Antibiotic use during the first episode of acute perianal sepsis: a still-open question
Article information
Abstract
Purpose
The role of antibiotics in preventing fistula formation following an initial abscess remains a subject of debate. This study compared the incidence of fistula in ano in patients experiencing their first episode of acute perianal sepsis, with and without antibiotic therapy, and evaluated the prevalence of fistula in ano necessitating surgical intervention at 1 year.
Methods
This retrospective cohort study was conducted at a tertiary care hospital with a dedicated proctology department. All patients who presented to the emergency proctology unit with a first episode of acute perianal sepsis were eligible for inclusion.
Results
This study included 276 patients. At 1 year, fistula formation was identified in 65.6% of all patients, 54.0% of those who had received antibiotics, and 75.0% of those who had not (P<0.001). This finding remained significant after weighted propensity analysis (odds ratio, 0.53; 95% confidence interval, 0.31–0.92; P=0.025).
Conclusion
The rate of fistula formation was relatively high in this study. However, it was lower among patients with perianal sepsis who were treated with antibiotics, although a causal relationship could not be established. Prolonged follow-up is needed to clarify the role of antibiotic therapy in preventing or delaying fistula development in patients with acute perianal sepsis.
INTRODUCTION
Perianal sepsis is a prevalent condition that causes pain and can lead to serious infectious complications. Treatment typically involves an emergency incision to drain the abscess and alleviate discomfort. If a fistula is present, surgical intervention in one or more stages is necessary to prevent recurrence [1]. Given the high incidence of fistula formation following acute perianal sepsis—70% to 90% as reported in a 1961 study by Park [2] and 81.3% in a more recent series [3]—the standard protocol at our institution in France includes performing an emergency incision under local anesthesia without debridement to relieve the tension abscess. This is followed by a secondary exploration under anesthesia, using endoanal ultrasound (EAUS) or magnetic resonance imaging (MRI) if the fistula is not clinically apparent, to address the underlying fistula, particularly in cases of recurrent sepsis [4–6]. According to most national and international guidelines, antibiotics are generally not prescribed in routine practice. However, they are considered for patients who are immunocompromised or have conditions such as diabetes mellitus, heart valve implants, or associated cellulitis [7–11].
One meta-analysis [12] and a randomized trial [13] suggested that administering antibiotics following the drainage of a perianal abscess might reduce the risk of developing an anal fistula. However, the meta-analysis included only 6 studies, 4 of which were retrospective, encompassing 64 to 299 patients. Only 1 of these studies followed patients for more than 12 months. In fact, only 2 randomized studies have assessed the impact of combining antibiotics with incisional drainage to prevent the emergence or continuation of an underlying fistula. The findings of these studies were inconsistent [13, 14], and ultimately, they reported a very low incidence of fistula compared to that found in Parks' original study [2].
Thus, the question of whether antibiotic therapy following incisional drainage of an initial episode of perianal sepsis can prevent the development of a fistula remains pertinent. This study aimed to assess the rate of anal fistula observed at 1 year in a tertiary proctology center among patients experiencing their first episode of acute perianal sepsis, without an apparent fistula, and to evaluate the impact of combining incisional drainage with antibiotic therapy on the development of fistula in ano.
METHODS
Ethics statement
The study was approved by the Ethics Committee of Groupe Hospitalier Diaconesses Croix Saint-Simon (No. 2022-A00577-36). In compliance with French regulations concerning retrospective studies, we obtained statements from patients indicating that they did not object to clinical research.
Study protocol and patient selection
The study included all patients who presented to the emergency unit of the proctological department at Diaconesses Croix Saint-Simon Hospital (Paris, France) with a first episode of perianal sepsis between March 15, 2019, and March 15, 2021. We excluded patients under 18 years of age and those with a history of Crohn disease, hidradenitis suppurativa, pelvic radiotherapy, or anal suppuration. Additionally, patients presenting with cellulitis, intramural abscess, sexually transmitted infections, pilonidal sinus, an obvious fistula, or those who had undergone surgery within the previous 48 hours were also excluded.
The source population was derived from the hospital's proctology emergency registry, which consists of prospectively recorded, computerized patient files that were reviewed retrospectively. Patients who had not undergone surgery and had either no known follow-up or a follow-up period of less than 1 year at Diaconesses Croix Saint-Simon Hospital were contacted by telephone between March 14, 2022, and April 8, 2022, in order to determine whether they had undergone surgery or imaging procedures at other hospitals within at least 1 year after their emergency visit.
The rate of anal fistula at 1 year was determined by objective evidence, which included the visualization of an internal orifice, an external orifice, and a fistulous pathway during a physical examination, an imaging procedure (EAUS or MRI), or in the operating room. Patients who were lost to follow-up were considered not to have a fistula if it had not been diagnosed in our hospital or another facility. The fistula identified in the operating room was classified according to the system proposed by Garg [15]: grade I (intersphincteric or low trans-sphincteric with a single external orifice) or grades II, III, IV, V for conditions such as horseshoe-shaped passages, intrasphincteric supralevator extensions, multiple tracts, high trans-sphincteric, suprasphincteric, or extrasphincteric fistulas. Antibiotics were prescribed by a general practitioner either before the emergency proctology visit or during the visit itself. The prescribing physician selected the appropriate antibiotic(s) and treatment duration based on the medical conditions and findings from the physical examination. If multiple antibiotics were administered before and/or after the emergency visit, the total duration of antibiotic therapy was calculated by adding the longest duration of an antibiotic regimen before the emergency visit to the longest duration after the visit. Decisions regarding incisional drainage under local, locoregional, or general anesthesia, or for a single or multistage surgical procedure, were made by the proctology emergency practitioner either at the emergency visit or during follow-up.
Statistical analysis
Categorical variables are reported as number (%) and quantitative variables as median (interquartile range, IQR). Categorical variables were analyzed using the Fisher exact test, incorporating Monte Carlo simulation for comparisons involving more than 2 groups. Quantitative variables were assessed using the nonparametric Mann-Whitney test for 2-group comparisons and the nonparametric Kruskal-Wallis test for comparisons across more than 2 groups.
Two methods were employed to compare patients who had received antibiotic therapy either before or immediately after their emergency visit with those who had not, using the propensity score to balance their baseline characteristics. The propensity score was derived from logistic regression based on 12 variables: age, sex, body mass index, active smoking, diabetes, immunosuppression, time from symptom onset to emergency visit, emergency incision, emergency suppuration, and imaging procedures (EAUS, MRI). For the first method, patients were matched 1:1 based on their propensity score. The maximum acceptable distance for matching was set at “0.2 × the standard deviation of the logit-transformed propensity score” [16]. For the second method, inverse probability of treatment weighting was applied.
The balance between the samples was assessed by calculating the standardized mean differences for each characteristic used in the propensity score calculation. We then directly compared the proportion of patients with fistulas between the groups using each method. To incorporate all relevant observations into the analysis, missing data for certain covariates used in the propensity score calculation (active smoking, diabetes, body mass index, EAUS, MRI) were imputed using the k-nearest neighbor method. These variables exhibited less than 10% missing data and were considered "missing at random," as it was reasonable to assume that the missingness of the data did not influence their likelihood of being collected. All analyses were performed using R ver. 4.1.2 (R Foundation for Statistical Computing). A P-value of less than 0.05 was considered statistically significant.
RESULTS
Between March 15, 2019, and March 15, 2021, there were 7,107 visits to the emergency unit of the proctology department. Of these, 813 patients presented with perianal sepsis. However, 537 did not meet the inclusion criteria (Fig. 1). Consequently, 276 patients, with a mean age of 42 years (range, 35–55 years), including 202 men (73.2%), were included in the study population (Table 1).
Flowchart of the study.
Comparison of demographic and clinical categories according to the overall use of antibiotics
During an emergency proctology visit, incisional drainage under local anesthesia was performed on 155 patients (56.2%), with or without concurrent antibiotic therapy. Before their emergency visit, 68 patients (24.6%) had been treated with antibiotics for a median duration of 6 days (IQR, 2.5–6.5 days). The management of oral antibiotic therapy varied, with 49 cases where it was stopped, 19 where it continued, and 56 where it was newly prescribed. Consequently, a total of 124 patients (44.9%) received antibiotics, of whom 102 (82.3%) were treated for at least 3 days. The primary antibiotics administered were amoxicillin/clavulanic acid in 56.5% of cases, metronidazole in 21.0%, and pristinamycin in 17.7%. Subsequently, 178 patients (64.5%) underwent surgery within an average of 21.0 days (IQR, 10.0–46.5 days) following their emergency visit, with fistulas identified in 169 patients (94.9%). At the 1-year mark, 44 patients (17.8%) had no recorded follow-up, including 37 (13.4%) who had no further follow-up after their initial emergency visit and were therefore presumed not to have a fistula. At the 1-year follow-up, fistulas were detected in 181 of 276 patients (65.6%) through EAUS, MRI, and/or intraoperative physical examination. Of these cases, surgical drainage was performed in 169 (93.4%). Notably, 12 patients either declined surgery or were lost to follow-up despite having a scheduled operation.
Univariate analysis indicated that patients who had received prior antibiotic therapy were less likely to have a surgical indication established during their first emergency visit compared to patients not receiving prior antibiotic therapy (50.0% vs. 63.8%, P=0.027). Similarly, at EAUS, fistulas were found less frequently in patients who had received prior antibiotic therapy (43.8% vs. 74.0%, P=0.010). The 1-year fistula rate was significantly lower in patients who had received antibiotics, regardless of whether a local incision had been made (Table 2). At the 1-year follow-up, patients who had received antibiotics underwent surgery less frequently compared to those who had not received antibiotics (68.4% vs. 84.7%, P=0.004). The rate of intraoperative fistula identification was similar whether patients received antibiotics or not (97.0% vs. 93.7%, P=0.486), and the fistulas identified were not significantly less likely to be grade I in patients who received antibiotics (26.2% vs 36.5%, P=0.168). Overall, the 1-year rate of fistula identification was lower in patients who received antibiotics (54.0% vs. 75.0%, P<0.001) (Table 3).
Comparison of 1-year fistula rates according to incision and antibiotic therapy
Univariate analysis according to antibiotic use
Propensity analysis, employing either 1:1 matching or inverse probability of treatment weighting, corroborated these findings: antibiotics significantly reduced the incidence of fistulas at 1 year (Table 4). When using the propensity score balanced groups method, the standardized mean difference for each covariate was less than 0.1, with the exceptions of immunodeficiency and EAUS, which only applied in the matching method. In the weighted cohort, antibiotic therapy was linked to a reduction in fistula cases at 1 year, with an odds ratio of 0.53 (95% confidence interval, 0.31–0.92; P=0.025).
Propensity analysis for fistula development according to the use of antibiotic therapy
DISCUSSION
Our study showed that the 1-year rates of fistula were lower in patients who received antibiotics. This finding remained significant after conducting a weighted propensity analysis, which indicated that antibiotic therapy could reduce the occurrence of fistulas by 47%. Although this study does not meet the criteria of a randomized clinical trial and has some limitations, the use of propensity analysis on a large cohort is a valid approach. This method balances specific patient characteristics, approximating the effects of randomization in terms of the likelihood of receiving antibiotics and undergoing surgery [17]. The study design did not permit the establishment of a causal relationship between the use of antibiotics and the formation of fistulas. An outstanding question is how antibiotics might be linked to a reduced incidence of fistula in ano at the 1-year follow-up. It has been suggested that gut-derived bacteria may contribute to the development of perianal fistulas [18]. Toyonaga et al. [3] observed that digestive tract bacteria were more commonly found in abscesses associated with fistulas, whereas skin bacteria were predominant in abscesses without fistulas. Given the potential benefits of antibiotic therapy, the selection of an appropriate antibiotic and the duration of treatment become important considerations. Antibiotics effective against digestive tract bacteria, such as gram-negative bacilli and anaerobes—including amoxicillin with clavulanic acid, third-generation cephalosporins, quinolones, and/or imidazoles—are recommended [13, 19]. However, the optimal duration of antibiotic therapy remains unclear, with few high-quality studies available to define an evidence-based duration, which may range from 5 to 10 days following drainage [12]. Despite current guidelines not recommending the use of antibiotics, their use was frequent in this series, with 44.9% of patients receiving antibiotics either before (24.6%) or after (27.5%) their visit to the emergency unit of the proctology department. The patterns of antibiotic prescription varied widely in both type and duration, with 17.7% of patients receiving a regimen for less than 3 days. We found no significant differences between the various categories of antibiotics used.
Fewer than half of the patients were exposed to antibiotics during the study, and after 1 year, a fistula had been identified in 65.6% of all patients. However, this rate of fistula identification at 1 year might still be underestimated due to the absence of systematic imaging or examination in the operating room, the diagnostic pathway for fistula in ano, and the relatively short follow-up period. This high rate of anal fistula is consistent with the findings of Parks [2] and Toyonaga et al. [3], supporting the natural history of fistula in ano through cryptoglandular infection and emphasizing the importance of investigating the presence of a fistula in any case of perianal sepsis, even during the initial episode. The rate of intraoperative identification of fistulas was 95% in our expert center, where all procedures are performed by senior surgeons, never by residents. Identifying the correct internal opening is indeed the most significant prognostic factor in reducing the risk of recurrence after anal surgery [20]. Therefore, we believe that even for the initial episode of perianal sepsis, surgery should be conducted under optimal conditions by experienced senior surgeons, not by residents alone.
The quality of the drainage should also be discussed. In most cases, we performed an emergency incision under local anesthesia to provide immediate relief without debridement of the abscess cavity. It has been suggested that the persistent inflammation from the abscess may continue to infect one of the pectineal glands and the underlying fistula tract, leading to epithelialization and/or the formation of granulation tissue. This process can prevent the healing of the fistula tract and contribute to the persistence of a fistula that is prone to recurrence [21]. Consequently, some authors suggest that the fistula may result from insufficient or delayed drainage. Observing that some abscesses heal completely without recurrence [22], they recommend simple incisional drainage [18, 23], potentially avoiding fistulotomy in over 60% of patients—specifically, those without a confirmed fistula diagnosis. However, our series suggests that antibiotics, rather than simple incisional drainage, may influence the rate of fistula formation. Clearly, antibiotics reduce signs of infection and inflammation, making it more challenging to clinically detect a fistula. The critical question is whether antibiotic therapy prevents fistula formation—and thus its clinical detection—or not. This question is worth considering, especially in light of our findings that evidence of fistula formation, either through EAUS or MRI, was less common in patients who received antibiotics. With a 1-year follow-up, identifying fistulas in the long term is beyond the scope of our study. Additionally, although patients followed up for hidradenitis suppurativa or recurrent suppuration were excluded, other forms of purely cutaneous suppuration could not be identified.
In the debate over the use of antibiotics for preventing fistulas, it is crucial to distinguish between the first episode of acute perianal sepsis and recurrent episodes, as the likelihood of an underlying fistula is significantly higher in recurrent cases. Therefore, antibiotic recommendations should vary between the initial episode and subsequent recurrences. To prevent the development of a chronic fistula, it is particularly important to identify which patients should receive an optimal course of antibiotics in conjunction with drainage during their first episode of perianal sepsis. It is hoped that the ATLAS (Antibiotic Treatment Following Surgical Drainage of Perianal Abscess) study will provide the necessary information, although long-term follow-up is essential to prevent delayed diagnosis [24].
In our expert center, a fistula was identified intraoperatively in 95% of patients presenting with their first episode of perianal sepsis. This observation underscores the continued high prevalence of fistula-associated perianal sepsis at the 1-year follow-up. While a direct causal relationship could not be established, our data revealed a 47% reduction in fistulous pathways among patients who received antibiotics during their first episode of acute perianal sepsis. Further prolonged follow-up and clinical controlled trials are needed to elucidate the potential role of antibiotics in either preventing or delaying the development or diagnosis of fistulas in these patients.
Notes
Conflict of interest
No potential conflict of interest relevant to this article was reported.
Funding
None.
Acknowledgments
The authors thank Drs. Patrick Atienza, Elsa Lambrescak, Anne-Carole Lesage, Jean-David Zeitoun, Anna Cabral Cordeiro, Sarah Taieb, Sihem Bendaoud, Anthony Caldiero, Charles Houdeville, Vincent Nguyen, Keopiseth Pich, Vireak Kann, and Régis Garrigue for their work in proctologic emergency.
Author contributions
Conceptualization: SB, IE; Data curation: SB, AE, SYD, MT, JB, DB; Formal analysis: SB, DL, IE; Investigation: SB; Methodology: SB, DL, IE; Project administration: SB, IE; Resources: DB, IE; Software: DL, DB, IE; Supervision: IE; Validation: all authors; Visualization: all authors; Writing–original draft: SB, DL, IE; Writing–review & editing: SB, DL, AE, IE. All authors read and approved the final manuscript.