Letters Section Editor: Stephen J. Lurie,
MD, PhD, Senior Editor.
To the Editor: Dr Joyce and colleagues1 used a cross-sectional design, in which there were
different participants at each cost-sharing level and in each year. This approach
is significantly less rigorous than longitudinal approaches that study temporal
changes in behavior with concurrent controls.2,3 The
use of the essentially cross-sectional design in this study is particularly
concerning because of potential selection biases from health plan adverse
selection; ie, when patients or employers with greater anticipated health
expenditures deliberately choose plans with lower levels of cost sharing,
and vice-versa. Such biases would amplify any expected cost differences between
plans with varying levels of cost sharing. As the authors acknowledge, the
lack of drug plan choice among employees of employers in this study provides
only limited protection for this type of selection bias.
John Hsu, Mary Reed. Financial Consequences of Drug Benefit PlansFinancial Consequences of Drug Benefit Plans. JAMA. 2003;289(4):423–424. doi:10.1001/jama.289.4.423-a