Letter to Editor About “Outcomes of clinical decision support systems in real-world perioperative care: a systematic review and meta-analysis”

Abstract

Dear Editor, I read with great interest the recent systematic review and meta-analysis examining the impact of clinical decision support systems (CDSS) on perioperative care[1]. Given the growing adoption of digital tools throughout the surgical pathway, the authors have tackled a timely and important topic. I would, however, like to discuss several points in greater depth. Although the review aims to reflect the “real-world” performance of perioperative CDSS, 75% of the included studies were conducted in the United States, with minimal representation from low- and middle-income countries. This imbalance limits the generalizability of the conclusions and obscures context-specific barriers such as infrastructure constraints, workforce training needs, and interoperability challenges, factors that can substantially affect CDSS effectiveness in resource-limited settings[2]. A more explicit discussion of how this skewed distribution influences the interpretation of the findings and hampers global implementation efforts would strengthen the review. The interventions covered in the review range from simple alert systems to advanced machine-learning-based decision tools; the outcomes assessed also span guideline adherence, medication errors, hemodynamic parameters, postoperative complications, and economic endpoints. Such broad heterogeneity makes it difficult to interpret pooled effect estimates. While the authors appropriately used narrative synthesis when meta-analysis was not feasible, the review would benefit from a more rigorous classification framework for CDSS. This could help clarify which categories of interventions are most, or least, effective in perioperative practice. Most included studies were observational, many of them retrospective, making them susceptible to confounding from clinician behavior, institutional protocols, and surveillance bias. CDSS implementation often coincides with concurrent quality-improvement initiatives, complicating causal attribution[3]. Although the authors applied RoB2 and NOS tools, a deeper examination of how high or unclear risk of bias shapes confidence in the reported improvements in guideline adherence and reductions in medication errors would add value. In the absence of clear sensitivity analyses stratified by study quality, it is difficult to judge the robustness of the positive findings. The review reports improvements in several perioperative process outcomes but no significant reduction in postoperative mortality. This apparent discrepancy merits careful contextualization. Mortality is influenced by a wide array of patient, procedural, and institutional factors, and in short-term or single-center CDSS evaluations, it may be too infrequent or indirect endpoint to detect meaningful differences. The authors should emphasize that validating CDSS impact on such outcomes requires large, adequately powered multicenter trials with standardized endpoints, rather than implying a lack of therapeutic effect. Given the high cost of developing, deploying, and maintaining perioperative CDSS, it is striking that only a small number of included studies examined economic implications. Although the authors call for more research, the review could have further underscored that cost savings observed in isolated settings may not be transferable or scalable across institutions with different baseline practices or financial structures. Overall, this review provides a valuable synthesis of emerging evidence on perioperative CDSS. We appreciate the authors’ contribution to this evolving field. Future research should prioritize multicenter prospective studies, standardized reporting frameworks, and the integration of human-factors and economic analyses. Only through such comprehensive evaluation can the perioperative community accurately determine the role of CDSS in improving patient safety and supporting clinical decision-making. Ethical approval Not applicable. Consent Not applicable.

Ähnliche Arbeiten