Purpose: Assessing the impact of health conditions in national surveys is important for policy-relevant cost-effectiveness analyses. We compare estimates of the impacts of eleven chronic health conditions in the same survey respondents using six commonly used indexes of health-related quality of life (HRQoL).
Method: The National Health Measurement Study is a cross-sectional telephone survey of 3844 US adults aged 35-89 that includes the items required to calculate the EQ-5D, HALex, HUI2, HUI3, SF-6D, and QWB-SA. Respondents were also asked whether they had ever been diagnosed with each of eleven different health conditions: sleep disorder, stroke, depression, back pain due to herniated disk, coronary heart disease (CHD), respiratory disease (COPD), ulcer, arthritis, diabetes, thyroid condition, and eye disease. Data were stratified into ages 35-64 and 65-89 years. Within each age stratum, we performed survey-weighted regression analyses for each index using indicator variables for all eleven health conditions, with age and gender as independent variables in the equation. Coefficients and associated standard errors for each condition represent condition impacts adjusted for other conditions, age, and gender.
Result: Condition impacts for the conditions vary by age strata and index. In the younger age stratum, the largest impacts were for herniated disk (mean impact = -0.11) and depression (-0.10); in the older age stratum, the largest impacts were for stroke (-0.10) and depression (-0.10). Smallest impact was for thyroid disorder (-0.01) in younger, and ulcer (-0.004) in older respondents. Within the conditions with large impacts, the ratio of maximum impact to minimum impact across the indexes ranged from 1.4 (arthritis) to 5.3 (COPD). Impacts measured by the HUI3 are generally largest and those for SF-6D smallest. For example, the estimated loss in undiscounted QALYs over 10 years for depression in those aged 65-89 ranges from 0.68 QALYs using the QWB-SA to 1.73 QALYs using the HUI3. Results are limited by lack of information on the severity of the self-reported conditions.
Conclusion: Health condition impact estimates can vary substantially across commonly used HRQoL indexes. This probably is due to both different scaling and differential sensitivity to dimensions of health across the indexes. Therefore, computed QALYs in reference case cost-effectiveness analyses may differ substantially by the HRQoL index used to measure outcomes.
Candidate for the Lee B. Lusted Student Prize Competition