University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Burden of Illness in Hemophilia A: Taking
the Patient’s Perspective

by
Jiat Ling Poon

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

December 2013

ii

To Granddad, Mama, Papa and Jiat Sing for always being there
and
In memory of Dr Kathleen Johnson

iii
Acknowledgements

I would like to extend a very special thank you to both my advisor Dr Jason Doctor, and to Dr
Mike Nichol for your mentorship, advice and encouragement. More importantly, I cannot thank
you both enough for taking me under your wings and looking out for me the past 16 months.

To the members of my dissertation committee, Dr Neeraj Sood, Dr Geoff Joyce and Dr Sue
Ingles, thank you for your support and invaluable comments.

I would also like to thank the HUGS group without which my dissertation would not be possible.
Your support and friendship over the past 3 years, especially through the bad times, was very
much appreciated.

To the professors within the USC Pharmaceutical Economics and Policy program, thank you for
challenging me and helping me prepare for the road ahead. Special thanks to Dr Jeff McCombs
for all the “have you got five minutes?” office visits.

For being there to bounce research ideas off of, for being there when I needed a hug or a cry
and for everything else in between, thank you to my big sisters Mimi, Jae Kyung and Jenny. For
always being just a text away even though we’re an ocean apart, thank you Lisa, Rada and Mila!
I love you girls!

Finally, thank you to my family for all your love, encouragement and support. I love you all so
much!
iv
Table of Contents
Acknowledgements........................................................................................................................... iii
List of Tables ..................................................................................................................................... vi
List of Figures .................................................................................................................................... vi
Abstract ............................................................................................................................................. 1
Chapter 1: Introduction ...................................................................................................................... 2
Background ............................................................................................................................................... 2
Overview of Hemophilia A ........................................................................................................................ 4
Clinical Manifestation and Management .............................................................................................. 4
Burden of Illness.................................................................................................................................... 7
Conclusion ............................................................................................................................................... 10
Figures ..................................................................................................................................................... 11
Chapter References ................................................................................................................................. 12
Chapter 2: Quality of Life in Hemophilia A: Hemophilia Utilization Group Study Va (HUGS Va) ........... 14
Abstract ................................................................................................................................................... 14
Background ............................................................................................................................................. 16
Materials and Methods ........................................................................................................................... 18
Study Design and Data Collection ....................................................................................................... 18
Eligibility Criteria ................................................................................................................................. 19
HrQoL Instruments .............................................................................................................................. 20
Statistical Analysis ............................................................................................................................... 23
Results ..................................................................................................................................................... 23
Discussion................................................................................................................................................ 27
Limitations........................................................................................................................................... 31
Conclusions ......................................................................................................................................... 32
Tables and Figures .................................................................................................................................. 33
Chapter References ................................................................................................................................. 38
Chapter 3: Longitudinal Changes in Health-Related Quality of Life for Chronic Diseases: An Example in
Hemophilia A ................................................................................................................................... 40
Abstract ................................................................................................................................................... 40
Introduction ............................................................................................................................................ 42
Methods .................................................................................................................................................. 45
Data ..................................................................................................................................................... 45
HrQoL instruments .............................................................................................................................. 46
Analysis ............................................................................................................................................... 47
Addressing missing data ..................................................................................................................... 49
Results ..................................................................................................................................................... 49
Discussion................................................................................................................................................ 51
Tables and Figures .................................................................................................................................. 55
Chapter References ................................................................................................................................. 59
Appendix ................................................................................................................................................. 60
Multivariate Multilevel Model ............................................................................................................ 60
Chapter 4: Societal Preferences for Parental Willingness-To-Pay in Children’s Hemophilia Treatment 62
Abstract ................................................................................................................................................... 62
Introduction ............................................................................................................................................ 64
Methods .................................................................................................................................................. 66
Experiment 1 ....................................................................................................................................... 67
v
Experiment 2 ....................................................................................................................................... 71
Results ..................................................................................................................................................... 72
Experiment 1 ....................................................................................................................................... 72
Experiment 2 ....................................................................................................................................... 73
Discussion................................................................................................................................................ 75
Tables and Figures .................................................................................................................................. 79
Chapter References ................................................................................................................................. 85
Chapter 5: Modelling Lost Productivity Costs: A Comparison of Treatment Types in Severe Hemophilia
A ..................................................................................................................................................... 86
Abstract ................................................................................................................................................... 86
Introduction ............................................................................................................................................ 88
Methods .................................................................................................................................................. 89
Influence Diagrams ............................................................................................................................. 89
Hemophilia Model .............................................................................................................................. 90
Hemophilia Data ................................................................................................................................. 92
Model Inputs ....................................................................................................................................... 93
Asthma Data ........................................................................................................................................ 94
Results ..................................................................................................................................................... 95
Discussion................................................................................................................................................ 97
Tables and Figures ................................................................................................................................ 101
Chapter References ............................................................................................................................... 105
Chapter 6: Conclusion .................................................................................................................... 106

vi
List of Tables
Table 2.1: Baseline characteristics of study population ............................................................................. 33
Table 2.2: Mean scores for quality of life and self-reported joint pain ...................................................... 34
Table 2.3: Mean scores for quality of life and self-reported motion limitation ......................................... 35
Table 3.1: Baseline Demographics .............................................................................................................. 55
Table 3.2: Study population grand means over time .................................................................................. 55
Table 3.3: Adults - Results from multivariate multilevel model (N=136) ................................................... 56
Table 3.4: Children - Results from multivariate multilevel model (N=125) ................................................ 57
Table 4.1: Sociodemographic characteristics ............................................................................................. 79
Table 4.2: Experiment 1 – Estimates from conditional logit model ............................................................ 79
Table 4.3: Experiment 1 – Willingness-to-pay estimates ............................................................................ 80
Table 4.4: Experiment 2 – Logistic regression of willingness-to-pay $0 against any additional amount ... 80
Table 4.5: Experiment 2 - Multinomial logistic regression of median willingness-to-pay against
willingness-to-pay below and above median amount ................................................................................ 81
Table 5.1: Characteristics of model population by disease group ............................................................ 102
Table 5.2: Results from hemophilia models ............................................................................................. 103
Table 5.3: Results from asthma model ..................................................................................................... 104

List of Figures
Figure 1.1: Simplified Coagulation Cascade ................................................................................................ 11
Figure 2.1: Mean percent clinically measured joint ROM limitation by severity of self-reported joint pain
in adults (N=91) and children (N=55) .......................................................................................................... 36
Figure 2.2: Mean percent clinically measured joint ROM limitation by severity of self-reported motion
limitation in adults (N=91) and children (N=55) ......................................................................................... 37
Figure 3.1: Adults – Study population mean (SD) follow-up HrQoL scores (N=136) .................................. 58
Figure 3.2: Children – Study population mean (SD) follow-up HrQoL scores (N=125) ............................... 58
Figure 4.1: Experiment 1 - Introduction to task .......................................................................................... 82
Figure 4.2: Experiment 1 – Example of discrete choice experiment question ........................................... 82
Figure 4.3: Experiment 2 – Survey question ............................................................................................... 83
Figure 4.4: Experiment 1 – Relative preferences for features of treatment .............................................. 83
Figure 4.5: Experiment 2 – Distribution and cumulative distribution of willingness-to-pay responses ..... 84
Figure 5.1: Illustration of Influence Diagram ............................................................................................ 101
Figure 5.2: Study Model ............................................................................................................................ 101

1
Abstract
“Burden of illness” is the negative impacts that an illness has on patients and their families, and
may include physical and psychological functioning, social stigma and economic burden. Burden
of illness is a constant concern for patients and their families, especially if the disease is rare
and also of a lifelong, chronic nature. This is despite the fact that, for many rare and chronic
diseases, affected individuals are able to lead relatively normal, productive lives if diagnosed
early and provided with appropriate disease management.

In order to help stakeholders identify and address the needs of patients with one such rare and
chronic disease, hemophilia A, this dissertation examines the burden of illness from the
patients’ perspective, in terms of its impact on patients’ health-related quality of life (HrQoL)
and economic burden. In the first study (chapter 2), HrQoL of patients with hemophilia A and
how it relates to the physical manifestations of the disease, such as joint pain and motion
limitation is described. The second study (chapter 3) examines the causes of fluctuations in
HrQoL over time, both within and between individuals with a stable, chronic disease. The third
study (chapter 4) describes a discrete choice experiment conducted to determine societal
preferences and willingness-to-pay for providing hemophilia treatment for pediatric patients in
order that they may have improved outcomes. Finally, in the fourth study (chapter 5), a
decision model is developed, comparing two treatment types to minimize the productivity
losses for caregivers of pediatric patients due to missing work or school because of their chronic
disease, using severe hemophilia A as an example. The methods employed here may also have
broader applications in research for other chronic, rare diseases.
2
Chapter 1: Introduction

Background
Rare diseases are a group of conditions with a very low prevalence in the population. In the
United States (US), they are defined as being “a disease or condition that affects fewer than
200,000 people in the United States” [1]. The European Union defines rare diseases as those
that “affect no more than 5 per 10,000 persons in Europe” [2]. Current estimates place the
number of known rare diseases to be between 5,000 and 8,000, the majority of which (80% or
more) are genetic in origin. Though individually rare, this group of diseases collectively affects
the health of millions of people globally, creating substantial burden of illness for both patients
and their families.

“Burden of illness” is a term that encompasses the negative impacts that an illness has on
patients and their families. It includes not just the physical and psychological aspects brought
about by the symptoms of the disease, but also social stigma that patients may experience and
economic burden, for both patients and their families. Economic burden includes direct
healthcare costs especially that not covered by health insurance, and just as importantly, the
economic impact (indirect costs) on patients and their families due to lost productivity, lost
wages and difficulty finding appropriate employment or remaining employed [1, 3]. Burden of
illness is a constant concern for patients and their families and this is especially so if the disease
is also of a lifelong, chronic nature. This is despite the fact that, for many rare and chronic
3
diseases, affected individuals are able to lead relatively normal, productive lives if diagnosed
early and provided with appropriate disease management.

This dissertation will focus on examining the burden of illness from the patients’ perspective, in
terms of economic burden as well as the physical and psychological impacts on patients in one
such example of a rare and chronic disease, hemophilia A. In chapter 2, we describe the health-
related quality of life of patients with hemophilia A and how this relates to the physical
manifestations of the disease, such as joint pain and motion limitation. In chapter 3, we
examine if health-related quality of life within and between individuals with a stable, chronic
disease fluctuates over time and if these fluctuations correspond with the occurrence of acute
events. In chapter 4, a discrete choice experiment is conducted to determine societal
preferences and willingness-to-pay for providing hemophilia treatment for pediatric patients in
order that they may have improved outcomes. Finally in chapter 5, we develop a decision
model comparing two treatment types to minimize the productivity losses for pediatric patients
and caregivers due to missing work or school in chronic diseases because of symptoms or
complications brought about by the disease, using severe hemophilia A as an example.

This dissertation will provide insight into the needs of and burden of disease on individuals with
hemophilia A. It will help stakeholders involved in hemophilia care, such as providers, decision-
and policy-makers identify and address the needs of people with hemophilia. At the same time,
the methods employed may have broader applications in research for other chronic, rare
diseases.
4
Overview of Hemophilia A
Clinical Manifestation and Management
Hemophilia is a group of congenital bleeding disorders caused by a lack of blood clotting
factors, that occurs with a frequency of approximately one in 10,000 births. This results in
persons with hemophilia having an impaired blood clotting process. Genetically, hemophilia is
caused by the transmission of a recessive, mutant clotting factor gene on the X-chromosome
and as a result, most commonly occurs in males, although it may also affect females. The most
common form of hemophilia is hemophilia A, which affects about 80-85% of the hemophilia
population and occurs in about one in every 5,000 male births [4] .

When an injury to a blood vessel occurs and the blood vessel wall is disrupted, it stimulates a
hemostatic response at the site of the injury. Hemostasis is the process of blood clot formation.
The central feature of this process is the coagulation cascade, shown in Figure 1.1 [5], which is
carefully regulated to initially stop bleeding through formation of a clot, followed eventually by
the removal of the clot (lysis) and tissue remodeling. Hemophilia A is caused by a deficiency in
coagulation factor VIII, which has a role in the activation of factor X, ultimately leading to the
formation of the blood clot. A person with hemophilia A has reduced levels of factor VIII in their
plasma, resulting in delayed clot formation and prolonged bleeding. The normal plasma level of
factor VIII in a person is defined as 100% (100U/dL, units per deciliter, of plasma), with the
normal range being between 40-180%. Persons with hemophilia A are classified as having mild
(5-40%), moderate (1-5%) or severe (<1%) hemophilia, based on their plasma factor VIII levels.
5
Diagnosis of hemophilia A is performed by measuring the level of factor VIII activity in the blood
[4].

The major clinical manifestation of hemophilia A is the occurrence of bleeding episodes and its
sequelae, both of which require lifelong management. Bleeding episodes may only present
themselves after the age of one in children, after they have become more mobile. Persons with
hemophilia may experience prolonged bleeding after sustaining an injury or after dental and
surgical procedures. In more severe cases, they may also experience spontaneous bleeding.
Bleeding in hemophilia most commonly occurs internally, into the muscles and joints, especially
into the knees, ankles and elbows and to a lesser extent into the hips, shoulders and larger
muscles. Recurrent bleeds into the joints may eventually lead to chronic hemophilic
arthropathy, resulting in both acute and chronic joint pain and a reduction in joint range of
motion [6]. Hemophilia arthropathy is a major cause of morbidity, affecting everyday function
and contributing to disability. Although relatively rare, persons with hemophilia can also be at
risk for spontaneous, life-threatening bleeds into the central nervous system, gastrointestinal
tract or neck and throat.

The extent and severity of bleeding in hemophilia A is generally correlated with plasma factor
VIII levels and hence disease severity. Persons with severe hemophilia tend to bleed frequently
with minimal, unrecognized trauma (spontaneous bleeding), predominantly into the joints and
muscles. Moderate hemophilia patients may experience occasional spontaneous bleeding and
joint and muscle bleeds with trauma. Mild hemophilia patients experience severe bleeding with
6
major trauma. All hemophilia patients, regardless of disease severity, may have excessive or
prolonged bleeding with surgery, including dental procedures.

The principal goal of hemophilia management is the prevention of bleeding, in order to prevent
future complications involving joint and tissue damage. Management focuses on timely and
adequate administration of factor concentrates (factor replacement therapy) to minimize the
effects of bleeding. In hemophilia A, the factor concentrate that is administered is factor VIII.
For severe hemophilia patients, it has, since the 1950s, become increasingly common practice
to manage hemophilia using prophylactic treatment, particularly in Europe [7]. Prophylactic
treatment is the regular infusion of factor concentrates in order to prevent bleeding episodes
and their consequences [8]. In contrast, on-demand treatment is the practice of infusing factor
episodically to treat an on-going bleed. In Europe, the trend is to initiate prophylaxis at an early
age in children with hemophilia, with several studies showing the benefits of doing so, before
frequent bleeds and subsequent joint deterioration occurs [7, 9, 10]. In older patients, the
outcomes of patients in whom prophylaxis was initiated at a later age were found to be inferior
to those given early prophylaxis because damage to already affected joints at the onset of
prophylaxis progressed irrespective of further prophylactic treatment [7]. Despite this, regular
prophylactic therapy is still found to reduce the rate of joint deterioration when compared to
on-demand therapy in adults and children [11]. In both mild and moderate hemophilia patients,
the treatment of choice is on-demand factor replacement therapy to manage acute bleeding
episodes.

7
Burden of Illness
As a rare but chronic disease, the management of hemophilia A is associated with high costs,
whether for patients, the healthcare system or society [12]. These costs may be both direct and
indirect.

The primary therapeutic management of hemophilia A, factor replacement therapy, incurs high
costs, whether administered prophylactically or on-demand, due to the high costs of factor
concentrates. Use of factor thus comprises the greatest proportion of not just medication, but
also overall direct medical costs of hemophilia. Direct medical costs include the cost of visits to
treatment centers, hospitalization and emergency room visits, laboratory tests and medication
costs for both factor concentrate and non-factor medications. A previous study on the cost of
hemophilia found that 20 to 30 times the resources available for the health care of an average
person in developed countries are utilized for the average hemophilia patient. This far exceeds
the estimates of 2 to 3 times the resources required. In developing countries, this relative
proportion of spending is even greater [13]. Clotting factor costs alone, as previously
mentioned, make up the bulk of hemophilia A management costs. Globe et al. [14] estimated in
1995 that the annual medical care cost of hemophilia A was US$139,102 per person, with
clotting factor consumption accounting for between 45% (for mild hemophilia) to 83% (for
severe hemophilia) of total costs. In 2012, Guh et al. [15, 16] estimated total cost per
hemophilia A patient across all disease severities to be US$148,215 and US$162,054 in
Medicaid beneficiaries and patients with employer-sponsored insurance respectively.
Combining hemophilia A and B patients, the costs were US$124,700 for Medicaid beneficiaries
8
and US$144,306 for those with employer-sponsored insurance, with factor concentrates costs
accounting for 86% and 85% of total costs respectively.

In addition, when considering the burden of illness of hemophilia A, indirect costs also play a
substantial role. Such costs arise primarily from losses in productivity such as
underemployment, absenteeism from work or school, as well as unpaid caregiver costs. Steen-
Carlsson et al. [17] in comparing prophylaxis and on-demand factor replacement therapy in
adults found that cost of productivity loss in prophylaxis was less than half of that in on-
demand therapy. Similarly, Johnson et al. [18] also observed similar results in a hemophilia A
population in the US, estimating indirect costs to be US$16,952 and US$8,867 annually for
severe hemophilia patients receiving on-demand and prophylaxis therapy respectively. Mild
and moderate hemophilia A patients sustained indirect costs of US$5,195 and US$9,043
annually, respectively. Although these costs are but a small fraction of total direct and indirect
costs due to hemophilia A, due to the high costs of factor concentrate, it nevertheless has a
substantial impact on patients and/or their families, especially when we consider the burden of
illness of the disease from the patient’s perspective.

Somewhat less quantifiable than the cost of illness is the impact of the disease on patients
physically and psychosocially. This includes impairments in health-related quality of life
(HrQoL), the consequences of pain and arthropathy, as well as lifestyle restrictions. Several
studies in Europe, Canada and the United States have evaluated HrQoL in specific
subpopulations of persons with hemophilia A. Persons with hemophilia generally suffer greater
9
impairments to the physical aspects of HrQoL than to the mental aspects. Lindvall et al. [19]
found in Sweden that persons with hemophilia aged 35 to 64 had significantly poorer physical
functioning than the general Swedish male population, although their mental functioning were
comparable. This impairment of physical functioning was greater for those with severe
hemophilia than those with mild or moderate hemophilia. Klamroth et al. [20] also found
hemophilia A to negatively impact physical, but not mental HrQoL when compared to European
and US population norms. Additionally, they also showed the physical burden of hemophilia A
to be greater than that of chronic back pain, diabetes and rheumatoid arthritis, but with less
mental burden. The presence of target joints was also negatively associated with HrQoL. Molho
et al. [21] also reported similar findings from France, observing that severe hemophilia patients
perceived poorer physical HrQoL than mental and those requiring orthopedic treatment
reported significantly poorer HrQoL than those who did not. Even among mild hemophilia
patients, the development of hemophilic arthropathy after repeated bleeding into the joints
had a negative association with physical HrQoL [22].

Concerns about bleeding episodes and the resultant pain and arthropathy have previously led
practitioners to advise against physical activities for hemophilia patients [23], leading to
lifestyle restrictions. Although it has now been found that regular exercise may be beneficial for
reducing the frequency of joint bleeds, reducing arthropathy-related pain and increasing joint
range of motion for hemophilia patients [24], practitioners while encouraging regular physical
activity, have nevertheless also discouraged participation in vigorous contact sports [25]. This
inability to participate in contact (or team) sports has led to considerable disappointment
10
among patients and may also have an impact on their self-esteem and social interactions,
affecting their psychosocial functioning [26].

Conclusion
Chronic diseases, especially those which are also rare, are potentially of great burden for
patients, their families and society. Using one such disease, hemophilia A, as an example, this
dissertation will examine burden of illness, from the standpoint of physical and psychological as
well as indirect economic burden, on patients. The methods employed may then be applied to
other rare and/or chronic diseases which collectively affect millions of patients globally, to
provide insight into the burden of illness on those patients.

11
Intrinsic Pathway Extrinsic Pathway
Common Pathway
Vascular surface

XII XII

X

XI

IX IX

VIII
X
VII VII

Tissue
thromboplastin
Prothrombin
Thrombin
Xa + V
Fibrinogen
Fibrin
monomer
Fibrin
polymer
XII

XIII

Stable fibrin
(Clot formation)
Figures

Figure 1.1: Simplified Coagulation Cascade

[5] Adapted from Vieira RL. Rash—Petechiae and Purpura. In: Fleisher GR, Ludwig S, et al., eds. Textbook of pediatric emergency
medicine. 6th ed. Wolters Kluwer/Lippincott Williams & Wilkins Health, 2010.

12
Chapter References
1. IOM (Institute of Medicine). Rare Diseases and Orphan Products: Accelerating Research and
Development. Washington, DC: The National Academies Press, 2010.
2. Aymé S, Rodwell C. 2012 Report on the State of the Art of Rare Disease Activities in Europe of
the European Union Committee of Experts on Rare Diseases. 2012.
3. McGuire T, Wells KB, Bruce ML, et al. Burden of illness. Ment Health Serv Res 2002; 4: 179-85.
4. Srivastava A, Brewer AK, Mauser-Bunschoten EP, et al. Guidelines for the management of
hemophilia. Haemophilia 2013; 19: e1-47.
5. Fleisher GR, Ludwig S. Textbook of pediatric emergency medicine. 6th edn. Philadelphia: Wolters
Kluwer/Lippincott Williams & Wilkins Health, 2010.
6. Choiniere M, Melzack R. Acute and chronic pain in hemophilia. Pain 1987; 31: 317-31.
7. Nilsson IM, Berntorp E, Lofqvist T, Pettersson H. Twenty-five years' experience of prophylactic
treatment in severe haemophilia A and B. J Intern Med 1992; 232: 25-32.
8. Ljung R. Prophylactic therapy in haemophilia. Blood Rev 2009; 23: 267-74.
9. 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. Br J Haematol 1999; 105:
1109-13.
10. Lofqvist T, Nilsson IM, Berntorp E, Pettersson H. Haemophilia prophylaxis in young patients--a
long-term follow-up. J Intern Med 1997; 241: 395-400.
11. Aledort LM, Haschmeyer RH, Pettersson H. A longitudinal study of orthopaedic outcomes for
severe factor-VIII-deficient haemophiliacs. The Orthopaedic Outcome Study Group. J Intern Med 1994;
236: 391-9.
12. Escobar MA. Health economics in haemophilia: a review from the clinician's perspective.
Haemophilia 2010; 16 Suppl 3: 29-34.
13. Schramm W, Berger K. Economics of prophylactic treatment. Haemophilia 2003; 9 Suppl 1: 111-
5; dicussion 6.
14. 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.
15. 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: 276-83.
16. 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: 268-
75.
17. Steen Carlsson K, Hojgard S, Glomstein A, et al. On-demand vs. prophylactic treatment for
severe haemophilia in Norway and Sweden: differences in treatment characteristics and outcome.
Haemophilia 2003; 9: 555-66.
18. The 2nd National Conference on Blood Disorders in Public Health. Burden of Illness: Direct and
Indirect Costs Among Persons with Hemophilia A. Atlanta, GA, USA.
19. Lindvall K, Von Mackensen S, Berntorp E. Quality of life in adult patients with haemophilia - a
single centre experience from Sweden. Haemophilia 2012; 18: 527-31.
20. Klamroth R, Pollmann H, Hermans C, et al. The relative burden of haemophilia A and the impact
of target joint development on health-related quality of life: results from the ADVATE Post-Authorization
Safety Surveillance (PASS) study. Haemophilia 2011; 17: 412-21.
21. Molho P, Rolland N, Lebrun T, et al. Epidemiological survey of the orthopaedic status of severe
haemophilia A and B patients in France. The French Study Group. . Haemophilia 2000; 6: 23-32.
13
22. Walsh M, Macgregor D, Stuckless S, Barrett B, Kawaja M, Scully MF. Health-related quality of life
in a cohort of adult patients with mild hemophilia A. J Thromb Haemost 2008; 6: 755-61.
23. Weigel N, Carlson BR. Physical activity and the hemophiliac: yes or no? American corrective
therapy journal 1975; 29: 197-205.
24. Buzzard BM. Sports and hemophilia: antagonist or protagonist. Clin Orthop Relat Res 1996: 25-
30.
25. Ross C, Goldenberg NA, Hund D, Manco-Johnson MJ. Athletic participation in severe hemophilia:
bleeding and joint outcomes in children on prophylaxis. Pediatrics 2009; 124: 1267-72.
26. Von Mackensen S. Quality of life and sports activities in patients with haemophilia. Haemophilia
2007; 13 Suppl 2: 38-43.

14
Chapter 2: Quality of Life in Hemophilia A: Hemophilia Utilization
Group Study Va (HUGS Va)

Published as:
Quality of Life in Hemophilia A: Hemophilia Utilization Group Study Va (HUGS Va). Poon JL, Zhou
ZY, Doctor JN, Wu J, Ullman MM, Ross C, Riske B, Parish KL, Lou M, Koerper MA, Gwadry-Sridhar
F, Forsberg AD, Curtis RG, Johnson KA. Haemophilia. 2012 Sep;18(5):699-707

Abstract
Aims: This study describes health-related quality of life (HrQoL) of persons with hemophilia A in
the United States (US) and determines associations between self-reported joint pain, motion
limitation, and clinically evaluated joint range of motion (ROM), and between HrQoL and ROM.
Methods: As part of a two-year cohort study, we collected baseline HrQoL using the SF-12
(adults) and PedsQL (children), along with self-ratings of joint pain and motion limitation, in
persons with factor VIII deficiency recruited from six Hemophilia Treatment Centers (HTCs) in
geographically diverse regions of the US. Clinically measured joint ROM measurements were
collected from medical charts of a subset of participants. Results: Adults (N=156, mean age:
33.5±12.6 years) had mean physical and mental component scores of 43.4±10.7 and 50.9±10.1,
respectively. Children (N=164, mean age: 9.7±4.5 years) had mean total PedsQL, physical
functioning, and psychosocial health scores of 85.9±13.8, 89.5±15.2, and 84.1±15.3,
respectively. Persons with more severe hemophilia and higher self-reported joint pain and
motion limitation had poorer scores, particularly in the physical aspects of HrQoL. In adults,
significant correlations (p<0.01) were found between ROM measures and both self-reported
measures. Conclusion: Except among those with severe disease, children and adults with
hemophilia have HrQoL scores comparable to those of the healthy US population. The physical
15
aspects of HrQoL in both adults and children with hemophilia A in the US decrease with
increasing severity of illness. However, scores for mental aspects of HrQoL do not differ
between severity groups. These findings are comparable to those from studies in European and
Canadian hemophilia populations.

16
Background
Hemophilia is a rare, chronic, inherited disease primarily affecting males, characterized by a
deficiency in a specific clotting factor, resulting in the inability of the blood to clot normally [1].
Hemophilia A is characterized by a deficiency of factor VIII, with an incidence in the United
States (US) of 1 in every 5,000 male births [1]. Major clinical manifestations of hemophilia are
bleeding episodes, most frequently into knee, ankle and elbow joints. Repeated bleeding may
eventually lead to chronic hemophilic arthropathy, resulting in both acute and chronic joint
pain and a reduction in joint range of motion [2]. Bleeding-related arthropathy is a major cause
of hemophilic morbidity, affecting everyday function and contributing to disability. These
clinical manifestations negatively impact not only the physical functioning of persons with
hemophilia, but may also affect their mental and social health and functioning, leading to an
impaired health-related quality of life (HrQoL).

Most studies of the relationship between hemophilia and HrQoL have used generic HrQoL
measures. These include the Medical Outcomes Study Short Form-36 (SF-36) [3-9] and Short
Form-12 Health Survey (SF-12) [10, 11] for adults, PedsQL for children [9, 12], and the Health
Utilities Index (HUI) [13] for both adults and children. More recently, several hemophilia-
specific HrQoL instruments have been developed and validated. These include Hemofilia-QoL
[14] and Hemolatin-QoL [15] for adults and Haemo-QoL [16] and the Canadian Hemophilia
Outcomes – Kids Life Assessment Tool (CHO-KLAT) for children [12]. Generic HrQoL instruments
are used to assess general health across different health problems and summarize a spectrum
of domains of health or quality of life applicable to different impairments, illnesses, patients
17
and populations [9], providing a common metric with which to compare HrQoL between
diseases. Disease-specific instruments are designed to identify constructs that are unique and
important in the disease state or patient population for which they were intended, such as
changes due to therapy, side effects, and other disease-specific complications.

Previous studies examining the impact of hemophilia on patients’ HrQoL and its relationship
with clinically evaluated disease manifestations were conducted mostly in Europe and Canada.
Most of these studies had small sample sizes and generally focused on specific subpopulations
such as adults only [3, 8], adults with specific hemophilia severity [4, 6, 7], hemophilia patients
who required a total knee arthroplasty [10], hemophilia patients with inhibitors [11], or
patients who were part of validation studies for hemophilia-specific instruments [9]. In these
studies, individuals with hemophilia in the various subpopulations studied reported poorer
HrQoL than the general population in their respective study countries, particularly in the
physical aspects, although they were comparable in the mental aspects [3-6, 8, 10, 11].
However, because these studies examined specific hemophilia subpopulations, it is difficult to
fully characterize the HrQoL of the hemophilia A population. Additionally, differences in
treatment practices, payment patterns, and individual demographic characteristics among
geographic regions may all result in variation in HrQoL between patients, making results from
Europe or Canada less generalizable to a US population.

Previously published US studies have compared the HrQoL of individuals in association with
hemophilia treatment adherence [17] or evaluated the HrQoL of children with hemophilia as
18
part of the development and validation of a hemophilia-specific HrQoL instrument for children
[18]. Like the European and Canadian studies, these studies had small sample sizes, pooled
hemophilia A and B participants, or were conducted in specific subpopulations, posing a
challenge for those wishing to understand how having hemophilia A affects HrQoL in general.

To fully characterize the HrQoL of US individuals with hemophilia A, this study included both
adults and children with all hemophilic disease severities. It aims to understand how the US
adult and pediatric hemophilia A populations compare to the general healthy US population or
to other chronic disease populations and to learn whether there are similarities to the results of
the European and Canadian studies previously reported in the literature.

The objectives of this study are to describe the health status of individuals with factor VIII
deficiency in the US and to describe its association with physical manifestations of the condition
such as joint pain and motion limitation.

Materials and Methods
Study Design and Data Collection
The Hemophilia Utilization Group Study Part-Va (HUGS Va), was a two-year, multicenter
observational cohort study conducted among six Hemophilia Treatment Centers (HTCs) located
in geographically diverse regions of the US. The study methods and baseline characteristics of
the study population have been previously reported [19]. Data were collected at initial
interview from participant self-report, health care providers, and patient chart reviews.
19
Information regarding clinical aspects of the disease, such as treatment regimen, arthropathy
and comorbidities were collected, as well as information regarding HrQoL and the economic
consequences of having hemophilia.

Study participants were enrolled at each HTC in accordance with study enrolment criteria
following informed consent from adult patients and from parents of minor children. A total of
329 participants (164 adults and 165 children) with factor VIII deficiency were recruited into the
HUGS Va study between July 2005 and July 2007.

The University of Southern California (USC) served as the data and coordinating center, and the
study protocol was approved by the Institutional Review Board of USC (IRB number: HS-
046012) and that of each participating HTC.

Eligibility Criteria
The inclusion criteria for participation in HUGS Va included: (1) ages 2-64 years; (2) factor VIII
level ≤ 30% with or without a history of inhibitor; (3) received at least 90% of hemophilia care at
the participating HTC; (4) obtained care at the HTC within two years prior to study enrolment
and (5) English speaking. Individuals determined to be cognitively impaired or having an
additional bleeding disorder were excluded from participation.

20
HrQoL Instruments
Adult general health was assessed using the SF-12 Health Survey Version 1, which has been
used in previous hemophilia studies [10, 11]. HrQoL of participants younger than 18 years was
assessed using the PedsQL™ 4.0 generic core scales. The PedsQL has previously been used in a
hemophilia population in Canada to validate the hemophilia-specific CHO-KLAT [9, 12]. Generic
HrQoL instruments were used instead of a hemophilia-specific tool in order to provide a basis
for comparison with other disease populations. As validated hemophilia-specific HrQoL
instruments were not available at the time of data collection, we were unable to administer
both generic and disease-specific instruments in tandem. Differences in HrQoL may not always
be statistically significant but may nevertheless be clinically meaningful [20]. Therefore, we
refer to the Minimally Clinically Important Difference (MCID), which is defined as: “the smallest
difference in score in the domain of interest which patients perceive as beneficial and which
would mandate, in the absence of troublesome side effects and excessive cost, a change in the
patient’s management.” [20]. In the absence of a measured MCID, clinical significance can be
considered to be half the standard deviation of the mean normalized scores [21].

SF-12 Health Survey Version 1
The SF-12 is an abbreviated, 12-item version of the widely-used SF-36 generic questionnaire
derived from the Medical Outcomes Study and is designed to reduce respondent burden while
accurately reproducing the scores of the SF-36 [22]. The instrument assesses eight specific
dimensions of HrQoL: physical functioning, role-physical, bodily pain, general health, vitality,
social functioning, role-emotional, and mental health, to yield two summary scores, physical
21
component score (PCS-12) and mental component score (MCS-12). Both summary scores were
calculated using the 1998 US scoring algorithm, which is norm-based and standardized to the
1998 US general population to have a mean score of 50 and standard deviation of 10. Scores
above or below 50 can be interpreted as being better than or worse than the general US
population norm, respectively. The SF-12 has been shown to be able to discriminate well
between groups of patients who differ in severity of physical and mental health [22], making it
appropriate for use in identifying discrete impairments in HrQoL within our study population. In
this context, MCID is considered to be a difference in score of 5.0 points.

PedsQL 4.0 Generic Core Scales
The PedsQL is a generic, non-disease specific HrQoL instrument developed in the US for
children and adolescents [23]. It has been shown to be valid and reliable for both self and
parent-proxy administration [24] and has also been found to be sensitive to differences
between disease types and between severity groups within the same disease [25]. Parents of
participants aged 2-7 years and participants aged 8-17 years (or their parents) self-administered
the questionnaire in our study. The PedsQL consists of 23 items that assess four subscales of
functioning: physical, emotional, social, and school-related. These contribute to a total score as
well as to two summary scores: physical health and psychosocial health. The physical health
summary score is comprised of the physical functioning subscale, while the psychosocial
summary scale is comprised of the emotional, social, and school functioning subscales. Scores
are reported on a scale of 0 to 100, with a higher score indicating better HrQoL. The total,
physical health and psychosocial health scores for a healthy pediatric population have been
22
determined to be 83.8±12.7, 87.5±13.5 and 81.9±14.1, respectively [25]. The assessment of
HrQoL in children should ideally be through self-report. However, in situations where this is not
possible due to the child being too young or ill, parent-proxy reports are generally used. Across
the PedsQL scales, moderate to good inter-rater agreement has been shown between patients
and proxies [24], allowing for parent-proxy scores to be used. The MCID of the PedsQL has been
determined to be 4.4 for child self-report and 4.5 for parent proxy-report[26]. In this study,
because the majority of pediatric responses were obtained by parent-proxy report (95.7%), an
MCID of 4.5 is used.

Joint Pain and Motion Limitation
Joint pain was determined by self-report using a 5-point scale, ranging from “1: No pain” to “5:
Severe pain all the time”. Similarly, motion limitation was self-reported using a 4-point scale,
ranging from “1: No limitation” to “4: Severe limitations”.

Range of Motion
Joint range of motion (ROM) measurements were abstracted from participants’ clinical chart.
Measurements based on methods developed by the American Academy of Orthopedic
Surgeons were made on 10 joints (bilateral hips, knees, shoulders, elbows and ankles) by a
physical therapist or trained healthcare provider. For each participant, an overall joint index
was calculated by taking the sum of all flexion and extension measurements on the 10 joints.
The total joint ROM limitation was calculated as the difference between the normal overall
joint index value (1690 degrees) and the participant’s overall joint index. From this, the percent
23
overall joint limitation was determined by taking the participant’s total joint ROM limitation
and dividing it by the overall normal joint index (1690 degrees), then multiplying the fraction by
100 [27]. Negative values of percent overall joint limitation indicate that the participant has
excessive joint mobility relative to normal.

Statistical Analysis
Descriptive statistics were computed to characterize the study population. One-way analysis of
variance (ANOVA) between subjects with post-hoc testing using Scheffe’s multiple comparison
procedure for unbalanced sample sizes was used to identify differences in SF-12 or PedsQL
scores across subgroups of interest. The subgroups were self-reported joint pain levels, self-
reported motion limitation levels, hemophilic severity, and prophylaxis status for participants
with severe hemophilia. Analyses were conducted separately for adults and children. Spearman
correlation coefficients were computed to examine the correlation between clinically measured
percent ROM limitation and self-reported pain or motion limitation, as well as between
clinically measured percent ROM limitation and HrQoL scores. All analyses were conducted
using SAS version 9.2 (SAS Institute, Cary, NC).

Results
Of the 329 participants (164 adults and 165 children), 156 adults completed the SF-12 and the
self-reported pain and motion limitation evaluations. Among children, 164 (157 parent-proxies,
7 self-completed) complete records were available for both the PedsQL and self-reported pain
evaluation, while 163 (156 parent-proxies, 7 self-completed) complete self-reported motion
24
limitation evaluations were available. Clinical ROM measurements were obtained from the
medical charts of 146 participants (91 adults, 55 children) for whom measurements had been
recorded within 6 months of their initial HUGS Va interview.

Baseline characteristics of the study population are summarized in Table 2.1. Characteristics of
interest include mean ages, hemophilic severity, treatment type, mean HrQoL scores, and mean
clinical ROM measurements for both adults and children.

Adult participants had lower physical component scores (PCS-12) (mean: 43.4±10.7) than the
general US population, with PCS-12 scores decreasing with increasing hemophilic severity
(Table 2.2). The mean mental component score (MCS-12) of 50.9±10.1 was comparable to the
general US population. No significant differences in HrQoL were found between individuals with
severe hemophilia on prophylaxis (mean PCS-12: 43.1±9.9; mean MCS-12: 52.0±9.9) and on-
demand treatment (mean PCS-12: 40.4±10.5; mean MCS-12: 50.2±10.9).
Among pediatric participants, the mean total PedsQL score was 85.9±13.8; the physical
functioning subscale score was 89.5±15.2 and the psychosocial health subscale score was
84.1±15.3. As observed among the adults, PedsQL scores among the children generally
decreased with increasing hemophilic severity, and no significant differences were found
between severe hemophilia patients on prophylaxis (mean total PedsQL: 84.1±14.2) and on-
demand treatment (mean total PedsQL: 86.5±12.4).

25
Using the self-reported joint pain and motion limitation measures, one-third of adults (30.1%)
reported some pain in at least one joint some of the time even when not bleeding, while many
children (43.6%) reported no pain in any joints. Similarly, half (49.7%) the adults reported
motion limitation affecting their daily activities while many children (54%) reported no such
limitation.

The mean HrQoL scores at baseline, categorized by self-reported joint pain and motion
limitation, are presented in Tables 2.2 and 2.3, respectively. Due to the small sample size of
children who indicated joint pain as “4: Pain most of the time” (N=9) and “5: Severe pain all the
time” (N=1), these two groups were combined for analysis. Similarly, due to the small sample of
children indicating motion limitation as “3: Limitation affects activities” (N=14) and “4: Severe
limitations” (N=2), these two groups were also combined for analysis.

In adults, PCS-12 decreased as the severity of joint pain increased (Table 2.2). There were also
statistically significant (p<0.05) and minimally clinically important differences in mean PCS-12
score between groups with little or no joint pain and those with more severe joint pain.
Similarly, there was a downward trend in mean PCS-12 scores with increasing motion limitation
(Table 2.3), as well as statistically significant (p<0.05) and minimally clinically important
differences between motion limitation severity groups. Such findings were not observed with
MCS-12 scores.

26
In children, mean total PedsQL scores as well as mean scores across all PedsQL subscales
decreased as the severity of joint pain increased (Table 2.2), with statistically significant
(p<0.05) and minimally clinically important differences between groups with no or little joint
pain and those with more severe joint pain. With increasing motion limitation, mean total
PedsQL as well as all PedsQL subscale scores also decreased (Table 2.3). Statistically significant
(p<0.05) and clinically important differences were also observed in the total PedsQL, physical
functioning, psychosocial health, and social functioning scores.

The mean percent of ROM limitation, grouped by severity of self-reported joint pain and
motion limitation, are shown in Figures 2.1 and 2.2, respectively, for both adults and children.
In general, clinically measured ROM limitation increased as pain or motion limitation increased;
adults are generally more severely affected than children. A significant and positive correlation
was observed between clinically evaluated joint ROM limitation and self-reported joint pain in
the combined subsample, with correlation coefficient ρ=0.50 (p<0.0001). While a significant,
positive correlation remained within the adult subsample, coefficient ρ=0.28 (p=0.0081), it was
not observed in the subsample of children. The same positive correlation was observed in the
relationship between joint ROM limitation and self-reported motion limitation in the combined
sample, ρ=0.63 (p<0.0001), and within the adult subsample, ρ=0.48 (p<0.0001). There was also
a significant negative correlation between ROM limitation and the adult PCS-12, ρ= -0.34
(p=0.0009). A significant positive correlation was noted between joint ROM limitation and social
functioning, ρ=0.29 (p=0.045) in children. Correlations between joint ROM limitation and all
other HrQoL scores were not found to be significant.
27
Discussion
This study used generic instruments to measure HrQoL, allowing comparison of the scores of
participants with hemophilia A to those of the general healthy US population as well as to other
disease populations reported in the literature. For a rare disease like hemophilia, which
generally results in studies with small sample sizes, this study captures the HrQoL of a larger
population than those of previous published reports. Studies on HrQoL in hemophilia using
generic measures would help providers, decision- and policy-makers understand how the needs
of individuals with hemophilia differ from those individuals with other chronic disease
conditions, allowing stakeholders to identify and address the specific needs of people with
hemophilia.

In adults, considering the minimally clinically important difference (MCID) on the SF-12 to be
half the standard deviation of the mean normalized scores [21], the study population had MCS-
12 similar to the US general norm (mean: 50±10), but poorer physical functioning (PCS-12).
Stratified by disease severity, PCS-12 was lower in study participants with severe hemophilia
compared to the US general norm (mean: 50±10). Similarly, the PCS-12 scores of the study
population declined as expected with increasing pain or motion limitation. In order to place
HrQoL in hemophilia in the context of other chronic diseases, US studies were identified that
also used the SF-12. With the exception of the group experiencing the most severe joint pain or
motion limitation, the hemophilic population had better MCS-12 and PCS-12 than obese
subjects with two or more of the following: diabetes, hypertension and hyperlipidemia [28],
and individuals with type 2 diabetes [29]. When compared to a group of male asthmatics
28
between the ages of 18 and 64, the adult hemophilic population also reported better MCS-12,
but poorer PCS-12, except for the group reporting no pain or motion limitation [30].

When comparing the pediatric study sample to published reports of a sample of California
children defined as healthy, children with hemophilia had a HrQoL better than or comparable to
the healthy sample (mean: 83.8±12.7), as measured by the PedsQL subscales across all
hemophilic severity levels. Among children with mild or no joint pain or motion limitation,
HrQoL scores in the hemophilic sample were higher than those found in the healthy, non-
hemophilic population [25]. With the exception of those who reported severe joint pain or
motion limitation, children with hemophilia also reported higher PedsQL scores than published
reports of children with diabetes, end-stage renal disease, gastrointestinal conditions, cardiac
diseases, asthma, cancer, psychiatric disorders or rheumatoid conditions [25].

The hemophilia A population in this study, both adults and children, displayed mental
functioning scores (MCS-12 or psychosocial functioning) comparable to or better than the
individuals who were healthy (mean: 50±10 and 81.9±14.1, respectively) or had other chronic
diseases [25, 28-30]. Due to the physical manifestations of hemophilia, however, poorer
physical functioning among those with greater morbidity was observed as compared to the
healthy population. Despite this, the study population generally reported better physical
functioning than either the general population or most other chronic disease populations
reported in the literature [25, 28-30]. In participants with greater morbidity, these findings are
similar to those of the European, Canadian and US studies mentioned previously, which found
29
hemophilia patients to have similar mental, but poorer physical functioning compared to
healthy populations [3-6, 8, 10, 11]. One possible explanation for these findings could be that
persons with hemophilia may have adapted to their life-long condition and are able to lead
lifestyles that are minimally disruptive of their quality of life. Since this study population is
drawn from those receiving care from HTCs, it could also be attributed to the comprehensive
care that they receive, thereby helping manage their disease in a more predictable way
compared to patients with other conditions, like asthma. Further studies of the non-HTC
population and populations across several countries could help researchers identify factors that
influence the HrQoL of persons with hemophilia A. Such studies would help determine whether
the similar patterns in mental and physical scores among hemophilia patients and the general
population that has been observed in Europe, Canada, and the US are driven by the same
sociodemographic, health care delivery and clinical characteristics.

Prophylactic factor replacement therapy is commonly practiced by patients with severe
hemophilia in order to prevent bleeding episodes and has been shown to achieve better joint
health maintenance outcomes compared to on-demand therapy [31, 32]. One study has shown
that initiating prophylaxis before the age of three has also been shown to result in better
HrQoL, especially in the physical domains, than if prophylaxis is initiated after age three [7]. The
same study showed both mental and physical HrQoL of early prophylaxis patients to be
comparable to that of the general population [7]. The lack of significant or clinically important
differences in HrQoL between the two groups may be due to the inability of a cross-sectional
study to identify differences between groups or may be a result of the large variation in the
30
overall length of time participants received prophylactic treatment (mean years on prophylaxis
for adults: 6.8±6.8 years). In addition, due to the length of time required to develop severe joint
disability, differences between children using prophylaxis (mean years on prophylaxis: 5.9±4.1
years) compared to those using on-demand treatment may not yet be quantifiable. A
longitudinal comparison of the two groups may enhance understanding of the differences in
HrQoL experienced by those using prophylaxis or on-demand regimens and should be a subject
of future analysis.

In this study, the SF-12 and PedsQL were used to measure HrQoL. In addition to the SF-12,
other generic HrQoL measurement alternatives in the adult population have also demonstrated
good discrimination between various subpopulations of persons with hemophilia. The SF-36
discriminates well between severe hemophilia patients of different age groups and between
those who required orthopedic treatment or treatment for joint pain and those who did not [4],
as well as between hemophilic severities and between patients and the healthy population [3].
The Health Utilities Index, a generic instrument that measures utilities, has also demonstrated
validity in hemophilia patients with hepatitis and/or HIV [13]. In children, the PedsQL previously
exhibited high correlations with disease-specific instruments like the CHO-KLAT and HaemoQoL
[9, 12], although all three measures were unable to distinguish between patients with
moderate and severe disease, which the authors attributed to the small study sample and the
hypothesis [12].

31
The strong correlations found between self-reported joint pain, self–reported motion
limitation, and clinically measured ROM limitation suggest a potential clinical role for the use of
self-reported joint pain and motion limitation scales. In the absence of a physical therapist or a
health professional trained in ROM measurement, these scales may be of value in assessing
clinical trends over time or between clinical evaluations.

Limitations
The results of this study indicate that many persons with hemophilia treated in HTCs, with a few
exceptions, are able to maintain a HrQoL comparable to that of healthy individuals in the US.
This result is encouraging, but because the study sample did not include individuals treated
outside of the HTC setting, the findings may not be representative of all persons with
hemophilia in the US. Additional studies that include both HTC and non-HTC treated
populations may help us discern whether the high HrQoL scores reported in our study may be
attributed in part to difference in standard of care between HTCs and non-HTCs. Despite this
limitation, the study sample is representative of the regional populations served by these study
sites.
A second limitation is the use of the SF-12, which yields aggregate scores that provide some
insight but are unable to generate reliable domain scores that can be compared to other HrQoL
instruments such as the eight SF-36 domains [22]. However, the SF-12 is a shorter
questionnaire than the SF-36, which shortens the time needed to administer the questionnaire
and also reduces respondents’ burden.

32
Lastly, clinical ROM measurements were available for only half the sample population, as this
data was collected through chart reviews and these specialized measurements were not
available for all study subjects during the data collection period.

Conclusions
The HUGS Va study provides one of the largest and most representative studies of HrQoL
among both children and adults with Factor VIII deficiency treated in HTCs in the US. The
findings indicate that children and adults with hemophilia have HrQoL scores comparable to
those of the healthy US population, except among those individuals with manifestations of
severe hemophilic disease. Future studies are necessary to elicit the factors that influence
HrQoL in persons with Factor VIII deficiency in the US. Future studies should also examine how
HrQoL in this population changes over time and is influenced by the model of care, especially by
treatment regimen and by the onset of acute medical events.

33
Tables and Figures
Table 2.1: Baseline characteristics of study population
Characteristics
Total
Mean (SD)
Hemophilia Severity, Mean (SD)
Mild Moderate Severe p-value
Adults N=157 N=42 N=16 N=99

Age 33.5 (12.6) 38.4 (14.2) 34.2 (12.9) 31.3 (11.4) 0.0087
On prophylaxis (%) 40 (25.5) 1 (3.4) 1 (6.3) 38 (38.4) -
Range of motion
limitation
a.b
10.2 (9.1) 5.4 (4.5) 12.3 (14.7) 12.2 (12.0) 0.0070
Adult SF-12 measures
PCS-12 43.4 (10.7) 46.7 (10.8) 46.5 (11.5) 41.5 (10.2) 0.0140
MCS-12 50.9 (10.1) 51.1 (9.0) 50.0 (11.4) 50.9 (10.4) 0.9275
Children N=165 N=36 N=20 N=109
Age 9.7 (4.5) 9.7 (4.7) 8.7 (3.7) 9.9 (4.6) 0.5630
On prophylaxis (%) 91 (55.2) 2 (5.6) 1 (5) 88 (80.7) -
Range of motion
limitation
a,c
-2.8 (4.5) -3.6 (4.1) -3.3 (2.3) -2.5 (4.7) 0.7592
Children PedsQL measures
Total score 85.9 (13.8) 90.8 (12.2) 85.6 (13.4) 84.3 (14.1) 0.0473
Physical functioning 89.5 (15.2) 95.2 (7.3) 90.5 (13.0) 87.3 (17.0) 0.0238
Psychosocial health 84.1 (15.3) 88.4 (16.0) 82.9 (15.1) 82.8 (15.0) 0.1541
Emotional functioning 82.2 (19.1) 86.3 (21.9) 80.8 (18.9) 81.1 (18.1) 0.3548
Social functioning 87.7 (16.9) 92.1 (15.6) 84.8 (17.3) 86.8 (17.1) 0.1886
School functioning 81.6 (17.3) 85.9 (17.1) 82.9 (18.0) 79.9 (17.1) 0.2145
a
Clinically evaluated joint range of motion using methods developed by American Academy of Orthopedic Surgeons; Negative
scores indicate excess joint mobility relative to normal.
b
Adults, N=91: Mild, N=27; Moderate, N=5; Severe, N=59
c
Children, N=55: Mild, N=12; Moderate, N=2; Severe, N=41
34
Table 2.2: Mean scores for quality of life and self-reported joint pain
1. No pain
2. Pain with
joint bleed
3. Some pain
sometimes
4. Pain most
of the time
5. Severe
pain all
the time
p-value
Mean (SD)
Adults N= 15 N= 36 N= 49 N= 42 N= 21
PCS-12 53.9 (6.2)
a
48.1 (9.7)
a,b
44.8 (8.7)
b,c
38.8 (9.4)
c,d
32.1 (8.8)
d
<0.0001
MCS-12 53.4 (7.7)
a
54.7 (8.6)
a
49.1 (10.3)
a
49.4 (10.0)
a
49.0

(12.4)
a
0.0682
Children N= 72 N= 59 N= 24 N= 9
*
N= 1
*

Total PedsQL 90.8 (10.6)
a
87.0 (11.1)
a,b
77.0 (12.1)
b
65.6 (23.2)
c
<0.0001
Physical
functioning
summary
95.7 (7.3)
a
90.6 (11.8)
a
80.2 (16.0)
b
56.9 (25.8)
c
<0.0001
Psychosocial
health

88.0 (13.7)
a
85.2 (13.4)
a,b
75.4 (13.6)
b,c
70.3 (24.9)
c
<0.0001
Emotional
functioning
86.4 (17.9)
a
83.3 (18.3)
a,b
71.7 (15.9)
a,b
70.5 (26.7)
b
0.0017
Social
functioning
91.3 (14.1)
a
88.4 (13.7)
a
83.3 (18.6)
a
68.8 (30.6)
b
0.0004
School
functioning
86.1 (14.5)
a
83.0 (15.3)
a,b
71.0 (19.8)
b,c
66.9 (23.4)
c
0.0001
*
Levels 4 and 5 are combined for children due to small sample sizes
a-d
: Means with the same letters across pain severity levels are not significantly different using Scheffe’s multiple comparison
procedure (P<0.05)
Adult MCID= 5.0, Children MCID= 4.5
35
Table 2.3: Mean scores for quality of life and self-reported motion limitation

1. No
limitation
2. Limitation
with joint
bleed
3. Limitation
affects
activities
4. Severe
limitations
P-value
Mean (SD)
Adults (N= 156) N= 26 N= 41 N= 76 N= 13
PCS-12

53.8 (5.3)
a
47.5 (8.7)
a
40.2 (8.8)
b
30.3 (9.4)
c
<0.0001
MCS-12 53.0 (8.4)
a
52.7 (9.6)
a
49.5 (10.0)
a
48.6 (14.3)
a
0.2076
Children (N= 163) N= 88 N= 59 N= 14
*
N= 2
*

Total PedsQL 89.2 (11.4)
a
84.6 (12.2)
a
71.7 (21.8)
b
<0.0001
Physical functioning
summary
94.0 (9.0)
a
87.9 (13.1)
a
66.1 (28.4)
b
<0.0001
Psychosocial health
summary
86.5 (14.2)
a
82.9 (14.5)
a,b
74.9 (20.6)
b
0.0171
Emotional functioning 84.8 (18.4)
a
79.8 (19.1)
a
76.6 (22.0)
a
0.1462
Social functioning 90.5 (13.8)
a
87.8 (15.8)
a
73.0 (26.4)
b
0.0007
School functioning 84.1 (16.9)
a
80.3 (16.0)
a
73.7 (22.0)
a
0.0780
*
Levels 3 and 4 are combined for children due to small sample sizes
a-c
: Means with the same letters across motion limitation severity levels are not significantly different using Scheffe’s multiple
comparison procedure (P<0.05)
Adult MCID= 5.0, Children MCID= 4.5

36
Figure 2.1: Mean percent clinically measured joint ROM limitation by severity of self-reported joint
pain in adults (N=91) and children (N=55)

a
Lines denote standard deviation
Adults: No pain (N=9), Only when bleeding (N=22), Some of the time (N=28), Most of the time (N=19), Severe pain all the time
(N=13)
Children: No pain (N=23), Only when bleeding (N=17), Some of the time (N=12), Most of the time (N=2), Severe pain all the time
(N=1)

3.8
9.1
11.0 11.0
13.4
-3.7
-3.2
-1.6
-2.8
10.8
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
1. No pain 2. Only when
bleeding
3. Some of
the time
4. Most of the
time
5. Severe pain
all the time
Mean % joint range of motion limitation
a

Adults
Children
37
Figure 2.2: Mean percent clinically measured joint ROM limitation by severity of self-reported motion
limitation in adults (N=91) and children (N=55)

a
Lines denote standard deviation
Adults: No limitation (N=12), Only when bleeding (N=23), Affects activities (N=52), Severe limitation (N=3)
Children: No limitation (N=28), Only when bleeding (N=20), Affects activities (N=6), Severe limitation (N=1)

2.6
7.0
13.2
14.7
-3.8
-3.0
0.5
10.8
-10.0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
1. No limitation 2. Only when
bleed
3. Affects
activities
4. Severe
limitation
Mean % joint range of motion limitation
a

Adults
Children
38
Chapter References
1. Centers for Disease Control and Prevention. Hemophilia, Data and Statistics. 2013.
2. Choiniere M, Melzack R. Acute and chronic pain in hemophilia. Pain 1987; 31: 317-31.
3. Miners AH, Sabin CA, Tolley KH, Jenkinson C, Kind P, Lee CA. Assessing health-related quality-of-
life in individuals with haemophilia. Haemophilia 1999; 5: 378-85.
4. Molho P, Rolland N, Lebrun T, et al. Epidemiological survey of the orthopaedic status of severe
haemophilia A and B patients in France. The French Study Group. . Haemophilia 2000; 6: 23-32.
5. Trippoli S, Vaiani M, Linari S, Longo G, Morfini M, Messori A. Multivariate analysis of factors
influencing quality of life and utility in patients with haemophilia. Haematologica 2001; 86: 722-8.
6. Walsh M, Macgregor D, Stuckless S, Barrett B, Kawaja M, Scully MF. Health-related quality of life
in a cohort of adult patients with mild hemophilia A. J Thromb Haemost 2008; 6: 755-61.
7. Khawaji M, Astermark J, Von Mackensen S, Akesson K, Berntorp E. Bone density and health-
related quality of life in adult patients with severe haemophilia. Haemophilia 2011; 17: 304-11.
8. Klamroth R, Pollmann H, Hermans C, et al. The relative burden of haemophilia A and the impact
of target joint development on health-related quality of life: results from the ADVATE Post-Authorization
Safety Surveillance (PASS) study. Haemophilia 2011; 17: 412-21.
9. Young NL, St-Louis J, Burke T, Hershon L, Blanchette V. Cross-cultural validation of the CHO-KLAT
and HAEMO-QoL-A in Canadian French. Haemophilia 2011.
10. Schick M, Stucki G, Rodriguez M, et al. Haemophilic; arthropathy: assessment of quality of life
after total knee arthroplasty. Clin Rheumatol 1999; 18: 468-72.
11. Brown TM, Lee WC, Joshi AV, Pashos CL. Health-related quality of life and productivity impact in
haemophilia patients with inhibitors. Haemophilia 2009; 15: 911-7.
12. Young NL, Bradley CS, Wakefield CD, Barnard D, Blanchette VS, McCusker PJ. How well does the
Canadian haemophilia Outcomes-Kids' Life Assessment Tool (CHO-KLAT) measure the quality of life of
boys with haemophilia? Pediatric Blood & Cancer 2006; 47: 305-11.
13. Barr RD, Saleh M, Furlong W, et al. Health status and health-related quality of life associated
with hemophilia. Am J Hematol 2002; 71: 152-60.
14. Arranz P, Remor E, Quintana M, et al. Development of a new disease-specific quality-of-life
questionnaire to adults living with haemophilia. Haemophilia 2004; 10: 376-82.
15. Remor E, Young NL, Von Mackensen S, Lopatina EG. Disease-specific quality-of-life measurement
tools for haemophilia patients. Haemophilia 2004; 10 Suppl 4: 30-4.
16. Bullinger M, von Mackensen S, Fischer K, et al. Pilot testing of the 'Haemo-QoL' quality of life
questionnaire for haemophiliac children in six European countries. Haemophilia 2002; 8 Suppl 2: 47-54.
17. du Treil S, Rice J, Leissinger CA. Quantifying adherence to treatment and its relationship to
quality of life in a well-characterized haemophilia population. Haemophilia 2007; 13: 493-501.
18. Manco-Johnson M, Morrissey-Harding G, Edelman-Lewis B, Oster G, Larson P. Development and
validation of a measure of disease-specific quality of life in young children with haemophilia.
Haemophilia 2004; 10: 34-41.
19. Broderick CR, Herbert RD, Latimer J, Curtin JA. Fitness and quality of life in children with
haemophilia. Haemophilia 2010; 16: 118-23.
20. Jaeschke R, Singer J, Guyatt GH. Measurement of health status. Ascertaining the minimal
clinically important difference. Control Clin Trials 1989; 10: 407-15.
21. Norman GR, Sloan JA, Wyrwich KW. Interpretation of changes in health-related quality of life:
the remarkable universality of half a standard deviation. Med Care 2003; 41: 582-92.
22. Ware J, Jr., Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales
and preliminary tests of reliability and validity. Med Care 1996; 34: 220-33.
39
23. Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric Quality of Life
Inventory version 4.0 generic core scales in healthy and patient populations. Med Care 2001; 39: 800-12.
24. Varni JW, Limbers CA, Burwinkle TM. Parent proxy-report of their children's health-related
quality of life: an analysis of 13,878 parents' reliability and validity across age subgroups using the
PedsQL 4.0 Generic Core Scales. Health Qual Life Outcomes 2007; 5: 2.
25. Varni JW, Limbers CA, Burwinkle TM. Impaired health-related quality of life in children and
adolescents with chronic conditions: a comparative analysis of 10 disease clusters and 33 disease
categories/severities utilizing the PedsQL 4.0 Generic Core Scales. Health Qual Life Outcomes 2007; 5:
43.
26. Varni JW, Burwinkle TM, Seid M. The PedsQL as a pediatric patient-reported outcome: reliability
and validity of the PedsQL Measurement Model in 25,000 children. Expert Rev Pharmacoecon Outcomes
Res 2005; 5: 705-19.
27. Soucie JM, Cianfrini C, Janco RL, et al. Joint range-of-motion limitations among young males with
hemophilia: prevalence and risk factors. Blood 2004; 103: 2467-73.
28. Sullivan PW, Ghushchyan V, Wyatt HR, Wu EQ, Hill JO. Impact of cardiometabolic risk factor
clusters on health-related quality of life in the U.S. Obesity (Silver Spring) 2007; 15: 511-21.
29. Sundaram M, Kavookjian J, Patrick JH, Miller LA, Madhavan SS, Scott VG. Quality of life, health
status and clinical outcomes in Type 2 diabetes patients. Qual Life Res 2007; 16: 165-77.
30. Eisner MD, Ackerson LM, Chi F, et al. Health-related quality of life and future health care
utilization for asthma. Ann Allergy Asthma Immunol 2002; 89: 46-55.
31. Fischer K, van der Bom JG, Molho P, et al. Prophylactic versus on-demand treatment strategies
for severe haemophilia: a comparison of costs and long-term outcome. Haemophilia 2002; 8: 745-52.
32. Manco-Johnson MJ, Abshire TC, Shapiro AD, et al. Prophylaxis versus episodic treatment to
prevent joint disease in boys with severe hemophilia. N Engl J Med 2007; 357: 535-44.

40
Chapter 3: Longitudinal Changes in Health-Related Quality of Life for
Chronic Diseases: An Example in Hemophilia A

Abstract
Background: Patients with well-managed chronic diseases such as hemophilia maintain a stable
health state and health-related quality of life (HrQoL) that may be affected by acute events.
Longitudinal HrQoL assessments analyzed using multivariate multilevel (MVML) modelling can
determine the impact of such events on individuals (within-person effect) and identify factors
influencing within-population differences (between-person effect). Objectives: To disaggregate
the within- and between-person factors influencing HrQoL over time in persons with
hemophilia A. Methods/Design/Participants: Using data on 136 adults and 125 children from a
two-year, observational cohort study of burden of illness in US hemophilia A patients, MVML
modelling determined the effect of time-invariant (sociodemographic and clinical
characteristics) and time-varying factors (bleeding frequency, emergency room (ER) visits and
missed work/school days) on within- and between-person HrQoL changes. HrQoL was assessed
using the SF-12 (adults) and PedsQL (children) at baseline, then every 6 months. Results: In
children, psychosocial functioning within- (p<0.0001) and between-persons (p<0.0001) was
reduced by each additional bleed and missed day (Within-person: p=0.0089; Between-person:
p=0.0060). Within-person physical functioning was reduced by each additional bleed
(p<0.0001), ER visit (p=0.0284) and missed day (p=0.0473). Between-persons, additional missed
days (p<0.0001) significantly decreased physical functioning. In adults, each additional missed
day reduced SF-12 mental (p=0.0025) and physical (p=0.0093) component scores. Each
additional bleed also decreased PCS significantly (p=0.0093). Conclusions: HrQoL was
41
significantly impacted by the frequency of missing additional work/school days and
experiencing additional bleeds relative to the study population (between-person), and to the
individual’s experience of acute events across time (within-person). MVML modelling identified
small but significant within- and between-person changes in HrQoL with each additional event
which, if frequent, could have a large cumulative impact. This study would help researchers and
healthcare providers understand the factors mediating HrQoL in hemophilia A, allowing for
better care management at the individual patient and disease-population levels.

42
Introduction
In stable chronic disease, well-managed patients typically maintain a relatively constant health
state over a period of time, as measured by disease-specific clinical and symptoms
manifestation indicators. Over the same period, patient-reported outcomes (PROs) such as
health-related quality of life (HrQoL) are also expected to remain relatively stable in the
absence of acute exacerbations of the disease. As a result, observing the same patient at two
discrete time points might lead to conclusions that there were no significant changes in clinical
outcomes or PROs over time.

Although the individual patient’s clinical condition is not expected to deviate significantly over
time from their chronic baseline, their condition is expected to fluctuate as time progresses, in
response to acute clinical exacerbations or external changes. This in turn leads to “within-
person” deviations in HrQoL from the chronic baseline. While it is important to be able to
identify within-person fluctuations in HrQoL, the observed variation does not necessarily
generalize to any differences that may exist between individuals with the same disease. It will
also be of interest to identify such “between-person” effects of influencing factors on HrQoL.

In order to identify such within- and between-person effects of factors influencing HrQoL,
repeated measurements of the same patients must be made across a period of time. Even if a
significant pattern of growth or decline over time is not demonstrated, using longitudinal data
would allow for the within- and between-person effects to be disaggregated. This would help
researchers and healthcare providers understand the factors mediating HrQoL, allowing the
43
latter to better manage patient care both on the levels of an individual patient and disease-
population.

To illustrate the partitioning of between- and within-person effects of clinical measures or
symptoms manifestation on HrQoL over time, we consider a sample of persons with hemophilia
A. Hemophilia A is a rare, chronic, X-linked disease primarily affecting males, characterized by a
deficiency in clotting factor VIII, resulting in the inability of the blood to clot normally. In the
United States (US), it affects one in every 5000 male births [1]. The major clinical manifestation
of hemophilia A is bleeding episodes, most frequently into knee, ankle and elbow joints.
Bleeding may occur due to trauma, or in more severe forms of the disease, may occur
spontaneously. Persons with hemophilia A are treated with either regular prophylactic infusions
of clotting factor concentrates or on-demand infusions during episodes of acute bleeding
events. During such acute episodes, patients may experience pain in the joints that may cause
some degree of motion limitation but otherwise, are able to lead active and productive lives [2].
In hemophilia A, we can see that patients experience long periods of symptom-free clinical
stability, punctuated by acute bleeding events which may be accompanied by joint problems.
We expect that HrQoL in hemophilia A patients will be associated with the frequency of
symptoms and environmental changes and thus will maintain a stable chronic baseline, with
fluctuations due to acute events.

Several studies have previously captured longitudinal HrQoL data in persons with bleeding
disorders, including hemophilia A, over periods ranging from 18 months to 5 years [3-5].
44
However, none have attempted to describe changes in HrQoL due to corresponding changes in
level of symptoms or environmental circumstances. Over 18 months, Gringeri et al [3] found
HrQoL to remain stable in a group of 41 hemophilia A and B patients with inhibitors when
measured at 6-month intervals and they therefore reported only baseline HrQoL results.
Similarly, over a 5-year period with annual follow-up, Lindvall et al [4] found no statistically
significant changes in HrQoL among a group of 10 patients with severe hemophilia. Solovieva
[5] observed a statistically significant change in physical functioning and vitality on the SF-36
scale in a group of 150 patients with various bleeding disorders over 3 years. However, the
study only observed the patients at two time points (year zero (baseline) and 3 years later) and
was not designed to observe any fluctuations in HrQoL that might have occurred in the
intervening period.

Over time, HrQoL is hypothesized to vary between-persons due to inter-individual differences
in disease manifestation and time-invariant sociodemographic characteristics. We also
hypothesize that there are fluctuations in HrQoL within-person, dependent on acute episodes
of symptom manifestations such as bleeding and its associated sequelae, the frequencies of
which vary over time. To address these hypotheses, we employed multivariate multilevel
modelling on longitudinal data collected by the Hemophilia Utilization Group Study Part Va
(HUGS Va) over a two-year period, to estimate between- and within-person associations
between level of symptoms and associated sequelae, and HrQoL.

45
Methods
Data
The Hemophilia Utilization Group Study Part-Va (HUGS Va), was a 2-year, multicenter
observational cohort study conducted among hemophilia A patients from six Hemophilia
Treatment Centres (HTCs) located in geographically diverse regions of the US. Details about the
study methods and study inclusion criteria have previously been reported [6, 7]. Informed
consent and data were collected at initial interview from participant/parent self- or proxy-
report and patient chart reviews completed by healthcare providers. Information regarding
clinical aspects of the disease, such as treatment regimen, arthropathy and comorbidities, as
well as HrQoL and the economic consequences of having hemophilia A were collected. Periodic
participant follow-up surveys were used to collect data on time lost from work, school or usual
activities, disability days, healthcare utilization and clinical outcomes, including bleeding
episodes, joint pain and motion limitation. Follow-up interviews were administered monthly in
the first year and semi-annually in the second year. HrQoL was measured at baseline and every
six months throughout the two-year follow-up. Baseline HrQoL and its association with joint
pain and motion limitation in the HUGS Va population has previously been reported [7]. This
analysis includes 261 participants (136 adults and 125 children) who completed their initial
interview and at least one follow-up, out of a total of 329 participants (164 adults and 165
children) with factor VIII deficiency who were recruited into the HUGS Va study between July
2005 and July 2007.

46
The University of Southern California (USC) served as the data and coordinating center. The
study protocol was approved by the Institutional Review Board of USC (IRB number: HS-
046012) and that of each participating HTC.

HrQoL instruments
Adult general health was assessed using the SF-12 Health Survey Version 1, which has been
used in previous hemophilia studies [6-9]. This is a 12-item version of the widely-used SF-36
that assesses eight specific dimensions of HrQoL: physical functioning, role-physical, bodily
pain, general health, vitality, social functioning, role-emotional, and mental health, to yield two
summary physical component (PCS-12) and mental component (MCS-12) scores [10]. The
minimal clinically important difference (MCID) in SF-12 score is considered to be 5.0 [10]. HrQoL
of participants younger than 18 years was assessed using the PedsQL™ 4.0 generic core scales
[11], a generic, non-disease specific HrQoL instrument developed in the US for children and
adolescents shown to be valid and reliable for both self and parent-proxy administration. It is
also sensitive to differences between disease types and between severity groups within the
same disease [11-13]. In this study, parents of participants aged 2-17 years proxy-administered
the questionnaire. The PedsQL consists of 23 items that assess four subscales of functioning:
physical, emotional, social and school-related. These contribute to two summary scores:
physical and psychosocial health, as well as a total score computed by taking the mean of the
two summary scores. The MCID of the PedsQL is 4.5 [14].

47
Generic HrQoL instruments were used instead of a hemophilia-specific tool in order to provide
a basis for comparison with other disease populations. Validated hemophilia-specific HrQoL
instruments were not available at the time of data collection.

Analysis
HrQoL was measured at times 0, 6, 12, 18 and 24 months. Baseline (0 month) and
measurements obtained within a 6 month window (3 months pre- and post-) of each follow-up
time point were included for analysis. The outcomes of interest for adults were the SF-12 PCS-
12 and MCS-12. For children, the outcomes of interest were the two summary scores of
physical health and psychosocial health from the PedsQL.

In order to disaggregate within- and between-person effects of patient characteristics and
outcomes on HrQoL, both time-varying and time-invariant covariates need to be considered.
Time-varying covariates include the frequency of bleeding episodes, emergency room visits and
missed days of work or school due to hemophilia. For each event, the total number of
occurrences between each follow-up period, as determined by the time between each HrQoL
measurements, were summed and attributed to that period. Time-invariant covariates include
hemophilia severity, treatment type, patient or parent employment status and health insurance
coverage.

Multivariate multilevel (MVML) modelling was performed using SAS 9.3 (SAS Institute Inc., Cary,
NC, USA) to characterize longitudinal within- and between-person variations in HrQoL. Data
48
from longitudinal studies such as HUGS Va have a hierarchical structure, where measurements
at each follow-up time point are nested within individual subjects. Due to repeated
measurements on the same subjects, the resultant correlation of residuals in the model must
be controlled. Multilevel models represent the data in hierarchical levels, controlling residual
correlations and eliminating the need to aggregate data or analyze each level separately. At the
same time, it estimates each individual subject’s growth trajectory (within-person effects) while
also determining how individual growth parameters vary as a function of differences between
subjects in background characteristics (between-person effects) [15, 16].

The analysis is conceptualized in two stages. In stage 1, the within-person model is considered,
where the HrQoL of a participant is a function of his systematic growth curve plus random error
[15-17]. The time-varying covariates for each participant center on his cross-time average
(mean frequency of event for each participant in each six-month period), representing the rate
of change of HrQoL with every unit increase in the covariate above the individual-mean at each
time point. Including the time-varying covariates of interest in this stage of the analysis
estimates the within-person effects of these covariates. Time-invariant covariates do not need
to be included in this stage of the analysis as they remain constant over time within each
participant.

In stage 2, we are interested in the between-person effects of how HrQoL varies as a function
of clinical or sociodemographic differences between individuals. These would include the time-
invariant covariates and the grand-mean (mean frequency for the entire study population
49
across two years) centered cross-time averages of each participant’s time-varying covariates.
The equations for the MVML model are included in the appendix.

Addressing missing data
To account for missing data due to loss of follow-up in both the outcomes of interest (HrQoL
scores) and time-varying covariates of interest, multiple imputation was employed using
IVEware version 0.2 (Survey Methodology Program, Survey Research Center, Institute for Social
Research, University of Michigan) [18]. A multivariate imputation procedure is performed
sequentially on each time-varying variable, conditioned on the subject’s age and race as well as
all observed time-varying variables. Imputed continuous variables are assumed to follow a
normal distribution and count variables are assumed to follow a Poisson distribution.

Results
Of the HUGS Va study population, 261 participants (136 adults, 125 children) had baseline and
at least one follow-up HrQoL measurement. The mean ages of adults and children in the study
are 33.1±12.0 and 9.6±4.4 respectively. The majority of the study population is White/Non-
Hispanic (75.0% adults; 68.0% children) and have severe hemophilia A (64.7% adults; 67.2%
children). Most participants or their parents have part- or full-time employment (62.5% adults;
73.6% children) and have some form of health insurance (91.9% adults; 100% children). (Table
3.1)

50
Over two years, wide variations were observed between each 6-month follow-up period in
mean number of bleeding episodes, ER visits, and missed days of work or school due to
hemophilia A (Table 3.2). Mean MCS-12 and PCS-12 for the entire adult study population at
baseline were 50.8±10.2 and 43.0±10.8 respectively, remaining fairly constant throughout the
duration of follow-up (Figure 3.1). Mean psychosocial and physical functioning scores for the
pediatric study population were 83.0±15.7 and 88.0±16.4 respectively, with greater variation in
physical functioning scores than psychosocial functioning scores over time (Figure 3.2).

From the MVML models in the adult study population (Table 3.3), employment status was
found to have a significant influence on MCS-12 (p=0.0160 for full-time employed vs
unemployed) as did between-person differences in missed work/school days (p=0.0026) and
within-person changes in bleeding frequency (p=0.0243) and ER visits (p=0.0350). The
interactions of employment status with within-person changes in bleeding episodes (p=0.0069
for part-time vs unemployed); employment with within-person changes in missed days
(p=0.0309 for full-time vs unemployed); insurance status with within-person changes in ER visits
(p=0.0160 for no insurance vs public insurance); and within-person changes in missed days with
hemophilia severity (p=0.0161) significantly moderated MCS-12. PCS-12 was significantly
influenced by insurance status (p=0.0047 and p=0.0007 for full- and part-time employed vs
unemployed respectively), hemophilia treatment type (p=0.0424), both between- and within-
person variations in missed work/school days PCS-12 (Between-person: p<0.0001; Within-
person: p=0.0064) and between-person differences in bleeding frequency (p=0.0093). The
interaction of between-person differences in missed days with within-person changes in
51
bleeding episodes (p=0.0223), as well as of between-person differences in bleeding episodes
with within-person changes in ER visits (p=0.0046) significantly influenced PCS-12.

In the pediatric study population (Table 3.4), the MVML model found hemophilia severity
(p=0.0139), insurance status (p=0.0047) and parental employment status (p=0.0496 for full-
time vs unemployed) to have significant influence on psychosocial functioning. Between-
(p<0.0001) and within-person (p=0.0013) variations in bleeding frequencies and both between-
(p=0.0043) and within-person (p=0.0023) variations in days missed from school/work also had
significant impact on psychosocial functioning. Several interaction terms were also found to
moderate psychosocial functioning. These included the interaction of disease severity with
treatment type (p=0.0404), employment status with within-person changes in missed days
(p=0.0035 for full-time vs unemployed), and between-person differences in bleeding episodes
and within-person changes in missed days (p=0.0003). Physical functioning was significantly
influenced by between-person differences in missed days from school/work (p<0.0001) and by
within-person changes in bleeding frequencies (p=0.0031) and ER visits (p=0.0095). The
interaction of within-person changes in bleeding frequencies with hemophilia severity also
significantly influenced physical functioning (p=0.0125).

Discussion
In this hemophilia study population, experiencing additional missed work/school days and
bleeding episodes relative to the study population, and themselves across time significantly
impacted HrQoL for both adults and children, affecting their ability to maintain an active and
52
productive lifestyle as experienced by those with stable, well-controlled hemophilia. With
longitudinal monitoring of patients’ HrQoL, it may be possible to obtain a reflection of a
patient’s disease control. Additionally, acute events for children were found to impact mental
aspects of HrQoL to a greater extent than physical aspects, whereas the reverse is true in
adults. An explanation is that acute events restrict the activities of children, reducing their
social interactions with peers and consequently, potentially having a negative impact on their
self-esteem, unlike adults who may with maturity, be better at managing such effects. In adults,
new acute events may exacerbate prior joint damage, leading to greater decrements in physical
aspects of HrQoL. Fluctuations in HrQoL were also found to be greater over time among
children, but less so among adults, who with age, may again have become better at managing
symptoms and problems. Another explanation may have to do with an interaction between
HrQoL instrument and age, as children could not complete the SF-12 and the PedsQL although
well-validated, may have different sensitivity to change.

Employing MVML modelling allowed for the disaggregation of both between-person and
within-person changes in HrQoL allowing for the identification of small but significant changes
that occur with each additional acute event (bleeds/ER visits/missed days). Individually, each
additional event decreased mental or physical HrQoL by only a small amount, but if frequently
occurring within a time period, the cumulative impact on patients’ HrQoL could be large, such
that there may exist clinically important differences in HrQoL either within-person across time
periods or between-persons. The MVML model also identified the impact of sociodemographic
differences between individuals and identified sociodemographic characteristics that associate
53
(interact) with the frequency of acute events to moderate HrQoL. Of interest is the interaction
between employment and missed days, which had an opposing effect on MCS-12 and
psychosocial functioning of adults and children, respectively. In adults working full-time, each
additional missed day of work reduced MCS-12 significantly, while in children whose parent
worked full-time, each additional missed day of school increased psychosocial functioning
significantly. A possible explanation is that adults may have concerns about taking sick days off
work and how that may affect their employment whereas such concerns do not exist for
children. Taken together, sociodemographic differences, the frequency of acute events and the
interactions between them have an additive effect on HrQoL of the individual patient both
relative to others with hemophilia as well as to themselves over time.

The methodology employed in this study may be applied to other chronic rare diseases. Using
generic measures of HrQoL such as the SF-12 and PedsQL allows HrQoL to be compared across
different disease populations. Additionally, studies conducted in rare diseases typically have
small sample sizes, leading to difficulties when trying to select a suitable method of statistical
analysis. Here, with a relatively small hemophilia population, the longitudinal study design
collected data at five time points, increasing the sample size and power for analysis. By applying
MVML modelling to the data, we were able to identify both time-invariant and time-varying
factors that influenced changes in HrQoL. This suggests that such a method, where
measurements are made across more than two time points, may also find application in studies
of longitudinal change in HrQoL in other chronic/rare disease populations that like hemophilia,
have long periods of stability with occasional acute exacerbations. This could serve as an
54
alternative way for healthcare providers to monitor a patient’s disease control through
monitoring their HrQoL.

This analysis was restricted by the small study sample size of 136 adults and 125 children,
although increasing the number of measurements increased overall sample size and power. As
mentioned previously however, this is a typical occurrence in studies of rare diseases and to our
knowledge this is one of the larger studies conducted in a hemophilia A population. Further
validation of the model and estimates obtained should be performed in additional hemophilia
populations and the feasibility of applying the methods to other disease populations explored.
An additional limitation of this study is the time of measurement of HrQoL which were
performed at fixed time points and not precisely captured at the exact time of an acute event
occurring. Despite this, significant changes in HrQoL due to acute events were identified which
suggest that the true impact of experiencing an acute event on HrQoL may be underestimated.
Future studies should attempt to measure HrQoL at or as close as possible to the time of event
occurrence, although this may prove to be a challenge.

Understanding how HrQoL is influenced by both sociodemographic and clinical factors over a
prolonged period time would help researchers and healthcare providers identify the factors
mediating HrQoL in patients with stable chronic diseases, such as hemophilia A. This would
allow providers to tailor patient education and management strategies, allowing for better care
management both at the individual patient and disease-population levels.

55
Tables and Figures
Table 3.1: Baseline Demographics
Variables, N (%)
All Patients
(N=261)
Adults
(N=136)
Children
(N=125)
Mean Age at Baseline (SD) 21.8 (15.0) 33.1 (12.0) 9.6 (4.4)
Race
White
Other

187 (71.7)
74 (28.3)

102 (75.0)
34 (25.0)

85 (68.0)
40 (32.0)
Hemophilia Severity
Mild/Moderate
Severe

89 (34.1)
172 (65.9)

48 (35.3)
88 (64.7)

41 (32.8)
84 (67.2)
Treatment Type
Prophylaxis
On-Demand

110 (42.2)
151 (57.8)

36 (26.5)
100 (73.5)

74 (59.2)
51 (40.8)
Health Insurance Type
Private Insurance
Public Insurance
No Insurance

174 (66.7)
76 (29.1)
11 (4.2)

82 (60.3)
43 (31.6)
11 (8.1)

92 (73.6)
33 (26.4)
0 (0.0)
Employment Status
Full-time
Part-time
Unemployed

116 (44.4)
61 (23.4)
84 (32.2)

60 (44.1)
25 (18.4)
51 (37.5)

56 (44.8)
36 (28.8)
33 (26.4)

Table 3.2: Study population grand means over time
Follow-up Period
(Mean±SD)
Number of Bleeds ER Visits
Missed Days of
Work/School
Adults (N=136)

6 months
12 months
18 months
24 months
10.4±14.8
8.0±10.5
10.1±15.7
6.5±11.6
0.9±1.4
0.5±1.2
0.4±0.8
0.3±0.8
12.0±33.9
11.8±38.0
6.9±21.3
5.2±23.3
Children (N=125)

6 months
12 months
18 months
24 months
4.8±10.7
3.8±6.7
4.7±12.2
7.1±13.5
0.6±1.6
0.3±1.1
0.5±1.3
0.3±0.8
2.5±4.7
4.8±17.9
3.6±13.2
4.5±8.6

56
Table 3.3: Adults - Results from multivariate multilevel model (N=136)
Mental Component Score Physical Component Score
β SE p-value β SE p-value
Intercept 47.4 2.0 35.7 2.0
Time -0.1 0.2 0.6355 -0.1 0.2 0.5383
Hemophilia Severity:
Mild/Moderate
9.3 5.8 0.1093 -6.8 5.8 0.2424
Treatment Type: On-demand
Therapy
1.2 1.9 0.5378 2.1 1.9 0.2613
Insurance (Ref= Public)
Private Insurance
No Insurance

1.5
-0.6

1.8
2.9

0.4004
0.8240

3.8
-0.2

1.8
2.9

0.0322
0.9575
Employment Status
(Ref=Unemployed)
Full-time
Part-time

-0.7
5.1

1.9
2.1

0.0160
0.7217

5.3
7.2

1.9
2.1

0.0047
0.0007
Between-Person Effects
Bleeding Episodes
ER Visits
Days missed from work/school

0.008
-1.4
-0.1

0.09
1.4
0.04

0.9250
0.3063
0.0026

-0.2
-0.1
-0.2

0.09
1.3
0.04

0.0093
0.9427
<0.0001
Within-Person Effects
Bleeding Episodes
ER Visits
Days missed from work/school

-0.00004
0.3
-0.003

0.05
0.4
0.02

0.9993
0.4817
0.8603

-0.04
0.3
-0.04

0.03
0.3
0.01

0.2170
0.3652
0.0064
Interactions of Interest
Severity and Treatment (Severe
hemophilia and receiving
prophylaxis)
Employment and W/in Bleeds
Full-time
Part-time
Employment and W/in Missed
days
Full-time
Part-time
ER Visits and Insurance
Private Insurance
No Insurance
Days missed and Severity:
Mild/Moderate Hemophilia
Between-person missed days
and Within-person bleeding
episodes
Between-person bleeding
episodes and Within-person ER
visits

-8.1

0.01
-0.3

-0.1
0.03

-0.9
-3.2

0.09
-

-

6.0

0.07
0.1

0.1
0.03

0.6
1.3

0.04
-

-

0.1796

0.8818
0.0069

0.0309
0.4391

0.1718
0.0160

0.0161
-

-

7.9

-
-

-
-

-
-

-
-0.0003

0.07

6.0

-
-

-
-

-
-

-
0.001

0.03

0.1888

-
-

-
-

-
-

-
0.0223

0.0046

57
Table 3.4: Children - Results from multivariate multilevel model (N=125)
Psychosocial Functioning Physical Functioning
β SE p-value β SE p-value
Intercept 75.4 3.1 82.5 3.3
Time 0.2 0.3 0.4069 -0.2 0.3 0.4675
Hemophilia Severity: Mild/Moderate -8.8 3.5 0.0139 -2.4 3.7 0.5239
Treatment Type: On-demand Therapy -10.7 7.3 0.1421 -7.0 7.6 0.3565
Insurance (Ref= Public)
Private Insurance

6.4

2.2

0.0047

4.5

2.4

0.0588
Employment Status (Ref=Unemployed)
Full-time
Part-time

-4.5
-0.4

2.3
2.5

0.0496
0.8754

1.0
4.2

2.4
2.6

0.6868
0.1117
Between-Person Effects
Bleeding Episodes
ER Visits
Days missed from work/school

-0.8
-0.7
-0.6

0.2
1.4
0.2

<0.0001
0.6025
0.0043

-0.2
0.04
-1.4

0.2
1.4
0.2

0.2425
0.9804
<0.0001
Within-Person Effects
Bleeding Episodes
ER Visits
Days missed from work/school

-0.2
-0.4
-0.7

0.1
0.4
0.2

0.0013
0.3611
0.0019

-0.5
-1.4
-0.3

0.2
0.5
0.2

0.0035
0.0095
0.0666
Interactions of Interest
Severity and Treatment (Severe
hemophilia and receiving
prophylaxis)
Employment and W/in Missed days
Full-time
Part-time
Between-person bleeding episodes
and within-person missed days
Within-person bleeding episodes and
Severity: Mild/Moderate

-16.4

0.7
-0.3
0.06

-

8.0

0.3
0.3
0.02

-

0.0404

0.0035
0.3955
0.0003

-

-14.2

-
-
-

-0.4

8.4

-
-
-

0.2

0.0909

-
-
-

0.0125
Figure 3.1: Adults – Study population mean (SD) follow-up HrQoL scores (N=136)

Figure 3.2: Children – Study population mean (SD) follow-up HrQoL scores (N=125)

50.8
(10.2)
49.5
(10.3)
50.8
(10.7)
49.2
(10.6)
50.0
(10.5)
43.0
(10.8)
43.6
(11.3)
43.7
(11.9)
43.2
(11.6)
42.4
(11.5)
38
40
42
44
46
48
50
52
Baseline 6 mths 12 mths 18 mths 24 mths
SF-12
Mental Physical
83.0
(15.7)
81.5
(13.2)
79.1
(14.8)
80.7
(16.2)
83.1
(13.5)
88.0
(16.4)
82.8
(18.8)
82.0
(20.6)
84.4
(15.5)
84.4
(16.0)
74
76
78
80
82
84
86
88
90
Baseline 6 mths 12 mths 18 mths 24 mths
PedsQL
Psychosocial Physical
59
Chapter References
1. Centers for Disease Control and Prevention. Hemophilia, Data and Statistics. 2013.
2. World Federation of Hemophilia. Guidelines for the Management of Hemophilia. World
Federation of Hemophilia, 2005.
3. Gringeri A, Mantovani LG, Scalone L, Mannucci PM. Cost of care and quality of life for patients
with hemophilia complicated by inhibitors: the COCIS Study Group. Blood 2003; 102: 2358-63.
4. Lindvall K, Von Mackensen S, Berntorp E. Quality of life in adult patients with haemophilia - a
single centre experience from Sweden. Haemophilia 2012; 18: 527-31.
5. Solovieva S. Clinical severity of disease, functional disability and health-related quality of life.
Three-year follow-up study of 150 Finnish patients with coagulation disorders. Haemophilia 2001; 7: 53-
63.
6. Zhou ZY, Wu J, Baker J, et al. Haemophilia utilization group study - Part Va (HUGS Va): design,
methods and baseline data. Haemophilia 2011; 17: 729-36.
7. Poon JL, Zhou ZY, Doctor JN, et al. Quality of life in haemophilia A: Hemophilia Utilization Group
Study Va (HUGS-Va). Haemophilia 2012; 18: 699-707.
8. Schick M, Stucki G, Rodriguez M, et al. Haemophilic; arthropathy: assessment of quality of life
after total knee arthroplasty. Clin Rheumatol 1999; 18: 468-72.
9. Brown TM, Lee WC, Joshi AV, Pashos CL. Health-related quality of life and productivity impact in
haemophilia patients with inhibitors. Haemophilia 2009; 15: 911-7.
10. Ware J, Jr., Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales
and preliminary tests of reliability and validity. Med Care 1996; 34: 220-33.
11. Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric Quality of Life
Inventory version 4.0 generic core scales in healthy and patient populations. Med Care 2001; 39: 800-12.
12. Varni JW, Limbers CA, Burwinkle TM. Impaired health-related quality of life in children and
adolescents with chronic conditions: a comparative analysis of 10 disease clusters and 33 disease
categories/severities utilizing the PedsQL 4.0 Generic Core Scales. Health Qual Life Outcomes 2007; 5:
43.
13. Varni JW, Limbers CA, Burwinkle TM. Parent proxy-report of their children's health-related
quality of life: an analysis of 13,878 parents' reliability and validity across age subgroups using the
PedsQL 4.0 Generic Core Scales. Health Qual Life Outcomes 2007; 5: 2.
14. Varni JW, Burwinkle TM, Seid M. The PedsQL as a pediatric patient-reported outcome: reliability
and validity of the PedsQL Measurement Model in 25,000 children. Expert Rev Pharmacoecon Outcomes
Res 2005; 5: 705-19.
15. Bryk AS, Raudenbush SW. Application of Hierarchical Linear-Models to Assessing Change.
Psychol Bull 1987; 101: 147-58.
16. MacCallum RC, Kim C, Malarkey WB, KiecoltGlaser JK. Studying multivariate change using
multilevel models and latent curve models. Multivar Behav Res 1997; 32: 215-53.
17. Curran PJ. Have multilevel models been structural equation models all along? Multivar Behav
Res 2003; 38: 529-68.
18. Raghunathan TE, Lepkowski JM, Van Hoewyk J, Solenberger P. A Multivariate Technique for
Multiply Imputing Missing Values Using a Sequence of Regression Models. Survey Methodology 2001;
27: 85-95.

60
Appendix
Multivariate Multilevel Model

The multivariate multilevel model is conceptualized in two stages. In stage 1, we consider the within-
person model, which is expressed as
𝑦 𝑖 𝑡 = 𝛽 0 𝑖 + � 𝛽 𝑝 𝑖 𝑥 𝑝 𝑖𝑡 𝑃 𝑝 = 1
+ 𝑟 𝑖𝑡
where 𝑦 𝑖 𝑡 is the HrQoL outcome of interest for individual i, at time t. 𝛽 0 𝑖 is the within-person intercept
for individual i, reflecting the individual’s true value of HrQoL at baseline. 𝛽 𝑝𝑖
is the regression
coefficient of 𝑦 𝑖 𝑡 on the p
th
time-varying covariate x for individual i, centered on the individual’s cross-
time average, representing the rate of change of HrQoL with every unit increase in covariate x at each
time point above the individual-mean . 𝑟 𝑖 𝑡 is the time- and individual- specific residual. Time-invariant
covariates are not included in stage 1.

In stage 2, we consider the between-person effects of how HrQoL varies as a function of hemophilia
severity, treatment type, employment status and health insurance coverage. They also include time-
invariant, grand-mean centered cross-time averages of each individual’s time-varying covariates.
Mathematically, this is expressed as:
𝛽 0 𝑖 = 𝛾 0 0
+ � 𝛾 0 𝑞 𝑤 𝑖𝑞 𝑄 𝑞 = 1
+ 𝑢 0 𝑖
𝛽 𝑝 𝑖 = 𝛾 𝑝 0
+ � 𝛾 𝑝 𝑞 𝑤 𝑖𝑞
𝑄 𝑞 = 1
+ 𝑢 𝑝 𝑖

61
the γs represent the fixed coefficients for the regression of 𝛽 0 𝑖 and 𝛽 𝑝𝑖
from stage 1 on the stage 2 time-
invariant covariates 𝑤 𝑖 𝑞 , 𝑢 0 𝑖 and 𝑢 𝑝𝑖
are the associated stage 2 residuals.
The stage 1 and 2 equations when combined reduce to:
𝑦 𝑖𝑡 = 𝛾 0 0
+ � � 𝑤 𝑖 𝑞 � 𝛾 0 𝑞 + � 𝛾 𝑝 𝑞 𝑥 𝑝 𝑖𝑡 𝑃 𝑝 = 1
�
𝑄 𝑞 = 1
� + � 𝛾 𝑝 0
𝑥 𝑝 𝑖𝑡 𝑃 𝑝 = 1
+ � 𝑢 0 𝑖 + � 𝑢 𝑝 𝑖 𝑥 𝑝 𝑖𝑡 𝑃 𝑝 = 1
+ 𝑟 𝑖𝑡 �

62
Chapter 4: Societal Preferences for Parental Willingness-To-Pay in
Children’s Hemophilia Treatment

Abstract
Background: Hemophilia is a bleeding disorder which if poorly controlled, may result in joint
damage due to frequent bleeds and a restriction of physical activities or a normal lifestyle. Due
to its treatment regimen, it is also a costly disease. Aims: To elicit the relative preferences of
society towards hemophilia treatment attributes and outcomes, and to elicit from the societal
perspective, willingness-to-pay for a child to obtain hemophilia treatment that would allow him
to engage in his desired level of leisure activities and sport, while minimizing adverse bleeding
events. Methods: Two experiments were conducted, the first employing a discrete choice
experiment (DCE) where an online convenience sample evaluates treatment options that vary
according to: Treatment frequency, bleeding frequency, level of physical activity and additional
monthly insurance premium. Conditional logit regression was used to determine preferences
and willingness-to-pay. Experiment 2 supplements experiment 1 by eliciting median willingness-
to-pay for hemophilia treatment from a representative US sample and identifying
sociodemographic characteristics of participants likely to have greater willingness-to-pay for
treatment. Results: Experiment 1 had 125 participants who identified having zero bleeding
episodes per month and being able to participate in contact sports as the most important
attributes of hemophilia treatment. Willingness-to-pay for treatment that would reduce bleeds
from more than 3 a month to zero was $208.64/month in additional insurance premium and
willingness-to-pay for being able to participate in contact sports over low intensity activities
63
was $224.81/month in additional premiums. Experiment 2 had 1027 respondents who
identified median willingness-to-pay to be between $100 and $199, with household income,
level of education and home internet access being factors that significantly influenced a
participant’s likelihood of being willing-to-pay for treatment. Conclusions: Experiment 1
demonstrated the face validity and feasibility of using DCEs to elicit preferences and
willingness-to-pay for hemophilia treatment, although the results will require further validation
in a hemophilia population. Both experiments showed that participants are willing to trade-off
monetary payments in order to obtain clinical and physical benefits for hemophilia patients,
especially those with higher household incomes and the more highly educated or well-
informed, as identified in experiment 2.

64
Introduction
In hemophilia, it is well understood that the main symptom of the disease is bleeding, especially
into the muscles and joints. Recurrent bleeds lead to progressive joint damage, causing chronic
hemophilic arthropathy which is a major cause of morbidity affecting everyday function [1]. To
avoid the chronic complications of hemophilic arthropathy and other acute consequences of
bleeding episodes, the principal goal of hemophilia management is to prevent the occurrence
of bleeds.

In order to prevent and minimize the effects of bleeding, patients receive infusions of clotting
factor concentrates. In particular, severe hemophilia patients are encouraged to practice
prophylactic factor replacement therapy, which is the infusion of factor following a regular
schedule [2]. Current prophylactic factor replacement regimens for hemophilia require patients
to receive infusions either every other day or every three or four days. Newer longer-acting
products currently in the regulatory approval process are recommended by their manufacturers
to be administered once a week. The high cost of coagulation factors and the need for
frequent, regular administration however, may lead to financial constraints or restrictions from
insurance providers, preventing patients from receiving prophylaxis.

Due to concerns about bleeding episodes and its consequences, up to the early 1970s, physical
activity was discouraged in hemophilia patients for fear that this may increase the risk of
bleeding and subsequent arthropathy [3]. In recent decades however, it has been found that
regular exercise for hemophilia patients can help build muscle strength, increase joint range of
65
motion, reduce pain related to hemophilic arthropathy and reduce the frequency of joint
bleeds [4]. As a result, patients are now encouraged to participate in some form of regular
physical activity.

In their recommendations on medical conditions affecting sports participation published in
2008, the American Academy of Pediatrics gave a “qualified yes” recommendation to sports
participation for persons with bleeding disorders, including hemophilia. However, they also
cautioned that the safety of the sport for each child should be evaluated by a physician prior to
participation [5]. Some practitioners while encouraging regular physical activity, have
nevertheless also discouraged participation in vigorous contact sports [6]. This inability to
participate in contact (or team) sports has led to considerable disappointment among patients
and may have an impact on their self-esteem and social interactions [7].

Due to the nature of the disease, its resultant complications, restrictions on activities and high
costs of treatment, patients and their parents/caregivers may sometimes have to make trade-
offs between the type of hemophilia treatment received and the activities patients can
participate in. At the same time, they have to make a decision about whether and how much
they are willing to pay to reap the benefits of their child’s treatment and desired level of
activity. Willingness-to-pay for a commodity (in this case hemophilia treatment) indicates the
utility and satisfaction that an individual expects to gain from the particular treatment [8]. It
also serves as a reference for determining the burden of illness of the disease as perceived by
66
the population in question. Eliciting willingness-to-pay and preferences allows us to determine
the value of hemophilia treatment to patients or society, depending on the perspective taken.

This study aims firstly to elicit the relative preferences of society for the frequency of treatment
and acceptable frequency of bleeding episodes for a child with hemophilia, as well as the level
of physical activity that the child is able to participate in. Secondly, the study also aims to elicit
from the societal perspective, willingness-to-pay for a child to obtain hemophilia treatment that
would allow him to engage in his desired level of leisure activities and sport, while minimizing
adverse bleeding events.

Methods
In order to achieve the study aims, two experiments were conducted. Experiment 1 aims to
identify the relative preferences of study participants towards a set of hemophilia treatment
attributes, as well as their willingness-to-pay for improvements in such attributes, using a
discrete choice experiment (DCE). As DCEs are elaborate and time-intensive, an online
convenience sample of participants from the US population was used in this experiment. To
supplement the results obtained in experiment 1, experiment 2 was conducted in a
representative sample of the US population. Experiment 2 aimed to determine willingness-to-
pay for hemophilia treatment that would offer improvements on the attributes described in
experiment 1, and identify sociodemographic characteristics of the representative US
population that are likely to have greater willingness-to-pay for treatment.

67
Experiment 1
In Experiment 1, a DCE was conducted, which is a stated-choice survey method that simulates
choice behavior in respondents by requiring them to evaluate and make trade-offs between
attributes of hypothetical products [9]. As with many real-life situations, respondents have to
consider several different factors simultaneously, in order to make a decision based on the
“complete package”. DCEs have been used widely for healthcare applications including in
hemophilia patients [10, 11] and are particularly suited for use in this study as they allow for
the evaluation of preferences for hypothetical hemophilia treatment options which may not yet
be commercially available.

In order to obtain a US societal perspective, respondents in this study were obtained from the
non-hemophilia population by online convenience sampling. Before performing the DCE task,
respondents were presented with a brief description of hemophilia and its symptoms (Figure
4.1). To elicit preferences based on a parental perspective, respondents were asked to imagine
that they were the parents of a child or adolescent with hemophilia. In addition to the DCE task,
sociodemographic information of the respondents was also collected. The Institutional Review
Board of the University of Southern California approved the study protocol (HS-13-00371).

Development of DCE and Sample size calculation
In the DCE, respondents were presented with ten choice questions, each containing three
alternatives described by four attributes of treatment. For each choice question, respondents
were asked to select the alternative that they preferred. The four attributes of interest are
68
further described by levels as follows: Dosing Frequency (Every other day, Every 3 or 4 days,
Once weekly); Desired Level of Physical Activity (Low intensity, Non-contact sports, Contact
sports); Bleeding Frequency (Zero bleeds per month, 1 or 2 bleeds per month, 3 or more bleeds
per month); and Additional Monthly Insurance Premium ($0, $50, $100, $200, $300). The
attribute levels for “Dosing Frequency” were derived from current prophylaxis treatment
practices as well as treatment recommendations for forthcoming commercial treatment
options. Attribute levels for “Desired Level of Physical Activity” were derived from the
Paediatric Version of the Haemophilia Activities List (PedHAL), a hemophilia-specific tool that
measures self-perceived limitation of activities in daily life in children and adolescents with
hemophilia [12]. “Bleeding Frequency” and “Additional Monthly Insurance Premium” were
derived from data collected in the Hemophilia Utilization Group Study Part Va (HUGS Va) [13].
In HUGS Va, mean monthly bleeding frequency for pediatric subjects was 2.7±5.9 bleeds per
month, with a median of zero bleeds per month. Choice sets were generated using the %MKTEX
macro in SAS 9.3 (SAS Institute Inc., Cary, NC, USA) [14]. An example choice set is shown in
Figure 4.2.

To determine the minimum sample size (n) required, the number of choice questions per
respondent (t), maximum number of attribute levels (c) and number of alternatives in each
choice question (a) are required. In this study, c=5, a=3 and t=10. Using the equation
𝑛𝑡 𝑎 𝑐 ≥ 500
to solve for n, we require a minimum sample of 84 subjects [15]. To improve power, a sample
of 125 complete responses was targeted.
69
Participant Recruitment
In order to obtain a US societal perspective for evaluating willingness-to-pay, participants were
recruited via the Amazon Mechanical Turk (MTurk) web service. This is a crowdsourcing service
that enables researchers and businesses located in the US to use human intelligence to perform
tasks. Tasks may include surveys, audio transcribing or evaluating websites among others.
Workers browse among existing tasks and complete them for a small monetary
payment. MTurk has previously been used in discrete choice experiments, although not for
health-related scenarios [16]. The advantages of MTurk are that data can be obtained relatively
quickly and inexpensively, test-retest reliability are comparable to traditional survey methods,
and wages offered have no effect on the quality of work in survey tasks [17].

For this study, workers were restricted to those with internet protocol addresses within the US.
Payments for completing tasks were transferred to workers’ Amazon payment accounts. Each
approved response was paid $0.30 with an additional 10% service fee paid to Amazon.

Data Analysis
An individual’s utility U for a product alternative j can be described by the following utility
function
𝑈 = � 𝛽 𝑘 𝑥 𝑗𝑘
+ 𝛽 𝑝𝑟𝑖𝑐𝑒
𝑥 𝑝 𝑟 𝑖𝑐𝑒
+ 𝜀 𝑖𝑗 𝐾 𝑘 = 1

where utility is the sum of the products of the parameter estimates β for each attribute k and
the attribute level x for each attribute k and the error term ε
ij
[9].
70
Dummy variables were used to describe levels of “Dosing Frequency” (reference: “Every other
day”), “Desired Level of Physical Activity” (reference: “Low intensity”) and “Bleeding
Frequency” (reference: “Zero bleeds per month”). The parameter estimates β for the dummy
variables represent the marginal value of a change from the reference to the given level and
indicate the relative importance of each attribute level, with larger values of β reflecting more
preferred attributes. The parameter estimate β for the continuous variable “Price” indicates the
value of every $1 change in additional cost per month. A conditional logit model was used to
estimate the parameters of the utility function. All statistical analyses were carried out using
Stata 10.0 (StataCorp LP, College Station, TX, USA).

Willingness-to-pay for a change in an attribute/level is the marginal rate of substitution of a
particular attribute/level for money, here represented by additional insurance premium. The
willingness of an individual to pay for an attribute level compared to its reference level is the
ratio of the parameter estimate for that level to the price parameter estimate (ie. – 𝛽 𝑘 / 𝛽 𝑝𝑟𝑖𝑐𝑒 ).
The ratio obtained is thus the difference in utility of the attribute levels valued in dollars (utility
difference per dollar) [18]. In this way, estimating willingness-to-pay allows for the comparison
of attributes in common dollar units.

In order to estimate the 95% confidence intervals for the willingness-to-pay estimates, the
Krinsky Robb parametric bootstrap method with 10000 repetitions was used [19]. This method
assumes that the estimated coefficients are joint normally distributed but do not have to be
symmetrically distributed, and yields confidence intervals that are defined in all samples.
71
Experiment 2
A second experiment was conducted to supplement the results obtained in experiment 1, in
order to obtain an estimate of the median willingness to pay for hemophilia treatment in a
representative sample of the US general population. Respondents were provided with a short
description of hemophilia and its symptoms (Figure 4.3), and then posed a multiple choice
question to determine the range of additional insurance premium they were willing to pay per
month in order for their child to receive hemophilia treatment.

Participant Recruitment
The survey was administered by Knowledge Networks (KN), which provides a national omnibus
survey service (KN/Quickview
TM
) conducted online [20]. KN maintains a probability-based panel
of households that is statistically representative of the US population and continuously
maintained by using the United States Postal Service’s Delivery Sequence File. Households are
selected from a random-digit-dialling frame and panelists participate in surveys using their
home-based computer or interactive televisions provided by KN. KN/Quickview is administered
weekly to assigned adults and has a 60-70% completion rate.

Along with the responses to the questions posed in the experiment, KN/Quickview also
provides twenty demographic variables and panelist specific survey weights to generate results
that are representative of the US population.

72
Data Analysis
Logistic regressions were performed to identify characteristics of participants that would make
them (a) more likely to be willing to pay additional premium for treatment compared to those
who were not willing to pay any additional amount and (b) more likely to be willing to pay more
or less than the median amount compared to those who were willing to pay the median
amount. All statistical analyses were carried out using Stata 10.0 (StataCorp LP, College Station,
TX, USA).

Results
Experiment 1
Using the Amazon MTurk service, 137 participants were recruited of which 125 participants
who completed the entire survey were included in the analysis. Participants had a mean age of
32.7±12.5 years (Range: 18 – 73), were predominantly white (80.0%) and 57.6% were male
(Table 4.1).

Table 4.2 shows the results of the conditional logit model, with larger values indicating more
preferred features of treatment than smaller values. The parameter estimates appear to have
face validity, with less frequent dosing, more vigorous physical activity, less frequent
occurrence of bleeding and lower additional monthly insurance premium costs preferred.
Figure 4.4 provides a visual depiction of the relative preferences of each feature under
consideration, in which the parameter estimates were rescaled such that 10 denotes the most
preferred feature and zero the least preferred. The two most preferred features were to have
73
zero bleeding episodes per month and be able to participate in contact sports, while the least
preferred feature was the ability to participate in low intensity physical activity.

Table 4.3 presents the marginal willingness-to-pay estimates for features of the hypothetical
hemophilia treatment being evaluated. In considering the dosing frequency of the treatment,
participants were willing to pay $89.03 and $115.06 more per month to obtain a treatment that
would require administration only every 3 or 4 days, or require administration once weekly
respectively, instead of every other day. In order that their child with hemophilia would be able
to participate in either non-contact sports or contact sports instead of low intensity physical
activity, participants were willing to pay $111.18 and $224.81 more in monthly insurance
premium respectively. For a treatment that would be able to reduce frequency of bleeding
from one or twice a month to zero times a month, participants were willing to pay an estimated
$120.52 more a month in insurance premium. Correspondingly, participants were willing to pay
an estimated $208.64 more in monthly insurance premium in order that their child would be
able to receive a treatment that would result in zero bleeds a month instead of three or more
bleeds a month.

Experiment 2
The KN/Quickview survey was assigned to 2000 panelists of which 1056 (52.8%) responded. Of
the respondents, an additional 29 (2.8%) panelists were excluded from the analyses due to
incomplete responses. Those excluded were not statistically different from panelists in the
included sample.
74
After applying the survey weights, the representative sample had mean age of 47.0±0.7 years,
were predominantly white (66.0%) and 48.7% were male. Eighty-nine percent had at least a
high school education and 53.8% were either employed or self-employed (Table 4.1). The
median additional willingness-to-pay was in the range of $100 to $199, for treatment that
would allow a child with hemophilia to experience fewer bleeding episodes and yet be able to
participate in any physical activities he/she desires. An estimated 20.8% of the representative
sample was not willing to pay any additional amount per month, while 29.1% was willing to pay
$400 to $499 additional per month (Figure 4.5).

From the logistic regression comparing panelists who were not willing to pay any additional
amount against those who were willing to pay at least some additional amount per month, the
odds were 2.41 times as large that a person with a Bachelor’s degree or higher was willing to
pay at least some amount than the odds that a person with less than a high school education
would be willing to do the same (p=0.042). The odds of a person with greater than $100,000 in
annual household income being willing to pay at least some amount was 2.68 times as large
than the odds of those with less than $25,000 in annual household income (p=0.011). For
panelists with internet access at home, the odds of them being willing to pay at least some
amount was 1.71 times as large than the odds of those without internet access (p=0.047).
(Table 4.4)

Multinomial logistic regression estimated that the odds of panelists with household income
greater than $100,000 being willing to pay less than the median estimated willingness-to-pay of
75
between $100 and $199 were 0.35 times as low as those with household income under $25,000
(p=0.019). The odds of those with household internet access being willing to pay less than the
median estimated willingness-to-pay was 0.49 times as low as those without internet access
(p=0.044). (Table 4.5)

Discussion
This study aimed to elicit society’s willingness-to-pay for a child with hemophilia to obtain
treatment, and the relative importance of treatment attributes and outcomes of hemophilia
care. Of the three treatment attributes considered, bleeding frequency and level of activity
were identified to be the most important considerations. This suggests that for a population
somewhat unfamiliar with the disease, the primary concern is for patients to have access to
care that would prevent adverse events while allowing for freedom of activity. Dosing
frequency was considered the least important attribute of treatment in this study, supporting
results from previous DCEs where it was the second least important treatment attribute (out of
12 attributes) for hemophilia patients with inhibitors in the US [10] and was also not among the
top three concerns for both patients and providers alike (out of six attributes) in an Italian
hemophilia population [11].

In experiment 1 the highest monetary value was placed on being able to participate in contact
sports over low intensity activities. Reduction of hemophilia symptoms was also important,
with having zero bleeds assigned the second highest monetary value compared to having three
or more bleeding episodes per month. Despite the general population in this study not having a
76
deep understanding of the risks and complications of hemophilia, study participants were
nevertheless willing to trade-off increased monetary payments for hemophilia treatment that
could reduce symptoms and allow patients to lead a more active lifestyle that would prevent
further complications of the disease. In addition, the convenience afforded by less frequent
dosing was also preferred by participants, who were willing to pay additional premium for
treatment that would require their child to not have to receive factor infusion every other day,
as is the current practice for prophylaxis.

DCEs allow respondents to simultaneously consider various attributes of the treatment options
available, allowing for the quantification of their preferences towards the entire “treatment
bundle” as well as individual levels within each attribute. Although this study was conducted in
a non-patient population, the results have shown face validity and it has demonstrated the
feasibility of using DCEs to determine the preferences and willingness-to-pay of the general
population towards hemophilia treatment. The results obtained can be used to further refine
the attributes and levels included in the DCE, before the instrument is employed in a patient
population to determine preferences and willingness-to-pay from the patient’s perspective.

In experiment 2, in order to supplement the results from experiment 1, a median estimate is
obtained from a representative sample of the US population of willingness-to-pay for a
treatment that would improve outcomes and increase physical activity levels by unspecified
levels for children with hemophilia. The median estimate obtained is approximately of the same
order of magnitude as that obtained in experiment 1, supporting the results of the DCE.
77
However, a direct comparison of the results in both experiments is not possible as the
treatment scenarios in both experiments were not described similarly. In experiment 2, the
estimate obtained of $100 to $199 in additional health insurance premium per month
translates to an additional $1,200 to $2,388 in health insurance expenses per year. It is
unsurprising therefore, that those with greater household income had greater willingness-to-
pay. Those more highly educated and with internet access at home also had greater willingness-
to-pay. This is likely attributed to them having greater access to information regarding the
disease, its treatment and consequences which enable them to make more informed decisions
regarding the level of care required in hemophilia.

Clinical research in the past decades demonstrated the benefits of prophylactic factor
replacement therapy for severe hemophilia patients, as well as encouraged physical activity in
patients to reduce bleeding episodes and prevent joint deterioration. To our knowledge, this is
the first study that has attempted to elicit the preferences of the general population with
regards to their willingness-to-pay to provide costly hemophilia treatment for patients, so that
they may be able to achieve relatively normal levels of physical activity while minimizing
adverse events. This in turn would also allow patients to lead relatively normal, productive
lives. The estimates obtained show that from a societal perspective, people are willing to trade-
off monetary payments in order to obtain clinical and physical benefits for hemophilia patients,
especially among those with higher household incomes and the more highly educated or well-
informed.

78
These experiments were conducted in the general population, which may or may not have
included hemophilia patients or their caregivers. For policymakers and payers, this gives
valuable insight into how society values hemophilia treatment, which would be useful for
developing future policy and payment decisions. For patients and providers however, the
societal perspective might not wholly reflect the concerns of this population. It would be
interesting to repeat the experiments in a hemophilia patient population to determine if
patient preferences mirror that of society. Repeating the experiments in a patient population
would also serve to validate the survey instruments developed in this study.

79
Tables and Figures
Table 4.1: Sociodemographic characteristics
Experiment 1 Experiment 2
Weighted Proportions
(95% Confidence Interval)
N = 125 N = 1027
Mean Age (SD) 32.7 (12.5) 47.0 (0.66)
Gender (%)
Male
Female

72 (57.6)
53 (42.4)

48.7 (44.9 – 52.5)
51.3 (47.5 – 55.1)
Race (%)
White
Black
Hispanic
Other

100 (80.0)
6 (4.8)
4 (3.2)
15 (12.0)

66.0 (62.1 – 70.0)
11.5 (8.7 – 14.2)
14.7 (11.6 – 17.8)
7.8 (5.6 – 10.1)
Employment Status
Employed
Unemployed
Retired

-
-
-

53.8 (50.0 – 57.6)
27.8 (24.2 – 31.5)
18.4 (15.8 – 21.0)
Education (%)
Less than high school
High school
Some college
Bachelor’s degree or higher

3 (2.4)
12 (9.6)
48 (38.4)
62 (49.6)

11.3 (8.5 – 14.1)
30.1 (26.6 – 33.5)
28.8 (25.4 – 32.3)
29.8 (26.5 – 33.2)
Annual Income (%)
<$25,000
$25,000 – $49,999
$50,000 - $74,999
$75,000 - $99,999
≥$100,000

51 (40.8)
39 (31.2)
24 (19.2)
9 (7.2)
2 (1.6)

18.6 (15.5 – 21.7)
23.4 (20.2 – 26.6)
18.9 (15.9 – 21.8)
15.6 (12.8 – 18.5)
23.5 (20.4 – 26.6)

Table 4.2: Experiment 1 – Estimates from conditional logit model
Estimates Parameter Estimate Standard Error p-value
Dosing Frequency (reference: Every other day)
Every other day (calculated)
Every 3 or 4 days
Once weekly

-0.784
0.342
0.442

-
0.081
0.078

-
<0.0001
<0.0001
Level of Activity (reference: Low intensity)
Low intensity (calculated)
Non-contact sports
Contact sports

-1.291
0.427
0.864

-
0.087
0.083

-
<0.0001
<0.0001
Bleeding Frequency (reference: Zero bleeds)
Zero bleeds/month (calculated)
1 or 2 bleeds/month
3 or more bleeds/month

1.264
-0.463
-0.801

-
0.074
0.079

-
<0.0001
<0.0001
Cost -0.004 0.0004 <0.0001

80
Table 4.3: Experiment 1 – Willingness-to-pay estimates
Estimate 95% Confidence Interval
*
Dosing Frequency (reference: Every other day)
Every 3 or 4 days
Once a week

$89.03
$115.06

46.12 - 140.21
72.70 - 166.73
Activity Level (reference: Low intensity)
Non-contact sports
Contact sports

$111.18
$224.81

67.44 - 159.09
177.89 - 283.70
Bleeding Frequency (reference: Zero bleeds)
1 or 2 bleeds/month
3 or more bleeds/month

(-$120.52)
(-$208.64)

(-165.06) - (-81.75)
(-265.43) - ( -163.30)
*
95% confidence interval estimated using Krinsky Robb parametric bootstrap with 10,000 repetitions

Table 4.4: Experiment 2 – Logistic regression of willingness-to-pay $0 against some additional amount
Odds Ratio 95% Confidence Interval p-value
Age 0.99 0.98 – 1.01 0.424
Gender (reference: Male)
Female

1.08

0.71 – 1.63

0.730
Race (reference: White)
Black
Hispanic
Other

0.65
0.78
1.11

0.36 – 1.20
0.42 – 1.45
0.46 – 2.65

0.170
0.424
0.821
Education (reference: Less than high school)
High school
Some college
Bachelor’s degree or higher

1.11
1.42
2.41

0.53 – 2.33
0.65 – 3.07
1.03 – 5.65

0.781
0.377
0.042
Employment Status (reference: Employed)
Unemployed
Retired

0.70
0.85

0.41 – 1.21
0.45 – 1.60

0.202
0.613
Household Income (reference: <$25,000)
$25,000 - $49,999
$50,000 - $74,999
$75,000 - $99,999
≥$100,000

1.02
0.87
1.12
2.68

0.56 – 1.88
0.44 – 1.73
0.51 – 2.47
1.25 – 5.72

0.940
0.695
0.785
0.011
Household Internet Access (reference: No)
Yes

1.71

1.01 – 2.90

0.047

81
Table 4.5: Experiment 2 - Multinomial logistic regression of median willingness-to-pay against
willingness-to-pay below and above median amount
Multinomial
Odds Ratio
95% Confidence Interval p-value
Below Median
Age 1.01 0.99 – 1.02 0.363
Gender (reference: Male)
Female

1.04

0.66 – 1.65

0.868
Race (reference: White)
Black
Hispanic
Other

0.69
1.94
1.32

0.31 – 1.50
0.83 – 4.54
0.48 – 3.61

0.347
0.129
0.590
Education (reference: Less than high school)
High school
Some college
Bachelor’s degree or higher

0.80
0.71
0.62

0.31 – 2.05
0.28 – 1.79
0.24 – 1.60

0.644
0.465
0.323
Employment Status (reference: Employed)
Unemployed
Retired

1.11
0.79

0.60 – 2.06
0.40 – 1.57

0.731
0.504
Household Income (reference: <$25,000)
$25,000 - $49,999
$50,000 - $74,999
$75,000 - $99,999
≥$100,000

0.81
0.93
0.81
0.35

0.36 – 1.81
0.40 – 2.18
0.34 – 1.93
0.15 – 0.84

0.610
0.869
0.637
0.019
Household Internet Access (reference: No)
Yes

0.49

0.24 – 0.98

0.044
Above Median
Age 1.00 0.98 – 1.01 0.694
Gender (reference: Male)
Female

1.06

0.68 – 1.64

0.809
Race (reference: White)
Black
Hispanic
Other

0.44
2.70
1.46

0.20 – 1.64
1.22 – 5.98
0.53 – 4.03

0.052
0.015
0.467
Education (reference: Less than high school)
High school
Some college
Bachelor’s degree or higher

0.90
0.93
0.90

0.36 – 2.28
0.37 – 2.31
0.36 – 2.27

0.829
0.876
0.829
Employment Status (reference: Employed)
Unemployed
Retired

0.99
1.19

0.54 – 1.81
0.63 – 2.28

0.975
0.591
Household Income (reference: <$25,000)
$25,000 - $49,999
$50,000 - $74,999
$75,000 - $99,999
≥$100,000

0.89
1.07
1.20
1.26

0.39 – 2.00
0.46 – 2.49
0.50 – 2.86
0.54 – 2.97

0.772
0.879
0.682
0.589
Household Internet Access (reference: No)
Yes

0.70

0.35 – 1.42

0.324

82
Figure 4.1: Experiment 1 - Introduction to task

Introduction to Task
Regular exercise is encouraged for hemophilia patients to help build muscle strength, increase joint range of
motion, reduce joint pain and the frequency of joint bleeds. However, vigorous contact sports are generally
discouraged. This may lead to disappointment among patients and may affect their interactions with friends.
In this study, we want to determine parents’ willingness to pay for hemophilia treatment that will allow their child
to participate in their desired physical activity.
There are TEN questions in this survey, each with THREE scenarios. For each question, please select the scenario
that appeals to you most.

Figure 4.2: Experiment 1 – Example of discrete choice experiment question
Treatment A Treatment B Treatment C
Treatment type Home-infusion by injection
every 3 or 4 days
Home-infusion by injection
every other day
Home-infusion by
injection once a week
With this treatment, my
child should be able to
participate in…
Non-contact sports (such as
running, tennis and cycling)
Low intensity activities (such
as going out with friends
and playing outside alone or
with others)
Contact sports (such as
football, soccer, basketball
and karate)
With this treatment, it is
acceptable if my child
has at most…
3 or more bleeds per month 1 or 2 bleeds per month Zero bleeds per month
In order to receive this
treatment, I am willing
to pay an additional…
$50 per month for my
child's health insurance
$0 per month for my child's
health insurance
$300 per month for my
child's health insurance

If I had to choose one scenario, I would prefer…

83
Figure 4.3: Experiment 2 – Survey question

Imagine you have a child with a rare disease which causes frequent bleeding that over time may result in severe
joint damage. Due to the risk of bleeding, your child is unable to participate in vigorous contact sports including
football, soccer, basketball and hockey. In order to prevent bleeding, your child requires medication twice a week,
given by self-injection at home. This medication only helps to prevent bleeding and does not cure the disease.

How much would you be willing to pay for a treatment that would allow your child to have fewer bleeding
episodes and yet be able to participate in any physical activities he/she desires, including vigorous contact sports?
Please indicate the amount that you would be willing to pay per month in addition to your current insurance
premium so that your child could participate in these activities:

I am not willing to pay any additional money (1)
$1-$99 per month (2)
$100-$199 per month (3)
$200-$299 per month (4)
$300-$399 per month (5)
$400-$499 per month (6)

Figure 4.4: Experiment 1 – Relative preferences for features of treatment

Note: Figure reflects estimated coefficients as described in Table 4.2 excluding the cost coefficient, rescaled from 0 (least
preferred) to 10 (most preferred)

84
Figure 4.5: Experiment 2 – Distribution and cumulative distribution of willingness-to-pay responses

85
Chapter References
1. Choiniere M, Melzack R. Acute and chronic pain in hemophilia. Pain 1987; 31: 317-31.
2. Ljung R. Prophylactic therapy in haemophilia. Blood Rev 2009; 23: 267-74.
3. Weigel N, Carlson BR. Physical activity and the hemophiliac: yes or no? American corrective
therapy journal 1975; 29: 197-205.
4. Buzzard BM. Sports and hemophilia: antagonist or protagonist. Clin Orthop Relat Res 1996: 25-
30.
5. Rice SG, American Academy of Pediatrics Council on Sports M, Fitness. Medical conditions
affecting sports participation. Pediatrics 2008; 121: 841-8.
6. Ross C, Goldenberg NA, Hund D, Manco-Johnson MJ. Athletic participation in severe hemophilia:
bleeding and joint outcomes in children on prophylaxis. Pediatrics 2009; 124: 1267-72.
7. Von Mackensen S. Quality of life and sports activities in patients with haemophilia. Haemophilia
2007; 13 Suppl 2: 38-43.
8. Fayers PM, Machin D. Quality of life the assessment, analysis, and interpretation of patient-
reported outcomes. 2nd ed. Chichester ; Hoboken, NJ: J. Wiley,, 2007: p. cm.
9. Louviere JJ, Hensher DA, Swait JD. Stated choice methods : analysis and applications. Cambridge,
U.K. ; New York: Cambridge University Press, 2000.
10. Brown TM, Pashos CL, Joshi AV, Lee WC. The perspective of patients with haemophilia with
inhibitors and their care givers: preferences for treatment characteristics. Haemophilia 2011; 17: 476-
82.
11. Mantovani LG, Monzini MS, Mannucci PM, et al. Differences between patients', physicians' and
pharmacists' preferences for treatment products in haemophilia: a discrete choice experiment.
Haemophilia 2005; 11: 589-97.
12. Groen WG, van der Net J, Helders PJ, Fischer K. Development and preliminary testing of a
Paediatric Version of the Haemophilia Activities List (pedhal). Haemophilia 2010; 16: 281-9.
13. Zhou ZY, Wu J, Baker J, et al. Haemophilia utilization group study - Part Va (HUGS Va): design,
methods and baseline data. Haemophilia 2011; 17: 729-36.
14. Kuhfeld W. Marketing research methods in SAS. Cary, NC, USA: SAS Institute Inc., 2010.
15. Orme B. Getting Started with Conjoint Analysis: Strategies for Product Design and Pricing
Research. Madison, WI: Research Publishers, LLC, 2005.
16. Hoffenson S, Frischknecht BD, Papalambros PY. A market systems analysis of the U.S. Sport
Utility Vehicle market considering frontal crash safety technology and policy. Accident; analysis and
prevention 2013; 50: 943-54.
17. Buhrmester M, Kwang T, Gosling SD. Amazon's Mechanical Turk: A New Source of Inexpensive,
Yet High-Quality, Data? Perspect Psychol Sci 2011; 6: 3-5.
18. Louviere JJ, Towhidul I. A comparison of importance weights and willingness-to-pay measures
derived from choice-based conjoint, constant sum scales and best-worst scaling. Journal of Business
Research 2008; 61: 903-11.
19. Hole AR. A comparison of approaches to estimating confidence intervals for willingness to pay
measures. Health economics 2007; 16: 827-40.
20. GfK Knowledge Networks. KN/QuickviewTM Specs. GfK Custom Research North America, 2013.

86
Chapter 5: Modelling Lost Productivity Costs: A Comparison of
Treatment Types in Severe Hemophilia A

Abstract
Background: Childhood chronic conditions have the potential to incur high lost productivity
costs throughout a patient’s lifetime. Effective therapeutic management of the disease can
reduce productivity losses and burden of illness for patients. Objectives: To construct a decision
model comparing treatment types, in order to minimize productivity losses in chronic diseases,
using pediatric severe hemophilia A patients as example and applying the resultant model to a
pediatric asthma population. Methods: Two influence diagrams were developed, comparing
prophylactic and on-demand factor replacement therapy to determine their effect on cost of
productivity losses in severe hemophilia A patients. Conditional probabilities for model inputs
were derived from the Hemophilia Utilization Group Study Part-Va and included hemophilia-
related annual bleeding episodes, emergency room visits, hospitalizations and days missed
from work/school. Costs considered included cost due to productivity losses and caregivers’
cost for the worker population. The resultant model was also applied to a pediatric asthma
population using publicly available data from the National Health Interview Survey 2011.
Results: Cost of productivity losses due to prophylaxis varied between $42.88 and $59.39, and
that for on-demand varied between $48.20 and $66.76 depending on the model developed.
Productivity losses due to on-demand therapy were consistently greater than those due to
prophylaxis, by between 3.9% to 19.2%. Modelling the asthma population yielded estimates of
a similar magnitude, with productivity losses in those on maintenance therapy estimated at
approximately $62, and $59 for those not receiving maintenance therapy. Conclusions: The
87
influence diagram developed was found to be applicable to the parental worker population for
both pediatric severe hemophilia A and pediatric asthma. Differences in lost productivity costs
between treatment types in both diseases were found to be small. In hemophilia, prophylaxis
therapy yielded smaller productivity losses, reducing burden of illness for patients, compared to
on-demand therapy. This model may be useful in guiding treatment decisions based on each
treatment’s impact on burden of illness for patients and their caregivers. However, further
validation of the model in other pediatric chronic diseases with management strategies similar
to hemophilia is required.

88
Introduction
Productivity losses due to lost or impaired ability to work or engage in leisure activities as a
result of disease-related morbidity, and costs associated with lost economic productivity in
cases of mortality due to disease contribute to the cost of a disease as indirect costs [1].
Quantifying indirect costs due to a disease involve the computation of non-medical economic
losses such as days missed from work or school, caregivers’ cost, premature retirement and
death. Incurring high indirect costs due to disease can be a burden for patients and their
families due to wage losses but at the same time, is also a burden for employers and society
due to losses in economic productivity.

Chronic conditions, particularly those that develop from childhood, have the potential to incur
high lost productivity costs throughout a patient’s lifetime. This includes caregivers’ cost and
parental lost wages in childhood, and actual and potential future wage losses for the patient
when they transition to adulthood, due to missing work or school days. This contributes to the
total burden of disease for patients.

Effective management of the disease could reduce the number of missing work or school days
experienced by patients, reducing costs due to productivity losses. This is especially true in
chronic diseases where with proper management, patients are able to lead functional,
productive lives. To evaluate the impact of different treatment options on lost economic
productivity, it is important to determine each treatment’s influence on the relationship
between disease symptom manifestations, healthcare service utilization and indirect costs due
89
to missing work or school days. As an example, a population of hemophilia A patients is
considered.

Hemophilia management focuses on timely and adequate administration of factor concentrates
to minimize the effects of bleeding. In severe hemophilia patients, patients are encouraged to
use prophylactic factor replacement therapy, the regular infusion of factor concentrates in
order to prevent bleeding episodes and their subsequent consequences, including joint damage
[2]. In contrast, on-demand treatment is the practice of infusing factor episodically to treat an
ongoing, active bleed.

The objective of this study was to develop an influence diagram that could compare treatment
types for children, to minimize the cost of productivity losses in chronic diseases in the worker
population, as a result of disease symptom manifestations and healthcare service utilization,
thus minimizing burden of illness for patients. As an example, data from a study cohort of
severe hemophilia A pediatric patients was employed to compare prophylactic factor
replacement therapy to on-demand therapy. The model was then populated with data from a
pediatric asthma population to demonstrate its applicability to other disease states.

Methods
Influence Diagrams
Influence diagrams are graphical representations of complex decision-making models that
describe the probabilistic dependencies between variables and decisions based on uncertain
90
information, which can then be solved [3]. Using an influence diagram helps to frame the steps
involved in the decision process allowing decision-makers, whether patients or healthcare
providers, to make informed choices. The decision pathway of an influence diagram consists of
an acyclic and directed network of nodes and arcs. There are three types of nodes, each
associated with a variable: decision nodes, chance nodes and a value node, which are
commonly represented by squares, circles and diamonds, respectively. Influences or
relationships between nodes are represented by connecting arcs. There are two types of arcs:
informational arcs which lead into decision nodes and conditional arcs which lead into chance
and value nodes [4, 5].

Figure 5.1 illustrates the basic structure of an influence diagram. Decision node a represents
the decision to be made by the decision-maker. In this context, the decision-maker could be the
patient, healthcare provider, payer or policymaker. Chance nodes b, c, d, e and f, represent
random variables, each associated with a set of conditional probabilities derived from the
conditional probability distributions of its direct predecessor node(s) (eg. e from b; f from c and
d) which lead into (influences) the chance node through conditional arcs [eg. (b,e), (c,f) and
(d,f)]. The value node g represents the objective to be maximized in expectation [3-5].

Hemophilia Model
In order to explore the relationship between symptom manifestations (acute events),
healthcare service utilization and lost productivity costs due to missing work or school days, the
following model as shown in Figure 5.2 was developed. This general model was developed and
91
populated using data from a hemophilia A population, but may be applicable to chronic disease
states which develop in childhood and continue throughout the patient’s life.

The model begins at the “Treatment” decision node where two treatments are considered. The
type of treatment chosen can influence the probability of a disease-related clinical event
occurring, as depicted by the “Event” node. This in turn may lead to a visit to the emergency
room (“ER visit” node) or lead to the patient missing school (“Child missed days” node). The
emergency room visit may also result in a hospitalization (“Hospitalization” node) which in itself
also leads to the patient missing school. Parents too may incur missed work/school days due to
their child’s illness (“Parent missed days” node). Children who have to stay home due to illness
may also require other services, including the help of a caregiver (“Others” node). All this will
lead to an incurred indirect cost (“Cost” node). The objective of this model is then to minimize
this cost in expectation, in order to compare the two treatment types.

As an example, this study will use data from the Hemophilia Utilization Group Study Part Va
(HUGS Va) to populate the model. The treatments to be compared in this example are
prophylactic and on-demand factor replacement therapy in severe hemophilia A pediatric
patients. The event of interest is the frequency of bleeding episodes experienced by the patient
in a one year period. The objective is to minimize the costs incurred due to missing work or
school days in a year due to hemophilia in parents of children with hemophilia A. For
constructing the model, potential productivity losses of pediatric participants aged between 2
and 17 years old will be attributed to the wage loss for the parent of the child missing work due
92
to their child’s hemophilia. Additional costs to be considered include the cost of hiring
caregivers.

Hemophilia Data
Data for populating the influence diagram were obtained from HUGS Va, a 2-year, multicenter
observational cohort study conducted among six Hemophilia Treatment Centers (HTCs) located
in geographically diverse regions of the US. Details about the study methods have been
previously reported [6]. Data were collected at initial interview from parent proxy-report and
patient chart reviews completed by healthcare providers, and included clinical information and
HrQoL outcomes. Periodic participant follow-up survey provided data on time lost from work,
school or usual activities, disability days, healthcare utilization and outcomes including bleeding
episodes, joint pain and motion limitation. Follow-up interviews were administered monthly in
the first year and semi-annually in the second year. More specifically, participants were asked
at each follow-up the number of bleeding episodes, emergency room visits and hospitalizations,
and the number of missed days from work or school for both child and parent, due to
hemophilia A they had experienced since their last follow-up. Information about the need for
caregivers since the last follow-up was also collected.

Study participants were enrolled at each HTC in accordance with study enrolment criteria
following informed consent from parents of minor children. This model will be developed using
data provided by 75 children with severe hemophilia A who have complete follow-up data out
93
of a total of 165 children (all severities) with factor VIII deficiency who were recruited into the
HUGS Va study between July 2005 and July 2007.

The University of Southern California (USC) served as the data and coordinating center, and the
study protocol was approved by the Institutional Review Board of USC (IRB number: HS-
046012) and that of each participating HTC.

Model Inputs
For this study, two separate hemophilia models were constructed, both comparing the
treatment arms of “Prophylaxis” and “On-demand” factor replacement therapy. The first
(Model A) considered bleeding events to be a count variable and the conditional probabilities
for the expected frequency of bleeding events were derived using Poisson regression. The
second model (Model B) considered bleeding events to be a binary variable and logistic
regression derived the conditional probabilities for at least one bleeding event occurring. In
both models, the conditional probabilities for the occurrence of ER visits, hospitalizations and
need for caregivers were derived using logistic regression. Conditional probabilities for missed
days of work or school were derived using Poisson regression.

To attribute the cost of compensation lost due to missing work, cost of hourly compensation
lost was assumed to be the mean total compensation cost for civilian workers of $31.09/hour
as published by the US Department of Labor’s Bureau of Labor Statistics (BLS)[7]. As an
additional analysis, cost of compensation was alternatively calculated by categorizing parents
94
into one of eight occupational groups according to their declared occupation, derived in
accordance with the BLS’s classification of occupational and industry groups, with the
compensation imputed for each group derived from the hourly employer cost for employee
compensation for each group. These compensation rates ranged from $23.27/hour to
$51.36/hour [7]. The mean hourly cost of hiring caregivers was derived from the caregivers’
compensation cost as reported by the BLS. Employment rates were obtained from the HUGS Va
data as well as from national employment rates provided by the BLS [8].

Asthma Data
To demonstrate the applicability of the influence diagram to other chronic diseases, Model B
was populated using publicly available data from pediatric asthma patients collected as part of
the 2011 National Health Interview Survey (NHIS). The two treatment arms being compared are
“Treatment” and “No treatment” with prescription asthma controller medication. The NHIS is a
cross-sectional household interview survey that collects information on the health of the
civilian non-institutionalized population of the US. From each sample household, a sample child
is randomly selected and basic information on health status, healthcare services and health
behavior is collected [9]. The NHIS sample child public use file includes questions identifying
whether a child has asthma, had experienced an asthma episode in the past year, and had
visited an ER due to asthma in the past year. Data on hospitalizations, prescription medication
use, missed days of school and need for caregivers are also available. As with the HUGS Va data,
conditional probabilities were derived using Poisson or logistic regressions. Cost of hourly
compensation lost was assumed to be the mean total compensation cost for civilian workers of
95
$31.09/hour as published by the BLS. Employment rates were estimated from the NHIS data as
well as from national employment rates. The 2011 NHIS sample comprised of 1241 children
with active asthma.

All statistical analyses for the derivation of model inputs were carried out using SAS version 9.3
statistical software (SAS Institute, Cary, NC). The influence diagram was developed using GeNie
2.0 (Graphical Network Interface), a freeware provided by the Decision Systems Laboratory at
the University of Pittsburgh.

Results
The mean age of HUGS Va participants included in the model was 9.3±4.8 years, and 74.7% of
participants’ parents were employed. Of the 75 participants, 86.7% were on prophylactic factor
replacement therapy. Sixty-three participants (84.0%) experienced at least one bleeding
episode, 21 (28.0%) visited the ER at least once, 8 (10.7%) were hospitalized due to hemophilia
and 27 (36.0%) required the services of a nurse or paid caregiver in a 12 month period.
Participants experienced a mean of 5.6±7.7 bleeding episodes annually and missed school an
average of 5.6±11.9 days per year, with their parents missing work an average of 3.1±9.8 days
per year due to their child’s hemophilia. (Table 5.1)

From Model A, using BLS’s mean total compensation cost, the expected cost of productivity
losses for the study population if on-demand therapy was chosen would be $3,828 per year, or
$51.04/patient/year. In comparison, the expected productivity losses if prophylaxis was chosen
96
would be $3,216/year ($42.88/patient/year), a 16.0% reduction in losses from on-demand
therapy. Using the mean compensation cost based on parental occupational category, choosing
on-demand therapy resulted in expected productivity losses of $4,866/year
($64.88/patient/year), compared to $3,947/year ($52.61/patient/year) if prophylaxis were
chosen, a difference of 18.9%. (Table 5.2)

In Model B, where bleeding events were categorized as a binary variable instead of actual
number of bleeds experienced, using BLS’s mean total compensation cost estimated the
expected cost of productivity losses to be $3,615/year ($48.20/patient/year) and $3,475/year
($46.33/patient/year) for on-demand and prophylaxis therapy respectively, a difference of
3.9%. Using mean compensation cost based on parental occupation, the respective estimates
for on-demand and prophylaxis were $4,545/year ($60.60/patient/year) and $4,336/year
($57.81/patient/year), a difference of 4.6%. (Table 5.2) For both Models A and B, estimating the
model using national employment rates instead of that of the study population did not change
the percentage differences in estimates between on-demand and prophylaxis treatment.

The 2011 NHIS data included 1241 children with asthma, with mean age 9.7±4.8 years, and
81.3% of parents of patients were employed. Of the study population, 567 (45.7%) had been on
prescription medication for their condition in the past 12 months. Fifty-three percent (n=655)
had experienced at least one asthma episode, 250 (20.2%) visited the ER due to asthma, 409
(33.0%) had been hospitalized for any cause and 19 (1.5%) had required a nurse or paid
caregiver at least once, in the past 12 months. Mean missed days of school was 4.6±10.2 days
97
per year and in the absence of additional information from the NHIS, mean missed days of work
for parents was assumed to be the same as that of their child. (Table 5.1)

Applying the NHIS data to Model B, the expected cost of productivity losses if patients chose to
be on prescription medication was $76,816 per year for the study population, or $61.90 per
patient per year. Expected cost of productivity losses if patients chose not to be on prescription
medication was $73,649/year ($59.35/patient/year), which is a 4.3% reduction in productivity
losses compared to those on treatment with prescription medication. (Table 5.3)

Discussion
This study has shown that in hemophilia A, expected productivity losses for parents of children
with hemophilia are greater for those practicing on-demand factor replacement therapy
compared to prophylactic factor replacement therapy. This supports the use of prophylaxis in
severe hemophilia A, to reduce burden of illness and indirect costs due to lost productivity for
patients. However, for both treatment types, the expected value of annual productivity losses
per patient due to hemophilia was small. This is regardless of whether losses were estimated by
considering the self-reported employment status of study participants or using national
employment rates. Using both these methods of determining employment status yielded very
similar estimates.

In both Models A and B, the expected value of productivity losses when calculated using mean
compensation rates were consistently lower than that calculated using parental occupational
98
categories, with the difference varying between 19% to 22% depending on the model and
treatment type. However, both methods of calculation utilized national estimates of
compensation rates and more exact estimates would require data in which the actual incomes
of participants are reported. In the absence of such data however, the estimates obtained in
this study provide us with a range of values of expected productivity losses to guide our
decision making.

In Model A, whether a bleeding event occurred in a one year period was described by a binary
variable (Yes/No) whereas in Model B, the actual number of bleeding events was used to derive
the conditional probabilities that populate the model. By collapsing count data into a binary
variable, the complexity of the information about how treatment type influences outcomes
(bleeds) and losses in productivity are lost. This is reflected in the percentage differences in
expected productivity losses between prophylaxis and on-demand treatment, which varies by a
greater extent in Model A (16% – 19%) than in Model B (4% - 5%). Collapsing the data
underestimates the differences between treatment types and where possible, should be
avoided.

Due to the nature of the NHIS data available, Model B was used to obtain the estimates of
expected productivity losses due to asthma. As with the hemophilia models, the expected value
of productivity loss due to either treatment arm was small. In a Finnish study from 2001,
calculated productivity losses of parents/caregivers for children with asthma was also
determined to be small, estimated at $31.58/child/year for children receiving maintenance
99
therapy with inhaled corticosteroids and $34.52/child/year for children not receiving
maintenance therapy [10]. Differences in the exact values of productivity losses between the
two studies could be attributable to differences between the Finnish and US populations,
inflation and assumptions made with regards to missed days of work for parents in the NHIS
population. However, the similarity of the magnitude of the expected values of productivity
losses between this study and the Finnish study does serve to demonstrate the applicability of
the influence diagram developed. For the current study, it is also interesting to note that
contrary to expectations, productivity losses were slightly greater for patients who were
receiving prescription medication for their condition, than those who were not. A possible
explanation for this could be that as discussed previously, the unavailability of information
regarding the exact frequency of occurrence of asthma episodes resulted in a loss of
information for populating the model. In addition, assumptions that were made regarding
hospitalizations and frequency of parental missed days could also have confounded the
estimates.

In this study, an influence diagram was developed to evaluate the effect of a treatment decision
and its resultant outcomes on productivity losses. The model describes a complex series of
events both graphically and mathematically, helping decision makers visualize the problem at
hand. The estimates of expected productivity losses obtained also help to guide the decision-
making process for patients and their parents/caregivers. In this study, this model has shown to
be applicable to both a hemophilia and asthma population. However, the small sample size
available from the hemophilia data and the nature of the NHIS asthma data makes it difficult to
100
validate the model and determine the robustness of the estimates obtained. Further validation
of the model is required using larger sample sizes and data with more well-defined frequency of
events. When validated, the model can be used to compare new or existing treatment types
and their impact on productivity losses for patients and their caregivers in other chronic disease
states besides those described here.
101
Tables and Figures
Figure 5.1: Illustration of Influence Diagram

[17] Adapted from Howard RA, Matheson JE. Influence Diagrams. Decision Analysis 2005; 2: 127-43.

Figure 5.2: Study Model

a
c
f
d
e b
g
(a,b)
(a,d)
(e,g)
(a,c)
(b,e)
(d,f)
(f,g)
(c,f)
102
Table 5.1: Characteristics of model population by disease group
Characteristics
Hemophilia
N=75
Asthma
N=1241
Mean age (SD) 9.3 (4.8) 9.7 (4.8)
Parental Employment Status (%)
Full-time employed
Part-time employed
Unemployed
Student
32 (42.7)
24 (32.0)
18 (24.0)
1 (1.3)
866 (69.8)
143 (11.5)
232 (18.7)
-
Treatment Type (%)
Prophylaxis/ Taking prescription medication
On-demand/Not taking prescription medication

65 (86.7)
10 (13.3)

567 (45.7)
674 (54.3)
Mean annual bleeding episodes (SD, Range)
5.6
(7.7, 0 – 49.0)
-*
Acute event(s) in 12 months (%) 63 (84.0) 655 (52.8)
ER visit(s) in 12 months (%) 21 (28.0) 250 (20.2)
Hospitalization(s) in 12 months (%) 8 (10.7) 409 (33.0)**
Mean annual missed days of school
(SD, Range)
5.6
(11.9, 0 – 83.6)
4.6
(10.2, 0 – 240)
Mean annual parental missed days of work
(SD, Range)
3.1
(9.8, 0 – 76.5)
4.6***
(10.2, 0 – 240)
Caregiver/Nurse needed in 12 months (%) 27 (36.0) 19 (1.5)
*Data unavailable
**All-cause hospitalizations
***Data unavailable; Assumption: Parental missed days is equal to child missed days

103
Table 5.2: Results from hemophilia models

Prophylaxis On-Demand
Percentage change
(Prophylaxis to On-
demand)
Study Population
(US$/year)
Per Patient
(US$/patient/year)
Study Population
(US$/year)
Per Patient
(US$/patient/year)
Model A
Study employment rate
Mean compensation rate*
Parental occupation**

3216
3947

42.88
52.63

3828
4866

51.04
64.88

+16.0
+18.9
National employment
rate
Mean compensation rate*
Parental occupation**

3278
4045

43.71
53.93

3918
5007

52.24
66.76

+16.3
+19.2
Model B
Study employment rate
Mean compensation rate*
Parental occupation**

3475
4336

46.33
57.81

3615
4545

48.20
60.60

+3.9
+4.6
National employment
rate
Mean compensation rate*
Parental occupation**

3549
4454

47.32
59.39

3696
4672

49.28
62.29

+4.0
+4.7
*Mean compensation rate: Assume mean total compensation cost for civilian workers of $31.09/hour [7]
**Parental occupation: Compensation rate ranges between $23.37/hour to $51.26/hour depending on parent’s occupation [7]

104
Table 5.3: Results from asthma model

Prescription medication No prescription medication
Percentage
change
(Prescription
to none)
Study Population
(US$/year)
Per Patient
(US$/patient/year)
Study Population
(US$/year)
Per Patient
(US$/patient/year)
Study employment rate
Mean compensation rate*

76816

61.90

73649

59.35

(-4.3)
National employment
rate
Mean compensation rate*

77115

62.14

73946

59.59

(-4.3)
*Mean compensation rate: Assume mean total compensation cost for civilian workers of $31.09/hour [7]
105
Chapter References
1. Gold MR. Cost-effectiveness in health and medicine. New York: Oxford University Press, 1996.
2. Ljung R. Prophylactic therapy in haemophilia. Blood Rev 2009; 23: 267-74.
3. Bielza C, Gómez M, Shenoy PP. A review of representation issues and modeling challenges with
influence diagrams. Omega 2011; 39: 227-41.
4. Howard RA, Matheson JE. Influence Diagrams. Decision Analysis 2005; 2: 127-43.
5. Shachter RD. Evaluating Influence Diagrams. Oper Res 1986; 34: 871-82.
6. Zhou ZY, Wu J, Baker J, et al. Haemophilia utilization group study - Part Va (HUGS Va): design,
methods and baseline data. Haemophilia 2011; 17: 729-36.
7. Bureau of Labor Statistics UDoL. Employer Costs for Employee Compensation - June 2013. 2013.
8. Bureau of Labor Statistics UDoL. 2013 Employment Situation. 2013.
9. CDC/National Center for Health Statistics. About the National Health Interview Survey. 2012.
10. Korhonen K, Reijonen TM, Remes K, Malmstrom K, Klaukka T, Korppi M. Reasons for and costs of
hospitalization for pediatric asthma: a prospective 1-year follow-up in a population-based setting.
Pediatric allergy and immunology : official publication of the European Society of Pediatric Allergy and
Immunology 2001; 12: 331-8.

106
Chapter 6: Conclusion

The burden of illness of a disease, particularly a chronic disease, involves not just the economic
burden of direct medical costs, but also indirect economic burdens that may result due to
disease-related lost productivity and caregivers’ costs. In addition, it also involves
unquantifiable costs due to the physical and psychological burden imposed on the patient for
having to cope with the effects of their disease. This dissertation focused on examining the
burden of illness from the patients’ perspective, focusing on the physical and psychological
burden of chronic disease for patients, eliciting the value patients and their caregivers place on
treatment and costs due to productivity losses, using a rare and chronic disease, hemophilia A,
as an example.

Through these studies, we found that although the main impact on HrQoL for hemophilia
patients is on their physical functioning and not so on psychosocial functioning, except among
those with severe disease, hemophilia patients have HrQoL that is comparable to the healthy
US population. HrQoL was also found to be significantly impacted by the frequency of missing
additional work/school days and experiencing additional bleeds relative to the study population
and to the individual’s experience of acute events across time. Although the impact of each
event on HrQoL is small, frequent occurrence of acute events could have a large cumulative
impact on reducing the HrQoL of patients. The multivariate multilevel model employed in
chapter 3 to identify the impact of acute events on HrQoL could have potential applicability in
other chronic diseases with disease characteristics similar to that of hemophilia.
107
By eliciting the preferences and willingness-to-pay of society for hemophilia treatment, we
showed that individuals from households with higher incomes and who were more highly
educated and well-informed were more willing to trade-off monetary payments for clinical and
physical benefits for hemophilia patients. The feasibility and face validity of using discrete
choice experiments for eliciting preferences was also demonstrated.

Finally, in attempting to determine differences in lost productivity costs between hemophilia
treatment types for a pediatric population with working parents, the influence diagram
developed in chapter 5 found only small productivity losses for both prophylaxis and on-
demand treatment, with smaller productivity losses attributable to prophylaxis treatment.
Applying the model to an asthma population also found small productivity losses and small
differences between treatment types. This model may be useful in guiding treatment decisions
based on each treatment’s impact on burden of illness for patients and their caregivers.

The results of these studies provide insight for stakeholders involved in hemophilia care,
including patients, payers and healthcare providers, into the needs of and burden of disease on
the hemophilia population. This would guide disease management at both the individual
patient and disease-population levels. At the same time, the methods employed, such as
multivariate multilevel modeling, discrete choice experiments and the use of influence
diagrams may find broader applicability in research for other chronic, rare diseases.

Abstract (if available)

Abstract "Burden of illness" is the negative impacts that an illness has on patients and their families, and may include physical and psychological functioning, social stigma and economic burden. Burden of illness is a constant concern for patients and their families, especially if the disease is rare and also of a lifelong, chronic nature. This is despite the fact that, for many rare and chronic diseases, affected individuals are able to lead relatively normal, productive lives if diagnosed early and provided with appropriate disease management. ❧ In order to help stakeholders identify and address the needs of patients with one such rare and chronic disease, hemophilia A, this dissertation examines the burden of illness from the patients' perspective, in terms of its impact on patients' health-related quality of life (HrQoL) and economic burden. In the first study (chapter 2), HrQoL of patients with hemophilia A and how it relates to the physical manifestations of the disease, such as joint pain and motion limitation is described. The second study (chapter 3) examines the causes of fluctuations in HrQoL over time, both within and between individuals with a stable, chronic disease. The third study (chapter 4) describes a discrete choice experiment conducted to determine societal preferences and willingness-to-pay for providing hemophilia treatment for pediatric patients in order that they may have improved outcomes. Finally, in the fourth study (chapter 5), a decision model is developed, comparing two treatment types to minimize the productivity losses for caregivers of pediatric patients due to missing work or school because of their chronic disease, using severe hemophilia A as an example. The methods employed here may also have broader applications in research for other chronic, rare diseases.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Three essays on emerging issues in hemophilia care

PDF

Discriminating changes in health using patient-reported outcomes

PDF

Understanding the value of hemophilia therapies

PDF

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

PDF

Economic aspects of obesity

PDF

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

PDF

Value in health in the era of vertical integration

PDF

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

PDF

Essays in pharmaceutical and health economics

PDF

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

PDF

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

PDF

Healthcare provider recommendations: a panacea to improving influenza vaccination rates?

PDF

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

PDF

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

PDF

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

PDF

Health-related quality of life correlates of the treatment decision making process of newly diagnosed prostate cancer patients

PDF

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

PDF

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

PDF

Essays on the economics of infectious diseases

Asset Metadata

Creator Poon, Jiat Ling (author)

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

School School of Pharmacy

Degree Doctor of Philosophy

Degree Program Pharmaceutical Economics and Policy

Publication Date 10/29/2013

Defense Date 10/02/2013

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

Tag burden of illness,health related quality of life,hemophilia A,indirect costs,OAI-PMH Harvest,patient reported outcomes,willingness-to-pay

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Doctor, Jason N. (committee chair), Nichol, Michael B. (committee member), Sood, Neeraj (committee member)

Creator Email jiatling@gmail.com,jiatlinp@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-340757

Unique identifier UC11295405

Identifier etd-PoonJiatLi-2115.pdf (filename),usctheses-c3-340757 (legacy record id)

Legacy Identifier etd-PoonJiatLi-2115.pdf

Dmrecord 340757

Document Type Dissertation

Format application/pdf (imt)

Rights Poon, Jiat Ling

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