A simple clinical–laboratory risk score for preoperative identification of ovarian torsion: a retrospective diagnostic study

Timely diagnosis of ovarian torsion remains challenging due to nonspecific clinical presentation and limitations of imaging. Delayed recognition may result in ovarian loss. Simple, readily available tools to support early risk stratification are therefore needed. In this retrospective diagnostic study, we analyzed women presenting with pelvic pain and an adnexal mass at a tertiary center between 2015 and 2023. Patients with surgically confirmed ovarian torsion ( n = 62) were compared with non-torsion controls ( n = 126). Candidate predictors were evaluated using multivariable logistic regression. A clinical–laboratory risk score was developed based on independent predictors and assessed for discrimination and calibration. Internal validation was performed using bootstrap resampling. Ovarian mass size, eosinophil count, and MPV remained independent predictors of ovarian torsion. Together with nausea/vomiting, these variables were incorporated into a 5-point ovarian torsion risk score (OTRS). The model demonstrated good discriminative performance (area under the curve [AUC] 0.912, 95% CI 0.871–0.953). At a threshold of ≥ 4 points, the score yielded high specificity (98.4%) and a strong positive likelihood ratio (LR + 29.25), consistent with a rule-in diagnostic tool, although sensitivity was limited (46.8%). Internal validation suggested stable model performance. This simple clinical–laboratory risk score, based on routinely available parameters, may assist clinicians in identifying patients at high risk of ovarian torsion and support timely surgical decision-making, particularly in diagnostically challenging cases. The score appears most useful as a rule-in aid rather than an exclusion tool. However, the model should not be used as a standalone diagnostic tool, and external validation in independent cohorts is required before clinical implementation.