Discordance in self-report and observation data on mistreatment of women by providers during childbirth in Uttar Pradesh, India.

Background: The study aims to assess the discordance between self-reported and observed measures of mistreatment of women during childbirth in public health facilities in Uttar Pradesh, India, as well as correlates of these measures and their discordance.

Methods: Cross sectional data were collected through direct observation of deliveries and follow-up interviews with women (n = 875) delivering in 81 public health facilities in Uttar Pradesh. Participants were surveyed on demographics, mistreatment during childbirth, and maternal and newborn complications. Provider characteristics (training, age) were obtained through interviews with providers, and observation data were obtained from checklists completed by trained nurse investigators to document quality of care at delivery. Mistreatment was assessed via self-report and observed measures which included 17 and 6 items respectively. Cohen's kappas assessed concordance between the 6 items common in the self-report and observed measures. Regression models assessed associations between characteristics of women and providers for each outcome.

Results: Most participants (77.3%) self-reported mistreatment in at least 1 of the 17-item measure. For the 6 items included in both self-report and observations, 9.1% of women self-reported mistreatment, whereas observers reported 22.4% of women being mistreated. Cohen's kappas indicated mostly fair to moderate concordance. Regression analyses found that multiparous birth (AOR = 1.50, 95% CI = 1.06-2.13), post-partum maternal complications (AOR = 2.0, 95% CI = 1.34-3.06); new-born complications (AOR = 2.6, 95% CI = 1. 96-4.03) and not having an Skilled Birth Attendant (SBA) trained provider (AOR = 1.47, 95% CI = 1.05-2.04) were associated with increased risk for mistreatment as measured by self-report. In contrast, only provider characteristics like older provider (AOR = 1.03, 95% CI = 1.02-1.05) and provider not trained in SBA (AOR = 1.44, 95% CI = 1.02-2.02) were associated with mistreatment as measured through observations. Younger age at marriage (AOR = 0.86, 95% CI = 0.78-0.95) and provider characteristics (older provider AOR = 1.05, 95% CI = 1.01-1.09; provider not trained in SBA AOR = 0.96, 95% CI = 0.92-0.99) were associated with discordance (based on mistreatment reported by observer but not by women).

Conclusion: Provider mistreatment during childbirth is prevalent in Uttar Pradesh and may be under-reported by women, particularly when they are younger or when providers are older or less trained. The findings warrant programmatic action as well as more research to better understand the context and drivers of both behavior and reporting.

Trial Registration: CTRI/2015/09/006219. Registered 28 September 2015.