Piperacillin To Prevent Cholangitis after Endoscopic Retrograde Cholangiopancreatography

Methods

Our trial was done at the Academic Medical Center in Amsterdam, the Netherlands, which is a tertiary referral center for ERCP. Patients were eligible for study entry if they were suspected of having either biliary tract stones or distal common bile duct stricture. Patients were excluded if they were 18 years of age or younger, had a known allergy to penicillin, had had fever during the 7 days before ERCP (reported body temperature >38.0 °C), or had had direct cholangiography or had used antimicrobial drugs during the 7 days before ERCP.

Eligible patients who gave informed consent were randomly assigned to receive either a single 4-g dose of piperacillin (in 100 mL of saline solution) or placebo (100 mL of saline solution) that was identical in appearance to the piperacillin. Consecutively numbered bottles containing either piperacillin or placebo were prepared in the hospital pharmacy and were assigned in ascending order to consecutive patients. The study medications were administered intravenously during a 10- to 15-minute period approximately 30 minutes before ERCP was done. Patients, investigators, and staff members involved in the management of patients during and after ERCP were unaware of treatment allocations. All ERCP procedures were done with therapeutic intent, but in many instances, ERCP was also considered to be an important diagnostic procedure. The study was approved by our institutional review board.

After ERCP, patients who had been referred to our center returned to their local hospitals for observation (≥ 24 hours) and further management. Patients were allowed to return home on the day of the procedure only if no abnormalities had been found and no interventions had been done during ERCP.

After ERCP, the responsible physicians were asked to keep records of their patients' body temperatures until discharge or until 48 hours after ERCP. No antibiotic agents were to be given as prophylaxis after the procedure, and a blood culture was to be done if symptoms of cholangitis occurred. Further management of each patient, however, was left to the attending physician. Follow-up information during the first 2 days after ERCP was obtained for every patient from the responsible physician through a structured telephone interview. Seven days after ERCP, each patient was contacted directly. If the patient was still in the hospital at that time, additional information was obtained from the attending physician.

The primary end point of the trial was the occurrence of acute cholangitis within 1 week after ERCP. Acute cholangitis was considered to be present if fever had developed (body temperature >38.0 °C), if clinical symptoms were severe enough to warrant the initiation of antibiotic therapy, and if no clear cause for the fever outside of the biliary tract was identified. If a diagnosis of cholangitis was considered, the patient or the responsible physician was questioned about symptoms that might indicate a biliary tract infection (such as jaundice or right upper abdominal pain) and symptoms that could indicate infections outside of the biliary tract (particularly urinary and respiratory tract infections). After 48 hours, by which time most study participants had left the hospital, the requirement for a recorded body temperature greater than 38.0 °C was dropped from the definition of acute cholangitis. If there was any doubt as to whether acute cholangitis should be diagnosed, expert opinion was obtained from three independent gastroenterologists who were blinded to treatment assignments and who decided by consensus. The severity of cholangitis was classified as mild, moderate, or severe by using the grading system of Cotton and colleagues [11]. In patients who developed acute cholangitis after ERCP was done a second time during the follow-up period, prophylaxis was not considered to have failed.

Our analysis was done on the basis of the intention-to-treat principle. In the main analysis, we compared the numbers of patients in each study group who had developed acute cholangitis within 7 days after ERCP. The effect of piperacillin was also assessed in two major subgroups, which were defined on the basis of the patients' indications for ERCP: biliary tract stones or distal common bile duct stricture. We calculated P values by using the chi-square test with the Yates correction or by using the two-tailed Fisher exact test if the expected value in one of the cells was less than 5. Calculations were done with Epi Info statistical software, version 5 (Centers for Disease Control and Prevention, Atlanta, Georgia). The CI for the absolute risk reduction was calculated using the software program Confidence Interval Analysis (BMJ Publishing Group, London, United Kingdom).

Before the trial began, we estimated the sample size. Estimates of the incidence of cholangitis were based on information obtained in a pilot study. If 12% of the patients in the placebo group and 6% of the patients in the treatment group were to develop cholangitis, 690 patients would need to be enrolled in the study (α = 0.05, β = 0.8). When the incidence of cholangitis proved to be much lower, an independent monitoring committee advised early cessation of the trial. A new sample size calculation based on the actual incidence rates indicated that approximately 6000 patients would need to be randomly assigned to treatment to make the projected outcome significant according to conventional statistical standards. Both the practical difficulties of continuing the study and the limited relevance of the projected low absolute risk reduction contributed to the monitoring committee's recommendation.

Results

Between April 1991 and June 1994, 562 patients were entered into the study. Eleven patients who met major exclusion criteria were inadvertently assigned to treatment and were excluded from the analysis. The characteristics of the remaining 551 patients were similar in the two study groups (Table 1). The median time between the administration of the study medication and the start of ERCP was 13 minutes in both groups (95% range, 3 to 90 minutes). Study medication caused no side effects in the piperacillin group; a mild rash was attributed to the study medication in 1 patient in the placebo group. Diagnoses made on the basis of the ERCP findings were similar in the two groups, as were the numbers and types of interventions done during ERCP (Table 2). The median duration of ERCP was 34 minutes for both groups (95% range, 10 to 86 minutes). A total of 19 complications occurred during ERCP (3.4%). Substantial bleeding from the sphincterotomy site was seen in 7 patients; the common bile duct was perforated with a guide wire in 6 patients; a gastrointestinal perforation was made with the endoscope in 4 patients; bradycardia developed in 1 patient, making it necessary to end the procedure; and the release mechanism of a metal self-expanding stent became lodged in the bile duct of 1 patient.

Complete follow-up information about the occurrence of fever, the use of antimicrobial drugs, and the indication for using antimicrobial drugs was available for all patients for the first 7 days after ERCP. The overall mortality rate during the 1-week follow-up period was 1% (three patients in the piperacillin group and two patients in the placebo group died). One patient in the placebo group died of cholangitis and severe septicemia. In this patient, who had pancreatic cancer, the biliary tract could not be drained and cholangitis developed after a second ERCP. Two patients died as a result of the cancer that had prompted ERCP, one patient had a fatal cerebral hemorrhage, and the cause of death was unclear in one patient.

Twelve patients (4.4%) in the piperacillin group and 17 patients (6.0%) in the placebo group developed acute cholangitis during follow-up (Table 3). All cases of cholangitis were diagnosed in the hospital. The relative risk for the development of acute cholangitis in patients who received prophylaxis with piperacillin was 0.73 (95% CI, 0.36 to 1.51; P > 0.2). The absolute risk reduction in the piperacillin group was 1.6% (CI, −5.3% to 2.1%). The relative risk for developing acute cholangitis within the first 48 hours after ERCP was 0.43 (CI, 0.15 to 1.21; P = 0.16). In 1 patient in the piperacillin group, cholangitis was classified as severe. This patient was admitted to the intensive care unit for treatment, after which he completely recovered.

The effect of prophylaxis with piperacillin was assessed separately for the two subgroups of patients that had had different indications for ERCP. Acute cholangitis developed in 3 of the 130 patients in the piperacillin group and in 6 of the 156 patients in the placebo group who had had ERCP because biliary tract stones had been suspected (relative risk, 0.60 [CI, 0.15 to 2.35]; P > 0.2). Acute cholangitis developed in 8 of the 106 patients in the piperacillin group and 6 of the 88 patients in the placebo group who had had ERCP because distal common bile duct stricture had been suspected (relative risk, 1.11 [CI, 0.40 to 3.07]; P > 0.2).

Fever was reported in 71 patients (32 receiving piperacillin and 39 receiving placebo). In 42 of these patients, cholangitis was not diagnosed because the other criteria for cholangitis were not met. In 8 patients, the fever was focused outside of the biliary tract.

Adverse events other than acute cholangitis occurred in 21 patients during the follow-up period. Ten patients developed acute pancreatitis. Three patients had significant bleeding from the sphincterotomy site (1 of the 3 also had pancreatitis). Intraabdominal biliary leakage, intestinal perforation, acute cholecystitis, and biliary colic were each diagnosed in 1 patient. Five patients had complications that could not be directly related to ERCP (for example, cerebral hemorrhage).

During follow-up, 53 patients had a second ERCP because the first procedure had failed or had been only partially successful. The median period between the two procedures was 5 days. Additional antibiotic prophylaxis given after the procedure was inadvertently administered in the referring hospital to 44 patients (16 in the piperacillin group and 28 in the placebo group), primarily because of concern about insufficient drainage of a contaminated biliary tract or because of complications during ERCP. A second ERCP or additional prophylaxis interfered with follow-up in 82 patients (15%).

Blood cultures were obtained from 20 patients who developed fever after ERCP. Escherichia coli was found in 3 patients (1 of whom received piperacillin), Enterococcus faecalis was found in 1 patient who received placebo, and Aeromonas species was found in 1 patient who received piperacillin. In 15 patients, the blood cultures showed no growth. Susceptibility testing for piperacillin was done in three instances and showed one E. coli strain resistant to piperacillin (in the piperacillin group).

Discussion

Our study indicates that single-dose antibiotic prophylaxis with piperacillin can, at best, marginally reduce the incidence of ERCP-induced cholangitis in patients suspected of having either biliary tract stones or distal common bile duct stricture. We included patients with strictures (almost invariably malignant) as well as patients with stones because the expected rate of cholangitis after ERCP is relatively low in both of these groups (approximately 7% and 2%, respectively) [2, 12] and the need for antibiotic prophylaxis in these patients is thus uncertain. Furthermore, ERCP is often used as a diagnostic tool with which to differentiate between the two conditions when ultrasonographic examination is equivocal. Although our hospital is a tertiary referral center for ERCP, our study group was well defined and our patients were similar to those seen in other institutions.

Our primary outcome measure was designed to assess a clinically relevant complication of ERCP. If patients develop fever shortly after ERCP, the infection is considered to be in the biliary tract until it is proven to be elsewhere. In such patients, physicians must decide—on the basis of each patient's severity of symptoms and general condition—whether antibiotic treatment is indicated. We chose to count only those ERCP-induced episodes of (presumed) cholangitis that prompted the attending physician to initiate antibiotic treatment. This does not include all patients who had a brief episode of chills and increased body temperature after ERCP, nor is it limited to the few patients who developed severe septic shock.

Do our results prove that prophylaxis with piperacillin is incapable of reducing the incidence of ERCP-induced cholangitis? The relative risk indicates a 27% risk reduction, albeit with a wide CI. The point estimate of absolute risk reduction, however, is only 1.6%. In other words, 63 patients would need to receive prophylaxis to prevent one case of cholangitis. Moreover, all of the cases prevented were mild or moderate in severity. Thus, although we cannot confidently rule out a small benefit of routine antibiotic prophylaxis in patients suspected of having biliary tract stones or distal strictures, this possibility must be weighed against the disadvantages of overtreating almost all (98.4%) of the patient population. Economic cost, adverse side effects, and the emergence of resistant bacteria associated with the unnecessary use of antibiotics are important drawbacks of routine prophylaxis.

If we consider only the first 48 hours after ERCP, the possibility that piperacillin will provide meaningful protection appears to be slightly greater because the absolute risk reduction (2.4%) and the statistical strength of the association (P = 0.16) are somewhat favorable compared with those for the 7-day study period as a whole (although this is far from statistically convincing). Considering that protection within the first 48 hours was a secondary end point of our trial and that no previous study has shown antibiotic prophylaxis to be advantageous, we believe that there is an insufficient basis for recommending the routine use of prophylaxis with piperacillin.

In this study, we used piperacillin as prophylaxis before ERCP. Piperacillin is a ureidopenicillin that is effective against the pathogens that most commonly cause cholangitis after ERCP, especially E. coli and Klebsiella species [13, 14]. Piperacillin is excreted well into unobstructed bile ducts, but it does not enter the bile if an obstruction is present [15, 16]. However, the biliary concentration of an antibiotic is probably not an important determinant of the drug's efficacy as prophylaxis [17]. Since the start of our trial, some reports have indicated that gram-negative bacterial resistance to piperacillin is emerging [10, 18]. In our study, one strain resistant to piperacillin was identified.

Several placebo-controlled trials [5-8] have investigated the efficacy of antibiotic prophylaxis in preventing ERCP-induced cholangitis. These studies used various antibiotic agents (ciprofloxacin, tetracycline, and cefotaxime), but none showed antibiotic prophylaxis to have a significant advantage over placebo according to conventional statistical standards. A total of 349 patients are described in these four reports. Five of the 174 patients in groups receiving placebo and 2 of the 175 patients in groups receiving prophylaxis developed cholangitis. The diversity of patients included and the study methods used, however, make these data difficult to interpret.

Considering the results of our own trial, the results of previous trials, and the previously mentioned disadvantages of overtreatment, we believe that routine antibiotic prophylaxis for all patients who have ERCP cannot be justified. Antibiotics, however, may still have a place in the prevention of ERCP-induced cholangitis in high-risk patients, such as those in whom complete drainage is not obtained. Incomplete drainage is the single most important risk factor for cholangitis after ERCP [4]. In some patients (such as those with primary sclerosing cholangitis or cholangiocarcinoma), failure to completely drain the biliary tract can be anticipated and antibiotic prophylaxis may be effective. The increased baseline risk of these patients would result in a higher absolute risk reduction, even if the relative risk is unchanged. The relevant question of whether cholangitis can be prevented by starting antibiotic therapy after ERCP in patients with incomplete biliary drainage remains unanswered.

Some limitations of our study should be noted. First, our study may have produced a false-negative outcome (a type II error), and our definition of cholangitis may not have been 100% sensitive and specific; these limitations have been addressed above. Second, further ERCP procedures and the use of additional antibiotic prophylaxis interfered with follow-up in 82 patients (15%). Many of these patients were considered to be at high risk for cholangitis after ERCP because of insufficient drainage of the biliary tract. It is possible that some of them would have developed cholangitis if these additional interventions had not been done. If, in a worst-case scenario, we were to regard all of these patients as having developed cholangitis, the relative risk would change marginally, to 0.78 (CI, 0.55 to 1.09). Although this is reassuring, it is not formal proof that these additional interventions did not bias our results. If a patient is considered, after ERCP, to be at high risk for cholangitis (for example, because of insufficient drainage), it seems advisable to start antibiotic therapy immediately after the procedure and to plan for a second ERCP. A recent study [19] indicated that prolonged antibiotic prophylaxis in patients in whom complete drainage is not obtained can reduce the incidence of cholangitis after ERCP.

We conclude that single-dose antibiotic prophylaxis with piperacillin does not substantially reduce the incidence of ERCP-induced cholangitis in patients who have ERCP for suspected biliary tract stones or distal common bile duct stricture. Given the low baseline incidence of cholangitis after ERCP in these patients and the relatively mild course of disease, the routine use of prophylaxis with piperacillin cannot be justified.

Dr. Dankert: Department of Medical Microbiology, Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands.

Dr. van Leeuwen: Division of Gastroenterology and Hepatology, University of Alabama at Birmingham, 404 Lyons-Harrison Research Building, 701 South 19th Street, Birmingham, AL 35294.