Incidence and predictors of surgical site infection following cesarean section in North-west Ethiopia: a prospective cohort study

A prospective cohort study was conducted among pregnant women who had a caesarean section between March 28, 2019, and August 31, 2019, at Debre Markos Referral Hospital (DMRH). DMRH located, in the East Goiiam zone, Amhara regional state of Ethiopia. It is one of the centers where pregnant mothers can receive CS services. It provides services with three units (labor, maternity, and gynecology). It also has outpatient clinics (antenatal care, family planning). All pregnant women admitted to the DMRH labor ward who underwent an elective or emergency CS were eligible for enrolment within 24-h post-CS and followed for 30 days to detect SSI, in accordance with the Centers for Disease Control and Prevention (CDC) Classification [6]. Pregnant women who were died during the surgical procedure or immediately after the procedure or surgical procedures performed outside DMRH were excluded from the study.

The sample size was computed using R statistical software Version 3.6.10 based on the sample size determination formula for survival analysis. For the first objective, the required sample size was determined by considering the following assumptions based on the single population proportion. Sample size (n) is calculated by taking the proportion (P) 10.9% obtained from similar published articles in Tanzania [24], 95% confidence level, 5% margin of error (d) and used a single population proportion formula: \( n=\frac{{\left( Z\alpha {/}_2\right)}^2\Big(p\left(1-p\right)}{d^2} \). This calculation yielded a N₁ = 150.

For the second objective, the sample size was calculated using Stata™ Version 14.1 by considering two-tailed significant level (α) of 0.05 and power 80%, and a 95% confidence interval

Where e = required number of the event, p = expected cumulative incidence of SSI following CS based on the previous studies, N = total sample size, 5% non-response rate. This calculation yielded a sample size of 513 participants. Accordingly, it is recommended to use the largest sample size obtained out of all calculations in order to accommodate all study objectives. Therefore, 513 individuals were required for the study. Patients were serially enrolled until the desired sample size was reached.

The dependent variable of this study was the incidence of SSIs following CS. The independent variables were socio-demographic variables like maternal age, marital status, residence, maternal educational status, occupational status, religion, and antenatal care. Relevant maternal medical history like diabetes mellitus, renal disease, anemia, HIV, bronchial asthma, previous history of CS, and/or hypertensive disorder were captured. Surgical intervention related variables were considered such as types of CS (elective or emergency), types of incision (vertical, horizontal), types of skin suturing (interrupted, subcuticular), premature rupture of membranes, number of peravaginal examination, blood loss, duration of procedures, anesthetic techniques (general, epidural), indication for CS, gestational age, blood transfusion, and/or antibiotic used.

Data were collected using a structured questionnaire that was recorded at each follow-up visit. Training on the objectives of the study and how to diagnose SSI clinically was given for data collectors and supervisors for 2 days. In addition to the patient interviews, baseline information was also retrieved from a variety of sources: medical records, antenatal cards, surgical notes, structured interviews, and clinical examinations [33]. On the first day after CS during the inpatient stay, all data related to the surgical procedure and post-surgical management were mined from the surgical notes. Demographic and clinical data were obtained from antenatal cards, medical records, structured interviews, and clinical examinations. The ward health care provider inspected the wounds daily and checked the patients for any signs of SSI during the inpatient stay. The women were followed for 30 days to detect SSI in accordance with the Centers for Disease Control and Prevention (CDC) Classification. Patients who were missed the appointed scheduled were contacted by telephone in order to determine whether they had SSI or not. A patient was considered lost to follow-up after five unsuccessful attempts to reach them by telephone during the follow-up period.

Patients were considered to have SSI following CS if they met the following definition: Involving skin and subcutaneous tissue at surgical site with any one of the following: purulent discharge or organisms isolated from fluid/tissues of superficial incision or at least one sign of inflammation (pain, fever, localized swelling, induration, dehiscence, overlying skin changes and exudative purulent discharge) or wound deliberately opened by the surgeon for drainage or surgeon declares that the wound is infected [6].

Data were cleaned, coded, and entered into EpiData™ Entry software Version 4.1 and exported into R version 3.61 for analysis. A descriptive analysis of categorical variables was performed through frequency tables, and interquartile ranges and medians were computed for continuous variables. The time for the development of SSI was estimated using the Kaplan-Meier (KM) method. The log-rank test was utilized to compare the estimated survival curve of patients based on categorical variables. The proportional hazard assumption (PHA) was checked using graphs and scaled Schoenfeld residuals tests. Both the graph and figure suggested that there is no evidence of non-proportional hazards for the remaining covariates (S1 Table and S2 Fig). Variables with p-values less than 0.2 at bi-variable analysis were screened for multi-variable Cox regression model. The goodness of fit was checked by the Cox Snell residual test. The final results were taken as significant at P < 0.05. Adjusted Hazard Ratio (AHR) with 95% Confidence Interval (CI) were used to report strength of the association between outcome variable and predictors.

About 520 pregnant women were undertook CS delivery between March 28, 2019 and August 31, 2019. All of these 520 pregnant women were serially enrolled and followed for 30 days to detect the development of SSI. Of these 520 CS, 422 (81.16%) were categorized as emergency CS procedures. A majority of the respondents,488 (93.8%) had ante-natal care (ANC) with a median of 4 visits (IQR:3–4) (Table 1). The age of the study subjects was ranged from 18 to 49 years with a mean age of 27.4 ± 4.8 years. The gestational age at caesarean delivery ranged between 30 and 44 weeks and a total of 485 (93.3%) neonates were delivered at term gestational age with the remaining cases delivered as 14 (2.7%) post-term and 21(4%) pre-term neonates (Table 2).

All pregnant women had received pre-operative antibiotic prophylaxis before undergoing CS. Ceftriaxone is the most prescribed antibiotic prophylaxis (91.9%) followed by ampicillin (2.7%).

In this study, 183 (35.2%) pregnant women underwent CS with an indication of fetal distress (Table 3). The mean gestational age of the study participants who were developed SSI was 39.6 weeks (SD ± 4.62). The median follow-up time is different among women with SSI and without (8 days’ vs 29 days) (Table 4).

During the study period, 132 (25.4%) patients had developed SSI with an incidence rate of 11.7 per 1000 (95% CI: 9.8, 13.9) persons/days. Among patients who had develop SSI, about 111(84%) had emergency CS procedure. The rate of SSI was 37/132 (28%) for patients who had rupture of membrane prior to CS procedure. Majority of study participants who had horizontal skin incision experienced SSI (124/132; 93.9%). Patients with SSI had longer hospital stays than those without a SSI (12.7 ± 6.9 days’ vs 4 ± 1.7 days). Due to the high prevalence of censoring in this cohort the median survival time could not be estimated. As indicated in Fig. 1, the incidence of SSI following CS is not statistically significant difference between the urban and rural residents. The survival time was different based on the surgical procedures performed (elective or emergency) (Fig. 2).

The multivariable Cox analysis documented that women who are not able to read and write were 1.32 times more likely to develop SSI compared to their literate counterparts (AHR = 1.30, 95% CI, 1.19, 2.11). The incidence of SSI among women who did not have antenatal follow up during pregnancy is 2-folds higher risk relative to those who had antenatal follow up (AHR = 2.16, 95% CI: 1.05, 4.53). Women who had previous history of CS were 21% more risk for SSI (AHR = 1.21, 95%CI: 1.11, 2.31). In our study, the incidence of SSI following CS was significantly different between HIV positive and HIV negative women. The incidence of SSI among HIV positive women is 39% higher compared to HIV negative women (AHR = 1.39,95% CI: 1.21, 2.57). The details of bivariable and multivariable output is presented in Table 5.

Our study has limitations. Some of the cases during the study period were not followed up due lost to follow up which could likely influence the calculated rate of surgical site infection. Further the study only occurred in a single hospital which may have a fairly homogenous catchment of women. Additionally, the DMRH is a centrally based hospital receiving clients what are likely at a higher risk than the population at large, which limits the extrapolation of the findings beyond similar contexts. In addition, microbiological isolation was not conducted in this study.