ESTIMATING PREFERENCE WEIGHTS IN OBESITY USING THE EQ-5D AND SF-6D

Purpose:

The literature is silent on preference weights (utilities) for obese patients, yet these would be useful when conducting economic evaluations. The purpose of this research is to use a nationally representative dataset to estimate utilities for levels of obesity using the EQ-5D and SF-6D.

Methods:

We used the 2003 Medical Expenditure Panel Survey data to estimate utilities among the adult obese population. The dataset includes one health-state utility measure, the EuroQoL (EQ-5D); and one measure of health status/quality of life, the Short-Form 12 (SF-12). To calculate each individual’s SF-6D utility from the SF-12, we used the published algorithm, which incorporates 7 of the 12 items. Each utility represents the individual’s health state, scored on a scale where death is equal to zero and perfect health is equal to one. We classified subjects into six obesity categories based on Body Mass Index (BMI): underweight, normal, overweight, class I, class II, and class III obesity; and estimated the mean utility within each category. We also examined subgroup variability among the obese population. Twenty-four subgroups were defined based on four age groups, three races, and gender. Finally, we ran two separate multivariate linear regression models, each with one of the two estimated utilities as the dependent variable and the six BMI categories as the primary independent variables of interest. We adjusted for potential confounding factors to assess the impact of various predictors on the utilities.

Results:

The mean utility score among the obese population (BMI≥30) was 0.785 (0.781-0.790) from the SF-6D and 0.837 (0.832-0.843) from the EQ-5D. Throughout these analyses, the EQ-5D provided consistently higher utilities than the SF-6D. The correlation between the two measures was strong. (q=0.760) The subgroup analysis revealed that the utilities were significantly higher in younger age groups, males, and Hispanics. Compared to individuals with normal weight, the results suggest that the utilities for those with class I obesity were decreased by 0.01 (p=0.000), while for those with class III obesity utilities decreased by 0.05 (p=0.000).

Conclusion:
This study provides two single utility measures for categories of obese population, stratified by age, gender, and race. These utilities will be useful when conducting economic evaluations of obesity-specific interventions; and can be used in league tables that compare utilities of patients with other diseases.