Identification of predictive factors for surgical site infections in gastrointestinal surgeries: A retrospective cross-sectional study in a resource-limited setting

Introduction

A surgical site infection (SSI) is a frequently encountered nosocomial infection that typically develops within 30 days of surgery. In cases where an implant is used, the timeframe for SSI occurrence can extend up to one year.¹ The estimated incidence of SSI is 0.5% to 3% worldwide, with a higher incidence reported in low-income countries, where SSI is estimated to be the most common healthcare-associated infection.^2,3 In addition to the socioeconomic status, surgeries that involve the gastrointestinal (GI) appear to have a higher SSI incidence, with reports indicating a 12%-30% incidence rate of such cases. The associated expenditure of increased hospitalization (7-11 folds), mortality, and morbidity (2-11 folds) force a higher emphasis on detecting such patients earlier in the course of their illness and identifying patients with a higher risk of developing SSI to improve the quality of care and minimize the cost.^4,5

Various factors have been studied concerning SSI, which can extend from socioeconomic status to the preoperative settings and surgical approaches.⁶ Certain non-modifiable risk factors include age, gender, immunosuppression, diabetes mellitus, obesity, or active smoking. Additionally, the pre-operative preparation, operative time, and intra-operative techniques may impact the development of SSI, which is seen at a higher rate in emergent and septic surgeries.^3–5 SSI can be attributed to microorganisms that are derived from the patient’s skin flora or from the surrounding environment.¹ In either scenario, the adherence of microorganisms to the surgical instruments can contaminate the incision. Contaminated surgical procedures pose an increased risk, particularly when multidrug-resistant microorganisms are involved.⁶

Previous monocentric and retrospective studies in Yemen reported SSI rates of 2.2% and 31.7%.^7,8 However, there is limited information available about the extent of SSI and its predictive factors in low-income countries, such as Yemen.^7,8 This study aimed to investigate the SSI rate and its predictive factors among Yemeni patients who underwent GI surgeries in a resource-limited setting.

Methods

Results

Multivariate logistic regression revealed the following independent risk factors

Illiteracy (OR:40.31; 95% CI: 9.54-170.26), current smoking (OR: 21.15; 95% CI: 4.63-96.67), diabetes (OR: 5.07; 95% CI: 2.27-11.35), leukocytosis (OR: 2.62; 95% CI: 1.24-5.53), hypoalbuminemia (OR: 3.70; 95% CI: 1.35-10.16), contaminated and dirty wounds (OR: 6.51; 95% CI: 1.62-26.09), longer operative duration (OR: 1.02; 95% CI: 1.01-1.03), emergency operations (OR: 12.58; 95% CI: 2.91-54.30), and administering antibiotics before 1 h of operation (OR: 3.01; 95% CI: 1.28-7.10) were independent factors for SSI (all p-value<0.05, Table 5). The prediction model’s total ROC curve was 0.946, which was much higher than the NNIS score (0.660) (Figure 1).

Table 5. Multivariate analysis of risk factors associated with surgical site infection.

Predictor	Estimate	SE	Z	P value	OR	95%CI
						Lower	Upper
Education level	3.6966	0.73508	5.0288	<0.001	40.31	9.54351	170.26
Khat chowing	0.6876	2.97657	0.2310	0.817	1.99	0.00582	679.66
Smoking	3.0518	0.77527	3.9365	<0.001	21.15	4.62903	96.67
Hypertension	-0.8014	0.84596	-0.9473	0.343	0.45	0.08548	2.36
Diabetes	1.6240	0.41073	3.9539	<0.001	5.07	2.26816	11.35
Shaving time	0.5350	0.41789	1.2802	0.200	1.70	0.75272	3.87
Temperature	0.2583	0.39358	0.6563	0.512	1.29	0.59864	2.80
WBC (109/L)	0.9643	0.38089	2.5316	0.011	2.62	1.24327	5.53
Hospital stays	0.2963	0.47340	0.6260	0.531	1.34	0.53180	3.40
Albumin	1.3094	0.51494	2.5429	0.011	3.70	1.35007	10.16
Wound class	1.8735	0.70827	2.6451	0.008	6.51	1.62461	26.09
Blood loss	0.6588	2.99194	0.2202	0.826	1.93	0.00549	680.54
Operative time	0.0214	0.00644	3.3313	<0.001	1.02	1.00886	1.03
ASA	-0.2291	0.58612	-0.3909	0.696	0.79	0.25212	2.51
NNIS score	-0.0705	0.85520	-0.0825	0.934	0.93	0.17435	4.98
Operative type	2.5323	0.74601	3.3944	<0.001	12.58	2.91575	54.30
Antibiotic time	1.1032	0.43687	2.5253	0.012	3.01	1.28013	7.10
Accuracy: 0.905; Specificity: 0.966; Sensitivity: 0.593; AUC: 0.946. Nagelkerke R square: 0.648 Significance of the model <0.001

Figure 1. The receiver operating characteristic curve of the prediction model compared with the National Nosocomial Infections Surveillance risk index in the validation cohort.

Abbreviations: AUROC: Area under the receiver operating characteristic curve; NNIS: National Nosocomial Infections Surveillance.

Discussion

The improved access to healthcare, increased population age, and increased complexity of surgical interventions and patients’ conditions shed light on the importance of managing post-operative complications. Despite the precautions and the hygienic approach implemented to limit the incidence of SSI, it still represents one of the most common post-operative complications. Such infections result in an increased healthcare expenditure, and worsened mortality and morbidity.¹⁰ This predicament can be especially disadvantageous for low-income nations, where providing healthcare is already a daunting task due to constrained resources, indigent communities, and elevated levels of antimicrobial resistance.¹¹

Among the 525 enrolled patients, the incidence of SSI within 30 days after surgery was 16.4%, which is in line with previous reports from developing countries, such as Saudi Arabia, with a rate of 16.3%.¹⁴ However, earlier studies showed much higher rates of SSI affecting up to one-third of the patients in Yemen.⁸ In contrast, more recent reports from Yemen have demonstrated a lower incidence of SSI, with a rate of 12.7% among patients who underwent gastrointestinal procedures.⁸ Our findings, which showed a slightly higher rate of SSI, could be partially attributed to the larger number of complicated cases or complex oncological procedures performed at our tertiary teaching hospital.

Several studies have been conducted to evaluate the link between putative risk variables and SSI in GI surgical operations. However, there is a large range of variation in the variables analyzed and the proportional effect of these factors on individual outcomes. To address this issue, we comprehensively studied the preoperative and operational risk variables in GI operations associated with the development of postoperative SSI. Hamza et al. and Lakoh et al. carried out similar investigations.^6,12 This study found that illiteracy, current smoking status, DM, leukocytosis, hypoalbuminemia, contaminated and dirty wounds, longer operative time, emergency operations, and longer time between administering antibiotics and operation were predictors for the development of SSI. Most of the potential predictive factors included have been previously reported as risk factors in other studies with a variety of reports and different levels.^6,12

The relationship between age and SSI risk is complex and not well understood. While some studies have reported an increased rate of infection in older patients, others have observed a favorable trend with increasing age. For instance, Kaye et al. demonstrated a 1.2% decrease in SSI risk for each additional year after 65 years of age.¹³ Nevertheless, these findings were demonstrated consistently, as a higher rate of SSI was observed in the older population.^14,15 Typically, with increasing age, there is an accumulated risk of developing comorbidities and immune dysfunction, which may lead to an increased likelihood of SSI. However, our study’s findings revealed no association between age and the development of SSI. This divergence may be attributed to variations in age categorization, as the majority of patients (67%) in this study were younger than 65 years.

The present study reveals a significant association between the level of literacy and the incidence of SSI. Specifically, illiterate patients were 40 times more susceptible to SSIs than educated patients. These results are in accordance with previous research conducted by Mezemir et al. and Baker et al.^16,17 Notably, a high prevalence of limited health literacy among adults in our country may adversely impact health outcomes. For example, patients with limited health literacy may experience difficulty in comprehending complex health information, may exhibit non-compliance with postoperative instructions, and may not adequately prepare for surgery. These factors may increase the risk of SSIs and other adverse outcomes, highlighting the potential health inequality in providing care and education for illiterate patients. Therefore, it is critical to improve health literacy among patients, particularly those with limited education, to potentially reduce the incidence of SSIs and enhance surgical outcomes.

This study did not find a significant association between unmodifiable risk factors, such as gender, BMI, residency, number of comorbidities (hypertension, history of malignancy, CRF, liver and lung diseases), perioperative blood transfusion, and SSIs in multivariate analysis. Although these social determinants are important factors that may contribute to patient outcomes, there is a lack of consensus on their association with SSI occurrence in the literature. For example, Deng et al. found that male sex and a greater number of comorbidities were associated with SSI occurrence.¹⁸ Additionally, Li et al. reported that ascites, bleeding diathesis, history of lung disease, radiotherapy, chemotherapy, chronic steroid use, and weight loss were associated with SSI occurrence.¹⁹ In contrast, Mezemir et al. did not find an association between gender, BMI, and SSI occurrence, which was similar to our study.¹⁶ These discrepancies may be attributed to sample size and demographics variation across studies, as well as variations in the documentation and management of patient comorbidities. The use of more objective measures, such as preoperative laboratory and radiologic values, may provide a better understanding of the association between comorbidities and SSI occurrence. Distinctly, in this study, DM and hypoalbuminemia had 5- and 3.7 times higher chances of developing SSIs, respectively. This association was observed in prior studies,^11,20 as hyperglycemia has been shown to impair WBC functions, leading to decreased immunity.²¹ On the other hand, reduced serum albumin levels are often associated with malnutrition or chronic wasting diseases.¹¹

Our study revealed that smoking was strongly associated with a 21-fold increased risk of developing SSIs compared to non-smokers. The vasoconstrictive and toxic effects of smoking are known to impede tissue oxygen delivery and hinder the healing process, thus contributing to the development of SSIs. These findings align with previous reports by Mawalla et al. and Billoro et al.^22,23

Regarding Khat (Catha edulis) chewing, its role in SSI occurrence remains uncertain. Our study observed a 1.99-fold increase in SSI occurrence among Khat chewers, although this association was not statistically significant. Currently, there is a lack of published studies specifically investigating the relationship between SSI and Khat chewing. However, Misha et al. found no association between Khat chewing and SSI occurrence in their regression analysis.³ Nevertheless, Khat chewing has been linked to various gastric issues (e.g., intestinal obstruction, and gastritis).²⁴ Furthermore, long-term Khat consumption poses a risk of developing severe complications including hepatitis, hepatic fibrosis, and cirrhosis in advanced stages.²⁵ Future prospective and more inclusive studies are recommended to investigate this issue, particularly in our country where the traditional use of these plants is widespread.

The settings of operation can significantly impact the development of SSI. Prior research has suggested that the degree of intraoperative wound contamination is indicative of SSI occurrence.^26,27 We found that contaminated and dirty wounds were 6.51 times more likely to develop SSI, which is consistent with other studies.^26–28 Furthermore, our study revealed that the prevalence of SSI in large, pancreatic, and biliary surgeries was lower than in intestinal procedures; although this association was not statistically significant. Similarly, other studies have also reported a higher incidence of SSI in the intestinal tract.^11,29 However, our findings were inconsistent with the literature documenting pancreatic and biliary leaks as independent risk factors for SSI occurrence.²⁷ This discrepancy could be attributed to the low number of cases involving biliary and pancreatic procedures, with most of them undergoing simple operations. Therefore, further prospective studies with a larger number of cases are necessary to clarify this issue.

Our study also found that emergency operations were 12.58 times more likely to result in SSI, consistent with other studies.^6,11 In addition, leukocytosis was found to be a predictor for the development of SSI, which aligns with previous research.³⁰ Additionally, prolonged operative time was recognized as an independent factor for SSI development in other studies, as it increases the risk of infection due to extensive surgical procedures and incisions, prolonged anesthesia, blood loss, and weaning antimicrobial prophylaxis concentration.^6,31 Furthermore, administering antibiotics one hour before operation has been reported as a predictor for SSI in previous studies.^23,31 In this study, it was observed that longer operation durations and administration of antibiotics more than one hour before the operation increased the likelihood of SSI by 1.02 times and 3.01 times, respectively. These findings are consistent with previous studies.^23,31

In this study, we investigated the microorganisms responsible for SSIs and their susceptibility to commonly prescribed prophylactic antibiotics. We found that the most common organisms isolated from infected wounds were Gram-negative bacteria, with extended-spectrum β-lactamase-producing E. coli being the most prevalent. Mawalla et al. reported a different outcome compared to this finding, as their studies indicated a higher presence of Gram-positive bacteria, including Staphylococcus aureus.²² In contrast, studies have reported similar findings to ours, demonstrating a higher occurrence of Gram-negative bacteria in infected abdominal wounds.^26,32 Furthermore, our findings revealed a high prevalence of multi-resistant pathogens in relation to commonly prescribed prophylactic antibiotics, which may serve as an explanation for the elevated rate of deep SSI observed in our study. Hence, there is a need to consider appropriate prophylactic antibiotics, especially for high-risk patients.

The NNIS risk index is a widely recognized framework for assessing and predicting the likelihood of SSI.⁹ Within our study, two elements of the NNIS exhibited statistical significance (operative time and wound class). However, upon conducting multivariate analysis, the overall NNIS model did not yield statistical significance. Moreover, when comparing the predictive accuracy, our developed model outperformed the NNIS model. Acutely, the performance of the NNIS model in this study showed poor predictive performance for the SSI occurrence as determined by the ROC curve. These results align with previous findings reported by Zhang et al.¹¹

Conclusions

This study highlights significant predictors of SSI, including illiteracy, active smoking, DM, leukocytosis, hypoalbuminemia, contaminated and dirty wounds, longer operative time, emergency operations, and extending antibiotic prophylaxis duration. Escherichia coli was the most common pathogen and had a high rate of multidrug-resistant strains. Identifying these risk factors can help surgeons adopt appropriate measures to reduce SSI and improve the quality of surgical care, especially in a resource-limited setting with no obvious and strict policy for reducing SSI.