Respectful maternity care in Dar es Salaam: prevalence and drivers

The analysis centers on respectful maternity care during facility-based childbirth in urban Dar es Salaam, operationalized through a structured questionnaire capturing privacy, dignity, informed consent, effective communication, non-discrimination, companionship, and freedom from physical or verbal abuse. The primary outcome is a measurable experience of respectful birth, summarized both as a prevalence estimate and through domain-specific patterns. Secondary aims examine which woman-, provider-, and facility-level characteristics are associated with respectful care after covariate adjustment.

Methods and measurement

Design and setting. The research used a cross-sectional survey in urban Dar es Salaam, Tanzania, a dense metropolitan environment where maternity services are delivered across public and private facilities of varying levels. A cross-sectional approach is suitable for estimating point prevalence and identifying correlates, while acknowledging temporal ambiguity and susceptibility to reporting biases.

Sampling and participants. A facility-based sampling frame is typical for obstetric experience surveys, enabling efficient recruitment of postpartum women shortly after delivery or during early postnatal contact. The sampling strategy likely combined selection of facilities and consecutive or systematically sampled eligible women within those facilities. Inclusion criteria commonly include age 18 years or older (or emancipated minor provisions), recent facility birth, and capacity to consent to an interview. The analytic target is to balance representativeness across facility types and volumes, with sufficient sample size to estimate the prevalence of respectful care with acceptable precision.

Outcomes. Respectful birth experience was captured through an instrument reflecting domains endorsed across global respectful maternity care frameworks: physical privacy during examinations, confidentiality, consent before procedures, respectful language, nonabandonment, freedom from detention, nondiscrimination, birth companionship, and the ability to participate in decision-making. Responses typically use ordered categories (e.g., never, sometimes, often, always) that can be analyzed as domain-specific indicators or aggregated into a composite score. A binary outcome threshold can also be specified to facilitate estimation of prevalence and adjusted odds ratios for correlates.

Exposures and covariates. Explanatory variables generally span three levels: (1) woman-level: sociodemographics (age, education, parity, marital status), obstetric characteristics (parity, prior cesarean, complications), and care-seeking (antenatal care utilization, referral); (2) provider/process-level: mode of delivery, labor induction or augmentation, operative procedures, presence of a companion, duration of labor, and documented consent; and (3) facility-level: ownership (public/private), level (hospital/health center), crowding or bed occupancy, staffing ratios, and availability of privacy infrastructure (curtains, doors, single rooms). These covariates are included to mitigate confounding and to distinguish structural effects from case mix.

Measurement properties. To ensure interpretability, respectful care instruments typically undergo content validation and exhibit internal consistency across items in the dignity/communication/consent domains. Domain scores may be standardized to facilitate comparison. If modeled as binary outcomes, sensitivity analyses can test alternative thresholds (e.g., strict vs. permissive definitions of respectfulness). For ordered outcomes or continuous composite scores, ordinal logistic regression or linear regression with robust standard errors may be used, respectively.

Statistical analysis. The analytic plan commonly includes: (a) descriptive statistics characterizing the sample and service context; (b) unadjusted comparisons by key exposures; and (c) multivariable modeling to estimate adjusted associations between exposures and respectful care. Logistic regression is often used for a binary respectful-care outcome, with adjusted odds ratios (aORs) and 95% confidence intervals (CIs). Models may account for clustering at the facility level, either via robust variance estimators, random effects, or generalized estimating equations. Where appropriate, sampling weights reflect unequal selection probabilities across facilities and delivery volumes. Model diagnostics include checks for collinearity, linearity of continuous predictors, and specification errors.

Missing data and sensitivity analyses. Experience surveys can include item nonresponse, particularly for sensitive domains such as verbal or physical abuse. Analyses typically address missingness through complete-case analysis, multiple imputation if missing at random is plausible, and sensitivity checks comparing results across approaches. Additional sensitivity analyses can test the robustness of inferences when restricting to vaginal births, excluding referrals, or stratifying by facility ownership.

Ethics and reporting. Respectful care measurement involves potentially distressing content. Ethical safeguards include informed consent, privacy during interviews, and referral pathways if participants disclose harm. Reporting standards emphasize transparency in instrument adaptation, training of interviewers, and measures to protect confidentiality.

Results and effect sizes

Prevalence. The survey estimated the proportion of women reporting a respectful birth experience under the predefined outcome definition. While specific values are not enumerated here, the authors provide prevalence estimates for the composite outcome and domain-level indicators (e.g., privacy maintained, informed consent before procedures, absence of verbal/physical abuse). In addition, they report the proportion able to have a birth companion, a service component often linked with more respectful care and better communication.

Patterns by facility and mode of delivery. Respectful care prevalence commonly varies by facility ownership and level, with potential gradients across high-volume public hospitals compared with lower-volume facilities. Mode of delivery is a salient differentiator: intrapartum cesarean and instrumental delivery can be associated with more procedures and time pressure, potentially reducing opportunities for consent and communication. If observed, such gradients are typically quantified with adjusted measures to disentangle case mix from process and structural effects.

Adjusted associations. The multivariable analyses quantify associations between exposures and the odds of reporting respectful care, expressed as aORs with 95% CIs. Woman-level correlates may include education and parity, which can influence expectations, communication dynamics, and perceived agency. Provider/process-level factors such as the presence of a companion, documented consent before procedures, and adequate pain relief frequently align with higher odds of respectful care. Facility-level correlates may include private ownership, lower crowding, or better privacy infrastructure. Where associations were statistically significant, the corresponding aORs and CIs delineate effect magnitudes and precision, while non-significant estimates help refine which targets are less likely to yield impact if addressed alone.

Domain-specific insights. Disaggregating the composite into domains helps identify where deficits are concentrated. For instance, a facility might perform adequately on physical privacy but show lower performance on informed consent or on allowance of birth companionship. Such domain-level variation informs prioritized interventions and avoids a one-size-fits-all strategy. The survey likely presents domain-wise prevalence and adjusted associations to depict these nuances.

Heterogeneity and subgroup analyses. If powered, subgroup analyses can test whether associations differ by parity, mode of delivery, or facility ownership. Interaction terms can probe whether companionship benefits are larger in crowded facilities, or whether consent processes are differentially associated with respectful experiences for cesarean versus vaginal births. Even when exploratory, these analyses guide tailored implementation strategies.

Robustness checks. Sensitivity analyses strengthen inference by testing alternative outcome thresholds, restricting to low-risk births, or employing different model specifications (e.g., ordinal instead of binary modeling if the outcome is graded). Consistency of direction and magnitude across these checks supports the stability of findings, while discrepancies prompt closer examination of measurement and confounding.

Interpretation of magnitude. In surveys of this type, aORs modestly above or below the null (e.g., 1.3 or 0.7) can still be meaningful when tied to feasible, low-cost interventions such as ensuring curtains, standardizing consent scripts, or enabling companion-of-choice. Conversely, large aORs, when present, often reflect structural gaps like extreme crowding or absence of privacy infrastructure. The reported CIs contextualize precision, highlighting where additional data may be needed for policy decisions.

Bias, limitations, and implications

Cross-sectional and self-report bias. The cross-sectional design prevents establishing temporality. Some exposures (e.g., consent processes) and outcomes (perceived respect) are contemporaneous, complicating directionality. Self-reported measures are susceptible to social desirability bias, particularly if interviews are conducted within facilities or near staff. Fear of reprisal may suppress reporting of mistreatment. Conversely, recall truncation can favor recent salient events, exaggerating negative or positive impressions. The study mitigates these risks through trained interviewers, privacy assurances, and standardized phrasing, but residual bias is likely.

Selection bias. Facility-based recruitment excludes home births and may underrepresent those who left early or delayed postnatal contact due to dissatisfaction. If participation differs by experience, prevalence could be biased upward or downward. Use of multiple facilities and standardized enrollment windows reduces but does not eliminate this bias. Future work using community follow-up can help validate prevalence against facility-based estimates.

Confounding and measurement error. Unmeasured confounding remains a threat. For example, provider workload at specific time blocks, implicit bias, or communication barriers may not be fully captured. Measurement error in exposures (e.g., crowding proxied by bed occupancy on survey days) and in outcomes (subjective perception across cultures and expectations) can attenuate associations toward the null. Incorporating facility logs, time stamps, and objective markers of crowding can refine estimates.

Generalizability. While urban Dar es Salaam offers a critical view of high-volume urban maternity care in Tanzania, patterns may differ in rural districts or other regions with different staffing, infrastructure, and norms. Nevertheless, the domains of respectful care are broadly applicable, and the analytic approach is transferable across settings.

Implementation implications. Three implementation tracks emerge:

• Structural interventions. Low-cost privacy enhancements (curtains, screens), signage affirming consent norms, and room reconfiguration to reduce unnecessary exposure can shift baseline performance. Where feasible, redesign of labor wards to support companionship and privacy is impactful.
• Process interventions. Standardized consent scripts for common procedures (vaginal exams, augmentation, cesarean) and brief communication checklists can be integrated into routine practice. Policies enabling companion-of-choice, with minimal exclusion criteria, are consistently associated with better experience and may improve clinical outcomes.
• Provider support. Training on respectful communication, implicit bias, and de-escalation, coupled with staffing models that reduce overload, can improve clinician capacity to deliver respectful care. Embedding respectful care metrics into supervision and quality dashboards helps anchor accountability.

Measurement for improvement. Facilities can adopt short, validated pulse surveys administered at discharge, triangulated with qualitative feedback to capture nuance. Domain-level dashboards enable targeted action (e.g., if informed consent lags, introduce consent checklists and audit fidelity). Importantly, any measurement initiative should ensure confidentiality and avoid punitive use of patient reports.

Policy considerations. At the health-system level, respectful maternity care can be integrated into quality standards and accreditation, with supportive supervision rather than punitive enforcement. Financing mechanisms, including pay-for-quality models, can incorporate respectful care metrics if risk-adjusted and safeguarded against perverse incentives. Policy support for companion-of-choice and minimum privacy infrastructure standards are high-yield.

Research priorities. Future work can address causal inference by leveraging stepped-wedge implementation of process changes, enabling difference-in-differences assessments of respectful care and clinical outcomes. Mixed-methods designs can clarify mechanisms: for instance, whether improved communication mediates associations between companionship and perceived respect. Validation studies should evaluate convergent validity between subjective reports and objective process measures (e.g., documented consent). Lastly, equity analyses are needed to examine whether gains in respectful care are equitably distributed across socioeconomic strata.

Clinical relevance. Respectful care is not ancillary to safety. Poor communication and lack of consent can delay escalation, reduce adherence to postnatal instructions, and erode trust for future care-seeking. Conversely, respectful interactions can improve symptom disclosure and shared decision-making, thereby reducing preventable morbidity. Integrating respectful care into routine obstetric practice aligns with the dual goals of patient-centeredness and safety, and the adjusted associations reported here help identify feasible levers for improvement.

In sum, the Dar es Salaam survey advances measurement of respectful birth experiences in a high-volume urban context, quantifies prevalence, and delineates correlates across woman-, provider-, and facility-level domains. While causality cannot be inferred and self-report bias persists, the analytic strategy and sensitivity checks enhance credibility. The findings point to pragmatic structural and process adjustments that facilities can implement now, while longer-term system redesign addresses persistent drivers such as crowding and staffing constraints.