Determinants of perceived quality of obstetric care in rural Tanzania: A cross-sectional study

Background: Patients’ reported opinions of the health system need to be understood in order to provide patient-centered care. We investigated determinants of women’s ratings of the quality of care during their most recent facility delivery. Methods: We conducted a census of all deliveries in the 6 weeks to 12 months preceding the survey, in villages served by 24 primary care clinics in rural Pwani Region, Tanzania. Women who had delivered children in a study facility were included in this analysis (n = 855). We interviewed women about demographic and obstetric factors and the quality of their obstetric care using a structured questionnaire. We created a composite index of perceived quality from six quality questions. We also assessed the functioning of the local health clinic using structured surveys. We used a multi-level model to analyze factors associated with women’s rating of the quality of care during delivery. Results: 14% of respondents rated the overall quality of care received during delivery as excellent. Women who listened to the radio daily reported lower quality composite scores (β: -0.99, p < 0.001). Women who reported receiving more services in ANC had higher quality scores (β: 0.46, p = 0.001), as did women receiving more delivery services (β: 0.55, p < 0.001). Women who reported disrespect and abuse during delivery had significantly lower quality scores (β: -4.13, p < 0.001). Conclusions: A woman's expectations and prior and current experiences influence her perception of the quality of care she received. Health facility characteristics did not influence ratings of overall quality. Focusing on improving the process rather than inputs of service delivery during ANC visits and delivery may increase perceived quality of delivery care in low-resource settings.

This analysis utilized baseline survey data collected for a cluster-randomized study of a quality improvement intervention for maternal and newborn health in four districts of Pwani Region, Tanzania: Bagamoyo, Kibaha Rural, Kisarawe, and Mkuranga (ISRCTN17107760). Pwani region is a primarily rural region in eastern Tanzania, with most of the population employed in small-scale subsistence farming or unskilled manual labor [6]. The Ministry of Health assigns each village to a primary health care facility called a dispensary. Dispensaries offer outpatient services including reproductive and child health services and uncomplicated deliveries. The dispensaries included in this study were staffed by medical attendants, nurses, and clinical officers. Health centers and hospitals are the next two tiers of healthcare in Tanzania and both offer inpatient and outpatient services. District and regional hospitals, as well as some health centers, offer comprehensive emergency obstetric and newborn care, including caesareans and blood transfusions. Health centers and hospitals both serve as referral centers for the lower-level health facilities [8, 15]. The study sites are 24 study dispensaries and the villages served by those facilities (i.e. villages officially designated to be in the facility’s catchment area). Facilities in the four districts were eligible for the study if they were government-managed primary care facilities, with at least one medically trained staff member (e.g. doctor, clinical officer, or nurse), were actively providing delivery services, and did not have an additional, ongoing large maternal and newborn health quality improvement project. From these, the six dispensaries with the highest volumes of deliveries were selected for inclusion in each district. The population-based survey was conducted between February 13 and April 28, 2012 as part of the baseline assessment for the study.We conducted a full census of all households in the study areas (30,076 households) to identify women who delivered between six weeks and one year prior to interview and were at least 15 years of age. Identified women were invited to participate in a structured interview. All women who provided written consent, or in the case of minors under 18 years of age, their assent and guardian consent, were interviewed. Participants were included in the current analysis if they delivered their most recent child in one of the 24 facilities included in the study (Figure 1). Study population, Tanzania, 2012. ϕ Of the women who delivered in a non-study facility, 59.8% delivered in a hospital, 18.5% delivered in a health center, and 21.7% delivered in a dispensary. The survey and consents were developed in English, translated to Swahili, back-translated and pre-tested to ensure accuracy. Detailed data collection methods have been previously reported [16]. We also conducted an assessment of the 24 primary care facilities serving the study population from December 5, 2011 to May 15, 2012 utilizing a structured questionnaire that assessed human resources, infrastructure, and services available at the facility. The survey was adapted from the needs assessment created by the Averting Maternal Death and Disability Program (AMDD) and the UN system that has been previously used in more than 30 countries, including Tanzania [17]. During the same time period we administered structured job satisfaction surveys to all health workers and clinical vignettes to all health workers trained to provide deliveries (nurses and clinical officers). The study was approved by the ethical review boards at Columbia University in New York and in Tanzania by the Ifakara Health Institute and the Tanzanian National Institute for Medical Research. The outcome measure, patient-perceived quality of care, was created from women’s responses to six questions regarding aspects of the quality of care of their most recent delivery. These questions assess technical and non-technical aspects of quality of care [18]. Respondents rated each element of care using a five-level likert scale (excellent, very good, good, fair, poor). The scores were added to create a summative composite index with possible range from 6–30. We calculated Cronbach’s α to asses internal consistency of the scale. The questions were: (1) During your delivery, how would you rate your experience of being greeted and talked to respectfully? (2) How would you rate the knowledge and competence of health workers at this facility for this delivery? (3) During your delivery, how would you rate the experience of how clearly health care providers explained things to you? (4) During your delivery, how would you rate the cleanliness of the rooms inside the facility, including toilets? (5) How would you rate the quality of the drugs and modern equipment available at this facility (where you delivered)? (6) During your delivery, how would you rate the privacy you were given? Our conceptual framework is informed by prior research that showed that patients’ perception of the quality of care received is dependent on both their expectations of care and the content of their care (Figure 2) [12, 14, 19]. Patients’ expectations in turn may be influenced by their social and economic standing, education, past health care experiences, self-perceived health and well-being and the opinions of their community members [13]. We assessed demographic and household variables including age, education (any secondary education versus less education as women with secondary education have been shown to be more selective users of health care) [20], occupation (farmer or homemaker versus skilled or student), woman as head of household as a measure of greater control of own health decisions, and exposure to radio (at least once per week versus less) as a measure of exposure to media. We also assessed each woman’s current health status using the EQ-5D (EuroQol Group, Rotterdam, Netherlands) as a measure of her general health status, expecting that women who are more ill may have higher expectations for health care than healthier women [11]. For each woman, we constructed an index of relative wealth using principal components analysis of a set of 18 questions on ownership of household assets [21]. We compared women in the top 20% of wealth against all other women, as wealthier women may have greater means to seek good health care and this in turn may shape their expectations of care in their local facility. Conceptual framework for the determinants of perceived quality of care. A woman’s past experiences and prior exposure to the health system were measured through her number of prior deliveries, if she had at least one prior delivery in a health facility (as a measure of her exposure to institutional delivery care), if she had ever had a child who died as a newborn (as a measure of a strong negative prior experience), if she attended antenatal care (ANC) for more than 3 visits for the delivery under study (as a measure of the intensity of her recent exposure to maternal health care), and the services she reported receiving in ANC (a standardized index of 8 items: weighed, height, blood pressure, urine sample, blood sample, malaria prophylaxis, tetanus vaccine, and iron supplements). In our study area, most women who deliver in their local dispensary also receive ANC there. Following Donabedian’s model, we assessed the structure and process of care using facility data and maternal self-report [12]. To assess the structure, or inputs, we evaluated whether the facility had access to clean water, the number of health workers available, and the facility average health worker self-assessed confidence in obstetric, newborn, and HIV skills. We assessed health worker scores on clinical vignettes as a measure of the knowledge and competence of facility staff [22]. We developed an index of essential maternal and newborn equipment, supplies, and drugs using the Tanzanian Ministry of Health required list, previously reported indices, and an expert review panel [8, 23]. Structural indicators were measured at a single point in time between December 5, 2011 and May 15, 2012, but these are slow-changing and so likely prevailed during her delivery. The process indicators at the facility are proxies for the services she may have received and include a standardized index of basic emergency obstetric functions provided in the past three months and a standardized index of the number postnatal services routinely provided (11 services included). We assessed the content of care the woman received during her delivery using an additive index of self-reported receipt of nine recommended services during delivery and immediately postpartum. These services were: mother checked, baby checked, uterotonic received, and mother given advice on: immediate feeding, exclusive breastfeeding, umbilical cord care, washing hands, immunization, and how to avoid chilling the baby. Other delivery variables included report of delivery complication and report of any disrespect or abuse. Women were asked if they experienced disrespect or abuse during delivery, and the terms were not further defined. The report is therefore their perception of what it means to experience disrespect or abuse. Outcomes of the visit included loans or sales of assets to pay for the delivery and child survival. Financial hardship related to delivery may negatively influence her rating of quality. Univariate statistics were calculated for individual and clinic-level characteristics and data were examined for variable distribution, outliers, and missingness. As noted above, covariates were categorized based on our conceptual framework (i.e., factors that influence expectations, secondary education, high wealth quintile). We standardized all indices in order to aid in interpretation of the regression model. We conducted bivariate regressions with each potential covariate and the perceived quality index to guide the decision of which covariates would be included in the final multi-variable model. In order to assess the contributions of both individual and facility factors to women’s rating of quality we estimated three separate 2-level linear random intercept models: a null model without covariates, a model including individual-factors (and a district fixed effect), and a final model including individual, facility, and district covariates. We then calculated the proportion of the variation in perceived quality that was due to individual and random effects versus facility-level effects. All analyses were conducted using Stata 12.1 (StataCorp LP, College Station, USA).

N/A

• Dima Health