Monitoring malaria using health facility based surveys: Challenges and limitations

Background: Health facility data are more readily accessible for operational planning and evaluation of disease control programmes. The importance, potential challenges and limitations of using facility based survey as an alternative tool for monitoring changes in local malaria epidemiology were examined. Methods: The study involved six areas within the administrative divisions of The Gambia. The areas were selected to reflect socioeconomic and malaria transmission intensities across the country. The study design involved an age stratified cross sectional surveys that were conducted during the wet season in 2008 and in the 2009 during the dry season. Participants were patients attending clinics in six health centres and the representative populations from the catchment communities of the health centres. Results: Overall participants’ characteristics were mostly not comparable in the two methodological approaches in the different seasons and settings. More females than males were enrolled (55.8 vs. 44.2 %) in all the surveys. Malaria infection was higher in the surveys in health centres than in the communities (p < 0.0001) and also in males than in females (OR = 1.3; p < 0.001). Males were less likely than females to sleep under an insecticide treated net in the communities (OR = 1.6; 95 % CI 1.3, 1.9) and in the health centres (OR = 1.3; 95 % CI 1.1, 1.5). Representativeness of the ethnic groups was better in the health centre surveys than in the community surveys when compared to the 2003 national population census in The Gambia. Conclusion: Health facility based survey though a potential tool for monitoring changes in the local epidemiology of malaria will require continuous validation of the facility and participants sociodemograhic characteristics as these may change over time. The effects of health seeking practices on service utilization and health facility surveys as an approach will also need continuous review. The study was carried out in The Gambia located in the western-most part of West Africa. The geography of the area is typified by sub-sahelien savannah vegetation with distinct dry and wet seasons. The mean annual rainfall ranges between 920 and 1450 mm with mean temperature ranging from 23 to 27 °C along the coast and 24 to 32 °C in other parts of the country. The landscape is dominated by the river Gambia and its flood plains. The country Gambia has an estimated population of 1.8 million and the main stay of the economy is agriculture and tourism [12–14]. In the Gambia, malaria transmission is seasonal and restricted to a solo short rainy season which typically lasts from June to October. There is a strong association between the River Gambia and malaria transmission. The alluvial bank of the river is prone to flooding and the resultant marshy vegetation and mangrove swamps generate suitable mosquito breeding sites. The important malaria vectors are Anopheles gambiae sensu stricto, Anopheles melas and Anopheles arabiensis, all members of the Anopheles gambiae. The dominant malaria parasite species is P. falciparum but P.malariae and P.ovale are also present [14, 15]. Recent data suggests that almost all infections contain P. falciparum species [16]. This is structured into primary, secondary and tertiary levels of health care. The primary level of health care targets are settlements often with a population of few hundreds or more people. The secondary level of care is provided by major and minor health centres as used in this study as well as private clinics and is supported by a number of clinical dispensaries. Registered and enrolled nurses and other auxiliary medical staff man the health centres. Services provided by health centres include out-patient, maternal and child welfare clinics, and inpatient care but often on a smaller scale for minor health centres. Some major health centres may have resident medical doctors. Tertiary level of health care is delivered by four government hospitals and supported by the Medical Research Council (MRC) hospital [12, 13]. As far as malaria treatment is concerned all health facilities in the Gambian health care system have the capacity to diagnose malaria following the introduction of the rapid malaria diagnostic tests. Access and equity to health facilities and malaria treatment is ensured by the availability of minor health centres and community health workers who provide basic services countrywide. In addition, there are private or traditional health care providers but majority of patients first seek treatment from the formal health system and those who prefer alternative providers usually do so for cost and failure to improve after visiting the formal health system [17]. The study involved six areas within six local government districts with at least one study area selected from each of the five administrative divisions of The Gambia (Fig. 1). The sites were selected to mirror the scale of malaria transmission intensities in the country [14–16]. They included two coastal areas, two areas from mid-country and two areas from the east. For each pair of areas, one was on the north and one on the south bank of the river. They included a mix of semi-urban and rural areas, and each area had a health centre as shown in Fig. 1. The design has been previously described [10, 18]. Locations of the six study areas and villages across the Gambia. They included a mix of semi-urban and rural areas and each area has a health centre (italics) for recruitment of participants. The map in figure 1 is from another journal, the reference [18] is to this journal This included both sexes of all age groups stratified into five age categories: < 2, 2–5, 6–12, 13–25, >25 years. The inclusion of a wide range of participants’ ages was to enable variations in age pattern of malaria infection in the study area to be detected. The ages were stratified into five categories mainly for reasons of logistics and representativeness. A series of age stratified cross sectional surveys were then conducted during the wet season of 2008 (September–November) and during the dry season of 2009 (March–May). Participants were patients attending the six health centres and the representative populations from catchment villages of the health centres. For each of the six study areas, the most centrally located health centre was used for the study for increasing representativeness. Participants’ selection criteria included all persons, both male and female living in the area. They must have been resident for a minimum of 4 weeks at the time of the survey. They or their parents/guardians (for subjects below 16 years) must have agreed to provide witnessed individual/parental informed consent, applicable assent and willingness to follow all study protocol requirements. The catchment villages of each of the health centres were listed and the Epi-Info 6 random list generator was used to randomly select two villages from each catchment area of a health centre. The village where the health centre was located was conveniently added to the other two villages for the community survey. A list of all compounds in the three villages were combined into a single list and a total of 120 compounds selected at random. Individuals in the compounds were asked to converge at a predetermined location on survey days. All participants who consented were enrolled consecutively into the study according to their age groups until the required number was obtained for each agegroup. For each participant enrolled, a study questionnaire was administered to collect basic information on sociodemographic, clinical and socioeconomic characteristics. The same villages and study procedures were used for both the wet and dry season surveys. To ensure that data from the community surveys represented the spectrum of malaria transmission in The Gambia, sites were selected from both banks of the River Gambia, the coastal areas , and the middle and eastern part of the country. All patients attending the selected health centres during the study period were eligible and were invited to participate regardless of their clinical presentation. Those from whom voluntary informed consent was obtained were enrolled consecutively into the study according to their age groups until the required number per age group per health centre was achieved. The entire data were collected concurrently with the community surveys. A structured questionnaire was administered to collect study information from each participant at the health centres. The questionnaire captured information on socio-demographic characteristics, current signs and symptoms and anti-malarial measures. To ensure that data from the health centre surveys represented that from the catchment communities and the spectrum of malaria transmission in The Gambia, continuous enrolment of participants into health centre surveys was adopted. This was based on previous malaria prevalence estimates. For a given prevalence estimate p and precision d, the 95 % confidence interval around p is p ± d, where d = 1.96x√ (p (1-p)/n). We evaluated n for a range of p values to give a suitable value of d. Different sample sizes with precision ranging from d = 0.05 (5 %) to 0.1 (10.0 %) for prevalence estimates of 12, 25, 30, 35 and 61 % that included the two extremes of parasite prevalence estimates (12 & 61 %) reported earlier in The Gambia [16, 19]. We took into consideration the range of parasite prevalence documented in different age groups at different times over recent years in The Gambia. We assumed 25 % to be the average current parasite prevalence in children less than 6 years of age . Therefore, a sample size of 120 per age group per area would provide over 80 % power at a 5 % level of significance to detect an 18 % difference in parasite prevalence (risk ratio =1.72) between any two age groups or between study sites. It would also provide estimates to within +/- 8.0 % of the true value. The total sample size per area or per health centre was calculated to be 600 (120 x 5 age-groups) people. All study data were captured using standard study forms designed specifically for this study. Only designated, trained study staff completed these forms. All completed forms were checked for internal consistency and all queries were resolved routinely in the field. Data were double entered and validated in OpenClinica database which is Good Clinical Practice (GCP) compliant. The completed datasets were verified and cleaned. Initial analyses checked logical inconsistencies, data completeness and quality. All statistical analyses and estimates were computed using STATA (2012 StataCorp) software according to a predefined analytical plan. All statistical analysis, estimations and hypotheses testing were based on parametric methods and were two sided with statistical significance level set at p-value of ≤ 0.05. All participants were enrolled consecutively according to their age groups until the required number was obtained to ensure representativeness. The same villages, health centres and study procedures were used for the surveys in the two seasons. Both the qualitative and quantitative discrepancies of the malaria parasite results were reviewed by a senior microscopist who was not associated with the study. In addition, the senior microscopist read 10 % randomly selected negative slides. For the haemoglobin test, proper training and handling of the equipment, and regular calibrations and standardization of the Hemocue® Photometer were ensured. The duplicate optical density (OD) of the ELISA results were averaged and normalised against a positive control. The cut-off for sero-positivity was mean plus three standard deviations (Mean +3SD) of the non-immune controls. All data management processes followed the standard operating procedures of the data management department of the Medical Research Council Unit in The Gambia. The scientific justification for this protocol was the need to document the current malaria situation in The Gambia and to help to identify the strengths and weaknesses of community and health facility approaches to malaria data collection. The Gambia Government and the Medical Research Council Unit Joint Ethics Committee gave approval to this research protocol.