The role of anesthetic management in surgical site infections after pediatric intestinal surgery

Structured Summary

Background:

Although surgical site infections (SSIs) remain a significant health care issue, a limited number of studies have analyzed risk factors for SSIs in children, particularly the role of intraoperative anesthetic management. Pediatric patients are less likely to have major adult risk factors for SSIs such as smoking and diabetes. Thus children may be more suitable as a cohort for examining the role of intraoperative anesthetics in SSIs.

Aim:

We examined an association between SSI incidence and anesthetic management in children who underwent elective intestinal surgery in a single institution.

Methods:

We performed a retrospective study of 621 patients who underwent elective intestinal surgery under general anesthesia between January 2017 and September 2019, with primary outcome as the incidence of SSI. We compared patients who were dichotomized according to the median of the sevoflurane dose. We used propensity score (PS) pairwise matching of these patients to avoid selection biases. PS matching yielded 204 pairs of patients.

Findings:

We found that higher doses of sevoflurane were associated with higher incidence of SSIs (9.8% vs. 3.9%, p = 0.019). We adjusted for intraoperative factors that were not included in the PS adjustment factors, and multivariate regression analysis after PS matching showed compatible results (odds ratio: 2.58, 95% confidence interval: 1.11–6.04, p=0.028).

Conclusion:

Higher doses of sevoflurane are associated with an increased odds of SSI after pediatric elective intestinal surgery. A randomized controlled study of volatile anesthetic-based versus intravenous anesthetic-based anesthesia will be needed to further determine the role of anesthetic drugs in SSI risk.

Introduction

Surgical site infections (SSIs) are the most common healthcare-associated infections in surgical patients, with substantial morbidities and mortality [1, 2]. Intestinal surgery is associated with the highest rate of SSIs [3]. Although SSI prevention bundles have been widely introduced to attenuate the incidence of SSIs [4], it remains high in intestinal surgery [5], indicating the need for additional strategies to decrease SSIs.

The World Health Organization (WHO) published expert recommendations for perioperative management to mitigate SSIs [6, 7]. The guideline recommended antimicrobial prophylaxis, glycemic control, maintenance of normothermia [8, 9], adequate oxygenation[10], and skin preparation [11]. In addition, multiple studies have identified patients’ underlying diseases (diabetes, obesity, etc.) and smoking as risk factors [12]. Blood transfusion [13], surgical duration, and operator have been also identified as risk factors [14].

Despite the fact that anesthesia is an essential component of surgery and many studies have examined the association between anesthetics and immune responses [15–18], only a limited number of clinical studies have reported on the association between anesthetics and SSIs in adult patients [19, 20].

General anesthesia is administered by volatile anesthetics (VAs) and intravenous anesthetics. We previously showed that use of VAs was associated with a reduction in granulocyte phagocytosis in pediatric patients undergoing cardiac catheterization, whereas propofol-based intravenous anesthesia was not [21]. In vivo mouse experiments showed that VA exposure exacerbated bacterial loads in the wound compared to the intravenous anesthesia [21, 22]. Because patients’ factors are often very heterogeneous, it is important to evaluate the impact of anesthetic drugs in the population with appropriate adjustment for potential confounding factors. For this purpose, it may be suitable to assess the effects of anesthetics in children because they are less likely to have confounders such as diabetes and smoking. Here, we examined the incidence and characteristics of SSI after intestinal surgery in children as well as the clinical association between SSIs and intraoperative anesthetic management in a single institution.

Methods

Study cohort and patient background

The protocol was approved by the Institutional Review Board at Boston Children’s Hospital. Informed consent was waived by the Institutional Review Board. We identified patients who underwent elective intestinal surgery at Boston Children’s Hospital from January 2017 to September 2019. The patients received similar SSI prevention bundles and postoperative SSI surveillance was conducted by the Infection Prevention and Control Department at Boston Children’s Hospital. We excluded patients who underwent endoscopic examination, endoscopic extraction with no wounds, or urgent surgery. Surgical sites were classified according to the name of the procedure in the AIMS as follows: upper gastrointestinal tract (esophagus to duodenum), lower gastrointestinal tract (small intestine to rectum), and others (e.g., extensive and unclassifiable). Preoperatively patients received skin preparation either with povidone-iodine (betadine) alone, chlorhexidine gluconate (CHG) plus alcohol, betadine plus alcohol, or all three antiseptics. CHG was not applied to patients under 2 months of age. The antimicrobials used for prophylactic administration were selected according to our institutional guideline and administered within 60 minutes prior to skin incision. We used the electronic medical record and the Anesthesia Information Management System (AIMS) to obtain intraoperative information such as medications and blood transfusion as well as American Society of Anesthesiology (ASA) physical status classification. Of the medications used perioperatively, the dosage in milligrams per kilogram of body weight was calculated for intravenous drugs, and the end tidal concentration per minute for inhaled (volatile) anesthetics or oxygen was summed up by the anesthesia duration (minutes). In this cohort, sevoflurane was largely used. Dose of sevoflurane was dichotomized into low-dose and high-dose at the median dose. In addition, to account for patients who received both sevoflurane and isoflurane, the total amount of individual volatile anesthetics was calculated by multiplying isoflurane dose by 1.5 to convert to its equipotent sevoflurane dose based on the minimal alveolar concentration (MAC) of these two anesthetics. Fentanyl and morphine were used as opioids, but the morphine dose was converted to an equipotent fentanyl dose using the formula of fentanyl(mcg/kg) = morphine (mcg/kg))/100 [23]. The incidence of SSIs in the patients had been all prospectively documented. We defined SSI and categories for surgical wound using the Centers for Disease Control and Prevention (CDC) definitions [24]. The SSI classification (superficial incisional, deep incisional, organ space) was also recorded. Discharge from wound or drain underwent bacterial cultures. The operative type was categorized into 2 groups: Laparoscopic or Open surgery. There were 23 operators in total.

Statistical analysis

Univariate analysis

We used chi-square tests to compare proportions for categorical variables (such as ASA class), and t-tests or Mann–Whitney U tests to compare averages or medians for continuous variables.

Propensity score analysis

We used 1:1 propensity score matching to compare the SSI outcomes between the low-dose and high-dose sevoflurane groups. We used a logistic regression model to calculate propensity scores (PS). The predictor variables were sex, age, weight, race, ASA class, surgical wound class, preoperative skin antiseptic, surgeon, antibiotic prophylaxis, and operative type (laparoscopic or open). Each patient in the lower sevoflurane dose group was matched with a patient in the higher dose group with the closest estimated PS within a caliper (≤0.2 of the pooled standard deviation of estimated logits) using the nearest-neighbor method without replacement. We calculated the C-statistic using the area under the receiver operating characteristic curve to evaluate the predictive ability of the model. We calculated standardized differences to examine the balance in the baseline variables of patients between the two study groups. We used t-tests to compare averages of continuous variables and the chi-squared test to compare proportions of categorical variables.

After PS matching, we performed multivariable logistic regression analyses to adjust for additional intraoperative factors, specifically anesthesia duration and blood transfusion. All hypothesis tests had a two-sided significance level of 0.05. All statistical analyses were conducted using Stata/MP 15.0 (StataCorp; College Station, TX, USA). P< 0.05 was considered statistically significant.

Sensitivity analysis

To test the robustness of the results of PS matching analysis by accounting for the possibility of unmeasured confounders, we performed PS adjustment and stabilized inverse probability of treatment weighting (IPTW). For PS adjustment, PS themselves were used as a covariate in the regression model, where the outcome of SSI was regressed against the two sevoflurane group assignments and the estimated PS. IPTW is calculated from PS and allows construction of a pseudo-data set adjusted for measured confounders to estimate the average treatment effect over the marginal distribution of measured confounders in the entire cohort. In both analyses, we adjusted for the duration of anesthesia and presence or absence of blood transfusion.

Results

In total, 630 patients were enrolled and 621 were analyzed, excluding 9 patients who underwent endoscopic resection. Of 621 cases, 39 (6.3%) met criteria for SSI. In univariable analyses, gender and high-dose sevoflurane were associated with SSIs (Table 1). Opioid dose was not different between the groups. Distribution of surgical procedures by site is shown in Supplemental Table 1. Surgical site was not associated with the incidence of SSI (Table 1). Intraoperative temperature did not differ between patients with and without SSI.

Table1.

Comparison of patient characteristics between patients with and without surgical site infection (SSI) using univariate analysis

	Patient without SSI (N=582)		Patient with SSI (N=39)		P value
	n	%	n	%
Sex Female	351	60.3	16	41.0	0.02
Age (Month), mean (SD)	106.2	(98.2)	105.3	(93.2)	0.65
Weight (kg), mean (SD)	31.3	(27.6)	27.3	(20.6)	0.95
BMI (kg/m2), mean (SD)	19.71	(5.8)	18.55	(3.9)	0.37
Race					0.94
White	351	60.3	26	66.7
Black or African American	43	7.4	2	5.1
Asian	22	3.8	9	23.1
Others	166	28.5	10	25.6
ASA					0.48
1	16	2.7	1	2.6
2	250	43.0	14	35.9
3	242	41.6	15	38.5
4	66	11.3	8	20.5
5	8	1.4	1	2.6
Wound Class					0.29
1	170	29.2	17	43.6
2	408	70.1	22	56.4
3	2	0.3	0	0.0
4	2	0.3	0	0.0
Skin preparation agent					0.47
CHG-alcohol (Chloraprep TM)	149	25.6	9	23.1
Betadine	339	58.2	28	71.8
Betadine and alcohol	32	5.5	0	0.0
All	9	1.5	0	0.0
None	53	9.1	2	5.1
Procedure type					0.06
Laparoscopic	171	29.4	6	15.4
Open	411	70.6	33	84.6
Surgical site					0.754
Upper GI	55	91.7	5	8.3
Lower GI	506	93.9	33	6.1
Others	21	95.5	1	4.6
Prophylactic antibiotics	544	93.5	38	97.4	0.50
Propofol (mg/kg), median (IQR)	2.6	(0, 4.7)	2.9	(0.88, 5.98)	0.45
ET sevoflurane, median (IQR)	266.2	(90.3, 456.5)	391.0	(211.2, 538.2)	0.026
ET isoflurane, median (IQR)	0.0	(0, 45.8)	0.0	(0, 87.6)	0.97
*Opioids (mcg/kg), median (IQR)	3.7	(2.1, 6.9)	4.3	(2.6, 6.74)	0.58
Duration of surgery(min), median (IQR)	166	(107, 265)	199	(155, 236)	0.12
Average temperature (°C) , median (IQR)	36.3	(35.8, 36.8)	36.2	(35.9, 36.7)	0.87
Minimum temperature, median (IQR)	35.3	(34.7, 35.9)	35.4	(34.6, 36.2)	0.69

SD, standard deviation; ASA, American Society of Anesthesiologists physical status classification; CHG, chlorhexidine gluconate; GI, gastrointestinal; IQR, interquartile range, ET, end tidal

^*Total amount of fentanyl and morphine converted to fentanyl using the formula of fentanyl (mcg/kg) = morphine (mcg/kg)/100 [Lee et al. Pediatr Anaesth 2016;26: 644].

Table 2 shows the patient characteristics in the two sevoflurane dose groups (high-dose group, low-dose group) before and after PS matching. The cut-off value for the two groups, or the median sevoflurane dose, was 272.5 (%*min) (Supplemental Table 2). It has been shown that the MAC of volatile anesthetics varies with age [25]. The MAC of sevoflurane was reported to be 2.6% at 25 years of age and 2.1% at 40 years of age [26]. The MAC of newborns and infants under 6 months was 3.2%, and the MAC of older children up to 12 years of age was only slightly lower, 2.5% [27]. An increase in age of 10 years results in a 7.2% decrease in MAC [28]. We combined these reports with the present results to obtain the age-corrected ET_sevo (% × min). The median age-corrected sevoflurane dose was 205.5 (% × min) (Supplemental Table 2). After PS matching, 204 pairs were analyzed. Preoperative confounders used to predict PS were appropriately assigned with less than 10% IPTW. Isoflurane doses were lower in the high sevoflurane dose group. Propofol doses varied slightly after PS matching (SD 15%). Opioid dose was higher in the low sevoflurane dose group. Operators were also equally distributed among the two sevoflurane dose groups.

Table2.

Characteristics of the patients in the two sevoflurane dose groups

	Before matching					Propensity score-matched patients
	Lower dose Sevoflurane (N=311)		Higher dose Sevoflurane (N=310)		SD (%)	Lower dose Sevoflurane (N=204)		Higher dose Sevoflurane (N=204)		SD (%)
	n	%	n	%		n	%	n	%
Male sex	128	41.2	126	40.6	1.0	84	41.2	83	40.7	1.0
Age (Montd), mean (SD)	100.41	(98.8)	111.92	(96.6)	11.8	110.6	(93.5)	110.9	(97.8)	−0.3
Weight (kg), mean (SD)	28.58	(26.4)	33.57	(27.8)	−18.4	32.6	(26.9)	31.8	(26.4)	2.8
Race
White	190	61.1	187	60.3	1.6	130	63.7	130	63.7	0.0
Black or African American	22	7.1	23	7.4	−1.3	10	4.9	12	5.9	−4.3
Asian	14	4.5	9	2.9	8.5	10	4.9	6	2.9	10.1
Others	85	27.3	91	29.4	−4.5	54	26.5	56	27.5	−2.2
ASA
1	9	2.9	8	2.6	1.9	4	2.0	7	3.4	−9.1
2	109	35.0	155	50.0	−30.6	92	45.1	87	42.6	4.9
3	124	39.9	133	42.9	−6.2	91	44.6	96	47.1	−4.9
4	60	19.3	14	4.5	46.9	14	6.9	14	6.9	0.0
5	9	2.9	0	0.0	24.4	3	1.5	0	0.0	17.3
Wound Class
1	108	34.7	79	25.5	20.3	53	26.0	56	27.5	−3.3
2	199	64.0	231	74.5	−23.0	150	73.5	148	72.5	2.2
3	2	0.6	0	0.0	11.4	1	0.5	0	0.0	9.9
4	2	0.6	0	0.0	11.4	0	0.0	0	0.0	-
Skin preparation agent
CHG-alcohol (Chloraprep TM)	68	21.9	90	29.0	−16.5	57	27.9	65	31.9	−8.6
Betadine	192	61.7	175	56.5	10.8	125	61.3	116	56.9	9.0
Betadine and alcohol	7	2.3	25	8.1	−26.5	6	2.9	6	2.9	0.0
All	5	1.6	4	1.3	2.7	2	1.0	4	2.0	−8.2
None	39	12.5	16	5.2	26.2	14	6.9	13	6.4	2.0
Procedure type
Laparoscopic	73	23.5	104	33.5	−22.5	61	29.9	56	27.5	5.4
Open	238	76.5	206	66.5	22.5	143	70.1	148	72.5	−5.4
Intraoperative factor
Patient with blood transfusion	17	5.5	8	2.6	14.7	4	2.0	6	2.9	−6.3
Regional anesthesia	262	84.2	246	79.4	12.7	163	79.9	168	82.4	−6.3
Propofol (mg/kg), mean (SD)	4.5	(7.7)	4.4	(5.5)	2.2	5.2	(8.2)	4.2	(5.2)	15.0
ET isoflurane, mean (SD)	108.5	(168.9)	18.9	(74.1)	68.7	128.4	(164.2)	15.0	(65.3)	90.8
ET O2, mean (SD)	13052.8	(8649.3)	15658.5	(7481.4)	−32.2	13613.1	(8195.9)	15381.3	(7521.8)	−22.5
Opioids (mcg/kg), mean (SD)	7.6	(9.2)	5.2	(6.8)	29.7	6.6	(7.9)	5.2	(6.9)	18.9
Duration of anesthesia, mean (SD)	269	(162.8)	333	(146.8)	−41.7	286.2598	(156.9)	324.3824	(142.9)	−25.4

SD, standard deviation; ASA, American Society of Anesthesiologists physical status classification; CHG, chlorhexidine gluconate; ET, end tidal

Table 3 shows the percentage of SSI rates in the two groups and the results of the χ² tests. The SSI rate was 3.9% in the low-dose group and 9.8% in the high-dose group (P-value 0.019).

Table 3.

Comparisons of outcomes between higher dose and lower dose sevoflurane groups in all patients and propensity score-matched patients

	All patients					Propensity score-matched patients
	Lower dose Sevoflurane (n = 311)		Higher dose Sevoflurane (n = 310)			Lower dose Sevoflurane (n = 204)		Higher dose Sevoflurane (n = 204)
	n	%	n	%	P-value	n	%	n	%	P-value
Surgical site infection	12	3.9	27	8.7	0.01	8	3.9	20	9.8	0.02

After adjustment for blood transfusion and the duration of anesthesia, which are the major intraoperative risk factors for SSIs (Table 4), patients receiving the high-dose sevoflurane had a significantly higher odds of SSI (OR 2.58, 95% confidence interval [CI]: 1.11–6.04; P-value =0.028). To further clarify the role of volatile anesthetics, we also adjusted for the opioid dose. The adjusted odds of SSI remained higher in the high dose sevoflurane group (OR=2.51, 95% CI: 1.07–5.92; P-value = 0.035).

Table 4.

Multivariable logistic regression analysis for surgical site infection in propensity score-matched patients

	Surgical site infection rates
	Odds Ratio	95 % CI	P value
Sevoflurane
Lower dose	Ref
Higher dose	2.58	1.11–6.04	0.03
Duration of anesthesia (min)	1.00	0.998–1.00	0.54
Blood transfusion
No	Ref
Yes	1.26	0.15–10.78	0.83

CI, confidence interval

Results of the sensitivity analysis for these PS-matched analyses are presented in Table 5. Both PS adjustment and IPTW were adjusted for blood transfusion and duration of anesthesia, and the OR for the high-dose sevoflurane group were 3.097 (95% CI:1.41–6.81) and 3.47 (95% CI: 1.65–7.29), p-values 0.005 and 0.001, respectively.

Table 5.

Propensity score adjustment and stabilized inverse probability of treatment weighting analyses for SSIs

	Surgical site infection rates
	PS adjustment			stabilized IPTW
	Odds ratio	95% CI	P-value	Odds ratio	95% CI	P-value
Higher-dose Sevoflurane	3.097	1.41–6.81	0.005	3.47	1.65–7.29	0.001

PS, propensity score; IPTW, inverse probability of treatment weighting; CI, confidence interval

When analyzed using the two groups defined by total VAs (both sevoflurane and isoflurane), which were dichotomized at the median doses. 188 pairs were selected through 1:1 PS matching. The incidence of SSIs was significantly higher in the higher-dose group (9.57% vs. 3.72%, p=0.023) and was similar in sensitivity analyses in the two groups (OR:2.86 [95% CI: 1.20–6.79]; p = 0.017 for PS adjustment, OR: 3.04 [95% CI:1.26–7.37]; p = 0.014 for IPTW). However, after further adjustment for anesthesia time and blood transfusion, there was no significant difference in SSI rate between the two total VA dose groups after PS matching (OR 2.11 [95% CI 0.77–5.83]); p = 0.15).

Figure 1 shows the breakdown of the bacterial culture results by SSI type: 22 specimens were positive among patients with superficial incisional (SI) SSI, and 20 were positive for the combination of deep incisional (DI) and organ/space SSI. More Staphylococci were detected in the SI-SSI, and a greater proportion of Escherichia coli and Candida were detected in the DI and organ/space SSI. Enterococci were detected most frequently in the SSIs at both superficial and deep SSIs.

Figure 1. Result of bacterial cultures sent from pediatric patients with SSI after intestinal Surgery.

Bacterial culture results of discharge from wound or drain. Because all detected organisms were tabulated, cases with suspected multiple infections were included. The number of Enterococcus is the total of detections of Enterococcus faecalis and Enterococcus faecium. All Staphylococcus species were Staphylococcus aureus except for one case of Staphylococcus epidermidis in SI.

Abbreviations SI, Superficial incisional; DI, Deep incisional; E. coli, Escherichia coli

Discussion

In this study, we investigated the incidence of SSIs in children after elective intestinal surgery among groups dichotomized by sevoflurane dose during anesthesia. The results demonstrated that the incidence of SSIs was significantly higher in the group that received higher doses of sevoflurane for the maintenance of anesthesia.

Few clinical reports have compared the effects of volatile and intravenous anesthetics on SSIs. Two studies examined the role of anesthesia in SSIs after colon surgery by randomizing into volatile anesthetic and intravenous anesthetic-based groups. Koo et al. showed that the use of volatile anesthetics was associated with higher incidence of SSIs than intravenous anesthetics [20]. In contrast, Shimizu et al showed that the use of volatile anesthetics was associated with lower incidence of SSIs [29]. The difference between the two studies may be influenced by differences in SSI rates and sample sizes. For example, the latter study reported a very high rate of SSIs with a much smaller sample size compared with the former study. Our study was retrospective and different from the previous two studies that prospectively randomized into two anesthetic groups.

In pediatric patients, there are fewer apparent preoperative confounders, such as diabetes, smoking, and abnormal body mass index (BMI) compared with adult patients. In this respect, a pediatric population allows evaluation of the effect of anesthetics on SSIs with less confounding factors, and the result might be applicable to the intraoperative management of adults as well. We have adjusted the PS to take into account the reports of fewer risk of SSIs in laparoscopic surgery than open surgery [30, 31].

Among the pre- and intra-operative factors we studied, intraoperative blood transfusion [13, 32] and operative duration [12] are well-recognized risk factors for SSIs. We used the total dose of sevoflurane, and it is intuitive that the sevoflurane amount will increase as the duration of surgery increases. Therefore, we adjusted for blood transfusion and duration of anesthesia. The SSI rate remained still higher in the high-dose sevoflurane group after this adjustment, and two different sensitivity analyses confirmed that this result was robust.

Several effects of VAs on the immune system have been reported so far [33–37]. Previously, we showed in murine models of SSI that longer-exposure to VA was associated with higher bacterial loads than to propofol, suggesting that selection of anesthetic drugs could affect the course of infection [21]. Bacterial loads following SSIs were significantly higher in mice receiving a long (6-hour), but not a short (2-hour) VA exposure. Propofol exposure did not affect bacterial loads. Also, we found that VA exposure attenuated the recruitment and phagocytic capacity of neutrophils in murine experimental abdominal sepsis and skin inflammation models [22]. Neutrophils are initial responders in surgical procedures during which anesthesia is provided, and their adequate function is critical to control SSIs. These findings seem to substantiate the findings of the present study.

Of the VAs, sevoflurane accounted for the majority of management in this cohort, although it was used in conjunction with isoflurane in some patients, when the dose of sevoflurane was relatively low as reflected in the variability after PS matching. Thus, we also converted isoflurane dose to its equivalent sevoflurane dose based on the MAC. The incidence of SSIs was assessed in the high-dose and low-dose VA groups. The incidence of SSIs was significantly higher in the high-dose VA group after PS matching and in the sensitivity analysis, but there was no significant difference when adjusted for the blood transfusion and anesthesia duration. It is unclear if there is any difference between isoflurane and sevoflurane with respect to SSI risk. Isoflurane and sevoflurane are structurally similar and derived from ether as their prototype. However, there is a growing literature suggesting that their targets are overlapped but not completely the same. We previously showed that VAs targeted a number of non-canonical molecules in immune cells including Toll-like receptor (TLR)2, TLR4, Ras-related protein 1 (Rap1) and β2 integrins [38]. Only sevoflurane interacted with TLR2. The attenuation of TLR1/TLR2 activation was seen under sevoflurane exposure but not under isoflurane exposure [39]. Similarly, only sevoflurane blocked Rap1. In this study, we were not able to evaluate the independent effect of isoflurane, but the differences in the mechanism of action of the two drugs should be considered. When patients were dichotomized based on the total VA dose, fewer pairs were available for analysis after after PS matching. In parallel, the number of patients with SSIs in low dose and high dose VA groups was only 8 and 17, respectively. A logistic regression analysis adjusted for three variables with this sample size may not be a good fit.

Regarding results of bacterial culture, a higher proportion of Staphylococci was detected in superficial SSI than in deep incisional and organ /space SSI, and more Escherichia coli, Pseudomonas aeruginosa, and Candida were detected in deeper samples than in superficial SSI. Lake et al. previously reported the distribution of pathogens associated with pediatric SSIs using the National Healthcare Safety Network database [40]. Although the study did not show result by SSI type, Escherichia coli has been shown to be the most common SSI-causing bacteria after abdominal surgery. Compared to the report by Lake et al, the percentage of Escherichia coli of the total detected bacteria was slightly lower in our series, while the percentage of Enterococcus and Pseudomonas aeruginosa, Staphylococci and Candida was higher. It is difficult to compare our result with those of multi institutional databases with different target surgeries and facility types, so further studies are needed.

We should note the following limitations. First, in children, it is possible that some of the underlying diseases are congenital, which may affect immune functions. In adults, the rate of SSIs is generally higher when there is a background of infectious or inflammatory diseases [41]. Second, we did not examine the use of drains, or the type of sutures and fluid balance. Although excessive volume may be associated with serosal edema and potential problems with bowel healing and motility, because of concerns about the accuracy of in-out fluid balance documentation, we did not evaluate fluid balance as a predictor in this study. Third, PS analyses can be biased owing to unmeasured confounders. Fourth, this was not a randomized controlled study to evaluate the role of anesthetics. A randomized trial of intravenous anesthetics versus VA would be necessary to confirm our findings.

Conclusions

We have shown that higher doses of sevoflurane were associated with higher rate of SSIs. Anesthesia is an important procedure in surgery, and its association with SSIs may change the view of perioperative management. Future randomized controlled studies are needed to validate our findings.

Abbreviations:

SSI

Surgical site infection

GI

Gastrointestinal

WHO

World Health Organization

VA

Volatile anesthetic

CHG

Chlorhexidine gluconate

AIMS

Anesthesia Information Management System

ASA

American Society of Anesthesiology

CDC

Centers for Disease Control and Prevention

IPTW

Inverse probability of treatment weighting

OR

Odds ratio

CI

confidence interval

PS

Propensity matching

SI

Superficial incisional

DI

Deep incisional

BMI

Body mass index

TLR

Toll-like receptor

Rap1

Ras-related protein 1

MAC

Maximum alveolar concentration