This observational study takes advantage of a natural experiment and comprehensive health care data in the province of British Columbia to examine the effects, intended and unintended, of a policy of differential cost-sharing (reference pricing) for prescription drugs implemented on Jan 1, 1997. Two classes of drugs, ACE inhibitors (ACEIs) and calcium-channel blockers (CCBs), are studied. Outcomes include utilization of several categories of health care, total costs of care, and adverse outcomes (cardiovascular disease hospitalizations and mortality). The underlying study population includes approximately 445,000 persons aged 65 years of age and over on Jan 1 1995. Of these, about 160,000 appear to have received one of the cost-sharing drugs during the baseline period and would therefore be included in the analyses. Interrupted time-series analyses are proposed to control for secular trends in utilization and event rates among persons. Three control groups will also be studied: historical controls (persons using the same drugs, but identified and followed beginning one year earlier); concurrent controls on a non-cost-sharing drug (diltiazem); and external controls, drawn from users of the same cost-sharing drugs but from another province, Ontario, where cost-sharing was not introduced. Primary analyses are conducted for the entire population of users of these medications. For several policy-related reasons, relatively small fractions of these patients (slightly less than 25%) appear to have switched to low cost alternative ACEIs or CCBs or to have stopped their previously used medications altogether. Thus, effects on utilization or increasing rates of adverse events may be difficult to detect in the entire population. Secondary analyses are therefore proposed that will compare post-implementation rates in """"""""switchers"""""""" versus """"""""non-switchers"""""""". The investigators recognize that these secondary analyses introduce the strong likelihood of confounding by indication (switchers are very likely to differ in risk from non-switchers) and they propose to use comorbidity adjusters (a pharmacy-based chronic disease score, and a diagnosis-based casemix adjustment measure), and multiple patient and physician characteristics. These confounders will be adjusted for using a propensity score (propensity to switch or stop the cost-sharing drug). Additional analyses will examine possible changes in prescribing patterns for persons initiating ACEI or CCB treatment.
Showing the most recent 10 out of 34 publications
