Individual and household characteristics of persons with Plasmodium falciparum malaria in sites with varying endemicities in Kinshasa Province, Democratic Republic of the Congo

The DRC Health Monitoring Information System maintains a database on both malaria case counts and total population at the level of the health area. Health areas (Aire de Santé) are the smallest administrative units used for health surveillance by the DRC government. In Kinshasa Province, the 406 health areas are nested inside 35 health zones. The national health surveillance database was used to calculate the prevalence of malaria in each health area from two selected health zones in Kinshasa Province, Maluku and Lingwala, for the year 2014. These two health zones were selected based on accessibility year-round by the research team and to cover the urban/rural gradient. Health area prevalence estimates were partitioned into quartiles and randomly selected one health area from the quartiles to ensure that sampling sites came from different levels of malaria endemicity. Two health areas (Bu and Kimpoko) were chosen from the Maluku 1 health zone and one health area was selected from Lingwala, for a total of three health areas. The Maluku health zone is a rural zone while the Lingwala health zone is located in metropolitan area of Kinshasa City.

Health area maps were used in conjunction with Google Earth to identify all village sites within the selected two health areas of the Maluku 1 health zone. Three village sites within each health area were selected for inclusion in the study. Selected sites needed to meet the following eligibility criteria: (1) the households within each site were accessible by either vehicle or a short walk from the vehicle; (2) the population in each site was not transient; (3) there was an availability of sufficient numbers of individuals within the site to give informed consent. The Voix du Peuple health area located in the Lingwala health zone was chosen to provide a low malaria prevalence site of surveillance. Thus, seven sites with a range of malaria endemicities were selected for surveillance. These sites are located in Kinshasa Province of the DRC; three located in the interior of the province (#1–3 in the Bu health area), three near the Congo River (#4–6 in the Kimpoko health area), and one in the Kinshasa City metropolitan region (#7 in the Voix du Peuple health area) (Fig. 1).

For each of the seven surveillance locations, two geographic coordinates within the site were selected, a beginning and end point. Households were then randomly recruited, starting from the household nearest the start point coordinate and moving towards the household nearest the end point coordinate. Geographic coordinates of households that consented to participate in the study were recorded using GPS devices (Garmin Dakota 20 GPS). Study enrollment took place from February through May, 2015 during the rainy season.

The study team visited each randomly selected household to assess whether the household met enrollment criteria and to obtain informed consent, or parental assent for minors < 18 years of age, in Lingala and French. Informed consent and assent was given not only for collection of survey information but for collection and storage of biological specimens for future studies. The study population included both sexes, pregnant women and children, and complied with all pertinent ethical guidelines for studies in these populations. Individuals were eligible for inclusion if they had permanent residence within the study site and were willing and able to provide informed consent. Criteria for exclusion included evidence of impaired judgement for informed consent, an inability to understand one of the languages used by the study team, or serious concurrent illness. If a given household did not meet the inclusion criteria, it was replaced by the next closest household to the corresponding coordinates.

During the consent and assent process, each subject was told about the study, explaining that this would be a 2-year research study composed of both active and passive surveillance. For the 2-year study, each household would be visited by a study team two times per year for the active surveillance part of the study. One visit will occur in the dry season and one in the rainy season.

At the time of enrollment, the study team administered a questionnaire to each subject, requiring approximately 30 min. The questionnaire based upon the malaria module of the Demographic and Healthy Surveys (MEASURE-DHS, Rockville, MD), was administered. This questionnaire collects basic demographic information (sex, age) as well as information on bed net presence, age and usage (Additional file 1). Additional items in the questionnaire examine housing characteristics (construction materials), wealth (item ownership, electrification) and proximity to malaria risk factors (i.e. water). Residents also reported information on basic health outcomes related to malaria, such as having prior malaria diagnoses or the presence of fever.

Subsequent to the collection of blood for the RDT, four dried blood spots (DBS), each containing approximately 50 µl of whole blood collected by finger-prick, was prepared on Whatman 3 MM filter paper (Fisher Scientific, Fair Lawn. NJ). The completed questionnaire and the filter paper containing the DBS were labelled with the subject’s code. After air-drying for at least 15 min or until dry at ambient temperature, the filter paper was placed in a ziplock bag-containing desiccant. DBS were then transported to the study office in Kinshasa and stored at – 20 °C until shipment to UNC-Chapel Hill for further testing.

In addition, at the time of enrollment, participants were told that any time they will experience fever or any symptoms related to malaria, they should visit the identified local study health centre for malaria diagnosis and treatment if needed. This is part of the passive surveillance of the study intended to estimate malaria incidence in the cohort. At the clinic during passive surveillance, all subjects had axillary temperature readings performed in addition to an RDT and hemoglobin. Even though microscopy was not part of the study design, any time smears were made and read, the results were recorded. Treatments given according to the National Malaria Programme guidelines were recorded.

Participants were instructed that whenever he/she subjectively experienced any adverse health event, or any other participant of his/her family experienced an adverse event, they were to report to the local health centre (clinic) or directly to the staff of the study. All adverse events, including malaria, will be reported to the study team in Kinshasa, and all serious events will be reported immediately to IRBs (KSPH and UNC).

At the end of the enrollment process, a study card was provided to participants in order to be presented anytime they visited a study clinic and during the subsequent household visits. This allows for the longitudinal, active and passive surveillance data to be accurately collected by clinic staff and community health workers. The completed questionnaire was labelled with the subject’s bar code and was subsequently double-entered and checked for consistency.

Subsequent to the enrollment data collection process, DNA was extracted from individual 6 mm DBS punches using Chelex as previously described [23]. The DNA was tested using a duplex real-time PCR assay targeting the P. falciparum-specific lactate dehydrogenase (pfldh) and human beta-tubulin (HumTuBB) genes, with a lower limit of detection of 5–10 parasites/µl (Additional file 2). All assays were performed in duplicate. Standards were made from cultured FCR3 strain P. falciparum, serially diluted in human whole blood after parasite density determination by three independent observers. The standards were prepared as 50 µl DBS using Whatman 3 MM filter paper and underwent DNA extraction using the same methods described above. Isolates that failed to amplify HumTuBB in duplicate or had a single positive replicate with cycle threshold value (C_T) > 38 were excluded from analysis. Isolates were considered to be pfldh PCR positive if both replicates amplified or if a single replicate amplified with C_T ≤ 38, as previously described [24]. Before proceeding to data analysis, it was first confirmed that both replicates of standards containing 10 parasite/µl and greater successfully amplified. Regular quality control checks were performed to ensure consistent PCR performance across plates, with monitoring of mean C_T values, standard curve R² values, and standard and negative control amplification by an independent observer, who also confirmed accurate entry of PCR results. While more sensitive PCR assays are available, a single-copy gene assay was chosen to improve the accuracy of the quantitative PCR results. Most single copy gene assays have limits of detection similar to the pfldh assay employed in this study. More sensitive assays typically target multi-copy genes or repetitive elements of the P. falciparum genome. While the presence of multiple targets allows for improved sensitivity, the number of gene copies can vary between parasites, making accurate quantitation impossible.

Only the baseline data collected at the time of enrollment is presented. Descriptive statistics at the individual and household level were generated. The relationship between demographic and behavioral characteristics and the presence of malaria parasites was assessed via Pearson’s Chi square test. The threshold for statistical significance was p = 0.05. A kappa statistic was calculated to determine concordance between RDT and PCR results for study individuals [25].

Rapid diagnostic test and PCR positive counts were generated at the household level as was the total population living in each household. To protect participant confidentiality and produce maps of malaria in the study sites, the total counts of RDT and PCR malaria and household populations were smoothed using kernel density estimation. In kernel density estimation a window (or kernel) estimates density at the window centroid using the values of all events (RDT or PCR or population counts) within the window [26]. Observations located closer to the window centroid are given greater weighting in the calculation of the density estimate. The kernel moves across the study area and calculates a density estimate for all locations in the seven study sites. The smoothed kernel for RDT and PCR was then divided by the kernel for population to generate an estimated surface of malaria prevalence in each of the study sites. A kernel of 100 m was used to capture heterogeneity in malaria prevalence across the small and compact study sites.

Chi square tests and calculation of the kappa statistic were completed in R (CRAN, R Foundation for Statistical Computing, Vienna, Austria). Mapping was completed in ArcGIS 10.4 (ESRI, Redlands, California).