Evaluation of SSI risk prediction model after spinal surgery: A systematic review and critical appraisal

This study aimed to systematically review and critically evaluate the risk of bias and applicability of surgical site infection (SSI) risk prediction models after spinal surgery. China National Knowledge Infrastructure, Wanfang Database, China Science and Technology Journal Database (VIP), SinoMed, PubMed, Web of Science, The Cochrane Library, Cumulative Index to Nursing and Allied Health Literature, and Embase were searched from inception to April 10, 2025. The prediction model risk of bias assessment tool-artificial intelligence (AI) and transparent reporting of a multivariable prediction model for individual prognosis or diagnosis-AI were used to assess the quality of the included studies, and RevMan software was used to perform a meta-analysis of the odds ratio values for certain model predictors. A total of 37 studies were included, identifying 43 predictive models. The incidence of SSI after spinal surgery ranged from 1.5% to 50%. Among these, 11 studies focused solely on model development, 4 studies included external validation, 22 studies were only internally validated, and 1 study was both internally and externally validated. The area under the curve values ranged from 0.610 to 0.991. The meta-analysis of high-frequency predictors identified statistically significant factors, including diabetes, age, surgery duration, albumin, body mass index, drainage time, smoking history, and American Society of Anesthesiologists score. All studies were rated as having a high risk of bias, primarily due to poor reporting related to study participants and the analysis domain. The evaluation using the prediction model risk of bias assessment tool indicated a considerable risk of bias in current predictive models for postoperative SSI after spinal surgery. Although the predictive model for SSI after spinal surgery is generally acceptable, most studies have methodological flaws. Moreover, studies with larger sample sizes and multicenter external validation are necessary to enhance the robustness of predictive models.