From January 2015 to June 2019, a historical cohort study was performed on a total of 92 patients (n = 92) admitted to a tertiary west London hospital for TOA. Patients were identified through a search of the hospital’s electronic database, using diagnostic codes for TOA. Patients were included if a TOA was confirmed either radiologically or surgically. A total of 370 cases were reviewed to confirm eligibility, of which 278 patients were excluded as they had past medical history of TOA rather than an active encounter or did not have radiological or surgical confirmation of the diagnosis. Patients who had secondary TOA due to pre-existing intra-abdominal pathology were included in our analysis and discussion.

Following institutional approval from the audit team (Imperial NHS Trust service evaluation project reference number 404), data on demographics, clinical management, investigations, and patient outcomes were extracted from the hospital’s electronic database. As the data was retrospectively collected and did not affect the included patients care no further ethical approval was required. Additionally, consent was waived by the ethics committee as the data collected from the hospital records were anonymized.

Demographic data collected included: age, body mass index (BMI), ethnicity, parity, and presence of comorbidities including endometriosis, intrauterine contraceptive device (IUCD) use, previous PID, immunocompromised state and smoking status (See Underlying data). ¹⁰ TOA data included its size, laterality, high vaginal swab (HSV) cultures and presence of any other intra-abdominal pathology.

Patients’ records were reviewed for signs and symptoms of PID on presentation including lower abdominal pain, abnormal vaginal discharge, fever (defined as a temperature ≥ 38°C), white cell count (WCC) and CRP at admission (See Underlying data). ¹⁰ Microbiology results of any microscopy, culture and sensitivity were also noted.

After confirmation of TOA, all patients were started on the hospital’s broad spectrum intravenous antibiotic regime for PID. Those that did not show improvement or who grew organisms with specific sensitivities were discussed with microbiology and had their antibiotic regimen tailored.

Based upon the management approach, all patients were classified into two groups. Group one consisted of patients treated successfully with intravenous antibiotics only (either short course or long course including OPAT). Successful treatment was defined as good clinical and biochemical response to antibiotics with down trending inflammatory markers. Conversely, Group two consisted of patients requiring drainage in addition to intravenous antibiotics due to lack of improvement or worsening of symptoms or inflammatory markers or recommendation on radiology report. Decision for the route of drainage was made on an individual patient basis by a multidisciplinary team involving gynecologists, radiologists, and general surgeons. The drainage approach included: (i) radiologically guided drainage, or (ii) laparotomy/laparoscopy. Laparotomy or laparoscopy involved a combination of washout, salpingectomy, adnexectomy, hysterectomy with bilateral salpingo-oophorectomy or bowel resection.

The main outcome measures were length of hospital stay (LoS), length of antibiotic treatment (LoA) and need for reintervention, including further antibiotics or drainage. Successful treatment was defined as clinical and biochemical improvement without recurrence of symptoms, further intervention, or hospital readmission.

Data was analysed using IBM SPSS Statistics version 25 (SPSS Inc, Chicago, IL, USA). Kolmogorov-Smirnov tests were used to assess for normality. Univariate analysis involved Mann-Whitney-U tests for comparison of medians of continuous variables and Chi-squared tests of independence for comparison of categorical variables to establish predictive risk factors for requiring additional drainage. Multivariate analysis involved a binary logistic regression model for any predictive factors found to be significant on univariate analysis. Statistical significance was defined as p < 0.05.

Demographic characteristics, clinical and biochemical laboratory values on admission between the two groups were compared and presented in Table 1a and 1b. Univariate analysis found that patients who required drainage (Group 2) were more likely to be febrile (>38°C) on admission (p = 0.038) compared to patients treated with antibiotics only (Group 1). Patients in Group 2 had a significantly higher median age than those in Group 1 (42 vs 44 years, p = 0.041). Size was also found to be a significant predictor of requiring drainage, with the median TOA size of 8 cm in Group 2 compared to 6.2 cm in Group 1 (p = 0.0001). Positive high vaginal swab (HVS) cultures were also more likely in patients who required drainage (p = 0.027). Although not statistically significant, all three positive blood cultures were in Group 2. There are no significant differences in parity, BMI, ethnicity, PID risk factors and admission inflammatory markers between the two groups. All patients received imaging with half receiving multiple imaging modalities. 74 (80%) patients underwent transvaginal/abdominal ultrasounds, 58 (63%) underwent CT, and 10 (11%) underwent MRI.

A binary logistic regression model adjusting for age, fever, high vaginal swab, and TOA size for multivariate analysis is shown in Table 2. Variables found to be significant independent predictors of successful treatment with antibiotics only on multivariable analysis were age < 35 (OR: 0.89, CI: 0.82-0.97) and TOA size < 6 cm (OR: 0.17 CI: 0.04-0.64). A pyrexia ≥ 38°C on admission significantly predicted the need for additional drainage (OR: 3.82, CI: 1.01-8.12). Although HVS was significant on univariate analysis, this was not significant on multivariate analysis.

Comparison of clinical outcomes between the two groups is presented in Table 3. Patients in Group 2 had a significantly prolonged inpatient stay, length of antibiotics and need for re-intervention compared to Group 1 (p <0.001). In Group 2, the median length of time between admission and decision for surgical drainage was two days (IQR 1-5 days).

Those that had drainage within 72 hours of admission had a trend towards shorter LoA and LoS compared to patients receiving drainage after 72 hours, although this did not reach statistical significance. There was no difference in need for re-intervention between the two subgroups. 3 (3.26%) patients had malignant histology at surgery (all colorectal or intestinal). 69 patients received post-treatment follow-up imaging ( Table 4). Patients in Group 2 were more likely to show no improvement or worsening of the TOA than those in group 1 (p = 0.0251)

In this cohort, 23 (25%) patients received OPAT ranging between 5-50 days ( Table 5). In Group 1, OPAT was used to successfully manage 11 (21%) patients, with a median total antibiotics’ duration of 47 days (IQR: 35-56 days) and a median OPAT length of 25 days (IQR: 16-35.5) The average age of this group was 40 years (± 9 years), with an average TOA size of 7.0 cm (±1.3 cm), and 45% had bilateral abscess. Among those patients who required invasive intervention, 7 (58%) patients had pre drainage OPAT, and 5 (42%) had post drainage OPAT. There is no significant difference in the length of OPAT days between Group 1 and Group 2 (25 vs 26 days, p = 0.853). There was no significant difference in OPAT days between those that received OPAT pre and post drainage (p = 0.749), although numbers are small.

Increased age was the only demographic risk factor between the treatment groups in our study. Other studies have investigated possible risk factors for needing early drainage, but the literature is heterogeneous and often sample sizes used are too small to reach statistical significance. Chan et al. found that BMI ≥ 24.9 had an increased risk of needing drainage. ¹¹ Fouks et al. created a risk assessment tool to predict patients that will need drainage based on four variables: age > 35 years, TOA size ≥ 7 cm, bilateral abscess and WCC > 16 cells/mm ³. ¹² Similarly, we found that women older than 35 years are more likely to need drainage. This could be due to associated comorbidities, more aggressive microorganisms, delay in diagnosis and treatment, and increased likelihood of a secondary TOA in older patients.

Although TOA size has been well-recognised as a risk factor for failing antibiotic management requiring early drainage, the recommended size threshold varies between 3-10 cm across multiple studies and international guidelines. ^{7 , 9 , 11 , 13 – 17} For example, French guidelines recommend all TOA ≥ 3 cm must be drained either radiologically as a first-line treatment or surgically due to higher risk of treatment failure and serious complications with antibiotics alone. ⁹ In our cohort, we found a TOA of 6 cm to be the maximum size of TOA for successful management with antibiotics only. The ability of antibiotics to penetrate large abscesses effectively is known to be limited because of reduced vascular supply, as well as the effects of encapsulation and acidity. ¹⁸ Benefits of early drainage also include the ability to identify causative organisms and sensitivities to better target antibiotic regimes. In our study group, a variety of organisms were grown from vaginal swabs, blood and aspirate cultures ( Table 6). Many patients required discussion with microbiologists to guide antibiotic treatment in the event of unsuccessful empirical management of PID.

Studies have also found a raised WCC ¹⁵ or leucocyte to neutrophil ratio ¹⁹ or CRP ¹⁵ to be predictors for the need for drainage. We found no correlation between WCC or CRP and the need for drainage but found pyrexia ≥ 38°C to be significantly associated with the need for surgical drainage, which corroborates with other groups. ^{11 , 15 , 17}

Our study observed a significantly longer duration of inpatient stay and antibiotics use for patients that received invasive intervention (Group 2) compared to those who had medical treatment only (Group 1). This is consistent with findings by Chan et al. and Habboub et al., ^{11 , 20} which may be reflected by increased clinical severity in patients in Group 2. On the contrary, Perez et al. found that patients who had drainage had a significantly shorter LoS compared to the antibiotic only group, ¹³ however their group was small and only included unilateral abscess and decision for drainage was within 6-12 hours of antibiotics.

Furthermore, we found that patients who received additional drainage were more likely to require readmission or further reintervention. This is consistent with the increased antibiotic duration and length of stay observed in these patients. In terms of methods of drainage, we found no difference in management outcomes when comparing surgical and radiological drainage. Differentiating a TOA from other intra-abdominal pathologies may pose a diagnostic challenge for radiologists. Nonclinical resolution of TOA despite broad spectrum intravenous antibiotics with or without drainage should prompt suspicion of alternate pathology causing a secondary TOA, especially non-gynecological malignancy. In our cohort, 3 patients (3%) were found to have colorectal or caecal malignancy. We could not find related incidences of colorectal or gastrointestinal malignancy presenting as TOA in the literature and it is possible that this is an under-recognised underlying pathology.

Jaiyeoba et al. suggested that failure of response to antibiotics within 48-72 hours (as characterised by persistent fever and increasing leukocytosis) should be considered for surgical drainage ²¹ and this is also highlighted in the BASHH guidance. ⁸ Our study found that patients who failed medical treatment received a median of two days (IQR 1-5 days) of antibiotics before a decision for invasive intervention was made, which is similar to the duration of 4.0±2.1 days by Chan et al. ¹¹ In our study, 58% of patients who received invasive intervention within 72 hours of admission, justified by either poor clinical response to antibiotics or radiologists’ recommendation. We found a trend towards reduction in length of antibiotics (24 vs 40 days) and length of stay (11 vs 17 days) in the subgroup that was drained within 72 hours compared to those that were drained >72 hours from admission, but this did not reach statistical significance. Our work suggests that there may be scope to recommend early intervention in specific groups of patients who have clinically severe disease whilst recognising there may be delays in accessing radiological drainage. In addition, surgical concerns about operating on friable tissues in the presence of sepsis can result in inappropriate perseverance with medical management with changing antibiotic regimens despite lack of initial clinical or biochemical improvement.

Although the general fear of operating in the acute phase of a TOA is the friability of tissues and bleeding, it is easier to operate on acute adhesions than on dense and vascular chronic adhesions, and Reich et al. describes the recommended technique of careful blunt dissection using a probe or aqua dissection. ²² Reich et al. also argues that patients with fertility desires should be managed with early surgical drainage to reduce the risks to fertility, as evidenced by Elmoghazy et al. who found that on second look laparoscopy the incidence of extensive adhesions with bilateral tubal block was significantly higher in the radiologically compared to the surgically drainage group. ⁴ A review on the fertility outcomes of different management of TOA concluded that with medical management alone the reported pregnancy rates were 4-15% and with the addition of laparoscopic drainage within 24 hours of antibiotics the reported pregnancy rates were 32-63%. ²³

OPAT is a safe, evidence-based, and cost-effective regimen for the management of patients with a wide range of infections including complex deep-seated infection ⁵ and can be a valuable management option in patients not fit for surgery. In our cohort of patients, OPAT was used in the management of about a quarter of our patients. To our knowledge, our study is the first published study looking at the use of OPAT specifically in the management of TOA. The decision for the use of OPAT was made by an MDT involving gynecologists, infectious disease specialists, interventional radiologists and our established regional OPAT service.

In the literature, the role of OPAT specifically for the management of TOA is unclear, and further research should aim to compare the efficacy of inpatient vs outpatient antibiotic therapy, as well as identify a subset of patients who may benefit most from OPAT. Interestingly, we identified three patients in our cohort who had TOA as the first presentation of colorectal or caecal malignancy. Two out of three patients had received OPAT for their presumed primary TOA. Although numbers are small, this suggests that all patients considered for OPAT should be managed with a high index of suspicion, whilst considering further investigations to exclude non-gynaecological malignancy, especially in those who did not respond to long-term antibiotics therapy. Hatcher found an 88% success rate of OPAT in the management of intraabdominal infection, they do not comment on the reason for failure for the 12% of patients and it would be interesting to know if they identified underlying malignancies as a cause. ²⁴

The study has several limitations. Firstly, data was collected retrospectively using hospital electronic records and may reflect retrospective case ascertainment. The relatively limited sample size captured the clinically unwell patients admitted with tubo-ovarian abscess. Secondly, potential selection bias may occur with the decision for surgical intervention in patients with TOA driven by decision of the clinical team based on personal clinical experience and judgement rather than based on hospital protocols. However, we believe the effect of heterogeneity was limited since the study is based on a tertiary hospital and decision making was largely confined to a number of experienced consultant gynaecologists. We did not make any specific conclusions on the use of OPAT in TOA since the number of cases were limited.

Open Science Framework: Tubovarian abscess, https://doi.org/10.17605/OSF.IO/JYA8D ¹⁰

This project contains the following underlying data: TOA data file: Excel Sheet containing anonymized data of patients

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).