Strengths and Limitations
To our knowledge this is the first published study on the association between timing of IOL and maternal and newborn outcomes in low-risk pregnancies in Sri Lanka. It is also the first study from a setting with limited resources reporting on the use of a prospective individual-patient database to analyse practices and outcomes of IOL.24 This study contributes to current international and local debate on the appropriateness of IOL near term. These study findings are extremely relevant locally both for clinicians, researchers and policy makers, as IOL at 40 GW is a common practice in Sri Lanka and has a significant economic impact on the health system and healthcare resources.
We acknowledge some limitations of this study. As an observational study, we could only assess associations between IOL and birth outcomes and not causation. Generalizability of study results may be limited by the characteristics of the local context and population in this single centre study. Larger sample sizes are required to detect significant differences in rare adverse events including stillbirth or maternal or perinatal death. Although gestational age was mostly determined by ultrasound examination, for 12% of the included women gestational age was estimated by menstrual dating.
Socio-cultural background and women’s empowerment may have affected both requests for induction and the type of care offered by physicians. Specifically, early induction (IOL at 40 GW) occurred more often in women with a high level of education. Unmarried women, still subjected to social stigma in Sri Lanka,29 were significantly more represented in the group undergoing IOL at 41 GW. Thus, numbers of CS and neonatal complications may have been influenced by socio-economic status. Other authors have described similar results, where unmarried women could have limited access to care29 while higher social status or economic condition is related to an increasing medicalization of birth.30,31 However, in our study, since these imbalances among groups affect results in different directions, there may be limited risk of bias.
Though results were corrected for confounding, we cannot exclude that induced women may have differed on characteristics not captured or not reported in the data collection form (such as unreported small for gestation foetuses, mild oligohydramnios, etc.). We were not able to explore specific practices related to IOL (such as safe use of uterotonics, appropriate maternal-foetal monitoring or CS indications), therefore we cannot exclude a difference among the groups for these variables. We had no information on the level of women’s participation in the decision process during labour care, nor specific choices, inclinations or skills of operators which may have had a substantial role in the differences observed.27, 32-34 Notably, most of the evidence that we actually rely on may have some of these biases. Observational studies may not capture these aspects of care, while RCT, even though controlling these with randomization, may suffer from study effect.
Finally, another limitation related to the database is the absence of timing for risk factor onset. Hence it was impossible to differentiate between high-risk pregnancy (with risk factors before 40+0 GW) and low risk women at 40+0 GW that developed complications due to prolonged pregnancy (after 40+0 GW). A sensitivity analysis was performed to assess this limitation and results showed that it did not affect the overall findings.