correspondence Dr Sodiomon Bienvenu Sirima, Centre National de Recherche et de Formation sur le Paludisme, Ministère de la Santé, 01 BP 2208, Ouagadougou 01, Burkina Faso, West Africa. Tel: +226 32 4695/96; Fax: +226 31 0477; E-mail: email@example.com
In rural, malaria-endemic Burkina Faso, we evaluated the impact of the use of pre-packaged antimalarial drugs (PPAM), by mothers in the home, on the progression of disease in children from uncomplicated fever to severe malaria. In each village of one province, a core group of opinion leaders (mainly older mothers) was trained in the management of uncomplicated malaria, including the administration of PPAM. Full courses of antimalarial (chloroquine) and antipyretic (aspirin) drugs were packaged in age-specific bags and made widely available through community health workers who were supplied through the existing drug distribution system. Drugs were sold under a cost-recovery scheme. Local schoolteachers conducted surveys in a random sample of 32 villages at the end of the high transmission seasons in 1998 and 1999. Disease history and the treatment received were investigated for all children under the age of 6 years having suffered from a fever episode in the previous 4 weeks. ‘Uncomplicated malaria’ was defined as every episode of fever and ‘severe malaria’ as every episode of fever followed by convulsions or loss of consciousness. During the study period, 56%[95% confidence interval (CI) 50–62%] of 3202 fever episodes in children under 6 years of age were treated promptly by mothers with the pre-packaged drugs made available by the study. A total of 59% of children receiving PPAM were reported to have received the drugs over the prescribed 3-day period, while 52% received the correct age-specific dose. PPAM use was similar among literate (61%) and non-literate mothers (55%) (P = 0.08). The overall reported risk of developing severe malaria was 8%. This risk was lower in children treated with PPAM (5%) than in children not treated with PPAM (11%) (risk ratio = 0.47; 95% CI 0.37, 0.60; P < 0.0001). This estimate of the impact of PPAM was largely unchanged when account was taken of potential confounding by age, sex, maternal literacy status, year or village. Our findings support the view that, after appropriate training and with adequately packaged drugs made available, mothers can recognize and treat promptly and correctly malarial episodes in their children and, by doing so, reduce the incidence of severe disease.
Delivery of prompt and adequate treatment for uncomplicated malaria at the community level remains a key strategy to reduce the burden of malaria in sub-Saharan Africa (WHO 1993) where, in the past, treatment for uncomplicated malaria has rarely been sought at health facilities and treatment provided at home has often been inappropriate or delayed (Dabis et al. 1989; Deming et al. 1989; Ruebush et al. 1995; Munguti 1998) This strategy is based on the assumption that adequate treatment, delivered at home by mothers soon after the appearance of symptoms, will result in a reduction in malaria morbidity and mortality (Greenwood et al. 1991) with a very low cost-effectiveness ratio (Goodman & Mills 1999; Goodman et al. 1999). Several studies have shown that community-based management of fever episodes in children can improve treatment compliance (Greenwood et al. 1991; WHO/TDR 1997). One randomized controlled trial, performed in Ethiopia, found evidence of a reduction in child mortality after an intervention to encourage prompt and adequate home treatment of suspected malaria (Kidane & Morrow 2000). However, an earlier observational study, performed in Kenya, found no evidence of a reduction in child mortality following an intervention to make chloroquine (CQ) readily available at the household level (Spencer et al. 1987). Thus, the evidence regarding the impact of prompt, presumptive household treatment of suspected malaria on mortality is generally weak.
To improve treatment of uncomplicated malaria at the household level, a strategy has been designed and introduced in several provinces of Burkina Faso since 1994. As part of this strategy, health personnel and members of the community have been trained to use pre-packaged antimalarial drugs (PPAM) supplied to health centres and villages. A first study, performed in 1994–95 and based on routine health system data (Pagnoni et al. 1997), provided circumstantial evidence of a beneficial effect of the intervention on severe malaria morbidity. The present study represents a continuation of that work, with the aim of evaluating the impact of the strategy on severe malaria morbidity.
Materials and methods
Study area and population
The intervention was implemented in the province of Boulgou, in south-eastern Burkina Faso. The province covers an area of 6992 km2 and has an estimated population of about 415 000 inhabitants, of whom approximately 75 000 are children younger than 6 years, living in 375 villages and hamlets. The majority of the population of the province lives by subsistence farming and belongs to three main ethnic groups, Bissa (65%), Mossi (20%) and Peulh (9%). The most widely spoken language is Bissa. Malaria is hyperendemic in the province. The transmission pattern is seasonal, reaching a peak in the second half of the year coinciding with the rainy season, which lasts from June to October. The average annual rainfall over the years 1996–2000 was 780 mm. Plasmodium falciparum is the main parasite responsible for malarial infections. CQ is the first line treatment for malaria with a clinical failure rate varying between 6% and 16% in Burkina Faso (Guiguemde et al. 1994).
The implementation of the strategy followed the steps described previously (Pagnoni et al. 1997). Information/sensitization meetings and training sessions were organized in all villages of the province. Training was provided to several categories of individuals. At health centres, nurses (as future trainers) and drug store managers were trained. At the community level, community health workers (CHWs) and village opinion leaders (mainly older mothers) were trained. The training of community members focused on the recognition of the symptoms of uncomplicated malaria, those of gravity, and those warranting referral to the health centre, and the use of PPAM. CHWs and village opinion leaders were also trained to act as intermediaries between the health services and other members of the community and to provide advice on treatment and referral when needed. Emphasis was placed on the responsibility of the primary caretaker of the sick child at household level. As in the vast majority of cases, the primary caretaker is the mother, the term ‘mother’ will be used in the text to refer to this individual.
In accordance with the Bamako Initiative policy in force in the country, all health centres in Boulgou have an essential drug store within their premises. In June 1998, they were supplied free of charge with a first stock of CQ, aspirin, labels and plastic bags to initiate the study. The drugs were then packaged locally by the drug store managers. Four different colour coded packs were produced, each containing a full course of treatment targeted at one of four age groups (0–6, 7–11, 12–35 and 36–83 months, see Figure 1). Each pack also contained a label with pictorial instructions on how to administer the drugs. CHWs were then provided with a seed stock of PPAM free of charge. The CHW sold the drugs to mothers at a price previously agreed with the local health management team, calculated to allow for full recovery of the drugs' purchase costs, for a 10% incentive margin for the CHW and, in some cases, for an incentive for the drug store manager in charge of packaging the drugs. The price of PPAM varied between 0.05 and 0.12 US$, depending on the age group, compared with a total of 0.7 US$ for the CQ and paracetamol routinely sold in the drug stores in packages of 20 tablets each. With the sale of the first stock of drugs, the CHW set up a revolving fund that allowed him/her to replenish the stock by buying new pre-packaged drugs at the health centre drug store. In order to avoid the use of PPAM by adults, sufficient quantities of CQ and aspirin in 1000 tablet tins were also delivered to CHWs.
The nurses in the health centres were responsible for the supervision of the CHWs, and were in turn supervised by the project team in collaboration with the District Health Team. The project team also visited randomly selected CHWs at regular intervals.
In two successive years 1998 and 1999, cross-sectional surveys were conducted in October, towards the end of the high transmission season, in the same 32 randomly selected villages. The survey personnel visited all households in which a child aged less than 6 years was living, and a standard questionnaire was administered to all mothers identified as having a child who had suffered from fever during the previous 4 weeks. The following operational definitions were adopted: uncomplicated malaria as every episode of fever or ‘hot body’ presumably as a result of malaria (memalé in the Bissa language), severe malaria as every episode of uncomplicated malaria followed by convulsions or loss of consciousness (benoyaba in Bissa). To minimize biases in answers, non-medical persons who had not participated in the implementation of the strategy (local schoolteachers) were employed to conduct the household surveys. The socio-economic status of the family, the occurrence of uncomplicated malaria alone or with the later appearance of severe malaria, the prompt use of PPAM (within 24 h from onset of symptoms), and the correctness of the treatment with regards to the dose per age and the duration were noted. Children whose disease started abruptly with convulsions or loss of consciousness, not preceded by uncomplicated malaria, or who were unable to take drugs per se from the start of the episode were excluded from the evaluation.
Data were checked and then entered onto a computer using epi-info (version 6.04) (CDC, Atlanta, USA and WHO, Geneva, Switzerland). Data analyses were performed using Stata version 7.0 (http://www.stata.com). The proportion of children progressing from uncomplicated to severe malaria was compared in children treated promptly with PPAM and the remaining children. Crude and adjusted (Mantel–Haenszel) risk ratios with 95% confidence intervals (CI) were computed. Statistical tests used to compare the two groups included the chi-square test and the Kruskal–Wallis test. Conditional logistic regression was used to control potential confounding variables and to look for interactions.
Ethical clearance and informed consent
The study received ethical clearance from the Ministry of Health, Burkina Faso. Village meetings were held prior to the start of data collection during which the objectives of the study were explained. During the cross-sectional surveys, verbal consent was obtained from household heads before administration of the questionnaire.
During the two surveys we indentified 3202 mothers with a child under 6 years of age who had suffered an uncomplicated malaria episode during the 4 weeks preceding the interview (1551 in the 1998 survey, 1651 in 1999). Information on the prompt use of PPAMs was obtained from all of them. The vast majority of the interviewed mothers (99.3%) were housewives and subsistence farmers; 292 (9.1%) were literates. The mean age ± SD of recruited children was 2.4 ± 1.4 years; the male:female sex ratio was 1.07. Data on use of PPAM are summarized in Table 1. In all 1806 episodes (56%; 95% CI 50–62%) were treated prompthy with PPAM. There was strong evidence that use of PPAMs varied between villages (minimum = 30%, maximum = 84%; P < 0.001), and this variation correlated with ratio of CHWs:population in the village (correlation coefficient = 0.9; P < 0.0001). The use rates of PPAM were very similar in the 2 years (58% and 55%; P = 0.11). Over the 2-year period, there was no evidence that use of PPAM varied with the age of the child (P = 0.22) or the sex of the child (P = 0.31). PPAM use appeared to be similar among literate (61%) and non-literate mothers (55%) (P = 0.08).
Table 1. Use of pre-packed antimalarials (PPAM) by age and sex of the child and educational level of the mother and year
Number of children with uncomplicated malaria
Number (%) receiving PPAM
Age in months
Sex of the child
Detailed information on mothers' use of the PPAM, such as duration and correctness of the dose, was available for 1583 of the 1806 episodes treated with PPAM (88%). Fifty-nine per cent of treated children were reported to have received the drugs over the prescribed 3-day period, while for 18% of children the administration of drugs was stopped earlier, and for 23% it extended beyond the third day. Fifty-two per cent of children treated with PPAM received the correct age-specific therapeutic unit, while 31% were treated with a unit targeted at a younger child (thus resulting in under dosing) and 17% with a unit prepared for an older child (thus resulting in an overdose of drugs).
The overall reported risk of developing severe malaria was 8% (246 of 3202). This risk of progression was lower in children treated with PPAM (5%) than in children not treated with PPAM (11%) (Table 2: risk ratio = 0.47; 95% CI 0.37, 0.60; P < 0.0001). The corresponding odds ratio is 0.44 (95% CI 0.34, 0.58). After taking into account whether or not the child was treated with PPAM, there was no evidence that risk of progression to ‘severe malaria’ was associated with the sex of the child (P = 0.54), the literacy status of the mother (P = 0.37) or the year of the survey (P = 0.29). There was evidence that the risk of progression varied with the age of the child (P = 0.008), with the risk peaking in the second 6 months of life (Table 2). The estimate of the impact of PPAM (the risk ratio) did not change when account was taken separately of potential confounding by age, sex, maternal literacy status or year (Table 2). In addition, we used conditional logistic regression to control simultaneously for any confounding effects of these four variables and village. This analysis produced an adjusted estimate of the odds ratio of 0.47 (95% CI 0.35, 0.64; P < 0.0001), very similar to the crude estimate (0.44), indicating the absence of any important confounding by any of these variables. Nor was there evidence that any of these variables modified the effect of PPAM use (P > 0.3 in all cases).
Table 2. Risk of progression from uncomplicated malaria to ‘severe malaria’ by age and sex of the child, literacy status of the mother and calendar year and by whether or not the child received PPAM
Crude risk ratio or Mantel–Haenszel risk ratio adjusted for the variable in the left hand column.
0.47 (0.37, 0.60)
Age in months
0.47 (0.36, 0.60)
0.47 (0.37, 0.60)
0.47 (0.37, 0.60)
0.47 (0.36, 0.60)
Finally, there was no evidence that the risk of progression to ‘severe malaria’ differed between children who received the correct rather than an insufficient age-specific dose of PPAM (5.3% and 6.2%, respectively; P = 0.63), or between children treated over the correct rather than an incorrect number of days (4.9% and 6.9%, respectively; P = 0.23).
We evaluated the impact of prompt home diagnosis and treatment of uncomplicated fever cases with a full course of PPAM on the risk of progression to some severe forms of disease. We did not perform a randomized controlled trial, but rather an observational study based on mothers' recall. Our results suggest that in our study population the prompt use of PPAM reduced the risk of developing severe forms of disease including loss of consciousness and/or convulsions by about 50%, and that the data are compatible with reductions ranging from about 35% to 65%. This finding is very unlikely to be as a result of chance (P < 0.0001).
The study has a number of limitations. First, evidence for the positive impact of PPAM is drawn from the reported incidence of disease and reported use of PPAM based on mothers' recall. A potentially serious source of bias is the fever episodes which were not reported by mothers. If mothers were more likely to forget (and thus not report) those fever episodes which were mild and for which no treatment (including PPAM) was given than those for which treatment was given (including PPAM), then our results will overestimate the impact of PPAM. To explore the potential impact of this bias we considered an ‘extreme’ situation in which mothers forgot 20% of mild episodes which were not treated with PPAM, but remembered all mild episodes treated with PPAM. Under this scenario, the risk ratio would become 0.57 (instead of 0.47) but remains highly and statistically significant (P < 0.0001). For the association to lose statistical significance we would have had to have missed more than 40% of mild episodes in untreated children without missing any in treated children. We think that such an extreme scenario is very unlikely. Restricting the analysis to children reported to have had a fever episode in the 3 days preceding the interview (reducing the likely extent of any such recall bias) produced a risk ratio estimate for PPAM use of 0.41, similar to the estimate obtained when all children were included.
There is also potential for both non-differential misclassification and systematic recall bias with respect to disease severity and PPAM use among reported episodes. Non-differential misclassification will tend to mask any impact of PPAM on disease severity. It is therefore unlikely to account for the impact we have observed. If mothers whose children developed severe disease were less likely to remember using PPAM than mothers of children with uncomplicated disease, this could account for our findings. This seems unlikely. On the contrary, one might expect mothers of children who were seriously ill to recall better the events surrounding the episode. Furthermore, although it has been shown that the traditional definitions of disease we used largely coincide with uncomplicated and cerebral malaria, respectively (Bonnet 1986; Faisang 1986; Coulibaly et al. 1991), none of the children had parasitological confirmation of disease. Although malaria is considered to account for a large proportion of febrile diseases, especially during the rainy season, other febrile illnesses certainly occur in the community, such as acute respiratory diseases. Thus, our study is likely to have included children who were not actually ill with malaria. If these children were treated in the same way as genuine cases, then inclusion of these cases will have tended to mask any impact of PPAM on risk of progression to severe malaria and certainly could not account for the impact we observed. Finally, the study does not provide any information on the impact of PPAM on clinical forms of severe malaria other than the cerebral syndrome, for example, those characterized by severe anaemia, which are known to be important in a rural setting (Snow et al. 1997; Modiano et al. 1998, 1999).
A number of confounding factors may have influenced our estimate of the impact of PPAM on severe malaria morbidity, including the educational and/or socio- economic status of mothers. We controlled for the effects of age, sex, maternal literacy and calendar year and found no indication that any of these factors confounded the association between use of PPAM and risk of severe disease. We cannot, however, rule out the possibility that mothers and children using PPAM differ with respect to other risk factors for severe disease from mothers and children who do not.
The treatment of fever episodes with PPAM was, in general, well accepted by the communities in our study (overall utilization rate of 56%), by both literate and non-literate mothers and for children of all age groups within the range studied. The large variation in PPAM utilization rates observed between villages can be largely explained by variations in the ratio of CHWs:population. Villages with a high ratio had a high PPAM utilization rate. The overall rate recorded in this study is slightly higher than the utilization rate of 46% we reported in the earlier study of the same strategy performed in another province of Burkina Faso (Pagnoni et al. 1997). Previous studies have shown an association between a low educational level of the mother and a low compliance with non-prepackaged home treatment for malaria (Mulumba et al. 1990). We believe that in our study, the simplicity of the pre-packaged treatment and the training received by mothers and CHWs helped to minimize the barriers to home treatment for non-literate persons. The low cost of the PPAM, calculated to cover the purchase costs of the drugs and to allow for a small profit margin for the CHW, and their wide availability in every village may also have contributed to a relatively high proportion of mothers accepting the strategy. This proportion might have been further increased if a higher number of CHWs per village had been used. Although not specifically investigated in this study, we expect that the direct costs incurred by the families in this study were not substantially different from those incurred during our first study of the strategy.
Despite the pre-packaging of the drugs, 17% of the children were treated with a greater than recommended dose of antimalarial. This may not be a major concern with CQ, which has relatively few serious adverse reactions and is very cheap. It will be more of a concern in settings where the first line drug is something else, such as sulphadoxine–pyrimethamine or combination therapy with an artemisinin drug. An improved system of pre-packaging the drugs (e.g. blister-packaging with separate sections for each day of treatment) might help to reduce this problem and improve overall compliance with the correct regimen.
The treatment received by children who did not receive PPAM has not been evaluated. It is possible that some of these children received effective antimalarial treatment. However, the quality of antimalarial treatment administered at home in the absence of PPAM has been shown to be poor in other settings (Deming et al. 1989; Ruebush et al. 1995), and the proportion of children treated correctly was only 3% in our baseline study of knowledge, attitudes and practices of mothers in another province of Burkina Faso (Pagnoni et al. 1997). The use of effective treatments other than PPAM, if widespread, would be expected to reduce the observed impact of PPAM.
The results of this study are consistent with those of an earlier study of the same strategy using routine data. They show a strong association between the use of PPAM and reduced risk of one of the commonest forms of severe malaria. If PPAM really do reduce the risk of progression to the severe forms of malaria that we studied, it is very plausible that they reduce mortality. In Ethiopia, in a randomized controlled trial, training mothers to recognize presumptive malaria episodes and treat them with an appropriate course of CQ was associated with a 40% reduction in all-cause child mortality (Kidane & Morrow 2000). Despite the limitations of our observational study design, our findings appear compatible with those from Ethiopia. They support the view that, after appropriate training and with adequately packaged drugs made available, mothers can recognize and treat promptly and correctly malarial episodes in their children and, by doing so, reduce the incidence of severe disease and thus mortality.
We thank the population of Boulgou province and all the CHWs who participated in the study, and all the schoolteachers and health staff of Tenkodogo district for their assistance in the implementation of the study. We also thank Dr Saïdou Ouili, Past Director of C.N.R.F.P. Ouagadougou, Burkina Faso, Dr Jane Kengeya-Kayondo, Manager of the Task Force on Malaria Home Management, OMS/TDR Geneva Switzerland, and the Italian Co-operation for their technical and financial support. This investigation received financial support from the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases (TDR).