Burden of malaria is higher among children in an internal displacement camp compared to a neighbouring village in the Democratic Republic of the Congo

This study was carried out in the IDP camp of Bilobilo, established near the village of Mubi, in Walikale district, North Kivu province, Eastern DRC. A map of the country and North Kivu province is shown in Additional file 1. This region of the country has faced years of violent conflict. Recently, in the spring of 2012, a group of armed Rwandan rebels caused the displacement of over 200,000 resident Congolese, leading to the establishment of the temporary IDP camp in this study [16]. A recent investigation of mortality in the Walikale region found a high crude mortality rate and that the main cause of death of children under five was due to fever/malaria [4].

Written, informed consent was provided by the parent/guardian for all study participants. Ethics approval for the study was obtained from Comité d’Éthique du Nord Kivu (Université Catholique du Graben, ref 002/TEN/2012), the University of Alberta Human Research Ethics Board (ref Pro00055619), and regionally from the Médecin Chef de Zone, within the DRC Ministry of Health.

In order to compare Plasmodium falciparum prevalence between displaced children and a control group, we selected the IDP camp of Bilobilo and the neighbouring village of Mubi. The camp consists of temporary housing structures built from tarpaulin and thatch covering an area of approximately 1 km². It is located approximately 5 km from the village of Mubi along a road, which links North Kivu to Kisangani, a major city in Orientale province. The community of Mubi has a linear configuration, with households situated along the major road. The sites are at the same altitude, share the same climate, and both IDP camp residents and residents of Mubi village are ethnically Nande. Both the camp and Mubi are located within savannah-type ecology adjacent to a large and sparsely populated equatorial forest. Both the camp and village are also near the Lowa River, the main water source for its inhabitants. This area receives rainfall for 5 months during the wet season which is from November to March. We performed two cross-sectional observational studies: (1) community-based surveys of the IDP camp and the village; and (2) a survey of febrile children presenting to a health clinic that treats both IDP camp residents and village residents. At the time of the clinic survey, the median duration of displacement for the families living in the IDP camp was 8 months and at the time of the community-based survey the median duration of displacement for these families was 12.5 months.

The first survey was a community-based cross-sectional observational study. Community health workers, trained in malaria rapid diagnostic testing as previously described [17], visited a random sample of 200 temporary shelters within the Bilobilo IDP camp. Random sampling of temporary households in the camp was performed using a census created by non-governmental organizations providing services in the area. All households on the census were eligible for inclusion. For the control group, the community health workers partnered with a polio vaccination drive organized by the Ministry of Health to survey 200 households in the village of Mubi. Households were sampled using methodology modified from the Expanded Programme on Immunization (EPI) coverage survey method, World Health Organization (WHO) [18]. Because of a relatively linear arrangement of households along the principal road in the village, a random number, n, was generated and the nth household along the road was sampled. In each residence, one child under five was randomly selected and tested for P. falciparum infection using a HRP2-based rapid diagnostic test (Paracheck-Pf^® kit; Orchid Biomedical Systems, Goa, India). All participants who tested positive for P. falciparum were treated with the artemisinin-based combination therapy according to WHO recommendations [19].

The second survey was a clinic-based cross-sectional observational study. A convenience sample of 100 children presenting to the clinic for management of febrile illness from the IDP camp were tested for P. falciparum using the same HRP2-based rapid diagnostic test. Over the same time period, a convenience sample of 100 children (controls) from the village of Mubi presenting to the same clinic for management of febrile illness were enrolled. All participants who tested positive for P. falciparum were treated with artemisinin-based combination therapy according to WHO recommendations [19].

A standard sample size calculation indicated that 93 patients would be needed in each group in order to detect an absolute difference of 20 % in malaria infection rates, assuming a malaria prevalence of 50 % among febrile children presenting from the community; with a power of 80 % and a significance level of α = 0.05.

In order to determine a wealth index, ownership of household assets were coded as binary variables and weighted using principal component analysis (PCA). This method, modified from Filmer and Pritchett [20] has been validated as a measure of household consumption and poverty and has been used in previous studies in sub-Saharan Africa (SSA) [21]. Eleven assets or household characteristics from the questionnaire were included to determine a wealth index; this index was then used to divide the cohort into wealth quintiles. The measures that we included were: household characteristics (electricity, brick vs mud and wattle construction) and asset ownership (bicycle, motor vehicle, radio, telephone, television, refrigerator, chicken, cow and goat).

Where multiple potential confounding variables were present, multivariable logistic regression models were used to determine independent predictors of important outcomes. Statistical analyses employed Stata 14.1 and SPSS Statistics version 19.

Two hundred children from the IDP camp and 200 from the control village were surveyed between July 15 and July 22, 2013 (see Additional file 2 for complete data set). Participant characteristics are shown in Table 1. Maternal education and household wealth were lower among families that live in the IDP camp (Table 1). Self-reported exposure to community violence among families living in the IDP camp was common including: theft (50 %), physical assault (9 %), sexual assault (5.5 %) and knife and gunshot injury (both at 0.5 %).

The point-prevalence of P. falciparum infection was 35/200 (17.5 %) among IDPs and 15/200 (7.5 %) for the control group (OR 2.6; 95 % CI 1.3–4.1; P = 0.0095) (Fig. 1). Of note, these children were minimally symptomatic and had not presented for medical care, but were actively identified in the community at large. Because RDT positivity may persist for up to 1 month after successful treatment of infection [22], this comparison was repeated after exclusion of participants with a history of febrile illness within the past month, and found similar results (OR 3.0; 95 % CI 1.4–6.3; P = 0.003). Factors associated with P. falciparum infection in the cohort included bed net ownership: P. falciparum was detected in 18/202 (8.9 %) children from households owning a bed net compared to 32/195 (16 %) without (OR 0.50; 95 % CI 0.27–0.92; P = 0.024). Age, sex, household size (number of children under five), maternal education, and household wealth were not statistically significantly associated with P. falciparum infection in the cohort. In a multi-variable logistic regression model accounting for possible confounding between covariates, IDP camp residence remained the only significant independent predictor of P. falciparum positivity (aOR 2.6; 95 % CI 1.2–5.7; P = 0.013).

Household bed net ownership was lower in the IDP camp compared to the controls (34 vs 68 %, P < 0.001). Bed net use by the index child the night prior to the survey was also lower among the IDPs than controls (25 vs 56 %, P < 0.001, Fig. 2). Factors associated with bed net ownership in the cohort included higher household wealth index; 59/76 (78 %) of the households from the wealthiest quintile owned a bed net compared to 58/150 (39 %) of households in the poorest quintile (P < 0.001). Maternal education was not associated with bed net ownership (P = 0.242) or use (P = 0.207).

Between January 5 and February 16, 2013, 100 children from the IDP camp and 100 children from the control village presenting for management of febrile illness were included (see Additional file 3 for complete data set). Clinical characteristics, listed in Table 3, show that severity of malaria among children from the IDP camp was similar or greater than control children from the village. As in the community-based survey, significant differences in household wealth and maternal education were observed among IDP camp residents. Comparison of exposure to community violence also showed significant differences between the two groups, with theft, physical and sexual assault higher among displaced families (Table 2).

The prevalence of malaria among febrile children from the IDP camp was 78/100 (78 %) compared to 39/100 (39 %) for the village controls (OR 5.5; 95 % CI 3.0–10.3; P < 0.001). To adjust for potential false-positive test results due to recent infection, this comparison was repeated after exclusion of children with a history of recent febrile illness, with similar results (OR 6.5; 95 % CI 3.1–13.6; P < 0.001). Factors associated with malaria infection in the clinic-based cohort included bed net ownership and use. Malaria was detected less frequently in children from households owning a bed net (39/96 (41 %) vs 77/103 (75 %); OR 0.23; 95 % CI 0.13–0.42; P < 0.001), and less frequently if the child was reported to sleep under the bed net (28/81 (35 %) vs 87/116 (75 %); OR 0.18; 95 % CI 0.095–0.33; P < 0.001). Age, sex, household size (number of children under 5), maternal education, and household wealth were not statistically significantly associated with malaria infection in the cohort. In a multivariable model adjusting for potential confounding effects, IDP camp residence was an independent risk factor (aOR 2.7; 95 % CI 1.1–6.6; P = 0.027) and sleeping under a bed net was an independent protective factor (aOR 0.25; 95 % CI 0.10–0.60; P = 0.002) for malaria infection.

Bed net ownership and use were significantly lower for the children from the IDP camp compared to the children from the village (21 vs 75 %, P < 0.001, and 16 vs 65 %, P < 0.001), respectively (Fig. 2). As in the community-based survey, higher household wealth and maternal education were associated with higher rates of bed net ownership (Fig. 2, P < 0.01 for all comparisons).