| Characteristic | Overall N = 2641 |
Emergency N = 1381 |
Elective N = 1261 |
p-value2 |
|---|---|---|---|---|
| Age, years | 31.0 (28.0–34.0) | 30.0 (27.0–33.0) | 32.5 (29.0–35.0) | <0.001 |
| Anaesthesia type | 0.7 | |||
| Spinal/SAB | 250 (97.7%) | 130 (97.0%) | 120 (98.4%) | |
| General | 6 (2.3%) | 4 (3.0%) | 2 (1.6%) | |
| Other | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| Unknown | 8 | 4 | 4 | |
| Payment category | 0.3 | |||
| Staff | 25 (9.5%) | 10 (7.2%) | 15 (11.9%) | |
| Corporate | 12 (4.5%) | 6 (4.3%) | 6 (4.8%) | |
| Student | 11 (4.2%) | 8 (5.8%) | 3 (2.4%) | |
| Private | 216 (81.8%) | 114 (82.6%) | 102 (81.0%) | |
| Length of stay, days | 4.0 (3.0–5.0) | 4.0 (3.0–5.0) | 4.0 (3.0–5.0) | 0.3 |
| Previous CS | 121 (45.8%) | 27 (19.6%) | 94 (74.6%) | <0.001 |
| Fibroid uterus | 11 (4.2%) | 8 (5.8%) | 3 (2.4%) | 0.2 |
| High EBL (≥500 mL) | 96 (36.4%) | 41 (29.7%) | 55 (43.7%) | 0.019 |
| Adhesions | 1 (0.4%) | 0 (0.0%) | 1 (0.8%) | 0.5 |
| Prolonged surgery (>60 min) | 100 (37.9%) | 51 (37.0%) | 49 (38.9%) | 0.7 |
| Subcuticular wound closure | 255 (96.6%) | 132 (95.7%) | 123 (97.6%) | 0.5 |
| 30-day SSI | 119 (45.1%) | 61 (44.2%) | 58 (46.0%) | 0.8 |
| 1 Median (Q1–Q3); n (%) | ||||
| 2 Wilcoxon rank sum test; Fisher’s exact test; Pearson’s Chi-squared test | ||||
Post-discharge Surgical Site Infection After Caesarean Section: A 30-Day Active Surveillance Cohort Study from a Ghanaian Primary Hospital
Abstract
Background: Surgical site infection (SSI) after caesarean section (CS) is a leading cause of maternal morbidity in sub-Saharan Africa. Routine surveillance confined to the inpatient period may substantially underestimate the true SSI burden.
Methods: We conducted a 30-day active surveillance cohort study of all CS procedures performed at KNUST University Health Service, Kumasi, Ghana, in 2024. SSI was defined as a diagnosis of surgical wound infection recorded in the facility’s clinical information system within 30 days of the operative date. The primary outcome was the proportion of SSIs detected after hospital discharge. Secondary analyses examined predictors of SSI using multivariable Firth-penalised logistic regression and time-to-SSI using Kaplan-Meier estimation with log-rank testing.
Results: Among 264 CS procedures (126 elective, 138 emergency), the 30-day SSI rate was 45.1% (95% CI 39.0–51.3%). Of 119 SSI cases, 118 (99.2%; 95% CI 95.4–100.0%) were detected after hospital discharge, with a median time to diagnosis of 7 days (IQR 5–10). Discharge-based surveillance would have captured only 0.8% of infections; extending follow-up to 14 days post-discharge captured 90.8% (95% CI 84.1–95.3%). All SSI-related return visits presented to General Service out-patient department. In multivariable analysis, younger maternal age was the only independent predictor of SSI (OR 0.92 per year; 95% CI 0.87–0.97; p = 0.004), an effect not explained by parity. No significant difference in SSI timing was observed between elective and emergency CS (log-rank p = 0.615).
Conclusion: Post-caesarean SSI is overwhelmingly a post-discharge phenomenon in this setting. A minimum 14-day post-discharge surveillance window is required to capture over 90% of cases. These findings support integration of structured wound surveillance into routine out-patient department services.
Keywords
surgical site infection, caesarean section, post-discharge surveillance, Ghana, maternal morbidity
1 Results
1.1 Cohort Characteristics
A total of 264 CS procedures (126 elective [47.7%], 138 emergency [52.3%]) performed on 264 unique patients were included in the analytic cohort. All procedures were from unique patients; no repeat CS within the study period were identified, confirming independence of observations. Table 1 summarises the cohort characteristics stratified by CS type.
The median maternal age was 31 years (IQR 28–34). Emergency CS patients were slightly younger than elective patients (median 30 vs. 32 years). Spinal anaesthesia (SAB) was used in the majority of procedures in both groups. Median length of stay was 4 days (IQR 3–5) and did not differ meaningfully between CS types.
1.2 SSI Outcomes and Surveillance Gap
The 30-day SSI rate was 45.1% (119/264; 95% CI 39.0–51.3%). Of 119 SSI cases, 118 (99.2%; 95% CI 95.4–100.0%) were detected post-discharge; only one SSI was identified during the inpatient admission. Accordingly, discharge-based surveillance would have captured 0.8% of all SSIs. Table 2 presents the surveillance gap metrics overall and by CS type.
| SSI Outcomes by Caesarean Section Type | ||||
| CS type | n | SSI n (%) | Post-discharge n (%) | Median days to SSI (IQR) |
|---|---|---|---|---|
| Elective | 126 | 58 (46.0%) | 58 (100.0%) | 6 ( 5–9 ) |
| Emergency | 138 | 61 (44.2%) | 60 (98.4%) | 8 ( 5–11 ) |
The median time from CS to SSI diagnosis was 7 days (IQR 5–10), spanning 3–6 days after the median discharge date. The timing distribution was similar for emergency and elective procedures. Figure 1 illustrates the distribution of SSI diagnosis by day and the cumulative detection curve.
1.3 Surveillance Horizon Analysis
Table 3 reports the cumulative SSI capture rate and exact 95% binomial CIs at each pre-specified surveillance horizon. A surveillance window of 7 days post-operation captured 52.1% (95% CI 42.8–61.3%) of all SSIs; extending to 14 days captured 90.8% (95% CI 84.1–95.3%). Complete ascertainment (100%) was achieved only at 30 days.
| Surveillance horizon (days) | SSIs captured (n) | Cumulative capture (95% CI) |
|---|---|---|
| 0 | 0 | 0.0% (95% CI 0.0–3.1%) |
| 3 | 0 | 0.0% (95% CI 0.0–3.1%) |
| 5 | 45 | 37.8% (95% CI 29.1–47.2%) |
| 7 | 62 | 52.1% (95% CI 42.8–61.3%) |
| 10 | 89 | 74.8% (95% CI 66.0–82.3%) |
| 14 | 108 | 90.8% (95% CI 84.1–95.3%) |
| 21 | 115 | 96.6% (95% CI 91.6–99.1%) |
| 30 | 119 | 100.0% (95% CI 96.9–100.0%) |
1.4 Patient Return Pathway
Figure 2 shows the service through which post-discharge SSIs were encountered and the cumulative capture yield at each surveillance horizon. Of 208 SSI-related return visits with linked service data, 200 (96.2%) presented to General Service OPD; 7 (3.4%) presented to Antenatal Service and 1 (0.5%) to OBGY Service. No post-discharge SSIs were identified through any active outreach or structured follow-up programme; all were passive re-presentations.
1.5 Predictors of 30-day SSI
Results of the multivariable logistic regression are presented in Table 4. With an EPV of 23.8, the model was adequately powered; the Hosmer–Lemeshow test indicated good fit (p = 0.387).
Younger maternal age was the only statistically significant predictor of SSI (OR 0.92 per year; 95% CI 0.87–0.97; p = 0.004). CS type, previous CS, estimated blood loss ≥500 mL, and prolonged operative duration were not independently associated with SSI in this cohort.
| Predictor | OR (95% CI) | p-value |
|---|---|---|
| CS type: Emergency (ref: Elective) | 0.86 (0.47–1.57) | 0.62 |
| Age (per year) | 0.92 (0.87–0.97) | 0.004 |
| Previous CS | 1.03 (0.56–1.91) | 0.92 |
| High EBL (≥500 mL) | 1.45 (0.86–2.45) | 0.17 |
| Prolonged surgery (>60 min) | 1.50 (0.90–2.52) | 0.12 |
| EPV, events per variable; OR, odds ratio; CI, confidence interval. Adhesions excluded due to quasi-complete separation (n = 1 case). Length of stay excluded as endogenous. Hosmer–Lemeshow p = 0.387. | ||
1.6 Age Gradient in SSI Risk
Figure 3 illustrates the SSI rate by maternal age group and the model-predicted probability across continuous age. The highest observed SSI rate was in the 25–29 year age group (60%; 95% CI 48–71%), with a progressive decline in older groups; the ≥35 year group had the lowest rate (30%; 95% CI 19–44%). The OR of 0.92 per year corresponds to approximately 57% lower odds of SSI over a ten-year increase in maternal age (0.9210 ≈ 0.43).
1.6.1 Parity Sensitivity Analysis
Parity data were successfully linked for 208 of 264 participants (78.8%) via the obstetric delivery register. In the parity-adjusted model, the age OR changed minimally from 0.934 (main model, complete cases) to 0.942 (parity-adjusted model), with no parity category reaching statistical significance (all p > 0.5). The shift in the age p-value from 0.004 (main model, n = 264) to 0.08 (parity model, n = 208) is attributable to the reduction in sample size from missing parity data, not to confounding. The age-SSI association therefore operates independently of parity.
1.7 Time to SSI by CS Type
The Kaplan–Meier cumulative incidence curves for emergency and elective CS are shown in Figure 4. Both curves rise sharply between days 4 and 10 — exclusively in the post-discharge period. The log-rank test found no significant difference in SSI timing between CS types (p = 0.615), indicating that urgency of procedure does not materially influence the temporal pattern of wound infection.
2 Discussion
2.1 Principal Findings
This study demonstrates that post-caesarean SSI is overwhelmingly a post-discharge phenomenon at KNUST-UHS. Fewer than 1% of 30-day SSIs would be captured by discharge-based surveillance, and the median time to SSI diagnosis was three days after the median discharge date. These findings reveal a substantial surveillance gap that, if unaddressed, renders facility-level SSI rates meaningless for quality benchmarking or infection prevention.
The cumulative detection analysis provides actionable thresholds: a 14-day post-discharge surveillance window captured 91% of SSIs, while a 7-day window captured 52%. From a programmatic standpoint, a structured two-week post-discharge wound review — achievable within routine OPD services — would recover the vast majority of cases without requiring active community outreach.
2.2 Comparison With Existing Literature
High post-discharge SSI fractions have been reported in comparable African settings. Sway et al. found that over 60% of post-CS SSIs in a Ugandan referral hospital occurred post-discharge [@sway2019africa], and a systematic review by Allegranzi et al. noted that SSI rates in LMICs are consistently underestimated due to insufficient post-discharge follow-up [@allegranzi2011burden]. The near-total post-discharge fraction observed in the present study (99.2%) is higher than most published estimates, likely reflecting the very short length of stay (median 4 days) that leaves essentially no inpatient window for SSI to manifest.
The overall 30-day SSI rate of 45.1% is considerably higher than estimates from passive surveillance studies in comparable settings (typically 5–20%) but consistent with reports from active surveillance programmes, which consistently identify two- to five-fold more SSIs than passive methods [@wilson2017active]. Because the analytic cohort represented the near-complete CS population at this facility (100.8% of the delivery register), selection or documentation bias are unlikely explanations. The high rate is more plausibly explained by the completeness of case-finding through the electronic clinical information system and the long surveillance window.
2.3 Younger Maternal Age as an Independent Risk Factor
The independent association between younger maternal age and SSI risk (OR 0.92 per year; p = 0.004) — equivalent to approximately 57% lower odds per decade of age — is a notable finding. Crucially, adjustment for parity did not attenuate this effect (OR 0.942 in the parity-adjusted model), indicating that the mechanism is not simply one of primigravid labour difficulty. Potential explanations include differential wound-healing capacity, immune response maturity, or socioeconomic factors associated with younger age that influence post-discharge care-seeking and wound hygiene. This finding warrants prospective investigation and may inform targeted post-discharge follow-up of younger parturients.
2.4 Patient Pathway: Implications for Surveillance Design
The finding that 96.2% of post-discharge SSIs were re-encountered at General Service OPD — rather than the obstetric or gynaecology clinic — has direct implications for surveillance design. Active SSI case-finding at this facility need not involve a dedicated maternity follow-up programme; instead, it can be embedded in the General Service OPD workflow through a simple query of wound infection diagnoses linked to the CS register. This passive active-surveillance hybrid would be low-cost and operationally feasible within the existing health information infrastructure.
2.5 Strengths and Limitations
Strengths: This study benefits from near-complete population coverage (100.8% of delivery register CS), a validated electronic data source, a 30-day surveillance window aligned with NHSN standards, and the use of Kaplan–Meier methods and multivariable regression appropriate for the data structure. The sensitivity analysis confirming no repeated patients supports the validity of the independence assumption in regression models.
Limitations: First, the SSI outcome is derived from diagnoses recorded at re-presentation to this facility. Patients who sought care at another facility, a pharmacy, or who did not seek care at all are not captured; the true SSI burden is therefore likely higher than reported. Second, wound infection diagnoses were based on free-text clinical records rather than standardised CDC criteria, which require microbiological confirmation or explicit clinician attestation. The inability to classify SSIs as superficial, deep, or organ-space limits comparability with international benchmarks. Third, the single-centre design and single-year study period limit generalisability and the ability to examine seasonal trends. Fourth, parity data were missing for 19.4% of participants, though the parity sensitivity analysis used conservative complete-case methods and the primary age finding was robust.
3 Conclusion
Post-caesarean SSI at KNUST University Health Service is overwhelmingly detected after hospital discharge, with discharge-only surveillance capturing fewer than 1% of infections. A minimum 14-day post-discharge surveillance window is required to capture over 90% of cases. Because returning patients present almost exclusively through General Service OPD, surveillance can be embedded in existing OPD workflows at low incremental cost. Younger maternal age is an independent predictor of SSI, an association not explained by parity, and may warrant targeted post-discharge wound review for younger parturients. These findings provide a data-driven basis for extending post-CS wound surveillance beyond the inpatient period in similar LMIC settings.
References
Appendix
Table S1. Parity-adjusted sensitivity model
| Predictor | OR (95% CI) | p-value |
|---|---|---|
| Emergency CS (ref: Elective) | 0.79 (0.44–1.43) | 0.45 |
| Age (per year) | 0.94 (0.88–1.01) | 0.08 |
| Primiparous (ref: Nulliparous) | 0.82 (0.40–1.68) | 0.59 |
| Multiparous (ref: Nulliparous) | 0.82 (0.37–1.80) | 0.62 |
| Grand multipara (ref: Nulliparous) | 0.93 (0.22–3.81) | 0.91 |
| High EBL (≥500 mL) | 1.25 (0.70–2.24) | 0.45 |
| Prolonged surgery (>60 min) | 1.43 (0.80–2.56) | 0.23 |
| Parity linked from obstetric delivery register (±14-day window). Previous CS omitted due to collinearity with parity. Reference category for parity: Nulliparous. | ||