Quality of and barriers to routine childbirth care signal functions in primary level facilities of Tigray, Northern Ethiopia: Mixed method study

Background Efforts to expand access to institutional delivery alone without quality of care do not guarantee better survival. However, little evidence documents the quality of childbirth care in Ethiopia, which limits our ability to improve quality. Therefore, this study assessed the quality of and barriers to routine childbirth care signal functions during intra-partum and immediate postpartum period. Methods A sequential explanatory mixed method study was conducted among 225 skilled birth attendants who attended 876 recently delivered women in primary level facilities. A multi stage sampling procedure was used for the quantitative phase whilst purposive sampling was used for the qualitative phase. The quantitative survey recruitment occurred in July to August 2018 and in April 2019 for the qualitative key informant interview and Focus Group Discussions (FGD). A validated quantitative tool from a previous validated measurement study was used to collect quantitative data, whereas an interview guide, informed by the literature and quantitative findings, was used to collect the qualitative data. Principal component analysis and a series of univariate and multivariate linear regression analysis were used to analyze the quantitative data. For the qualitative data, verbatim review of the data was iteratively followed by content analysis and triangulation with the quantitative results. Results This study showed that one out of five (20.7%, n = 181) mothers received high quality of care in primary level facilities. Primary hospitals (β = 1.27, 95% CI:0.80,1.84, p = 0.001), facilities which had staff rotation policies (β = 2.19, 95% CI:0.01,4.31, p = 0.019), maternal involvement in care decisions (β = 0.92, 95% CI:0.38,1.47, p = 0.001), facilities with maternal and newborn health quality improvement initiatives (β = 1.58, 95% CI:0.26, 3.43, p = 0.001), compassionate respectful maternity care training (β = 0.08, 95% CI: 0.07,0.88, p = 0.021), client flow for delivery (β = 0.19, 95% CI:-0.34, -0.04, p = 0.012), mentorship (β = 0.02, 95% CI:0.01, 0.78, p = 0.049), and providers’ satisfaction (β = 0.16, 95% CI:0.03, 0.29, p = 0.013) were predictors of quality of care. This is complemented by qualitative research findings that poor quality of care during delivery and immediate postpartum related to: work related burnout, gap between providers’ skill and knowledge, lack of enabling working environment, poor motivation scheme and issues related to retention, poor providers caring behavior, unable translate training into practice, mismatch between number of provider and facility client flow for delivery, and in availability of essential medicine and supplies. A facility based cross-sectional explanatory sequential mixed method study was conducted among recently delivered mothers in primary health care facilities of Tigray regional state, Northern Ethiopia. Tigray is the northern most region of Ethiopia with an estimated total population of 5,247,005 with 21.2% of the population living in urban areas and 50.3% being female [24]. The maternal and new born health care services in the region are provided mainly by emergency obstetric surgeons and obstetricians at hospitals and health centers by midwives, nurses, and health officers. The service is given free of charge in all public health facilities on a seven day per week 24 hours a day basis. As of 2016, in Tigray Region, there were 2 comprehensive specialized hospitals, 15 general hospitals, 23 primary hospitals, 214 health centers, and 718 health posts [25, 26]. a sample size calculation for the quantitative study component among recently delivered mothers was determined by a single population proportion with 95% confidence interval, margin of error (d) of 5% and taking 54.06% prevalence (P) of overall quality of delivery care in Arbaminch, south Ethiopia public health facilities [27]; design effect of 2 and adding 10% for non-response rate. A total of 881 mothers received routine intra-partum and immediate postpartum care signal functions from 40 primary level care facilities. Additionally, a total of 225 skilled birth attendants (SBAs) working in the study facilities at the time of data collection were included. A multi-stage sampling procedure was adopted to select the districts and primary level health facilities from each district. In the first stage, three of the seven zones were selected randomly. In the second stage, nine of the 22 districts were chosen and 6 primary hospitals were randomly selected. Thereafter, all health centers with their respective catchment primary level hospitals were included with the total sample size being distributed over each of the health facilities proportionate to their sample considering average number of deliveries per facility per month. All SBAs in the study were enrolled. Finally, all eligible recently delivered women were chosen by a systematic random method until the required sample size was achieved. A referred mother requiring care at a higher level facility for further management and/or delivered by cesarean section were excluded from this study. Client exit interview tracer indicators for routine childbirth care signal functions (care that should be provided for all mothers and newborns) utilized self-administered questionnaires, facility inventories, and interviews of providers to collect the quantitative data. A 40-item knowledge tests, as well as satisfaction of health workers and facility readiness surveys (i.e., availability of infrastructure; essential medications and commodities; guidelines; staff) were conducted. Tracer indicators for facility readiness were used from the WHO Service Availability and Readiness Assessment (SARA) list, previously reported indices [28]. Twelve data collectors and three supervisors worked as data collection teams. Data collectors had previous research experience and trained for two days. Data for this quantitative study was collected between July to August 2018. We developed semi-structured questionnaires to conduct key informant interviews (KII) and Focus Group Discussions (FGD). Participants for FGDs and KIIs were selected purposively. The key informant participants were medical directors from each of the primary level health facilities, woreda, and regional health bureau maternal and child health experts and unit head of maternity wards. We assumed that these KII could better inform us of barriers to provide childbirth care than other health workers. A total of twelve KIs were conducted. Probing questions were used for a better understanding where necessary. Each participant was interviewed individually at his/her place of work, with interview duration ranging from 20 to 35 minutes by a team of trained data collectors. After training, three researchers from the College of Health Sciences at Mekelle University and Tigray Health Research Institute conducted qualitative data collection from the 9th to 29th of April 2019. The semi-structured interview guide used can be found in S1 Appendix. After the interviews, three FGD were held for SBAs working at intra-partum and immediate postpartum care ranging from 60 to90 minutes. One interviewer and note taker were involved. Skilled birth attendants with clinical work experience of six months and below were excluded from the qualitative study. All interviews and discussions were audio recorded, then transcribed verbatim in Tigrigna (the local dialect) by two independent investigators. A third investigator checked the consistency of the transcripts and verified the transcripts by listening to the tapes again. They were subsequently translated into English prior to analysis. The primary outcome investigated was quality of routine childbirth and immediate postpartum care. It was measured as a continuous variable constructed as a composite variable from the total of 32 standards of quality process of care indicators. The routine intra-partum and immediate postpartum care signal functions used in this study are grounded in validated indicators in the Tigray regional state context. Detail of the measurement and validated tool findings is found in the recent article submitted for publication [29]. Principal component analysis (PCA), the most common technique of creating a single or composite quality index, which is a variable reduction method to obtain a smaller set of uncorrelated variables from a large list of correlated variables, was used. Each component is a linear combination of the observed variables optimally weighted to account for the maximum amount of variance [30]. Therefore, quality measures reflect the minimum standards of routine intra-partum and immediate postpartum care, irrespective of the type of health facilities where the delivery service is performed. According to the PCA, QoC was defined as a binary variable of “low” to “high” on a continuous scale from 0 to 100. If a mother’s review received 75% and above, it was termed as high QoC, and otherwise received low QoC. Details of the PCA tool for measuring QoC is found in S2 Appendix. The providers’ satisfaction variable was classified as “satisfied” (providers scored 75th percentile and above), whereas below the 75th percentile was considered “not satisfied”; facility readiness was categorized as adequately ready at the 75th percentile and above and below was considered inadequately ready). Details of the PCA tool for measuring providers’ satisfaction is found in the S3 Appendix. Knowledge of providers on intra-partum and immediate postpartum care signal functions was determined using a set of 31 multiple choice questions and 9 true or false questions. Each correct answer was valued at one point, and a wrong answer attracted no points. Questions that were not answered were treated as wrong answers. Ultimately, participants were evaluated out of 100, and grouped as either sufficient knowledge (median or higher) or insufficient knowledge (less than median value). First, we entered the data in to EPI data, cleaned and analyzed it using SPSS™ version 21 software. Descriptive statistics were used to summarize the characteristics of delivered mothers, facilities, and providers. Characteristics of the study population were presented with mean and standard deviation for variables with normal distribution. The normality of distribution of quantitative variables was tested by Kolmogorov–Smirnov test. We used linear regression analysis to assess the association between quality of care and explanatory variables. Simple linear regression analyses were conducted and those independent variables with p value of ≤ 0.25 were considered for multiple linear regression with the forward likelihood ratio method. Finally, statistical significance was considered if p < 0.05. Furthermore, an index score of PCA was done after checking the suitability of the data. The correlation coefficient was set at a cut-off point of 0.4 or above. The Kaiser-Meyer-Oklin value, which was used to assess sampling adequacy, was set at a cut-off point of 0.5 [30], while the Bartlett’s test of sphericity was used to support the factorability of the correlation matrix. Furthermore, a scree plot tests and eigenvalue of over 1.0, which represents the total variance explained by a factor, were used to inspect the plotting of each eigenvalue of the factors to find a point at which the shape of the curve changes direction and becomes horizontal. All factors above the break in the plot and/or with eigenvalues over 1.0 were retained for further analysis. Lastly, further analysis was done using the Vari-max method to minimize the number of variables with high loadings on each factor. Two researchers independently reviewed the audio recorded comments line- by- line and then agreed on a set of codes; broadly categorized into those related to the quantitative checklist and codes for other emerging issues. Both researchers then jointly coded all the open-ended comments. In cases where disagreements arose between researchers, further discussion took place until consensus was achieved. The data analysis was carried out in three stages. First, familiarization involving reading and re-reading the transcripts to aid understanding of the data. Second, organizing and coding the data. The coding was determined based on the quantitative results, to aid understanding how the quantitative findings were manifest. The coding was done using Atlas ti™7.5 software. Third, data from each code point were reviewed and summarized to reduce the number of words without losing the content or context of the text and to ensure contents were internally consistent. Then content analysis and triangulation of data were done through a continuous back and forth interpretation of findings. The study protocol was approved by the Institutional Research Review Board of Mekelle University’s College of Health Sciences and Community Services Ethical Review Committee (ERC 1436/2018). Permission was obtained from all relevant authorities in the Tigray Regional Health Bureau and health facilities. Informed consent was obtained from all participants prior to enrollment in the study. Parental or legal guardian consent was obtained for participants who were under 18 years of age. Data collection was conducted confidentially while data was de-identified and de-linked with storage in a secure location.