Effects of the WHO Labour Care Guide on cesarean section in India: a pragmatic, stepped-wedge, cluster-randomized pilot trial

The study addresses the challenge of rising cesarean section rates worldwide, many without clear medical indication, which exposes women and babies to unnecessary risks. Ensuring high-quality intrapartum care is crucial to reducing maternal and neonatal morbidity and mortality, particularly in low- and middle-income countries. The WHO historically recommended partograph use to guide labor care, but its implementation has been inconsistent due to training, workload, supply, and policy barriers. In 2018, WHO updated intrapartum care recommendations, redefining active labor at 5 cm cervical dilation and abandoning the 1-cm-per-hour rule and alert/action lines. In response, the WHO developed the Labour Care Guide (LCG) to support evidence-based, respectful, woman-centered care, including monitoring labor progress and supportive care practices. The research question was whether implementing an LCG strategy (provider training plus monthly audit and feedback using Robson classification) reduces overuse of cesarean section—particularly in low-risk women (Robson Group 1)—and improves intrapartum care processes and outcomes compared with routine care using the simplified partograph.

Evidence shows cesarean rates have increased markedly since 1990, often without medical indication. The WHO partograph, supported by a multicenter trial in the 1990s, optimized intrapartum interventions and outcomes but suffers from poor real-world implementation due to systemic barriers. WHO’s 2018 recommendations emphasized a more individualized approach to labor progress (active phase at 5 cm dilation, removal of the 1-cm/hour expectation). The LCG (2020) was co-designed with providers and tested for usability across countries, aiming to strengthen both clinical monitoring and supportive care (companionship, mobility, oral intake, preferred birth position, pain relief). A WHO research prioritization highlighted the need to test LCG implementation strategies and evaluate effects on maternal and perinatal outcomes. Prior effectiveness data are sparse; one single-center randomized trial in India (Pandey et al., 2022) reported large reductions in cesarean use with LCG versus modified partograph, as well as reduced oxytocin use, but broader generalizable evidence from randomized designs is limited.

Design: Pragmatic, stepped-wedge, cluster-randomized pilot trial in four public tertiary hospitals (clusters) in Karnataka, India, conducted 1 July 2021–15 July 2022. One hospital transitioned from control to intervention every 2 months, with a 2-week transition period, following a concealed, computer-generated random sequence. Setting and participants: Hospitals provided comprehensive emergency obstetric care, had >4,000 annual births and ≥30% cesarean rates. Eligible participants were all women giving birth at ≥20 weeks’ gestation at participating hospitals during the study. Data were collected from admission for childbirth until discharge, transfer, death, or 7 days after admission. Control: Usual intrapartum care, including the WHO simplified partograph; standardization training on partograph and dissemination of WHO intrapartum recommendations were provided at trial start. Intervention (LCG strategy): Co-designed implementation comprising (1) provider training—2-day workshops for all labor ward staff (train-the-trainer model), followed by 8 weeks of low-dose, high-frequency, case-based and bedside teaching cycles; refresher sessions as needed; immediate replacement of partographs with LCG and supportive supervision; and (2) monthly audit and feedback meetings on cesarean rates using WHO Robson classification, with brief training on data interpretation and using audit and feedback to optimize practice. Facilities addressed anticipated implementation barriers via policy/protocol revisions, labor ward reorganization and addressing supply constraints. Logbooks, tracking sheets and site visits monitored fidelity. Outcomes: Primary outcome was cesarean section among women in Robson Group 1 (nulliparous, singleton, term, cephalic, spontaneous labor). Secondary outcomes included other intrapartum interventions (cesarean in other Robson groups, augmentation, artificial rupture of membranes, episiotomy, operative vaginal birth), maternal severe morbidity (e.g., third-/fourth-degree tears; postpartum hemorrhage requiring balloon tamponade/surgery; suspected/confirmed infection requiring therapeutic antibiotics), fetal/neonatal outcomes (stillbirths—antepartum/intrapartum, Apgar <7 at 5 minutes, resuscitation/ventilation, NICU use, HIE admissions, composite neonatal morbidity, neonatal and perinatal death), hospital stay durations, and women’s experience of care (surveyed among a sub-sample of Group 1 or 3 women in the last 15 days of each 2-month period, with liveborn babies, age ≥18, consenting). Sample size: Based on pre-trial data (≈24,000 births/year; overall cesarean ≈44%; Robson Group 1 cesarean ≥40%), the trial was designed for 92% power to detect a 25% relative reduction in Robson Group 1 cesareans (40%→30%) assuming ICC=0.02, cluster autocorrelation=0.90, average 300 women/cluster/time period, coefficient of variation of cluster size=0.60. Data collection and management: Trained staff abstracted non-identifiable individual-level data into REDCap via tablets. Data validation, cleaning algorithms, audits, and site monitoring were used to ensure quality. Women’s experience surveys were interviewer-administered in Kannada, Hindi or Marathi by day 7 postpartum or at discharge. Statistical analysis: Intention-to-treat (by planned exposure). For binary outcomes, generalized estimating equations with modified Poisson and log link, exchangeable correlation, with Manck and DeRouen bias correction and degrees-of-freedom adjustment appropriate for few clusters. Models included treatment indicator (control vs intervention) and categorical time period. Results reported as relative risks (RRs) with 95% CIs; for continuous duration outcomes, mean differences in days. ICC for the primary outcome estimated during control period. No multiplicity adjustment for secondary outcomes. Ethics: Conducted per Declaration of Helsinki, Ottawa Statement for cluster trials, and GCP. Approvals from Alfred Hospital HREC (737/20) and Indian institutional/state committees; trial registered (CTRI/2021/01/030695). Waiver of individual consent for routine clinical data; separate informed consent for experience surveys. Independent DSMC oversight.

• Participants: 26,331 women gave birth to 26,595 babies across four hospitals; analysis included 11,517 women (control) and 14,814 (intervention); 1,080 women in transition excluded from main analysis. Baseline characteristics were similar between periods; Robson Group 1 represented 30.8% (control) and 29.0% (intervention).
• Primary outcome: Cesarean in Robson Group 1 reduced from 45.2% (1,602/3,543) in control to 39.7% (1,709/4,302) in intervention; crude absolute difference −5.5%; RR 0.85 (95% CI 0.54–1.33), P=0.1088. ICC during control: 0.015 (95% CI 0–0.043).
• Maternal process-of-care: • Cesarean in Robson Groups 1 and 3: 30.9% control vs 26.9% intervention; RR 0.81 (0.59–1.11). • Cesarean in Groups 1,2,3,4,5: 51.3% both periods; RR 0.92 (0.78–1.10). • Cesarean (all women): 50.5% control vs 50.7% intervention; RR 0.91 (0.71–1.15). • Augmentation with oxytocin during spontaneous labor: 27.3% control (2,273/8,318) vs 9.3% intervention (912/9,764); crude absolute difference −18.0%; RR 0.34 (0.01–15.04) with wide CI reflecting variability. • Artificial rupture of membranes: 6.7% control vs 5.7% intervention (RR not estimated). • Episiotomy (vaginal births): 55.0% control vs 65.9% intervention; RR 0.99 (0.73–1.35). • Operative vaginal birth: 1.96% control vs 2.63% intervention; RR 1.12 (0.13–9.36). • Time from admission to birth (days): mean 0.30 control vs 0.34 intervention; difference 0.05 (−0.31 to 0.41). • Time from birth to discharge (days): mean 3.52 control vs 3.29 intervention; difference −0.23 (−1.30 to 0.84).
• Maternal outcomes: Very low prevalence and no clear differences: • Third-/fourth-degree tears: 0.22% control vs 0.12% intervention; RR 0.51 (0.01–29.16). • PPH requiring balloon tamponade/surgery: 0.40% vs 0.19%; RR 0.38 (0.00–84.07). • Infection requiring therapeutic antibiotics: 0.46% vs 0.77%; RR 2.12 (0.06–70.96).
• Fetal/neonatal outcomes: No clear differences: • Stillbirth: RR 0.97 (0.43–2.19); neonatal death: RR 1.31 (0.37–4.71); perinatal death: RR 1.06 (0.41–2.73). • Apgar <7 at 5 min: RR 1.17 (0.86–1.59); bag-mask ventilation: RR 1.21 (0.08–18.75); mechanical ventilation: RR 1.29 (0.36–4.66). • NICU >48 h: RR 1.14 (0.47–2.79); NICU for HIE: RR 0.40 (0.04–3.74); composite neonatal morbidity: RR 1.11 (0.32–3.79).
• Women’s experience (Robson 1 or 3): No meaningful differences in satisfaction or respectful care domains; e.g., labor companion reported by 83.9% control vs 76.9% intervention; being offered pain relief was low in both (5.2% vs 15.3%); overall satisfaction very high (97.6% vs 99.3%).
• Adverse events: Control period had 5 maternal deaths, 196 neonatal deaths, 367 stillbirths; intervention period had 13 maternal deaths, 200 neonatal deaths, 449 stillbirths; none judged related to the intervention.

Introducing the WHO Labour Care Guide through a structured implementation strategy (training plus audit and feedback) in high-volume Indian public hospitals yielded a non-significant 5.5% crude absolute reduction in cesareans among low-risk (Robson Group 1) women and a notable crude reduction in oxytocin augmentation, without differences in maternal or neonatal adverse outcomes or women’s reported experiences. These findings suggest potential for the LCG strategy to decrease intrapartum intervention overuse, aligning with WHO recommendations that emphasize supportive, individualized care and more realistic expectations of labor progress. While smaller and underpowered for definitive effectiveness estimates—particularly for rare adverse events—the pilot’s direction of effect and lack of apparent harm support progression to larger, definitive stepped-wedge cluster trials. Contextual factors (busy tertiary settings, existing respectful maternity care initiatives) may influence both baseline practices and effect sizes, and future research should explore generalizability and optimal implementation supports across diverse settings.

This pragmatic stepped-wedge pilot trial demonstrates that an implementation strategy for the WHO Labour Care Guide is feasible in large public hospitals and may reduce cesarean section among low-risk women and decrease oxytocin augmentation, without evidence of harm. Although not powered for definitive conclusions, the findings support conducting larger, multi-cluster effectiveness trials in settings with high cesarean rates to robustly estimate impacts on cesarean use and rarer maternal and neonatal outcomes. Future research should incorporate strategies to strengthen supportive care components (e.g., pain relief options, shared decision-making on birth position) and consider antenatal engagement to prepare women for new care practices.

• Limited number of clusters (n=4) and variability across hospitals/time periods led to wide confidence intervals and imprecise estimates for several outcomes.
• GEE-based corrections for small numbers of clusters yield approximate inferences; results should be interpreted cautiously.
• Intervention lacked an antenatal component; better preparation of women for supportive care options may enhance effects.
• Potential for some women to have given birth more than once during the 54-week study period cannot be excluded.
• Women’s experience data may be influenced by social/courtesy bias despite independent interviewing.
• Conducted in large tertiary public hospitals within one Indian state with existing quality initiatives (LaQshya) and specific staffing models; generalizability to other contexts may be limited.