost-
effectiveness of adding a statin adherence program to a lipid cascade screening strategy from a
US societal perspective. Sensitivity analyses of transition probability parameters, QALY
assumptions and cost data showed that while the ICER for Lipid Screening + AD compared with
Lipid Screening is sensitive to some model parameters, the result is robust at a US willingness-
to-pay of $150,000/QALY or less.

110
While analyses of genetic cascade screening in European countries have suggested that
such a strategy is cost-effective and should be used to diagnose FH, the results of this study show
that Genetic Screening compared with Lipid Screening produces only a very small increase in
QALYs at a cost of over $5,000 per person and is not cost-effective in the US. European CEAs
have been able to incorporate more definitive data collected from national screening programs
using both genetic testing and LDL-C level measurement. In the US, reliable figures regarding
FH disease progression, FH cascade screening disease diagnosis probabilities, and costs for FH
screening and management are rare. European models were based on FH mutation detection
sensitivities of validated genetic testing kits for well-identified population-specific mutations or
actual data from screening programs. Many of these programs are government run or subsidized,
making cost figures more available and genetic testing costs relatively inexpensive compared
with the US, where no guidelines are in place for such screening and comprehensive FH
sequencing of index cases is suggested before mutation detection. Finally, most published
European CEAs present ICER results for genetic screening compared with no population
screening, rather than the alternative lipid screening scenario, driving up the apparent
effectiveness of genetic screening.
Currently, LDL-C testing is reliable, extremely inexpensive, and able to identify
individuals with high cholesterol due to both genetic and lifestyle factors. A genetic cascade
screening approach alone, without strict guidelines for continued cholesterol testing, could
overlook non-FH patients that develop high cholesterol later and delay these at-risk individuals
from receiving early lipid-lowering treatments. Until the price of genetic testing decreases
dramatically in the US or the literature regarding FH-related mutations improves, genetic
cascade screening is not likely to be a cost-effective alternative to lipid cascade screening

111
strategies currently in use in the US. Rather, funds could be directed towards more regulated
lipid cascade screening programs and statin adherence measures to improve health outcomes for
all individuals with high cholesterol.
4.4.2. Study limitations
This study is mainly limited by the lack of US data regarding the FH patient population.
Because there are few long-term cohort studies of FH and no clinical trials regarding FH
diagnosis or treatment, the CVD and stroke risks calculated for the Markov model are
approximations and may not fully reflect real-world risks faced by relatives of FH index cases.
The limited risks available in published literature only allowed for a greatly simplified Markov
model of heart disease progression, which does not include separate states for secondary events
and treatment. Additionally, this study examines disease progression in a male population only,
and may not be directly generalizable to the entire population. It is possible that genetic cascade
screening would be more cost effective in females, who, absent FH, have a lower relative risk of
cardiovascular disease. However, this study made reasonable assumptions given what has been
published and reported to address growing concerns for the FH population.
Further, there are gaps in knowledge regarding genetic screening and FH disease
management in the US. No genetic cascade screening programs have been reported in the US, so
information regarding figures needed for such a screening strategy was estimated from European
data. It is possible that the FH populations between Europe and the US are different and that
screening and treatment programs will have different effectiveness. Due to the lack of
information on differences between clinical outcomes for patients with FH and those with non-
FH high CHD risk, it is unknown whether lipid-lowering therapies would effectively mitigate FH

112
beyond LDL-C reduction. Variability between health-state utilities and costs for patients with FH
versus those with non-FH high LDL-C could also exist, but has not been addressed in the US.

4.5. Conclusions
This study is the first to use a CEA to evaluate the appropriateness of genetic cascade
screening and a statin adherence program for the FH population in the US. The ICER results
support implementation of enhanced lipid cascade screening with statin adherence measures,
while showing that genetic cascade screening is not cost-effective at a US willingness-to-pay
threshold of $150,000/QALY or less. Using a US societal perspective and the model parameters
assumed for this study, the results suggest that better statin adherence management is more
effective in improving health outcomes for FH individuals than a genetic diagnosis approach,
which would rely on currently expensive genetic sequencing and testing with variable sensitivity
for disease diagnosis. While the ICER results were robust in sensitivity analyses, more research
regarding the US FH population and a better understanding of the disease will be required to aid
in future efforts to improve FH diagnosis and management.

113
4.6. Chapter references
1. Sharma P, Boyers D, Boachie C, et al. Elucigene FH20 and LIPOchip for the diagnosis of
familial hypercholesterolaemia: a systematic review and economic evaluation. Health
Technology Assessment. 2012;16(17):1-266.
2. Alonso R, Defesche JC, Tejedor D, et al. Genetic diagnosis of familial
hypercholesterolemia using a DNA-array based platform. Clinical Biochemistry
2009;42(9):899-903.
3. Austin MA, Hutter CM, Zimmern RL, Humphries SE. Familial hypercholesterolemia and
coronary heart disease: a HuGE association review. American Journal of Epidemiology.
2004;160(5):421-429.
4. Heidenreich PA, Trogdon JG, Khavjou OA, et al. Forecasting the future of cardiovascular
disease in the United States: a policy statement from the American Heart Association.
Circulation. 2011;123(8):933-944.
5. Kong MH, Fonarow GC, Peterson ED, et al. Systematic review of the incidence of
sudden cardiac death in the United States. Journal of the American College of
Cardiology. 2011;57(7):794-801.
6. Sullivan PW, Ghushchyan V. Preference-Based EQ-5D index scores for chronic
conditions in the United States. Medical Decision Making. 2006;26(4):410-420.
7. Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA Guideline on the
Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in
Adults: A Report of the American College of Cardiology/American Heart Association
Task Force on Practice Guidelines. Journal of the American College of Cardiology. 2013.
8. Leren TP, Finborud TH, Manshaus TE, Ose L, Berge KE. Diagnosis of familial
hypercholesterolemia in general practice using clinical diagnostic criteria or genetic
testing as part of cascade genetic screening. Community Genetics. 2008;11(1):26-35.
9. Ned RM, Sijbrands EJ. Cascade Screening for Familial Hypercholesterolemia (FH).
PLoS Currents. 2011;3:RRN1238.
10. Marang-van de Mheen PJ, ten Asbroek AH, Bonneux L, Bonsel GJ, Klazinga NS. Cost-
effectiveness of a family and DNA based screening programme on familial
hypercholesterolaemia in The Netherlands. European Heart Journal. 2002;23(24):1922-
1930.
11. Nherera L, Marks D, Minhas R, Thorogood M, Humphries SE. Probabilistic cost-
effectiveness analysis of cascade screening for familial hypercholesterolaemia using
alternative diagnostic and identification strategies. Heart. 2011;97(14):1175-1181.
12. Oliva J, Lopez-Bastida J, Moreno SG, Mata P, Alonso R. [Cost-effectiveness analysis of
a genetic screening program in the close relatives of Spanish patients with familial
hypercholesterolemia]. Revista espanola de cardiologia. 2009;62(1):57-65.
13. Wonderling D, Umans-Eckenhausen MA, Marks D, Defesche JC, Kastelein JJ,
Thorogood M. Cost-effectiveness analysis of the genetic screening program for familial
hypercholesterolemia in The Netherlands. Seminars in Vascular Medicine. 2004;4(1):97-
104.
14. Chapman RH, Ferrufino CP, Kowal SL, Classi P, Roberts CS. The cost and effectiveness
of adherence-improving interventions for antihypertensive and lipid-lowering drugs.
International Journal of Clinical Practice. 2010;64(2):169-181.

114
15. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on
Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult
Treatment Panel III) final report. Circulation. 2002;106(25):3143-3421.
16. Neil A, Cooper J, Betteridge J, et al. Reductions in all-cause, cancer, and coronary
mortality in statin-treated patients with heterozygous familial hypercholesterolaemia: a
prospective registry study. European Heart Journal. 2008;29(21):2625-2633.
17. Kronmal RA, Cain KC, Ye Z, Omenn GS. Total serum cholesterol levels and mortality
risk as a function of age. A report based on the Framingham data. Archives of Internal
Medicine. 1993;153(9):1065-1073.
18. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-effectiveness in health and
medicine. New York: Oxford University Press; 1996.
19. Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics--2013 update:
a report from the American Heart Association. Circulation. 2013;127(1):e6-e245.
20. Blake GJ, Ridker PM, Kuntz KM. Potential cost-effectiveness of C-reactive protein
screening followed by targeted statin therapy for the primary prevention of
cardiovascular disease among patients without overt hyperlipidemia. The American
Journal of Medicine. 2003;114(6):485-494.
21. Martin C, Vanderpump M, French J. Description and validation of a Markov model of
survival for individuals free of cardiovascular disease that uses Framingham risk factors.
BMC Medical Informatics and Decision Making. 2004;4:6.
22. Heart Protection Study Collaborative Group. Statin cost-effectiveness in the United
States for people at different vascular risk levels. Circulation. Cardiovascular quality and
outcomes. 2009;2(2):65-72.
23. Greving JP, Buskens E, Koffijberg H, Algra A. Cost-effectiveness of aspirin treatment in
the primary prevention of cardiovascular disease events in subgroups based on age,
gender, and varying cardiovascular risk. Circulation. 2008;117(22):2875-2883.
24. Grieve R, Hutton J, Green C. Selecting methods for the prediction of future events in
cost-effectiveness models: a decision-framework and example from the cardiovascular
field. Health Policy. 2003;64(3):311-324.
25. D'Agostino RB, Sr., Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for
use in primary care: the Framingham Heart Study. Circulation. 2008;117(6):743-753.
26. Kapur NK, Musunuru K. Clinical efficacy and safety of statins in managing
cardiovascular risk. Vascular Health and Risk Management. 2008;4(2):341-353.
27. Jones P, Kafonek S, Laurora I, Hunninghake D. Comparative dose efficacy study of
atorvastatin versus simvastatin, pravastatin, lovastatin, and fluvastatin in patients with
hypercholesterolemia (the CURVES study). The American Journal of Cardiology.
1998;81(5):582-587.
28. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term
persistence in use of statin therapy in elderly patients. JAMA : The Journal of the
American Medical Association. 2002;288(4):455-461.
29. Caspard H, Chan AK, Walker AM. Compliance with a statin treatment in a usual-care
setting: retrospective database analysis over 3 years after treatment initiation in health
maintenance organization enrollees with dyslipidemia. Clinical Therapeutics.
2005;27(10):1639-1646.

115
30. Jones PH, Davidson MH, Stein EA, et al. Comparison of the efficacy and safety of
rosuvastatin versus atorvastatin, simvastatin, and pravastatin across doses (STELLAR*
Trial). The American Journal of Cardiology. 2003;92(2):152-160.
31. Lee JK, Grace KA, Taylor AJ. Effect of a pharmacy care program on medication
adherence and persistence, blood pressure, and low-density lipoprotein cholesterol: a
randomized controlled trial. JAMA : The Journal of the American Medical Association.
2006;296(21):2563-2571.
32. Beck JR, Kassirer JP, Pauker SG. A convenient approximation of life expectancy (the
"DEALE"). I. Validation of the method. The American Journal of Medicine.
1982;73(6):883-888.
33. Kannel WB, Sorlie P, McNamara PM. Prognosis after initial myocardial infarction: the
Framingham study. The American Journal of Cardiology. 1979;44(1):53-59.
34. Weinblatt E, Shapiro S, Frank CW, Sager RV. Prognosis of men after first myocardial
infarction: mortality and first recurrence in relation to selected parameters. American
Journal of Public Health and the Nation's Health. 1968;58(8):1329-1347.
35. Sacco RL, Wolf PA, Kannel WB, McNamara PM. Survival and recurrence following
stroke. The Framingham study. Stroke. 1982;13(3):290-295.
36. Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, Warlow CP. Long-term
survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke.
1993;24(6):796-800.
37. Torrance GW. Measurement of health state utilities for economic appraisal. Journal of
Health Economics. 1986;5(1):1-30.
38. Dolan P. Modeling valuations for EuroQol health states. Medical Care.
1997;35(11):1095-1108.
39. Clearfield MB, Amerena J, Bassand JP, et al. Comparison of the efficacy and safety of
rosuvastatin 10 mg and atorvastatin 20 mg in high-risk patients with
hypercholesterolemia--Prospective study to evaluate the Use of Low doses of the Statins
Atorvastatin and Rosuvastatin (PULSAR). Trials. 2006;7:35.
40. Fryback DG, Dasbach EJ, Klein R, et al. The Beaver Dam Health Outcomes Study: initial
catalog of health-state quality factors. Medical Decision Making. 1993;13(2):89-102.
41. Coleman CI, Straznitskas AD, Sobieraj DM, Kluger J, Anglade MW. Cost-effectiveness
of clopidogrel plus aspirin for stroke prevention in patients with atrial fibrillation in
whom warfarin is unsuitable. The American Journal of Cardiology. 2012;109(7):1020-
1025.
42. Hanmer J, Lawrence WF, Anderson JP, Kaplan RM, Fryback DG. Report of nationally
representative values for the noninstitutionalized US adult population for 7 health-related
quality-of-life scores. Medical Decision Making. 2006;26(4):391-400.
43. Litterman NK, Rhee M, Swinney DC, Ekins S. Collaboration for rare disease drug
discovery research. F1000Research. 2014;3:261.
44. Physician Fee Schedule Search. http://www.cms.gov/apps/physician-fee-
schedule/search/search-results.aspx?Y=0&T=0&HT=0&CT=0&H1=99214&M=1
(accessed June 2013).
45. Huse DM, Russell MW, Miller JD, et al. Cost-effectiveness of statins. The American
Journal of Cardiology. 1998;82(11):1357-1363.

116
46. Pignone M, Earnshaw S, Tice JA, Pletcher MJ. Aspirin, statins, or both drugs for the
primary prevention of coronary heart disease events in men: a cost-utility analysis.
Annals of Internal Medicine. 2006;144(5):326-336.
47. Russell MW, Huse DM, Drowns S, Hamel EC, Hartz SC. Direct medical costs of
coronary artery disease in the United States. The American Journal of Cardiology.
1998;81(9):1110-1115.
48. Goldman L, Weinstein MC, Goldman PA, Williams LW. Cost-effectiveness of HMG-
CoA reductase inhibition for primary and secondary prevention of coronary heart disease.
JAMA : The Journal of the American Medical Association. 1991;265(9):1145-1151.
49. Macroeconomics and Health: Investing in Health for Economic Development. In: Health
CoMa, ed. Geneva: World Health Organization; 2001.
50. Glick HA, Doshi JA, Sonnad SS, Polsky D. Economic evaluation in clinical trials. New
York, NY: Oxford University Press; 2007.

117
CHAPTER 5: Conclusion
This work aimed to generate evidence critical for efforts to improve the treatment and
outcomes of individuals affected by hereditary diseases. In particular, the current studies
examined data on hemophilia and familial hypercholesterolemia (FH) to provide new insights
into the nuanced impact of treatment decisions and adherence on clinical, quality of life (QoL),
and economic outcomes among clinically different patient subgroups in two hereditary
conditions with different severity of outcomes and cost and ease of treatment. The analyses in
hemophilia confirmed that this is a costly condition, in which healthcare expenditures, treatment
patterns, and clinical outcomes can differ significantly between adults and children and between
individuals with different types of hemophilia even in limited sample sizes. Moreover, poor
adherence to prophylaxis has a significant and negative impact on clinical outcomes, which are
worse in routine clinical practice than those observed in clinical trials. Finally, economic
modeling showed that implementing a statin adherence program is cost-effective among
individuals with FH.
It is evident that the consequences of poor adherence are major issues in many common
diseases, especially chronic conditions with lifelong burdens. FH and hemophilia represent two
different types of chronic and hereditary conditions with varying prevalence, diagnosis rate,
onset of clinical symptoms, and ease and cost of pharmaceutical treatment. The findings from
analyses in both conditions highlight an unmet need to refine treatment approaches for individual
differences in diverse hereditary disease populations. More research is still needed to inform
improvements to hereditary disease management and outcomes and to guide patients, caregivers,
healthcare practitioners, and other health system influencers in efforts to optimize patient
experiences. The types of data and methods employed in the current studies may have broader

118
applications for other hereditary, chronic diseases. Future work in these diseases should expand
on the results presented here to examine the impact of variations among patient subgroups on
experiences beyond clinical outcomes in larger study samples, determine long-term treatment
adherence patterns, better understand the drivers of treatment decisions and adherence, and
estimate the budget impact of suboptimal treatment patterns in routine clinical practice.

Abstract (if available)

Abstract Hereditary diseases are generally chronic and impose a lifetime of burden of illness. The majority of these diseases are rare conditions that individually affect only a small proportion of the population. Many do not have effective treatments, and even among those with disease-specific pharmaceutical interventions, treatment regimens may not be optimal. It has been established that approximately 50% of patients with chronic conditions do not adhere to their prescribed treatment. For individuals with hereditary or chronic diseases, this eventually leads to poor clinical and quality of life (QoL) outcomes, increased morbidity, and unnecessary healthcare expenditures over the course of a lifetime. Further, while individually uncommon, together there are thousands of rare diseases that create significant burden of illness for over 350 million people globally, but thorough understanding of real-world treatment patterns and patient experiences is limited by the geographic dispersion and low prevalence of individuals with these conditions. In order to highlight unmet treatment needs among individuals with hereditary diseases and advance efforts to optimize treatments for patient experiences, this dissertation aims to better understand variations in clinical and economic outcomes, examine the impact of treatment decisions and adherence on clinical and economic outcomes, and assess the potential for cost-effective treatment optimization among patients with two different examples of hereditary diseases. The studies here report findings from hemophilia and familial hypercholesterolemia (FH), which represent two conditions with varying prevalence, diagnosis rate, severity and onset of clinical symptoms, and ease and cost of pharmaceutical treatment. ❧ The first study used prospective longitudinal data from the Hemophilia Utilization Group Studies (HUGS) to determine societal burden of illness, including direct and indirect costs and annualized bleed rate (ABR), for persons with hemophilia B in the United States (US) and to conduct subgroup analyses, which found that costs and ABR differed significantly among subgroups by hemophilia severity and treatment regimen. The second study employed the relatively large hemophilia sample sizes and rich datasets from HUGS in regression analyses to evaluate the covariate-adjusted impact of adherence to prophylaxis on ABR among persons with hemophilia A or B in the US, and to identify other socio-demographic and clinical variables significantly associated with bleeding episodes. The results showed a statistically significant impact of adherence on bleeding episodes even after controlling for many patient and disease characteristics. While this relationship between adherence and outcomes was found in all subgroups by age and hemophilia type, the impact of adherence was not significant among children. The third study used decision tree and Markov modeling to show that implementing a statin adherence program in addition to currently recommended lipid-based cascade screening for FH diagnosis and management is cost-effective compared with genetic cascade screening or lipid-based screening alone for individuals with high cholesterol and a family history of FH or heart disease in the US. ❧ Together, the studies demonstrated that hereditary diseases can be costly conditions, in which healthcare expenditures, treatment patterns, and clinical outcomes differ significantly between clinically meaningful patient subgroups. Further, adherence was a common issue across different types of hereditary diseases regardless of the severity of outcomes and cost or ease of treatment. The results also suggested that management of these conditions could be tailored to meet individual needs and clinical goals and that improving treatment adherence could be cost-effective to society. The types of data and methods employed here may by appropriate for broader research regarding treatment optimization in other chronic and hereditary conditions.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Three essays on emerging issues in hemophilia care

PDF

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

PDF

Understanding primary nonadherence to medications and its associated healthcare outcomes: a retrospective analysis of electronic medical records in an integrated healthcare setting

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

US biosimilar market entry and uptake: US commercial payer preferences in access decision making

PDF

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

PDF

The impact of Patient-Centered Medical Home on a managed Medicaid plan

PDF

Evaluating approaches to reduce inappropriate antibiotic use in the United States

PDF

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

PDF

Essays on the economics of infectious diseases

PDF

Medication compliance to statin therapy and its impact on disease outcomes in type 2 diabetes

PDF

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

PDF

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

PDF

Three essays in health economics

PDF

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

PDF

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

PDF

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

PDF

Effects of a formulary expansion on the use of atypical antipsychotics and health care services by patients with schizophrenia in the California Medicaid Program

PDF

Discriminating changes in health using patient-reported outcomes

Asset Metadata

Creator Chen, Christina Xiaoyue (author)

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

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Pharmaceutical Economics and Policy

Publication Date 07/21/2016

Defense Date 05/12/2016

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

Tag adherence,burden of illness,clinical outcome,cost-effectiveness,familial hypercholesterolemia,hemophilia,hereditary disease,OAI-PMH Harvest,Quality of life,rare disease

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Nichol, Michael B. (committee chair), Doctor, Jason N. (committee member), Joyce, Geoffrey F. (committee member)

Creator Email christinachen8@gmail.com,xiaoyuec@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-272338

Unique identifier UC11279394

Identifier etd-ChenChrist-4572.pdf (filename),usctheses-c40-272338 (legacy record id)

Legacy Identifier etd-ChenChrist-4572.pdf

Dmrecord 272338

Document Type Dissertation

Format application/pdf (imt)

Rights Chen, Christina Xiaoyue

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