* Candidate for the Lee B. Lusted Student Prize Competition
Purpose: This study examined the underlying factor structure of health-related quality of life (HRQoL) domains across five preference-weighted indexes of HRQoL: SF-6D, EQ-5D, HUI2, HUI3 and QWB-SA.
Method: Data came from the National Health Measurement Study (NHMS) a telephone survey of a nationally representative sample of 3844 non-institutionalized adults aged 35-89 residing in the continental US conducted in 2005-06. Data for all five indexes were collected from each survey respondent and arrayed as categorical scores for each domain included in each index (e.g., EQ-5D has 5 domain scores with 3 categories each). We used oblique categorical exploratory factor analysis to explore the underlying dimensions of these data while accounting for the complex survey design of the NHMS. Confirmatory factor analysis was used to assess model fit. Item response theory was applied to explore the amount of information HRQoL domains contribute across the continuum of the underlying latent dimensions.
Result: The first 3 eigenvalues (13.7, 2.2, 1.7) were greater than one, hence 3 main dimensions of HRQoL domains were identified: “physical function,” “psychosocial function” and “pain.” Fourteen out of 26 HRQoL domains contributed to the physical dimension, and 6 domains contributed to each of the other two dimensions. The 3 dimensions were correlated: 0.42 (physical, psychosocial), 0.54 (physical, pain), 0.43 (psychosocial, pain). Average loading on the physical dimension was 0.71 with smallest being 0.26 (HUI3 Hearing) and largest 0.91 (QWB-SA Physical Activity). On the psychosocial dimension, the average loading was 0.77 with smallest 0.59 (HUI3 Cognition) and largest 0.90 (SF-6D Social Function). Average loading on the pain dimension was 0.82; although primarily defined by pain domains, extreme loadings on this factor were 0.62 (SF-6D Vitality) and 0.94 (EQ-5D Usual Activities). This factor structure fit the HRQoL data well (CFI = 0.97, TLI = 0.98, RMSEA = 0.045).
Conclusion: Collectively, HRQoL domains of five commonly-used HRQoL indexes capture 3 underlying latent dimensions of HRQoL, physical and psychosocial function and pain. Some HRQoL domains, in particular domains of sensation on HUI3 (Hearing, Speech, Vision), SF-6D Vitality, and HUI3 Cognition, displayed relatively more unique variance not captured by the 3 main identified dimensions. HRQoL domains contributed most information at the lower to middle range of the latent continuum of dimensions and least information at the high end.
Purpose: Most prominent national EQ-5D valuation studies employ the Time Trade-Off (TTO) preference elicitation method, using sequential choice tasks to model a trade-off between health and longevity. In the US, UK and Danish EQ-5D valuation studies, the first choice task was the most significant, categorizing health states as being worse than death (wtd), equal to death (etd) or better than death (btd). Adding the special case of respondents unwilling to trade (utt) any lifetime to achieve perfect health, valuations fall into four qualitatively distinct categories. The purpose of the study was to test to which extent the final tariffs are determined by these categories; how much is gained by continuing the TTO task after the initial choice.
Method: We performed separate analyses on TTO data from the US, UK and Danish valuation studies using two transformations of negative utilities corresponding to the US and UK methods. Original valuation study exclusion criteria were used. TTO values were collapsed onto four equidistant points corresponding to the four distinct categories (utt, btd, etd, wtd). Averages for each directly measured health state were correlated between full and collapsed TTO values. We calculated Pearson’s correlations between predicted values based on full and collapsed TTO for all 243 health states using both the D1 and N3 regression specifications.
Result: Pearson’s r between averaged health states for full and collapsed TTO were >.998 (p<.001) for all three datasets. Correlations between predicted values for all 243 states based on full and collapsed TTO using the same transformation and regression model were all >.999 (p<.001). Using three national valuation datasets, two different transformations of negative utilities and two different regression specifications, the patterns of distances between averaged health state values were conserved (r2>.998 for all measures) when collapsing the TTO onto four points.
Conclusion: There was no gain from continuing the valuation task after the initial choice. The resulting population tariffs appear to be determined by patterns in the proportion of the population electing each of the four categories. Depending on whether we consider the first choice task alone to be a valid measure of health state severity, the TTO method could be radically simplified, revised or should possibly be abandoned.
Purpose: The person tradeoff is widely used as a measure of health, but unlike the standard gamble does not have a basis in utility theory to support its use. In this study we identify preference conditions in utility theory that guarantee the validity of person tradeoffs. We then test these preference conditions underlying the PTO in an experiment.
Method: One-hundred thirteen students from the University of Southern California and Erasmus University Rotterdam, served as subjects in an experiment that tested the PTO assumption of Marginality (the social decision maker has equal preference between policies that, for each individual, have the same expected benefit) and Additivity (a policy whereby half the persons improve from health state A to B and half from B to C is equal in preference to a policy whereby half move from A to C and the other half make no improvement). Two Marginality and one Additivity question were employed in the experiment.
Result: With respect to Marginality, approximately 40% of the subjects satisfied this condition in each test separately. Seventy percent of the subjects who satisfy marginality in the first test also satisfied marginality in the second test. Consequently, about 25% of the subjects satisfy marginality in both tests. With regard to Additivity, subjects were generally not indifferent between policies for which improvements were additive. A majority of subjects, 70% (p < 0.001) preferred to give some health improvement to everyone rather than give a subset of individuals a large improvement and everyone else no improvement.
Conclusion: The findings of this study put in doubt the PTO requirement of Additivity of preference, and provide only limited support for Marginality. We hope that this paper will help to clarify what is involved in using the PTO and will foster future empirical research into its validity.
Purpose: QALY weights for the EQ-5D have typically been obtained through time trade-off (TTO) surveys using a sub-set (n=17 or n=43) of the 243 health states. Discrete choice experiments (DCEs) are an alternative, potentially more flexible approach. We explore the development of EQ-5D algorithms based on both approaches and the impact of selection of health states in each.
Method: In phase one, respondents recruited via an on-line panel (n=228) were randomly allocated to three DCE designs: A (health states from the original UK EQ-5D valuation set); B (excluded health states which combined Mobility Level 3 was with Usual Activities Level 1 or Self Care Level 1); and C (no restrictions on the selection of health states). A pooled model and separate models for each design were estimated. In phase two an on-line panel sample (n=1000) completed the DCE for the preferred design. Each choice set presented two health profiles (EQ-5D state and survival duration), and death. In phase three, a population sample (n=417) completed a computer based TTO task and the DCE. The TTO incorporated 198 of the 243 health states (the remainder excluded as implausible). Respondents were randomly assigned to value 11 health states and the worst health state. Separate models were estimated for the DCE and the TTO.
Result: Comparison of the models suggest that the different choice experiment designs lead to differences in parameter estimates as well as differences in variance (scale). Response rates were higher for Design A, but there is a considerable loss of statistical efficiency arising from the reduced set of health states included in the experimental design. For the TTO data, models were estimated based on the functional form for the published algorithm for the EQ-5D, and allowing for interaction terms. While relatively few interaction terms are significant, a likelihood ratio test demonstrates that inclusion of interaction terms improves the fit of the model.
Conclusion: While the results for both approaches are broadly consistent with the previously published EQ-5D algorithms, there are some important differences. The DCE approach allows us to explore interactions between health states and duration that cannot be estimated with the TTO approach. In both approaches a design that allows for more extensive coverage of the EQ-5D space is appropriate.
Purpose: The aim of this study was to describe how diabetes complications influence the quality of life of persons with diabetes.
Method: We mailed a questionnaire to 1,000 persons with diabetes type 1 and 2 in
Result: In multivariate analyses the strongest determinants of reduced mobility were neuropathy (OR: type 1 diabetes 10.67, OR: type 2 diabetes 11.44), and ischemic heart disease (OR type 1 diabetes: 22.35, OR type 2 diabetes 2.06). “Fear of hypoglycaemia” was a strong predictor (OR type 1 diabetes: 4.12, OR type 2 diabetes 5.71) of anxiety and depression. For patients without complications, the EQ-5D index was 0.90 (type 1 diabetes) and 0.85 (type 2 diabetes). For patients with complications, the EQ-5D index was 0.68 (type 1 diabetes) and 0.73 (type 2 diabetes). In linear regression the factors with the greatest negative impact on the EQ-5D index were ischemic heart disease (coefficient -0.181 for type 1 diabetes), stroke (– 0.291 for type 1 diabetes and – 0.135 for type 2 diabetes), neuropathy (-0.358 for type 1 diabetes and -0.187 for type 2 diabetes) and fear of hypoglycaemia (-0.078 for type 2).
Conclusion: Diabetes complications may have a considerable impact on several dimensions of quality of life, and the impact may be substantial. The strongest determinants of reduced quality of life in diabetics were ischemic heart disease, stroke and neuropathy. The complexity of the disease means that several dimensions need to be considered when priorities are set for diabetes interventions.
Purpose: Women have longer life expectancies than men at all ages. The epidemiology of many chronic diseases differs between men and women. We ask here whether there are gender differences in health-related quality of life (HRQoL) measured by five HRQoL indexes commonly used in US nationally representative data sets.
Method: We analyzed data on African Americans and Caucasians aged 35-89 from 4 surveys using 1 or more of the indexes: Joint Canada/United States Survey of Health, Medical Expenditure Panel Survey, National Health Measurement Study, and US Valuation of the EuroQol EQ-5D Health States Survey. Gender differences measured by SF-6D, EQ-5D, HUI2, HUI3 and QWB-SA were estimated for indexes common to each dataset and in a pooled dataset using survey-weighted least-squares regression. Differences in HRQoL were estimated with and without adjustment for sociodemographic and socioeconomic (SES) indicators.
Result: Distributions of age, race and education were similar between men and women across the 4 studies. In the age range analyzed, women were twice as likely than men to report being ‘widowed, divorced or separated’ and 1.5 times more likely to be in the lowest income category (<$20,000). Women had lower HRQoL scores than men on all HRQoL measures prior to adjustment for marital status, education and income (p<0.05). The average HRQoL difference (men minus women) was 0.023 before and 0.014 after adjustment. Average differences (favoring men) by index were: 0.027 (SF-6D), 0.018 (EQ-5D), 0.024 (HUI2), 0.024 (HUI3), and 0.03 (QWB-SA). Marital status and income had similar and the largest impact on the estimated differences. Simultaneous adjustment for all covariates substantially reduced the differences with QWB-SA and SF-6D scores being least affected and HUI3 most. Differences between black men and women were much larger than for whites on all measures and persisted after adjustment.
Conclusion: Older women in the