Background

Increasing access to skilled care during childbirth is a key strategy for reducing maternal and perinatal mortality and morbidity [1]. Mistreatment, however, is highly prevalent in health facilities globally and ranges from subtle negligence and abandonment to overt verbal or physical abuse [2,3,4,5,6,7]. The Bowser and Hill framework is commonly cited to describe seven categories of mistreatment during childbirth [4, 6, 8,9,10,11,12]. These include physical abuse, non-consented care (including denial of birth companionship), non-confidential care, non-dignified care (including verbal abuse), discrimination based on specific attributes, abandonment or denial of care and detention in facilities.

These forms of mistreatment are major human rights violations and discourage women from seeking care during birth [3, 4, 8, 9, 13,14,15,16,17,18,19]. In addition, mistreatment has been shown to negatively affect clinical outcomes [2, 6, 19, 20]. Mistreatment can occur at the level of interaction between women and health care providers (HCP) or may result from health system failures, including supply constraints or non-enabling facilities [3, 12, 21].

Ethiopia has a low facility-based birth rate of only 26% [22]. Systematic analyses of studies on mistreatment in Ethiopia have shown that among those who utilize facility-based care, almost half (49.4%) are experiencing neglectful or abusive care, with 13.6% reporting physical abuse and 16.4% abandonment [23]. Lack of training on interpersonal communication, poor working environments and high workloads are among the drivers of mistreatment [6, 13, 24]. In the long-run such mistreatment tends to be accepted as the ‘norm’ both by women and HCPs and may not be raised as concern [4, 9, 12, 25,26,27].

Although birth companions provide emotional, psychological and social support, birth companions were not widely allowed in Ethiopia. Recent studies have shown that the presence of a birth companion in the labor ward is associated with improved outcomes for both mother and baby, including increased spontaneous vaginal births, shortened labor time and higher Apgar scores [28]. In fact, women who felt that they received dignified and supportive care reported fewer newborn complications [19].

To address mistreatment in health care, the Federal Ministry of Health of Ethiopia (FMoH) focused on motivating health professionals to become Compassionate, Respectful and Caring (CRC) as one of the four priority agendas of the Health Sector Transformation Plan [29]. Although there is increasing documentation of effective interventions to reduce mistreatment globally [14, 30], there are no studies that have shown effective interventions to reduce mistreatment in maternity care in Ethiopia.

The Institute for Healthcare Improvement (IHI)Footnote 1 in partnership with the FMoH, integrated an approach to institutionalize respectful maternity care (RMC) into a large-scale maternal newborn health (MNH) focused quality improvement program. The RMC intervention aimed to empower HCPs through a life-testimonial video-based training that also contained participatory discussion and reflection, didactic sessions on communication skills and onsite coaching to devise local solutions that can enhance RMC. This study describes its development and implementation and measures its effectiveness using programmatic data.

The MNH Quality Improvement Collaborative aims to reduce maternal and neonatal deaths by 30% through the introduction of a district-wide quality improvement (QI) program. Details of its design are published elsewhere [31]. In brief, the approach brings teams together from all facilities in the district for district-level improvement collaborative work for a period of 12–15 months. Target indicators representing key evidence-processes of MNH were selected by national and regional MNH leadership based on country-wide priorities, including two RMC-focused measures related to birth companionship and privacy. The program commenced in April 2016 with a facility leadership training in QI to build leadership buy-in, followed by baseline assessment of key MNH-care inputs, processes and outcomes using data from the previous 12 months (July 2015 to June 2016) to determine areas for improvement. During the first learning session, in November 2016, QI teams from all facilities in the district were convened, trained in QI methods and presented with the baseline findings. QI teams were then supported to design QI projects to fill gaps identified in the baseline assessment. These teams pursued collective aims of improving MNH by generating and testing new ideas, using multiple Plan-Do-Study-Act (PDSA) cycles in their local facilities during the action period between two learning sessions [32]. Intensive QI coaching and MNH clinical mentorship occurred during each action period (Fig. 1). When training gaps were found during baseline assessment, HCPs received basic emergency obstetric and newborn care (BEmONC) training which included RMC orientation.

Fig. 1
figure 1

District-wide quality improvement approach of IHI

Full size image

IHI provided the QI training, the baseline assessment tool and supported facilities in conducting baseline assessment. Coaching visits to the facility teams were done by IHI senior project officers, while BEmONC training was provided by Ethiopian Midwives Association in collaboration with IHI. In close consultation with the regional health bureaus, three districts were targeted as prototype districts in Tigray, Oromia and Southern Nations, Nationalities, and People’s (SNNP) Regions (Table 1).

Table 1 Health facilities by residing population, birth volume and providers who participated in the training session, in the three districts
Full size table

RMC intervention

Design of the RMC videos

We conducted a focus group discussion (FGD) with IHI senior project officers who had first-hand experiences as HCPs in rural settings. Focus group participants drew upon their experiences with supporting and listening to pregnant women during community engagement activities. They also had witnessed as clinical mentors and HCPs disrespect and abuse that women faced first-hand. In the FGD, we explored the current state of RMC-related issues in the program-supported districts. These findings were consolidated and key themes identified. Three testimonial scripts were written to capture these key themes, depicting a woman with an uncomplicated birth, another one with referral and emergency care, and an adolescent pregnant woman with preterm labor (Additional file 1). The scripts were three to four minutes long and translated into Amharic. Volunteer actresses were then trained to perform the scripts in video testimonials to protect confidentiality.

Delivery of the RMC training module

The videos were shown to participants during the second learning session. Participants in the three districts are described in Table 1. The session was attended by multidisciplinary health professionals, including facility leadership, MNH clinical providers, data managers and health extension workers. The three videos were followed by participatory reflection and discussion. Participants were asked to reflect on the videos using questions depicted in Additional file 2.

After discussion, a short presentation followed on the prevalence of mistreatment in Ethiopia and a skills-building session on empathic communication and relationship development with women (Additional file 2). Following the session, participants returned to their QI teams to develop new ideas or local solutions to enhance RMC in their facilities using multiple PDSA cycles. The importance of testing changes to enhance RMC was reinforced by facility coaches during coaching visits in the action period. These visits also helped to collect data and assure data quality. A minimum of three coaching visits happened per facility in each action period.

Methods

Data collection

Measures targeted for improvement through this intervention were privacy and birth companions offered during labor and childbirth. Monthly programmatic data indicating the percentage of births with privacy maintained and birth companion offered were collected from the 17 health centers and three primary hospitals in the three districts from November 2016 until January 2019 for a total of 27 months. Data were sampled from 30 births in the previous month that had been monitored using the FMoH-adopted Safe Childbirth Checklist (SCC). Systematic sampling was used for facilities that have higher numbers of births (Additional file 3- SCC). For facilities with lower births (30 or less), all SCCs filled-in during the past month were reviewed. IHI senior project officers collected the data and entered these into the program database as part of their routine work. Even though RMC training addressed all seven categories of mistreatment, only births with privacy and birth companion offered were documented in the database. Hence, we used these two measures to assess effectiveness of the training module. We only registered whether a companion during birth was offered and not whether a birth companion was actually present during birth.

New ideas tested in facilities were extracted from routine QI coach documentation and evaluated based on quantitative criteria for “success” based on run chart rulesFootnote 2 [32]. Those with higher degrees of success and with an RMC focus were extracted for this analysis.

Data analysis

We conducted an interrupted time series and regression analysis using STATA version 13 to analyze the effectiveness of the intervention. In the regression analysis, we analyzed short-term effects of the intervention which measure the first 10 to 11 months following the training (February/March to December 2017 during which direct project support occurred), while long-term effects measure the impact of the intervention after direct support ended. We used the Bowser and Hill mistreatment categories to label a ‘change idea’ as having a component that aims to enhance RMC. We presented the new change ideas implemented in the facilities that successfully enhanced the experience of care for women as results.

Results

Quantitative results on privacy and birth companion offered during labor and birth

A total of 23,129 births took place during the 27 months from November 2016 to January 2019 in the 20 health facilities. On average, in each of 17 health centers 34 births were attended per month. In the three primary hospitals on average 96 births occurred per month (Table 1).

Figure 2 shows an interrupted time series for the percentage of sampled births offered a birth companion by district. Figure 3 shows an interrupted time series analysis for the sampled percentage of births where privacy was maintained by district. The first vertical lines indicate when the second learning session in which RMC was introduced. This was conducted in the district of Oromia in February 2017 and in Tigray and SNNP in March 2017. Baseline data were collected from November 2016 (when the SCC was introduced) until February/March 2017. The second vertical line depicts the end of the direct project support (the fourth learning session) in December 2017, while data collection continued until January 2019.

Fig. 2
figure 2

Percentage of births offered companionship in three regions in Ethiopia

Full size image
Fig. 3
figure 3

Percentage of births with privacy maintained in three regions in Ethiopia

Full size image

As shown in Table 2, regression analysis in Tigray showed a significant increase in percentage of births with companion and privacy during the project period (increment trend of 18% from the baseline trend of reduction of 27%, p < 0.001) which was maintained in the long-term, a year after the end of the direct project support (increment trend of 27%, p < 0.001). In SNNP, there was a significant short-term effect on the percentage of births with companion and privacy (increment trend of 26% from the baseline increment trend of 1%, p = 0.02) while this was not reliably sustained in the long-term (with reduction trend of 1%, p = 0.94). In Oromia, the short-term effect was not as remarkable (with reduction trend of 46% from the baseline reduction trend of 78%, p = 0.02) while the long-term effect was significant with an increased percentage of births with companion and privacy (increment trend of 77%, p = 0.002).

Table 2 Regression analysis of births with companion offered and privacy maintained by region
Full size table

Interventions deployed to enhance RMC

In addition to the outcome data, we assessed the new ideas tested in the facilities as part of improving the overall MNH quality of care. Out of a total of 73 new ideas tested by the QI teams, 27 were related to RMC, and among these, 23 met the criteria for inclusion in the change package. A list of successfully tested new ideas related to improving the experience of care of women is shown in Table 3. In all three districts, pregnant women conferences, where pregnant women come together monthly for group counselling, were modified to include discussions on new efforts to maintain women’s privacy during labor and birth and encourage women to bring a birth companion. A tour of the labor ward and a coffee ceremony were also incorporated into pregnant women conferences. To ensure privacy of women during birth, health facilities developed new ideas such as using screens, including those that are made from locally available materials.

Table 3 RMC related new ideas tested in the health facilities
Full size table

During reflection and discussion following the videos, participants noted that the videos reflected the reality of women’s experience of care during birth and helped them see their care from their patients’ perspectives. Examples of disrespect mentioned in the videos that resonated with HCPs, included lack of cleanliness in the facilities and failure of HCPs to introduce themselves to women. Discussions with participants further revealed that testimonial videos appealed to their feelings and allowed them an opportunity to be ‘in their patients’ shoes’ and empathize with women.

Discussion

Our study revealed that RMC-focused interventions embedded into a district-wide QI approach led to significant improvements of the two measures of RMC in Tigray and SNNP. Significant improvements were noted after 13 months of the end of the project in Tigray and Oromia. The combination of RMC training sessions, embedded into the QI initiative, helped HCPs both understand and address some of the systemic issues that contributed to mistreatment.

Failure to allow a family companion during institutional childbirth is one of the deterrents to utilization of maternity care in Ethiopia and other low- and middle-income countries [4]. HCPs who had received RMC training specifically recognized the importance of encouraging family support and companionship. This led to improvement in the practice of offering birth companionship in the intervention sites.

Less notable results were seen in both privacy and birth companion option data in Oromia from October to December 2017. This may have been due to civil unrest that took place during this time, affecting short-term effects. In SNNP, long-term effects may have been affected by supply shortages of SCC; in cases where data were not recorded, it was assumed that services were not offered.

Previous studies of RMC-related interventions have shown the importance of a multifaceted approach, including training on RMC and addressing barriers of RMC [14, 30, 33]. Studies in countries such as Kenya and Tanzania, using pre- and post-comparative evaluation designs, showed reductions in mistreatment, ranging from 7 to 66% [14, 33]. We were not able to show specific reductions in mistreatment. Our analysis, however, showed significant increase in births where privacy was maintained and birth companionship was offered following RMC training.

Strengths and limitations

This study adds to the limited literature on successful strategies to improve RMC in sub-Saharan Africa in a sustainable manner. Our study also has important policy and programmatic implications in that QI methods can be applied to identify and address root causes of disrespect in MNH. In combination with focused RMC skills building, QI could address individual and system level barriers to RMC. As the intervention districts were distributed over the three agrarian regions of Ethiopia, findings may be generalizable to other agrarian contexts.

Our study has some important limitations. Ideally, we would have liked to interview women who gave birth recently as key informants to develop the video scripts. The videos, however, were made in the initial stage of the project when we had not yet built the necessary level of trust with the local communities and were in a relative position of power to patients. Therefore, we opted to interview project staff who had extensive clinical mentorship, patient support, and community engagement experience as key informants to develop personas to characterize a sample of positive and negative experiences with the health system. Scripts were developed based on reported issues and mistreatment witnessed or experienced in health facilities, which have later been confirmed by HCPs. Scripts of persona testimonials were developed based on these findings and volunteer actors recited the scripts to protect confidentiality of actual users. As the project matured, we were able to supplement the videos by inviting community representatives to learning sessions to share their reflections on the video content, their personal experiences and priorities. This process further helped us confirm that the experience portrayed in the videos had actually reflected the reality.

Since our analysis is based on available programmatic data, we could not use all seven Bowser and Hill’s categories of mistreatment. That would have required to interview women and observe their interactions with HCPs. That said, privacy and support by a companion were identified as priority problems by women in Ethiopian settings and hence selected by FMoH to be included in the Ethiopian SCC [13, 25, 34,35,36]. Despite this limitation, the RMC training module was specifically designed to address all seven mistreatment categories. In this study, we focused on ensuring privacy (non-confidential care) and allowing family companionship (non-consented care) due to availability of these data. Other domains of RMC may have been incorporated into change ideas across any of the target indicators. Further study exploring impact of interventions in all domains of RMC is warranted. Validated tools that help to measure women-centered maternity care are becoming increasingly available and can be used in the evaluation of intervention implementation [21].

Another important limitation is that the SCC showed only if a birth companion was offered but not their actual presence. Hence, we were not able to indicate how many of those offered have actually had a birth companion during labor and childbirth. It would be important to include a variable that measures the actual presence of a birth companion in the SCC and not just the proposition of such options. In addition, as data were collected from the SCC, shortages of the form in some health facilities led to lower measurement of coverage even if privacy and birth companionship were offered, contributing to underestimate impact. Finally, without a comparison district, it is possible that the results were related to other factors, including the national initiative on CRC [29] and attributing results to the RMC approach alone is difficult.

Conclusion

This study suggests that integrating RMC training into a QI program is effective in improving RMC. Use of testimonial videos are especially helpful as they appeal to the heart and remind HCPs of their moral obligation to treat women with dignity. Embedding this intervention within an ongoing QI effort enabled HCPs to look deeper into the care process and to reform it in ways that are genuinely family- and women-centered. Complementing RMC training with onsite coaching helped in institutionalizing RMC in the targeted facilities. These interventions could be replicated in similar settings to ensure women get RMC they deserve. Further studies would be useful to evaluate impact of such interventions on person-centered maternity care comprehensively. It would especially be important to include views and experiences of childbearing women directly in making the videos. Finally, it would also be important to add a variable on the actual presence of a birth companion in the SCC to enable measurement of actual presence of birth companionship.