Risk factors for maternal mortality during labour induction in an urban Kenyan tertiary hospital: A case-cohort study

Introduction: Maternal and Perinatal Death Surveillance and Response (MPDSR) aims to reduce maternal and perinatal mortality through ongoing monitoring and retrospective data analysis. The Kiambu County Referral Hospital (KCRH) MPDSR committee observed an increase in maternal mortality with the use of misoprostol for labour induction. An administrative decision was implemented in June 2019 to replace misoprostol with dinoprostone for labour induction. Here, we sought to investigate the factors associated with maternal mortality and the role of induction agents, as well as the effectiveness of the MPDSR at KCRH. Methods: We analysed all the 58 maternal mortality cases that occurred during the 36-month study period between January 2018 and December 2020 against a random sub-cohort of 232 controls. Multiple logistic regression, adjusted for intensified MPDSR activities and labour induction agents, was employed to identify factors affecting maternal mortality. Results: In multivariable analysis, postpartum haemorrhage (PPH) was strongly associated with higher odds of maternal death (aOR: 14.4; 95%CI: 2.72–75.8; p = 0.002), while gestational age ≥38 weeks was associated with significantly lower odds (aOR: 0.16; 95%CI: 0.05–0.54; p = 0.003). Use of misoprostol for induction was associated with a four-fold higher odds of maternal death compared to dinoprostone (aOR: 4.95; 95% CI: 0.56–43.8; p = 0.15), although this difference was not statistically significant. Unemployment, single marital status, eclampsia, gestational hypertension, vaginal delivery, and maternal pyrexia were associated with either higher or lower odds of maternal death, but none of these associations were statistically significant. Intensified MPDSR activities were not associated with a significant change in maternal death risk (aOR: 1.07; 95% CI: 0.32–3.55; p > 0.9). Conclusion: Although enhancing MPDSR activities did not reduce maternal mortality, inducing labour with misoprostol showed higher odds of maternal mortality compared to dinoprostone, though this association did not reach statistical significance. Our findings raise a safety concern consistent with the initial safety concern that motivated this study, and further post-marketing surveillance of labour induction methods is required to establish whether this reflects a true causal relationship. This work also underscores the urgent need to strengthen prevention and management of PPH in resource-limited settings while also ensuring careful evaluation of induction agents.