Vervolganalyses

Vervolganalyses, in research, are analyses conducted after the primary analysis or after data collection has ended to investigate additional outcomes, longer-term effects, or performance over time. They may be pre-specified as part of a study protocol or conducted post hoc to explore new questions arising from the data.

Common purposes include assessing durability of treatment effects, long-term safety, progression, or real-world effectiveness in follow-up periods (for example, six months, one year, or multiple years). Methods vary with data: repeated measures and longitudinal analyses (linear mixed models, generalized estimating equations) for continuous or ordinal outcomes; time-to-event or survival analyses (Kaplan-Meier, Cox proportional hazards) for incidents; and mixed-models for trajectory analysis.

Key methodological considerations include handling missing data due to drop-out, censoring in time-to-event data, multiplicity from multiple timepoints or endpoints, and potential biases from non-random loss to follow-up. Pre-specification in the study protocol helps mitigate bias; when not pre-specified, post hoc analyses should be labeled exploratory with cautious interpretation.

Reporting focuses on the follow-up interval, sample size at follow-up, attrition reasons, the analytic method, and the robustness of findings under different assumptions about missing data. Vervolganalyses complement the primary analysis by providing information on longer-term outcomes but are generally considered less definitive without prior planning and adequate follow-up.

Used across fields such as clinical trials, epidemiology, psychology, and health services research, they help determine whether early benefits persist, fade, or change over time and identify late-emerging risks.