Assessing the impact of complex health systems strengthening program on maternal health care utilization in fragile and conflict-affected states: evidence from the Democratic Republic of the Congo

doi:10.21203/rs.3.rs-4688043/v1

Download PDF

Research Article

Assessing the impact of complex health systems strengthening program on maternal health care utilization in fragile and conflict-affected states: evidence from the Democratic Republic of the Congo

https://doi.org/10.21203/rs.3.rs-4688043/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

The maternal mortality ratio in the Democratic Republic of the Congo (DRC) remains among the highest in the world. However, to date, only limited evidence is available on the impact of maternal health care interventions on service quality and utilization in the DRC (11). The purpose of the study was to investigate the impact of a large, complex health systems strengthening program on maternal health care utilization in DRC.

Methods

The study used difference-in-differences models based on data collected from population-based household surveys at baseline (2014) and endline (2017) in treatment and matched comparison areas in three provinces. The dependent variables consist of the use of antenatal care (ANC) and facility-based delivery. Qualitative analyses were also incorporated in the study to explore perceptions of maternal health care offered and strategies to improve care.

Results

The health systems strengthening program was found to have had a positive impact on the probability of a woman delivering in a health care facility. There was not a significant impact on the receipt of at least four ANC consultations. The qualitative results indicated that uptake of ANC consultations in program areas was challenged by distance to facilities and the time involved in ANC, limited community outreach, and cultural norms.

Conclusions

The research points to the sample population’s high reliance on health centers and posts that provide basic maternal services and low accessibility to hospitals offering lifesaving comprehensive emergency obstetric care. The findings also highlight the need for formative research to inform interventions so that they better match social norms and address contextual needs.

maternal health care

Democratic Republic of the Congo

impact evaluation

fragile states

Although recent global estimates suggest that there has been substantial progress over the past two decades in reducing maternal mortality, the maternal mortality ratio (MMR) in low-income countries remains high, particularly in fragile and conflict-affected states (FCS) in sub-Saharan Africa (SSA), where MMR estimates in 2017 ranged from 401 per 100,000 live births (Ethiopia) to 1,150 per 100,000 live births (South Sudan) (1).

To accelerate progress, quality improvement strategies in combination with health systems innovations and behavioral change communication strategies have been implemented in FCS in SSA. These strategies have several objectives, including to: improve access to quality antenatal care; increase rates of institutional deliveries for better access to intrapartum care, including delivery assisted by a trained birth attendant, and postpartum care; and enhance the structural capacity of facilities via the increased availability of lifesaving drugs and improved facility conditions, including the availability of emergency obstetric care (2)

However, barriers to effectively scale up maternal health interventions are particularly challenging in FCS due to the disruptive impact of conflict and political instability on health care delivery systems (2). These effects occur through several mechanisms, including destruction of and breakdowns in health care infrastructure, disruption of supply chains, problems in recruiting and retaining qualified health workers, and interference in financial and management systems, all of which hinder service quality (1)(3). Moreover, several demand-side challenges also play a role in constraining access to services, including the effects of crises on household incomes and the ability to pay for services, travel barriers due to security concerns, and loss of female autonomy associated with the normalization of violence against women (4).

A case in point is the Democratic Republic of the Congo (DRC), a country that has experienced chronic conflict and political instability in recent decades (5). Conflict in the DRC has been pronounced since the Rwanda genocide in 1994 destabilized eastern and central Africa. Conflict involving armed groups and militias operating in the eastern part of the country and long-standing political tensions at the Rwanda border continue to this day. Ethnic conflict as well as banditry, particularly near mining operations, also plague the country(6)(7). Conflict has had devastating impacts on the country’s economy and development, and poverty is widespread and extreme. In 2020, gross national income per capita was USD 497 and about 64% of Congolese lived below the poverty line, with individuals living in rural areas more likely to be impoverished than individuals in urban areas (World Data Atlas, 2020). These factors, along with poor governance, have contributed to underinvestment in the health system, problems in the recruitment and distribution of health workers, and disruptions in the procurement and distribution of medicines, supplies, and equipment.

In recent decades, progress has been made in reducing maternal mortality despite the country’s fragility and instability. However, the MMR remains among the highest in the world. In 2017, the MMR was estimated to be 473 per 100,000 live births (8)(1). Paradoxically, high maternal mortality persists despite the DRC having a relatively high rate of institutional deliveries (around 80 percent according to 2013-14 Demographic and Health Survey (DHS) and the 2017-18 Multiple Indicators Cluster Survey) (9)(10). This mismatch between high rates of institutional delivery and the high burden of maternal mortality in the DRC indicates that there is a critical gap in service quality, as many maternal deaths are a result of preventable complications during pregnancy and childbirth. If services are delayed, inadequate, unnecessary, or harmful, then there are missed opportunities to prevent maternal mortality (11).

Previous research that has examined the utilization of antenatal care and skilled birth attendants in the DRC suggested that context-specific interventions are needed to make further progress on expanding access to services (4). However, to date, only limited evidence is available on the impact of maternal health care interventions in the DRC on service quality and utilization (11).

The purpose of the study is to assess the impact of a large, complex health systems strengthening program in the DRC that included a major maternal health care programmatic component.

The Accès aux Soins de Santé Primaires (ASSP) program was implemented by IMA World Health and other partners from 2012 to 2018 to improve reproductive, maternal, neonatal and child health (RMNCH) in 52 health zones in Maniema, Tshopo, Nord Ubangi, Kasai, and Kasai Central provinces with a total estimated population of nine million.

Health zones cover an average population of 110,000 and include a central health zone office and at least one general reference hospital located in more populated settings. Health zones are divided into health areas, each containing, in theory, at least one health center responsible for providing a minimum package of services as defined by the government. Larger health areas may also have health posts to supplement health care access. Private facilities are also available, but less so in remote areas. While geographical terrain varies across provinces, distances to health facilities can be long and transport is generally poor, especially in rural areas.

ASSP was a complex program, consisting of population, health, nutrition, and water and sanitation programmatic interventions, as well as health systems strengthening interventions. Maternal heath was one of the priority areas of the program, and to improve the availability and quality of maternal health services, several facility- and community-based strategies were used. These strategies aimed to:

Improve the financial sustainability of service delivery by supporting the establishment of user fee schedules that included exemptions for the poor and other vulnerable populations.
Help address cultural barriers and improve outreach by training community leaders and community health workers (CHWs) to sensitize the population about the benefits of antenatal care (ANC) and facility deliveries and identify and refer newly pregnant women to health facilities(12).
Improve the availability and quality of services by renovating 177 facilities and constructing 109 new health structures, which constituted about 10% of all health area facilities in program-supported health zones. Newly constructed facilities were always at the same location as the facilities that were replaced, thereby not affecting the physical distance between households and the closest health facility. The program also provided equipment, medication, and training for health providers to improve the readiness to deliver the Ministry of Health (MOH) paquet complémentaires d’activités (package of complementary activities) including basic emergency obstetric and new-born care (B-EmONC) in at least four health centers and comprehensive emergency obstetric and new-born care (C-EmONC) in the reference hospital in each health zone receiving program assistance.

During the period when the program was designed, the DRC had been experiencing relatively peaceful conditions compared to earlier years. However, soon after the startup of the program, conditions deteriorated due to several factors, including 1) increased political conflict and demonstrations associated with national elections, which resulted in widespread tension and causalities, 2) an intensification of rebel attacks by militia groups in eastern DRC, and 3) persistent occurrences of fraud and corruption starting at the national level. These factors caused disruptions in a) the delivery of equipment and supplies, b) provision of capacity building activities for health managers and providers, and c) the ability of Kinshasa-based ASSP staff to supervise program-supported areas. Moreover, civil unrest that occurred in 2017 in two program provinces in central DRC, Kasai and Kasai Central, led to a months-long suspension of activities in program-supported health zones. These contextual issues likely had significant impacts on the ability to achieve the program’s expected results by affecting service availability and quality.

Theory

To assess the impacts of the program, the study drew on Andersen’s Behavioural Model of Health Care Utilization (12). According to Andersen, health care utilization is influenced by pre-disposing, enabling, and need factors. Pre-disposing factors include characteristics such as parity, sex, age, and marital status, while enabling factors include health system and organization factors, such as quality of care, price of care, wealth, education, distance, and availability of transport. Since the program was expected to improve the quality of health care and make user fees affordable, households were hypothesized to increase the utilization of maternal health care services because of the program interventions. All pregnant women were assumed to have a need for ANC and facility-based delivery assistance. Contextual factors such as political instability and conflict, transportation infrastructure, and seasonality were classified as part of the external environment.

Research design

The study used a mixed methods research design to investigate impact in three provinces: Maniema, Tshopo, and Nord Ubangi.

The evaluation used a quasi-experimental research design with constructed treatment and matched comparison groups to investigate impact in three provinces: Maniema, Tshopo, and Nord Ubangi. The primary outcome variables were the use of ANC and the use of delivery assistance in a health facility. For each outcome variable, a difference-in-differences (DID) multivariate estimation strategy was used by estimating Probit models using population-based survey data collected from women who delivered during a one-year period prior to the survey, as explained below. Each model controlled for potentially relevant confounding factors aligned with Andersen’s model.

The research design is depicted in Fig. 1. Baseline (2014) and endline (2017) household and health facility surveys were carried out in three provinces that received program support. The provinces were grouped into two sampling areas. The first sampling area (which is synonymous with a survey domain) consisted of health areas within Nord Ubangi and the second consisted of health areas within Maniema and Tshopo. For each sampling area, corresponding matched comparison groups consisting of randomly selected villages within matched health areas which did not receive the intervention package were also selected to reduce potential confounding and control for secular differences influencing the observed outcomes. Program-supported health areas in Nord Ubangi province were matched with non-program health areas in Sud Ubangi province. Program-supported health areas in Maniema and Tshopo provinces were matched with non-program health areas within each province.

To complement the population-based surveys, health facility surveys were also administered at both baseline and endline at the government health center or health post designated to serve each village selected for the household survey. The data collected included information on staffing, services available, infrastructure, equipment, availability of drugs and supplies, management and supervision, and distances to referral facilities. The survey was administered to the nurse in charge of each facility. For villages that were included in both the baseline and the endline survey, the health facility survey was administered at the same facilities in both waves.

The operational definitions of the outcome and explanatory variables used in the analysis are presented in Table 1. Due to delays in receiving the final approval of the research design from the sponsor, the baseline survey was conducted in April 2014, a year after maternal health activities had begun. Because of this issue, the outcome variables were restricted to the women’s most recent birth that occurred 12 to 24 months prior to the survey to reflect the true baseline situation prior to program interventions being introduced. The indicator of the recommended number of ANC consultations (four or more) was based on WHO’s guidelines that were in place prior to 2016, when WHO increased the number of recommended visits from four to eight. (13)

Table 1

Operational definitions of variables.
Variable	Definition
Outcomes
Four or more ANC visits	Coded 1 if woman who delivered a live birth 12 to 24 months prior to the survey received four or more ANC visits; coded 0 otherwise
ANC visit in first trimester	Coded 1 if woman who delivered a live birth 12 to 24 months prior to the survey received an ANC visit during the first trimester; coded 0 otherwise
Delivery in a health facility	Coded 1 if woman who delivered a live birth 12 to 24 months prior to the survey delivered in a health facility; coded 0 otherwise
Predisposing factors
Parity	Birth order of live birth
Age at birth	Coded 1 if woman was 20 to 34 years and 2 if women 35 years or older; coded 0 otherwise (less than 20 years)
Enabling factors
Sampling domain	Coded 1 if woman was from Nord/ Sud Ubangi; coded 0 if woman was from Maniema/ Tshopo
Peri-urban/rural	Coded 1 if woman was from a peri-urban area; coded 0 if woman was from a rural area
Distance	Number of kilometers between the household and the government health center or health post designated to serve the household’s community
Wealth	Coded 1 if woman was second to poorest wealth quintile, 2 if woman was middle wealth quintile, 3 if woman was in fourth wealth quantile, 4 if woman was in richest wealth quintile; coded 0 otherwise (poorest wealth quintile)
Educational attainment	Coded 1 if woman had some education, 2 if woman had completed primary schooling, 3 if woman had completed secondary schooling; coded 0 otherwise (no education)
Employment status	Coded 1 if woman not currently working but had worked in last 12 months, 2 if woman was currently working. Otherwise coded as 0 (not working in past 12 months)
World Bank PBF program	Coded 1 if health zone received cash support from program, 2 if health zone received performance-based financing payments under the World Bank program. Otherwise coded as 0 (not part of the program)
Mean cash expenditure (USD) on delivery	Average expenditure on delivery based on an exchange rate of 900 Congolese francs per dollar at baseline in 201) and 1,400 Congolese francs per dollar at endline in 2017

Pre-disposing explanatory variables included parity and age at birth, and enabling variables included sampling domain, peri-urban/rural status, household wealth, educational attainment, and employment status. Geo-coded data on the latitude and longitude of households and their designated public health facilities was used to calculate a proxy indicator of travel distances. In addition, to control for the influence of a World Bank-sponsored performance-based financing (PBF) project that was introduced in several matched comparison areas in 2016, the models also included a categorical indicator for the two modes of World Bank support: 1) cash support only or 2) the full PBF intervention, which involved contracts with government and faith-based facilities, health zone offices, and provincial health offices to provide heath or health management services in exchange for a cash payment that could be used to cover staff bonuses, operational costs, and small investments (14).

The 2014 baseline survey was also administered in the two remaining provinces that received program support, Kasai and Kasai Central. Unfortunately, civil unrest prevented revisiting these provinces for the endline survey, and as a result, the impact evaluation was carried out in the two remaining sampling areas (Nord/Sud Ubangi and Maniema/Tshopo). Figure 2 presents a map of the intervention and matched comparison areas. To compensate for the loss of households in Kasai and Kasai Central, additional households in both intervention and matched comparison areas were selected to preserve the target sample size (4,200 households at endline).

To complement the quantitative findings, we collected qualitative data from government health officials, program implementing partners, and beneficiaries to explore perceptions of program implementation, understand why selected strategies did or did not work, and identify perceived differences between planned and actual implementation and why these discrepancies occurred. Key informant and in-depth interviews were carried out in Kinshasa and the provinces of Nord Ubangi and Maniema. In each province, one higher performing health zone (where facility-based deliveries had increased during the ASSP program period) and one lower performing health zone (where facility-based deliveries had plateaued or decreased during the ASSP program period) was selected.

Key informants included MOH representatives at the national, provincial, and zonal levels, implementing partners at the national and provincial level, and representatives of the donor agency in Kinshasa. Purposive sampling was used based on expertise in maternal health and involvement in program activities, with an initial target to carry out 18 key informant interviews. Key informant interviews examined implementation of interventions since program inception, perceptions of activities, barriers to achieving outcome indicators, and recommendations to improve interventions.

In-depth interviews involved women who had delivered in a health facility or at home within the past six months, facility-based health providers, and CHWs. Based on facility records, we used purposive sampling to select women who had delivered in a facility, while women who delivered at home were identified through CHW records. The preliminary goal was to interview 18 women who had recently given birth. Topics examined included knowledge of ANC and maternity care, household decision-making regarding maternity care, timing of ANC and why visits were made, barriers and motivators to delivering in a facility or at home, perceptions of the quality of care, and the cost of maternity care.

Purposive sampling was used to select at least one facility-based nurse or midwife providing maternal health care in each sampled health zone, while random sampling was employed to identify 2–3 CHWs in each health zone. We examined program activities related to equipment and supplies distributed, training, workload, salary support, motivation, referral systems during complicated deliveries, and supervision, as well as perceptions of the program.

Further details on the study methods are described in the final evaluation report (15).

Quantitative results

The final analytic sample consisted of 564 and 873 women at baseline and endline, respectively. There were no significant baseline differences between the intervention and matched comparison groups in the observed individual- and household-level characteristics (Table 2).

Table 2

Comparison of individual- and household-level characteristics between intervention and matched comparison groups, 2014 baseline.
Variable	Intervention (n = 261)		Matched Comparison (n = 303)		Difference
	N	%	N	%	%	p-value
Mother
Age-group at birth
Less than 20	36	10.89	51	15.81	4.92	0.213
20–34	174	70.50	197	66.47	-4.03	0.529
35–49	51	18.61	55	17.72	-0.89	0.853
Education level
No education	83	29.78	95	30.24	0.46	0.947
Some primary	85	30.92	94	29.33	-1.59	0.793
Completed primary	87	36.99	109	39.36	2.37	0.739
Completed secondary	6	2.31	4	1.06	-1.25	0.445
Employment status
Not worked in last 12 months	22	6.37	37	11.65	5.28	0.202
Not currently working but worked in last 12 months	10	4.62	4	0.98	-3.64	0.023
Currently working	229	89.01	260	87.37	-1.64	0.757
Birth order
1	51	18.07	53	15.96	-2.11	0.580
2–3	89	30.04	105	34.02	3.98	0.520
4–5	66	27.43	66	22.79	-4.64	0.313
6+	55	24.46	79	27.23	2.77	0.597
Household
Sampling domain
Nord/Sud Ubangi	129	35.19	167	75.03	39.84	0.001
Maniema/Tshopo	132	64.81	136	24.97	-39.84
Setting
Peri-urban	36	9.98	21	2.76	-7.22	0.041
Rural	225	90.02	282	97.24	7.22
Distance (mean, SD)	39.97 (13.45)		36.14 (46.77)		-3.83	0.275
Wealth quintile
Low	29	10.24	22	7.31	-2.93	0.416
Low middle	61	19.66	77	24.16	4.50	0.458
Middle	62	26.44	67	25.52	-0.92	0.908
High middle	53	20.71	73	25.47	4.76	0.423
High	52	22.95	64	17.54	-5.41	0.403

Note: The results are weighted. The variable “education level” is missing one case, the variable “employment status” is missing two cases and the variable “wealth quintile” is missing four cases.

For ANC, the percentage of women who received four or more visits at baseline was found to be lower in the intervention areas than in the matched comparison areas (33.2 percent vs. 51.6 percent, p = 0.027) (Table 3). The percentage of women who reported an ANC visit during the first trimester was higher in the intervention areas than in the matched comparison areas (15.7 percent vs. 8.1 percent) but the difference was not statistically significant (p = 0.157).

Table 3

Percent of women delivering a live birth 12 to 24 months prior to the survey who received ANC services, delivered in a health care facility, and had a C-section, 2014 baseline and 2017 endline.
Variable	Intervention group				Matched comparison group
Variable	2014	2017	Absolute difference	p-value	2014	2017	Absolute difference	p-value
	%	%			%	%
Four or more ANC visits
Total	33.18	27.37	-5.81	0.291	51.60	39.41	-12.19	0.162
Nord/ Sud Ubangi	36.77	32.04	-4.73	0.451	49.11	40.73	-8.38	0.367
Maniema/ Tshopo	31.29	22.16	-9.13	0.215	59.92	36.02	-23.9	0.206
Overall N	222	398			231	428
ANC visit in first trimester
Total	15.70	12.74	-2.96	0.461	8.10	12.73	4.63	0.247
Nord/ Sud Ubangi	6.24	12.70	6.46	0.117	3.83	9.87	6.04	0.102
Maniema/ Tshopo	22.12	12.79	-9.33	0.170	22.56	24.10	1.54	0.811
Overall N	182	331			201	332
Delivery in a health care facility
Total	49.28	74.61	45.33	0.004	70.81	78.91	8.10	0.155
Nord/ Sud Ubangi	28.79	63.10	34.31	< 0.001	70.50	77.02	6.52	0.379
Maniema/ Tshopo	60.63	87.50	26.87	0.020	71.84	83.94	12.10	0.003
Overall N	255	418			299	447
C-sections
Total	2.39	4.51	2.12	0.269	0.79	2.78	2.00	0.207
Nord/ Sud Ubangi	0.00	2.09	2.09	0.177	0.00	1.17	1.17	0.257
Maniema/ Tshopo	3.68	7.24	3.57	0.254	3.48	7.05	3.56	0.528
Overall N	224	404			238	435

Note: p-values in bold font are statistically significant at the 0.05 level. The results are weighted.

The percentage of women who delivered in a health facility at baseline was found to be lower in the intervention group than in the matched comparison group (49.3 percent vs. 70.8 percent, p = 0.022). However, the figure for the intervention group masks a large difference between the two sampling areas, with Maniema/Tshopo having had a much higher percent of women who delivered in a health facility than Nord/Sud Ubangi (60.6 percent vs. 28.8 percent, p = 0.0288).

Over the study period, no significant changes were found in the percentage of women who received four or more ANC visits in either the intervention or the matched comparison areas (Table 3). The results for ANC use during the first trimester suggested an increase in Nord/Sud Ubangi, from 6.2 percent in 2012 to 12.7 percent in 2016 (p = 0.117), and a decrease in Maniema/Tshopo, from 22.1 percent in 2012 to 12.8 percent in 2016 (p = 0.170), although neither change was statistically significant.

For delivery assistance, the percentage of women who had a facility-based delivery significantly increased over the study period in the intervention areas (p < 0.01) only, from 49.3 percent to 74.6 percent (Table 3). The increases were statistically significant in the intervention areas for both Nord/Sud Ubangi (p < 0.001) and Maniema/Tshopo (p = 0.02). Overall, the percentage of women who reported having had a C-section was low (i.e., 4.5 percent in intervention areas and 2.8 percent in matched comparison areas at endline). However, for the overall sample, there were no significant differences in reported C-sections over the study period in either the intervention or the matched comparison areas.

As shown in Table 4, the increase in the percentage of women who delivered in facilities was due to increased uptake in the use of public hospitals and health centers. Among women in the intervention group, there was a 25.9 percentage point increase in the use of public facilities (statistically significant, p < 0.01). In the matched comparison group, the 8.8 percentage point increase was not statistically significant (p = 0.112).

Table 4

Percent distribution of women by place of most recent birth and type of attendant at delivery for the intervention and matched comparison groups, 2014 baseline and 2017 endline.
	Intervention group						Matched comparison group
Variable	2014		2017		Absolute difference		2014		2017		Absolute difference
	(N = 261)		(N = 424)				(N = 303)		(N = 449)
	N	%	N	%	%	p-value	N	%	N	%	%	p-value
Facility type ^ϯ
Public facility	132	48.50	319	74.36	25.86	0.004	199	69.70	362	78.49	8.79	0.112
Public hospital	0	0.00	44	11.34	11.34	0.000	6	6.88	19	6.26	-0.62	0.904
Public health center/post	132	48.50	275	63.02	14.52	0.127	193	62.81	343	72.24	9.43	0.130
Private facility	2	0.78	3	0.25	-0.53	0.211	2	1.12	1	0.42	-0.70	0.452
Other (incl. home)	121	50.72	96	25.39	-25.33	0.004	98	29.19	84	21.09	-8.10	0.155
Trained delivery attendant at delivery^ǂ
No	92	40.16	83	22.42	-17.74	0.065	61	25.33	62	11.77	-13.56	0.004
Yes	166	59.84	336	77.58			238	74.67	385	88.23

^ϯ Public hospitals includes general reference hospitals and secondary hospitals. Public health centers/posts include reference health centers, health centers, health posts, maternities. Private facilities include private hospitals/clinics, university clinics, specialty clinics, pharmacies, private medical practices, emergency nurses, and other private provider. Other includes deliveries that occurred in the woman’s residence, other residences, and other responses.

¥ Other includes village heads, traditional healers, village moms, and other responses (e.g., in-law, parents; husband).

Note: p-values in bold font are statistically significant at the 0.05 level.

Note

The results are weighted.

In both the intervention areas and the matched comparison areas, public health centers and posts were more frequently utilized for delivery assistance than public hospitals at both baseline and endline. For example, the results from the endline survey indicate that 11.3 percent of women in the intervention group and 6.3 percent in the matched comparison group delivered in a public hospital, compared to 63.0 percent and 72.2 percent, respectively, that delivered in a public health center or post.

To explore possible reasons for the relatively low reliance on hospitals, we also investigated the distances between communities and the closest referral hospital using self-reported data from respondents to the endline health center survey. The median distance between the closest health center and the hospital to which they usually referred patients was 18 kilometers in the intervention areas (with the maximum distance of 175 kilometers) and 25 kilometers in the matched comparison areas (with a maximum distance of 150 kilometers).

We also found that the percent of women who delivered in a public hospital were more likely to report a C-section than the percent of women who delivered in health center/post. For example, 19.0 percent of women delivering in a public hospital in intervention areas at endline reported a C-section, compared to 3.6 percent delivering in a health center/post (results not reported). In matched comparison areas, 27.3 percent of women delivering in a public hospital in intervention areas at endline reported a C-section, compared to 1.4 percent delivering in a health center/post.

Because the evaluated program was complex, we also explored how the program influenced intermediate outcomes that would be expected to affect service provision and utilization, such as the availability of equipment and supplies, based on data from the health facility surveys, as well as perceptions of various aspects of service quality based on data from the household survey. These results are not presented in this article, but can be found in the evaluation report (15). While intervention areas experienced increases between baseline and endline in the percentage of facilities with adequate equipment – defined as having all basic equipment stipulated as essential by the government (thermometers, stethoscopes, blood pressure cuffs, adult scales, pediatric scales, and timers) – the increases were not statistically significant, and improvements were also not observed in matched comparison areas. Intervention health facilities experienced improvements in drug supplies, but these changes were not found to be statistically significant. The availability of drugs was found to be low, with 45 percent of facilities or less having had all three tracer drugs in stock on the day of the survey.

The analysis of data collected through the household survey on the perceptions of sample women who delivered a live birth on various aspects of service quality present a different picture. The program was found to have had a positive and significant effect on perceptions of adequate cleanliness of the health facility (marginal effect: 12.21; p = 0.018). However, there were no statistically significant program impacts on patients’ perceptions of adequate availability of equipment, competence of the person assisting delivery, availability of drugs, or the way in which the health care provider explained the situation.

The program also introduced improved user fee guidelines that incorporate exemptions for the extreme poor and other vulnerable women, which could potentially influence service utilization. The analysis of out-of-pocket expenditure associated with facility-based delivery assistance during the past two years in constant US dollars (USD) indicated that in the intervention areas, the mean expenditure for delivery decreased over the study period in public health centers and health posts, from $7.9 at baseline to $6.3 at endline, but was not statistically significant. Changes in out of pocket expenditures in public hospitals and private health care facilities were also insignificant (15).

Table 5 reports the Probit model results on the impact of the program on maternal health care utilization. The marginal effect of the program was found to be positive and statistically significant for facility-based delivery (marginal effect = 19.5, p = 0.001). In the models based on the full sample and the sample from Nord/Sud Ubangi and Maniema/Tshopo, the program was not found to have had a statistically significant impact on either use of four or ANC visits or use of ANC care during the first trimester. All model estimates controlled for selected observed individual-, household-, and community-level characteristics, including the presence of the World Bank-support PBF program in matched comparison areas.

Table 5

Marginal effects of the program on antenatal care and delivery in a health care facility for the *most recent birth among* women who delivered 12 to 24 months prior to the survey.
Variable	N	Marginal effect	p-value
Four or more ANC visits
Total	1,272	-11.07	0.097
Nord/ Sud Ubangi	674	11.65	0.194
Maniema/ Tshopo	591	-11.57	0.154
ANC visit in first trimester
Total	1,040	-0.53	0.917
Nord/ Sud Ubangi	585	-0.22	0.971
Maniema/ Tshopo	449	-9.34	0.189
Delivery in a health care facility
Total	1,412	19.52	0.001
Nord/ Sud Ubangi	753	17.19	0.027
Maniema/ Tshopo	649	5.90	0.930

Note: p-values in bold font are statistically significant at the 0.05 level

Qualitative results

We interviewed 11 women who delivered in a facility and seven women who delivered at home within the past six months, seven facility-based health workers, and 10 CHWs.

Use of ANC services

Most women who did not attend four or more ANC visits mentioned distance and difficult terrain on route to the health facility, extensive wait time, and inadequate seating while waiting for ANC to start as barriers to participation. This informant said,

I preferred to start ANC at six months of pregnancy to avoid walking too much, because if you start ANC at three months and go into labor at ten months, you have to walk a lot. I started at six months to reduce going back and forth. (Woman from Nord Ubangi)

Some women mentioned that it takes an entire day to obtain facility-based ANC, preventing them from performing daily farming and household activities including childcare.

Most women from Maniema reported being visited by a CHW during pregnancy, while in North Ubangi, few women received a household visit. CHWs emphasized that it is difficult to conduct household visits during the rainy season when many families relocate to be near agricultural fields. In addition, CHWs reported that they are not remunerated for community activities but only receive per diem during training and special events such as vaccination campaigns. Data collected from facility-based health workers and CHWs revealed high attrition of CHWs and fewer active CHWs than planned by the program. Both active and former CHWs were predominantly male.

Women informants indicated that talking about the pregnancy before the pregnancy becomes visible can bring bad luck and jeopardize the pregnancy and considered it inappropriate to share information about a pregnancy with men. This informant stated,

When the community worker came to my home at the beginning of the pregnancy, he asked me if I was pregnant and when I was planning to go to ANC. I had difficulty giving an answer because, first, I was ashamed to tell a man that I am pregnant, and even more importantly, the pregnancy was not yet visible, and I wondered how he could ask me a question like that. It was my older sister who answered. It is not easy to respond to a man about pregnancy. (Woman from Nord Ubangi)

Other reasons women gave for not attending ANC during the first trimester included that services are the same even if they start later in the pregnancy and difficulties traveling long distances during the first trimester. The local belief that the fetus starts transforming into a human being later in the pregnancy may also influence the perceived need for ANC early on.

Women who attended four or more ANC visits reported that they followed health worker recommendations and valued the knowledge gained during ANC sessions, highlighting information on the evolution of the pregnancy, medical signs and symptoms that could signify risks of complications during childbirth, and the health of their baby. These informants appreciated the different components of ANC and the health workers administering the services, whom they described as well-qualified, courteous, and interested in the health of pregnant women. These women also mentioned that they live in proximity to the health center, ANC service fees are affordable, and they can decide on their own whether and when to attend sessions.

Women who delivered in facilities reported choosing a facility delivery because the birth attendants are skilled and respectful, childbirth is safer, they had had a positive previous facility-based delivery, health workers recommend facility deliveries during ANC visits, and the newly constructed or renovated maternity ensures privacy and a clean environment during childbirth. An informant said,

The birth attendant always assists my deliveries without any problems. Out of my four children, I have given birth in this health center since the delivery of my second child. I delivered my first child in the general hospital, but after discovering this woman (the HC birth attendant), I have never been anywhere else. She knows what she is doing, she does not insult women during childbirth, she speaks to us with respect. She is a really good woman. (Woman from Maniema)

Some informants mentioned that facility delivery fees had been reduced to an affordable price and payment is provided by their husbands, while for home deliveries gifts in-kind such as chickens are given, which is often the woman’s responsibility. This woman stated,

I prefer to go straight to the health center. It’s cheaper. When I gave birth, I only paid a thousand francs (.72 USD). This is what prompted me to go to the health center to give birth. When I paid the thousand francs, I started to feel sick, I was suffering from malaria, and I was given treatment for malaria. It was included as part of the thousand francs I had paid for the delivery. (Woman from Nord Ubangi)

Informants also stated that police demand large sums of money (up to 100,000 Congolese francs (CF), or 71 USD) from women who deliver at home. Nord Ubangi informants reported that the availability of female birth attendants 24-hours daily guided their decision to deliver in a facility.

In a peri-urban site in Maniema, we were unable to identify women who had delivered at home. In the other three sites, which were all in rural settings, most women who did not have a facility delivery had planned to give birth in a health structure. These women were working in the fields when contractions started or experienced contractions quickly and could not reach the health facility, faced long delays finding transport, or delivered on the way to the health structure. All these women received assistance from a health facility birth attendant or nurse during or shortly after the birth. The few women who chose not to deliver in the facility were not aware of the reduced fees, could not afford transport costs to the facility, or reported that the reduced delivery fees were not enforced by health care providers. Failure to honor reduced delivery fees was reported by the majority of informants, although the increase was generally only slightly higher than the official fee of 1,000 CF (0.72 USD). In Nord Ubangi, women also expressed concerns about delivering with a male attendant, not having suitable clothes to wear when returning home post-delivery, and the disposal of the placenta, which in the health facility is burned in an incinerator or buried. Informants reported that traditional tenets dictate that the placenta be buried near the household to protect the welfare and safeguard the origins of the child.

Several health worker informants indicated that the maternal health program had improved women’s desire to use maternal health services, citing more comprehensive ANC consultations and reduced delivery fees. In Nord Ubangi, health workers claimed that prior to the ASSP program ANC consultations had not been routinely available and most women gave birth at home because they did not understand the importance of delivering in the presence of a trained provider.

Health providers in both provinces reported regular medication stock outs, failure to receive supplies and equipment promised by program staff, and problems maintaining equipment provided by the program. Despite this, those supplies and equipment that ASSP provided were generally appreciated.

While all facilities had established a referral system for complicated cases, health workers consistently reported that the system exceeded the capabilities of the health center, and as a result, referrals were rarely made. Informants mentioned that transport for referral cases was the responsibility of the pregnant woman and her family, and that when made, referrals were sent to the reference hospital of the health zone.

Overall, the study results suggest that the program had mixed impacts on the use of maternal health services in the DRC, as the program was shown to have had a significant and positive impact on the propensity of delivering in a health care facility based on the results of the DiD model estimations, but not an impact on the propensity of using ANC.

The descriptive findings suggest that several enabling factors may have played a role on the use of delivery assistance. In the qualitative component, respondents in the intervention areas mentioned: their perceptions that birth attendants were well qualified, making delivery safer, improvements in health facility infrastructure, that fees for delivery assistance were considered affordable, and pressure from local authorities. The quantitative results showed increases in the percentage of women with the perception of improved availability of equipment and cleanliness in the facility, however findings from the facility surveys indicate that actual equipment availability did not change.

Except for the finding on actual equipment availability, these findings are consistent with information from the program’s routine monitoring system that show the extent to which various financing and quality improvement interventions were introduced and scaled-up during the study period. These strategies included user fee exemptions for the extremely poor and other vulnerable populations, training to increase provider skills, and improvements in facility infrastructure, equipment availability, and cleanliness. Moreover, the quantitative findings show a general trend in both the intervention and matched comparison areas towards increased facility deliveries, suggesting that in the DRC, where women are likely highly aware of the risks associated with childbirth, delivering in a health care facility is becoming a social norm, as is the case in many African countries (16). However, the MMR in the DRC is still among the highest in the world, raising concerns about the quality of care.

While improvements in birthing skills and environments, supply chains, and essential services are a central focus to maternal and new-born health in most countries, experts contend that this is only a partial answer to quality shortcomings, particularly in rural, low-volume facilities (11). Evidence suggests that women with choices of facility care are able to ascertain low quality maternity care, which is likely driving a shift in some African countries with improved transport systems for women to bypass rural facilities in favor of better equipped hospitals with more qualified staff (17)(18)(19). This is not the case in rural DRC, where health systems are weak and distances to facilities, especially hospitals, are typically long, routes can be hazardous and insecure, particularly for women, and transport poses a tremendous problem. These contextual factors may help explain the study findings on the relatively low use of hospitals offering C-EmONC and the relatively high reliance on health centers and posts where, at best, only basic obstetric services are available. The program aimed to ensure that these health structures, which are more accessible to target populations and are expected to manage the large proportion of deliveries that are uncomplicated, are sufficiently equipped and staffed to handle normal deliveries (10). However, health centers and posts are often situated in remote settings in which maintaining birth attendant skill levels and infrastructure readiness including the availability of essential drugs and equipment present major challenges, as does establishing effective referral systems and ensuring rapid access to lifesaving emergency services when unpredictable obstetric complications arise. In both the intervention and matched comparison areas, C-section rates were far below levels recommended by WHO to prevent maternal morbidity and mortality (20). The co-existence of poor-quality maternal services and lack of access to C-EmONC helps to explain the persisting high MMR in the DRC.

The descriptive analysis found that the proportion of women who received at least four ANC visits decreased in the intervention areas, from 33.2 percent to 27.4 percent. This was primarily due to drops in Maniema, from 31.3 percent at baseline to 22.2 percent baseline, as rates in Nord Ubangi were relatively stable over the study period. The use of ANC services also decreased in the matched comparison areas, from 51.6 percent to 39.4 percent.

One possible explanation for the lack of impact on ANC use concerns the presence of DFID’s predecessor health program, Access to Healthcare (ATH), which operated in Maniema/Tshopo until 2012, one year before the ASSP program was initiated, but not in Nord Ubangi. Improving the delivery of ANC services was a key aim of ATH. which offered ANC and other services free of charge and provided supplementary fixed and performance-based payments to motivate health workers to improve service provision, all of which may have contributed to higher ANC attendance. During ASSP, the number of women from Maniema/Tshopo who incurred any ANC expenses over the two-year recall period increased significantly, which may have played a role in decreased use of ANC services in that region. In contrast, changes in ANC costs between baseline and endline among women living in Nord and Sud Ubangi were minimal. In an extremely poor context such as rural DRC, fee increases may have impeded the use of ANC, a service which many women view as non-essential. Incentives offered to health workers to improve performance and free ANC services during the predecessor program may explain, at least in part, why the propensity to use ANC did not increase over the study period.

The qualitative results also shed light on potential reasons why the program was not found to have had a positive impact on ANC. While the program relied on CHWs to encourage women to attend ANC during the first trimester and uptake of four ANC visits during pregnancy, an insufficient number of trained and female CHWs limited outreach activities designed to encourage the use of ANC, particularly to women living in remote locations. In the DRC, CHWs are volunteers, with most CHWs in rural communities relying on farming for their livelihoods. Prevailing poverty and insecurity restrict farmers from adjusting time-sensitive agricultural activities, their first priority, and likely affect the ability of CHWs to fulfill responsibilities during critical farming periods. While research in fragile states highlights that CHWs require extra support to compensate for the special challenges they face carrying out their work, such assistance was not offered by the program (21). CHWs interviewed as part of the qualitative component of the study expressed dissatisfaction with their work due to limited remuneration and other forms of motivation, leading to high attrition. Our findings raise questions regarding the appropriateness of relying on volunteer CHWs to improve ANC participation, as well as the feasibility of retaining voluntary CHWs in an FCS context such as the DRC.

Women included in the qualitative study also complained about the time involved in traveling to the health centers and participating in ANC, with some claiming that participation generally consumes an entire day, requiring them to forgo livelihood and household activities. Our qualitative data suggest that the timing of the first ANC visit is affected by cultural prohibitions discouraging women from disclosing their pregnancy (especially to men) while it is not visible. These cultural norms appear to affect the ability of CHWs to effectively promote early initiation of ANC and thus uptake of participation in four ANC visits. A recent study in the DRC elucidated other factors related to violence and financial constraints that influence attendance of four ANC visits (4).

The study has several limitations. First, as mentioned in the methods section, the baseline survey was conducted in April 2014, well after maternal health activities had begun. Because the 2014 baseline estimates may not reflect the situation prior to ASSP interventions, we restricted the analysis to women who delivered 12 to 24 months prior to the baseline survey to generate cleaner estimates of program impact. However, this restriction reduced the power of the analysis. Second, due to political unrest during endline data collection, the analysis excluded the two provinces where the program had the highest level of activity, and according to the program’s routine program data, the most active uptake of some services, which may have dampened the estimates of program impact. Another limitation is that qualitative data was only collected at endline and did not include the matched comparison areas.

Despite these limitations, this study makes an important contribution to the evidence base on maternal health care in FCS. First, although fragility and conflict have received increased attention in recent years as key contextual issues that jeopardize maternal health outcomes in SSA countries (4) (8), the impact of maternal health care interventions in FCSs has been rarely explored through systematic research methods. To the best of our knowledge, this is the first study that has rigorously assessed the impact of a maternal health care program on the uptake of ANC and facility-based delivery in the DRC, a country that has been plagued by conflict for decades. Second, there have also been increased calls to “re-engineer” health systems as part of the strategy to make progress towards Universal Health Coverage targets, which include health coverage and quality goals for maternal health and other types of health services, “rather than just improving coverage for current implementation strategies” (8)(22). The health systems strengthening program was supported by the MOH and viewed as a key step in shifting from a humanitarian approach to a more integrated and sustainable health systems strengthening approach that can potentially be more effective than a vertical program approach in addressing underlying health systems problems.

This evaluation represents one of the few studies of the impact of a large, complex health systems strengthening program on the uptake of maternal health care services. Overall, the program had a mixed impact on the use of maternal health care services, improving utilization of facility-based deliveries, but not utilization of ANC services. The qualitative results suggest several enabling factors that may have been responsible for increased facility-based deliveries, including the perception that birth attendants were qualified and deliveries safer, improvements in health facility infrastructure, and perceived affordability of delivery assistance. However, expanding the use of maternal health services was also found to be challenged by supply shortages, an insufficient number of trained and female CHWs, and a lack of consideration of behavioral change strategies to address cultural practices and social norms related to pregnancy and childbirth. In addition, extremely challenging transportation systems and long distances to emergency care were found to be barriers that may explain low use of referral hospitals and rates of C-section deliveries. To decrease the persistently high MMR in the DRC, health systems strengthening programs should concentrate on developing strategies to overcome obstacles that women with obstetric complications or a history of complicated deliveries face accessing timely EmONC, as well as improving quality of care in reference hospitals to ensure that C-sections and other life-saving practices are available 24-hours a day.

These findings highlight the need for formative research to inform interventions so that they match social norms and better address the often complex and dynamic contextual needs in fragile states, as well as the importance of factors outside the health system, such as roads and other aspects of transportation, to improve access to comprehensive emergency obstetric essential care needed to save women’s lives. There were also substantial differences in the outcomes between the two regions, indicating the importance of considering variations in sociocultural systems and other contextual factors, including the history of prior interventions, when designing and implementing programs in different regions of the same country.

Ethics approval and consent to participate: The study received ethical approval from the Biomedical Institutional Review Board of Tulane University and the Office of the Ethical Committee of the University of Kinshasa’s School of Public Health. Informed consent was obtained from all study participants.

Availability of data and materials: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and confidentiality.

Competing interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the study.

Funding: The study was funded by the United Kingdom’s Department for International Development (which has since been replaced by the Foreign, Commonwealth, and Development Office) as an independent evaluation under the Access to Primary Care in DR Congo (ASSP) Project. The funders played no role in the writing of the manuscript or the decision to submit it for publication.

Authors' contributions: DRH and LSB conceptualised the study idea and devised a study design and analysis approach. DRH, LSC, and JW carried out the quantitative analysis. LSB and AY carried out data qualitative analysis. DRH, LSB, and LSC drafted the manuscript. HW and PL critically reviewed the manuscript for its intellectual content. All authors read and approved the manuscript for submission.

Acknowledgements: The authors are grateful to the following ASSP project staff for their assistance in the study: Larry Sthreshley, Bernard Ngoy, Scott Shannon, Mano Tayingi, and Marie-Claude Mbuyi.

Organization WH. Trends in maternal mortality 2000 to 2017: estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. 2019.
Nyamtema AS, Urassa DP, van Roosmalen J. Maternal health interventions in resource limited countries: a systematic review of packages, impacts and factors for change. BMC Pregnancy Childbirth. 2011;11:30.
Salama P, Spiegel P, Talley L, Waldman R. Lessons learned from complex emergencies over past decade. Lancet. 2004;364(9447):1801-13.
Ziegler BR, Kansanga M, Sano Y, Kangmennaang J, Kpienbaareh D, Luginaah I. Antenatal care utilization in the fragile and conflict-affected context of the Democratic Republic of the Congo. Soc Sci Med. 2020;262:113253.
Messner JJ, Haken N, Taft P, Blyth H, Maglo M, Murp C ea. Fragile States Index Annual Report 2019. The Fund for Peace; 2019.
Stearns J. Dancing in the Glory of Monsters: The Collapse of the Congo and the Great War of Africa. Philadelphia, PA: Perseus Books; 2011.
Zarocostas J. Mega-crisis in DR Congo. The Lancet. 2018;391(10118):297-8.
Graham W, Woodd S, Byass P, Filippi V, Gon G, Virgo S, et al. Diversity and divergence: the dynamic burden of poor maternal health. Lancet. 2016;388(10056):2164-75.
Ministère du Plan et Suivi de la Mise en œuvre de la Révolution de la Modernité (MPSMRM) MrdlSPMeI. Enquête Démographique et de Santé en République Démocratique du Congo 2013-2014. Rockville, Maryland, USA: MPSMRM, MSP et ICF International; 2014.
INS. Enquête par grappes à indicateurs multiples, 2017-2018, rapport de résultats de l’enquête. Kinshasa, République Démocratique du Congo. 2019.
Koblinsky M, Moyer CA, Calvert C, Campbell J, Campbell OM, Feigl AB, et al. Quality maternity care for every woman, everywhere: a call to action. Lancet. 2016;388(10057):2307-20.
Anderson RM. Revisiting the Behavioral Model and Access to Medical Care: Does it Matter? Journal of Health and Social Behavior. 2005;36(1).
Organization WH. WHO recommendations on antenatal care for a positive pregnancy experience. World Health Organization; 2016.
Publique MdlS. Manuel opérationnel du Financement Basé sur la Performance. 2016.
Team AS. Evaluation of the Impact of the ASSP (Accès aux Soins de Santé Primaires) Program in the Democratic Republic of the Congo. 2019.
Montagu D, Sudhinaraset M, Diamond-Smith N, Campbell O, Gabrysch S, Freedman L, et al. Where women go to deliver: understanding the changing landscape of childbirth in Africa and Asia. Health Policy Plan. 2017;32(8):1146-52.
Hanson K, McPake B, Nakamba P, Archard L. Preferences for hospital quality in Zambia: results from a discrete choice experiment. Health Econ. 2005;14(7):687-701.
Kruk ME, Mbaruku G, McCord CW, Moran M, Rockers PC, Galea S. Bypassing primary care facilities for childbirth: a population-based study in rural Tanzania. Health Policy Plan. 2009;24(4):279-88.
Kruk ME, Hermosilla S, Larson E, Mbaruku GM. Bypassing primary care clinics for childbirth: a cross-sectional study in the Pwani region, United Republic of Tanzania. Bull World Health Organ. 2014;92(4):246-53.
Organization WH. WHO Statement on Caesarean Section Rates. In: Research RHa, editor. 2015.
Raven J, Wurie H, Idriss A, Bah AJ, Baba A, Nallo G, et al. How should community health workers in fragile contexts be supported: qualitative evidence from Sierra Leone, Liberia and Democratic Republic of Congo. Hum Resour Health. 2020;18(1):58.
Syed SB, Leatherman S, Neilson M, Griekspoor A, Horemans D, Letaief M, et al. Improving quality of care in fragile, conflict-affected and vulnerable settings. Bull World Health Organ. 2020;98(1):2-A.

No competing interests reported.

Download PDF

Editorial decision: Revision requested
11 Jul, 2024
Editor assigned by journal
10 Jul, 2024
Submission checks completed at journal
10 Jul, 2024
First submitted to journal
04 Jul, 2024

You are reading this latest preprint version

Assessing the impact of complex health systems strengthening program on maternal health care utilization in fragile and conflict-affected states: evidence from the Democratic Republic of the Congo

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Background

Methods

Theory

Research design

Results

Quantitative results

Qualitative results

Use of ANC services

Discussion

Conclusions

Declarations

References

Additional Declarations

Status:

Version 1