Cost-Sharing for Prescription Drugs—Reply

In Reply: We agree with Ms Gibson and colleagues that one must be cautious in generalizing study results to different populations. However, one cannot compare their results with ours and conclude that their data provide evidence that the employed, unlike the poor and elderly, are only transiently affected by prescription cost-sharing. Methodological differences in the approaches taken may have contributed to apparent differences in findings. Our study measured the response to the policy implementation for each person in the study population, whereas the analysis by Gibson et al appears to have used an aggregate measure of prescription consumption for the study population in each quarter. When the underlying population remains constant, results for these 2 forms of analysis should be similar; however, this assumption requires validation. In the results of Gibson et al, the almost flat rate of consumption in the first 7 quarters preceding the co-payment policy followed by variation in utilization rates of 25% in the next 8 quarters suggests that other factors, including a possible change in the underlying population, may have been operative. This could bias the evaluation of the policy change because drug consumption is being compared among different groups.