Early detection of the existence or absence of treatment effect: a cumulative meta-analysis

Abstract Objective An unexpected promising effect of low-molecular-weight heparins (LMWHs) on survival in cancer patients was observed in early trials in post-hoc subgroup analyses but not found in more recent trials. In order to highlight a possible regression over time toward the lack of antitumoral effect of LMWHs, we performed a cumulative meta-analysis of survival data from RCTs. Study Design And Setting Medical databases were searched to identify randomized clinical trials comparing LMWHs to placebo or no treatment in cancer patients without VTE, or to VKA in cancer patients with VTE in overall survival. Cumulative hazard ratio (HR) was estimated after each study inclusion in chronological order. Results Twenty-three studies (12,970 patients) were included. The cumulative meta-analysis of the earlier trials showed a significant improvement in overall survival with LMWHs. This apparent benefit then gradually regressed over time toward an absence of effect of LMWHs on survival(HR 0.98 [95% CI, 0.93; 1.03]). Conclusion Despite supportive experimental data and early clinical findings, the promising antitumor effect of LMWHs in cancer patients gradually vanished over time toward a lack of impact on overall survival. This result suggests ‘p-hacking’ and selective reporting of positive results from post-hoc subgroup analyses in the early studies.