Platform evaluation and lessons learned with IT-supported patient recruitment

Recruiting participants for clinical trials is a challenge, often limited by inefficient methods for identifying eligible candidates. Clinical trial recruitment support systems (CTRSS) can enhance recruitment efficiency by leveraging data from electronic health records (EHR). We evaluated two CTRSS-recruIT and KAS+-to quantify their real-world impact on enrollment. Within the German MIRACUM consortium, we conducted a multicenter segmented interrupted time-series (ITS) evaluation across six university hospitals. Twenty-nine trials contributed 814 observation weeks spanning a control phase (usual practice) and an intervention phase (CTRSS active). Weekly enrollments were compared using segmented negative binomial mixed models estimating level and slope changes at implementation, complemented by Wilcoxon signed-rank tests. A criteria revision covariate captured iterative refinement of computable cohort definitions. In pooled ITS analyses, we observed neither a statistically significant immediate level change nor a statistically significant post-implementation slope change in weekly enrollments. Effects were highly heterogeneous across trials and specialties; implemented eligibility logic most often relied on diagnosis, age, and visit data, while laboratory values and structured observations were seldom available. Higher values of the criteria revision covariate were associated with higher recruitment in some settings, with the most exploratory signal observed at the single KAS+ site. Despite the potential benefits of CTRSS, their effectiveness depends on careful customization and implementation. Overall, our findings suggest that CTRSS impact is context-dependent and that measurable benefit is more likely driven by iterative computable phenotyping and local workflow fit rather than a simple "switch-on" effect. Context-specific adaptation and continual revision of eligibility logic appear critical to realizing benefit. Future studies should reduce trial heterogeneity, strengthen compliance incentives, and employ granular outcomes, such as identification-to-enrollment latency, to better capture workflow-mediated effects.