Implementing maternal and newborn health quality of care standards in healthcare facilities to improve the adoption of respectful maternity care in Bangladesh, Ghana and Tanzania: a controlled before and after study

[ad_1]

WHAT IS ALREADY KNOWN ON THIS TOPIC

Although respectful maternity care (RMC) is a crucial component of quality of care and a human right, there is evidence that many women do not access RMC, particularly in low-resource settings.
A large body of evidence describes the extent of RMC, and very few studies have evaluated interventions with robust methodologies to address RMC.

WHAT THIS STUDY ADDS

We performed a multicountry comparative before-and-after evaluation to measure the effect of implementing the maternal and newborn health (MNH) quality of care standards on RMC measures.
Implementing MNH quality standards under real-world health system conditions was associated with improvements in effective communication, respectful and dignified care measures and women’s satisfaction. In addition, it improved some contextual factors, enabling environments to support changes and improvements.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

This study suggests that healthcare facilities and systems in low and middle-income countries can accelerate RMC by implementing MNH quality standards and developing multilevel, context-specific interventions when adequate investment and support are provided.

Introduction

The current global agenda focus on the survival of women and their babies during childbirth and ensuring that they thrive and realise their full potential.1 Respectful maternity care (RMC) is a human right-based approach that can improve women’s pregnancy, labour and childbirth experience and address health inequalities.1 RMC refers to care organised for all women and provided to them in a manner that maintains their dignity, privacy and confidentiality; ensures freedom from harm and mistreatment and enables informed choice and continuous support during labour and childbirth.2 However, many women, particularly those in low and middle-income countries (LMICs), cannot access RMC.3 Many women experience poor quality of care (QoC) and treatment during childbirth, including disrespect and violations of their rights to privacy, informed consent and having a companion of choice during childbirth.3–6 These negative experiences of care can prevent women from seeking care in facilities during the postnatal period and for their subsequent deliveries.4 Additionally, disrespectful, abusive or neglectful care during childbirth may have direct adverse consequences for both the mother and infant.7

Women place a high value on RMC, and most healthcare providers would like to provide respectful, dignified and woman-centred care but may feel unable to do so due to resource constraints.8 Most research studies focus on identifying the extent and nature of gaps in providing RMC, and very few evaluate interventions to improve RMC. The latter often focus on training providers and fail to demonstrate a consistent sustained change over time.9–13 On the other hand, the literature suggests that complex, context-specific interventions targeting multiple levels of the health system are most likely to be effective in improving RMC.8 9 11 14–18 There is a need to advance from understanding the nature and extent of RMC gaps to developing and evaluating interventions designed to improve and sustain the adoption of RMC.9

In 2016, UNICEF/WHO published maternal and newborn health (MNH) QoC standards to improve the quality of maternal and newborn care, address health system inequities and strengthen accountability. The nine standards focused on providing evidence-based, safe care; experiencing dignified and respectful care for women and newborns and creating an enabling environment for such care.19 The standards envisioned experience of care in three domains: (a) effective communication, (b) social and emotional support and (c) respectful and dignified care. Although guidance to improve maternal and newborn care by implementing quality standards has been developed, no study has targeted improving RMC through a standard-based MNH QoC improvement pathway.20 The implementation of these standards was evaluated in seven intervention districts in Bangladesh, Ghana and Tanzania to inform the feasibility and effect of their institutionalisation within health systems. We present an evaluation of the effect of implementing the MNH QoC standards on RMC measures, focusing on effective communication, emotional support, respectful and dignified care and maternal satisfaction in each country.

Methods

Study design

We used a facility-based, controlled before and after design to measure the effect of implementing the MNH QoC standards on RMC measures. The evaluation was conducted in seven intervention districts in Bangladesh, Ghana and Tanzania. Eight adjoining districts with similar characteristics as the intervention districts were evaluated for comparison. Data were collected during two time periods: between October and December 2016 (baseline) and 18 months later, between July and November 2018 (endline) (figure 1). Multiple data collection methods were used: interviews with health workers (HWs), exit interviews with women and observation of woman–provider interaction during and after labour and childbirth. Standards for Quality Improvement Reporting Excellence (SQUIRE) guidelines were used for reporting the results.21

Figure 1

Timeline for the evaluation of the effect of implementing MNH quality of care standards implementation in healthcare facilities.

Setting and participants

Participating regions (the Rangpur Region of Bangladesh, the Upper East region of Ghana and the Njombe region of Tanzania) were purposefully selected and prioritised by the ministries of health (MOH) of each respective country. Box 1 describes the context of maternity care in each country. Facilities were selected from the public sector based on their designation to provide emergency obstetric and newborn care, maternity caseload, the absence of quality improvement (QI) interventions at baseline and catchment populations’ sociocultural case mix (online supplemental table S1). A total of 43 health facilities were included: 15 in Bangladesh, 16 in Ghana and 12 in Tanzania. Nineteen intervention facilities were selected to implement the NMH QoC standards. Individual characteristics of each healthcare facility, including the type of model of care, are described in online supplemental table S2. The intervention was assigned to UNICEF-focused districts based on criteria such as low coverage of health interventions (in Kurigram, Bangladesh), social disadvantages (in Bawku Municipal, Bolgatanga Municipal, Bongo District, Kassena Nankana and West District, Ghana) or poor MNH indicators (Ludewa and Wanging’ombe, Tanzania). Twenty-four control facilities were selected from adjoining districts with populations of similar demographic characteristics. Figure 2 describes participating healthcare facilities, women enrolled, HWs interviewed and woman–provider interaction observations organised by group, time of evaluation (baseline; endline) and country. The population of the intervention districts was approximately 2.8 million.22–24

Box 1

Context of maternity care in Bangladesh, Ghana and Tanzania

Bangladesh

The evaluation was conducted in the Ragpur region, which has a population of 2 069 273.
In 2016, the region had a maternal mortality ratio of 222 deaths per 100 000 live births and 37 neonatal deaths per 1000 live births.
The region has 69 hospitals and 2541 health centres, with 46% of births occurring in health facilities.
Kurigram district was selected to implement the intervention due to low coverage of health interventions, while Gaibandha and Lalmonirhat districts were selected as controls.

Ghana

The assessment was conducted in the Upper East Region, which has a population of 1 109 338.
In 2016, the maternal and neonatal mortality rates varied depending on the data source. According to the District Health Information Management System, there were 111 maternal deaths per 100 000 live births and seven neonatal deaths per 1000 live births. However, the 2017 Maternal Health Survey reported a mortality risk of 310 maternal deaths per 100 000 live births and 24 neonatal deaths per 1000 live births based on survey data collected in 2014 (Ghana Demographic and Health Survey (GDHS) 2014).
There are 164 healthcare facilities in the region and 1118 Community-Based Health Planning and Services, with 70% of births occurring in health facilities.
Due to social disadvantages, Bawku Municipal, Bolgatanga Municipal, Bongo District, Kassena Nankana and West District were selected to implement the intervention. In contrast, Builsa North District, Kassena Nanakana Municipal, Bawku West District, Talensi District, Gaibandha and Lalmonirhat districts were selected as controls.

Tanzania

The evaluation was conducted in the Njombe region, which has a population of 803 299.
In 2016, the region had a maternal mortality ratio of 101 deaths per 100 000 live births and 31 neonatal deaths per 1000 live births.
The region has 10 hospitals and 263 health centres, with 87% of births occurring in health facilities.
Ludewa and Wanging’ombe districts were selected to implement the intervention due to poor maternal and newborn health indicators, while Njombe and Makete districts were selected as controls.

Figure 2

Participating healthcare facilities, women enrolled, health workers interviewed, and woman-provider interaction observations organised by group and country.

Intervention

The intervention combined several strategies proposed to implement the MNH QoC standards based on QI frameworks that have been shown to change providers’ behaviours,25 26 institutionalise quality caregiving27 and address maternal and newborn care quality gaps.19 28 The main strategies used in the intervention group consisted of implementing the Every Mother Every Newborn QoC standards by (a) establishing and institutionalising QI teams and processes with the involvement of all-level leadership (MOH, district, facility and unit leaders) and (b) creating an enabling environment (including the development of infrastructure) to support the provision of quality care for mothers and newborns. Specific interventions were implemented as part of efforts to improve the care provided and experienced by mothers and caregivers. These included specific training of clinical staff in the intervention facilities on caregiving with compassion and respect as part of training on other QI interventions. Healthcare providers were also trained to engage women and their families in the decisions around the care they experienced, including seeking consent for interventions. Training also covered solicitation of client feedback and using the feedback to improve performance and quality.

In addition, to establish and institutionalise QI processes, QI teams defined change ideas, set objectives around these and implement them within Plan‒Do‒Study‒Act cycles. They use data for decision-making on quality and monitoring improvements.

Creating an enabling environment involved instituting structural changes to ensure better privacy, although temporary. In the case of the latter, some facilities procured curtains to provide separate enclosures for women in labour as part of ensuring visual privacy. Water, sanitation and hygiene as well as overall infection prevention and control measures, were identified as quality issues that affect client experience of care and were emphasised. For instance, it was considered that the cleanliness of the dedicated toilet for women within the maternity unit was a critical component of the experience of respectful care. Creating an enabling environment (eg, physical resources, human resources, policies, guidelines) to support quality care for mothers and newborns was also a key.

The intervention was pragmatically implemented in the context of routine intrapartum. Health facilities in the control group continued usual intrapartum care practice without introducing the MNH QoC standards.

Measures and data collection

RMC was measured along the domains included in the WHO framework to assess the experience of care in the context of healthcare services—effective communication, emotional support and respectful and dignified care—from the perspectives of women, HWs and observers.19 These were supplemented with input measures (eg, physical resources, human resources, policies, guidelines).29 Independent clinicians and social scientists were trained for data collection at a 7-day workshop facilitated by the study coordination. Data were collected in each facility with piloted standardised structured paper-based forms, translated to local languages and culturally adapted. Data collectors obtained consent and conducted 1827 exit interviews, interviewed 818 health personnel and directly observed 1512 woman–provider interactions. Given the consecutive invitation of all women, sometimes the same woman who participated in the observation participated in the exit interview. However, this was not a criterion for inclusion in the exit interview, and some women participated only in the observation or only in the exit interview.

Below is a summary of the standardised procedures for measurement and data collection, while detailed information has been published elsewhere.30

Exit interviews

All postpartum women recently discharged from the postnatal ward were invited to participate, regardless of age and perinatal outcome. Interviews were conducted in private rooms. The questionnaires included sections that explored women’s sociodemographic characteristics and their perceived experience of care during and around intrapartum care, including privacy, HWs’ attitudes, communication, responsiveness to women needs, respect for women’s preferences, satisfaction and experiences of disrespect and abuse.

HW interview

HWs providing care during labour and childbirth (specialists, doctors, midwives and staff nurses) were invited to participate in structured interviews to explore contextual factors potentially influencing practice change, such as formal RMC training, policies and procedures for addressing patients’ concerns, rights and identifying and reporting abuse.

Clinical observations

All women consecutively admitted for labour and childbirth during the data collection visits were invited to participate in the woman–provider interaction observation. The observation was initiated at reception and lasted until the immediate postnatal period. Shifts of independent, external observers covered 24 hours a day and 7 days a week. Each observer stayed with the same woman during the whole process. Data collectors used a structured observation checklist to assess communication between HWs and women, privacy, supportive care, labour and delivery ward layout, occupancy and availability of human resources in health.

Data management and statistical analysis

Data were collected in paper forms and entered into password-protected servers in each country. Study investigators conducted data quality assurance through supportive supervision with biweekly field visits to physically verify the completeness, accuracy and consistency of the data.

The analysis focused on reported and observed measures of RMC. It included several measures for each experience of care domain—effective communication, emotional support and respectful and dignified care. In addition, the experience of care measures was assessed by measuring overall satisfaction with care, HWs’ attitudes, overall hygiene, desire to return to the same health facility and recommendation of the facility to relatives/friends. Finally, we measured selected contextual factors that might enable or inhibit practice change.

We described women, HWs’ and observations’ baseline characteristics in each country using proportions to summarise categorical data, mean and SD or median and IQR according to data distribution for continuous data.

We compared temporal changes in RMC measures (measured as proportions with the exception of the measure ‘Number of HWs on duty’, which was measured on a numerical scale and summarised with means and SD) between the two groups using the difference-in-difference (DiD) analytical approach with models that included the main effects of group, time of evaluation and a two-way interaction term, separately for each country, adjusting for the cluster (facility) effect with robust SEs to correct for heteroscedasticity. This analytical method is used in quasiexperimental (nonrandomised) designs, where the two groups do not start at the same level at baseline. The DiD is implemented by computing two differences between groups: the first is the difference in the outcome variable between the two periods for each group. The second difference is the difference between the differences calculated for the two groups. The DiD estimate represents the differential improvements or declines in the outcomes of interest associated with the intervention. Significance was set at p=0.05, and 95% CIs were reported around estimates. Before conducting the analysis, we compared the characteristics of the groups at baseline and endline and did not find any substantial difference in participants’ characteristics (online supplemental table S4). The composition of the intervention and comparison groups was stable over time. In addition, the outcomes did not determine the selection of intervention districts. Intervention facilities were selected because the local government identified them as being located in areas with socioeconomic disadvantages and poor health indicators. Finally, we assumed that the intervention group’s outcomes would slowly improve (have a parallel trend), similar to the control group, given that RMC is on the international and national agendas. Stata V.14 (StataCorp, College Station, Texas) was used for the analyses.

Patient and public involvement

The data collection instruments were pretested in all three countries to assess their acceptability to women and to adapt them culturally based on their suggestions. We interviewed women and family members in all participating facilities to obtain their perspectives on the care they received during labour and delivery. Special efforts were made to ensure confidentiality by storing all paper forms in locked cabinets with face sheets separated from study forms; electronic data were deidentified using participant ID, and no other identifiers were included in the data set. Women and family members are thanked for their contributions in the acknowledgements of this publication.

Results

Characteristics of women, HWs and observations

Women who participated in the exit interviews were very similar in both groups in all three countries except for Bangladesh, where at baseline, more adolescents (32.5% vs 17.8%) and women with lower education levels were in the intervention group than in the control group (24.6% of women completed middle/high school vs 32.5%) (table 1).

Table 1

Baseline characteristics of participating women, health workers and observations by group and country

Interviewed HWs were similar in both groups at baseline, with a few exceptions. In Bangladesh, in the intervention group, HWs were slightly older (45.3% were ≥40 years old vs 36.2% in the control group), and there were fewer nurses compared with other healthcare cadres (47.2% vs 63.8% in the control group). In Tanzania, the intervention group had slightly fewer young HWs (26.8% were <30 years old vs 45.0% in the control group), more midwives (43.9% vs 30%) and fewer physicians (17.1% vs 27.5%). There were no differences between the intervention and control groups in Ghana, but compared with other countries, HWs were younger, more experienced, and the majority were midwives (65.9%).

The observations across all countries had similar baseline characteristics regarding the type of facility, day of the week, time of arrival (day/night) and woman’s age. A baseline difference was identified in Tanzania, where there were slightly more observations performed in hospitals than in health centres in the intervention group compared with the control group (72.3% vs 58.2%).

RMC measures

Table 2 describes the relative frequencies of each RMC measure for each group (intervention and control), time of evaluation (preintervention and postintervention) and country. Table 3 reports DiD estimates for each RMC measure and country.

Table 2

Frequency of RMC measures by group, time of evaluation and country

Table 3

Difference-in-difference estimates for each RMC measure by source of information and country

Bangladesh

In the intervention group, women reported a statistically significant reduction in physical abuse (DiD −5.2; 95% CI −9.0 to –1.4). In addition, although statistically non-significant, women reported a reduction in verbal abuse (DiD −8.2; 95% CI −16.6 to 0.09). The proportion of women reporting verbal abuse decreased from 8.7% at baseline to 0.9% at endline in the intervention group, while no changes were observed in the control group. Conversely, statistically non-significant improvements were observed in the control group for outcomes that had lower performance at baseline compared with the intervention group. These outcomes were friendly communication (DiD −11.4; 95% CI −48.5 to 26.5) and HW informing the woman of the findings (DiD −9.8; 95% CI −69.9 to 50.1). Other RMC measures showed similar results, as both groups showed enhancements (table 2). Across RMC measures, changes ranged from 8.7% to 22.9% in the intervention group, and 3.5% to 29.8% in the control group. Measures with modest changes typically had very high baseline rates. An exception was the measure ‘satisfaction with general hygiene’, which, despite low satisfaction rates at baseline (10.3% in the intervention group and 8.3% in the control group), had minimal improvements reported (15.3% in the intervention group and 16.6% in the control group).

A statistically significant improvement was shown in observer-reported communication of the delivery plan (DiD −33.8; 95% CI −62.9 to −4.6) in the control group. The proportion of HWs communicating the delivery plan decreased in the intervention group from 86.3% at baseline to 61.8% at endline, while in the control group, the proportion increased from 60.4% to 69.6%. On the other hand, observers reported statistically non-significant improvements in the intervention facilities in these measures: courteous communication between HWs and women (DiD 16.6; 95% CI −49.6 to 82.7), ensuring privacy during labour (DiD 17.7; 95% CI −28.9 to 64.3) and HWs informing findings to women (DiD 10.6; −33.7 to 60.8).

Ghana

Women in Ghana reported statistically significant improvements associated with the intervention across various measures of RMC. These improvements included HWs enabling women to ask questions and express concerns (DID 37.5; 95% CI 5.9 to 69.0), considering women’s concerns (DiD 14.9; 95% CI 4.9 to 24.9), being responsive when women asked for support (DiD 5.5; 95% CI 0.7 to 10.3), ensuring privacy (DiD 8.0; 95% CI 0.6 to 16.0), treating women with respect (DiD 5.5; 95% CI 1.0 to 9.9) and reducing verbal (DiD −8.0; 95% CI −12.1 to –3.8) and physical abuse (DiD −5.2; 95% CI −11.4 to –0.9). In addition, more women in the intervention group would select the current facility for another delivery (DiD 17.5; 95% CI 5.5 to 29.4) and they were satisfied with the attitude of health personnel (DiD 9.0; 95% CI 0.9 to 18.5). Women in the intervention group also reported enhanced satisfaction with overall hygiene (DiD 22.5; 95% CI −5.7 to 50.6), although this change was not statistically significant.

Two observer-reported measures showed further but still statistically non-significant improvements associated with the interventions: HWs informing women about the delivery plan (DiD 26.5; 95% CI −40.9 to 93.8) and ensuing privacy during initial examination (DiD 9.5; 95% CI −43.3 to 62.4). Other measures did not show differences, as changes were observed in both groups, including observer-reported privacy, which had notably low rates at baseline.

Tanzania

The intervention was associated with statistically significant improvements in women-reported privacy (DiD 24.2; 95% CI 0.2 to 48.3). Women reported other statistically non-significant improvements with the intervention: friendly communication (DiD 20.2; 95% CI −16.3 to 56.6), HW enabling questions and conerns (DiD 14.6; 95% CI −27.1 to 56.4), willingness to return for another delivery (DiD 18.8; 95% CI −21.9 to 59.5) and high satisfaction with care (DiD 17.0; −12.8 to 46.7). Additionally, the proportion of verbal abuse decreased in the intervention group from 7.1% at baseline to 2.3% at endline. In the control group, it decreased from 2.3% to 1.0%. Other measures did not show differences, mainly because rates improved in both groups or were already high at baseline.

No differences between groups were shown in most observer-reported RMC measures due to improvements in both groups or maintenance of high baseline rates. A statistically non-significant improvement was seen in observer-report privacy during the initial examination (DiD 12.4; −16.2 to 41.0) in the intervention group. Conversely, in the control group, positive trends were noted in observer assessments of HW informing women about the delivery plan (DiD -21.0 to −57.5,14.6) and providing support when women were in pain (DiD −18.2; −65.1 to 28.8).

Environmental factors potentially enabling or acting as barriers to RMC

In Bangladesh, similar trends with input measures were observed in both groups. The labour ward layout improved in both groups. Women’s satisfaction with labour ward toilet cleanliness showed minimal change, despite very low baseline rates (table 4). On the other hand, substantial and statistically significant improvements were reported by HWs in RMC training (DiD 59.0; 95% CI 33.4 to 84.6) and the availability of policy/procedures for both addressing patients’ concerns (DiD 46.0; 95% CI 4.7 to 87.4) and identifying/reporting abuse (DiD 45.9; 95% CI 19.9 to 71.8) (table 4).

Table 4

Environmental factors potentially enabling or acting as barriers to RMC by group, time of evaluation and country

In Ghana, there was a statistically significant increase in the mean number of staff on duty—mostly midwives (DiD 1.39; 95% CI 0.3 to 3.3)—in the intervention group, and a reduction of open-layout labour wards (DiD −56.0; 95% CI −112.5 to 0.4). At the same time, there was an increase in the intervention group in the proportion of observations, in which all labour ward beds were occupied (DiD 32.6; 95% CI 9.9 to 64.3). In addition, HWs were more likely to report improvements in the availability of both policies/procedures for addressing patient concerns (DiD 19.6; 95% CI −7.6 to 46.7) and clear policies on patients’ rights (DiD 19.2; 95% CI −4.5 to 42.8). There was no difference between groups in the availability of procedures for identifying and reporting abuse or training in RMC.

In Tanzania, the results suggest a favourable trend in the cleanliness of labour toilets in intervention facilities (DiD 21.0; 95% CI −18.1 to 60.0) and an increase in the occupancy of labour wards (DiD 22.0; 95% CI −46.4 to 90.4). No changes were observed in labour ward layouts and staff availability. Although statistically non-significant, an improvement was reported in intervention facilities in the availability of a process for identifying and reporting abuse (DiD 22.6; 95% CI −20.0 to 65.1). Conversely, greater improvements were reported in control facilities in the availability of clear policies on patient rights (DiD −13.6; −60.4 to 3.1).

Finally, we measured the impact of introducing the QoC standards on HWs’ self-assessed provision of RMC (online supplemental table S3). Tanzania was the only site where HWs in intervention facilities reported a statistically significant improvement over time on self-assessed provision of RMC (DiD 0.84; 95% CI 0.21,1.47).

Discussion

Implementing MNH quality standards for 18 months under real-world health system conditions was associated with some improvements in RMC measures. There was a larger trend in reducing physical and verbal abuse, enhancing privacy and increasing women’s satisfaction in the intervention facilities. However, other measures, such as effective communication and emotional support, showed no difference or varied substantially across countries. In most cases, where no differences were detected, both groups either improved or maintained high rates throughout the study. The availability of policies/procedures addressing patient concerns, patient rights and identifying/reporting abuse increased more in intervention facilities than in control facilities. Nevertheless, the results related to cleanliness, the availability of human resources and their training varied substantially across countries. Only in Ghana did the implementation of the MNH standards show consistent improvements over time across most domains and measures. Overall, women more often reported improvements in RMC than external observers.

Most of the body of evidence has focused on measuring RMC adoption gaps and validating RMC measurement methods.9 Non-comparative studies consistently suggest improvements in RMC with implemented interventions.15 31–33 However, we identified that RMC measures in intervention and control facilities tend to improve. This may be because the need for RMC is already clearly identified and puts pressure on the entire health system towards improvement or due to contamination, given the nature of the multilevel intervention.34 35 Either way, uncontrolled pre–post intervention studies fail to identify whether the improvements are associated with the intervention or secular trends. Very few comparative studies measured the effect of interventions to improve RMC and showed divergent results.11 17 36 Some authors attribute this to the multiple and complex challenges of implementing change in a low-resource setting and variations in measurement.9 17 The study that obtained similar results to ours is the one that implemented a complex strategy, which was designed in a participatory manner with multiple stakeholders and levels of leadership, specific to the context and supported structural improvements.11 Contrary to the most commonly used implementation strategy (the training of HWs), complex, multilevel, context-specific implementation strategies addressing a broad spectrum of barriers (including contextual factors) may be more effective in accelerating the adoption of RMC.16 37

Consistent with the literature, women reported more significant improvements in care experience compared with observers.38 39 This may suggest that women may have lower expectations of care experiences than observers or may indicate a possible social desirability bias. These findings further highlight the importance of the discussions around which source to use in measuring RMC objectively, considering potential biases and the cost of the different data collection approaches.

Some results are of particular interest. Intervention facilities in Ghana showed improvements of greater magnitude compared with other countries. The literature reports higher adoption of RMC in midwifery-led care services and lower workload.40 41 Given that the facilities in Ghana had the highest proportion of midwives and young personnel and that facilities in the intervention group significantly increased the available staff, the question arises as to whether the intervention could have better penetration among young midwives or new personnel. There remains a critical gap in women’s satisfaction with hygiene in Bangladesh and Tanzania. Knowing that women reported high satisfaction levels for other dimensions, these proportions may indicate that facility hygiene is a priority for women and was not addressed in some countries. Finally, emotional support did not improve because the adoption was already high at baseline in Tanzania, but other factors could have played a role in Bangladesh.

This study had several key strengths. It was a prospective multicountry comparative study conducted in 43 health facilities in three countries across two world regions. The intervention was implemented at a large scale (multiple districts) with the participation of numerous key stakeholders and substantial local input and leadership. The data collection involved mixed methods integrating the views of women, HWs and independent observers. It was a study conducted in real-life routine practice conditions, including diverse healthcare facilities, facilitating the generalisability or applicability of the findings to many similar settings around the world.42 It implemented the standards of care as a package using various standardised methods described in globally available guidelines that could facilitate their reproducibility. The inclusion of trained independent external observers was another strength.

Our study had some limitations. While a controlled before and after analysis is more robust than non-comparative studies, it does have limitations in controlling potential confounding.43 These effects were minimised by estimating DiD, which assumes that baseline rates are different and compares measure changes over time between groups to obtain an appropriate counterfactual for estimating a causal effect. Nevertheless, since the outcomes of interest are not typically collected in routine health information systems and collecting baseline primary data at multiple points in time was not feasible, we were unable to determine differences in pretreatment trends. Nonetheless, our analysis compared facilities and participants with many similar characteristics, for which parallel trends seem plausible. A ceiling effect was observed for some measures, benefiting the group with lower rates. Second, the country-specific number of clusters may not have been enough to detect some clinically significant point estimates as statistically significant. In addition, a few large facilities may have contributed most of the samples and had most of the improvements. Third, some high RMC rates could have been biased by the Hawthorne effect, which may have overestimated frequencies. We mitigated this with 2 weeks of continuous observation, which we considered sufficient time for participants to return to normal behaviours.

Conclusion

This study provides evidence that implementing MNH QoC standards could accelerate the improvement of some RMC measures in LMICs. Participatory designs are likely to encourage engagement, ownership, and capacity share at multiple levels, potentially driving systems strengthening to achieving universal health coverage with quality and respectful care. Using context-specific solutions may contribute to advancing RMC, provided there is adequate investment and support. The results suggest that a scale-up of implementing MNH QoC standards in LMICs and accelerating women’s access to RMC is feasible and must be a desired goal.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

The study protocol received approval from institutional review boards in the respective countries (ICDDRB in Bangladesh-PR-16024-1/6/2016, Ghana Health Service NHRCIRB226-6/4/2016, NIMR Tanzania-NIMR/HQ/R.8a/Vol. IX/2176-11/4/2016). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

This work was conducted under the UNICEF–BMGF MNCH partnership. We appreciate the contributions from the management and staff of our collaborating institutions at the International Centre for Diarrhoeal Disease Research, Bangladesh, Navrongo Research Centre of the Ghana Health Service, National Institute of Medical Research, Tanzania and the UNICEF country offices in the three countries. We acknowledge the valuable contribution of multiple stakeholders, including global/international experts and professionals in maternal and newborn health, ministries of health and partners in Bangladesh, Ghana and Tanzania. Thanks to Dr. Patience Afulani who provided feedback during the publication development process. We also thank the women and health workers who participated in this study.

[ad_2]

Source link