Study sites
Seasonal malaria chemoprevention was implemented in Kita district in 2014. Kita is located approximately 180 km north of the capital Bamako in the western Kayes Region. The health infrastructure includes a district hospital and 47 community health centres providing primary health, including curative, preventive, social, and promotional services for a catchment area of up 15 km, usually composed of several villages or quartiers. At the village level, community health centres are supported by community health workers who are members of the community chosen by community and who have received shorter training than professional workers to provide basic mother and child health services. In 2014, the district population was estimated at 516,649 inhabitants with about 77,497 children aged 3–59 months in 336 villages/quartiers. Bafoulabe, a neighbouring district in the same Kayes Region, with highly seasonal rainfall and malaria transmission similar to Kita, was chosen as the comparison district because it is adjacent to Kita, with similar geographic, social and demographic characteristics. In 2014 the rainfall was about 854 mm in the area of Bafoulabe and 839 mm in Kita. SMC was not implemented in Bafoulabe in 2014 due to funding constraints. Bafoulabe has a district hospital and 24 community health centres; in 2014 the population was estimated at 152,976 inhabitants with 22,946 children aged 3–59 months in 336 villages/quartiers. In 2013, parasitaemia prevalence among children ages 6–59 months in the Kayes Region, where Bafoulabe and Kita are located, was 36.9% according to the recent Demographic and Health Survey [11]. Both districts received a universal bed net campaign in April 2014 (one insecticide-treated bed net per two people) and it was expected that levels of bed net coverage would be similar in the two districts. Indoor residual spraying was not done in any of the two districts nor in other districts in the Kayes Region.
Study design
A non-randomized, pre-post design was used, with an intervention district (Kita) where SMC was implemented through the health system, and a comparison district (Bafoulabe) where SMC was not implemented. SMC implementation consisted of the administration of SP + AQ at monthly intervals in children 3–59 months in August, September, October, and November 2014. During each round, children aged 3–11 months received 75 mg of AQ given once daily for 3 days plus a single dose of 250/12.5 mg of SP, while children aged 12–59 months received 150 mg AQ base given once daily for 3 days and a single dose of 500/25 mg of SP. The single dose of SP was given only on the first day, at the same time as the first dose of AQ. Tablets were crushed into water and sugar added and administered under direct observation by the health care worker. Children were observed for 30 min after drug administration, and drugs were re-administered if vomiting occurred during this period. The first day of treatment (SP + AQ) was administrated by health workers at fixed points in the village (health centres where possible or at a central point in villages where there was no health centre). The doses of AQ for the second and third days were given to parents to be administrated at home. The administration of the first day’s treatment was recorded on a special SMC card that was given to all children receiving SMC during the first round. The SMC implementation in Kita was carried out by health staff in the district, including 588 heath workers (nurses and community health workers) organized into 133 teams of 2–6 health workers each; 37 teams were based in health centres and 96 teams were deployed to fixed distribution sites in surrounding villages. The same locations were used at each of the four SMC rounds. Children between 3 and 59 months of age are eligible to receive SMC if they are afebrile, have no history of allergy to SP or AQ, not under co-trimoxazole prophylaxis for HIV infection or exposure.
Seasonal malaria chemoprevention activities were funded by the US President’s Malaria Initiative (PMI) through the Maternal and Child Survival Programme (MCSP) of the Save the Children. The National Malaria Control Programme and regional and national health directorates, in collaboration with PMI and Save the Children, led the planning and implementation of SMC. A research team of the Malaria Research and Training Centre of the University of Science Techniques and Technologies of Bamako that was not involved in the implementation conducted the surveys.
Assessment of the impact of SMC on malaria, anaemia and molecular markers of the resistance to SP and AQ
Baseline and follow-up, cross-sectional, household surveys were carried out in the intervention and comparison districts to assess the impact of SMC on malaria parasitaemia, fever, malaria illness, and anaemia. The baseline survey was performed in July 2014 prior to the start of SMC implementation and the post-intervention (follow-up) survey took place in December 2014. Thirty clusters (villages or quartiers), 15 in each district, were selected using a random sampling with probability proportional to the size of the population. In each cluster, a sample of about 38 children aged 3–59 months were randomly selected and surveyed using the WHO EPI method [12]. Households were selected first and a child in target age group was selected from each of these households. For better comparability the pre-and post-surveys were conducted in the same localities (clusters) were used for the two surveys. After informed consent was obtained, a brief clinical examination (including history of fever and temperature measurement) was performed and a finger prick blood sample was collected for a blood smear, rapid diagnostic test (RDT) (in case of fever or symptoms suggestive of malaria), measurement of haemoglobin concentration, and filter paper for molecular markers of SP and AQ resistance. Children with a positive RDT were treated with artemether–lumefantrine, in line with current national case management guidelines. Children with moderate severe anaemia (haemoglobin < 8 g/dL) or with signs of severe malaria were referred to the nearest health centre for management. Malaria infection was defined as presence of malaria parasitaemia by blood smear. Fever was defined as axillary temperature ≥37.5 °C, and malaria illness was defined as fever and presence of asexual malaria parasitaemia by blood smear.
Assessment of adherence to SMC and frequency of adverse events
Cross-sectional surveys were carried out 4–7 days after each of the four rounds of SMC to assess caregivers’ adherence to the administration of SMC drugs and determine the frequency of adverse events in the intervention district of Kita. Adherence surveys were conducted in 10 clusters (localities) selected randomly among the 15 already selected for the pre- post-intervention surveys. The same localities (clusters) but different households were surveyed after each round. The main indicators assessed through caregiver interview and verification of the SMC card included: the proportion of children who received the second and third dose of AQ at home, the proportion who spat out or vomited the treatment, the nature and frequency of the adverse events, and opinions of parents/caregivers regarding the intervention.
Assessment of SMC coverage
During the post-SMC survey in Kita, 15 additional clusters (villages/quartiers) were added to the 15 surveyed at baseline, yielding a total of 30 clusters for the coverage survey in Kita. Parents or guardians of children in the target age range at the time of the SMC administration were questioned using standardized questionnaires about receipt of SMC. In addition, information from SMC cards was recorded for each child, when available. Interviews in each household were conducted by two trained interviewers in the local language. Primary indicators assessed from the coverage survey included administration of each day of the 3-day SMC treatment course during each of the four rounds and the use of insecticide-impregnated bed nets (ITN) during the previous night. Completed questionnaires were verified at the end of each day by a supervisor and corrected if necessary. Coverage of SMC at each round was defined as the proportion of the children aged 3–59 months at the time of SMC who received the 3 days’ treatment of SMC during that specific round. Full SMC coverage was defined as the proportion of children aged 3–59 months who received the complete 3-day treatment course during all four rounds of SMC.
Impact on malaria morbidity using health records data
The study team initially intended to use routine data reported from health centres in the intervention and the comparison districts to assess trends in malaria cases, but these data were too incomplete. For example, data from several weeks (up to 3 months) were not reported for several health centres. For this reason, the impact on malaria morbidity was assessed using routine data on confirmed malaria cases extracted from the registers by the research team in nine of the 47 community health centres in Kita and 7 of the 24 health centres in Bafoulabe; these health centres were randomly selected from among the 15 health areas of pre- post-intervention surveyed in each of the two districts. These data on confirmed malaria, uncomplicated and severe cases for 2013 and 2014 were collected from registers in February 2015.
Laboratory analysis
Rapid diagnostic test (RDT)
Malaria RDTs were performed in the field according to the manufacturers’ recommendation using the same RDTs used in the primary health centres (SD Bioline Plasmodium falciparum). These RDTs can detect infection with Plasmodium by detecting histidine-rich protein 2 antigen specific to P. falciparum (PfHRP2). RDT results were given to caregivers of children who were tested; those testing positive received malaria treatment according to national guidelines in Mali.
Haemoglobin measurement
HemoCue® machines (Angelholm, Sweden) were used to measure haemoglobin (Hb) concentration in children using capillary blood obtained by finger prick.
Thick and thin blood smears
Thick and thin blood smears were prepared in the field for participating children in the pre-post intervention surveys. Thin smears were fixed in methanol. All slides were air dried and stained using Giemsa stain 3% for 45–60 min the field, and transferred to the Malaria Research and Training Centre (MRTC) laboratory for reading and storage. The stained thick and thin blood films were examined microscopically at a magnification of 1000× to identify the parasite species and to determine the parasite density. Parasite density was assessed by experienced certified microscopists unaware of the intervention and comparison groups, by counting the number of asexual parasites in a set number of white blood cells (typically 200) with a hand tally counter according to established standard operating procedures, assuming 8000 white blood cells (WBC)/µL.
Dried blood spots and polymerase chain reaction
Filter paper samples were air dried and stored individually in small Ziploc bags with a desiccant and transferred to the MRTC laboratory in Bamako for analysis. Molecular analysis was performed on samples from children whose blood smear showed P. falciparum parasitaemia ≥160/uL (4 parasites/200 WBC) by blood smear. Samples were analysed by nested polymerase chain reaction (PCR) and/or PCR-restriction fragment length polymorphism (RFLP) for mutations at codons 51, 59 and 108 of the dhfr gene, 437 and 540 of the dhps gene, mutations at codon 76 in the P. falciparum chloroquine transporter gene (pfcrt), and at codon 86 of the P. falciparum multidrug resistance gene one (pfmdr1) according to published methods [13, 14]. Cases of mixed infection (wild type and mutant) were categorized as mutant. Quintuple mutant was defined as the presence of the three dhfr mutations (N51I, C59R and S108N) and the two dhps mutations (A437G and K540E).
Sample size
Household survey
Based on a meta-analysis of SMC that found a reduction of parasitaemia prevalence of 53% in trial settings [7], and assuming a 40% reduction under programmatic conditions for this study. A sample size of 556 children per district (total of 1112 children) was required to detect a 40% difference in parasitaemia between the SMC and non-SMC districts at follow-up (15 and 25% parasitaemia, respectively), with an alpha = 0.05, power of 80%, a design effect of 2.0, and 10% non-response.
Adherence surveys
The adherence surveys were powered to estimate an adherence rate (completing all three doses of SP − AQ) of 80%, within a 95% confidence limit of ±10%. Assuming a design effect of 2 and a 10% non-response rate, 156 children were required after each round. To achieve this sample size, 10 clusters were selected in Kita, and 20 households per cluster were visited and surveyed after each SMC round.
Coverage survey
Assuming a SMC coverage of 50%, with an alpha of 5%, design effect of 2 and non-response rate of up to 10%, a total of 840 children were required for the coverage survey. To achieve this sample size, 30 clusters were randomly selected in Kita and 38 children were surveyed in each cluster.
Management and analysis of data
Survey data were collected on standardized forms, entered into a MS Access database, and analysed using Stata (version 12.1). As described above, data from health facility registers were extracted from the registers in selected heath centres in intervention and comparison districts. Analyses from household surveys were adjusted for cluster sampling, using the survey commands in Stata. Logistic regression was used for binary outcomes of anaemia and parasitaemia, and used a difference-in-differences approach, which can be useful in estimating impact from observational study designs [15]. These regression analyses were adjusted for potential confounding variables, age, gender, ITN use, and the cluster design. The level of significance was set at 5%.