3 Cystic Fibrosis Clinic, University of Alberta Hospital, Edmonton, Alberta, Canada

Transcription

1 HEALTH-RELATED QUALITY OF LIFE FOR ADULT AND PEDIATRIC CYSTIC FIBROSIS PATIENTS: A REGRESSION APPROACH TO ASSESSING THE IMPACT OF RECOMBINANT HUMAN DNASE Jeffrey A. Johnson 1,2, Melinda Connolly 1, Peter Zuberbuhler 3, Neil Brown 3,4, Joan Tabak 3 Working Paper Institute of Health Economics, Edmonton, Alberta, Canada 2 Faculty of Pharmacy & Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada 3 Cystic Fibrosis Clinic, University of Alberta Hospital, Edmonton, Alberta, Canada 4 Division of General Internal Medicine, Faculty of Medicine, University of Alberta, Edmonton, Alberta, Canada Legal Deposit 1999 National Library Canada ISSN

2 ACKNOWLEDGEMENT Supported by a grant from the ACCP Research Institute (1997 ACCP-Pharmacia & Upjohn Pharmacoeconomics Research Award) and by the Institute of Health Economics. Dr. Johnson is a Population Health Investigator with the Alberta Heritage Foundation for Medical Research (AHFMR).

3 TABLE OF CONTENTS LIST OF TABLES...iv LIST OF FIGURES...v ABSTRACT...1 Objectives...1 Methods...1 Results...1 Conclusions...2 I. INTRODUCTION...3 II. BACKGROUND Epidemiology of CF HRQOL in Cystic Fibrosis...5 III. METHOD Patients Surveys Measures...11 a. Adult...11 b. Pediatric...12 c. Demographics Clinical Data Analysis...14 IV. RESULTS Pediatric CF Patients...15 a. Pediatric Baseline Respondent Sample...15 b. Pediatric Baseline Health Status...16 c. Pediatric Follow up Adult CF Sample...20 a. Adult Baseline Respondent Sample...20 b. Adult Baseline Health Status...21 c. Adult Follow-up...26 V. DISCUSSION...28 VI. CONCLUSION...33 VII. REFERENCES...35 Institute of Health Economics Working Paper 99-8 ii

4 LIST OF TABLES TABLE Pediatric Demographic Statistics TABLE Pediatric Resource Utilization Baseline TABLE Pediatric Health Status Baseline TABLE Pediatric Resource Utilization Follow-up TABLE Pediatric Health Status Change over 1 year TABLE OLS Regression Results (Dependent variable: 1998 FSII[R]) TABLE Adult Demographic Statistics TABLE Adult Resource Utilization Baseline TABLE Adult Health Status (SF-36) Baseline TABLE Adult Resource Utilization Follow-up TABLE Adult Health Status Change over 1 year TABLE OLS Regression Results (Dependent variable: 1998 EQ-5D Index) Institute of Health Economics Working Paper 99-8 iii

5 LIST OF FIGURES FIGURE Distribution of VAS Scores for CF Children and Adolescents FIGURE QOL Scores for Pediatric CF Patients Hospitalized in previous 12 months FIGURE FSII(R) Distribution of Scores Pediatric CF Patients FIGURE Adult Health Status (EQ-5D) Baseline FIGURE Distribution of VAS Scores CF Adults (Age 18-59) & Alberta General Population Norms (Age 18-59) FIGURE SF-36 Mean Scores for Adult CF Patients FIGURE SF-36 Mean Scores by Household Income Adult CF Patients FIGURE SF-36 Composite Scores by Household Income Adult CF Patients Institute of Health Economics Working Paper 99-8 iv

6 ABSTRACT Objectives The objectives of this study were: 1) to examine the general health-related quality of life (HRQOL) of adult and pediatric cystic fibrosis (CF) patients using available generic instruments; and 2) to determine important predictors, including the use of recombinant human DNase (rhdnase), of one-year change in general health status in patients with CF. Methods Two generic HRQOL instruments were mailed to the 59 adults attending the adult CF clinic (University of Alberta Hospital, Edmonton, Alberta): the SF-36 and the EuroQol (EQ-5D). Three generic HRQOL instruments were mailed to the parents of the 88 children attending the pediatric CF clinic: the Functional Status II[R] (FSII[R]), the RAND Child Health Status, and the EuroQol Visual Analog Scale (EQ-VAS). Baseline surveys were sent in June 1997 and followup surveys in June Demographic information was also requested on the survey and clinical data was abstracted from the CF clinic charts. Univariate analyses were used to determine relationships between demographic and clinical parameters and baseline health status. Multivariate regression models were used to assess the relationships between clinical and demographic variables with one-year change in health status. Results Both adults with CF and parents of children with CF reported lower HRQOL scores compared to the general population norms for the generic instruments. Differences were seen in the FSII[R] scores between children who were hospitalized and those who were not. For adults, differences in SF-36 Mental Composite Summary (MCS) scores were found for the three household income groups, with lower incomes having lower scores (p=0.006). Comparisons with clinical indicators showed that Body Mass Index (BMI) was positively correlated with the FSII[R] score in the pediatric group, whereas, in the adult group, forced expiratory volume in one second (FEV1) % predicted was positively correlated with the SF-36 Physical Composite Summary (PCS) and EQ-VAS scores. Institute of Health Economics Working Paper

7 Regression analyses indicated that the 1998 FSII[R] scores for the pediatric group were positively associated with baseline health status, but not with other demographic or clinical variables. For adults, age (p=0.010), FEV1 % predicted (p=0.005), hospital admission (p=0.023) and the use of rhdnase (p=0.026) were all positively associated with 1998 EQ-5D Index scores, while BMI was negatively associated (p=0.025) with the 1998 EQ-5D Index score, after controlling for the baseline 1997 EQ-5D scores. However, the size of the regression coefficients would indicate that only baseline 1997 EQ-5D scores and hospital admission in the year were quantitatively importantly associated with 1998 EQ-5D scores. Conclusions As these questionnaires have not been used in the CF population before, important baseline information was generated regarding the general health status of CF patients in Alberta. While there exists a statistically significant association between use of rhdnase and one-year change in HRQOL in adults, the magnitude of this relationship may not be clinically important. Further evaluation of the relationship between hospital admission and change in HRQOL is required. In pediatric patients, assessments of child health may be limited when using parents as proxies. Institute of Health Economics Working Paper

8 I. INTRODUCTION The impact of chronic diseases is often assessed through mortality statistics and clinical measures of physiologic functioning. More recently, however, increasing attention is being paid to the personal burden of illness and the impact on quality of life when living with a disease. Instruments designed to measure health-related quality of life (HRQOL) have been used in a variety of clinical trials and cross-sectional studies involving people with chronic diseases. Consumers, governments, and payers in health care today are looking to health care providers to evaluate medical therapies using outcome measures that are more global and incorporate the person s own perception of their health status and functioning. The purpose of this study was to examine health-related quality of life (HRQOL) for patients with cystic fibrosis (CF). Little work has been done in assessing HRQOL for people with CF, but there exists an increased importance in studying this area. There are approximately 3,000 people with CF in Canada and 330 in Alberta. New therapies such as recombinant human DNase (rhdnase) are becoming available that can potentially impact on the health status and HRQOL of patients with CF. The objectives of this study were: to examine the general HRQOL of adult and pediatric CF patients using available generic instruments; and, to determine important predictors, including the use of rhdnase, of one-year change in general health status in patients with CF. II. BACKGROUND 1. Epidemiology of CF CF is an inherited, multi-system disorder affecting approximately one in every 2500 to 3000 live births, and is one of the most common lethal inherited disorders among Caucasians with 3.3% being carriers of the defective gene. The basic defect in CF is abnormal chloride ion Institute of Health Economics Working Paper

9 transport, which results in increased viscosity of mucus secretions. Because chloride ion channels are so widespread in the body, the physiologic impact of CF affects numerous systems. CF is a chronic disease that is quite variable with respect to both the systems involved and the severity of the symptoms. The variability is due in part to the various genetic mutations, but also to environmental and treatment factors. Patients with CF generally experience a slow deterioration over time, although there is much variability between and within CF patients. Assessment of the efficacy and effectiveness of treatment strategies for CF necessarily takes a long time because of this slow and variable progression. Pulmonary disease is primarily responsible for the morbidity and mortality in patients with CF. Increased viscosity and decreased mucocilliary clearance of sputum leads to chronic bacterial colonization, usually initially with Staphylococcus aureus and subsequently with Pseudomonas aeruginosa. The range of the disease s impact on people with CF includes an asymptomatic level with no outward appearance of ill health, to having symptoms of chronic obstructive pulmonary disease, with the production of large amounts of sputum, wheezing, dyspnea and limited exercise tolerance. Other systems affected by the defective gene include the digestive, musculoskeletal and reproductive systems. Complications of CF, apart from the pulmonary symptoms, include pancreatic insufficiency and malabsorption of fat and malnutrition in 85% of patients, diabetes mellitus, biliary cirrhosis, subacute bowel obstructions, arthritis, and infertility. Epidemiologic studies of Canadian CF patients indicate that the most significant predictor of mortality was FEV1, but age, sex, and weight are also important predictors (Kerem et al, 1992; Corey and Farewell, 1996). Age and sex adjusted relative risk of death within two years is estimated to be 2.0 for each decrement in the forced expiratory volume in one second (FEV1) of 10 percent below the predicted value (Kerem et al, 1992). A similar doubling of the risk of death were seen among patients 10 years younger than other patients with the same FEV1, and among females as compared to male patients (Kerem et al, 1992). Also an important predictor of mortality of CF patients in Canada is colonization status with Ps. aeruginosa and Burkolderia cepacia (previously Ps. cepacia) (Corey and Farewell, 1996). Institute of Health Economics Working Paper

10 As pulmonary disease is the leading cause of sickness and death in this patient population, therapies aimed at ameliorating pulmonary infections and improving lung functioning are of primary importance (Ramsey 1996). One such therapy, recombinant human DNase (rhdnase), has received considerable attention recently. rhdnase acts to reduce the viscosity of the pulmonary secretions by breaking down the high-molecular weight DNA that is present in high concentrations in the sputum of CF patients. In the major phase III clinical trial, rhdnase administered in aerosolized form once or twice daily over 24 weeks resulted in significant reductions in the number of exacerbations of respiratory symptoms requiring parenteral antibiotics and significant improvements in pulmonary function measures when compared to placebo (Fuchs et al, 1994). 2. HRQOL in Cystic Fibrosis HRQOL has been evaluated in CF patients using generic instruments, such as the Quality of Well-Being Scale (QWB) (Czyzewski et al, 1994, 1998; Kaplan et al, 1989; Munzenberger et al, 1999; Orenstein et al, 1989, 1990, 1991), the Nottingham Health Profile (NHP) (Busschbach et al, 1994; Caine et al, 1991; Congleton et al, 1996, 1997, 1998; Dennis et al, 1993; Keilen et al, 1994; Mannes et al, 1997), the Child Health Questionnaire (CHQ) (Powers, 1998), the RAND Functional Status Index (Shepherd et al, 1992), and the Sickness Impact Profile (SIP) (de Jong et al, 1997). These investigations have evaluated interventions such as outpatient antibiotic therapy and lung transplantation. Standard-gamble (SG) and time trade-off (TTO) techniques have been used in only a limited number of CF patients, as has the EuroQol visual analog scale format, in evaluating the HRQOL before and after bilateral lung transplantation in patients with CF (Busschbach et al, 1994). Although very preliminary, this study showed the feasibility of these preference-based measures in this patient population. Using the QWB on 44 CF patients aged 7 to 36 years, Orenstein et al (1989) found that QWB was significantly positively correlated with three pulmonary function test results (FEV1, FEF25-27%, PEFR) and exercise performance (peak VO2), while negatively correlated with age (Orenstein 1989). After controlling for disease severity (FEV1), the relationship between QWB and age became nonsignificant. In a second study, Orenstein et al (1990) applied the QWB in 28 patients with CF before and after a two-week course of oral ciprofloxacin used to treat Institute of Health Economics Working Paper

11 pulmonary exacerbations. They found significant positive correlations between changes in QWB and pulmonary function tests (FEV1, FVC, SaO2) (Orenstein 1990). Czyzewski et al (1994) used the QWB to collect health status ratings for adolescents with CF from the parents and adolescents. The results indicated, for both parents and adolescents, that the QWB did not correlate significantly with spirometry scores, but that clinical health status (represented by the NIH score) was significantly positively correlated with QWB (Czyzewski 1994). Parental reports of general pulmonary factor, weight factor, disability factor, and impact on family were all significantly negatively correlated with QWB, while adolescents reports of psychosocial factor and socialization were significantly negatively correlated with QWB (Czyzewski 1994). From what appears to be the same data set, Czyzewski et al (1998) provided the following results: QWB reported by parents and adolescents was significantly positively related to the NIH score; QWB reported by parents was statistically negatively related to the Child Behavior Checklist, Youth Self-Report, and the Impact on Family; QWB reported by adolescents was statistically positively correlated with weight (Czyzewski 1998). More recently, the responsiveness and validity of the QWB was evaluated in a sample of 20 children and adolescents experiencing acute exacerbations (Munzenberger et al., 1999). The QWB was administered immediately before and after hospital admission and at 6- and 12-month follow-ups. Results indicated that the overall QWB score, the physical and social function scores, and the symptom-problem complex scores were responsive to the changes before and after acute treatment. However, only the symptom-problem complex scores changed over the 12-month follow-up. When compared to pulmonary function (i.e., FEV1, FEV25-75) in these pediatric patients, QWB scores were weakly correlated (r=0.35). There were no significant correlations when changes in QWB scores were considered (Munzenberger et al., 1999). The Nottingham Health Profile (NHP) has also been used in several studies to assess the HRQOL in people with CF. Part 1 of the NHP has six dimensions: energy, pain, emotion, sleep, social isolation, and physical mobility, with higher scores indicating worse states of health. Congleton et al (1996, 1997) used the NHP with six additional questions related to CF in a group of patients 16 years of age and older. When compared with normal scores, men had significantly worse scores in the area of pain, energy, and social isolation, while women had significantly Institute of Health Economics Working Paper

12 worse scores in pain, emotion, and sleep (Congleton 1996). The NHP dimensions were also compared to clinical variables. All six dimensions were significantly negatively correlated with FEV1 % predicted and time able to work, and significantly positively correlated with shortness of breath, sputum volume, time on treatment, time in hospital (Congleton 1996). BMI was significantly negatively correlated with the energy, sleep, and physical mobility dimensions (Congleton 1996). Congleton et al (1998) obtained follow-up scores for 152 of the original 204 patients surveyed at an interval of 1, 2, or 3 years. The only dimension that showed a significant change over time was emotion, which showed a small improvement. There was also a significant deterioration in the CF-specific total score (Congleton 1998). There was no overall difference in change in NHP scores between the patients whose FEV1 declined and those who did not (Congleton 1998). The NHP has also been used to assess HRQOL in CF patients before and after lung transplantation. Caine et al (1991) and Busschbach et al (1994) found significant improvements for the physical mobility and energy dimensions after transplant (Caine 1991; Busschbach 1994). Dennis et al (1993) found that after transplant, there were significant improvements in physical mobility, energy, sleep, social isolation, and emotional reactions (Dennis 1993). Mannes et al (1997) compared four groups of people with end-stage lung disease, one group being CF, and found that the mobility and energy dimensions improved after lung transplantation (Mannes 1997). Keilen et al (1994) compared CF patients awaiting transplants with people with eating disorders and people with angina, and found that the patients with CF had a lower level of mobility, but were less impaired in the areas of emotional reaction and social isolation than the other groups (Keilen 1994). The Sickness Impact Profile (SIP) was used by de Jong et al (1997) to evaluate the relationship between HRQOL and pulmonary function, exercise capacity, and dyspnea. Their results showed no relationship between HRQOL and pulmonary function (FEV1, IVC), a significant negative relationship with exercise capacity, and a significant positive relationship with the dyspnea score (de Jong et al 1997). Powers (1998) used the Child Health Questionnaire (CHQ) to assess to differences among adolescent, mother, and father perceptions of child health in 24 adolescents with CF (Powers Institute of Health Economics Working Paper

13 1998). All three groups reported significant positive relationships between pulmonary function and perceptions of general health and physical health, but pulmonary function did not relate to the psychosocial, emotional, or behavioral health scales (Powers 1998). Shepherd et al (1992) used 12 questions on functional status from the RAND Health Insurance Study to assess the health status of 37 adults with CF compared with 46 controls (Shepherd 1992). The functional status of the CF group was significantly lower than that of the comparison group, but within the CF group, the Functional Status Index (FSI) had a significant positive correlation with self-rated health, current life satisfaction, acute symptom index, disability days, and hospital days, while it had a significant negative correlation with slow vital capacity (Shepherd 1992). More recently, studies assessing the effect of rhdnase on HRQOL have been reported. Assessments of CF-related symptoms and overall well-being were included in a large phase III clinical trial of rhdnase (Fuchs et al, 1994). However, the instrument used to assess well-being and symptom control was not fully described by the authors. Several additional studies appear to have used the same unidentified HRQOL instrument (Ramsey et al, 1993; Ranasinha et al, 1993; Wilmott et al, 1996). It consisted of five questions about general well-being (feeling, energy, physical activity, appetite, sleep pattern) and four on CF-related symptoms (ease of sputum expectoration, cough frequency, cough severity, congestion). The answers were ranked on a five-point Likert scale. However, the reliability and validity of this measure have not been reported. Results from the use of this particular questionnaire have been divided. Fuchs et al (1994) found that patients receiving rhdnase once daily reported a small improvement in well-being over baseline compared to patients receiving placebo; patients receiving twice daily rhdnase did not differ from placebo-treated patients. Patients receiving rhdnase in both groups reported significant improvements in CF-related symptoms when compared to the placebo group. Wilmott et al (1996), on the other hand, found no difference in HRQOL scores between the placebo and rhdnase treatment groups in the amount of improvement in cough and congestion, dyspnea at different levels of effort, activity limitation, emotional well-being, fatigue, days of restriction to bed, or general health perceptions. Ramsey et al (1993) found that the frequency Institute of Health Economics Working Paper

14 and magnitude of improvement (from baseline to Day 10) across all HRQOL questions was consistently greater among the patients receiving rhdnase than those receiving placebo. Ranasinha et al (1993) found that administration of rhdnase did not improve overall HRQOL, but there were significant improvements for rhdnase in individual components (general wellbeing, cough frequency, chest congestion). Heijerman et al (1995) used a different unidentified questionnaire that contained six questions, with responses placed on a visual analog scale, in the following areas: patients awareness of expectoration, chest congestion, tenacious sputum, exercise tolerance, physical handicap, and general well-being. They found that, for 12 adults with moderate to severe CF, treatment with rhdnase (2.5mg daily for six weeks) significantly improved the CF-related symptoms (perception of expectoration, airway congestion, and tenacious sputum) (Heijerman 1995). However, there were no significant improvements in general well-being, exercise tolerance, or physical limitations reported by these patients. Shah et al (1995) used yet another unidentified HRQOL questionnaire that covered eight domains: cough, congestion, dyspnea, physical functioning, emotional well-being, fatigue, staying in bed, and general health perception. They found a non-statistically significant improvement in the HRQOL score (Shah 1995) in patients receiving rhdnase twice daily. Two investigations on the HRQOL impact of rhdnase have been presented in abstract form only. Munzenberger et al (1996) reported significant improvements in QWB scores for 30 patients with mild CF (FVC > 70%) treated for two weeks with rhdnase. This improvement in HRQOL was observed despite a lack of significant change in pulmonary function tests. Mellis (1997) reported statistically significant improvements in both spirometry and HRQOL, using the SF-36, following three month s treatment with rhdnase in 69 patients with mild to moderate CF (FVC > 40%). However, Mellis noted improvements in SF-36 total scores; it is not clear in the abstract what was meant by this. To increase the sensitivity to detect subtle changes in the symptomatology of the disease under study, it is generally suggested to also include a disease-specific HRQOL measure (Tullis and Guyatt, 1995; CCOHTA, 1997). Discussions with colleagues and other researchers in this area indicated that there are currently two different groups developing CF-specific instruments. Institute of Health Economics Working Paper

15 The first instrument is being developed by researchers in Toronto, and is specifically for adult patients with CF (Dr. Elizabeth Tullis, personal communication). The CFQoL Group in France has recently developed a French HRQOL instrument called the Cystic Fibrosis Questionnaire (CFQ). There are three versions: a parent form for children aged 8-13 (CFQ-Child P), a child form for ages 8-13 (CFQ-Child C), and a self-completed form for teenagers and adults (CFQ- 14+) (Henry et al, 1997; 1998; 1998). Initial testing for validity and reliability has been done (Henry et al, 1997). The questionnaires have been translated to German and Spanish, with pilot tests run for each (Henry et al, 1998). The three versions of the CFQ have recently been translated for use in North America (Quittner et al., 1998). Plans are currently underway to conduct further psychometric testing of the CFQ versions at multiple sites in the US (Quittner & Munzenberger, 1999). III. METHOD 1. Patients All patients registered at the CF Clinic, University of Alberta Hospital, Edmonton were initially included. Because of the wide age range of treated patients, it was important to recognize the differences in impact on HRQOL that the disease and treatments may have on patients with CF. For this reason, the HRQOL assessments were divided into adult and pediatric groups, as defined by their attendance at the CF Clinics: pediatrics (aged 0-17 years) (N=88) and adults (aged 18+ years) (N=59). 2. Surveys In a baseline, cross-sectional survey, general health status measures were mailed in June Mailed packages included a cover letter, health status measures, and a postage-paid return envelope. All mailings were distributed through the CF Clinic office at University of Alberta Hospital. Non-responders to this initial mailing were telephoned to encourage response. Oneyear follow-up surveys were sent in June 1998 to all respondents of the baseline cross-sectional survey. Again, telephone follow-up at 2 weeks after the initial mailing was used to increase follow-up response rate. Institute of Health Economics Working Paper

16 3. Measures a. Adult Two generic health status instruments were mailed to adult CF patients: the SF-36 and the EuroQol (EQ-5D). The SF-36 is the most commonly used health status profile measure (Ware and Sherbourne, 1992). This instrument includes eight multi-item health concepts or dimensions: physical functioning (10 items), role limitations due to physical problems (4 items), social functioning (2 items), bodily pain (2 items), mental health (5 items), role limitations due to emotional problems (3 items), vitality (4 items) and general health perceptions (5 items). An additional item reflecting health transition, that is, changes in health status over a one-year period, is included in the 36-item instrument, but is not used in the aggregation of the eight multi-item scales (Ware and Sherbourne, 1992). For each dimension, item scores are coded, summed, and transformed on a scale of 0 to 100, with higher scores indicating better health status. The popularity of the SF-36 stems from its brevity (approximately 5-10 minutes to complete), and the extensive information regarding the psychometric properties of the health status measurements obtained with this instrument (Ware and Sherbourne, 1992). It has also been shown that the multi-item scales of the SF-36 load differentially on two general components of health status, that is physical and mental health constructs (Ware et al., 1993). Ware and colleagues (1994) provide further evidence and guidance for the construction of Physical Component Summary (PCS) and Mental Component Summary (MCS) scales from scores of the SF-36. Approximately 80 to 85 percent of the reliable variance in the eight scales of the SF-36 is accounted for by physical and mental components (Ware et al., 1994). It has been suggested by these investigators that these two summary scores from the SF-36 are useful for most purposes, such as evaluating specific treatment effects or assessing the general health of a population. The EQ-5D provides a multi-dimensional descriptive system of health status (Kind 1996). The 5 dimensions are: Mobility, Self-care, Usual Activity, Pain/Discomfort, and Anxiety/Depression. All dimensions, or attributes, are divided into three levels, reflecting no Institute of Health Economics Working Paper

17 problem, some problem, and extreme problem with the focus of that dimension. By applying a previously developed scoring system, based on population preferences for health states described by the EQ-5D system, a single, preference-based index is generated (Kind, 1996). For the purpose of this investigation, the time trade-off (TTO) based preference scoring function, derived from the general population in the United Kingdom (Dolan 1997), was applied. In addition to the multi-dimensional descriptive system, the EQ-5D also includes a visual analog scale (EQ-VAS) as a means of valuing the respondent s health state within the descriptive system. The endpoints of the EQ-5D VAS are labelled best imaginable health state and worst imaginable health state anchored at 100 and 0, respectively. Respondents are asked to indicate how they rate their own health state by drawing a line from an anchor box to that point on the VAS which best represents their own health on that day. These instruments were chosen for a number of reasons. First, both instruments can be selfadministered and can be completed in a relatively short period of time. Second, an abundance of evidence is available as to the measurement properties of both of the instruments. The EQ-VAS has been used to evaluate HRQOL in lung transplant recipients (Bussbach et al., 1994), and the SF-36 has been used to assess the impact of rhdnase in this patient population (Mellis 1997). Third, the combination provides both a profile measure (SF-36) which taps several distinct dimensions of HRQOL, including both physical and mental status and functioning, as well as an overall index measure (EQ-5D) which can be used to represent patients overall preference for their current health states. Such an index can also be incorporated in Quality Adjusted Life Year (QALY) calculations if required. b. Pediatric Health status measures in pediatric patients are less well developed than measures for general adult populations (Connolly and Johnson 1998). Of particular concern are issues related to comprehension of health status measures, and the use of parents as proxies. For the purpose of this investigation, three generic HRQOL instruments were mailed to the parents of 88 CF patients, with one parent being asked to respond on behalf of each child. The generic HRQOL instruments used in the pediatric population were: the Functional Status II[R] (FSII[R]), the RAND Child Health Status, and the EuroQol Visual Analog Scale (EQ-VAS). Institute of Health Economics Working Paper

18 Because the SF-36 has not been used with children, it would not be considered appropriate in this investigation. However, a related 7-item instrument, based on measures used in the RAND Health Insurance Experiment (Eisen et al, 1979; 1980) has been used in assessing the health status of children. The same researchers who developed the SF-36 developed the RAND Child Health Status instrument. The first question of the RAND Child Health Status instrument corresponds to the first question of the SF-36. This item is a global assessment of health status using a 5-point response format: excellent, very good, good, fair, poor (EVGFP). In addition, a complementary 14-item child health status measure, the Functional Status II[R] (FS II[R]) (Stein and Jessop, 1990; 1991), was used. This combination of short, selfadministered instruments has been used with good success in children with chronic disease (Lewis et al, 1989). There are a total of 21 items, which take approximately 10 minutes to complete, and was self-administered by the parents. In addition, the EQ-VAS was selfadministered by the parents to determine a rating of the child s overall health state. c. Demographics In addition to the standardized health status measures, demographic information was requested in the survey. This information included gender, date of birth, and household income which was collected in three categories (< $25,000; $25,000 to 50,000; > $50,000). The respondent was also asked if they had a family doctor and, if so, how many times they had seen their family doctor for any reason, as well as for problems directly related to their CF, in the previous 12 months. Information was obtained on whether they had received emergency room treatment for non-cf related problems in the previous 12 months. Information relating to hospital admissions was also collected, specifically if they had been admitted to a hospital for any reason, or alternatively admitted for problems directly relating to their CF, in the previous 12 months. 4. Clinical Data Clinical data was abstracted from the CF clinic charts for the period June 1997 to June 1998 for all respondents of the baseline cross-sectional survey. This information included height, weight, BMI, total number of hospital days due to CF, microbiology status (presence of Institute of Health Economics Working Paper

19 Pseudomonas aeurginosa, Burkeholderia cepacia, Staphylococcus aureus, and other), spirometry test scores (FEV1 %predicted, FVC %predicted), use of pancreatic enzymes, use of rhdnase, and use of antibiotics (prophylactic; exacerbation: oral, inhaled, iv). 5. Analysis Baseline health status scores were analyzed based on demographic and clinical parameters. Univariate analyses were used to determine relationships between demographic and clinical parameters and health status. Because of the relatively small sample size, nonparametric tests were used to assess statistical significance in the univariate analyses of health status scores. In addition, observed baseline health status scores were compared to published population normal scores for the instruments, where available. Univariate and multivariate analyses were also used to determine predictors of change in health status. One-year change scores were calculated for the health status measures by subtracting 1997 baseline scores from the 1998 follow-up scores. Again, for univariate comparisons of health status change scores, based on clinical and demographic data, Spearman correlations and Kruskal-Wallis ANOVA were applied for both adult and pediatric respondent samples. Multivariate regression models were used to assess the relationships between clinical and demographic variables and change in health status. For the pediatric sample, the FSII[R] was used as the dependent variable; for the adult sample, the EQ-5D derived index score was the primary dependent variable. Regression models were also evaluated using the SF-36 PCS and MCS as dependent variables. Analyses of covariance (ANCOVA) models were applied, using baseline (1997) health status as a covariate (Streiner and Norman, 1995). Potential predictor variables included: age, gender, height, weight, FEV1 % predicted, FVC % predicted, CF exacerbations, hospitalizations in , bacteriologic status (presence of Pseudomonas aeurginosa, Burkeholderia cepacia, or Staphylococcus aureus), and use of rhdnase. Because not all patients had multiple spirometry tests during the study period, only the first recorded FEV1 % predicted and FVC % predicted values were used. Institute of Health Economics Working Paper

20 IV. RESULTS 1. Pediatric CF Patients a. Pediatric Baseline Respondent Sample Baseline questionnaires were returned from the parents of 69 children (78% response rate). The respondent sample accurately represented the clinic population in terms of age and gender (Table 1). Household income categories were available only for the respondents of the questionnaires. Table 1 - Pediatric Demographic Statistics Clinic (N = 88) Respondents (N = 69) Male 47 (53%) 36 (52%) Female 41 (47%) 33 (48%) Age mean: 9.9 (SD 5.0) range: 0 17 mean: 9.3 (SD 4.8) range: 0 17 Age (58%) 43 (62%) Age (42%) 26 (38%) Household Income < $25,000 $25,000-50,000 > $50,000 no response n/a 18 (26%) 20 (29%) 21 (30%) 10 (15%) Resource utilization information was also obtained in the baseline survey regarding visits to a family doctor in the previous 12 months for problems directly relating to their CF, as well as admissions to the hospital for problems directly relating to their CF in the previous 12 months (Table 2). Institute of Health Economics Working Paper

21 Physician Visits in previous 12 months Hospital Visits in previous 12 months Table 2 - Pediatric Resource Utilization Baseline Category Respondents (N = 69) 0 times 1-2 times 3-5 times 6-10 times > 10 times no response yes no no response 24 (35%) 10 (15%) 12 (17%) 4 (6%) 5 (7%) 14 (20%) 23 (33%) 44 (64%) 2 (3%) B. PEDIATRIC BASELINE HEALTH STATUS Descriptive statistics for the various instruments appear below (Table 3). The FSII[R] and RAND Total score had internal consistency reliability coefficients of 0.76 and 0.84, respectively. The distribution of Visual Analog Scale (VAS) scores indicated poorer health for the adolescents (aged 12-17) compared to the children (aged 0-11), as shown in Figure 1. Table 3 - Pediatric Health Status Baseline Instrument/scale (range) Mean SD Cronbach s a FSII[R] (0-100) RAND Total (0-100) RAND EVGFP (1-5) EQ-VAS (0-100) Scores on all three scales were significantly lower for children who had CF-related hospital admission in the previous 12 months (Figure 2). There were no differences by gender or household income. At baseline, there were no significant correlations between health status measures and lung function (FEV 1 %predicted). BMI was positively correlated with the baseline FSII[R] score (Spearman s rho=0.255; p=0.049), but not with the other scales. Institute of Health Economics Working Paper

22 FIGURE 1 - DISTRIBUTION OF VAS SCORES FOR CF CHILDREN AND ADOLESCENTS Children (mean=79.9, sd=17.8) Adolescents (mean=75.7, sd=16.3) < VAS Score Categories FIGURE 2 - QOL SCORES FOR PEDIATRIC CF PATIENTS HOSPITALIZED IN PREVIOUS 12 MONTHS CF Hospitalization No Hospitalization p = * p = * 70 p = * RAND SCALE FUNCTIONAL STATUS II (R) EQ-VAS Institute of Health Economics Working Paper

23 When compared to previously published normal scores, the mean FSII[R] score was lower in this sample of children with CF (84.3) than scores for Well (96.0) and Sick (86.5) children (Stein and Jessop 1990) (Figure 3). Sick children were described by Stein and Jessop (1990) as those having chronic physical conditions. FIGURE 3 - FSII(R) DISTRIBUTION OF SCORES - PEDIATRIC CF PATIENTS Well (%)* (mean=96.0) Sick (%)* (mean=86.5) CF (%) (mean=84.3) % *Stein RE, Jessop DJ. Manual for the Functional Status II(R) Measure. PACTS Papers. Bronx, New York: Albert Einstein College of Medicine C. PEDIATRIC FOLLOW-UP Follow-up surveys were only sent to 67 parents. In the one-year follow-up period, two children died and, therefore, follow-up surveys were not sent to their families. Follow-up surveys were returned for 60 children (90% follow-up rate). Of the seven non-responders, two had moved to another province, one had a disconnected phone and could not be reached, and four did not answer their phone or return the clinic s call. To our knowledge all seven of these patients were still alive at the time of the one-year follow-up. The follow-up sample was 55% male with a mean age of 9.8 (4.9) years (range 2-18). The distribution for the household income was as follows: 23% (< $25,000), 22% ($25,000 50,000), 42% (> $50,000), 13% (no response). Resource utilization information was also Institute of Health Economics Working Paper

24 obtained in the follow-up survey regarding visits to a family doctor in the previous 12 months for problems directly relating to their CF, as well as admissions to the hospital for problems directly relating to their CF in the previous 12 months (Table 4). Physician Visits in previous 12 months Hospital Visits in previous 12 months Table 4 - Pediatric Resource Utilization Follow-up Category Respondents (N = 60) 0 times 1-2 times 3-5 times 6-10 times > 10 times no response yes no no response 22 (37%) 10 (17%) 14 (23%) 2 (3%) 2 (3%) 10 (17%) 16 (27%) 42 (70%) 2 (3%) Overall, there was no significant change in health status from 1997 to 1998 on any of the three measures (Table 5). However, parents reported larger change in health status on the FSII[R] for children who were hospitalized during 1998 (change = -10.7) than children not hospitalized (change = -0.9) (p=0.03). No other demographic or clinical variables were associated with a change in health status/hrqol in this sample of pediatric CF patients. Table 5 - Pediatric Health Status Change over 1 year Instrument/scale (range) Mean Change SD Range 95 % CI Lower Upper FSII[R] (0-100) to RAND Total (0-100) to EQ-VAS (0-100) to Multiple linear regression analyses indicated only baseline health status was associated with a positive 1-year variance in FSII[R] scores (Table 6). The R 2 value for this model was 0.457, Institute of Health Economics Working Paper

25 indicating that the variables explained approximately 46 percent of the variance in 1998 FSII[R] Index scores. (F=3.368; p=0.007) TABLE 6 - OLS REGRESSION RESULTS (Dependent variable: 1998 FSII[R]) b SE Sig. Lower 95% CI Upper Constant Age Gender BMI rhdnase (1=Yes, 0=No) FEV 1 %predicted Hospital Admission Pseudomonas FSII[R] Index ADULT CF SAMPLE A. ADULT BASELINE RESPONDENT SAMPLE Baseline questionnaires were returned from 39 adults (66.1% response rate). As in the pediatric sample, the adult respondent sample accurately represented the clinic population in terms of age and gender (Table 7). The three household income categories were available only for the respondents of the questionnaires. The largest number of people were in the lowest income group, then the middle income group, with the smallest number of people being in the highest income group. This distribution likely represents the functional limitations of CF, in that many adult patients are unable to obtain or hold gainful employment. Institute of Health Economics Working Paper

26 Table 7 - Adult Demographic Statistics Clinic (N = 59) Respondents (N = 39) Male 35 (59%) 23 (59%) Female 24 (41%) 16 (41%) Age mean: 30.9 (SD 9.6) range: Household Income < $25,000 $25,000-50,000 > $50,000 no response n/a mean: 32.5 (SD 10.6) range: (38%) 12 (31%) 9 (23%) 3 (8%) Resource utilization information was also obtained in the baseline survey regarding visits to a family doctor in the previous 12 months for problems directly relating to their CF, as well as admissions to the hospital for problems directly relating to their CF in the previous 12 months (Table 8). Physician Visits in previous 12 months Hospital Visits in previous 12 months Table 8 - Adult Resource Utilization - Baseline Category Respondents (N = 39) 0 times 1-2 times 3-5 times 6-10 times > 10 times no response yes no no response 13 (33%) 6 (16%) 5 (13%) 2 (5%) 2 (5%) 11 (28%) 12 (31%) 25 (64%) 2 (5%) B. ADULT BASELINE HEALTH STATUS The percentages of respondents with no problem, some problem, and extreme problem for the five EQ-5D dimensions are presented below (Figure 4). None of the respondents had a Institute of Health Economics Working Paper

27 problem in the area of self-care. In the areas of mobility, usual activities, and anxiety/depression, most people had no problem. In the area of pain/discomfort, the majority had some problem. 120% FIGURE 4 ADULT HEALTH STATUS (EQ-5D) - BASELINE 100% Extreme problem Some problem No problem 80% 60% 40% 20% 0% Mobility Self-Care Usual Activities Pain / Discomfort Anxiety / Depression Observed scores for the EQ-5D were generally lower than published normal scores for this instrument. When compared with the Visual Analog Scale (VAS) scores for Alberta general population norms, the mean score for the CF group was lower (74.4) than the Alberta general population (81.8). The sample scores were not age-adjusted, and since the general population values were from an older sample on average, the difference in score distribution may actually be greater than portrayed in this graph. (Figure 5). Institute of Health Economics Working Paper

28 FIGURE 5 - DISTRIBUTION OF VAS SCORES - CF A DULTS (AGE 18-59) & ALBERTA GENERAL POPULATION NORMS (AGE 18-59) CF Population (mean=74.4, sd=17.5) 25 General Population (Alta) (mean=81.8, sd=13.4) % < Descriptive statistics for the eight SF-36 dimensions are presented below (Table 9). Internal consistency reliability coefficients were all above Observed scores for the SF-36 were generally lower than published normal scores for the general U.S. population for this instrument (Figure 6), with the exception of the mental health index, where the mean score was slightly higher for the people with CF. TABLE 9 - ADULT HEALTH STATUS (SF-36) - BASELINE Dimension Mean SD Cronbach s a Physical Functioning Role Physical Pain Index General Health Vitality Role-Emotional Social Functioning Mental Health Index Institute of Health Economics Working Paper

29 FIGURE 6 SF-36 MEAN SCORES FOR ADULT CF PATIENTS Norm* (Age 18-54) CF (Age 18-59) Physical Functioning Role- Physical Pain Index Social Functioning Mental Health Index Role- Emotional Vitality *Ware JE, et al. SF-36 Health Survey Manual and Interpretation Guide. Boston, Massachusetts: The Health Institute, New England Medical Center General Health Perceptions At baseline, FEV 1 % predicted was significantly positively correlated with two scales at baseline: the PCS-36 score (Spearman s rho = 0.396; p=0.025) and the EQ-5D VAS score (Spearman s rho = 0.427; p=0.017). There were no significant correlations between health status measures and BMI. No significant differences were found to exist when baseline SF-36 and EQ-5D scores were compared by gender, with the exception of the SF-36 Bodily Pain dimension, where males (78.64; SD=20.42) reported less pain than female respondents (64.63; SD=20.66). No relationship was found to exist between age and any of the HRQOL measures. Individuals who had been admitted to a hospital tended to have lower scores in the general health status and pain dimensions of the SF-36, but the difference were not statistically significant. When compared by self-reported household income, significantly lower scores for the lowest income category were observed for four of the eight SF-36 dimensions: physical role limitation, emotional role limitation, social functioning, and mental health scale (Figure 7). Scores for the MCS of the SF-36 were also significantly lower for respondents reporting the lowest household income (Figure 8). The Physical Composite Score (PCS) has a similar Institute of Health Economics Working Paper

30 gradient, although differences were not statistically significant in this sample. When the other measures were compared by household income, significantly more respondents with lower household income indicated some problem on the Anxiety/Depression dimension of the EQ-5D (chi-squared = ; p=0.004). FIGURE 7 SF-36 MEAN SCORES BY HOUSEHOLD INCOME ADULT CF PATIENTS * * * > $50,000 (N = 9) $25,000 - $50,000 (N = 12) < $25,000 (N = 15) * * (p<0.05) 0 Physical Functioning Role- Physical Pain Index Social Functioning Mental Health Index Role- Emotional Vitality General Health Perceptions Institute of Health Economics Working Paper

31 FIGURE 8 - SF-36 COMPOSITE SCORES BY HOUSEHOLD INCOME -ADULT CF PATIENTS < $25,000 (N = 15) 80 $25,000 - $50,000 (N = 12) p= > $50,000 (N = 9) p=0.0061* Physical Component Scale (PCS) Mental Component Scale (MCS) C. ADULT FOLLOW-UP Follow-up surveys were returned by 32 people (82% follow-up rate). Of the seven people who did not respond to the second questionnaire, two had moved to another province, one had a disconnected phone and could not be reached, and four did not answer their phone or return the clinic s call. To our knowledge all seven of these patients were still alive at the time of the one year follow-up. The follow-up sample was 54% male with a mean age of 33.9 (12.4) years (range 0-60). The distribution for the household income was as follows: 31% (< $25,000), 28% ($25,000 50,000), 31% (> $50,000), 10% (no response). Resource utilization information was also obtained in the follow-up survey regarding visits to a family doctor in the previous 12 months for problems directly relating to their CF, as well as admissions to the hospital for problems directly relating to their CF in the previous 12 months (Table 10). Institute of Health Economics Working Paper

32 Table 10 - Adult Resource Utilization Follow-up Category Respondents (N = 32) Physician Visits in previous 12 months Hospital Visits in previous 12 months 0 times 1-2 times 3-5 times 6-10 times > 10 times no response yes no no response 8 (25%) 8 (25%) 2 (6%) 2 (6%) 2 (6%) 10 (32%) 7 (22%) 24 (75%) 1 (3%) TABLE 11 - ADULT HEALTH STATUS CHANGE OVER 1 YEAR Instrument/scale (range) Mean Change SD Range 95 % CI LOWER UPPER EQ-5D Index (0-1) to PCS-36 (0-100) to MCS-36 (0-100) to Overall, there was no significant change in health status from 1997 to 1998 (Table 11). However, multiple linear regression analyses indicated that age, FEV 1 %predicted, a hospital admission, and use of rhdnase were associated with a higher EQ-5D Index scores in 1998, after controlling for baseline EQ-5D index scores (Table 12). BMI was found to have a negative association with the EQ-5D Index scores. The R 2 value for this model was 0.734, indicating that the variables explained approximately 73 percent of the variance in 1998 EQ-5D Index scores. (F=3.591; p=0.017) Institute of Health Economics Working Paper