Abstract
OBJECTIVE: Tubo-ovarian abscesses (TOAs) cause significant morbidity. Surgical intervention is required if broad-spectrum intravenous antibiotics are unsuccessful. This study aimed to describe admission characteristics that predict failed medical management and to evaluate a previously developed risk score for predicting the need for surgical intervention in cases of TOA. DESIGN: Single centre, retrospective cohort study. SETTING AND PATIENTS: Patients admitted to a tertiary-level public teaching hospital with a radiologically or surgically proven TOA between January 1, 2012 and December 31, 2018. MEASURES: Demographic and clinical details were obtained from electronic clinical records. Medical treatment was considered "failed" when surgical intervention was required beyond 24 h of antibiotics. Multivariable analyses using logistic regression was used to determine predictors of failed medical management. Risk scores were calculated as per Fouks et al. and a receiver operating characteristic curve was constructed to assess correlation with outcomes. RESULTS: There were 425 patients and 522 admissions with TOA. In the first 24 h, 14% (72/522) of admissions were treated with a surgical intervention in addition to intravenous (IV) antibiotics, while 86% (450/522) were treated with IV antibiotics alone. In those treated with IV antibiotics alone, medical treatment was successful in 65% (293/450) of cases, with 35% (159/450) requiring additional surgical or radiological intervention prior to discharge. Variables independently associated with failed medical treatment were fever at admission (adjusted odds ratio [aOR] 1.72, 95% confidence interval [CI] 1.11-2.67), larger mean diameter of TOA (2% higher odds for every 1-mm increase in abscess size) and higher C-reactive protein value (1% higher odds for every unit increase) at admission. The area under the curve (95% CI) for Fouks et al. scoring system was 0.63 (0.58-0.68), indicating poor discriminatory ability. CONCLUSIONS: A third of TOAs managed medically required surgical intervention. Fever, higher inflammatory markers, and larger mass were predictive of requiring surgery. However, a scoring system using these variables had poor discriminatory ability to predict treatment failure. Prospective studies are needed to determine whether earlier recourse to surgery can improve outcomes.