THE CONSTRUCT VALIDITY OF THE RAND-12 AND ... - CiteSeerX

THE CONSTRUCT VALIDITY OF THE RAND-12 AND HEALTH UTILITIES INDEX MARK 2 AND MARK 3 IN HIGH-RISK PRIMARY-CARE PATIENTS Working Paper 04-02 David Feeny 1,2,3 Karen Farris4 Isabelle Côté 5 Jeffrey A. Johnson6 Ross T. Tsuyuki7 and Ken Eng1

1. Institute of Health Economics, Edmonton, Alberta, Canada 2. Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada 3. Health Utilities Incorporated, Dundas, Ontario, Canada 4. College of Pharmacy, University of Iowa, Iowa City, Iowa, USA 5. Innovus Research Inc., Burlington, Ontario, Canada 6. Department of Public Health Sciences, University of Alberta, Edmonton, Alberta, Canada 7. Heritage Medical Research Centre, Department of Medicine, University of Alberta, Edmonton, Alberta, Canada

Legal Deposit 2000

April 2004

National Library Canada ISSN 1481-3823

TABLE OF CONTENTS ABSTRACT...............................................................................................................................1 Background ..........................................................................................................................1 Objective ..............................................................................................................................1 Study Design and Setting.....................................................................................................1 Results..................................................................................................................................1 Conclusions..........................................................................................................................1 INTRODUCTION .....................................................................................................................2 METHODS ................................................................................................................................3 Patients and Procedures .......................................................................................................3 SF-12....................................................................................................................................3 RAND-12 HSI .....................................................................................................................4 HUI2 and HUI3....................................................................................................................4 A Priori Hypotheses.............................................................................................................5 Data Analysis .......................................................................................................................6 RESULTS ..................................................................................................................................6 DISCUSSION ............................................................................................................................8 REFERENCES ........................................................................................................................12

Institute of Health Economics Working Paper 04-02

i

LIST OF TABLES TABLE 1..................................................................................................................................16 Summary of A Priori Predictions About Relationships Between SF-12 and RAND-12 Physical and Mental Health Summary Scores and HUI2 and HUI3 Single-Attribute and Overall Utility Scores, Number of Medical Conditions, and Number of Prescription Drugs TABLE 2..................................................................................................................................19 Summary of Demographic and Clinical Characteristics of Patients at Baseline (n=154) TABLE 3..................................................................................................................................22 Physical and Mental Health Summary Score of the SF-12, RAND-12 and HUI2 and HUI3 Scores at Baseline (n=154) TABLE 4..................................................................................................................................24 Summary of Pearson Correlations and A Priori Predictions


ii

ACKNOWLEDGMENTS Financial support for the study of primary healthcare teams was provided by a grant from the Health Transition Fund Alberta Health and Wellness. Financial support for the analyses presented in this paper was provided by a grant from the Merck Company Foundation to the Institute of Health Economics. The HTF, Merck Company Foundation, and IHE played no role in the design, interpretation, or analysis of the project reported here and have not reviewed or approved of this manuscript. The authors gratefully acknowledge the input of Sherry L. Dieleman, Sandra Brilliant, Ross Bayne, Leslie Gardiner, Dr. David Moores, and Marj Sandilands to the Primary-Care Study. The authors also gratefully acknowledge the constructive suggestions made by two anonymous reviewers.

It should be noted that David Feeny has proprietary interest in Health Utilities Incorporated which distributes copyrighted Health Utilities Index materials

Address for Correspondence: David Feeny Institute of Health Economics #1200, 10405 Jasper Avenue Edmonton, AB T5J 3N4 Canada Telephone: 780 448 4881, ext. 236 FAX: 780 448 0018 E-Mail: [email protected]


iii

ABSTRACT Background The Short Form 12 (SF-12) is used in a number of primary-care settings. To date the RAND-12 Health Status Inventory (HSI), Health Utilities Index Mark 2 (HUI2) and HUI Mark 3 (HUI3) have not been as widely used in primary-care settings. Objective The objective was to examine the construct validity of the SF-12, RAND-12 HSI, HUI2, and HUI3 in the context of high-risk primary-care patients. Study Design and Setting The SF-12, HUI2, and HUI3 were administered to a cohort of high-risk primary-care patients. SF-12 and RAND-12 HSI summary scores for physical and mental health were generated. Single-attribute utility scores for each dimension of health status as well as overall health in HUI2 and HUI3 were also computed. A priori hypotheses were specified. Results In general, the relationships among SF-12/RAND-12 scores and HUI2 and HUI3 scores were consistent with construct validity. Thirty-eight of 76 of a priori predictions were confirmed. However, predictions about the correlations between the number of medical conditions and number of medications and the measures of health-related quality of life were, in general, not confirmed. Conclusions The RAND-12 and HUI2 and HUI3 appear to be useful among primary-care patients with diverse chronic conditions. An advantage of HUI is the availability of both single-attribute (domain) and overall scores. Further investigation of construct validity is warranted.


1

INTRODUCTION In recent years there has been growing interest in using measures of health status and healthrelated quality of life (HRQL) (also know as patient reported outcomes) as measures of outcome in research studies (Guyatt et al. 1993; McDowell and Newell 1996; Coons et al. 2000; Revicki et al. 2000). There has also been growing interest in using these measures in routine clinical care both for quality assurance purposes and to assist in patient management and the monitoring of patient HRQL (Edelman et al. 1999; Lubetkin et al. 2002, 2003; Macran et al. 2003; Rubenstein et al. 1995; Wasson et al. 1999; Welch 1999; McHorney and Tarlov 1995; McHorney and Bricker 2002; Detmar et al. 2002). Applications have included primary-care settings as well as a number of speciality clinics. Do these measures measure health status and HRQL? Are these measures valid? In the field of HRQL there are no gold standard measures. The assessment of validity therefore depends on assessing the extent to which a measure performs as it should if it truly measures what it says it does - construct validity. The assessment of construct validity for any measure involves the accumulation of results from applications in a variety of settings (McDowell and Newell 1996; Coons et al. 2000). Patients seen in primary-care settings have diverse problems and conditions. Therefore, disease-specific measures, while very useful for patients with those specific conditions, do not permit comparisons among patients. Being able to make comparisons is especially important in the context of primary healthcare reform and quality assurance. It is therefore important to assess the construct validity of generic measures of HRQL in primary-care settings. The Medical Outcomes Study Short-Form (SF) set of measures (SF-36, SF-12) have been widely used in routine clinical care and in research studies (Ware and Sherbourne 1992; Ware et al. 1993, 1996; Ware 1996). To date the RAND-12 Health Status Inventory (HSI) (Hays et al. 1993; Hays 1998; Hays and Morales 2001), Health Utilities Index Mark 2 (HUI2), and Health Utilities Index Mark 3 (HUI3) (Feeny et al. 1992, 1996, 2002; Furlong et al. 2001) have not been as extensively used in primary-care settings. The purpose of this study was to assess the construct validity of four widely used generic measures, SF-12, RAND-12, HUI2, and HUI3, in

2


the context of high-risk primary-care patients.

METHODS Patients and Procedures The main study was concerned with an evaluation of the provision of primary care by multidisciplinary teams of providers (physicians, nurses, and pharmacists) for patients in the community at high risk for medication problems (Côté et al. 2002; Farris et al. 2004). Criteria for identifying patients as high risk included (1) taking three or more routine medications per day, (2) having one or more chronic disease states that were poorly controlled, (3) having one or more chronic disease states that were not currently being treated, (4) having a dosage regimen that had changed more than four times in the previous 12 months, (5) taking medications that had a narrow therapeutic index, (6) having an identified drug-related problem or the potential for one, (7) having a history of noncompliance or (8) having experienced a recent decline in health status. The recruitment of the patients into the study occurred between October 1999 and March 2000 in Edmonton, Alberta. Patients were assessed at baseline (study entry), three, and six months. During the study 199 of the 299 patients who were asked to be part of the study were enrolled (Côté et al. 2002). The most prevalent medical conditions were hypertension (30.2%), osteoarthritis (26.4%), depressive disorder (20.3%), and osteoporosis (20.3%). Patients had diverse and numerous physical and/or mental health problems and conditions. SF-12 The SF-12 includes 12 items covering eight domains of health status: physical functioning, role-physical, bodily pain, general health, vitality, social functioning, role-emotional, mental health (Ware et al. 1996; Ware 1996). The physical component summary score (PCS) is intended to summarize physical functioning, role-physical, bodily pain, and general health domains while the mental component summary score (MCS) reflects vitality, social functioning, role-emotional, and mental health domains. Ware et al. 1993 consider physical and mental health to be distinct and therefore uncorrelated. Therefore the PCS and MCS summary scores are based on orthogonal factor rotation (Ware et al. 1996). More specifically, PCS is calculated Institute of Health Economics Working Paper 04-02

3

using positive weights for physical functioning, role-physical, bodily pain, and general health and negative weights for vitality, social functioning, role-emotional, and mental health. Similarly, MCS is calculated using positive weights for vitality, social functioning, roleemotional, and mental health and negative weights for physical functioning, role-physical, bodily pain, and general health. PCS and MCS scores range from 0 to 100 with higher scores indicating better health status. RAND-12 HIS The RAND-12 HSI utilizes the same health-status assessment questionnaire as the SF-12 (Hays et al. 1993). The scoring system for the RAND HSI, however, differs from the scoring system for the SF measures in two important ways (Hays 1998; Hays and Morales 2001). First, RAND-12 HSI permits correlation between physical and mental health (oblique factor rotation). Negative weights are not used in calculating the 2 summary scores. Second, scores are based on weights derived from item-response theory. The RAND-12 HSI generates two summary scores: the physical health composite (PHC) and the mental health composite (MHC) scores. (A comprehensive comparison of SF-12 and RAND-12 summary scores is beyond the scope of this paper and is reported separately in Blanchard et al. 2003). There is evidence that in a number of contexts (but not all), there are important differences between the SF-12 and RAND-12 summary scores; see (Simon et al. 1998; Birbeck et al. 2000; Nortvedt et al 2000; Wilson et al. 2000; Taft et al 2001; Johnson and Maddigan 2004; Cunningham et al. 2003). PHC and MHC scores range from 0 to 100 with higher scores indicating better health status. HUI2 and HUI3 Each HUI system includes a health-status classification system and multi-attribute utility function based on community preferences that is used to generate utility scores reflecting the HRQL of the health states (Feeny et al. 1992, 1996, 2002; Furlong et al. 2001). HUI2 includes 7 dimensions of health status: sensation (vision, hearing, and speech), mobility, emotion, cognition, self-care, pain, and fertility. (In most applications, including this one, fertility is omitted; thus HUI2 includes 6 dimensions.) There are 4 or 5 levels within each dimension ranging from highly impaired (unable to control or use arms and legs for mobility) to normal. Single-attribute utility scores for each dimension of health status are on a scale in which the most 4


highly impaired level is assigned a score of 0.00 and normal (no problem) is assigned a score of 1.00. Overall scores are on a scale in which the all-worst HUI2 health state has a score of -0.03, dead has a score of 0.00 and perfect health has a score of 1.00 (Torrance et al. 1996). Similarly, HUI3 includes vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain and discomfort. In HUI3 there are 5 or 6 levels per dimension. Single-attribute utility scores on a scale where 0.00 = most impaired and 1.00 = normal (Feeny et al. 2002). Overall HUI3 scores are on a scale in which the all-worst HUI3 state has a score of -0.36, dead = 0.00, and perfect health = 1.00. Patients self completed the SF-12, HUI2, HUI3, a self-report instrument on medication adherence (Morisky et al. 1986), a series of questions on their utilization of healthcare services, their satisfaction with the care they were receiving, and their socio-demographic characteristics. Questions on utilization, the use of medications, and medical conditions were taken from previous surveys, most notably items included in the Statistics Canada National Population Health Survey. In general, little trend in health status and HRQL was observed in the study. Data from the study are thus suitable for the investigation of cross-sectional construct validity but not suitable for the assessment of longitudinal construct validity and responsiveness. Because the number of patients with complete data is highest for the baseline assessment, the baseline data are used in the analyses reported in this paper. The study was approved by the Health Research Ethics Board, Panel B of the University of Alberta. A Priori Hypotheses Construct validity involves the accumulation of evidence over time and in diverse settings about the performance of a measure (Coons et al. 2000; McDowell and Newell 1996). The specification of a priori hypotheses is an important step in the process of assessing construct validity. In general, it is not difficult to provide an ex post rationalization for an observed relationship. Before examining the degree of association between the SF/RAND and HUI measures, individually five of the six authors (excluding the author who conducted the statistical analyses) specified the relationships they expected to observe. The authors included a currently practicing Institute of Health Economics Working Paper 04-02

5

clinician pharmacist, two health economists, and two pharmacist-health-services researchers (with previous clinical experience). Some authors were familiar with all of the measures; others were mainly familiar with the SF/RAND measures. Individual a priori hypotheses were compiled and the majority view was taken as a consensus set of hypotheses (summarized in Table 1) was formulated. The system suggested by Guyatt et al. 1987 was used to interpret the degree of association. Correlations of 0.00 through 0.19 were regarded as negligible or not correlated; 0.20 through 0.34 was regarded as weak; 0.35 through 0.50 was regarded as moderate; and > 0.50 was regarded as strong. Data Analysis SF-12 and RAND-12 HSI scores were calculated using the algorithms outlined by the respective developers for each scale (Ware et al. 1993; Hays 1998). Descriptive HUI2 and HUI3 data and scores were derived using standard HUI algorithms (Feeny et al. 1996, 2002). Pearson correlations between the physical and mental health summary scores and singleattribute, overall health-state utility scores for HUI, the number of prescription medications, and number of chronic conditions were calculated. Data analysis was done using SPSS Version 11 (SPSS Inc., Chicago, Illinois). RESULTS Complete SF-12, RAND-12, HUI2, and HUI3 data were available for 154 of the 199 patients evaluated at baseline. Basic demographic and clinical characteristics for the 154 patients at baseline are presented in Table 2. Available information for the 45 patients excluded because of missing data for one or more HRQL measure indicates, as one might expect, that those excluded had modestly lower health status than the 154 for whom there was complete data. The excluded group also tended to be older, have lower levels of education, and were less likely to be male and/or born in Canada. These differences were statistically significant (p < 0.03). Results reported here are suitable for assessing construct validity and the relationships among measures; these results are not necessarily representative of the cohort. Summary results for health status and HRQL at baseline for SF-12, RAND-12, HUI2, and HUI3 are presented in Table 3. As expected, given the high-risk nature of the cohort, all 6


measures identify problems with physical health. Mean (standard deviation) Canadian SF norms for 65-74 year olds (mean age of cohort is 67 years) are 47.2 (9.7) for PCS and 53.7 (8.3) for MCS [27]. The RAND12-HSI MHC scores are suggestive of mental health problems; the single-attribute HUI3 emotion scores also indicate mental health problems. The MCS scores do not. The mean HUI2 and HUI3 scores are well below scores observed in healthy populations and indicate a substantial burden of morbidity. The range of scores observed for each measure was large; floor and ceiling effects were not evident. The effects of the imposed orthogonality (and use of negative weights) between PCS and MCS scores in the SF-12 as compared to the RAND-12 approach that allows PHC and MHC to be correlated is evident in several results. HUI2 mobility and HUI3 ambulation are more strongly associated with MHC than with MCS (Table 4). The pattern is similar for HUI2 selfcare and HUI3 pain. HUI2 emotion and HUI3 emotion scores are more strongly related to PHC than they are to PCS scores. Results comparing the observed relationships (Pearson correlations) among SF-12, RAND12, HUI2, and HUI3 and whether or not the observed relationships conformed to the a priori predictions are provided in Table 4. (Similar patterns of relationships were observed using Spearman correlations; data not shown.) In general, the nature of relationships that we expected were observed. HUI2 sensation (vision, hearing, and speech), HUI3 vision, HUI3 hearing, and HUI3 speech were only negligibly related to the SF and RAND summary scores. HUI2 mobility and HUI3 ambulation were moderately or strongly related to the SF and RAND physical health summary scores and negligibly related to the mental health scores. HUI2 and HUI3 emotion scores were weakly or negligibly related to the SF and RAND physical health scores and strongly related to the mental health scores. A priori predictions were confirmed by the results in 38 of 76 comparisons (50 %). When predictions were not confirmed, they were off by one category (for instance, weak instead of negligible or moderate) in 31 cases and off by two categories in 7 cases. Agreement between observed and predicted correlations was particularly low for the HUI2 and HUI3 single-attribute utility scores. Agreement between predicted and observed was even lower for overall HUI scores, number of medical conditions, and number of prescription drugs. Agreement between Institute of Health Economics Working Paper 04-02

7

predicted and observed was also low the RAND-12 summary scores, the number of medical conditions, and number of prescription drugs. DISCUSSION Generic measures of health status and HRQL life are especially relevant for application in populations of patients with diverse physical and mental health problems. The high-risk primary care patient cohort recruited into this study is a relevant example. Although disease-specific measures of HRQL might have provided important information on the health status of selected patients, such measures would be inappropriate for generating summary scores at the group level. Results from this study provide evidence that the RAND-12, HUI2, and HUI3 have reasonable construct validity in such settings. The relationships observed among measures were generally consistent with a priori predictions. It should be noted, however, that the correlations between overall HUI scores and the number of medications and number of medical conditions were negligible or weak. Similarly, the correlations between the RAND and SF summary scores and the same variables were weak or negligible. Interpretation of these results is challenging. The number of prescription medications and number of medical conditions are widely used as surrogate measures of health status. In fact, this led to our a priori hypotheses were that we would observe moderate or weak correlations. Are the numbers of medications and conditions less useful than previously believed? Or did the study have insufficient variability (or power) to detect the hypothesized associations? It is important to note that taking three or more prescription medications (53% of patients qualified for enrollment in the study by this criterion; criteria were not mutually exclusive) and having poorly controlled chronic conditions (91% of patients) were among the inclusion criteria for recruiting patients into the study. The mean (standard deviation; minimum; maximum) number of prescription drugs was 5.0 (2.5; 0; 17) and mean number of medical conditions was 4.5 (1.4; 1; 6). Although the inclusion criteria may have censored out patients taking few prescription medications or having few medical conditions, it does not seem as if insufficient variability accounts for the low degree of association observed.

8


One possible explanation for the low correlations observed for the number of medical conditions is the lack of information on severity. Several medical conditions with minor burdens may represent a lower overall burden than one more severe medical condition. The number of medications may be related to severity because people with more severe conditions are often given more medications. The low correlations between RAND-12 and HUI3 scores and the number of prescription medications remains puzzling. One can speculate that the number of prescription medications somewhat analogously may be less informative than previously thought. One possibly relevant factor is that some of the prescription medications taken by patients in the study were for conditions for which there were few, if any, symptoms or burdens. Examples include anti-hypertensive and contraceptive medications; 43.2% of patients were taking anti-hypertensives while only 1.5% were taking contraceptives. Additional examination of the evidence in primary-care setting of the relationships among measures of health status, HRQL, the number of medications, and number of medical conditions appears to be warranted. Do sample sizes have to be large for the number of prescription medications or number of medical conditions to be reliable indicators of burden? The RAND-12, HUI2, and HUI3 appeared to accommodate possible interactions between physical and mental health problems more readily than did SF-12. Further investigation of the measurement properties of HUI2, HUI3, and the RAND-12 in the context of primary care appears to be warranted. Results reported here are consistent with results from previous studies using generic measures in primary-care settings (Edelman et al. 1999; Lubetkin et al. 2002, 2003; Macran et al. 2003; Rubenstein et al. 1995; Wasson et al. 1999; Welch 1999); but see also McHorney and Tarlov 1995; an oncology application is reported in Detmar et al. 2002. Macran et al. 2003 report that ceiling effect issues were less frequent with HUI and SF-12 than with the EuroQol EQ-5D. Lubetkin and Gold 2002 report less frequent ceiling effect problems with HUI than with SF-12 and EQ-5D. They also report that respondents felt that HUI made the most sense to respondents relative to the other two measures in terms of how they think about their health. Several study limitations should be noted. First, the analyses are for cross-sectional baseline Institute of Health Economics Working Paper 04-02

9

data only and provide no evidence on construct validity at other points of assessment or the ability of these instruments to capture important change in health status over time. (Results in Farris et al. indicate that the health status of the cohort of patients was maintained over time, so the study provided little opportunity to assess responsiveness.) Second, data was missing for 45 patients who, on average, probably experienced higher burdens of morbidity than those for whom we had complete data for all four instruments. It is useful to compare the extent of agreement between observed relationships and a priori predictions in this study with results reported for other studies. In a study examining the construct validity of HUI2, HUI3, and SF-36 in elective total hip arthroplasty (THA), Blanchard et al. 2004, using the same scheme for the interpretation of the degree of association as was used in this study, report a 75% (65 of 87) success rate for a priori predictions. Several factors may be responsible for their higher success rate observed. First, the patients in the THA study were a more homogenous group; all had osteoarthritis sufficiently severe to make them candidates for THA. In contrast, patients in the high-risk primary care study had diverse physical and mental health problems. Second, patients in the THA cohort were well known to two of the five investigators who formulated the a priori predictions. In contrast in this primary care study none of the investigators had provided direct care to the patients being studied. A priori predictions specified by Juniper et al. 1996 about cross-sectional relationships in the context of pediatric asthma were correct in 5 of 9 (55.6%) of cases. In a related paper Juniper et al. 1996 on caregiver=s burden for parents of pediatric asthmatics, predictions were correct in 3 of 6 cases (50%). The success rate with predictions results reported in this study is consistent with those reported in several other studies. A priori predictions were particularly weak for pain. Moderate relationships between HUI2 and HUI3 pain scores and the physical health summary scores (PHC, PCS) were expected but observed in only one case; instead strong associations were observed in three cases. One might speculate that the high prevalence of pain problems in this patient population (60% reported taking over the counter pain relief medications and 30% reported taking prescription pain relief medications, Table 2) and the association of pain with other aspects of physical and mental health implied that the pain burden was reflected not only in the pain item on the SF/RAND-12 10


but other items as well, generating a higher than expected correlation. In conclusion, the results reported here indicate that RAND-12, HUI2, and HUI3 have adequate construct validity in high-risk primary care patients. Of course, evidence on test-retest reliability among stable patients and responsiveness among patients who experience important change is also relevant and important. Further investigation of these three instruments in primary-care settings appears to be warranted.


11

REFERENCES 1. Birbeck GL, Kim S, Hays RD, Vickery BG Quality of Life Measures in Epilepsy. How Well Can They Detect Change over Time? Neurology 2000; 54: 1822-1827. 2. Blanchard C, Côté I, Feeny D. Comparing Short Form and RAND Physical and Mental Health Summary Scores: Results from Total Hip Arthroplasty and High-Risk PrimaryCare Patients. Int J Technol Assess Health Care 2003; 19:4: 1-7, in press. 3. Blanchard C, Feeny D, Mahon JL, Bourne R, Rorabeck C, Stitt L, Webster-Bogaert S. Is the Health Utilities Index Valid in Total Hip Arthroplasty Patients? Qual Life Res 2004; 13: 339-348. 4. Coons SJ, Rao S, Keininger DL, Hays RD. A Comparative Review of Generic Qualityof-Life Instruments. Pharmacoeconomics 2000; 17: 13-35. 5. Côté I, Farris KB, Feeny D, Johnson JA, Tsuyuki RT, Dieleman SL, Brilliant S, Bayne R, Gardiner L, Moores D, Sandilands M. AUsing Multi-Disciplinary Teams to Improve Primary Care: Quality of Medication Use in the Community,@ Institute of Health Economics Working Paper No. 02-01, February, 2002. 6. Cunningham WE, Nakazono TT, Tsai KL, Hays RD. Do differences in methods for constructing SF-36 physical and mental health summary measures change their association with chronic medical conditions and utilization? Qual Life Res 2003; 12: 1029-1035. 7. Detmar SB, Muller MJ, Schornagel JH, Wever LDV, Aaronson NK. Health-Related Quality-of-Life Assessments and Patient-Physician Communication: A Randomized Controlled Trial. J Am Med Assoc 2002; 288: 3027-3034. 8. Edelman D, Williams R, Rothman M, Samsa GP. A Comparison of Three Health Status Measures in Primary Care Outpatients.@ J Gen Intern Med 1999; 14: 759-762. 9. Farris K, Côté I, Feeny D, Johnson JA, Tsuyuki RT, Brilliant S, Dieleman SL. Using Multi-Disciplinary Teams to Enhance Primary Health Care: A Demonstration Project. Can Fam Physician, forthcoming. 10. Feeny D, Furlong W, Barr RD, Torrance GW, Rosenbaum P, Weitzman S. A Comprehensive Multiattribute System for Classifying the Health Status of Survivors of Childhood Cancer. J Clin Oncol 1992; 10: 923-928. 11. Feeny D, Torrance GW, Furlong W. Health Utilities Index. In: Spilker B. Ed., Quality of life and Pharmacoeconomics in Clinical Trials. Philadelphia, PA: Lippincott-Raven; 1996: 239-251. 12. Feeny D, Furlong W, Torrance GW, Goldsmith CH, Zhu Z, DePauw S, Denton M, Boyle 12


M. Multi-Attribute and Single-Attribute Utility Functions for the Health Utilities Index Mark 3 System. Med Care 2002; 40: 113-128. 13. Furlong WJ, Feeny DH, Torrance GW, Barr RD. The Health Utilities Index (HUI) System for Assessing Health-Related Quality of Life in Clinical Studies. Ann Med 2001; 33: 375-384. 14. Guyatt GH, Berman LB, Townsend M, Chambers L. A Measure of Quality of Life for Clinical Trials in Chronic Lung Disease. Thorax 1987; 42: 773-778. 15. Guyatt GH, Feeny DH, Patrick DL. Measuring Health-Related Quality of Life. Annals of Internal Medicine 1993; 118: 622-629. 16. Hays RD, Sherbourne CD, Mazel RM. The RAND 36-Item Health Survey 1.0. Health Econ 1993; 2: 217-227. 17. Hays RD. R36 H.S.I. RAND 36 - 36 Health Status Inventory. Orlando: The Psychological Corporation/Harcourt Brace & Company; 1998. 18. Hays RD, Morales LS. The RAND-36 Measure of Health-Related Quality of Life. Ann Med 2001; 33: 350-357. 19. Hopman WM, Towheed T, Anastassiades T, Tenenhouse A, Poliquin S, Berger C, Joseph L, Brown JP, Murray TM, Adachi JD, Hanley DA, Papadeimitropoulos E. Canadian Normative Data for the SF-36 Health Survey. Canadian Multicentre Osteoporosis Study Research Group. Can Med Assoc J 2000; 163: 265-271. 20. Johnson JA, Maddigan SL. Performance of the RAND-12 and SF-12 Summary Scores in Type-2 Diabetes. Qual Life Res, 2004; 13: 449-456. 21. Juniper EF, Guyatt GH, Feeny DH, Ferrie PJ, Griffith LE, Townsend M. Measuring Quality of Life in the Parents of Children with Asthma." Qual Life Res 1996; 5: 27-34. 22. Juniper EF, Guyatt GH, Feeny DH, Ferrie PJ, Griffith LE, Townsend M. Measuring Quality of Life in Children with Asthma. Qual Life Res 1996; 5: 35-46. 23. Lubetkin EI, Gold MR. Comprehensibility of Measures of Health-Related Quality of Life in Minority and Low-Income Patients. J Natl Med Assoc 2002; 94: 327-335. 24. Lubetkin EI, Gold MR. Areas of decrement in health-related quality of life (HRQL): comparing the SF-12, EQ-5D, and HUI3. Qual Life Res 2003; 12: 1059-1067. 25. Macran S, Weatherly H, Kind P. Measuring Population Health: A Comparison of Three Generic Health Status Measures. Med Care 2003; 41: 218-231. 26. McDowell I, Newell C. Measuring Health: A Guide to Rating Scales and Institute of Health Economics Working Paper 04-02

13

Questionnaires. Second Edition. New York: Oxford University Press; 1996. 27. McHorney CA, Tarlov AR. Individual-Patient Monitoring in Clinical Practice: Are Available Health Status Surveys Adequate? Qual Life Res 1995; 4: 293-307. 28. McHorney CA, Bricker DE. A Qualitative Study of Patients= and Physicians= Views about Practice-Based Functional Health Assessment. Med Care 2002; 40: 1113-1125. 29. Morisky DE, Green LW, Levine DM. Concurrent and Predictive Validity of a SelfReported Measure of Medication Adherence. Med Care 1986; 24: 67-74. 30. Nortvedt MW, Riise T, Myhr K-M, I. Nyland HI. Performance of the SF-36, SF-12, and RAND-36 Summary Scales in a Multiple Sclerosis Population. Med Care 2000; 38: 1022-1028. 31. Revicki DA, Osoba D, Fairclough D, Barofsky I, Berzon R, Leidy NK, Rothman M. Recommendations on Health-Related Quality of Life Research to Support Labeling and Promotional Claims in the United States. Qual Life Res 2000; 9: 887-900. 32. Rubenstein LV, McCoy JM, Cope DW, Barrett PA, Hirsch SH, Messer KS, T. Young RT. Improving Patient Quality of Life with Feedback to Physicians about Functional Status.J Gen Intern Med 1995; 10: 607-614. 33. Simon GE, Revicki DA, Grothaus L, Vonkorff M. SF-36 Summary Scores. Are Physical and Mental Health Truly Distinct? Med Care 1998; 36: 567-572. 34. Taft C, Karlsson J, Sullivan M. Do SF-36 Summary Component Scores Accurately Summarize Subscale Scores? Qual Life Res 2001; 10: 395-404. 35. Torrance GW, Feeny DH, Furlong WJ, Barr RD, Zhang Y, Wang Q. Multi-Attribute Preference Functions for A Comprehensive Health Status Classification System: Health Utilities Index Mark 2. Med Care 1996; 34: 702-722. 36. Ware JE, Sherbourne CD. "The MOS 36-Item Short Form Health Survey (SF-36)." Med Care 1992; 30: 473-483. 37. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey manual and interpretation guide. Boston: New England Medical Center, The Health Institute; 1993. 38. Ware JE. The SF-36 Health Survey. In Spilker B. Ed., Quality of Life and PharmacoEconomics in Clinical Trials. Second Edition. Philadelphia: Lippincott-Raven Publishers; 1996: 337-345. 39. Ware JE, Kosinski M, Keller SD.AA 12-Item Short-Form Health Survey: Construction of Scales and Preliminary Tests of Reliability and Validity.@ Med Care 1996; 34: 220233. 14


40. Wasson JH, Stukel TA, Weiss JE, Hays RD, Jette AM, Nelson EC. A Randomized Trial of the Use of Patient Self-Assessment Data to Improve Community Practices. Eff Clin Pract, Vol. 2, No. 1, January/February, 1999, pp 1-10. 41. Welch HG. Simple Strategies, Complex Issues. Eff Clin Pract 1999; 2: 44-46. 42. Wilson D, Parsons J, Tucker G. The SF-36 Summary Scales: Problems and Solutions. Sozial- und Praventivmedzin 2000; 45: 239-246.


15

Table 1. Summary of A Priori Predictions about Relationships Between SF-12 and RAND-12 Physical and Mental Health Summary Scores and HUI2 and HUI3 Single-Attribute and Overall Utility Scores, Number of Medical Conditions, and Number of Prescription Drugs

Pairs

Prediction

HUI2 Sensation - PCS, PHC

N, N

HUI2 Sensation - MCS, MHC

N, N

HUI2 Mobility - PCS, PHC

M, M

HUI2 Mobility - MCS, MHC

N, N

HUI2 Emotion - PCS, PHC

W, W

HUI2 Emotion - MCS, MHC

S, S

HUI2 Cognition - PCS, PHC

W, W

HUI2 Cognition - MCS, MHC

W, W

HUI2 Self-Care - PCS, PHC

W, W

HUI2 Self-Care - MCS, MHC

W, W

HUI2 Pain - PCS, PHC

M, M

HUI2 Pain - MCS, MHC

W, W

Overall HUI2 - MCS, MHC

W, W

Overall HUI2 - PCS, PHC

M, M

16


HUI3 Vision - PCS, PHC

N, N

HUI3 Vision - MCS, MHC

N, N

HUI3 Hearing - PCS, PHC

N, N

HUI3 Hearing - MCS, MHC

N, N

HUI3 Speech - PCS, PHC

N, N

HUI3 Speech - MCS, MHC

N, N

HUI3 Ambulation - PCS, PHC

M, M

HUI3 Ambulation - MCS, MHC

N, N

HUI3 Dexterity - PCS, PHC

W, W

HUI3 Dexterity - MCS, MHC

N, N

HUI3 Emotion - PCS, PHC

W,W

HUI3 Emotion - MCS, MHC

S, S

HUI3 Cognition - PCS, PHC

W, W

HUI3 Cognition - MCS, MHC

W, W

HUI3 Pain - PCS, PHC

M, M

HUI3 Pain - MCS, MHC

W, W

Overall HUI3 - MCS, MHC

W, W

Overall HUI3 - PCS, PHC

M, M


17

Number of Medical Conditions - PCS, PHC

M, M

Number of Medical Conditions - MCS, MHC

W, W

Number of Medical Conditions - Overall HUI2,

M, M

HUI3 Number of Prescription Drugs - PCS, PHC

M, M

Number of Prescription Drugs - MCS, MHC

W, W

Number of Prescription Drugs - Overall HUI2,

M, M

HUI3

Note: PCS = Physical Component Summary Score, SF-12; PHC = Physical Health Composite score, RAND-12; MCS = Mental Health Component Summary Score, SF-12; MHC = Mental Health Composite Score, RAND-12; S = Strongly Correlated (> 0.50); M = Moderately Correlated (0.35 -0.50); W = Weakly Correlated (0.20 - 0.34); and N = Negligible or not Correlated (0.00 - 0.19).

18


Table 2. Summary of Demographic and Clinical Characteristics of Patients at Baseline (n=154) N

Percent (%)

Age (years): mean ∀ 1 SD Sex, male

67 ∀ 17 75

49

Marital Status Single

65

42

Married / partner

89

58

Born-country Canada

129

84

First language English

137

89

Highest level of education completed Less than high school

30

20

High school / certificate / tech.

97

63

Baccalaureate university degree or higher

27

18

Retired

75

49

Recovering from illness

25

16

Caring for family

20

13

Working

11

7

Caring and working

11

7

6

4

Primary activity

Other Institute of Health Economics Working Paper 04-02

19

Missing

2

1

Income 0.50); Moderately Correlated (0.35 - 0.50); W = Weakly Correlated (0.20 - 0.34); and N = Negligible or not Correlated (0.00 - 0.19)

26


THE CONSTRUCT VALIDITY OF THE RAND-12 AND ... - CiteSeerX

THE CONSTRUCT VALIDITY OF THE RAND-12 AND ... - CiteSeerX

Suggest Documents