A validated predictive model for adnexal torsion pre-operative diagnosis

Purpose: To develop a simple predictive model for pre-operative diagnosis of adnexal torsion (AT). Methods: A retrospective cohort study with a retrospective validation, including 669 separate episodes of women who underwent laparoscopy due to a suspected AT between January 2011 and June 2020. We compared the pre-operative characteristics between women with surgically confirmed AT and those without. Results: The derivation cohort included 615 episodes of suspected AT. AT was surgically confirmed in 445 episodes (72%). The retrospectively collected validation cohort included 54 episodes, with 31 (57.4%) surgically confirmed AT. In a multivariate regression analysis, vomiting, neutrophils to lymphocytes ratio > 3.5 and sonographic finding of enlarged ovary were independently associated with AT [OR 95% CI 2.78 (1.21–6.36), 3.15 (1.42–6.97) and 2.80 (1.33–5.88), respectively]. In the derivation cohort, the PPV for AT diagnosis was 69.7%, 84.5% and 93.1% if 1, 2 and 3 risk factors were present, respectively. Retrospective validation analysis underlined a PPV of 67.6%, 82.6 and 66.6% for 1, 2 and 3 risk factors, respectively. Conclusion: We have developed and validated a simple predictive model for pre-operative diagnosis of AT, based on three parameters. Our model may assist clinicians while evaluating patients with suspected AT and improve pre-operative diagnosis.