Estimation of treatment effect adjusting for treatment changes using the intensity score method: application to a large primary prevention study for coronary events (MEGA study)

Abstract

The MEGA study was a prospective, randomized, open-labeled, blinded-endpoints study conducted in Japan to evaluate the primary preventive effect of pravastatin against coronary heart disease (CHD), in which 8214 subjects were randomized to diet or diet plus pravastatin. The intention-to-treat (ITT) analysis showed that pravastatin reduced the incidence of CHD (hazard ratio=0.67; 95 per cent confidence interval (CI): 0.49-0.91) and of stroke events, which was the secondary endpoint in the MEGA study (hazard ratio=0.83; 95 per cent CI: 0.57-1.21). Owing to considerable treatment changes, it is also of interest to estimate the causal effect of treatment that would have been observed had all patients complied with the treatment to which they were assigned. In this paper, we present an intensity score method developed for clinical trials with time-to-event outcomes that correct for treatment changes during follow-up. The proposed method can be easily extended to the estimation of time-dependent treatment effects, where the technique of g-estimation has been difficult to apply in practice. We compared the performances of the proposed method with other methods (as-treated, ITT, and g-estimation analysis) through simulation studies, which showed that the intensity score estimator was unbiased and more efficient. Applying the proposed method to the MEGA study data, several prognostic factors were associated with the process of treatment changes, and after adjusting for these treatment changes, larger treatment effects for pravastatin were observed for both CHD and stroke events. The proposed method provides a valuable and flexible approach for estimating treatment effect adjusting for non-random non-compliance.