P–344 Clarifying tubo-ovarian abscess management: a risk score for predicting antibiotic failure

Abstract

Abstract Study question Can antibiotic treatment failure of tubo-ovarian abscesses (TOA) be predicted based on clinical features at the time of diagnosis? Summary answer We propose a risk score including patient temperature, c-reactive protein and TOA size that could predict which patients are likely to fail parental antibiotic treatment. What is known already Current guidance is that the first line management of non-ruptured TOA is with parental antibiotics. However, it is reported that treatment failure rate is 20–30%. Alternative treatment modalities include radiological drainage or laparoscopic/open surgery. In patients who require intervention, outcomes, such as morbidity, length of hospital stay and fertility, are improved when this is performed early rather than later in their hospital admission. However, our current guidance is scant with regards to the decision making for interventional TOA management. Study design, size, duration This is a multicentre retrospective cohort study over 81 months (01/01/13- 30/09/19) identifying 214 consecutive patients admitted to hospitals in North-West London with diagnosed TOA. Participants/materials, setting, methods: Demographics, medical history, presenting symptoms, laboratory results, radiological findings, treatments administered, hospital length of stay and follow up data was collected. The patients were chronologically split with the first 150 being used for the development of our risk score. Univariate and bivariate analyses were employed to ascertain statistically significant variables in the failure of parental antibiotic. The remaining 64 patients were used for risk score validation. Main results and the role of chance Statistically significant variables were: temperature at admission (median= 37.1 °C vs 38.2 °C, p = 0.0001), C-reactive protein (CRP) at admission (151mg/L vs 243mg/L, p = 0.0001) and size of TOA (6.0cm vs 8.0cm, p = 0.0001). Those requiring intervention, stayed in hospital twice as long as those who did not (4 days vs 8 days, p < 0.001). A scoring system was formulated using the statistically significant variables. A score of ≥ 4 was associated with requiring radiological/surgical intervention (p < 0.001), with sensitivity 69% and specificity 88% (AUC 0.859) when tested on the validation cohort. Limitations, reasons for caution Being a retrospective study, which puts the data at risk of information and selection bias. Although there are merits to a multi-centre study, variation in patient management will invariably cause data heterogeneity. Wider implications of the findings: TOA patients may have their hospital management tailored early according to the postulated tool, alleviating uncertainty in their treatment as well as possibly reducing morbidity and length of hospital stay. Trial registration number Not applicable