Malaria rapid diagnostic test (HRP2/pLDH) positivity, incidence, care accessibility and impact of community WASH Action programme in DR Congo: mixed method study involving 625 households

Mixed method research comprising three distinct studies was conducted in two rural towns of Kongo Central province, DRC, from 1 January 2017 through March 2018. The main study, which involved 625 households (3,712 household members), consisted of an Action Research in which pre-(baseline) intervention and post-intervention surveys were carried out using the Malaria Indicator Questionnaire from the World Bank and WHO Malaria Programme in Africa and Madagascar [13, 21]. It was a 6-month prospective study comprising two interventions: community participatory WASH action and malaria education campaign. The latter intervention was implemented in both study sites (Fig. 1a). In addition, baseline and end-of-study malaria testing was performed among the survey responders (household heads).

The second study was a prospective hospital-based epidemiological study was conducted from 1 January through June 2017; medical records of patients were collected during the 6-month period, which was conducted using medical records of patients admitted from 1 January through 30 June 2017 at two randomly selected referral health settings.

In the main study, a two-stage cluster sampling technique was used to randomly select two study sites, at the ‘health zone’ and municipality (county) levels as shown in Fig. 1b. Loma county (WASH action site) was randomly selected in Mbanza-Ngungu health zone in the rural town of Mbanza-Ngungu, located at 154 km from the capital Kinshasa, with an area of 8,460 km². It has a population of 651,092. On the other hand, Quartier residentiel county (Control site) was randomly selected in Kasangulu health zone in the town of Kasangulu. Located at 33 km from the capital Kinshasa, Kasangulu has an area of 4,680 km² and a population of 194,190 inhabitants. In DRC, a health zone consists of primary operational units of the health system and, usually, it covers a population of 100,000–150,000 inhabitants in rural areas and 200,000–250,000 inhabitants in an urban area [22]. The following criteria were used to select the county where the study should be conducted: (1) having a referral health setting under the supervision of the health zone inspector, and (2) the health setting should have a well-handled patients’ records. Hospital-based epidemiological data were collected at each study site to determine malaria incidence.

In the main study, all households from the randomly selected study sites having at least three members were eligible. Considering a power of 80% (β = 0.80) for α value of 0.05, we expected to have at least 200 households participate in this research. Within each study site, blocks of 50 houses were created; thereafter, data collectors have randomly selected every second house on each street in the area of each block. Data collectors were public health nurses and doctors who served under the supervision of one of the authors (WR) who is professor at the Faculty of Medicine, University of Kinshasa in DRC.

Additionally, only households having at least three members were finally included in the study. In total, 625 households (3,712 individuals) were surveyed, including 316 (50.6%) from the WASH action site and 309 (49.4%) from the Control site.

Surveys were conducted simultaneously at both study sites at baseline and at the end of six-month intervention period, following a schedule that was announced to residents by local health zone staff. The French version of malaria indicator survey (MIS) questionnaire was used in this study. It is a validated questionnaire used by the National Malaria Programme of several French speaking African countries to estimate household malaria burden.

MIS comprises an informed consent form and questions related sociodemographic and anthropometric characteristics, household characteristics, past medical history, WASH status, malaria preventive measures, malaria status and care. All household heads participated in the baseline and post-intervention surveys. Additionally, home visits were undertaken by local collaborative research team and Health Zone staff to evaluate WASH status at home and in the living environment, and check indoor and outdoor preventive measures used by household members, and ascertain consistency of their use. A hand-held GPS GIS device was used to collect data on geospatial localization of mosquito breeding sites; that is to estimate the distance between residences and mosquito breeding sites (grassy area, stagnant water spot, garbage spot and river side). We assumed that when a residence was located at less than 200 m from a mosquito breeding site, household members were considered at high risk for malaria.

Anti-malaria interventions comprised the following actions: (1) community WASH action consisting of a weekly participatory hygiene and sanitation transformation (PHAST) was carried out only in WASH action site in order to clean the residential environment and eliminate mosquito breeding spots; (2) community anti-malaria education. The latter was implemented in both study sites after the baseline survey. Education sessions were organized in communities, schools and leaflets that display risk factors and behaviors were being distributed to the participants as well as household heads in each study site. The PHAST approach is a learning methodology commonly used to prevent a broad range of infectious diseases at community level, through improvement of hygiene behaviors and sanitation, and encourages a better community management of water and sanitation services. In this study, PHAST was extended to periodic cleansing of the living environment, by local volunteers and community members.

Participants underwent the rapid diagnostic test (RDT) for malaria at baseline and at the end of the study. Blood sample was collected via finger prick into Heparin-coated tube. RDT is a validated diagnostic test for malaria; it has high accuracy for malaria diagnosis; it has the advantages of rapid-detection and it is easy to use and cost-effective. Thus, it is useful diagnostic procedure in resource-limited and endemic areas for malaria, particularly. It can detect malaria parasite’s specific antigens in the blood: histidine-rich protein-2 (HRP2) and lactate dehydrogenase (LDH). The test allows to diagnose malaria caused by Plasmodium falciparum and other Plasmodium species [23]. All participants with a positive RDT at baseline received malaria treatment and were followed for six months.

Regarding the hospital-based epidemiological study, only patients admitted to internal medicine and pediatric departments o between 1 January through 30 June 2017, and whose records showed final diagnoses were included; records that showed no diagnosis were excluded. For medical records showing comorbidities, the first diagnosis was considered.

In general, measures that reduce outdoor and indoor mosquito population are believed to play a crucial role in reducing malaria prevalence, especially in malaria endemic countries. Thus, during WASH intervention, mosquito breeding sites, which are considered high risk areas, were targeted.

Households living in proximity (distance less than 200 m) to a river, still and stagnant water/grassy spots were considered high risk areas for malaria. On the other hand, residences located at higher altitude or far from river, grassy/stagnant water spots were at lower risk for malaria.

Outcome variables were the following: (1) prevalence of positive RDT among responders, (2) survey-based self-reported household incident malaria (number of doctor-diagnosed malaria cases in household members at any health setting), and (3) hospital-based malaria incidence from data registered at selected referral hospital at each study site. Data are presented as proportions for categorical variables, whereas means and their standard deviations are used for continuous variables. Comparisons within and between study groups were performed using paired t test (for incident malaria incidence cases) and chi-square test (for categorical variables); however, for categorical variables with repeated measures (RDT), McNemar’s test was used. All variables that showed a significant or marginally significant association with household malaria in the bivariate logistic regression analysis were subjected to a multivariate analysis model to determine predictors of malaria. Analyses were performed with the use of Stata software version 15.

Of the 625 respondents (household heads), 70.9% were females; overall mean age was 33.8 ± 8.8 years. The majority (66.2%) of the respondents were married men and women. Regarding education level, most respondents (64.6%) had high school level, followed by those with primary education level (13.6%), whereas 10.2% were illiterate. Mean household size was 5.9 ± 2.8 members, and almost 80% of them earned less than 100 US dollars a month. About 25% of households used appropriate/acceptable latrines. Baseline survey results also revealed that periodic sanitation intervention was not carried out in the residential area; this was supported by 97.6% of the respondents; 18.6% reported that the general sanitation status in their living environment was unacceptable (11.3% at control site vs. 25.6% at WASH action site) (Table 1).

Long-lasting insecticide-treated net (LLIN) was the most commonly used malaria preventive measure (55% of households), followed by mosquito repellent (15%), whereas indoor residual spraying (IRS) and the combination of LLIN and IRS accounted for 2% each. Strikingly, 24% of households did not use any of the measures as shown in Fig. 2.

Regarding malaria testing among the respondents, results showed that the prevalence of positive RDT increased in the post-intervention RDT at the Control site, but not significantly (25% in pre-test vs. 35% in post-test, p > 0.05), whereas a marked reduction was observed in the WASH-action site (38% vs. 20%, respectively; p < 0.05) (Fig. 3).

When comparing the malaria status before and after intervention, the disease incidence decreased among the respondents (96% vs. 52%, respectively) (not shown). Of the 96% of respondents who reported malaria event at baseline, only 66.5% received malaria care at a health setting. At household level, mean incident malaria cases decreased markedly in the WASH action site (2.3 ± 2.2 cases in pre-test vs. 1.2 ± 0.7 in post-test; p < 0.05) following 6-month sanitation intervention; however, no significant change was observed at the Control site (1.8 ± 0.5 vs. 2.0 ± 0.7, respectively) (Fig. 4).

Each study site was divided into areas of low and high risk for malaria as described in methods section. Results showed that household malaria incidence rate was markedly higher (83%) among households with residences located at high risk areas as compared to those living in low risk areas (83 vs. 17%, respectively; p < 0.001). Moreover, household malaria incidence rate was highest (60.9%) among households living in proximity to grassy and/or still and stagnant water spots (Fig. 5a, b).

The multivariate logistic regression analysis (with adjustment for age, gender and occupation) showed that household malaria was positively associated with household size (ORa = 1.39; 95%CI: 2.62–4.89; p < 0.05), proximity of residence to high risk area for malaria (ORa = 5.13; 95%CI: 2.29–8.07; p < 0.001) and the frequency of mosquito bites (2.68; 95%CI: 2.84–4.17; p < 0.05), whereas inverse association was found between household malaria and income status (OR = 2.37; 95%CI: 1.05–3.12; p < 0.05), also between household malaria and general WASH status in the residential environment (ORa = 4.10; 95%CI:2.11–7.08; p < 0.001) (Table 2).

Figure 6 shows the disease distribution at the two referral hospitals located at both study sites. Overall malaria incidence was high, 67.36% (1,108/1,645), followed by respiratory diseases (21.22%; 349/1,645) and other conditions, including gastrointestinal disorders (11.43%; 188/1,645) (Fig. 6a). Pediatric malaria was more frequent, 69.67%, whereas of the disease accounted for 30.33% of patients admitted to Internal Medicine Departments (Fig. 6b).

When considering malaria status at each study site, the rate was 61.35% at Loma county (WASH site) and 73.92% in Kasangulu (Control site) (Fig. 6c). Highest malaria incidence rates were observed between April (83.37%) and May (85.37%); it decreased in June 61.36%, which corresponds to the dry (cold) season (Fig. 6d).