Comparative effectiveness of surgery versus antibiotics in acute appendicitis: a systematic review

To be included in this review, studies had to require the same diagnostic certainty for all the patients eligible for inclusion in either treatment arm. This should lead to similar rates of acute appendicitis, complicated appendicitis and misdiagnosis between operative and nonoperative groups [9]. All studies required basic history and physical exam findings to identify possible appendicitis. Lab values, including WBC and CRP were reported by most studies. The study design for Hansson et al. [6] included a deliberate attempt not to establish requirements for diagnosis; however, patient history, physical exam, laboratory tests and imaging results were used according to established clinical practice. The authors explained that they chose to not require imaging in the diagnostic criteria to reduce additional testing costs [6].

Exclusion criteria were not discussed in 27.3% of the reviewed studies [5,6,9]. The majority of studies excluded patients with peritonitis or suspicion of perforation [3,7–9,11–13]. Histories of chronic intermittent abdominal pain or inflammatory bowel disease were also exclusion criteria in five studies [3,7,11–13]. Recent antibiotic use [7] and allergy to study antibiotics [3,7,9] were also the exclusion criteria in some studies.

It is known that imaging can improve the diagnostic accuracy of acute appendicitis [8,9,17]. Improvements in diagnostic imaging technology quality and availability allow for improved preoperative diagnosis and characterization of the severity of appendicitis, which can direct patient care [11]. The choice of diagnostic imaging technique varied among studies and was not explicitly required in two studies [6,9]. For the remainder of the studies, ultrasound and/or computed tomography (CT) were used to confirm the diagnosis of appendicitis and to identify perforation, phlegmon or the presence of fecalith. CT scan was explicitly required by Vons et al. [7] and Salminen et al. [13] and was conducted after patient consent was obtained in the former. It is known that the use of CT scan reduces the number of NA [4,21,22]. However the number of patients imaged by each modality is not provided to allow us to evaluate a difference in imaging preference and its effect on outcomes [3,5,11,12,14].

The modified Alvarado score was used as a diagnostic criterion in two studies. However, one study did not specify a cut-off score [9] and the other used a cut-off score of 5, which is ‘equivocal for acute appendicitis’ [3]. The use of such a scoring system can help to guide the physicians in the identification of those with probable (score: 8–9) and highly probable (score: >9) acute appendicitis for treatment [23,24]. A study conducted by Farhnak and Talaei-Khoei [25] was excluded from our review because the Alvarado score was used to stratify patients into treatment arms. Patients with Alvarado score: 5–7 were treated with antibiotics and observation and those with score: 8–10 were surgically managed. While this might be a means of reducing the rate of NA, it does not ensure a high rate of diagnostic accuracy for the antibiotic group [25].

Antibiotics regimens included one, two or three antibiotics. Duration of mandatory inpatient antibiotic management varied among studies from 0 to 2 days. The majority of protocols required a 1–2-day inpatient stay with intravenous antibiotics and transition to oral antibiotics prior to discharge [5,6,8,9,11–14]. Vons et al. [7] allowed patients to begin treatment with oral antibiotics if tolerated and DiSaverio et al. [3] discharged patients from the emergency department with oral antibiotics. Subjects were monitored as inpatients to observe failure of primary antibiotic therapy in 90.1% of studies [3,5–14]. Patients were discharged from the hospital when patients’ condition improved and oral antibiotics and diet were tolerated. Outpatient antibiotic regimens consisted of one to two antibiotics for total treatment duration of 4–10 days.

Patients within the operative management arms underwent surgery within 24 h of admission for all the studies. Use of laparoscopy ranged from 5.5 to 100% among studies that reported this operative modality [5,7,9,11–14]. Discrepancy in operative approach among studies likely relates both to difference in time periods and the countries in which the studies were conducted. Postoperative discharge criteria were not explicitly stated in the majority of studies evaluated.

For the purpose of our review, we have defined crossover patients as those patients who were initially assigned to antibiotic management yet required surgery during the index admission (or in the case of discharge from the emergency department, within the first 5 days as was the definition for DiSaverio et al. [3]). Among the studies included in this review, crossover did not necessarily indicate early failure of antibiotic management, because crossover to surgery could occur for a variety of reasons (patient choice, surgeon discretion and misdiagnosis of disease severity on presentation, among others).

Crossover rates ranged from 5 to 47.5% (Table 2). While the majority of patients required crossover for failure to improve with antibiotic management, there were additional reasons for patients to change treatment arms within the index admission. Two studies cited patients’ desire for surgery as a reason for crossover [5,6]. The crossover rate from primary antibiotics management to primary surgical management for Hansson et al. [6] was the highest at 47.5%. Within this study, patients were allowed to change study arms on request. Among all the study patients, the authors reported that 30.3% of patients preferred a different treatment than what they were assigned; this included 12.9% of patients assigned to primary antibiotic management. Although the authors reported a primary failure rate of antibiotic therapy as 9.2%, they also noted that surgeons made the decision to crossover to surgery in 31.6% of patients originally randomized to antibiotics. Due to this ambiguity, we allocated these 31.6% of patients who underwent surgery during the index admission to the crossover group. One solution to the problem faced by Hansson et al. [6], is a prospective patient choice design [11,12]. Minneci et al. used this study design for both pilot and then a larger study in pediatric patients. Patient and family satisfaction was high among both surgical and antibiotic management groups in each study, which the authors attributed to the positive benefits of active patient involvement in decisions about their care [11,12].

LOS is presented in Table 3. Many of the studies reported longer LOS for the antibiotic management arms. This was especially true among studies requiring two or more days of intravenous antibiotics with an early nothing by mouth (NPO) period [5,8–10,13]. LOS was shorter in studies with higher usage of laparoscopy and in fast-track antibiotic management, such as in DiSaverio et al. [3].

Patients were followed on average for 1–2 years to assess for recurrence. Recurrence rates ranged from 3.6 to 36.8%, excluding those with immediate failure of antibiotic treatment requiring crossover to appendectomy (during index admission) and time to recurrence ranged from 3.4 to 8 months. Hansson et al. [6] did not report an overall recurrence rate, but did note that a third of recurrences occurred within 10 days of hospital discharge, and the remaining two-thirds between 3 and 16 months. The majority of patients with recurrence were managed operatively; however, some patients were successfully managed a second time with antibiotics [5–6,14]. Park et al. [14] and Hansson et al. [6] reported that 13–22.2% of patients with recurrent appendicitis were managed successfully with a second round of antibiotics. Turhan et al. [5,14] also reported that 22.2% of patients with recurrent appendicitis underwent medical treatment. One of these patients had a third recurrence, which was treated with appendectomy [5].

Final pathology results were compared among the three groups: patients who underwent primary surgical management, patients who crossed over to surgery during the initial antibiotics phase and primary antibiotic patients with recurrent appendicitis. Among all groups who had surgery, appendicitis was found on pathology in 85.2–97.9% of patients (Table 2). Four studies did not report how many patients did not have appendicitis on final pathology. When reported, appendiceal inflammation documented on final pathology was present at higher rates in those who crossed over to surgery and those with recurrent appendicitis compared with those assigned to primary surgery management. Although many studies included ultrasound or CT for diagnosis, data correlating imaging modality and final pathology were not reported [3–5,11,12]. Accuracy of diagnosis was greatest in the study by Salminen et al. [13], which used CT as the required diagnostic imaging modality. In this study, patients in the surgery arm were correctly diagnosed 99.3% of time, crossover patients 100%, and antibiotic arm patients who later developed recurrence 90.9%. Vons et al. [7], which also used CT, did not report pathology of the surgical or crossover groups; however, diagnosis of appendicitis (based on pathology) for those with recurrent appendicitis was accurate in 88.9% [7]. When patients in these studies presented with recurrent abdominal pain, it was frequently presumed that they had recurrent appendicitis: however 10–11.1% of patients in the recurrence groups had NA. Use of diagnostic imaging was not reported among those with recurrent symptoms.

For those patients who underwent surgery in the study by Hansson et al. [6], efficacy of operative treatment was defined as either the correct pathology of appendicitis or another diagnosis requiring surgery. Among those who underwent surgery during the index admission, those in the surgery and crossover groups, 88% had pathology consistent with acute appendicitis and 1.2% had another diagnosis treatable by surgery, although the specific pathologies of the three patients in this category were not specified. For the purposes of our review, only diagnoses of acute appendicitis were considered efficacious.

Among all the included studies, the diagnoses of those without appendicitis included mesenteric adenitis [8,9], Campylobacter enteritis, terminal ileitis, ruptured ovarian cyst, pelvic inflammatory disease, Meckel’s diverticulum [9], granulomatous disease [11] and normal pathology [6,9].

Table 2 displays available data for rates of perforated appendicitis. Three studies presented data on complicated appendicitis, defined as presence of abscess, phlegmon or gangrene [3,12,14]. Rates were calculated using the number of patients with perforated or complicated appendicitis with respect to the number of patients undergoing surgery within a given group. Rates of perforated appendicitis ranged from 0.7 to 20% in the primary surgery group, 0–100% in the crossover group, and 0–31.3% in the recurrent appendicitis group. Eriksson and Granström [8] had one patient in the crossover group and the pathology of that patient was consistent with perforated appendicitis, giving a perforated appendicitis rate of 100%. This patient was noted to develop peritonitis during obligatory admission for antibiotics and was found to have appendiceal perforation on pathology [8]. It is not always possible to identify perforation on imaging, thus it is not known how many antibiotic-treated patients had perforated appendicitis at the time of randomization. Additionally, progression to perforation is a dynamic process and may occur between the time of imaging and randomization. Among the randomized controlled trials, the rates of perforated appendicitis should be equal between study arms. As more studies are conducted using modern diagnostic imaging tools, patients with significant perforation (i.e., not microperforations that are evident only on final pathology) may be identified prior to enrollment and not included. This will likely affect the rate of crossover in future studies as patients with clinically significant perforated appendicitis may not improve with antibiotics alone.

Rates of complications were calculated for patients assigned to surgical management and separately for those in the antibiotic group who required surgery (Table 2). Within the surgery group, complications occurred in 4.4–20.5% of cases versus 0–18.5% in the group of patients originally started on antibiotics but who crossed over to surgery. There was one death among all the study populations. The death occurred in a patient randomized to operative management and occurred on day 5 after surgery due to cardiomyopathy. The authors did not specify whether the patient’s heart disease decompensated secondary to perioperative stress [13]. Complications varied among studies, but were most commonly wound infections and intra-abdominal abscesses. Few patients required reoperations, and some of these were not complications, in particular two patients who required definitive operation for malignancy discovered at the initial surgery [5]. Complication rates for those with acute appendicitis on pathology versus those with not available (NA) were not reported.

Salminen et al. [13] and Hansson et al. [6] had the highest complication rates for surgery (20.5%) and antibiotic (16.2%) groups, respectively. Salminen et al. [13] reported a low laparoscopic surgery rate, which could explain the higher incidence of complications in this group. The authors noted in their discussion the majority of the morbidity within their surgery group was related to wound infections. They postulated that had their laparoscopic rates been higher, there would have been likely fewer complications [13]. Minneci et al. [12] reported a 100% laparoscopic rate and their rates of complication were 7.7%.

Patients in the antibiotic group reported less pain than those in the surgery group and used less narcotics [8]. Lower pain scores were reported as early as 12 h [8]. Pain levels reported at 1 month were not different among groups [6,7]. Sick days and days of work/school missed were reported by few studies (Table 3). When presented, patients in the antibiotic group used less sick leave and spent less time away from school or work compared with patients undergoing predominantly open appendectomy [3,6,9,11,13].

Cost information was presented in five studies (Table 4). Primary admission costs were higher among those in the surgery group. As noted in the ‘Methods’ section, costs were compared as a ratio of surgery to antibiotics. Surgery ranged from 1.29- to 2.05-times more expensive than primary antibiotic management [3,5,6,12,14]. The cost ratio was highest for Hansson et al. [6], likely in part due to longer LOS for both arms due to including obligatory intravenous antibiotics for 24 h in the antibiotic group and recovery after open appendectomy in the surgery group. Cost of antibiotic management reported by DiSaverio et al. [3] was low due to outpatient antibiotic management of acute appendicitis [3]. The Comparing Outcomes of Drugs and Appendectomy Trial is a trial in active enrollment and is designed to evaluate the effectiveness of mostly outpatient antibiotic management of acute appendicitis in a setting where laparoscopic surgery is much more common than in the European trials [26]. As treatment practices move toward outpatient management with antibiotics and earlier discharge of surgically managed patients [27,28], costs of each arm will likely decrease.