Pediatr Surg Int (2014) 30:587–592 DOI 10.1007/s00383-014-3506-x
ORIGINAL ARTICLE
Surgical site infections in neonates and infants: is antibiotic prophylaxis needed for longer than 24 h? Lan T. Vu • Eric Vittinghoff • Kerilyn K. Nobuhara Diana L. Farmer • Hanmin Lee
•
Accepted: 9 April 2014 / Published online: 9 May 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose The purpose is to determine whether use of perioperative antibiotics for more than 24 h decreases the incidence of SSI in neonates and infants. Methods We studied neonates and infants who had clean– contaminated or contaminated gastrointestinal operations from 1996 to 2006. Patient- and operation-related variables, duration of perioperative antibiotics, and SSI within 30 days were ascertained by retrospective chart review. In assessing the effects of antibiotic duration, we controlled for confounding by indication using standard covariate adjustment and propensity score matching. Results Among 732 operations, the incidence of SSI was 13 %. Using propensity score matching, the odds of SSI were similar (OR 1.1, 95 % CI 0.6–1.9) in patients who received B24 h of postoperative antibiotics compared to [24 h. No difference was also found in standard covariate adjustment. This multivariate model identified three independent predictors of SSI: preoperative infection (OR 3.9, 95 % CI 1.4–10.9) and re-operation through the same incision, both within 30 days (OR 3.5, 95 % CI 1.7–7.4) and later (OR 2.3, 95 % CI 1.4–3.8). L. T. Vu (&) H. Lee Department of Surgery, University of California, 513 Parnassus Avenue, HSW 1601, San Francisco, CA 94143-0570, USA e-mail: [email protected] E. Vittinghoff Department of Epidemiology and Biostatistics, University of California, San Francisco, USA K. K. Nobuhara Washington State Health Care Authority, Olympia, USA D. L. Farmer Department of Surgery, University of California, Davis, USA
Conclusion In clean–contaminated and contaminated gastrointestinal operations, giving [24 h of postoperative antibiotics offered no protection against SSI. An adequately powered randomized clinical trial is needed to conclusively evaluate longer duration antibiotic prophylaxis. Keywords Surgical site infections Neonates Infants Antibiotic prophylaxis Propensity score analysis
Introduction Surgical site infections (SSI) contribute to significant postoperative morbidity in children, leading to prolonged hospital stays for the patients and loss of work productivity for parents and guardians [1, 2]. Retrospective cohort studies have demonstrated rates of postoperative wound infections in neonates and infants between 0.7 and 16.6 % [3–6]. Well-documented deficiencies in neonates’ immunological defense mechanisms, such as impaired chemotaxis and phagocytosis, may place neonates at an increased risk for surgical site infection (SSI). In addition, infants also lack a full complement of immunoglobulins [7–10]. On the basis of this evidence, physicians have argued that neonates and infants should be treated differently from adults, and that the antibiotic prophylaxis guidelines for adults undergoing surgical procedures do not necessary apply to neonates and infants. Even though several retrospective studies and a multicenter, prospective study have identified contamination at operation and duration of the operation as risk factors for SSI in the general pediatric population, they have not addressed the issue of perioperative antibiotics. [5, 6, 11, 12]. It remains unclear if duration of antibiotic prophylaxis influences the development of SSI in the neonates and infants.
123
588
The primary objective of this study was to determine whether giving perioperative antibiotics for more than 24 h decreases the risk of SSI after gastrointestinal operations in neonates and infants. The study also describes the current use of perioperative antibiotics at a single tertiary institution and identifies potential risk factors for the development of SSI in this population.
Pediatr Surg Int (2014) 30:587–592
the patient’s admission history and physical by the presence of maternal fever greater than 38 °C during labor and one of the following: maternal tachycardia (greater than 100 beats/min), fetal tachycardia (greater than 160 beats/ min), uterine tenderness, or foul odor of the amniotic fluid. Maternal medical records were not separately reviewed. Outcome measurement
Methods Study design and data collection The study design was a retrospective chart review of clean– contaminated and contaminated thoracic and abdominal gastrointestinal operations on patients B12 months of age from January 1996 to July 2006 at University of California San Francisco Benioff Children’s Hospital. For this study, we combined the two wound classes of clean–contaminated and contaminated based on the findings of the Vu et al. [13] study in 2009, which demonstrated that there was poor correlation among pediatric surgeons on wound classification, specifically these two categories. The study design included all gastrointestinal operations during this time period to increase the power of the study. The calculated sample size needed to detect 50 % reduction of SSI (from 15 to 7.5 %) in patients receiving [24 h of antibiotics compared to those receiving B24 h of antibiotics was at least 350 patients in each group. We excluded patients who died on the same day of the operation and those who received long-term antibiotic prophylaxis for cardiac or urologic diseases. The study was approved by the UCSF Committee on Human Research. The following information was abstracted from the written and electronic medical records. The primary predictor of interest was the use of perioperative antibiotics, which was dichotomized and defined as B24 h postoperative and [24 h postoperative antibiotics. Patient-specific variables consisted of gender, age at time of operation [neonate (B30 days) or infant (31 days–12 months)], weight at the time of operation (kg), and documented infection at most 7 days before the operation. Operationspecific variables consisted of minimally invasive surgery (i.e., thoracoscopy and laparoscopy), duration of operation (min), re-operation using the same incision, duration between initial operation and re-operation (\30 or C30 days), and American Society of Anesthesiologists preoperative assessment (ASA) score. For neonates, we also collected information on concurrent cardiac anomalies, gestational age at birth, Apgar scores at 1 and 5 min, and maternal history of chorioamnionitis. We documented maternal history of chorioamnionitis if the clinical suspicion of chorioamnionitis was noted in the delivery note or
123
The outcome variable was the development of SSI within 30 days after the operation. SSI is defined by the Center for Disease Control and Prevention [14]. However, it was difficult to adhere to the published definition in a retrospective cohort study, where we had no control over how the outcome variable was documented. Therefore, for the purposes of our study, SSI included terms such as cellulitis, wound erythema, superficial wound infection, wound abscess, wound opening and drainage with positive bacterial cultures, fascial dehiscence, necrotizing soft tissue infection, and intra-abdominal abscess diagnosed by radiographic imaging, direct examination on re-operation, or positive bacterial cultures from drainage or re-operation. One of the investigators served as the primary data abstractor. The inter-rater reliability for the diagnosis of SSI was determined by calculating the kappa score. An additional abstractor independently reviewed a random sample of 40 charts, with oversampling of patients who were identified as having a SSI by the primary abstractor. The kappa for diagnosis of SSI was 0.90. The second abstractor had missed two diagnoses documented in outpatient clinic records. Statistical analysis Confounders in the association between longer duration postoperative antibiotics and SSI were controlled by two separate statistical methods: standard covariate adjustment and propensity score matching. In the first analysis, we used the multivariate logistic regression model that included variables with p \ 0.2. In the second analysis, we used a logistic model to estimate a propensity score for the use of postoperative antibiotics for [24 h compared to B24 h. Variables in the model included gender, age at time of operation (neonate or infant), weight, days to discharge (both log transformed to meet linearity assumptions), history of preoperative infection, minimally invasive versus open operation, site of operation, re-operation, duration of operation, and year of operation. The distribution of propensity scores between the two treatment groups is shown in Fig. 1. We then matched patients who received[24 h of postoperative antibiotics to comparable (similar propensity scores) patients who received B24 h of antibiotics. We examined the balance of pretreatment variables across
Pediatr Surg Int (2014) 30:587–592
589 Table 1 Types of operations Operations Esophageal anastomosis: primary or colon interposition Perineal anoplasty Posterior sagittal anorectoplasty Ostomy closure Gastrostomy Diverting ileostomy or colostomy Duodenoduodenostomy Pull-through for Hirschsprung’s disease Cholecystectomy Hepatic portoenterostomy Ladd’s procedure Excision of choledochal cyst Resection of intestinal stricture or atresia Primary or staged closure of gastroschisis Repair of omphalocele Resection of meconium pseudocyst
Table 2 Standard covariate adjustment Predictor variables
Adjusted OR
95 % confidence interval
p value
(0.8–2.2)
0.2
1.5
(0.95–2.5)
0.08
3.9
(1.4–10.9)
0.009
Preop Antbx or 24 h postop (reference) [24 h Postop Fig. 1 Distribution of propensity scores between the two treatment groups: a propensity scores of patients who had preoperative or at most 24 h of postoperative antibiotics. b Propensity scores of patients who had more than 24 h of postoperative antibiotics
1.4
Patient-related: ASA score: B 2 (reference) [2 Preoperative infection Operation-related:
matched sets using random effects procedures to account for the matching. In the final step, we used a normallogistic model to determine the association of duration of postoperative antibiotic use and SSI between the matched pairs. The method of propensity score matching has been used in previous cohort studies, mostly in the cardiology literature and more recently in the pediatrics literature, to control for confounding by indication [15–17]. All analyses were performed using STATA (version 9.2) software (StataCorp, College Station, Texas). Two-sided p values \0.05 were considered statistically significant.
Results We identified a total of 732 clean–contaminated or contaminated abdominal gastrointestinal operations from 1996 to 2006; 42 % of the operations were performed on neonates. We excluded 26 cases, where perioperative antibiotics were not given. The types of operations included in the study are shown in Table 1. Thirty-day follow-up was
Type of operation: Abdomen (reference) Thorax
0.1
Foreign body 2.7 Re-operation: none (reference)
(0.02–1.0)
0.05
(0.7–9.6)
0.1
B30 days
3.5
(1.7–7.4)
0.001
[30 days
2.3
(1.4–3.8)
0.001
Variables excluded from model for p value greater than 0.2: cardiac anomalies, gestational age at birth, mode of delivery, Apgar scores at 1 and 5 min, gender, ASA score, weight, age, need for intubation, days to discharge, minimally invasive surgery, year of operation, duration of operation
complete in 63.3 % of the patients; average follow-up time was 23.3 (±10) days. The overall incidence of SSI within 30 days of the operation was 13 % (97/732), including 9.1 % (31/342) of patients who received either B24 h of postoperative antibiotics and 16.9 % (66/390) for those who received [24 h of postoperative antibiotics. The analysis using standard covariate adjustment (Table 2) showed no significant association between duration of antibiotic use and SSI (OR
123
590
Pediatr Surg Int (2014) 30:587–592
1.4, 95 % CI 0.8–2.2). This model did identify three independent predictors of SSI, including history of preoperative infection and re-operation through the same incision, both within 30 days and later. In the matched analysis, propensity scores were within the tighter caliper of 0.1 for 82 % of the matched pairs (180/219). The distribution of pretreatment variables was similar in the two matched groups (Table 3). Again, we found no statistically significant association between
Table 3 Comparison of pretreatment variables between the two matched treatment groups Pretreatment variables
Patient-related Gender: female
Treatment A N (%) or mean (±SE)
Treatment B N (%) or mean (±SE)
100 (46)
94 (43)
Weight (kg)
4.3 (±0.1)
4.1 (±0.1)
Age (neonate)
94 (43)
100 (46)
Intubated at birth
74 (34)
79 (36)
Gestational age at birth \32 weeks
75 (34)
81 (37)
32–36 weeks
14 (6)
16 (7)
4 (2)
3 (1)
\4
56 (26)
65 (30)
4–6
11 (5)
[6
19 (9 %)
20 (9)
70 (32)
83 (38)
[36 weeks Apgar score at 1 min
9 (4)
Apgar score at 5 min \4 4–6 [6 Mode of delivery: vaginal Cardiac anomaly Maternal chorioamnionitis Preoperative infection
6 (3)
9 (4)
8 (4) 65 (30)
1 (0.5) 66 (30)
178 (81)
176 (80)
2 (0.9) 1 (0.5)
4 (2) 3 (1)
ASA: [2
117 (53)
125 (57)
Days to discharge
12.2 (±0.7)
12.7 (±0.6)
Operation-related Foreign body
3 (1)
6 (3)
26 (12)
23 (11)
Site of operation: abdomen (thorax)
200 (91)
204 (93)
Re-operation: none B30 days
166 (76) 8 (4)
160 (73) 10 (5)
Minimally invasive
[30 days Duration of operation (min) Year: 1996–1997 (1998–2006)
45 (21)
49 (22)
90.6 (±4.5)
97.3 (±4.5)
25 (11)
18 (8)
A preoperative or at most 24 h of antibiotics, B greater than 24 h of antibiotics
123
duration of postoperative antibiotic use and risk of SSI (OR 1.1, 95 % CI 0.6–1.9).
Discussion In this observational study, we found no statistical evidence that the use of more than 24 h of postoperative antibiotics among neonates and infants undergoing clean–contaminated or contaminated abdominal operations reduces the odds of SSI. Analyses using extensive covariate adjustment and pair matching on propensity scores gave qualitatively consistent results. The overall incidence of SSI in our study was higher than what has been reported in the literature for infants and neonates [5, 6, 11, 12]. One possible explanation is that we excluded clean cases, in which there is little rationale for longer use of antibiotics. In addition, the patients in our study who were at highest risk for SSI include those who undergo re-operation through the same incision and have a history of preoperative infection. Operation-related factors, such as wound contamination at operation and duration of the operation, have been documented as risk factors for SSI in the general pediatric population [5, 6, 11, 12]. However, the issue of perioperative antibiotics has not been adequately addressed. Previous studies did attempt to document preoperative antibiotic use and found that in clean surgical cases, preoperative antibiotics did not significantly change the incidence of postoperative SSI [11]. In addition, one randomized control trial found no difference in the incidence of SSI between infants who received a single dose of preoperative antibiotics and those who did not receive antibiotics for open pyloromyotomy (absolute risk reduction of 2.7 %; 95 % confidence interval between a reduction of 9.8 % to an increase of 4.5 %) [18]. In addition, Breckler et al. [19] studied 118 children who underwent colostomy closure for anorectal malformations. They found that 66 % of the patients received antibiotics for 24 h or less and 28 % of the patients received greater than 24 h of antibiotics. The overall incidence of SSI was 14 %. However, the sample size was too small to determine if the longer duration of antibiotics made a difference in the incidence of SSI. In response to concerns about theoretical immunosuppression in neonates and infants, the trend in pediatrics has been to overuse antibiotics, at least in part because unlike adults, who can communicate their symptoms, infants often lack any definite clinical findings. However, as the medical community has witnessed with adult patients, bacterial resistance has become a major public health issue. Methicillin-resistant Staphylococcus aureus, once a nosocomial infection, now is more prevalent in the community [20].
Pediatr Surg Int (2014) 30:587–592
Exposure to antibiotics at an early age can lead to the potential development of resistance at a later age for the patient and also the development of more virulent microflora in the hospital. In 2002, the Center for Disease Control launched the 12-Step Campaign to Prevent Antimicrobial Resistance to educate clinicians and provide strategies to change clinical practices, including antimicrobial prescribing [21]. Patel et al. [22] published a multicenter retrospective observational study of antibiotic use in four tertiary care NICUs and found that approximately 25 % of antibiotic courses and antibiotic days were considered inappropriate. In addition, the category of inappropriate use of cefazolin reflected prolonged surgical prophylaxis, which violated the CDC step 6 of ‘‘Practice antimicrobial control’’. The results of our study argue that increased duration of perioperative antibiotics for more than 24 h may not benefit neonates and infants undergoing clean–contaminated or contaminated thoracic or abdominal gastrointestinal operations. The baseline incidence of SSI was 9 % in patients who received only preoperative antibiotics, or at most 24 h of postoperative antibiotics. A clinically significant benefit from longer duration of antibiotics would include a greater than 50 % reduction in incidence. However, the lower limit of the 95 % confidence interval of 0.8 for the standard covariate adjustment analysis and 0.6 for the propensity score matching analysis in our study excluded a relative risk reduction of 50 %. Therefore, increasing the sample size would not change the results of this study. As this was an observational study, our results were potentially biased by treatment selection. The study used two statistical analysis methods to control for treatment selection bias: standard covariate adjustment and propensity score matching. However, both statistical methods are limited by their ability to control for only known and measured confounders, so that residual confounding cannot be ruled out. In addition, our cohort of patients included different types of operations and antibiotics. It was not feasible to examine only one type of operation, or to model the effect of different antibiotics, due to the lack of available numbers. This may have obscured a protective effect of using more than 24 h of postoperative antibiotics if the benefit is restricted to certain subtypes of operations or antibiotics. In addition, only 63.3 % of the patients had complete 30-day follow-up. This may have underestimated the true incidence of surgical site infections in our study population. However, we do not think the length of followup is biased toward one group of patient or another. Therefore, it should not affect the association between potential risk factors and development of surgical site infection. In conclusion, our observational study was unable to show that using more than 24 h of postoperative antibiotics
591
among neonates and infants undergoing clean–contaminated or contaminated thoracic and abdominal gastrointestinal operations reduces the risk of SSI. Despite careful adjustment for a range of risk factors using two statistical methods, residual confounding by indication cannot be completely ruled out. An adequately powered multicenter randomized trial is needed to definitively evaluate the benefit of longer duration antibiotic prophylaxis in this patient population.
References 1. Fry DE (2002) The economic costs of surgical site infection. Surg Infect 3(Suppl 1):37–43 2. Urban JA (2006) Cost analysis of surgical site infections. Surg Infect 7(Suppl 1):19–22 3. Duque-Estrada EO, Duarte MR, Rodrigues DM, Raphael MD (2003) Wound infections in pediatric surgery: a study of 575 patients in a university hospital. Pediatr Surg Int 19(6):436–438 4. Sharma LK, Sharma PK (1986) Postoperative wound infection in a pediatric surgical service. J Pediatr Surg 21(10):889–891 5. Bhattacharyya N, Kosloske AM (1990) Postoperative wound infection in pediatric surgical patients: a study of 676 infants and children. J Pediatr Surg 25(1):125–129 6. Davenport M, Doig CM (1993) Wound infection in pediatric surgery: a study in 1,094 neonates. J Pediatr Surg 28(1):26–30 7. Chiswick ML (1983) Infection and defences in neonates. Br Med J (Clin Res Ed) 286(6375):1377–1378 8. Miller ME (1969) Phagocytosis in the newborn infant: humoral and cellular factors. J Pediatr 74(2):255–259 9. Velilla PA, Rugeles MT, Chougnet CA (2006) Defective antigenpresenting cell function in human neonates. Clin Immunol 121(3):251–259 10. Kovarik J, Siegrist CA (1998) Immunity in early life. Immunol Today 19(4):150–152 11. Horwitz JR, Chwals WJ, Doski JJ, Suescun EA, Cheu HW, Lally KP (1998) Pediatric wound infections: a prospective multicenter study. Ann Surg 227(4):553–558 12. Casanova JF, Herruzo R, Diez J (2006) Risk factors for surgical site infection in children. Infect Control Hosp Epidemiol 27(7):709–715 13. Vu LT, Nobuhara KK, Lee H, Farmer DL (2009) Conflicts in wound classification of neonatal operations. J Pediatr Surg 44(6):1206–1211 14. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG (1992) CDC definitions of nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections. Am J Infect Control 20(5):271–274 15. Barrett CS, Jaggers JJ, Cook EF, Graham DA, Yarlagadda VV, Teele SA et al (2013) Pediatric ECMO outcomes: comparison of centrifugal versus roller blood pumps using propensity score matching. ASAIO J 59(2):145–151 16. Bird TM, Bronstein JM, Hall RW, Lowery CL, Nugent R, Mays GP (2010) Late preterm infants: birth outcomes and health care utilization in the first year. Pediatrics 126(2):e311–e319 17. Latib A, Maisano F, Bertoldi L, Giacomini A, Shannon J, Cioni M et al (2012) Transcatheter vs surgical aortic valve replacement in intermediate-surgical-risk patients with aortic stenosis: a propensity score-matched case–control study. Am Heart J 164(6):910–917 18. Nour S, MacKinnon AE, Dickson JA, Walker J (1996) Antibiotic prophylaxis for infantile pyloromyotomy. J R Coll Surg Edinb 41(3):178–180
123
592 19. Breckler FD, Rescorla FJ, Billmire DF (2010) Wound infection after colostomy closure for imperforate anus in children: utility of preoperative oral antibiotics. J Pediatr Surg 45(7):1509–1513 20. Moran GJ, Krishnadasan A, Gorwitz RJ, Forsheim GE, McDougal LK, Carey RB, Talan DA (2006) Methicillin resistant S. aureus infections among patients in the emergency department. N Engl J Med 355(7):666–674 21. Center for Disease Control and Prevention (2002) 12-Step program to prevent antimicrobial resistance in health care settings.
123
Pediatr Surg Int (2014) 30:587–592 Available from: http://www.cdc.gov/drugresistance/healthcare/ default.html 22. Patel SJ, Oshodi A, Prasad P, Delamora P, Larson E, Zaoutis T et al (2009) Antibiotic use in neonatal intensive care units and adherence with Centers for Disease Control and Prevention 12 Step Campaign to Prevent Antimicrobial Resistance. Pediatr Infect Dis J 28(12):1047–1051
ORIGINAL ARTICLE
Surgical site infections in neonates and infants: is antibiotic prophylaxis needed for longer than 24 h? Lan T. Vu • Eric Vittinghoff • Kerilyn K. Nobuhara Diana L. Farmer • Hanmin Lee
•
Accepted: 9 April 2014 / Published online: 9 May 2014 Ó Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose The purpose is to determine whether use of perioperative antibiotics for more than 24 h decreases the incidence of SSI in neonates and infants. Methods We studied neonates and infants who had clean– contaminated or contaminated gastrointestinal operations from 1996 to 2006. Patient- and operation-related variables, duration of perioperative antibiotics, and SSI within 30 days were ascertained by retrospective chart review. In assessing the effects of antibiotic duration, we controlled for confounding by indication using standard covariate adjustment and propensity score matching. Results Among 732 operations, the incidence of SSI was 13 %. Using propensity score matching, the odds of SSI were similar (OR 1.1, 95 % CI 0.6–1.9) in patients who received B24 h of postoperative antibiotics compared to [24 h. No difference was also found in standard covariate adjustment. This multivariate model identified three independent predictors of SSI: preoperative infection (OR 3.9, 95 % CI 1.4–10.9) and re-operation through the same incision, both within 30 days (OR 3.5, 95 % CI 1.7–7.4) and later (OR 2.3, 95 % CI 1.4–3.8). L. T. Vu (&) H. Lee Department of Surgery, University of California, 513 Parnassus Avenue, HSW 1601, San Francisco, CA 94143-0570, USA e-mail: [email protected] E. Vittinghoff Department of Epidemiology and Biostatistics, University of California, San Francisco, USA K. K. Nobuhara Washington State Health Care Authority, Olympia, USA D. L. Farmer Department of Surgery, University of California, Davis, USA
Conclusion In clean–contaminated and contaminated gastrointestinal operations, giving [24 h of postoperative antibiotics offered no protection against SSI. An adequately powered randomized clinical trial is needed to conclusively evaluate longer duration antibiotic prophylaxis. Keywords Surgical site infections Neonates Infants Antibiotic prophylaxis Propensity score analysis
Introduction Surgical site infections (SSI) contribute to significant postoperative morbidity in children, leading to prolonged hospital stays for the patients and loss of work productivity for parents and guardians [1, 2]. Retrospective cohort studies have demonstrated rates of postoperative wound infections in neonates and infants between 0.7 and 16.6 % [3–6]. Well-documented deficiencies in neonates’ immunological defense mechanisms, such as impaired chemotaxis and phagocytosis, may place neonates at an increased risk for surgical site infection (SSI). In addition, infants also lack a full complement of immunoglobulins [7–10]. On the basis of this evidence, physicians have argued that neonates and infants should be treated differently from adults, and that the antibiotic prophylaxis guidelines for adults undergoing surgical procedures do not necessary apply to neonates and infants. Even though several retrospective studies and a multicenter, prospective study have identified contamination at operation and duration of the operation as risk factors for SSI in the general pediatric population, they have not addressed the issue of perioperative antibiotics. [5, 6, 11, 12]. It remains unclear if duration of antibiotic prophylaxis influences the development of SSI in the neonates and infants.
123
588
The primary objective of this study was to determine whether giving perioperative antibiotics for more than 24 h decreases the risk of SSI after gastrointestinal operations in neonates and infants. The study also describes the current use of perioperative antibiotics at a single tertiary institution and identifies potential risk factors for the development of SSI in this population.
Pediatr Surg Int (2014) 30:587–592
the patient’s admission history and physical by the presence of maternal fever greater than 38 °C during labor and one of the following: maternal tachycardia (greater than 100 beats/min), fetal tachycardia (greater than 160 beats/ min), uterine tenderness, or foul odor of the amniotic fluid. Maternal medical records were not separately reviewed. Outcome measurement
Methods Study design and data collection The study design was a retrospective chart review of clean– contaminated and contaminated thoracic and abdominal gastrointestinal operations on patients B12 months of age from January 1996 to July 2006 at University of California San Francisco Benioff Children’s Hospital. For this study, we combined the two wound classes of clean–contaminated and contaminated based on the findings of the Vu et al. [13] study in 2009, which demonstrated that there was poor correlation among pediatric surgeons on wound classification, specifically these two categories. The study design included all gastrointestinal operations during this time period to increase the power of the study. The calculated sample size needed to detect 50 % reduction of SSI (from 15 to 7.5 %) in patients receiving [24 h of antibiotics compared to those receiving B24 h of antibiotics was at least 350 patients in each group. We excluded patients who died on the same day of the operation and those who received long-term antibiotic prophylaxis for cardiac or urologic diseases. The study was approved by the UCSF Committee on Human Research. The following information was abstracted from the written and electronic medical records. The primary predictor of interest was the use of perioperative antibiotics, which was dichotomized and defined as B24 h postoperative and [24 h postoperative antibiotics. Patient-specific variables consisted of gender, age at time of operation [neonate (B30 days) or infant (31 days–12 months)], weight at the time of operation (kg), and documented infection at most 7 days before the operation. Operationspecific variables consisted of minimally invasive surgery (i.e., thoracoscopy and laparoscopy), duration of operation (min), re-operation using the same incision, duration between initial operation and re-operation (\30 or C30 days), and American Society of Anesthesiologists preoperative assessment (ASA) score. For neonates, we also collected information on concurrent cardiac anomalies, gestational age at birth, Apgar scores at 1 and 5 min, and maternal history of chorioamnionitis. We documented maternal history of chorioamnionitis if the clinical suspicion of chorioamnionitis was noted in the delivery note or
123
The outcome variable was the development of SSI within 30 days after the operation. SSI is defined by the Center for Disease Control and Prevention [14]. However, it was difficult to adhere to the published definition in a retrospective cohort study, where we had no control over how the outcome variable was documented. Therefore, for the purposes of our study, SSI included terms such as cellulitis, wound erythema, superficial wound infection, wound abscess, wound opening and drainage with positive bacterial cultures, fascial dehiscence, necrotizing soft tissue infection, and intra-abdominal abscess diagnosed by radiographic imaging, direct examination on re-operation, or positive bacterial cultures from drainage or re-operation. One of the investigators served as the primary data abstractor. The inter-rater reliability for the diagnosis of SSI was determined by calculating the kappa score. An additional abstractor independently reviewed a random sample of 40 charts, with oversampling of patients who were identified as having a SSI by the primary abstractor. The kappa for diagnosis of SSI was 0.90. The second abstractor had missed two diagnoses documented in outpatient clinic records. Statistical analysis Confounders in the association between longer duration postoperative antibiotics and SSI were controlled by two separate statistical methods: standard covariate adjustment and propensity score matching. In the first analysis, we used the multivariate logistic regression model that included variables with p \ 0.2. In the second analysis, we used a logistic model to estimate a propensity score for the use of postoperative antibiotics for [24 h compared to B24 h. Variables in the model included gender, age at time of operation (neonate or infant), weight, days to discharge (both log transformed to meet linearity assumptions), history of preoperative infection, minimally invasive versus open operation, site of operation, re-operation, duration of operation, and year of operation. The distribution of propensity scores between the two treatment groups is shown in Fig. 1. We then matched patients who received[24 h of postoperative antibiotics to comparable (similar propensity scores) patients who received B24 h of antibiotics. We examined the balance of pretreatment variables across
Pediatr Surg Int (2014) 30:587–592
589 Table 1 Types of operations Operations Esophageal anastomosis: primary or colon interposition Perineal anoplasty Posterior sagittal anorectoplasty Ostomy closure Gastrostomy Diverting ileostomy or colostomy Duodenoduodenostomy Pull-through for Hirschsprung’s disease Cholecystectomy Hepatic portoenterostomy Ladd’s procedure Excision of choledochal cyst Resection of intestinal stricture or atresia Primary or staged closure of gastroschisis Repair of omphalocele Resection of meconium pseudocyst
Table 2 Standard covariate adjustment Predictor variables
Adjusted OR
95 % confidence interval
p value
(0.8–2.2)
0.2
1.5
(0.95–2.5)
0.08
3.9
(1.4–10.9)
0.009
Preop Antbx or 24 h postop (reference) [24 h Postop Fig. 1 Distribution of propensity scores between the two treatment groups: a propensity scores of patients who had preoperative or at most 24 h of postoperative antibiotics. b Propensity scores of patients who had more than 24 h of postoperative antibiotics
1.4
Patient-related: ASA score: B 2 (reference) [2 Preoperative infection Operation-related:
matched sets using random effects procedures to account for the matching. In the final step, we used a normallogistic model to determine the association of duration of postoperative antibiotic use and SSI between the matched pairs. The method of propensity score matching has been used in previous cohort studies, mostly in the cardiology literature and more recently in the pediatrics literature, to control for confounding by indication [15–17]. All analyses were performed using STATA (version 9.2) software (StataCorp, College Station, Texas). Two-sided p values \0.05 were considered statistically significant.
Results We identified a total of 732 clean–contaminated or contaminated abdominal gastrointestinal operations from 1996 to 2006; 42 % of the operations were performed on neonates. We excluded 26 cases, where perioperative antibiotics were not given. The types of operations included in the study are shown in Table 1. Thirty-day follow-up was
Type of operation: Abdomen (reference) Thorax
0.1
Foreign body 2.7 Re-operation: none (reference)
(0.02–1.0)
0.05
(0.7–9.6)
0.1
B30 days
3.5
(1.7–7.4)
0.001
[30 days
2.3
(1.4–3.8)
0.001
Variables excluded from model for p value greater than 0.2: cardiac anomalies, gestational age at birth, mode of delivery, Apgar scores at 1 and 5 min, gender, ASA score, weight, age, need for intubation, days to discharge, minimally invasive surgery, year of operation, duration of operation
complete in 63.3 % of the patients; average follow-up time was 23.3 (±10) days. The overall incidence of SSI within 30 days of the operation was 13 % (97/732), including 9.1 % (31/342) of patients who received either B24 h of postoperative antibiotics and 16.9 % (66/390) for those who received [24 h of postoperative antibiotics. The analysis using standard covariate adjustment (Table 2) showed no significant association between duration of antibiotic use and SSI (OR
123
590
Pediatr Surg Int (2014) 30:587–592
1.4, 95 % CI 0.8–2.2). This model did identify three independent predictors of SSI, including history of preoperative infection and re-operation through the same incision, both within 30 days and later. In the matched analysis, propensity scores were within the tighter caliper of 0.1 for 82 % of the matched pairs (180/219). The distribution of pretreatment variables was similar in the two matched groups (Table 3). Again, we found no statistically significant association between
Table 3 Comparison of pretreatment variables between the two matched treatment groups Pretreatment variables
Patient-related Gender: female
Treatment A N (%) or mean (±SE)
Treatment B N (%) or mean (±SE)
100 (46)
94 (43)
Weight (kg)
4.3 (±0.1)
4.1 (±0.1)
Age (neonate)
94 (43)
100 (46)
Intubated at birth
74 (34)
79 (36)
Gestational age at birth \32 weeks
75 (34)
81 (37)
32–36 weeks
14 (6)
16 (7)
4 (2)
3 (1)
\4
56 (26)
65 (30)
4–6
11 (5)
[6
19 (9 %)
20 (9)
70 (32)
83 (38)
[36 weeks Apgar score at 1 min
9 (4)
Apgar score at 5 min \4 4–6 [6 Mode of delivery: vaginal Cardiac anomaly Maternal chorioamnionitis Preoperative infection
6 (3)
9 (4)
8 (4) 65 (30)
1 (0.5) 66 (30)
178 (81)
176 (80)
2 (0.9) 1 (0.5)
4 (2) 3 (1)
ASA: [2
117 (53)
125 (57)
Days to discharge
12.2 (±0.7)
12.7 (±0.6)
Operation-related Foreign body
3 (1)
6 (3)
26 (12)
23 (11)
Site of operation: abdomen (thorax)
200 (91)
204 (93)
Re-operation: none B30 days
166 (76) 8 (4)
160 (73) 10 (5)
Minimally invasive
[30 days Duration of operation (min) Year: 1996–1997 (1998–2006)
45 (21)
49 (22)
90.6 (±4.5)
97.3 (±4.5)
25 (11)
18 (8)
A preoperative or at most 24 h of antibiotics, B greater than 24 h of antibiotics
123
duration of postoperative antibiotic use and risk of SSI (OR 1.1, 95 % CI 0.6–1.9).
Discussion In this observational study, we found no statistical evidence that the use of more than 24 h of postoperative antibiotics among neonates and infants undergoing clean–contaminated or contaminated abdominal operations reduces the odds of SSI. Analyses using extensive covariate adjustment and pair matching on propensity scores gave qualitatively consistent results. The overall incidence of SSI in our study was higher than what has been reported in the literature for infants and neonates [5, 6, 11, 12]. One possible explanation is that we excluded clean cases, in which there is little rationale for longer use of antibiotics. In addition, the patients in our study who were at highest risk for SSI include those who undergo re-operation through the same incision and have a history of preoperative infection. Operation-related factors, such as wound contamination at operation and duration of the operation, have been documented as risk factors for SSI in the general pediatric population [5, 6, 11, 12]. However, the issue of perioperative antibiotics has not been adequately addressed. Previous studies did attempt to document preoperative antibiotic use and found that in clean surgical cases, preoperative antibiotics did not significantly change the incidence of postoperative SSI [11]. In addition, one randomized control trial found no difference in the incidence of SSI between infants who received a single dose of preoperative antibiotics and those who did not receive antibiotics for open pyloromyotomy (absolute risk reduction of 2.7 %; 95 % confidence interval between a reduction of 9.8 % to an increase of 4.5 %) [18]. In addition, Breckler et al. [19] studied 118 children who underwent colostomy closure for anorectal malformations. They found that 66 % of the patients received antibiotics for 24 h or less and 28 % of the patients received greater than 24 h of antibiotics. The overall incidence of SSI was 14 %. However, the sample size was too small to determine if the longer duration of antibiotics made a difference in the incidence of SSI. In response to concerns about theoretical immunosuppression in neonates and infants, the trend in pediatrics has been to overuse antibiotics, at least in part because unlike adults, who can communicate their symptoms, infants often lack any definite clinical findings. However, as the medical community has witnessed with adult patients, bacterial resistance has become a major public health issue. Methicillin-resistant Staphylococcus aureus, once a nosocomial infection, now is more prevalent in the community [20].
Pediatr Surg Int (2014) 30:587–592
Exposure to antibiotics at an early age can lead to the potential development of resistance at a later age for the patient and also the development of more virulent microflora in the hospital. In 2002, the Center for Disease Control launched the 12-Step Campaign to Prevent Antimicrobial Resistance to educate clinicians and provide strategies to change clinical practices, including antimicrobial prescribing [21]. Patel et al. [22] published a multicenter retrospective observational study of antibiotic use in four tertiary care NICUs and found that approximately 25 % of antibiotic courses and antibiotic days were considered inappropriate. In addition, the category of inappropriate use of cefazolin reflected prolonged surgical prophylaxis, which violated the CDC step 6 of ‘‘Practice antimicrobial control’’. The results of our study argue that increased duration of perioperative antibiotics for more than 24 h may not benefit neonates and infants undergoing clean–contaminated or contaminated thoracic or abdominal gastrointestinal operations. The baseline incidence of SSI was 9 % in patients who received only preoperative antibiotics, or at most 24 h of postoperative antibiotics. A clinically significant benefit from longer duration of antibiotics would include a greater than 50 % reduction in incidence. However, the lower limit of the 95 % confidence interval of 0.8 for the standard covariate adjustment analysis and 0.6 for the propensity score matching analysis in our study excluded a relative risk reduction of 50 %. Therefore, increasing the sample size would not change the results of this study. As this was an observational study, our results were potentially biased by treatment selection. The study used two statistical analysis methods to control for treatment selection bias: standard covariate adjustment and propensity score matching. However, both statistical methods are limited by their ability to control for only known and measured confounders, so that residual confounding cannot be ruled out. In addition, our cohort of patients included different types of operations and antibiotics. It was not feasible to examine only one type of operation, or to model the effect of different antibiotics, due to the lack of available numbers. This may have obscured a protective effect of using more than 24 h of postoperative antibiotics if the benefit is restricted to certain subtypes of operations or antibiotics. In addition, only 63.3 % of the patients had complete 30-day follow-up. This may have underestimated the true incidence of surgical site infections in our study population. However, we do not think the length of followup is biased toward one group of patient or another. Therefore, it should not affect the association between potential risk factors and development of surgical site infection. In conclusion, our observational study was unable to show that using more than 24 h of postoperative antibiotics
591
among neonates and infants undergoing clean–contaminated or contaminated thoracic and abdominal gastrointestinal operations reduces the risk of SSI. Despite careful adjustment for a range of risk factors using two statistical methods, residual confounding by indication cannot be completely ruled out. An adequately powered multicenter randomized trial is needed to definitively evaluate the benefit of longer duration antibiotic prophylaxis in this patient population.
References 1. Fry DE (2002) The economic costs of surgical site infection. Surg Infect 3(Suppl 1):37–43 2. Urban JA (2006) Cost analysis of surgical site infections. Surg Infect 7(Suppl 1):19–22 3. Duque-Estrada EO, Duarte MR, Rodrigues DM, Raphael MD (2003) Wound infections in pediatric surgery: a study of 575 patients in a university hospital. Pediatr Surg Int 19(6):436–438 4. Sharma LK, Sharma PK (1986) Postoperative wound infection in a pediatric surgical service. J Pediatr Surg 21(10):889–891 5. Bhattacharyya N, Kosloske AM (1990) Postoperative wound infection in pediatric surgical patients: a study of 676 infants and children. J Pediatr Surg 25(1):125–129 6. Davenport M, Doig CM (1993) Wound infection in pediatric surgery: a study in 1,094 neonates. J Pediatr Surg 28(1):26–30 7. Chiswick ML (1983) Infection and defences in neonates. Br Med J (Clin Res Ed) 286(6375):1377–1378 8. Miller ME (1969) Phagocytosis in the newborn infant: humoral and cellular factors. J Pediatr 74(2):255–259 9. Velilla PA, Rugeles MT, Chougnet CA (2006) Defective antigenpresenting cell function in human neonates. Clin Immunol 121(3):251–259 10. Kovarik J, Siegrist CA (1998) Immunity in early life. Immunol Today 19(4):150–152 11. Horwitz JR, Chwals WJ, Doski JJ, Suescun EA, Cheu HW, Lally KP (1998) Pediatric wound infections: a prospective multicenter study. Ann Surg 227(4):553–558 12. Casanova JF, Herruzo R, Diez J (2006) Risk factors for surgical site infection in children. Infect Control Hosp Epidemiol 27(7):709–715 13. Vu LT, Nobuhara KK, Lee H, Farmer DL (2009) Conflicts in wound classification of neonatal operations. J Pediatr Surg 44(6):1206–1211 14. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG (1992) CDC definitions of nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections. Am J Infect Control 20(5):271–274 15. Barrett CS, Jaggers JJ, Cook EF, Graham DA, Yarlagadda VV, Teele SA et al (2013) Pediatric ECMO outcomes: comparison of centrifugal versus roller blood pumps using propensity score matching. ASAIO J 59(2):145–151 16. Bird TM, Bronstein JM, Hall RW, Lowery CL, Nugent R, Mays GP (2010) Late preterm infants: birth outcomes and health care utilization in the first year. Pediatrics 126(2):e311–e319 17. Latib A, Maisano F, Bertoldi L, Giacomini A, Shannon J, Cioni M et al (2012) Transcatheter vs surgical aortic valve replacement in intermediate-surgical-risk patients with aortic stenosis: a propensity score-matched case–control study. Am Heart J 164(6):910–917 18. Nour S, MacKinnon AE, Dickson JA, Walker J (1996) Antibiotic prophylaxis for infantile pyloromyotomy. J R Coll Surg Edinb 41(3):178–180
123
592 19. Breckler FD, Rescorla FJ, Billmire DF (2010) Wound infection after colostomy closure for imperforate anus in children: utility of preoperative oral antibiotics. J Pediatr Surg 45(7):1509–1513 20. Moran GJ, Krishnadasan A, Gorwitz RJ, Forsheim GE, McDougal LK, Carey RB, Talan DA (2006) Methicillin resistant S. aureus infections among patients in the emergency department. N Engl J Med 355(7):666–674 21. Center for Disease Control and Prevention (2002) 12-Step program to prevent antimicrobial resistance in health care settings.
123
Pediatr Surg Int (2014) 30:587–592 Available from: http://www.cdc.gov/drugresistance/healthcare/ default.html 22. Patel SJ, Oshodi A, Prasad P, Delamora P, Larson E, Zaoutis T et al (2009) Antibiotic use in neonatal intensive care units and adherence with Centers for Disease Control and Prevention 12 Step Campaign to Prevent Antimicrobial Resistance. Pediatr Infect Dis J 28(12):1047–1051
![[Surgical site infections: antibiotic prophylaxis in surgery].](/assets/img/hold.png)
