By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Adoption of the new antimalarial drug policy in Tanzania – a cross-sectional study in the community
Division of Clinical Pharmacology, Department of Laboratory Medicine, Karolinska Institutet at Karolinska University Hospital, Huddinge and Division of International Health (IHCAR), Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
Authors Jaran Eriksen, Division of Clinical Pharmacology, C1 68, Karolinska Institutet at Karolinska University Hospital, Huddinge, 141 86 Stockholm, Sweden. Tel.: +46-9-5858 1198; Fax: +46-8-5858 1070; E-mail: jaran.Eriksen@labmed.ki.se (corresponding author). Stephen E.D. Nsimba, Department of Clinical Pharmacology, Muhimbili University College of Health Sciences (MUCHS), Dar-es-Salaam, Tanzania. Omary M.S. Minzi, Department of Clinical Pharmacology, MUCHS, Dar-es-Salaam, Tanzania. Anku J. Sanga, Department of Sociology, University of Dar-es-Salaam, Tanzania. Max Petzold, Nordic School of Public Health, Gothenburg, Sweden and Division of International Health (IHCAR), Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden. Lars L. Gustafsson, Division of Clinical Pharmacology, Department of Laboratory Medicine, Huddinge University Hospital, Karolinska Institutet, Stockholm, Sweden. Marian Y. Warsame, Division of International Health (IHCAR), Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden. Göran Tomson, Division of International Health (IHCAR), Department of Public Health Sciences, Karolinska Institutet and Medical Management Centre (MMC), Karolinska Institutet, Stockholm, Sweden.
Objective To assess the diffusion of the change of first line antimalarial drug from chloroquine (CQ) to sulphadoxine/pyrimethamine (SP) at household level in a rural district of Tanzania less than a year after the policy implementation.
Methods Caretakers in 729 households were interviewed on knowledge of the new policy, home stocking of antimalarials, home-treatment practices of children younger than 5 years with fever, health-seeking behaviour and experience of SP. SP and CQ levels in blood were analysed from 328 children younger than 5 years in the households. Twelve focus group discussions (FGD) were performed with mothers, fathers and health workers.
Results About 51% of the population knew that SP was the first line antimalarial. Only 8% of mothers stocked antimalarials, and only 4% stated self-treatment as the first action. We estimated that 84% of the children who had had fever during the last 4 weeks sought care at public health facilities. SP was detectable in 18% of the total child population and in 32% of those with reported fever, CQ in only 5% and 7%, respectively. The FGDs revealed negative perceptions of SP and fear of severe adverse reactions with mass media reported as key informant.
Conclusion The policy had diffused to the communities in the sense that CQ had been changed to SP, which was well known as first line treatment. Moreover, there was a reported dramatic change from self-treatment with CQ to seeking care at public health facilities where SP was given under observation.
Every year about 2 million people die from malaria worldwide, 90% of them are children younger than 5 years living in sub-Saharan Africa (WHO 1993; Breman 2001). For decades, chloroquine (CQ) was used as first line treatment. Because of rapidly spreading resistance, many African countries changed their malaria treatment policies to the sulphadoxine/pyrimethamine (SP) drug combination. Malawi was the first in 1993, followed by South Africa and Kenya (Shretta et al. 2000). Tanzania, faced with an average CQ treatment failure rate of 42% already in 1999 (Kitua et al. 1999), changed its national malaria treatment policy from CQ to SP as first line treatment in August 2001, officially banning the use of CQ. The change of policy was a long and complex process involving key stakeholders and decision-makers at many different levels of society (Williams et al. 2004).
Inappropriate and extensive use of SP causes the drug to quickly loose efficacy, leaving more expensive drugs as the only treatment alternatives (Schapira et al. 1993; Ronn et al. 1996). A recent study in Tanzania shows that even before introducing SP as first line drug there was severe resistance at the molecular level, a possible sign that clinical resistance will develop fast (Eriksen et al. 2004). At the same time access to drugs, including antimalarials and antibiotics, close to the home is now advocated by many (WHO 2002, 2004; Sazawal & Black 2003). This has to be balanced with the risk of promoting spread of drug resistance (Kamya et al. 2002).
Combination therapy (CT), in particular artemisinin-based combinations (ACT), is currently the strongest option for improving cure of falciparum malaria (2003) but introducing this treatment strategy in low-income countries is a great challenge to their inadequate health systems, not only from a financial point of view (White et al. 1999; Bloland et al. 2000; Snow et al. 2003; Attaran et al. 2004). The policy cycle necessary for a rational policy formulation and implementation is complex and encompasses many steps, of which the implementation itself is a late stage of the process (Williams et al. 2004). This has previously been very poorly studied in Sub-Saharan Africa. We studied how communities in a Tanzanian rural district adopted the new policy less than a year after the implementation of a new first line antimalarial drug.
Material and methods
The study was conducted in Kibaha, a rural district 40 km northwest of Dar es Salaam, Tanzania. The district is holoendemic for malaria with a population of 132 443 people (NBoST 2003).
Selection of study population
The administrative structure of the Tanzanian districts formed the basis for a three-stage random sampling procedure used to select wards1, hamlets2 and households. The selection was made non-proportionally to the population size.
Of the total nine wards, four were classified as urban and five as rural. Two wards of each category were randomly chosen. In these, a total of 20 hamlets were randomly chosen. Assuming a home stocking frequency of antimalarial drugs in the homes of 60% (Mlavwasi 1994) and using a 5% significance level we calculated a needed sample size of 730 households (including 20% drop-outs). From each of the 20 hamlets 36 or 37 households were randomly selected. Although the wards in Kibaha are classified as either rural or urban, the difference is not as large as one might expect. Apart from the population density being higher in the urban wards, the major difference consists of the proximity to the main road from Dar es Salaam. The urban wards are all located along this road, receiving more merchandise and information from the large towns of Tanzania.
To complement the findings from the quantitative survey, 12 focus group discussions (FGDs) with six to 11 participants each were held, four with mothers, four with fathers and four with health workers. The mothers and fathers were divided into age groups of either 18–30 or 31–50 years. They had all completed primary education only. The participants of these groups were selected with the help of the ward secretaries. The health care workers groups were divided into one group of either clinical officers or assistant clinical officers, one with trained nurses and finally two groups consisted of either nurse assistants or mother and child health aids.
We used a semi-structured questionnaire (available on request from first author) consisting of closed- and open-ended questions addressing home stocking of antimalarial drugs, home-treatment practices of children younger than 5 years with fever, source and knowledge about antimalarials, especially SP, and knowledge of the recent policy change. The questionnaire was translated into Swahili and pre-tested. All main caretakers of households were interviewed and in households with children younger than 5 years with fever, his/her mother was interviewed if she was present. One part of the questionnaire concerned the last febrile episode of the child and was only addressed to households with children younger than 5 years with fever. A recall period of four weeks was used. In households with more than one such child (Table 1), the questions were asked concerning the child for whom the most recent episode of febrile illness was reported.
Table 1. Socio-demographic characteristics of interviewees in 729 households (397 households with under-five children and 332 households without underfives)
Respondents (n = 729)
* Includes incomplete primary education, complete primary education, secondary and higher education, adult education and education in Muslim primary school (madrasa).
Any kind of education*
Distribution of underfives (493 in a total of 729 households)
After obtaining verbal consent from mothers/guardians, capillary blood samples were collected from the underfive whose febrile episode the interview had concerned. Two blood samples of 100 μl each were collected on filter paper and analysed for the determination of CQ and sulphadoxine (S) in blood concentrations, the latter representing SP.
Focus group discussions:
After the household interviews, thematic guides for the FGDs were developed, including self-treatment practices with antimalarials, perceptions of SP and the antimalarial drug policy change. The discussions were moderated by one of the co-authors who is a sociologist (AS). Two note takers and one observer from the research team were also present (JE or SN). All discussions were tape recorded and later transcribed by two of the authors (AS and SN).
The household survey was performed from January to March 2002 and the FGDs in July and August the same year. The peak malaria seasons occur in November and June, respectively.
The blood samples were analysed for blood drug levels using two different HPLC methods for CQ (Minzi et al. 2003b) (at Division of Clinical Pharmacology, Huddinge, Sweden) and sulphadoxine (Bergqvist et al. 1987) (at Department of Clinical Pharmacology, Muhimbili University College of Health Sciences, Tanzania).
The responses to the closed and open-ended questions in the household interviews were coded, entered, processed and analysed using the Statistica® and Stata® software. Logistic regression with systematic inclusion and exclusion based on significances was performed with the following predictors: level of education of the interviewee, age of the interviewee, sex of the interviewee, whether there was an underfive living in the household, whether the household had an child younger than 5 years with a reported febrile episode during the last 4 weeks, whether the household was in a rural or urban location. In the sequel population estimates, obtained by weighted analysis to compensate for the non-proportional sampling, are reported. In the analysis 5% significance level was used and 95% confidence intervals are given in the results.
The FGDs were analysed for content by applying codes within the structure of the thematic guide. Additional themes that emerged from the material were also coded and analysed using the same approach. The data was then structured into categories and put into a grid before identifying variables and associations between variables (Green 1998; Hardon et al. 2001).
Participation was voluntary and informed consent was obtained from all household and FGD participants. Ethical approval was obtained from the human ethics committees of the Muhimbili University College of Health Sciences in Dar es Salaam, Tanzania and the Karolinska Institutet in Stockholm, Sweden (D-nr: 426/01). Approval was also obtained from the Kibaha district, Coast Region administrative authorities, ward secretaries and leaders of the hamlets.
We interviewed caretakers in 729 households, 397 of these were households with children younger than 5 years. The mean age of the interviewees was 39.3 years. On average each household had a family size of five persons (Figure 1). We found 493 children younger than 5 years in 397 households (Table 1), but from each household only the child with the most recent episode of febrile illness was included in the study. The number of boys and girls was equal.
Drug stocking, health seeking behaviour and self-treatment
Eight per cent (4.7–12.1) of all households presented home-stocked antimalarials. The most common were quinine, SP and CQ. As many as 42% of households stated having previously stocked antimalarials, the odds of this being higher in urban wards than rural ones [OR = 7.3 (4.2–12.8)] and in homes where the caretaker had some kind of formal education compared with those without formal education [OR = 3.0 (1.3–6.6)].
Blood samples were taken from one child younger than 5 years in 328 of the 397 households with young children (Table 2). It was possible to assay 321 blood samples for sulphadoxine (S). We estimated that 18% (11.3–24.4%) of the children had detectable S levels up to 332 μm (suggested therapeutic range = 100–300 μm) (Eriksen et al. 2004). Three per cent of the children had detectable S in the blood, but we were unable to quantify the levels of S as they were lower than the 25 μm quantification limit of the analytical method. Among the 320 samples assayed for CQ, it was estimated that only 5% (1.9–8.4) had detectable levels of the drug and only 1.5% had levels above the quantification limit of the method used (>100 nm), the highest detected level being 1803 nm (suggested therapeutic concentration around 1000 nm) (Eriksen et al. 2004). In 2% (−0.2–4.4) of the children both CQ and S could be detected.
Table 2. Detected sulphadoxine (S*) or chloroquine (CQ) in blood samples from 328† underfives selected from 729 households. In 116 underfives caretakers reported that a febrile episode had occurred within 4 weeks prior to interview
CQ detected in blood
S detected in blood
S and CQ detected in blood
No drugs detected in blood
The figures in the text referring to this table are a weighted population estimate.
* SP measured as sulphadoxine (S).
† Samples from 328 different children could be analysed (320 were assayed for CQ and 321 for S in their blood).
Total reported within last 4 weeks
0–14 days ago
15–28 days ago
Total reported more than 4 weeks ago
Children with recent febrile episode
A total of 160 children younger than 5 years were reported to have had fever episodes during the last 4-week period and 80% (68–91) of the population was estimated to have been given drugs at home; 75% analgesics and a small proportion [4% (0.2–7.2)] antimalarials. Most mothers (88%) reported also to have sought care outside the home, mainly from public health facilities (84% of those who sought care) (Table 3). There were no significant predictors for care-seeking behaviour. Among the children whose mothers stated to have sought care outside the home, 30% had detectable levels of SP in the blood (Table 3). Of the children who were reported to have had fever within the last 4 weeks, 32% (23.5–40.5) and 7% (2.4–11.6) had detectable levels of sulphadoxine and CQ, respectively (Table 2). At the health facilities, 70% claimed to have been given antimalarials. We found that 51% of those stating to have been given an antimalarial had detectable levels of SP or CQ in their blood. However, some reported to have been given amodiaquine or quinine (not detected by our analysis) which means the correlation between reported intake and blood drug levels could actually be higher. On the other hand, 24% of those who claimed not to have been given antimalarials had detectable levels of drug in their blood.
Table 3. Reported health seeking behaviour outside homes in underfive children (116 with reported febrile episode within 4 weeks prior to interview) compared to their blood concentrations of antimalarials
Type of care
CQ detected in blood
S detected in blood
S and CQ detected in blood
No drugs detected in blood
The figures in the text referring to this table are a weighted population estimate.
HF, health facility; CQ, chloroquine; S, sulphadoxine.
* Includes private dispensaries, medical laboratory, pharmacy and ordinary shop
Public HF (n = 90)
Private HF* (n = 8)
Did not seek care outside home (n = 18)
Total (n = 116)
Household and health staff knowledge of antimalarial policy change
In the households, 51% correctly stated that the current first line treatment against malaria was SP and 41% explained that CQ was not used any more because of drug resistance (using the word ‘sugu’ = resistant). The odds of knowing the current first line treatment against malaria were higher for residents of an urban ward [OR = 2.9 (1.5–5.6)], those who had completed primary school [OR = 5.0 (1.6–15.7)] or had a child younger than 5 years in their household who had had a febrile episode during the last 4 weeks [OR = 2.6 (1.3–5.3)]. Having a child younger than 5 years in the household, however, did not increase the odds of knowing the current malaria treatment.
Reported experiences of treatment with SP
From the household interviews we estimated that 74% of those who had an opinion of use of SP perceived it to be a good drug. The odds of having a positive perception of SP were lower in the urban wards [OR = 0.3 (0.1–0.6)]. Two per cent reported to have experienced minor skin reactions.
Focus group discussions
In all FGDs, almost all the participants claimed to know that the current first line antimalarial was SP. They also knew that the new drug had been implemented because of malaria resistance against CQ. Participants mentioned having heard about SP on the radio or from staff or posters at health facilities. The FGDs also revealed that CQ is not readily available but some participants said that they would like to continue using CQ, as they believed it to be effective.
Most participants in the FGDs with mothers and fathers expressed fear of using SP because of adverse reactions although few had actual experience of such reactions. Some people tried to avoid SP by asking to be prescribed second or third line drugs. This is supported by health worker statements such as:
‘…some of them lie that they are allergic to sulphur so they don't use SP but amodiaquine and quinine’. – Health worker
Most FGD participants had heard about people who died after using SP. Most commonly newspapers and radio were reported information sources.
Information presented by the media was taken seriously and several participants suggested that journalists need education about SP. Despite this, one participant said:
‘…if I put [give to a journalist] a photo of a child burnt by fire and say it was caused by SP, I know I will get money[…]we have to be very careful in what we read from these papers’. – Mother
In several of the FGDs mothers and fathers were upset and said they thought they were part of an experiment staged by the Tanzanian government to see how many people would die using SP:
‘…these drugs are on trial. The state wants to see how many people will be killed by this SP!’– Mother
Because of the fear of SP, health workers explained that they have to give the drug under observation in the health facility to avoid that the mothers throw away the tablets instead of giving them to their children:
‘…the first dose you are given to take when you are in the hospital[…]people think that we are forcing them to use SP’. – Health worker
The health workers did not think that the number of patient visits because of fever in children younger than 5 years had changed since the introduction of the new drug. At the same time there was never a shortage of the first line antimalarial in the health facilities.
The malaria policy making in Tanzania has been mentioned as a good example of an incrementalist model (Williams et al. 2004), meaning that the policy change was influenced by numerous factors, including scientific knowledge, consumer and media pressure and the interest and values of selected key stakeholders. Most of the data in that study (Williams et al. 2004) describes the theory of the policy cycle up to the policy implementation. Our study presents data from the implementation phase of the policy cycle, which is rarely studied. Diffusion of innovations in the health systems are dependent on factors such as perceptions of the innovation, characteristics of the individuals who may adopt the change and contextual factors within the organization and in the community (Berwick 2003). Limited data have suggested that 18–24 months are necessary to change policy (WHO 2003). This lead-time to policy change was slightly longer in Tanzania, where efficacy data had been compiled already early 1999 (Kitua et al. 1999) and the policy was implemented in 2001. We have not been able to find any literature on lead-time for the implementation itself, although at least 24 months has been suggested to be necessary (Williams et al. 2004). Our data was collected 6–11 months after the policy change, making the behavioural change we found seem quite dramatic.
We showed that SP, despite being previously reported to be poorly recognized and perceived too strong for children in this area (Tarimo et al. 2001), seemed to have been accepted by the population. The large proportion (51%) of the population estimated to know that SP was the first line malaria treatment and that SP dominated in the blood samples is an indication that the policy has diffused well to the communities in our study. Not surprisingly, the knowledge of the antimalarial policy was higher among those who lived in a so-called urban setting, those who were literate and those who had underfives in the households that had had fever during the last 4 weeks.
Home stocking of antimalarial drugs and self-treatment with antimalarials as first action when the child has a fever has decreased markedly when comparing our data to studies performed in the same district before the policy change. In those studies one-third of the households stocked the first line antimalarial at the time of the study (Nsimba et al. 1999) and 98% of children seeking care at primary health care facilities (PHF) had detectable CQ levels in their blood (Nsimba et al. 2002). The low overall number (less than one-third) of underfives in the households with detectable CQ or SP in their blood confirms the low self-treatment with these drugs in our study. In addition, few mothers of underfives stated to stock and self-treat their sick children with antimalarials and the FGDs revealed fear and negative perceptions of using SP.
We now report that an absolute majority of children with fever are taken to primary health care facilities but only one-third of those who reportedly sought care at PHF for a febrile episode during the last 4 weeks had detectable levels of CQ or SP in their blood. As the drugs have long half-lives and SP can be measured in the blood 6–8 weeks (Abdi et al. 1995) after intake, this means the children had not been given any of these two drugs. This could be because of reporting bias or because of the health staff diagnosing other illnesses or giving other treatments. Poor compliance seems a less probable reason as both mothers and health workers reported giving SP as DOT (direct observed treatment) in the PHF, but intake of the second or third line antimalarial (also available at the PHF) would not be detected in our analysis. Poor quality of the drugs given at the health facilities also seems an unlikely reason for the low blood drug levels as the Medical Stores Department (MSD) in Tanzania make quality assurance controls at both the Tanzanian FDA and Government Chemist on drugs from all their drug sources (L. Nderimo, personal communication, 2005). However, there have been reports of poor quality drugs from private pharmacies (Minzi et al. 2003a). Even though there was a discrepancy between the reported intake of antimalarials and the blood drug levels, the correlation was in fact good compared with a previous study in the same district reporting that 97% of those stating not to have taken CQ actually had detectable levels of the drug in their blood (Nsimba et al. 2002).
The Integrated Management of Childhood Illnesses (IMCI) for PHF includes treating all febrile illnesses with antimalarials, in contrast to what is reported here. Moreover, international organizations advocate effective and affordable treatment for malaria within 24 h of onset of illness (Amexo et al. 2004), implying that treatment in some cases must be available in or close to homes. In this study, many reported seeking care at health facilities for febrile illness, but a substantial proportion were not given drugs, according to the blood drug levels. This effect of the new policy leads to potential risk for the children and must be studied further. It must also be stressed that unrestricted access to antimalarials inappropriately used may increase drug selection pressure and promote the spread of resistance. Any new antimalarial drug policy must therefore balance these issues. However, with cases appearing at health facilities, quality of case management must be improved (WHO 2001; Kamya et al. 2002).
In some of the FGDs the participants stated that they thought the SP policy was part of an experiment launched by the Tanzanian government, indicating that the people's trust in the health authorities concerning the SP policy was quite low. In addition, all FGDs revealed negative perceptions to SP. Although few persons had experienced adverse reactions to SP, most people had heard about people who died from using the drug. However, the majority of the household interviewees perceived SP to be a good drug. This is supported by findings in Zambia where caregivers recognized the greater efficacy of SP but feared that its ‘potency’ could be dangerous (Bloland & Ettling 1999). Mass media was reported as an important source of information about the policy change. Indeed, mass media involvement, in addition to poster campaigns and health staff training, was an important part of the implementation process of SP as the first line antimalarial in Tanzania (Mbuyazi 2003). This use of social marketing has previously been found to be effective in campaigns for oral rehydration salts (Kenya et al. 1990) and bednets (Holtz et al. 2002). This could be one of the reasons for the seemingly high impact of the policy. Although the participants in the FGDs were critical to the information found in newspapers, they still paid a lot of attention to the information the mass media spread.
A limitation of this study is that it is cross-sectional and thus only gives information about the situation at a certain point in time. However, 4 months passed between the household survey and the FGDs, a factor that could have affected the awareness of the policy change. Although very few mothers stated visiting private facilities or drug shops, the exclusion of private drug sellers in our study was a missed opportunity to know their role in malaria management after the policy shift (Nsimba et al. 1999, 2002). We used a 4-week recall period and when comparing recall from the first 2 weeks with the two last weeks, no major differences were found. The random selection of the study population was non-proportional in population size, however this was compensated for by weighted analysis of the data. Another limitation is the method of interviews itself where we have to rely on the information given by the interviewees. In our study, the findings from the interviews sometimes conflict those from the blood level analysis.
The therapeutic life of SP is known to be short and the only present future option for effective malaria treatment is combination therapies involving the use of two or more blood schizonticidal antimalarials (Nosten & Brasseur 2002; Snow et al. 2003), preferably a combination where one of the drugs is an artemisinin derivative (ACT, artemisinin-based combination therapy) (Bloland et al. 2000; Attaran et al. 2004). One obstacle for implementing ACT in Africa is the cost as it is 10 times more expensive than SP (Attaran et al. 2004). An additional challenge for the health system is that compliance may be lower than with SP as ACT is usually administered as a six-dose treatment over 3 days (Snow et al. 2003). Implementation of a drug policy is a complex process and a study similar to ours can help understand its adoption in a local context. We suggest that future drug policy changes build in studies of policy implementation of this nature.
Administrative subdivision of a district.
Smallest administrative unit of a village.
We thank all the parents/guardians and health care providers for their co-operation and participation in the study. We are grateful to James Fulgence and Walter Msangi (Department of Clinical Pharmacology, Muhimbili University College of Health Sciences) for assisting in the collection of blood samples in the field and in analysis of sulphadoxine respectively and to Dora Semkwiji and Hekima Dossa (Sociology Department, University of Dar es Salaam) for note taking during the FGDs. We also appreciate Margarita Mahindi who gave expert advice on blood sample analysis at Karolinska Institutet, Karolinska University Hospital, Huddinge.