Postoperative Antibiotics Correlate with Worse Outcomes after Appendectomy for Nonperforated Appendicitis

Postoperative Antibiotics Correlate with Worse Outcomes after Appendectomy for Nonperforated Appendicitis Brian A Coakley, MD, Eric S Sussman, BA, Theodore S Wolfson, BA, Anil S Bhagavath, MD, Jacqueline J Choi, MD, Nalin E Ranasinghe, MD, Elizabeth T Lynn, MD, Celia M Divino, MD, FACS BACKGROUND: STUDY DESIGN: RESULTS: CONCLUSIONS: Acute appendicitis remains the most common cause of acute abdominal pain necessitating operative intervention. Although postoperative antibiotics are universally used for perforated appendicitis, no consensus exists on whether postoperative antibiotics are beneficial for preventing surgical site infections (SSIs) in nonperforated cases. We set out to determine how postoperative antibiotic therapy affects outcomes after appendectomy for nonperforated appendicitis. The medical records of 1,000 patients undergoing appendectomy for nonperforated appendicitis at The Mount Sinai Medical Center from January 2005 through July 2010 were retrospectively reviewed. In total, 728 cases contained sufficient follow-up data for analysis; 334 of these patients received postoperative antibiotics and 394 did not. There were no significant differences in patient demographics, medical comorbidities, American Society of Anesthesiologists (ASA) class, admission temperature, preoperative antibiotic treatment, operating room time, estimated blood loss, appendiceal diameter, or intraoperative transfusion between the two groups, although WBC was higher for patients receiving postoperative antibiotics (12.3 vs 14 cells/mm 3,p 0.001). Postoperative antibiotics did not alter the incidence of superficial SSIs, deep SSIs, or organ space SSIs (all p 0.1), but did correlate with higher rates of Clostridium difficile infection (p 0.02), urinary tract infection (p 0.05), postoperative diarrhea (p 0.001), and longer length of stay (LOS) (1.1 vs 2.4 days, p 0.001). Patients receiving postoperative antibiotics also showed trends toward higher readmission and reoperation rates (both p 0.06). Postoperative antibiotic treatment for nonperforated appendicitis did not reduce infectious complications and prolonged LOS while increasing postoperative morbidity. Therefore, postoperative antibiotics likely increase the treatment cost for nonperforated appendicitis while not adding an appreciable clinical benefit and, in some cases, actually worsening outcomes. (J Am Coll Surg 2011;213:778 783. 2011 by the American College of Surgeons) Acute appendicitis remains the most common cause of acute abdominal pain necessitating operative treatment. Up to 20% of the population will be affected over the course of their lifetime. 1 If left untreated, acute appendicitis can progress to gangrene of the appendiceal wall and eventual appendiceal perforation with abscess formation. Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein, Editor-in-Chief, has nothing to disclose. Abstract presented at the American College of Surgeons 97th Annual Clinical Congress, Surgical Forum, San Francisco, CA, October 2011. Received July 3, 2011; Revised August 24, 2011; Accepted August 24, 2011. From the Department of Surgery, The Mount Sinai Medical Center, New York, NY. Correspondence address: Celia M Divino, MD, FACS, Department of Surgery, One Gustave L Levy Place, Box 1259, New York, NY 10029. email: celia.divino@mountsinai.org Studies have shown that the stage of appendicitis at the time of operation alters the risk of developing postoperative surgical site infections (SSIs), with acute appendicitis conferring a 9% to 30% risk of superficial SSIs; cases with documented gangrene result in a substantially higher rate. 2-4 In an effort to not only reduce the risk of SSIs, but also bacteremia and sepsis, cases of perforated appendicitis are uniformly treated with a course of postoperative antibiotics. Conversely, supplemental postoperative antibiotics in nonperforated cases are commonly used, but this use remains controversial. Multiple studies have been performed documenting the ability of perioperative antibiotics in reducing the rates of SSIs after appendectomy. 4-6 In 2005, a Cochrane Collaboration review 1 of 45 studies covering 9,576 patients concluded that antibiotic prophylaxis is effective at preventing 2011 by the American College of Surgeons ISSN 1072-7515/11/$36.00 Published by Elsevier Inc. 778 doi:10.1016/j.jamcollsurg.2011.08.018

Vol. 213, No. 6, December 2011 Coakley et al Antibiotics after Appendectomy Worsen Outcomes 779 infectious complications after appendectomy, regardless of whether the dose is given pre-, peri-, or postoperatively. Specifically, the most commonly suggested regimen for appropriate antibiotic coverage is a single preoperative dose of any second-generation cephalosporin. However, virtually all patients undergoing appendectomy, at our institution and others, are given antibiotic prophylaxis preoperatively. Widespread debate persists as to whether any of these patients might also benefit from receiving postoperative antibiotics. Our intention was to compare outcomes of patients treated with antibiotics before and after appendectomy with those of patients who received only preoperative antibiotics. To accomplish this, we reviewed all cases of patients undergoing appendectomy, for which pathology confirmed the diagnosis of nonperforated appendicitis (NA) at The Mount Sinai Medical Center over a 5-year period. Specifically, we set out to determine how the administration of antibiotics after appendectomy for NA, in patients who had already received adequate preoperative antibiotic prophylaxis, affected the risk of developing both postoperative SSIs and antibiotic-related morbidity. METHODS After obtaining approval from the Institutional Review Board at The Mount Sinai Medical Center, the medical records of all patients undergoing appendectomy for NA at our institution from January 1, 2005 through June 31, 2010 were retrospectively reviewed. All patients found to have perforated appendicitis, based on final pathologic examination, and those patients lost to follow-up, were excluded from our analysis. Five patients were thought to have an occult perforation based on the surgeon s operative report, but were included in our cohort due to the pathologist s inability to locate a site of perforation. All patients found to have NA or a normal appendix on final pathologic examination were included. On review of individual medical records, information such as patient demographics, medical comorbidities, imaging data, laboratory values, preoperative antibiotic treatment, perioperative statistics, clinical stage of appendicitis at operation (catarrhal, suppurative, gangrenous, or possibly perforated), and pathologic examination results were recorded. Patient ethnicity was determined based on the patient s response alone and was included to see if there were differences in antibiotic prescribing patterns based on race. Intraoperative hypothermia was defined as a core body temperature of less than 35 o C on at least 2 consecutive vital sign assessments by the corresponding anesthesiologist. In addition, all data regarding postoperative antibiotic administration, postoperative length of stay, readmission, reoperation, time to outpatient follow-up, and development of any postoperative complications were collected. Specifically, all medical records were carefully reviewed for any evidence of superficial SSIs, deep SSIs, organ space SSIs, urinary tract infections, as well as diarrhea and Clostridium difficile infection. The Centers for Disease Control definitions for SSIs were used in all cases. Univariate and multivariate methods were used in comparing patients who did and did not receive postoperative antibiotics. Univariate categorical variables were compared using the chi-square test for large cell counts or Fisher exact test for small cell counts; continuous variables were compared with the 2-independent sample t-test with the Welch correction for unequal variances when applicable. To determine what variables were independently associated with outcomes, multivariate analysis was performed using logistic regression analysis, with the model of best fit determined by the backwards-stepwise regression method. Values of p 0.05 were considered statistically significant for all tests. However, all univariate variables with p values 0.1 were included in our multivariate analysis. All analysis was performed with SPSS (SPSS for Windows, Rel. 11.0.1. 2001.). All statistical methods were reviewed by a statistician. RESULTS In total, 1,144 patients underwent appendectomy at The Mount Sinai Medical Center between January 1, 2005 and July 31, 2010. One hundred forty-four cases were determined to be perforated; the other 1,000 appendectomies were for NA. Of these, sufficient follow-up data were unavailable for 272 patients, leaving 728 cases for inclusion in our analysis. All 728 patients received at least 1 dose of preoperative antibiotics. Within these 728 cases, 394 patients did not receive postoperative antibiotics and 334 were treated with antibiotics after surgery. As depicted in Table 1, there were no significant differences in multiple medical comorbidities (diabetes, liver disease, renal disease, hyperlipidemia, heart disease), patient demographics (age, sex, ethnicitiy), or American Society of Anesthesiologists (ASA) class between the 2 groups. Patients with chronic hypertension were more likely to be treated with postoperative antibiotics (p 0.05). We also saw no significant differences between the 2 groups regarding fever, time from admission to operation, multiple radiologic parameters (appendiceal diameter, presence of stranding, presence of free fluid), laparoscopic vs open operative approach, conversion of laparoscopic to open procedures, estimated blood loss, operating room time, inadvertent spillage of contaminated material, intraoperative transfusion rates, or intraoperative hypothermia.

780 Coakley et al Antibiotics after Appendectomy Worsen Outcomes J Am Coll Surg Table 1. Univariate Analysis of Preoperative Demographics and Statistics vs Postoperative Antibiotic Treatment Postoperative antibiotics (n 334) No postoperative antibiotics (n 394) p Value Variable Male:female 139:195 205:189 0.01 Mean age, y 27.7 29.6 0.2 Ethnicity, n (%) Caucasian 144 (43.1) 192 (48.7) 0.14 Black 29 (8.7) 39 (9.9) 0.6 Hispanic 131 (39.2) 133 (33.8) 0.14 Asian 12 (3.6) 11 (2.8) 0.7 Other 18 (5.4) 18 (4.6) 0.6 Comorbidities, n (%) Hypertension 39 (11.7) 29 (7.4) 0.05 Hyperlipidemia 30 (9.0) 25 (6.4) 0.2 Heart disease 24 (7.2) 17 (4.3) 0.1 Diabetes 11 (3.3) 11 (2.8) 0.7 Liver disease 8 (2.4) 3 (0.8) 0.07 Renal disease 4 (1.2) 1 (0.3) 0.2 Immunosuppressive disease 9 (2.7) 4 (1.0) 0.09 Immunosuppressive medication 10 (3.0) 6 (1.5) 0.2 Current smoker, n (%) 12 (3.6) 17 (4.3) 0.6 Median ASA class 1.50 1.51 0.8 Admission temperature, o C 37.0 36.5 0.01 WBC 14.0 12.3 0.0001 Neutrophilic shift ( 75%) 253/306 (82.7) 226/363 (62.3) 0.0001 Imaging Appendiceal diameter, mm 11.0 10.3 0.01 Presence of free fluid, n (%) 70/238 (29.4%) 57/270 (21.1%) 0.03 Preoperative antibiotics, n (%) Flouroquinolones 51 (15.3) 72 (18.3) 0.3 Cephalosporins 160 (47.9) 229 (58.1) 0.007 Beta-lactams 139 (41.6) 102 (25.9) 0.0001 Aminoglycosides 95 (28.4) 62 (15.7) 0.0001 Other Metronidazole 176 (52.7) 232 (58.9) 0.1 Clindamycin 14 (4.2) 12 (3.1) 0.4 Operative approach, n (%) 0.07 Open 24 (7.2) 16 (4.1) Laparoscopic 310 (92.8) 378 (95.9) Drain placement, n (%) 16 (4.8) 0 (0) 0.0001 Urinary catheter placement, n (%) 290/334 (87) 350/394 (89) 0.43 Operating room time, min 56 59 0.2 Estimated blood loss, ml 16 14 0.3 Intraoperative hypothermia, n (%) 24 (7.2) 33 (8.4) 0.6 Postoperative length of stay, d 2.38 1.14 0.0001 Time to follow-up, d 14.8 14.5 0.3 As a group, patients receiving postoperative antibiotics did present with a higher WBC (12.3 4.8 vs 14.0 4.6 cells/mm 3 ). All 16 patients in whom an indwelling drain was placed at the time of surgery were treated with postoperative antibiotics. Review of operative reports revealed that in 13 of these 16 cases, drains were placed due to the presence of turbid or murky fluid found in the right paracolic gutter, right lower quadrant, or pelvis. Five of these patients developed postoperative SSIs. There were also slight differences in the preoperative antibiotic treat-

Vol. 213, No. 6, December 2011 Coakley et al Antibiotics after Appendectomy Worsen Outcomes 781 Table 2. Univariate Comparison of Pathologic and Clinical Stage of Appendicitis vs Postoperative Antibiotic Treatment Postoperative antibiotics (n 334) No postoperative antibiotics (n 394) Variable n % n % p Value Surgical stage Normal 10 3.0 18 4.6 0.3 Early/catarrhal 95 28.4 190 48.2 0.0001 Suppurative 172 51.5 168 42.6 0.02 Gangrenous 52 15.6 15 3.8 0.0001 Perforated 5 1.5 3 0.8 0.5 Pathologic stage Normal 15 4.5 30 7.6 0.09 Early/catarrhal 81 24.3 161 40.9 0.0001 Suppurative 183 54.8 179 45.4 0.01 Gangrenous 55 16.5 24 6.1 0.0001 ment regimens between the 2 groups. Specifically, patients receiving postoperative antibiotics were more likely to have received beta-lactams and aminoglycosides and less likely to have been given cephalosporins before operation. However, all patients in both groups were determined to have received adequate preoperative antibiotic prophylaxis. Treatment with postoperative antibiotics correlated with a significantly longer postoperative length of stay (1.14 0.5 vs 2.38 0.6 days, p 0.0001). Patients receiving postoperative antibiotics were deemed to have a more advanced stage of disease intraoperatively by the treating surgeon (Table 2). According to phi coefficient analysis, our results indicate a strong correlation between the surgeon s determination of the stage of appendicitis and the pathologist s findings (p 0.0001). Treatment with postoperative antibiotics did not significantly reduce the risk of developing superficial SSIs, deep SSIs, organ space SSIs, or urinary tract infections (Table 3). Interestingly, patients receiving postoperative antibiotics were significantly more likely to sustain a postoperative urinary tract infection (p 0.05). Although Table 3 shows that patients treated with postoperative antibiotics had a higher rate of total infectious complications on univariate analysis, this difference did not remain significant after multivariate comparison. Postoperative antibiotics also strongly correlated with increased risk of Clostridium difficile infection (p 0.02) and postoperative diarrhea (p 0.0001) while showing trends toward higher rates of both readmission (p 0.06) and reoperation (p 0.06). There were several significant differences between patients who did and did not sustain postoperative infectious complications. On univariate analysis, open procedures, increased age, renal disease, hypertension, hyperlipidemia, heart disease, liver disease, and the presence of gangrene (intraoperatively or during pathologic examination) all correlated with a significantly higher risk of developing Table 3. Univariate Comparison of Rates of Postoperative Complications vs Postoperative Antibiotic Treatment Postoperative antibiotics (n 334) No postoperative antibiotics (n 394) Variable n % n % p Value Total infectious complications 37 11.1 24 6.1 0.02 Urinary tract infection 8 2.4 2 0.5 0.05 Postoperative diarrhea 27 8.1 6 1.5 0.0001 Superficial SSI 28 8.4 22 5.6 0.14 Deep SSI 4 1.2 1 0.3 0.19 Organ-space SSI 5 1.5 1 0.3 0.09 Clostridium difficile infection 5 1.5 0 0.0 0.02 Wound dehiscence 2 0.5 0 0.0 0.2 Ileus 9 2.7 5 1.3 0.3 Other 10 3.0 4 1.0 0.1 30-d reoperation 5 1.5 1 0.3 0.06 30-d readmission 11 3.3 5 1.3 0.06 SSI, surgical site infection

782 Coakley et al Antibiotics after Appendectomy Worsen Outcomes J Am Coll Surg Table 4. Multivariate Analysis of Independent Predictors for Postoperative Infectious Complications Factor p Value Heart disease 0.53 Hypertension 0.83 Liver disease 0.43 Immunosuppressive disease 0.61 Leukocytosis (WBC 12,000) 0.22 Fever 0.31 Free fluid (imaging) 0.72 Open procedure 0.0001 Gangrenous disease (intraoperative) 0.02 postoperative infectious complications. However, after multivariate analysis, only open surgery and gangrenous disease remained significant (Table 4). Both the surgeons and pathologists determinations of gangrenous disease correlated with postoperative SSIs on multivariate analysis. DISCUSSION The incidence of postoperative infection after appendectomy has been reported to range from 0% to 11%. 4,6,7 So, our rate of SSI development of 8.5% is consistent with ranges reported in previous studies. Many studies have analyzed the relative efficacy of different antibiotics in preventing postoperative SSIs, and others have attempted to identify the optimal dosing regimens for such antibiotics. However, few studies have assessed the clinical benefit, if any, that might be achieved by administering postoperative antibiotics to patients who had already received adequate preoperative antibiotic prophylaxis. Our literature review revealed only 3 published reports that attempted to address this question. In 1995, Liberman and colleagues 7 randomized 136 patients with NA to receive preoperative cefoxitin, preoperative cefotetan, or combine pre- and postoperative cefoxitin. Their analysis concluded that a single-dose of preoperative cefotetan constituted optimal prophylaxis for nonperforated cases. In 2005, Mui and associates 8 evaluated the outcomes of 269 patients with NA who had been randomized to 1 of 3 separate antibiotic prophylaxis regimens. Specifically, all patients received preoperative cefuroxime plus metronidazole before open appendectomy. Patients were then randomized to receive an additional 3 doses of both drugs, an additional 5 days of both drugs, or no additional prophylaxis. They concluded that all 3 regimens were equally effective because there were no differences in the rate of postoperative SSIs between any of the 3 groups. In 2010, Le and coworkers 9 retrospectively reviewed 507 patients with NA who had received adequate preoperative antibiotics, 63% of whom had received postoperative antibiotics. Their study concluded that the addition of postoperative antibiotics did not affect the rate of postoperative SSIs in their analysis (10% vs 9%, p 0.64). In our analysis, the addition of postoperative antibiotics did not reduce the rate of development of any postoperative SSI. In addition, patients in our study who received postoperative antibiotics were statistically more likely to develop a urinary tract infection or postoperative SSI. However, multivariate analysis confirmed that open appendectomy and gangrenous pathology were the only 2 factors that were independently predictive of sustaining postoperative SSIs. Interestingly, we observed a strong correlation between the surgeon and pathologist assessments as to the severity of appendicitis, a finding not observed in other series. 9,10 Not surprisingly, we observed statistically higher rates of postoperative diarrhea and Clostridium difficileassociated disease among the group treated with postoperative antibiotics. Both of these complications remained significant on multivariate analysis. Many patients sustaining these antibiotic-associated morbidities presented to the emergency department for evaluation, which contributed to the elevated rate of 30-day readmissions among patients receiving postoperative antibiotics (3.3% vs 1.3%, p 0.06). Although not quantified in our analysis, the increased rate of readmission, cost of the antibiotics themselves, and significantly prolonged postoperative length of stay (1.14 vs 2.38 days, p 0.0001) argue strongly that postoperative antibiotics may significantly increase the cost of care for patients with NA. Our study suffers from the same drawbacks as any other nonrandomized retrospective review. Although the 2 groups of patients were fairly well matched, there were differences between them, such as the prevalence of hypertension, sex distribution, and WBC. However, multivariate analysis enabled us to lessen the effects of these discrepancies and reduce the potential bias they could introduce when comparing the 2 groups. In addition, follow-up data was unavailable for 27.2% of our study pool, so these patients had to be excluded from our analysis. There were also differences between the 2 groups regarding preoperative antibiotic regimens, but all patients were judged to have received adequate antibiotic prophylaxis. The group receiving postoperative antibiotics did contain a higher percentage of patients with gangrenous disease. It remains possible that the addition of postoperative antibiotics in these patients therefore suppressed what otherwise may have been an even higher rate of SSIs. However, this question can likely be definitively answered only with a randomized clinical trial. Altogether, our results argue that the addition of postoperative antibiotics for NA does not result in reduced rates of postoperative SSIs. Further, postoperative antibiotics significantly increase the rate of postoperative morbidity in the form

Vol. 213, No. 6, December 2011 Coakley et al Antibiotics after Appendectomy Worsen Outcomes 783 of higher rates of antibiotic-associated diarrhea and Clostridium difficile-associated disease. Taken together, these findings indicate that postoperative antibiotics do not afford any clinical benefit to patients with NA and, in fact, may worsen outcomes while clearly increasing treatment costs. Because open appendectomy and gangrenous pathology were found to be independent predictors of postoperative SSIs, patients meeting these conditions may merit increased consideration for postoperative antibiotic prophylaxis by the treating surgeon. However, additional studies are needed to determine if the addition of postoperative antibiotics provides any clinical benefit even in these select cases. Author Contributions Study conception and design: Coakley, Choi, Divino Acquisition of data: Coakley, Sussman, Wolfson, Bhagavath, Choi, Ranasinghe, Lynn Analysis and interpretation of data: Coakley, Choi Drafting of manuscript: Coakley, Sussman, Wolfson, Bhagavath, Choi, Ranasinghe, Lynn, Divino Critical revision: Coakley, Divino Acknowledgment: The authors thank Kai Dallas, BS, of The Mount Sinai Medical Center (New York, NY), for assistance in performing statistical analysis of study data. REFERENCES 1. Andersen BR, Kallehave FL, Andersen HK. Antibiotics versus placebo for prevention of postoperative infection after appendectomy. Cochrane Database Syst Rev 2001;(3):CD001439. 2. Gottrup F, Hunt TK. Antimicrobial prophylaxis in appendectomy patients. World J Surg 1982;6:306 311. 3. Pieper R, Kage L, Nasman P. Acute appendicitis: a clinical study of 1018 cases of emergency appendectomy. Acta Chir Scand 1982;148:51 62. 4. Winsolow RE, Dean RE, Harley JW. Acute nonperforating appendicitis. Arch Surg 1983;118:651 654. 5. Bauer T, Vennits B, Holm B, et al; Danish Multicenter Study Group III. Antibiotic prophylaxis in acute nonperforated appendicitis. Ann Surg 1989;209:307 311. 6. Busuttil RW, Davidson RK, Fine M, Tompkins RK. Effect of prophylactic antibiotics in acute nonperforated appendicitis: a prospective, double-blind clinical study. Ann Surg 1981;194: 502 509. 7. Liberman MA, Greason KL, Frame S, Ragland JJ. Single-dose cefotetan or cefoxitin versus multiple-dose cefoxitin as prophylaxis in patients undergoing appendectomy for acute nonperforated appendicitis. J Am Coll Surg 1995;180:77 80. 8. Mui LM, Ng CS, Wong SK, et al. Optimum duration of prophylactic antibiotics in acute nonperforated appendicitis. Aust NZJSurg2005;75:425 428. 9. Le D, Rusin W, Hill B, Langell J. Post-operative antibiotic use in nonperforated appendicitis. Am J Surg 2009;198:748 752. 10. Bliss D, Mckee J, Cho D, et al. Discordance of the pediatric surgeon s intraoperative assessment of pediatric appendicitis. J Pediatr Surg 2010;45:1398 1403.