Two strategies for the delivery of IPTc in an area of seasonal malaria transmission in the Gambia: A randomised controlled trial

Background: The Expanded Programme on Immunisation (EPI) provides an effective way of delivering intermittent preventive treatment for malaria (IPT) to infants. However, it is uncertain how IPT can be delivered most effectively to older children. Therefore, we have compared two approaches to the delivery of IPT to Gambian children: distribution by village health workers (VHWs) or through reproductive and child health (RCH) trekking teams. In rural areas, RCH trekking teams provide most of the health care to children under the age of 5 years in the Infant Welfare Clinic, and provide antenatal care for pregnant women. Methods and Findings: During the 2006 malaria transmission season, the catchment populations of 26 RCH trekking clinics in The Gambia, each with 400-500 children 6 years of age and under, were randomly allocated to receive IPT from an RCH trekking team or from a VHW. Treatment with a single dose of sulfadoxine pyrimethamine (SP) plus three doses of amodiaquine (AQ) were given at monthly intervals during the malaria transmission season. Morbidity from malaria was monitored passively throughout the malaria transmission season in all children, and a random sample of study children from each cluster was examined at the end of the malaria transmission season. The primary study endpoint was the incidence of malaria. Secondary endpoints included coverage of IPTc, mean haemoglobin (Hb) concentration, and the prevalence of asexual malaria parasitaemia at the end of malaria transmission period. Financial and economic costs associated with the two delivery strategies were collected and incremental cost and effects were compared. A nested case-control study was used to estimate efficacy of IPT treatment courses. Treatment with SP plus AQ was safe and well tolerated. There were 49 cases of malaria with parasitaemia above 5,000/μl in the areas where IPT was delivered through RCH clinics and 21 cases in the areas where IPT was delivered by VHWs, (incidence rates 2.8 and 1.2 per 1,000 child months, respectively, rate difference 1.6 [95% confidence interval (CI) -0.24 to 3.5]). Delivery through VHWs achieved a substantially higher coverage level of three courses of IPT than delivery by RCH trekking teams (74% versus 48%, a difference of 27% [95% CI 16%-38%]). For both methods of delivery, coverage was unrelated to indices of wealth, with similar coverage being achieved in the poorest and wealthiest groups. The prevalence of anaemia was low in both arms of the trial at the end of the transmission season. Efficacy of IPTc against malaria during the month after each treatment course was 87% (95% CI 54%-96%). Delivery of IPTc by VHWs was less costly in both economic and financial terms than delivery through RCH trekking teams, resulting in incremental savings of US$872 and US$1,244 respectively. The annual economic cost of delivering at least the first dose of each course of IPTc was US$3.47 and US$1.63 per child using trekking team and VHWs respectively. Conclusions: In this setting in The Gambia, delivery of IPTc to children 6 years of age and under by VHWs is more effective and less costly than delivery through RCH trekking clinics. © 2011 Bojang et al. Study participants were recruited from villages and hamlets on the south bank of Upper River Region (URR), The Gambia. The study was conducted between September 2006 and June 2007. In this area, malaria transmission is highly seasonal, occurring mostly during and immediately after the rainy season (July–November) [21]. Usage of ITNs is relatively high with 58.5% of households having at least one ITN in a survey undertaken in 2006 [22]. At the time of this study, first-line treatment for uncomplicated malaria in The Gambia was chloroquine plus SP. In 2001, the PCR-corrected treatment failure rate in symptomatic children at day 28 after the start of treatment with SP and chloroquine was 6% in the central part of the country [23]. Since 2008, first-line treatment has been changed to lumefantrine-artemether (Coartem, Novartis). The study protocol and CONSORT checklist can be found in the supporting information (Text S1 and S2, respectively). The study was designed as an open-label, cluster randomised trial with the unit of randomisation being the catchment population of RCH trekking teams. There were 33 trekking clinics in the study area on the south bank of URR of which 27 were in rural areas and visited by trekking teams from health centres at Gambisara, Basse, or Fatoto. 26 rural trekking clinics were selected for inclusion in the study (one was excluded to give an equal number in each study arm). The catchment populations of these clinics were grouped into three strata on the basis of the number of children under 5 y of age in each catchment area, and the average distance from the nearest health centre. It was intended that only children under five would be enrolled. However at the time of enrolment about 16% of the children enrolled were more than 5 years. Hence the trial and the analysis were expanded to include children up to and including the age of 6. Within each stratum, clinic catchment areas were randomised, using permuted blocks using the random number generator in Stata version 10 (Stata Corporation), to receive IPTc delivered either by a VHW or through the RCH trekking team. The randomisation was well balanced with respect to distance from the nearest health centre, population of children up to 6 y of age, proximity to a main road, and quality of the road, factors that might influence access to the intervention and to malaria diagnosis and treatment. For all the villages in the study area, the nearest health centre could be reached in less than 4 h by local transport or was within a few hours walking distance. Following discussions with the National Malaria Control Programme (NMCP) and the URR district level health authorities, members of the study team, accompanied by the district health team, visited all villages in the study area, explained the purpose and methods of the study, and answered questions during meetings open to all villagers. During these meetings, consent was obtained from the elders of all villages in the study area to participate in the trial. Before the start of the study, field workers and investigators carried out a census in each of the 172 villages in the study area that agreed to participate in the trial. A list of all children living in these villages who would be aged between 6 and 72 mo at the time of the first treatment in September 2006 was obtained from this database to ensure that all the children in the right age group were considered for inclusion in the study. Families of prospective study participants were visited at home, where more information was provided about the trial and individual, written informed consent for a child to participate in the trial was obtained from parents or guardians. Exclusion criteria included known allergy to any antimalarial drug or the presence of acute or chronic, clinically significant pulmonary, cardiovascular, hepatic, or renal disease. A reporter employed by the project was based in each village and asked to keep records of all deaths, birth, immigrations, and emigrations that occurred during the study period. Village reporters were visited fortnightly by project field workers and their records collected and checked. Deaths that occurred at home were investigated with a standardised verbal-autopsy questionnaire and cause of death established whenever possible [24]. Drugs were delivered by either members of RCH trekking teams or by VHWs. Study participants were observed for approximately 30 min after dosing and, if vomiting occurred within this period, it was recommended that the child take another dose. RCH trekking teams visited each of the study villages approximately once a month to administer routine EPI vaccines and vitamin A. During the course of the intervention each trekking team in the RCH arm of the trial comprised an average of ten staff members, four of whom were provided by the study team to assist with the intervention. Trekking teams comprised staff with varying levels of training ranging from state enrolled nurses to volunteers. Each member of the team employed by the Ministry of Health received a salary supplement of US$28 per month (US$ 2006) (about 47% of their salary) during the 3 mo of the intervention as an incentive for the extra hours of work that administration of IPTc incurred. Trekking team drivers received a supplement of US$14 per month. All villages in The Gambia with a population of approximately 400 or more have provision for a VHW who receives limited training from the Ministry of Health on the recognition and treatment of common illnesses including malaria. VHWs are meant to be supported by their community and do not receive any payment from the Ministry of Health. VHWs retain a small stock of essential drugs, including antimalarials, but as these must be purchased from a government pharmacy they frequently run out of stock. Thus, for the period of the study, all VHWs in the study were supplied with antimalarials by the project for the period of the study. VHWs received US$11 per month (US$ 2006) during the period of drug administration and a driver with a supplement of US$14 per month to deliver drugs to the VHWs. VHW supervisors did not receive any additional payment. Before the start of the study, VHWs were trained for 5 d on how to distribute monthly IPTc using drug treatment charts, recognise signs and symptoms of malaria and their relation to severity of illness, and maintain a compliance ledger. Study participants from villages within the IPTc arm of the study but without a resident VHW received IPTc from the nearest village with a VHW. A system for recording drug administration that could be used by a VHW was devised, which consisted of a treatment card held by the mother or guardian of each child and an IPT register held by the VHW or RCH trekking team. The treatment card and the IPT register were labelled with the child’s, mother’s, and village’s names and compound, and with demographic surveillance and study numbers. A set of three sticky labels were preprinted with the child’s identifiers and placed in an envelope that was stapled onto the treatment card. The dosage of trial medication for each study participant was indicated on a largely pictorial treatment card, employed because of the low levels of literacy of some VHWs, using coloured circles and semicircles (a full circle for one tablet and semicircle for half a tablet). To ensure that a child did not receive more doses of IPTc than they should, blue-coloured treatment cards were issued to children in villages where IPTc was to be delivered by VHWs and pink-coloured cards to those allocated to receive medication from the RCH trekking clinic. When a child presented to the VHW or RCH trekking clinic for their monthly treatment, one sticky label was removed from the envelope and attached to the IPTc register, the child’s identity checked and trial medication SP (tablets containing 500 mg sulfadoxine/25 mg pyrimethamine) (Microlabs) and AQ (200 mg base tablets) (Microlabs) were given. A single dose of SP was given; children aged 3–11 mo received one half tablet and a whole tablet was given if a child was aged 1–6 y. AQ was given daily for 3 d: one quarter of a tablet if a child was aged 3–11 mo, one half of a tablet if aged 1–2 y, and a whole tablet if aged 3–6 y. SP and the first dose of AQ were given under direct observation by the VHW or member of the trekking team. The mother or carer was given tablets for the second and third dose of AQ in a small plastic bag and was asked to administer these on the following 2 d at home. Children were given their tablets crushed, suspended in water, and administered with a spoon. The solubility and drug content of the SP and AQ tablets was confirmed by analysis with high performance liquid chromatography undertaken at the London School of Hygiene & Tropical Medicine. To determine the prevalence of adverse events following drug administration, 15 children in each cluster selected from the participant database by simple random sampling were visited at home within a week of the start of each round of drug administration by a trained field worker who completed an adverse event questionnaire. Parents and guardians of study children were encouraged to take their child to the study nurse based in the village or to the Medical Research Council (MRC) clinic at Basse if the child became unwell between visits. Any adverse reaction was graded by trained field workers as mild (grade 1) if it was easily tolerated, moderate (grade 2) if it interfered with normal activity, and severe (grade 3) if it prevented normal activity and required treatment. In addition, records were kept of all deaths and hospital admissions among study children during the period of the trial. A serious adverse experience was defined as any event that was fatal, life threatening, disabling or incapacitating or resulted in hospitalisation, prolonged a hospital stay, or was associated with overdose (either accidental or intentional). A decision was made by the study physician as to whether a serious adverse event was likely to be drug related. Passive surveillance for malaria was maintained throughout the malaria transmission season (September to December). In both arms of the trial, study children had 24-h access to medical treatment provided by VHWs based in key villages or by study nurses based in the health centres within the study area. In both arms of the study, VHWs were asked to refer study participants who presented to them to the nearest health facility for evaluation. Each time a study patient presented to a health facility, axillary temperature was recorded using a digital thermometer and haemoglobin (Hb) concentration was measured using a portable haemoglobinometer (HemoCue AB). The Core Malaria Pf test (CORE Diagnostics), based on detection of circulating Plasmodium falciparum histidine-rich protein 2 antigen, which is sensitive and specific for detection of P. falciparum in whole blood samples, was used to diagnose malaria if fever (axillary temperature ≥37.5°C) or a history of fever within the previous 48 h was present. In such cases, a thick blood smear was also collected for subsequent confirmation of the diagnosis and estimation of parasite density. Study participants with documented fever or history of recent fever and malaria parasitaemia were treated with Coartem. Treatment of study participants seen at the health centres for other conditions was carried out in accordance with national guidelines. A cross-sectional survey was undertaken in all 26 clusters at the end of the malaria transmission season. Children in each cluster were selected for review by simple random sampling from the list of study participants. The families of children selected for review were visited at home, and follow-up visits made to find children who were not present at the first visit. The first 40 children to be located in each cluster were included in the survey. Children included in the survey were examined, their axillary temperature measured, and their height and weight recorded. A finger-prick blood sample was taken for preparation of thick smears and measurement of Hb concentration. The dates of IPT treatments received were recorded from the child’s identification (ID) card, and the mother or career was asked about reasons for missed treatments, about the acceptability of the most recent round of treatment, and about compliance with the unsupervised doses. The families of the same children were visited 2–3 wk later and asked questions about household assets, level of education and sources of income of the child’s primary care giver, the care giver’s perception of the relative wealth of the household, bednet use by the study child, and asked to inspect the place where the child usually slept and to inspect the condition of their net if a net was being used. In order to estimate efficacy of IPTc, a nested case-control approach was used. Cases were children who presented to one of the study health centres with an axillary temperature ≥37.5°C or a history of fever in the previous 48 h and who had malaria parasitaemia with a parasite density ≥5,000/µl on initial reading. Controls (initially two per case and subsequently one per case) were selected from among children who presented to the same health centre with a febrile illness in the same week that the case was detected, who had a negative blood film, and whose village of residence was in the same delivery arm (VHW or RCH) as the case child. Cases and controls were visited at home to inspect the place where the child slept, to ask the child’s mother or carer about bednet use at the time that the case child became sick, and about household assets. Dates of IPT treatment courses were obtained from the child’s ID card and cross-checked against study registers. Thick smears were stained with Giemsa stain and 200 high power fields (HPFs) were examined before a smear was declared negative. Parasite density was expressed per microlitre with the assumption that one parasite per HPF equals a density of 500 parasites per microlitre [25]. All slides were read independently by two laboratory technicians. If there was disagreement between their readings on parasite positivity or if the difference in the log10 densities recorded was more than 1.5, slides were read by a third technician. Agreement was reached among the three technicians after the slides had been rechecked. Discrepancies occurred mainly in smears with very low parasite densities. Hb concentration was measured during morbidity surveillance and at the survey at the end of the malaria transmission season using a portable haemoglobinometer (Hemocue AB). A standard ingredients approach was used to cost the value of each unit of input needed to deliver IPTc and to treat inpatient or outpatient malaria episodes [26],[27]. We report both financial costs (additional government expenditure) and economic costs (the costs of all resources needed to deliver IPTc regardless of funding sources). For both the VHW and RCH trekking team, IPTc delivery costs were broken down by categories including IPTc drugs, non-IPTc drugs, drug dispensing, drug distribution, supervision, training, and supplies. In addition, household costs of receiving IPTc were based on a questionnaire completed by a sub sample of 15 participants attending each of the 26 maternal and child health clinic sites (n = 390). Fieldworkers were asked to stagger their interviews over the course of the day to ensure that care givers accessing IPTc at various times could be included. To reflect potential health system savings associated with fewer malaria episodes, economic costs of treating malaria were based on detailed retrospective cost data from health facilities across primary, secondary, and tertiary level health facilities in the study area between 2006 and 2007. Resource use associated with personnel, materials and supplies, equipment, transport, utilities, and buildings were recorded. Costs were identified using information found in patient folders, facility stock records, activity data collected by the district medical team, discussions with health facility personnel (both medical and administrative), and components of the study budget. Potential savings to households (direct and indirect), such as reduced expenditure on transport and other out-of-pocket expenses and potential savings in time lost from seeking treatment were collected from the families of all study children who were admitted to a hospital or health centre and from the families of a subsample of 100 children treated as outpatients. Costs are reported in 2008 US$. In order to calculate the incremental cost-effectiveness ratios, the costs of delivering IPTc via RCH trekking teams were compared with the costs of delivery by VHWs and this difference in costs was then compared to the following additional effects: (i) number of malaria episodes averted; (ii) number of children receiving the first dose of all three treatments; and (iii) number of children who received the first dose of at least one treatment. For ethical reasons, no placebo group was included in the current study so it was not possible to compare cost-effectiveness of these two strategies to current practice. Net cost effectiveness was calculated for each delivery system by subtracting resources saved from the total programme compared to the relevant outcome measure. Data from participants were recorded on forms and were checked by field supervisors, the study physician, and the data manager for consistency and accuracy. All data were entered into an SQL database using MS Access front-end software. The accuracy of data input was checked and validated using customized validation programmes. In addition, source data verification was done by the data manager. The cleaned data were converted to Stata version 10 file (Stata Corporation) prior to analysis. The primary study endpoint was the incidence of malaria (documented fever [axillary temp ≥37.5°C] or a history of fever within the previous 48 h accompanied by asexual malaria parasitaemia at a density of ≥5,000 parasites/µl) observed during the study period. Secondary endpoints were the incidence of a febrile illness with parasitaemia at any density, coverage of three IPTc treatment courses, the proportion of children who received no IPTc, the incidence of adverse events reported after IPTc administration, the mean Hb concentration, and the prevalence of asexual parasitaemia in the cross-sectional survey undertaken at the end of the malaria transmission season. The study was designed to determine whether delivery of IPTc through RCH trekking clinics would be at least as effective in preventing malaria as delivery by VHWs, and the sample size was calculated on this basis. On the basis of previous studies carried out in The Gambia, we assumed that 75% coverage with three IPT treatment courses would be achieved in both arms, that the efficacy of IPT would be 90%, and that the incidence rate of malaria by passive case detection would be 0.2 per child per transmission season without IPT. The predicted cumulative incidence of malaria would therefore be 6.5%, and we assumed that malaria incidence in each cluster would range from 0% to 20%. With 26 clusters and a cluster size of 500 children per cluster, the study was designed to have 90% power to show (using a 95% confidence interval [CI]) that the difference in cumulative incidence between the delivery methods was not more than 3.5%. Similar calculations were done for coverage of IPTc, using a noninferiority margin of 10% to 15%, and for mean Hb at the end of the transmission season, using a margin of 0.5 g/dl. An analysis plan was agreed before the data were analysed. The primary analysis included all children in the correct age range who were present in the study area at the initial census and were issued with a malaria card, regardless of the number of treatments received. Characteristics of all children enrolled in the study were tabulated by study arm, overall estimates (not adjusting for clustering) are presented. For the primary outcome of malaria incidence during the study period, analysis was by intention to treat. Time at risk was calculated from September 11 to December 8 (the surveillance period) or until date of death or emigration. All episodes of malaria were included in the calculations; only two children had two episodes of malaria. The incidence of malaria in each study arm was calculated as the total number of cases divided by the total time at risk. The point estimate of the rate difference between the two arms was the difference between these rates and a 95% CI adjusting for clustering, and strata was calculated from the variances of the stratified ratio estimates [28]. An adjusted rate difference was calculated using a two-stage approach [29]. Both individual and village-level covariates (child’s age, sex and ethnicity, and village population and distance to nearest health centre) were included in a Poisson regression model, the adjusted rate difference and 95% CI were then estimated from the model residuals. Some covariates likely to be prognostic for malaria risk, for example net use, were only measured after the intervention period and on a subset of children and were therefore not adjusted for. For the analysis of the efficacy of IPTc against clinical malaria using the case-control approach, since some cases had parasitaemias <5,000/µl on definitive reading, two case definitions of malaria were used: a febrile illness accompanied by malaria parasitaemia of any density or by malaria parasitaemia at a density ≥5,000 per µl. The efficacy of an IPTc course given within the previous 28 d in preventing malaria was estimated as 100× (1−1/OR) in which OR is the odds ratio estimated by conditional logistic regression relating case/control status to IPTc. Bednet usage (defined as usually slept under an intact or treated net, and the net could be tucked under the mattress), age, ethnicity, self-reported wealth, and distance to the nearest health centre were considered as potential confounders. In a secondary analysis, interaction with bednet usage was assessed using a likelihood ratio test, to determine whether there was any difference in the benefit of IPTc in children using bednets compared to children not using nets. Interactions with delivery method and socioeconomic variables were also assessed. Coverage with IPTc was evaluated in the cross-sectional sample survey conducted at the end of the study. The proportions of children who had received each monthly treatment course, those who received three treatment courses, and those who received no treatment were tabulated. Differences in coverage estimates between trial arms were estimated by calculating the arithmetic mean of coverage proportions in each cluster, and conducting a two-way analysis of variance, allowing for stratification. To determine whether one delivery method was better at reaching particular subgroups of the population, the coverage indicators were tabulated, stratified by wealth, and by age of each child. Wealth quintiles were calculated as quintiles of the scores on the first component from principal components analysis of household asset ratings (based on household ownership of ten items), as recorded in the cross-sectional survey. Numbers of adverse events were tabulated by study arm, and risk differences with 95% CIs were calculated adjusting for clustering and strata using analysis of variance on the arithmetic cluster means, as described above. Outcomes collected at the December cross-sectional survey were also analysed using the analysis of variance method described above. The study was approved by the London School of Hygiene & Tropical Medicine ethics committee and by the joint MRC/Gambia Government ethics committee. The conduct of the trial was guided by a Data Safety and Monitoring Board. The trial was registered on the NCHS clinical trials database (number {"type":"clinical-trial","attrs":{"text":"NCT00376155","term_id":"NCT00376155"}}NCT00376155).