Compliance with the transition to AQ+SP bi-therapy
The transition to AQ+SP was well followed at Oussouye Hospital. This result can be attributed to the favourable context of prescription in place in a District Hospital. Indeed, nurses were trained and regularly supervised by the hospital physician. Moreover, stocks of CQ were removed and those of AQ+SP never experienced shortage.
These good practices contrast with those reported in 2004, in Thies, north-western Senegal, where nurses prescribed AQ or SP mono-therapy for 15% to 78% of presumptive malaria cases [
12]. Similarly, in Zambia in 2004 and in Kenya in 2006, only 13% and 34% of treated children received artemether/lumefantrine (A/L), the first line anti-malarial [
13,
14]. Even if these three studies reported prescription practices while the new therapy has been recommended for a few months, in Zambia, two years later, only 47% of treated children received A/L [
15]. In contrast to this study, health facilities studied were dispensaries. To explain their results, authors reported health system's dysfunctions such as the availability of previous recommended anti-malarials, shortages of new anti-malarials and a lack of medical staff's training and supervision. While "cascade" training had been planned at the time of new guidelines deployment, in Senegal [
12] and Kenya [
8], it was delayed at the lowest level of the health pyramid by reason of lack of Global Fund subventions. The facility's location on the health pyramid seems therefore to be a major component. The higher the facility is located up the pyramid, the more it benefits from a favourable context for good healthcare practices.
In addition, to justify non-recommended prescriptions, nurses from Thies [
12] and Kenya [
8] explained they deviated from guidelines in response to certain practical or clinical situations. These situations encompassed the risk of an excessive dose in case of repeated prescriptions of SP, the paediatric forms' convenience, patients' demands or adjusted prescriptions in case of high fevers or adverse events of AQ.
In Oussouye, informal interviews with nurses showed similar concerns and the extent to which guidelines describe standard clinical cases while nurses are confronted with various individual situations, which can lead to non-recommended prescriptions. In the paediatric consultation room, nurses preferred to prescribe AQ alone when SP had been taken in the previous month to adhere to the one-month minimum interval recommended by the Intermittent Preventive Therapy during pregnancy. When patients had started to self-medicate with CQ, AQ or AS, nurses could carry on with this treatment, conscious of some households' low incomes. In young children, they could choose to prescribe AS suppositories because of their convenience. In the adult consultation room, in cases of high fever, one particular nurse preferred to prescribe AS because of its effectiveness. Lastly, all nurses explained that pressure from parents can sometimes affect their prescribing. Parents wanted IM QN because it was perceived as being more effective or AQ drinkable suspension to store at home for treating fevers. Others refused AQ because of its adverse effects such as asthenia, stomach aches, nausea or itching.
To conclude, it seems there are two kinds of potential impediments to the compliance with guidelines, first linked to the running of the health facility or its level in the health pyramid, and second linked to medical staff's handling of situations that they encounter daily with their patients.
Low use of blood smears and over-prescriptions
At Oussouye Hospital, compliance with the transition to AQ+SP has to be balanced. Indeed, while a laboratory facility was available, a very small proportion of patients got a blood smear. In addition, three types of over-prescription were identified. Sixty five percent (1, 055/1, 627) of non-febrile children were treated. However, as no history of previous fever was consigned in consultation records, these over-prescriptions are probably over-estimated. Secondly, 22% (591/2, 726) of children diagnosed with a non-malaria illness were treated. These over-prescriptions explain why 57% (2, 789/4, 924) of children received an anti-malarial, while only 45% (2, 198/4, 924) were diagnosed with malaria. Lastly, and more surprisingly, 80% (377/474) of children received an anti-malarial despite their negative blood smear. These over-prescriptions explain why 84% (504/601) of children referred to the laboratory facility for a blood smear received an anti-malarial, while the positive blood smear percentage was only 21% (127/601).
In a context of RDT deployment, similar results were reported. In Zambia in 2006, about one year after their introduction, only 23% of patients with fever or previous fever were tested. Moreover, 36% of patients with a negative RDT were treated with an anti-malarial [
16]. In Tanzania in 2005, this proportion rose to 54% [
17]. Despite their many advantages, the use of RDT could therefore encounter similar problems to those identified with blood smears.
Considering observations made at Oussouye Hospital and informal interviews with its medical staff, three spheres of influence seem to lead to these discrepancies in guidelines. Firstly, an ambiguous guideline enacted by the hospital physician, recommending to treat fevers with a negative blood smear as "
a precaution" if no other illness was diagnosed, probably affected nurses' prescribing. This type of ambiguous guideline is officially found in other sub-Saharan countries [
7,
18]. In Zambia, Hamer
et al[
16] also pointed out the ambiguity of the guideline in the case of negative results. However, from the clinicians' point of view, it is justified by the risk of severe malaria in cases of false negative result.
Furthermore, the multiplicity of guidelines, with IMCI, could also contribute to over-prescriptions. Indeed, this strategy recommends presumptive treatments of fevers in children under five to reduce malaria mortality. Nurses could, therefore, have applied this principle to older children, considering that there is no age limit for the risk of severe malaria.
Alongside guidelines, dysfunctions at the laboratory facility also resulted in practices not complying with guidelines. Both nurses and laboratory technician pointed out the high workload required by blood smears, especially results reading time, and technical or material problems such as lack of supplies, microscope failures or power cuts. As a result, nurses confessed they were most of the time discouraged to request blood smears. Moreover, the laboratory technician explained that he was sometimes able to read blood smears only after the end of consultations leading nurses to prescribe without knowledge of their negative results. These dysfunctions have been reported in other sub-Saharan countries [
1,
7,
19]. At two Tanzanian regional hospitals, they led to both a low use of blood smears and a delay in reading 56% of referred patients' slides [
7].
While RDT do not require either microscope or electricity and hence avoid material problems, their usage still requires time, in particular at least 15 minutes before the results can be read. At dispensaries where the mean consultation time is often very short, in particular five minutes or less during the rainy season in Kenya or in Tanzania [
2,
20], it seems probable that nurses, who have to perform these tests, take shortcuts. In Senegal, some nurses considered the delay to obtain results as being still too long [
21].
Lastly, at Oussouye Hospital, nurses' own considerations seem to have played an important role in explaining both the low blood smear use and over-prescriptions. To explain low blood smear use, nurses blamed a lack of healthcare's organisation. The long period of time to obtain results led to an accumulation of patients at the laboratory and consequently delayed the end of consultations. Patients returning from laboratory also disrupted the clinic as they sometimes pushed in front of other waiting patients creating confusion and dissatisfaction. In addition, due to restricted laboratory opening times, nurses were not able to refer patients for a blood smear after 1 pm. In Zambia, Barat
et al[
1] also reported a very long consultation time, on average 57 ± 31 minutes, when blood smears were requested. Moreover, authors emphasized the absence of full-time laboratory facilities.
Another concern of Oussouye nurses was that patients refused to have a blood smear due to its cost (300 CFA or 0.46 Euro) and time involved. These refusals must be interpreted thoroughly. Indeed, patients build their own diagnosis and therapeutic approach through what they learn from different sources of information, in particular practices observed during consultations [
22]. Therefore, patients probably noticed that they had been given anti-malarials whether or not they had been referred and, worse, whatever the result of their blood smear. These practices probably did not convince them of the usefulness of this exam.
To justify over-prescriptions, nurses' own considerations encompassed their lack of trust in the reliability of negative blood smears due to laboratory dysfunctions and their perceptions of malaria. Indeed, beyond standard symptoms of malaria, all nurses described atypical malaria cases, such as "
non-febrile malaria" suspected in patients with headaches, chills, asthenia or vomiting, and "
hidden malaria" in which the blood smear is negative because parasites are not yet in the blood. Furthermore, nurses mentioned once again some situations encountered with their patients, such as parents' demands for an anti-malarial or previous self-medication with an anti-malarial or an antipyretic which can explain the absence of fever or the negativity of blood smear and, therefore, do not exclude an anti-malarial prescription. In Tanzania, Chandler
et al[
7] interviewed nurses who shared some of these considerations, such as the doubt about the reliability of negative blood smears, the notion of false negative results because of parasites sequestration and patients demanding an anti-malarial. Although blood smears are reliable when performed by a skilled technician, their sensitivity in routine practice can be mediocre [
2,
3,
23]. Besides technical or material problems, the lack of qualified laboratory staff is another frequent reason for unreliable blood smear results [
24]. In such a context, the lack of confidence expressed by nurses cannot be blamed.
Although RDT are considered reliable, nurses could also doubt their negative results. In Tanzania, within routine practice, their sensitivity was evaluated between 19% and 86% [
25]. In this country and in Sudan [
26], three types of error were frequently observed: inadequate capillary blood sample volume, incorrect waiting time to read results and low intensity positive results read as negative. Moreover, inadequate storage conditions, such as temperatures above 35°C or humidity, frequent in sub-Saharan Africa, can also affect the quality of RDT [
27].
To summarise, over-prescriptions seem to be explained by practical, technical or socio-cultural considerations more than by clinical considerations. Nonetheless, as an underlying concern, interviews with Oussouye nurses also revealed their fear of severe malaria in non-treated patients. In Tanzania, Chandler
et al[
7] revealed the same dogma: malaria is a frequent and fatal illness that nurses have to treat as a priority. This fear can be explained by personal experiences, campaigns against malaria highlighting the lethality of this disease, but also by the training of medical staff which put a slant on the risk of severe malaria. In order to rationalize prescriptions, it therefore seems important to consider this fear and appease it.