In this study, carried out in hospitals and health centers of four regions of Ethiopia, labors and births were observed. The analysis revealed the prevalence of RMC and mistreatment of women in hospitals and health centers and identified factors associated with the observed RMC and mistreatment of women.
Respectful maternity care
On average, a woman received two-thirds of the aspects of RMC assessed. We discuss some of the practices that were least likely to be observed in our study and showed significant variation between hospitals and health centers.
Allowing women to choose preferred birthing position
Providers’ practice of allowing women to choose their preferred birth positioning occurred at the lowest frequency of all the desired behaviors; only about two in five women in health centers and one in five women in hospitals were given choices for delivery position. Quality statement 6.2 of the WHO standards for improving quality of maternal and newborn care in health facilities states that every woman should receive support to encourage her to adopt the position of her choice during labor . Bohren et al’s  systematic review of barriers to institutional delivery found that being asked to adopt unfamiliar birthing positions and having no control over choice of birthing position are important reasons why some women prefer home deliveries. In our study, the practice of allowing preferred positions was significantly higher in health centers than in hospitals. A possible reason for this discrepancy may be the relatively higher client volumes and lower staff-to-patient ratios in hospitals, which may impede providers’ ability to offer more individualized care. The low level of practice of allowing women to choose their preferred birthing position could be attributed to the fact that facilities usually do not have physical structures for alternative birth positions (i.e., suitable delivery couches or floor space for squatting positions). For example, a study in Afar region in Ethiopia showed women preferred a sitting position for delivery but delivery beds that have space for a semi-sitting position were not available . Providers’ lack of training on alternate birth positions, particularly during their pre-service practicum, may also explain why some do not allow women to deliver in their preferred position. Health workers in a study in Bangladesh and Uganda reported that they had not been trained to deliver women in positions other than lying at their backs and thus did not feel confident to do so [28, 29].
A majority of women were permitted to take light food during labor and delivery, with health centers encouraging this more frequently than hospitals. The practice occurred much more frequently than in a previous study in Ethiopia in 2012 that reported only 40% of women were allowed food or fluid intake during labor and delivery . The reason for the higher rate in our study could be the result of exposure of providers to the in-service BEmONC training that includes an RMC session focused on interpersonal communication skill of providers, respecting culture, belief and values of clients .
Birth companions can improve experiences of women during labor and delivery; this is articulated in a statement by the World Health Organization . One of the promising findings of this study was health workers’ frequent practice of allowing a support person to be with women during labor. Four in five women were allowed to have a support person during labor, with no significant difference between health centers and hospitals. The finding was promising compared to another qualitative study, in Tanzania, that reported women felt ignored and neglected during child birth because family members or companions were not allowed to provide support . Similarly, a study conducted in Jordan also revealed that women felt dissatisfied with the health system when they were not allowed to have a support person in delivery room .
Provider and facility factors
Several socio-demographic and health facility factors were found to be related to observed RMC practices. First, the type of health worker was significantly associated with provision of RMC care; midwives were better RMC service providers compared to nurses, health officers and doctors perhaps because their training focuses primarily on maternity care. In Ethiopia MNH service is provided by midwives, nurses, health officers and doctors. A Cochrane review on midwife-led models of care for childbirth in high income countries showed that midwife-led care was beneficial particularly for normalizing and humanizing childbirth .
Surprisingly, male providers were observed engaging in RMC practices more frequently than female providers. This finding is difficult to interpret and runs counter to stereotype of women being more empathic and caring than men. A clue from a study of nurses’ abuse of patients in South Africa concluded that female nurses deployed violence against patients in their work as a means of creating social distance and maintaining fantasies of identity and power in their continuous struggle to assert their professional and middle class identity . A literature review on barriers to quality midwifery care discussed the triple burdens faced by female midwives: (1) reproductive (childbearing), (2) productive (economic), and (3) community management (e.g. unpaid work in support of the community). The effect of social, economic and professional barriers resulted in moral distress and burn out, which may have led to abusive behavior . The sex and professional disparity in the provision of RMC calls for strengthened intervention starting from teaching institutions, in-service training and health program administration to institutionalize provision of RMC by all providers male and female. This is also in line with MOH’s health sector transformational agenda of creating a caring, respectful and companionate health professionals .
The third factor that affected provision of RMC was the presence of birth companion. Women were more likely to receive RMC when birth companions were allowed in labor. Presence of birth companions helped the women receive emotional and physical support and comfort from their loved ones, and removed some of the burden from health workers. Respondents in studies in Tanzania discussed how birth companions assisted and encouraged women, because providers were absent [32, 37]. The WHO Safe Birth checklist also mentions companions in the context of calling providers for help when needed .
The final factor that showed a significant relationship to the provision of RMC services was implementation of SBM-R©quality improvement approach; facilities that implemented the approach showed higher level of RMC compared to those who did not. SBM-R© was one of the quality improvement approaches designed to promote RMC reviewed by Bowser and Hill in the 2010 landscape analysis exploring evidence for mistreatment of women in facility based childbirth . Integrating RMC in quality improvement approaches is important in order to improve care for women. Experience of care is an integral part of the WHO’s Quality of Care Framework for Maternal and Newborn Health  and RMC improves the experience of care.
Mistreatment of women
Article IV of the UN’s universal rights of childbearing women document states that every woman has the right to be treated with dignity and respect . In this study, more than a third of the women observed in delivery were not treated with respect, that is, they experienced at least one form of D&A, defined as physical abuse, verbal abuse, violation of privacy and abandonment. In observational studies, physical abuse (slapping/hitting) is expected to be low because of a potential observer effect. In this observational study however, the level of D&A was high compared to an exit interview of women conducted in four sub-counties and Nairobi, Kenya, which reported that 20% of women experienced any form of D&A . However, it was low compared to the prevalence of D&A found in a study using exit interviews conducted in four health facilities in Addis Ababa, Ethiopia, in which 98% of women reported at least one form of D&A [43, 44]. Given the similar cultural contexts, we believe that there might have been some observational effect reducing the prevalence from what it might have been had there been no observers, though one cannot rule out an actual effect of the intervention without further research designed to rule out observer effects.
Physical abuse (woman being slapped or hit) was reported in 9% of the observations. This is much higher than observations of care in Tanzania where 2.7% of women living with HIV and 4.7% of women who were not HIV positive were physically abused in labor . Levels of observed physical abuse in this study were also higher than those reported by four client exit interview studies in sub-Saharan Africa [43, 46]. The reason for high rates of physical abuse even in the presence of an external observer was unexpected and needs further investigation as to why health workers are committing such actions. Part of the reason could be rationalization of physical abuse by health providers, with the belief to ensure safety of newborn. In a qualitative study conducted among midwifery students in Ghana and health workers in Nigeria, some students and health workers mentioned it was necessary to hit women to gain compliance [47, 48].
In this study, eight percent of women were verbally abused by health providers. This was a little higher than an observational study in a hospital in Tanzania, where providers used non-dignified language with 5.6% and shouted at 6.6% of HIV-negative women while taking their medical history . An exit interview study conducted in Ethiopia and Kenya showed 14% women in Addis Ababa hospitals  and 18% of women in Kenya were verbally abused . Reasons for health providers verbally abusing laboring women were not explored in this study but qualitative study in Tanzania suggested coming too early or too late for delivery, wearing old dirty dresses and not pushing strongly were some of the reasons why women were verbally abused by providers . A study in Ghana with midwifery students revealed that both students and their preceptors do not know how to encourage women to push or to open their legs .
The rate of verbal abuse observed was less than in client exit interview reports  . Much work is needed to eliminate verbal abuse by health providers; treating every woman with respect and dignity is a human right issue.
Though there were factors found to be related to positive treatment of women in labor, assessment of socio-demographic and institutional related factors on the observed mistreatment of women showed that none of the hypothesized factors were significantly associated. This may be related to a greater emphasis on promoting positive behaviors in the quality interventions than on eliminating negative ones, though this requires some investigation. Because we generally think of positive and negative treatment of women as being inversely related to each other and doing one would negate the other, it seems that this was not necessarily the case. Some additional analysis of the relationship between RMC practices and mistreatment of women behaviors may provide useful insight to clinicians, trainers and policy makers.
Strengths and limitations
A strength of this study is that it is one of the few that has explored prevalence of mistreatment of women through observation. Most studies conducted on mistreatment of women used client exit interviews to measure mistreatment of women, which may underestimate prevalence due to recall bias. The data collectors who observed provider-client interaction observation were clinicians experienced in BEmONC services, or independent consultants who worked in universities or other health facilities outside their permanent work stations.
Another strength of this study was that it covered both hospitals and health centers in the four major regions of Ethiopia, which strengthens its ecological validity. The study also has a number of limitations. Its main limitation is the cross-sectional design, which precludes any conclusion of causal effect. We found associations between some provider and facility-related factors and RMC but cannot conclude that these factors caused RMC. Another study limitation was the possible Hawthorne effect, in which providers will show acceptable behavior during service provision because they know that they are being observed. This effect usually diminishes with each observation and each provider was observed more than once. Also, we can not ignore the potential measurement error caused by differences in understanding among observers. To minimize the potential measurment error, highly experienced assessors who were national trainers of BEmONC training, who received 5 days of training for the observer role and were actively supervised. Lastly, the observation tool used in this study was not validated in Ethiopia as was the tool recently developed in Ethiopia . However, the study team discussed each item in the tool with participants in the data collectors training. It was useful for the observation guides to collect information on both positive and negative behaviors.