Treatment of uncomplicated malaria at public health facilities and medicine retailers in south-eastern Nigeria

The study was undertaken in two study sites in Enugu State in south-eastern Nigeria: Enugu urban (comprising of Enugu East, Enugu South and Enugu North local government areas (LGAs)) and Udi LGA. Enugu urban is the largest predominantly urban area in Enugu State, and Udi LGA is rural. Malaria is endemic in Enugu State, and occurs all year round. The people of Enugu are of Igbo ethnicity and speak the Igbo language. The activities of the majority of the population include farming, fishing, wine tapping, and poultry keeping and rearing of domestic animals; the main agriculture season runs from November to February.

ACT was introduced into the study site in 2005 by the State Malaria Control Programme. The implementation package consisted of training health workers on symptomatic diagnosis, change in antimalarial policy and rational prescription of antimalarials and was accompanied by a community awareness campaign.

The study was undertaken at public primary health facilities, private sector pharmacies and patent medicine dealers (PMDs) in Enugu State, south-eastern Nigeria [17]. The term PMD refers to retail outlets that are licensed to sell over the counter pharmaceutical products, though often hold a wider range of stock, and typically have no formal training [18]. Pharmacies and PMDs are a major source of malaria treatment [14, 19]. These facilities are medicine retailers and do not routinely offer clinical care or diagnostic services. At the primary care level, presumptive treatment of malaria is recommended in febrile patients, as few public facilities offer malaria microscopy or RDTs. Primary health centres are usually staffed by community health officers and community health extension workers and supported by registered nurses and midwives [20].

A stratified multistage cluster survey was conducted between July and December 2009. The survey sampling was clustered in 16 randomly selected communities and stratified by type of facility: i) public facilities including primary health centres, dispensaries and health posts, and ii) pharmacies and PMDs. Within each community all public primary health centres were included due to their small number. There are a large number of pharmacies and PMDs, and these were randomly selected with probability proportionate to size assuming that a total of 80 (out of 298) medicine retailers could be visited given the financial resources and time available. All health workers within each facility responsible for prescribing or dispensing medicines were included in the study.

A survey sample of 20 patients per public facility was calculated to estimate the primary outcome, the proportion of febrile patients receiving the recommended treatment for malaria, with a precision of +/- 13%, assuming that the variability (intra-cluster correlation, ICC) in treatment between facilities is 0.3 [21]. For pharmacies and PMDs, 14 patients per facility allows the primary outcome to be calculated with a precision of +/- 6.6% assuming the same degree of variation. The estimates assume a prevalence of 50% for the primary outcome and give the maximum range for precision (if the observed prevalence by higher or lower than 50% then greater precision would be achieved). The sampling was based on an enumeration of health facilities and their staff conducted in April 2009.

In advance of the survey a field team visited each facility to explain the purpose of the survey to the head of the facility and obtain informed written consent. Informed consent was reconfirmed verbally on the day of the actual survey. The survey questionnaires were developed specifically for the study and pretested on a non-random sample of individuals with characteristics similar to those of the survey population but not chosen for inclusion in the survey. Survey teams were trained on procedures for conducting the survey and involved in the pretesting and revision of the questionnaires. Site supervisors monitored and supervised all aspects of data collection.

Data were collected using three structured approaches; a patient exit questionnaire, a health worker survey and a health facility audit. Written consent from patients and caregivers (who may or may not be accompanied by the patient) exiting the health facility was sought before screening to determine their eligibility to participate in the survey. An individual was considered eligible if s/he reported seeking treatment for a fever or if s/he had received an ACT. Treatment may be sought for themselves, a child or another person who is not present (the latter applies only at medicine retailers). Individuals that were exiting a health facility were assessed in turn until the patient quota was reached. All workers that were involved in prescribing or dispensing malaria treatment and were available at the time of the survey were invited to complete the health worker survey and written consent was obtained from all participants.

The patient exit questionnaire collected data on the patient's prior treatment seeking and use of anti-malarials, reasons for attendance, the consultation and diagnosis, prescriptions and medicines received, the cost of treatment seeking and the demographic characteristics of the patient. The health worker questionnaire captured data on their characteristics, access to in-service training and national malaria treatment guidelines, malaria knowledge and treatment practices. The health worker survey was conducted once all the patient exit questionnaires had been completed to ensure that the treatment received by patients was not influenced by the content of the health worker questionnaire and the patient exit data best reflects current prescribing practices. The health facility audit was conducted following the health worker survey and collected data on the characteristics of the health facility, diagnostic services, management and procurement of medicine, including the availability of ACT.

The treatment received by patients was assessed against the national malaria treatment guidelines, which recommends that patients with a fever are presumptively treated with an ACT, with the exception of pregnant women in the first trimester. The accuracy of the ACT dose provided to patients was assessed in accordance with dosage recommendations based on the patient's age and the type and composition of ACT received. Thus, the analysis takes into account that the correct number of tablets (or powder sachets) varies by brand, the amount of active ingredients contained in each tablet and whether they are co-formulated or co-blistered. Suspensions were excluded from the analysis on dosing. As patient age was used as a proxy for weight [11, 22] this may cause some error in estimating the accuracy of dosing among children, though this would not apply to adults. Patient knowledge on the dose regimen was ascertained by asking the patient or their caregiver to explain how and when the medicine should be taken. Knowledge was considered accurate if they reported the number of tablets (or powder sachets) which should be taken per day over 3 days that corresponds to the specific brand of ACT received, and the patient's age. Suspensions were excluded from the analysis due to the difficulties in accessing the accuracy of the correct dose.

Data were entered and verified using Microsoft Access 2007 (Microsoft Inc., Redmond, Washington) and analysed using STATA version 11.0 (STATA Corporation, College Station, Texas) that allows for complex survey design by identifying different probabilities of selection (sampling weights), clustering and stratification (applying the prefix svy) [23]. Thus, all percentages and odds ratios reported are population-average estimates which have been adjusted to take into account the stratification, clustering and sampling weights of the study design. The weights are equal to the inverse probability of being sampled and took into account the sampling probabilities at the facility, health worker and patient level. At the patient level, number of days it took to recruit patients was used to create a proxy for the volume of patients, with the less time indicative of a larger facility.

Treatment outcomes by strata were compared using the Rao and Scott chi-square correction [24]. Survey logistic regression was used to assess factors associated with receiving the recommended treatment. The following were investigated for their potential association: characteristics of the patient and health worker, patient consultation, and the resources available at the health facility (all factors are listed in Table six). Factors associated with receiving the recommended treatment were investigated in the multivariable model if the univariable association was statistically significant at the 10 percent level, or the odds ratio was less than 0.5 or greater than 1.5. Factors were retained in this multivariable model if they remained significantly associated at the 10% level of significance or with an adjusted odds ratio less than 0.5 or greater than 1.5. Models were compared using an adjusted Wald test. Pregnant women and children under the age of 6 months were excluded from the analysis because the national malaria treatment guidelines have alternative recommendations for these groups.

Ethical approval for this study was obtained from the ethics committees of University of Nigeria and London School of Hygiene and Tropical Medicine.

Data were collected from 100 health facilities and the analysis is based on exit data collected from 1,642 febrile patients attending public facilities and medicine retailers and 149 health workers (Figure 1). There was notable variation in the characteristics of patients attending the different types of health facility (Table 1). More than half (57%) of the patients treated at public health facilities were children, while 80% of the cases presenting at pharmacies and PMDs were adults. Treatment-seeking also varied by education levels and socioeconomic status (SES), with respondents surveyed at medicine retailers more likely to have tertiary education and be of a higher wealth quintile. At medicine retailers81% of patients reported it was the first time that they had sought treatment for this illness episode, and 43% had sought treatment on the same or day following the onset of symptoms. While at public facilities 61% of patients at public facilities were seeking treatment for the first time and the time before treatment was much longer, with only 16% seeking treatment on the same or day following the onset of symptoms. When asked about their choice of health facility, many respondents said that they had sought treatment at this facility for past illnesses (54%) and it was convenient (55%). In addition, patients at public health facilities often mentioned the lower cost of treatment, while the reputation of the provider and the availability of drugs were more often cited at medicine retailers.

The provision of basic equipment, such as weighing scales and thermometers was good in public health facilities, though more mixed in pharmacies and PMDs (Table 2). Very few health facilities offered malaria microscopy testing and none of the health facilities surveyed used RDTs. At the time of the survey, all health facilities reported that they had anti-malarials in stock, and field staff verified that ACT was in stock in 80% of health facilities. There was some variation by facility type, with 71% of public health centres and 89% of pharmacies and PMDs stocking at least one ACT. Two-thirds of health facilities had artemether-lumefantrine (AL) available, though other types of ACT were common in medicine retailers. Artemisinin monotherapy was available in 96% of medicine retailers while the vast majority (90%) of facilities also had SP as well as other types of anti-malarials available; these included less effective conventional treatments such as chloroquine and amodiaquine.

All the public facilities reported that ACT was available to patients free of charge. In pharmacies and PMDs the median price of an ACT was 600 Naira for an adult dose and 350 Naira for a child dose (which is approximately equivalent to USD $4.00 and USD $2.30). The median price of AL was higher at 750 Naira for an adult dose and 650 Naira for a child dose (equivalent to USD $6.00 and USD $4.30).

Just under half (44%) of workers in public facilities were community health extension workers, semi-skilled health workers trained in primary care, and while junior to the other cadres listed in Table 2 may prescribe treatment or undertake minor procedures [20]. In the medicine retailers the majority (83%) described themselves as patent medicine dealers or pharmacy attendants.

Knowledge of malaria treatment was variable, though better in public facilities, as 80% of workers in public facilities reported that ACT is the recommended treatment for uncomplicated malaria, compared to 62% of workers in pharmacies and PMDs. Moreover, less than one in three health workers surveyed had attended an in-service malaria training workshop over the past three years and relatively few (9%) had access to the malaria treatment guidelines (31% of public health workers and 5% of health workers at pharmacies and PMDs).

The nature of the patient's consultation differed by type of health facility (Table 3). In public health facilities 95% of respondents reported the health workers were told of the patient's symptoms, and 90% reported that they had told the health worker about the patient's fever. Patients reported to have been physically examined in 65% of cases, 50% had their temperature taken, though just 6% of patients were tested for malaria. At public facilities with microscopy testing available 21% of patients were tested for malaria. In pharmacies and PMDs patients were rarely examined (6%) or tested (<1%), though in 32% of cases health workers were told about the patient's symptoms and asked further questions.

The majority of patients attending public facilities had medicines prescribed and in 78% of cases the prescription was for an anti-malarial. ACT was prescribed to 34% of patients seeking treatment, though as many patients were prescribed SP, which is no longer recommended for treating malaria.

At pharmacies and PMDs, 15% of patients had a prescription and patients often asked for a specific medicine. At these facilities, 58% of patients attending asked for an anti-malarial. Patients often asked for SP (26%), though also requested ACT (16%) and artemisinin-monotherapy (12%). Almost all (96%) of those patients that asked for an anti-malarial also received the medicine they had requested.

Overall, the majority of patients received an anti-malarial, though ACT was received by only 22% of all patients attending health facilities and by 29% of children under five years of age (Table 4). SP is no longer recommended, though still frequently used, and 38% of patients had received this medicine. At public facilities, differences were observed between the proportion of patients that were prescribed and received antimalarials at facilities which had ACT in stock. The proportion of patients that received an antimalarial at public facilities was also low compared to the medicine retailers. There were, however, few differences between the proportions of patients receiving ACT and SP at public health facilities and medicine retailers, though patients were more likely to receive oral artemisinin monotherapy at medicine retailers than public facilities (14% compared to 2%, p < 0.001). Other anti-malarials, such as chloroquine, amodiaquine and quinine were rarely received by patients. By type of ACT, AL (44%) was most often dispensed and was widely used in the public sector. In medicine retailers, AL was regularly dispensed, though patients also received ASAQ and DHAPQ.

Two-thirds (66%) of all types of ACT dispensed were estimated to be in the correct dose, while 58% of ACT dispensed were in the correct dose and the patient (or their caregiver) accurately reported how the medicine should be taken (Table 5). Given the challenges in estimating the accuracy of ACT dosage in children, the results are also presented for febrile adults receiving ACT. Overall the results are reasonably similar, with 56% of ACT received in the correct dose and by patients that had accurate knowledge of how to take the medicine. Very few patients receiving an ACT were told of any side effects associated with the medicine.

The odds of a febrile patient receiving an ACT were significantly associated with whether the patient had a prescription, asked for an ACT, the patient's gender, and the education level of the patient (or their caregiver) (Table 6). Patients were also significantly more likely to receive an ACT at health facilities that were better equipped, and had one or more health workers that knew ACT was recommended for uncomplicated malaria. Patients that chose the health facility because it was convenient or relatively inexpensive were significantly less likely to receive an ACT. Of all the variables considered in the univariable analysis, patients asking for ACT had by far the highest odds ratio of 53.3 (15.9-179.1, p < 0.001). This variable remained highly significant in the multivariable model with an odds ratio of 55.5 (15.0-205.6, p < 0.001), though the other significant variables were the patient's gender, the education level of the patient (or caregiver), whether the facility had a thermometer available, and whether the facility had health workers that knew ACT was recommended.