Evidence on the Impact of Community Health Workers on Health Indicators in Congo: A Quasi-Experimental Temporal Study ()
1. Introduction
Human resources for health are a fundamental lever for improving access to primary health care and achieving universal health coverage goals in developing countries. The World Health Organization (WHO) estimates that there will be an overall gap of 18 million qualified health professionals to be filled by 2030, a deficit that particularly affects sub-Saharan Africa [1]
In this context, a renewed interest in the use of community health workers (CHWs) is evolving in several countries globally [2]-[5]. CHWs are a category of frontline health workers who are usually members of the communities they serve, receiving minimal health training to provide basic services in their communities [6]-[8]. They connect health services and communities, providing culturally appropriate health information, support, and services [9]. Their role is central to access to care, especially in vulnerable and disadvantaged communities. They are involved in a variety of public health activities, such as awareness-raising, early prevention and detection of diseases, distribution of medicines, home visits, and referrals to health facilities.
In the Republic of Congo, the health system is structured at three levels: strategic or central, intermediate or regional and operational or peripheral. The peripheral level is represented by the health district, composed of a district referral hospital linked to a network and several Integrated Health Centres (IHCs) [10]. In Congo, each IHC serves a health area with between 5000 and 15,000 inhabitants in rural and urban areas respectively; a Health District (HD) covers between 30,000 and 100,000 inhabitants in rural areas, and up to 300,000 in urban areas [11]. It should be noted that a health district is the area covered by a number of health centres. However, Congo experiences a persistent shortage of health workers, particularly in rural areas. The current density is about 1 health professional (doctor, nurse, midwife) per 1000 inhabitants, well below the WHO-recommended standard of 4.45 per 1000 inhabitants [12] [13]. This context limits the supply of health services and population access to healthcare. To address this issue, the Ministry of Health deployed a community health program since 2022 with 2652 CHWs, also known as community relays or associative agents [14]. These agents are trained to ensure the link between the population and the health facilities (FOSA) in health areas. It is expected that CHWs can help alleviate the shortage of health personnel, but it is crucial to verify whether their presence really translates into a significant improvement in the performance of health districts. In Congo, despite the expansion of the CHW program, we have not found any studies conducted to measure its real effect on the performance of health districts. This study was designed to fill this gap by documenting the impact of CHWs on key performance indicators of health services. It seeks to answer the following question: What is the impact of the introduction of community health workers on the performance of health districts in terms of access to and coverage of health services? To answer this question, the following hypothesis was formulated: the intervention of community health workers significantly improves the performance indicators of health districts such as vaccination coverage, antenatal consultations and antenatal follow-up. The results of this study will fill a gap in the existing literature regarding the real impact of CHWs on people’s access to care, guide decisions based on scientific evidence and potentially revise the intervention strategies implemented.
2. Methods
2.1. Design and Setting
A quasi-experimental ecological study based on a pre-post comparison model over a period of 24 months without a control group was conducted. Two periods were compared: the pre-intervention phase (before) T0, from July 1, 2021 to June 30, 2022, and the post-intervention phase (after) T1, from July 1, 2022 to June 30, 2023. Indeed, the Ministry of Health of Congo had introduced CHWs in all health districts of the country in July 2022. The study was conducted in the Republic of Congo, in the health districts. During the study period, the country had 52 health districts divided into 12 Departments of Care and Health Services (DCHS). A total of six (06) health districts were randomly selected for this study. This selection has been adjusted for geographical diversity and data availability. Thus, 6 following health districts, spread over 6 DCHS, were selected for the study at the rate of one SD per DCHS. It is the DS of Oyo in the DCHS of the Cuvette, of Gamboma in the DCHS of the Plateaux, of Kinkala in the DCHS of Pool, of Madingou in the DCHS of Bouenza, of Dolisie in the DCHS of Niari and the DS of Hinda-Loango in the DCHS of Kouilou. The total population of these 6 six DCHS was 355,349.00 in 2023.
2.2. Databases
Data used for this study are from the National Health Information Management System DHIS2 (District Health Information Software, version 2), officially deployed and used by the Ministry of Health and the population of Congo, for routine data collection and monitoring at the national level. DHIS2 is a widely used open-source platform, supported by the University of Oslo, that enables the capture, validation, analysis, and visualization of aggregated health data [15]. In the DHIS2, data are monthly reports at the Integrated Health Centers (IHCs) level and compiled at the Health District (HD) level. The primary data used to calculate the indicators for the two study periods, namely 24 months before and 24 months after the introduction of CHWs, were extracted from DIHS2 by five trained investigators, using the following criteria: department, health district and covered population. These data were used to create a secondary database using KoboCollect Box app. Data validation process was used to check completeness, identify outliers, and ensure data reliability. In the event of an anomaly, checks were carried out from the paper registers of the health centres.
2.3. Indicators
The indicators explored in this study included:
1) Antennal Care 1 (ANC 1) rate: ANC 1 is the first prenatal consultation a pregnant woman has with a healthcare professional, usually a doctor or midwife. Rate of ANC 1 was assessed as Number of pregnant women who had their first prenatal consultation/Number of expected pregnant women × 100
2) HIV testing rate assessed as: (number of people tested for HIV/Target population to be tested) × 100
3) HIV-positive women: assessed among pregnant women at ANC 1 as: (number (Number of pregnant women tested positive for HIV/Total number of pregnant women tested) × 100
4) Intermittent Preventive Treatment 2 (IPT2) rate: Intermittent Preventive Treatment (IPT) for malaria is a malaria prevention strategy that involves administering antimalarial drugs at regular intervals to prevent infection. IPT2 rate was assessed as: (Number of pregnant women who received the 2nd dose of IPT2/Pregnant women expected or who received IPT1) × 100
5) Antennal care 2 (ANC 2) rate: (Number of women who have completed 4 or more ANCs/Number of expected pregnant women) × 100
6) vaccination coverage for pentavalent 1: (Number of children who have received the 1st dose of pentavalent vaccine/Number of children < 1 year of age expected) × 100
7) Vaccination coverage for pentavalent 3: (Number of children who have received the 3rd dose of pentavalent vaccine/Number of children < 1 year of age expected) × 100
8) Vitamin A supplementation coverage (Number of children 6 - 59 months who received vitamin A/Number of children 6 - 59 months expected) × 100
9) Rate of curative consultations: (Total number of curative consultations carried out/Expected total population) × 100
10) Rate of hospitalization for malaria: (Number of malaria hospitalizations/Total all-cause hospitalizations) × 100
2.4. Statistical Analysis
A univariate analysis was first performed to assess whether CHWs had an impact on health indicators. Thus, each indicator was calculated and compared between the 24 months before and 24 months after the introduction of CHWs. All indicator values, expressed as percentages, were compared in terms of absolute differences in percentage points (simple difference between pre- and post-intervention proportions). These variations were assessed by repeated measures analysis of variance (ANOVA), appropriate for longitudinal data, to measure the degree of significance. Assumptions of normality were tested using the Shapiro-Wilk test, and the hypothesis of sphericity was tested using the Mauchly test. Secondly, to assess the real impact of CHWs on the health indicators, mixed effects linear regression was performed to control for potential confounding factors. The choice of this type of regression was justified, as each health district constitutes a geographical site where the data collected are hierarchical or correlated, i.e., several observations are often from the same people within the health districts, with measures before and after intervention. In this case, the observations were not independent. On the contrary, simple regression models that assume that the residuals are independent were not applicable because the standard errors would be too small and p-values incorrect. Indeed, mixed-effects regression introduced random effects to model correlations and hierarchies, in order to control potential confounding effects. Ultimately, linear regression with mixed effects makes it possible to estimate the average effect of the intervention, i.e., the introduction of CHWs, while integrating the unexplained variability between observation units through the introduction of random effects. The models were fitted using the restricted maximum likelihood (REML) method. Residual normality and homoscedasticity assumptions were verified by graphic inspection. The results are presented as estimated coefficients (β), with their 95% confidence intervals and corresponding p-values. Statistical analyses were performed using R software (lme4 package).
2.5. Variables
In this study, the variables included in the regression equation are as follows:
Dependent variables
The dependent variables consisted of the ten tracer indicators selected for the study, which were treated as continuous variables.
Primary variable of interest
The main variable of interest was time (before vs. after intervention), introduced as a fixed effect.
Covariates
These included: 1) seasonality (rainy vs. dry), 2) presence of NGO-supported health projects, 3) flood occurrence (binary variable), and 4) CHW-to-population ratio, all treated as fixed effects. These covariates were selected as adjustment variables. For instance, if health districts with NGO support have both a higher number of CHWs and better health indicators, adjusting for this variable helps avoid overestimating the effect of CHWs. Similarly, because health indicators vary by season, including seasonality as a covariate allows for the assessment of temporal effects on health outcomes.
Random effect
A random effect was introduced for health districts, to account for within-group correlation and inter-unit variability.
The following equation represents a linear fixed-effects regression model used to estimate the impact of community health workers (CHWs) on a given health outcome:
Yit = β0 + β1⋅ASCit + β2⋅ONGit + β3⋅Seasonit + β4⋅Floodit + β5⋅RatioCHWit + ui + εit
where:
- Yit: The dependent variable, representing the health indicator measured for area iii at time t. For example, this could be a consultation rate or morbidity indicator.
- β0: The intercept, representing the average value of YYY when all explanatory variables are zero.
- β1⋅ASCit: The effect associated with the variable ASCit, which reflects the presence or activity of community health workers (CHWs) in area i at time t. The coefficient β1 measures the average impact of this variable on YYY.
- β2⋅NGOit: The effect of NGO presence in area i at time t. This factor controls for parallel interventions that might influence health indicators.
- β3⋅Seasonit: The seasonal effect, for example distinguishing between dry and rainy seasons, which can affect health outcomes.
- β4⋅Floodit: The effect of flooding in area iii at time t, considered an environmental shock that may impact population health.
- β5⋅RatioCHWit: The effect of the relative density of community health workers in area i at time t.
- ui: The unobserved, area-specific effect, assumed to be constant over time (fixed effect).
εit: Random error term specific to each observation, assumed to be independent of others and drawn from the same distribution with consistent properties.
2.6. Ethical Issues
The use of data from DHIS2 was within a framework legally authorized by the Ministry of Health and Population, which is responsible for the collection and management of routine health data. The data used in this study was aggregated and anonymized, and did not contain any personally identifiable or individual information. A formal authorization for the use of the data has been obtained from the Minister of Health and Population. Data extraction was performed by the authorized agents of the Health Information Directorate of the Ministry of Health and Population), and the consolidated files were password-protected throughout the analysis process.
3. Results
3.1. Main Characteristics of Health Districts
Detailed data for each Health District (HD) are presented in Table 1. Data from six Health Districts (HDs) across six Departments of Care and Health Services (DCHS) were used, with one HD selected per DCHS. The total population was 1,122,580, with 62 Community Health Workers (CHWs), representing an average allocation of 10 CHWs per Health District (HD). Half of the health districts (50%) simultaneously benefited from health development support programs provided by Non-Governmental Organizations (NGOs).
Table 1. Description of the six health districts.
Department of care and health services |
Health district |
Number of CHWs |
Average total population |
NGOs health programs |
Flooding |
Dominant season |
Kouilou |
Hinda_Loango |
10 |
100,187 |
None |
Yes |
winters |
Bouenza |
Madingou |
8 |
254,402 |
None |
Not |
winters |
Niari |
Dolisie |
17 |
323,717 |
Yes |
Not |
winters |
Pool |
Kinkala |
5 |
102,866 |
None |
Not |
winters |
Plateaux |
Gamboma |
14 |
224,593 |
Yes |
Yes |
winters |
Cuvette |
Oyo |
8 |
116,815 |
Yes |
Yes |
winters |
3.2. CHWs and Health Indicators
Table 2 shows the cumulative averages of the monthly values of the health indicators before and after the intervention of CHWs, as well as the level of performance for each indicator in percentage points (pp). Significant increases were observed after the introduction of CHWs for the following indicators: ANC1 (+12.6 pp, p < 0.001); HIV testing (+29.7pp, p < 0.001); ANC4 (+5.6 pp, p = 0.002); TPI2 (+10.6 pp, p = 0.008); Pentavalent 1 (+28.9 pp, p < 0.001); Pentavalent 3 (+27.3 pp, p < 0.001); Vitamin A (+34.5 pp, p < 0.001); Curative consultations (+8.3 pp, p = 0.004). However, no significant difference was observed for HIV positivity rate and malaria hospitalizations between the pre- and post-intervention periods involving CHWs.
3.3. Effect of Community Health Workers on Health Indicators
The results for each indicator are presented in Table 3. The effect of CHWs on
Table 2. Comparative summary of the evolution of health indicators.
Indicator |
Average Front |
Average After |
Variation (Percentage points) |
p-value |
Coverage in CPN1 |
30.6 |
43.2 |
12.6 |
<0.001 |
HIV testing rates |
48.2 |
77.9 |
29.7 |
<0.001 |
HIV positivity rate |
2.8 |
5.2 |
2.4 |
0.188 |
Coverage in CPN4 |
17.6 |
23.2 |
5.6 |
0.002 |
TPI2 Coverage |
71.7 |
82.3 |
10.6 |
0.008 |
Vaccination coverage Penta 1 |
66.5 |
95.4 |
28.9 |
<0.001 |
Penta 3 Immunization Coverage |
61.7 |
89.0 |
27.3 |
<0.001 |
Vitamin A supplementation coverage |
64.0 |
98.5 |
34.5 |
<0.001 |
Rate of use of curative consultations |
19.3 |
27.6 |
8.3 |
0.004 |
Malaria hospitalization rate |
53.7 |
53.5 |
−0.2 |
0.957 |
Table 3. Effects on health indicators, final linear mixed regression model.
Variables |
CPN1 (β [95% IC]) |
p |
HIV testing (β [95% CI]) |
p |
HIV positivity rate (β [95% CI]) |
p |
CPN4 (β [95% IC]) |
p |
TPI2 (β [95% IC]) |
p |
After (ASC) |
+35.26 [18.55, 51.96] |
<0.001 |
+55.75 [33.72, 77.78] |
<0.001 |
+6.74 [−7.51, 21.00] |
0.351 |
+14.67 [5.57, 23.78] |
0.002 |
+11.66 [−14.23, 37.56] |
0.375 |
Time |
+2.97 [1.97, 3.98] |
<0.001 |
+5.32 [3.71, 6.94] |
<0.001 |
+0.46 [−0.26, 1.17] |
0.209 |
+2.03 [1.47, 2.59] |
<0.001 |
+2.39 [0.53, 4.25] |
0.012 |
OS Program* |
+16.75 [−39.43, 72.94] |
0.328 |
+14.29 [−95.47, 124.05] |
0.632 |
−4.71 [−30.67, 21.25] |
0.517 |
+3.28 [−51.60, 58.16] |
0.821 |
−3.22 [−28.39, 21.96] |
0.712 |
TSF Program** |
+44.28 [−11.91, 100.46] |
0.077 |
−28.36 [−138.12, 81.40] |
0.382 |
−1.07 [−27.03, 24.90] |
0.876 |
+18.08 [−36.80, 72.96] |
0.292 |
−16.28 [−36.18, 3.63] |
0.080 |
Flooding (Yes) |
+4.26 [−51.92, 60.45] |
0.775 |
+25.29 [−84.47, 135.05] |
0.426 |
−2.23 [−28.19, 23.73] |
0.747 |
−1.26 [−56.14, 53.62] |
0.930 |
– |
– |
Dry season |
+6.50 [2.32, 10.68] |
0.003 |
+1.69 [−3.83, 7.20] |
0.546 |
+2.05 [−1.52, 5.62] |
0.258 |
+1.53 [−0.75, 3.81] |
0.188 |
+4.74 [−1.74, 11.23] |
0.150 |
Period × Time |
−3.15 [−4.31, −1.99] |
<0.001 |
−4.86 [−6.39, −3.33] |
<0.001 |
−0.53 [−1.52, 0.46] |
0.289 |
−1.81 [−2.44, −1.17] |
<0.001 |
−1.61 [−3.41, 0.19] |
0.079 |
Variables |
Penta1 (β [95% CI]) |
p |
Penta3 (β [95% CI]) |
p |
Vitamins A (β [95% CI]) |
p |
Consultation curative (β [IC 95%]) |
p |
Malaria-related hospitalization (β [IC 95%]) |
p |
After (CHWS) |
+165.59 [116.76, 214.42] |
<0.001 |
+148.02 [106.52, 189.52] |
<0.001 |
+197.96 [137.11, 258.80] |
<0.001 |
+34.99 [21.48, 48.52] |
<0.001 |
+12.56 [−6.87, 31.99] |
0.203 |
Time |
+8.49 [6.04, 10.94] |
<0.001 |
+7.77 [5.69, 9.85] |
<0.001 |
+7.24 [4.19, 10.30] |
<0.001 |
+3.05 [2.37, 3.73] |
<0.001 |
+1.60 [0.62, 2.57] |
0.001 |
OS Program* |
+38.89 [−37.77, 115.55] |
0.161 |
+22.83 [−57.22, 102.88] |
0.345 |
+2.86 [−199.40, 205.11] |
0.957 |
+9.16 [−89.41, 107.73] |
0.728 |
+16.16 [−4.06, 36.38] |
0.075 |
TSF Program** |
−16.83 [−93.49, 59.83] |
0.445 |
−17.79 [−97.84, 62.26] |
0.440 |
−5.28 [−207.53, 196.98] |
0.921 |
−5.17 [−103.74, 93.40] |
0.842 |
+58.36 [38.14, 78.58] |
0.006 |
Flooding (Yes) |
+10.03 [−66.63, 86.69] |
0.630 |
+9.90 [−70.15, 89.95] |
0.648 |
−21.24 [−223.50, 181.01] |
0.696 |
+2.72 [−95.85, 101.29] |
0.916 |
−62.87 [−83.09, −42.65] |
0.006 |
Dry season |
+4.19 [−8.03, 16.42] |
0.499 |
+2.05 [−8.34, 12.44] |
0.697 |
+9.54 [−5.70, 24.77] |
0.218 |
+3.93 [0.54, 7.31] |
0.023 |
+0.36 [−4.51, 5.23] |
0.884 |
Period × Time |
−12.89 [−16.28, −9.50] |
<0.001 |
−11.56 [−14.44, −8.68] |
<0.001 |
−13.53 [−17.76, −9.31] |
<0.001 |
−3.43 [−4.37, −2.49] |
<0.001 |
−1.73 [−3.08, −0.38] |
0.012 |
*Interventions of the World Health Organization (WHO) Operational Strategy at the Health District Level **Interventions of the Non-Governmental Organization Terre Sans Frontières (TSF) at the Health District Level.
each health indicator is described below:
Antenatal care 1 (ANC1) coverage
CHWs significantly increased the coverage of first antenatal care visits (ANC1) by 35.3 percentage points (95% CI: 18.6 - 52.0; p < 0.001). The time factor was also significant, showing an average positive trend of 3.0 percentage points per month (95% CI: 2.0 - 4.0; p < 0.001). In addition, the dry season was associated with a significant increase in coverage (+6.5 percentage points; p = 0.003). No significant effects were observed for the other adjusted variables, such as specific programs or flooding. However, the interaction between the post-intervention period and time was negative and significant, indicating a gradual loss of momentum in the effect of CHWs on this indicator over time (−3.2 percentage points per month; 95% CI: −4.3 to −2.0; p < 0.001).
HIV Testing at ANC1
CHWs significantly increased the rate of HIV testing at first antenatal care visits (ANC1) by 55.8% (95% CI: 33.7 - 77.8; p < 0.0001). The time factor was also significant (p < 0.0001; 95% CI: 3.7 - 6.9), indicating an overall positive trend. In contrast, the significant negative interaction (CI 95: −6.4 - −3.3; p < 0.0001) indicates that the effect of CHWs in this indicator decreases over time. No statistically significant effects were observed for the other factors examined, including associated programs, seasonality, and flooding.
HIV positivity rate among pregnant women (ANC1)
CHWs did not have a significant impact on the HIV positivity rate among pregnant women at the first antenatal visit (ANC1). No other factors included in the analysis (complementary programs, weather, floods, and dry season) also showed a statistically significant effect. These results suggest that the integration of CHW does not directly influence the HIV positivity rate.
Antenatal care 4 (ANC14) coverage
CHWs was associated with a significant improvement in ANC4 coverage, estimated at 14.7% (IC 95: 5.6 - 23.8; p < 0.002). The time factor also shows a significant positive effect (IC 95: 1.5 - 2.6; p < 0.0001). The term interaction is negative and significant (CI95: −2.4 - −1.2; p < 0.0001;), reflecting a gradual weakening of the effect of CHW on this indicator. The other variables, such as non-governmental organization health programmes and dry season or floods did not have a significant effect.
Intermittent preventive treatment (IPT2) regimens for malaria coverage
CHWs did not significantly influence the coverage of intermittent preventive treatment (IPT2), although a marginal trend towards a gradual decrease in its effect was observed (Period After × Time interaction: β = −1.61; p = 0.079; IC 95: −3.4 - 0.2). However, an overall significant improvement in the indicator is found over time regardless of the intervention (β = 2.39; CI 95: 0.5 - 4.2; p = 0.012). No other factors (complementary programs and seasonality) had a significant effect.
Penta 1 vaccination coverage
CHWs had a very significant effect on Pentavalent 1 vaccination coverage, with an estimated increase of 165.6% (p < 0.0001; IC 95: 116.8 - 214.4). The time factor also reinforces this dynamic (IC 95: 6.0 - 10.9; p < 0.0001). Nevertheless, the significant negative interaction (p < 0.0001) suggests a reduction in effect over time. None of the other factors (programmes, season, floods) were significantly influenced by this indicator.
Pentavalent 3 immunization Coverage
Similar to the first dose, Pentavalent 3 coverage increased significantly with CHWs intervention, with an estimated effect of 148% (IC 95: 106.5 - 189.5; p < 0.0001). Time also plays a significant positive role (p < 0.0001; IC 95: 5.7 - 9.9). The negative period × time interaction is also significant (CI95: −14.4 - −8.7; p < 0.0001), indicating a gradual decrease in effect over time. No other variables analyzed showed a statistically significant effect.
Vitamin A supplementation coverage
CHWs were associated with a highly significant improvement in vitamin A supplementation coverage, estimated at 198% (95% CI: 137.1 - 258.8; p < 0.0001). The time variable significantly contributed to the observed improvement (p < 0.0001; 95% CI: 4.2 - 10.3). However, the interaction between period and time was negative and significant (95% CI: −17.8 to −9.3; p < 0.0001), suggesting a progressive reduction in the intervention effect over time. No other variables, such as associated programs, flooding, or dry season, showed statistically significant effects.
Use of services (Curative consultations)
CHWs had a significant positive impact on curative consultations, with an estimated increase of 35% (p < 0.0001; 95% CI: 21.5 - 48.5). Time was also strongly associated with an increase in consultations (95% CI: 2.4 - 3.7; p < 0.0001). The dry season showed a moderate but significant effect (95% CI: 0.5 - 7.3; p = 0.0233), while the period-by-time interaction indicated a significant decrease in effect over time (p < 0.0001; 95% CI: −4.4 to −2.5). None of the other adjusted variables had a significant effect on this indicator.
Malaria-related hospitalization
In contrast, the time factor (p = 0.0015; 95% CI: 0.6 - 2.6), TSF programs (p = 0.0064; 95% CI: 38.1 - 78.6), flooding (p = 0.0055; 95% CI: −83.1 to −42.7), and the interaction between period and time (p = 0.0124; 95% CI: −3.1 to −0.4) all exhibited statistically significant effects. These findings suggest that changes in hospitalization rates are more influenced by structural or environmental factors than by the community intervention itself.
4. Discussion
This study provides empirical evidence suggesting an improvement in several health district performance indicators after the introduction of Community Health Workers (CHWs). However, this relationship should be interpreted with caution. Overall, the introduction of CHWs appears to have contributed to a significant improvement in the majority of health performance indicators analyzed. This trend confirms observations in other African and international contexts, where CHWs have been shown to be effective vectors for expanding health care coverage [9] [16]-[18].
Significant improvements have been observed in indicators such as coverage of antenatal consultations (ANC1 and ANC4), vaccination (Pentavalent 1 and 3), vitamin A supplementation, and the rate of use of curative services. These results are consistent with the findings reported by Gilmore in a review of developing countries [19]. Moreover, Berini et al., in a review conducted in the United States, concluded on the same effects [20].
These improvements suggest that CHWs have been a catalyst for access to care [20]. Their proximity to communities, their knowledge of the field, their ability to establish a relationship of trust with the population, and their role as relays between households and health facilities may be factors that have facilitated this increase in indicators. This contribution is all the more relevant in a context marked by the chronic shortage of qualified personnel and the difficulty of access to health facilities in rural or peri-urban areas [4]. Indeed, by providing a link between the community and health care structures, CHWs have made it possible to reduce geographical and cultural barriers [9].
The effectiveness of CHWs on immunization coverage and antenatal care indicators is consistent with the results of similar studies conducted in Ethiopia or Rwanda, where the structured integration of CHWs has resulted in a rapid and sustained increase in attendance at basic services [16] [21]. This dynamic reveals not only their ability to mobilize communities, but also their potential to support the achievement of universal health coverage [2].
However, certain indicators, such as HIV seroprevalence rates and malaria-related hospitalizations, did not show any significant change after the introduction of Community Health Workers (CHWs). This observation can be explained by the nature of these indicators, which depend not only on community mobilization but also on structural factors within the health system. More specifically, several obstacles may have limited the potential impact of CHWs on more complex clinical outcomes. These include the limited availability of diagnostic tools at the community level (e.g., rapid HIV tests, rapid malaria diagnostic tests), irregular supply of medicines, particularly for antimalarial treatments, and insufficient capacity of referral facilities to manage severe or complicated cases. This suggests that, while CHWs play a vital role in detection, education and referral at the community level, their efforts alone are not sufficient to influence outcomes that depend on the functioning of higher levels of the healthcare system [3].
In addition, the analysis shows that the positive effect of Community Health Workers (CHWs) decreases over time across several indicators. The observed decline in CHW impact over time may be attributed to the weakening of key programmatic components necessary for sustaining their performance. Evidence from similar settings highlights the importance of continuous support mechanisms, including performance-based incentives, access to refresher training, and clearly defined career pathways [7] [22]. Without these structural supports, CHW motivation and service quality tend to deteriorate, ultimately reducing program effectiveness. These findings underscore the critical importance of ensuring long-term supervisory structures and support systems to preserve and build upon initial gains [3].
Finally, certain contextual variables (dry season, floods, parallel programs) also influenced the performance of the districts, reminding us that the action of the CHWs is part of a broader ecosystem, where several factors can potentiate or hinder the effects of their intervention.
This study is distinguished by its quasi-experimental before-and-after methodology, based on a longitudinal ecological design covering a period of 24 months without a control group. It is based on the use of routine data aggregated monthly, from the national health information system (DHIS2), and on adjusted statistical analyses taking into account several potentially confounding contextual factors.
However, several limitations should be acknowledged that may affect the robustness of the findings. First, the absence of a control group due to the nationwide simultaneous introduction of CHWs precluded the establishment of a valid comparison condition, thereby limiting the strength of causal inferences. Second, although the pre-intervention period spanned 12 months, allowing for the inclusion of a full seasonal cycle, challenges remain. A one-year baseline may be insufficient to fully capture long-term secular trends or interannual variability in health indicators. This limitation may compromise the stability of baseline trends and reduce the power of causal inference in interrupted time-series analyses. Longer pre-intervention periods would enable more robust trend estimation and improve the reliability of attributing observed changes to the intervention. Finally, while the model adjusts for key contextual and time-varying factors such as seasonality, presence of NGOs program, and CHWs density, residual confounding cannot be ruled out. Important contextual variables including socio-economic disparities and local variations in resource availability were not captured and may have influenced the outcomes. Therefore, the estimated effects of CHWs should be interpreted with caution, as they represent an approximation within a complex and heterogeneous real-world implementation context.
5. Conclusion
The study shows a positive impact of CHWs on several health indicators in health districts. Results are in favor of strengthening CHWs programs as levers to strengthen health indicators in health districts with a low ratio of health workers, to increase the chances of moving towards achieving the objectives of universal health coverage. In this regard, challenges related to governance and financing of national community health strategies, as well as the training and motivation of CHWs, must be addressed in order to stimulate and maintain the benefits of care in local communities.
Acknowledgements
We sincerely thank the Ministry of Health and Population of the Republic of Congo for granting access to the DHIS2 platform, which greatly facilitated data collection and analysis. We also wish to acknowledge all those who contributed to the completion of this work.
Author Contribution
D. R. N. M. and G. N. conceptualized the study and manuscript. D. R. N. M. conducted field data collection. D. R. N. M., J. A. N., and J. M. performed the statistical analyses. G. N., C. A. N., G. E. M., and P. L. contributed to supervision, database visualization, and manuscript review and editing. All authors read and approved the final version of the manuscript prior to submission.
Data Availability
The source data used in this study are available on the DHIS2 platform of the Ministry of Health and Population of the Republic of Congo, and can be obtained upon simple request addressed to the institution. However, the specific dataset used for the analysis and preparation of this manuscript is available from the corresponding author, R. D. M. N., upon reasonable request.