Principal findings
This largest prospective analytic study with excellent follow-up was unable to identify an additional significant risk if SAP was delayed after cord clamping. Our results show a basically unchanged SSI risk whether SAP is administered before or after clamping.
Of note, almost all estimates of the odds ratio were > 1, suggesting a numerical tendency to an increased risk in this group. Thus, the odds ratio of the main analysis does not rule out the possibility that the odds of an event may be increased by up to 36%. Many unmeasured confounders may have been present and the confidence intervals of the estimate are consistent with a potential increase in infection with administration after cord clamping.
In addition, our data show that factors other than timing of SAP administration were significantly associated with SSI risk. It was expected that increased procedure duration, increased BMI, high ASA score, contaminated wound and emergency operation were associated with an increased risk. The finding that age group of 30–40 years had a decreased risk in comparison to < 30 years may be due to a selection bias, as patients < 30 were more likely to have an emergent procedure and more frequently experienced premature rupture of the membranes. Of note, the choice of SAP agent did not play a significant role regarding SSI risk.
Results (internal and external validity)
The internal validity of our study was excellent, as hospitals throughout Switzerland participated, including smaller institutions (< 200) and large centers (> 500) beds. The multilevel analysis approach with clustering at the hospital level allowed us to control for potential variation in SSI rate and reporting between the centers.
Concerning external validity, the analysis of large prospective registries may be the ideal source for generating high-quality scientific data [27]. The overall SSI rate was low in our study (1.7%), compared to 4–5% reported by the WHO and the meta-analyses [2, 15] and up to 6–12% from a recent randomized controlled trial [28]. The difference could be explained by the fact that former studies included patients from developing countries in a different setting [2, 15]. In the latter study, the baseline characteristics suggested a relevantly higher BMI (+ 8 kg/m2) and a higher rate of patients had already had ruptured membranes, both important SSI risk factors after cesarean delivery [28]. As this is a national multi-center study with clustering at institutional level, we are convinced that our results can be transferred to similar health-care settings in resource rich countries, but this may not be the case for lower and middle income countries.
Even though other studies have shown a decreases SSI risk for SAP before incision, this was relatively small in absolute numbers: the Cochrane review reported 38/2531 mothers developed an endomyometritis before clamping compared to 70/2510 after clamping. This potentially increased risk was not confirmed by our study and must be contrasted with a potential harm to the neonate’s immune development [5,6,7,8,9]. Disrupted transmission of maternal Bacteroides strains to babies was even seen in vaginally delivered babies whose mothers underwent antibiotic prophylaxis [9].
Finally, the potentially very short time interval between the two strategies argue against a relevant differential risk.
Clinical implications
These results challenge the latest WHO recommendation which extends the time window of SAP from 60 min to 120 min prior to incision but does not consider administration after clamping [29].
Research implications
In future research, the principally unchanged or slightly elevated SSI risk for the mother has to be balanced with long-term neonatal outcomes/microbiome development.
Strengths and limitations
The main strengths of our study were the standardized evaluation of SSI cases by dedicated physicians, a post-discharge surveillance at 30 days, a mere < 10.4% loss of follow-up, routine on-site monitoring of the data collection quality, and a multilevel model that permitted adjustment for SSI variation across the institutions. The main limitation of this study, was that the exact timing of clamping was not available and therefore administration after incision served as a surrogate marker. We believe we have sufficient evidence that SAP administration after incision corresponds to “after clamping”. First, clamping is done quickly after incision, usually within 5 min [20]. This is reflected in Fig. 2, where SAP application after incision peaks between 5 and 10 min after incision, compatible with administration after clamping. Moreover, a supplementary analysis failed to identify an optimal timing window within +/− 60 min relative to incision. In addition, we performed two subgroup analyses that made post-incisional SAP administration even more likely to represent “after clamping” administration, and the results are in line with the main analysis.
Other limitations of the study were the lack of detailed information concerning the individual surgeons, type of operating room, and patient comorbidities/characteristics (such as diabetes, glycemic control, smoking, nutritional status, intraoperative core temperature, estimated blood loss, and oxygen saturation), preoperative skin preparation procedures and/or the presence of SSI intervention bundles. None of these variables are recorded in the Swissnoso database. In addition, data on antimicrobial treatment before and after surgery, group B streptococcal carriage, multiple pregnancy, and/or vaginal disinfection were not routinely recorded.
The study period extended over 10 years, with potential changes in the management of cesarean section. However, cesarean section rates in Switzerland during this period remained constant at ~ 33% [30, 31], and surgical technique has not significantly changed over this time period. Stratified analysis for the period before and after 2014 did not provide arguments for a significant confounding by the two periods.
The variable BMI was only available in 40% of patients. Increasing BMI is a well-known risk factor for SSI [28, 32, 33]. It might have been a confounding factor, but we performed several subgroup analysis with patients where BMI was available and the results were in line with the main analysis.