Fig. 3

Adjusted disutility of disease. Notes: Adjusted disutility in the figure is predicted disutility from the model averaged across all individuals in the NHWS, with the access and utilization variables set to their observed values. Error bars are 95% confidence intervals. Standard errors were calculated by using the simulation-based method described in Mandel et al. [30] In particular, coefficients from the regression model were simulated 1000 times using an asymptotic normal distribution. For each of the 1000 simulations, disutility was predicted for all survey respondents with a given medical condition. Mean disutility was then calculated across individuals for each draw and the mean, 2.5% quartile, and 97.5% quartile of mean disutility were used to generate the point estimates and 95% confidence intervals in the plot