Mediation analysis with a time-to-event outcome: a review of use and reporting in healthcare research

Abstract

BackgroundMediation analysis tests whether the relationship between two variables is explained by a third intermediate variable. We sought to describe the usage and reporting of mediation analysis with time-to-event outcomes in published healthcare research.MethodsA systematic search of Medline, Embase, and Web of Science was executed in December 2016 to identify applications of mediation analysis to healthcare research involving a clinically relevant time-to-event outcome. We summarized usage over time and reporting of important methodological characteristics.ResultsWe included 149 primary studies, published from 1997 to 2016. Most studies were published after 2011 (n = 110, 74%), and the annual number of studies nearly doubled in the last year (from n = 21 to n = 40). A traditional approach (causal steps or change in coefficient) was most commonly taken (n = 87, 58%), and the majority of studies (n = 114, 77%) used a Cox Proportional Hazards regression for the outcome. Few studies (n = 52, 35%) mentioned any of the assumptions or limitations fundamental to a causal interpretation of mediation analysis.ConclusionThere is increasing use of mediation analysis with time-to-event outcomes. Current usage is limited by reliance on traditional methods and the Cox Proportional Hazards model, as well as low rates of reporting of underlying assumptions. There is a need for formal criteria to aid authors, reviewers, and readers reporting or appraising such studies.Electronic supplementary materialThe online version of this article (10.1186/s12874-018-0578-7) contains supplementary material, which is available to authorized users.

An overview of systematic reviews found suboptimal reporting and methodological limitations of mediation studies investigating causal mechanisms

“…[81,82]. Our findings and those of others [78,80] suggest that existing guidelines such as the CONSORT for randomised controlled trials or STROBE for observational studies do not cover the…”

“…One-third of the systematic reviews reported limitations in synthesising primary mediation studies due to inadequate reporting. Most often, these reviews identified that the primary studies reported insufficient detail about the statistical analysis and did not report key to-event healthcare research was varied and suboptimal, and highlighted the need for formal guidance to improve reporting standards [78,80].…”

“…unique aspects of study design, analysis, and results that should be reported in a mediation study [78,80]. Requirements to justify the development of a reporting guideline have not been precisely defined [83].…”

“…Requirements to justify the development of a reporting guideline have not been precisely defined [83]. However, our overview has shown that inadequate reporting of primary mediation studies can stifle systematic reviews; and the accumulating evidence of inadequate reporting of mediation studies [75][76][77][78][79][80], suggests that a guideline may help improve reporting standards.…”

Performance of methods to conduct mediation analysis with time‐to‐event outcomes

“…Several authors have discouraged the use of the Cox PH model for traditional mediation analysis (Lange & Hansen, 2011;Lapointe-Shaw et al, 2018;Martinussen, Vansteelandt, Gerster, & Hjelmborg, 2011;VanderWeele, 2011) because the product and difference methods yield different indirect effect estimates when based on the Cox PH model (Gelfand et al, 2016;Tein & MacKinnon, 2003). These two methods are assumed to approximate each other only in case of rare outcomes (VanderWeele, 2011).…”