490 - Diagnostic Test Accuracy of History and Physical Examination for Pediatric Ovarian Torsion: Systematic Review/Meta-analysis

The diagnosis of ovarian torsion is difficult because of its low prevalence.  Delay in its diagnosis increases the risk of gonadal ischemia, increasing the odds of infertility. A timely workup begins with thorough history and physical examination (H&P); yet, the diagnostic test accuracy of an H&P has not been quantified.  This review will describe the diagnostic test accuracy of history and physical examination for ovarian torsion in pediatric patients.

Objective:

To calculate the sensitivity, specificity, likelihood ratios and diagnostic odds ratio

Design/Methods:

We performed a literature search on selected core databases: PubMed, Embase, CINAHL, Web of Science, and Cochrane, starting from year of 1993.  The search was initiated on 23 September 2022.  Our study population was females less than 18 years old who presented with undifferentiated abdominal pain, but were found to be at risk for ovarian torsion. Subjects were recruited from the emergency department or outpatient clinics.  Those females presenting with current pregnancy, recent miscarriage or abortion, or abuse or trauma were excluded. Ovarian torsion was diagnosed via surgical exploration.  We employed bivariate analysis to calculate our diagnostic test parameters.  Our review was registered with PROSPERO: CRD42021252224.

Results:

We screened 2602 articles, from which we identified 16 articles meeting our inclusion criteria.  Vomiting was the only element with statistical significance, where sensitivity (Sn) was 73%, specificity (Sp) was 68%, positive likelihood ratio (LR+) was 2.3, and negative likelihood ratio (LR-) was 0.40.  Contrary, nausea had poor diagnostic accuracy, where Sn was 62%; Sp was 54%; LR(+) was 0.72; and LR(-) was 0.72.  Pain characteristics, whether intermittent, diffuse, or peritoneal irritation, had poor diagnostic test accuracy, where their LR(+)s were 1.25, 0.88, and 1.50, respectively.  Their LR(-)s values were equaling approximately one.  Lastly, menarchal status was non-diagnostic, where its LR(+) and LR(-) were equaling approximately one.

Conclusion(s):

Among individual diagnostic elements, only vomiting was associated with a small increase in diagnostic accuracy.  None of the abdominal pain symptoms of intermittent, diffuse, or peritoneal had clinical diagnostic accuracy.  No single history or exam element would rule-in or rule-out ovarian torsion.