Objective
To create a risk scoring system comprised of clinical and radiological characteristics that can predict the likelihood of antibiotic treatment failure of tubo‐ovarian abscesses. The score should guide clinicians in identifying patients to whom early intervention should be offered instead of a prolonged trial of antibiotics.
Methods
A multicenter, retrospective cohort study carried out between January 1, 2013 and September 30, 2019, identified consecutive patients with tubo‐ovarian abscess. Using a chronological split, patients were allocated to two groups for the development and subsequent validation of the postulated scoring system. Univariate and bivariate analyses were performed to identify statistically significant variables for the failure of intravenous antibiotic treatment.
Results
In total, 214 consecutive patients with tubo‐ovarian abscesses were identified. Data from the first 150 patients were used for the development of the postulated scoring system; data from the subsequent 64 patients were used for validation. Statistically significant clinical features between those having successful and unsuccessful management were: temperature (median = 37.1℃ vs 38.2℃, P = 0.0001), C‐reactive protein (151 mg/L vs 243 mg/L, P = 0.0001), and tubo‐ovarian abscess diameter (6.0 cm vs 8.0 cm, P = 0.0001). These parameters were used to create a risk prediction score. A score of four or more was predictive of requiring surgical/radiological intervention of tubo‐ovarian abscess (P < 0.001). The score had a sensitivity of 69% and a specificity of 88%, with area under the curve (AUC) = 0.859.
Conclusion
Currently, there is no guidance for clinicians on when to operate on a tubo‐ovarian abscess. Our prediction score is simple, using only three easily obtained clinical characteristics.