Malaria prevalence and treatment of febrile patients at health facilities and medicine retailers in Cameroon

Authors


Corresponding Author Lindsay J. Mangham, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK. E-mail: Lindsay.mangham@lshtm.ac.uk

Abstract

Objective  To investigate the quality of malaria case management in Cameroon 5 years after the adoption of artemisinin-based combination therapy (ACT). Treatment patterns were examined in different types of facility, and the factors associated with being prescribed or receiving an ACT were investigated.

Methods  A cross-sectional cluster survey was conducted among individuals of all ages who left public and private health facilities and medicine retailers in Cameroon and who reported seeking treatment for a fever. Prevalence of malaria was determined by rapid diagnostic tests (RDTs) in consenting patients attending the facilities and medicine retailers.

Results  Among the patients, 73% were prescribed or received an antimalarial, and 51% were prescribed or received an ACT. Treatment provided to patients significantly differed by type of facility: 65% of patients at public facilities, 55% of patients at private facilities and 45% of patients at medicine retailers were prescribed or received an ACT (P = 0.023). The odds of a febrile patient being prescribed or receiving an ACT were significantly higher for patients who asked for an ACT (OR = 24.1, P < 0.001), were examined by the health worker (OR = 1.88, P = 0.021), had not previously sought an antimalarial for the illness (OR = 2.29, P = 0.001) and sought treatment at a public (OR = 3.55) or private facility (OR = 1.99, P = 0.003). Malaria was confirmed in 29% of patients and 70% of patients with a negative result were prescribed or received an antimalarial.

Conclusions  Malaria case management could be improved. Symptomatic diagnosis is inefficient because two-thirds of febrile patients do not have malaria. Government plans to extend malaria testing should promote rational use of ACT; though, the introduction of rapid diagnostic testing needs to be accompanied by updated clinical guidelines that provide clear guidance for the treatment of patients with negative test results.

Abstract

Objectif:  Investiguer la qualité de la prise en charge des cas de paludisme au Cameroun cinq ans après l’adoption de la thérapie à base de combinaisons d’artémisinine (ACT). Les modes de traitement ont été examinés dans différents types de services et les facteurs associés au fait d’être prescrits ou de recevoir un ACT ont étéétudiés.

Méthodes:  Une surveillance transversale en grappes a été menée auprès de personnes de tous les âges ayant déclaré avoir recherché un traitement pour de la fièvre dans des services de santé publics et privés et chez les détaillants de médicaments au Cameroun. La prévalence du paludisme a été déterminée par des tests de diagnostic rapide (TDR) chez les patients consentants, fréquentant les établissements et les détaillants de médicaments.

Résultats:  73% des individus ayant déclaré avoir recherché un traitement pour la fièvre ont été prescrits ou reçu un antipaludique et 51% ont été prescrits ou reçu un ACT. Le traitement prodigué aux patients différait significativement selon le type d’établissement: 65% des patients dans les services publics, 55% des patients dans les services privés et 45% des patients chez les détaillants de médicaments ont été prescrits ou reçu un ACT (P = 0.023). Les chances d’un patient fébrile d’être prescrit ou de recevoir un ACT étaient significativement plus élevées pour les patients qui ont demandé un ACT (OR = 24.1, P < 0.001), qui ont été examinés par un agent de santé (OR = 1.88; P = 0.021), qui n’avaient pas déjà recherché un antipaludique pour la maladie (OR = 2.29; P = 0.001) et qui ont recherché le traitement dans un service public (OR = 3.55) ou privé (OR = 1.99; P = 0.003). Le paludisme a été confirmé chez 29% des patients et 70% des patients avec un test négatif ont été prescrits ou reçu un antipaludique.

Conclusions:  La prise en charge des cas de paludisme pourrait être améliorée. Le diagnostic symptomatique est inefficace puisque les deux tiers des patients fébriles n’avaient pas le paludisme. Les plans du gouvernement d’étendre les tests du paludisme devraient promouvoir l’utilisation rationnelle de l’ACT, même si l’introduction des tests de diagnostic rapide doit être accompagnée de mises à jour des directives cliniques fournissant des orientations claires pour le traitement des patients avec des résultats de test négatif.

Abstract

Objetivo:  Investigar la calidad del manejo de casos de malaria en Camerún cinco años después de la adopción de la terapia de combinación con artemisinina (TCA). Se examinaron los patrones de tratamiento en diferentes clases de instalaciones y se investigaron los factores asociados con el habérsele prescrito o recibir TCA.

Métodos:  Se realizó un estudio croseccional entre individuos de todas las edades que reportaron haber buscado tratamiento para la fiebre al salir de instalaciones sanitarias públicas o privadas y minoristas de medicamentos en Camerún. La prevalencia de malaria se determinó mediante pruebas diagnósticas rápidas (PDRs) en pacientes que lo consintieron.

Resultados:  A un 73% de los individuos que reportaron haber buscado tratamiento para la fiebre se les prescribió o se les dió un antimalárico, y al 51% se les prescribió o se les dió TCA. El tratamiento entregado a los pacientes difería significativamente según el tipo de instalación: un 65% de los pacientes en instalaciones públicas, un 55% de los pacientes en instalaciones privadas, y un 45% de pacientes en minoristas de medicamentos se les había prescrito o habían recibido un TCA (P = 0.023). La probabilidad de que a un paciente febril se le hubiese prescrito o hubiese recibido TCA era significativamente mayor para el paciente que preguntaba por un TCA (OR = 24.1, P < 0.001), había sido examinado por un trabajador sanitario (OR = 1.88, P = 0.021), previamente no había buscado un antimalárico para su enfermedad (OR = 2.29, P = 0.001) y buscaba tratamiento en una centro sanitario público (OR = 3.55) o privado (OR = 1.99, P = 0.003). Se confirmó la presencia de malaria en un 29% de los pacientes y un 70% de los pacientes que tenían un resultado negativo fueron prescritos o recibieron un antimalárico.

Conclusiones:  Se podría mejorar el manejo de casos de malaria. El diagnóstico sintomático es ineficiente ya que dos tercios de los pacientes febriles no tenían malaria. Los planes del gobierno para ampliar la distribución de las pruebas de malaria deberían ir acompañados con la promoción de un uso racional de las TCA, acompañando la introducción de las pruebas rápidas con unas guías clínicas revisadas en donde se provea una guía clara para el tratamiento de pacientes con resultados negativos.

Background

Malaria is a major cause of morbidity and mortality and places considerable burden on health services in countries across sub-Saharan Africa. Symptomatic diagnosis of malaria is a routine practice in malaria endemic settings; though, recent guidelines from the World Health Organization recommend parasitological confirmation of suspected malaria cases in all patients before treatment, where testing facilities are available (WHO 2010a). Rapid diagnostic tests (RDTs) have attracted interest in recent years because of their high specificity and sensitivity and are suitable for resource-constrained settings as they require minimal infrastructure and training. Moreover, there is an economic case for introducing RDTs, compared with presumptive treatment, given the comparatively high cost of unnecessary treatment with antimalarials, such as artemisinin-based combination therapy (ACT), among those with non-malarial febrile illness (Lubell et al. 2007; Shillcutt et al. 2008).

The Cameroon government adopted the ACT, artesunate-amodiaquine (ASAQ), as the first-line treatment for uncomplicated malaria in 2004 and endorsed artemether lumefantrine (AL) as an alternative ACT in 2006 (Sayang et al.2009a). The National Malaria Control Programme (NMCP) organised workshops across all regions in Cameroon to inform health workers at public, mission and private health facilities of the policy change. Quinine and artemether injection are reserved for cases of severe malaria, and sulfadoxine-pyrimethamine for intermittent preventive treatment during pregnancy. The 2008 treatment guidelines for suspected cases of uncomplicated malaria lack coherence: confirmation using microscopy was advised in patients over 5 years; though, the treatment algorithm did not incorporate parasitological diagnosis and there was no advice on what action should be taken if the microscopy was negative. Moreover, while the guidelines noted that microscopy is necessary to reveal the presence of Plasmodium, they also stated that ‘a negative blood film or smear does not rule out the presence of malaria’ (Ministry of Public Health 2008). The guidelines note that pregnant women presenting with signs of malaria should be treated for severe malaria using quinine (Ministry of Public Health of the Republic of Cameroon 2008).

In August 2009, the Cameroon government announced their intention to promote the rational use of ACT using RDTs or microscopy in all cases of fever in patients over 5 years before treatment (Ministry of Public Health 2009).The NMCP is currently developing plans to pilot the introduction of RDTs in public health facilities in selected health districts across Cameroon.

Patterns of malaria treatment have been researched elsewhere in Africa and shown the extent to which malaria is over-diagnosed (Reyburn et al. 2007; Rowe et al. 2009; Juma & Zurovac 2011). Intervention studies have investigated the effect of improving microscopy and introducing RDTs on treatment (Ngasala et al. 2008; Skarbinski et al. 2009; Kyabayinze et al. 2010). Much less is known about malaria treatment practices in Cameroon. In the year following the introduction of ACT, Sayang et al. (2009b,c) found that few health workers were aware of the change in antimalarial drug policy, and facility records indicated that quinine was usually prescribed for adults and amodiaquine for children <5 years.

This study describes malaria case management in Cameroon 5 years after the adoption of ACT. Diagnosis and treatment patterns were examined in different types of facility, and the factors associated with being prescribed or receiving an ACT were investigated.

Methods

Study setting

The study was conducted in two sites in Cameroon: Yaoundé in the Centre region and Bamenda in the North-West region. In Yaoundé, the capital and predominately Francophone population, all five health districts were included. The Bamenda site encompassed urban Bamenda and five rural health districts within 21 km radius, which serve an Anglophone population. Facilities included public district hospitals and primary health centres; private health care facilities (including mission hospitals, mission health centres and private clinics); and medicine retailers (either pharmacies or drug stores). Many public and private facilities include a laboratory; though, malaria microscopy may be limited by the availability of reagents or trained staff. Pharmacies are present in both study sites and are licensed to sell prescription and over-the-counter medicines, although not to provide patient consultations or malaria testing. Drug stores are informal providers that typically sell over-the-counter medicines and are peculiar to the North-West and South-West regions (R. Hughes, C. I. R. Chandler, L. Mangham, W. Mbacham, unpublished data). Malaria is endemic in both study sites. Transmission in Yaoundé is perennial, and in the North-West region, peak transmission is between March and October.

Study design

A stratified multistage cluster survey was conducted between July and November 2009. The survey sampling was clustered in 20 randomly selected communities, defined as a natural grouping of health areas, stratified by site. Facilities were then selected, stratifying by type of facility: i) public facilities, ii) private facilities and iii) medicine retailers. Private facilities grouped both mission and for-profit facilities, which operate with considerable independence from the government. Within each community, all public and private facilities were included, while medicine retailers were randomly selected with probability proportionate to size assuming that a total of 100 medicine retailers could be visited.

The primary outcome was the proportion of individuals reporting seeking treatment for a fever that were prescribed or received an ACT. A survey sample of 12 patients per public facility and eight patients per private facility and medicine retailer was calculated to estimate the primary outcome with a precision of ±6.2% (private facility or medicine retailer) and ±8.6% (public facility), assuming that the intra-cluster correlation in treatment between facilities is 0.3 (Bennett et al. 1991). These precision estimates differ given the different sample sizes per type of facility and assume a prevalence of 50% for the primary outcome to give the least amount of precision. The sampling was based on an enumeration of health facilities conducted in February–May 2009.

Survey activities

A field team visited each facility to explain the purpose of the survey to the head of the facility or medicine retailer and obtain informed consent. Data were collected using four structured approaches: 1) patient exit questionnaire, 2) malaria testing, 3) health worker survey and 4) facility audit. The survey questionnaires were developed specifically for the study and pretested on a non-random sample of individuals at facilities not selected for inclusion in the study. Field teams were trained on procedures for conducting survey activities, including administering RDTs, and were involved in the pretesting and revision of the questionnaires. The field team comprised of researchers without a medical background but with laboratory expertise and masters’ degrees in public health or microbiology. A quality assurance officer supervised all aspects of data collection.

The primary outcome was measured through the exit questionnaire which collected data on the patient’s previous treatment seeking, their consultation and diagnosis, prescriptions and medicines received, the cost of treatment seeking, and demographic characteristics. Individuals exiting the facility or medicine retailer were invited to participate, asked to give written consent and were considered eligible if s/he reported seeking treatment for a fever (for themselves, a child or a patient not present) or if s/he had received an ACT. Patients with signs of severe malaria were excluded by the field teams. Patients who were present at the facility or medicine retailer were also asked whether they were willing to be tested for malaria using a RDT (SD Bioline Malaria Ag Pf/pan) to determine the prevalence of malaria and the proportion of patients receiving appropriate treatment. Treatment was considered appropriate if the patient’s test result was positive and s/he had been prescribed or had received an ACT, or if the test result was negative and s/he had not been prescribed and had not received an antimalarial. Patients were told the test result, and anyone with a positive result was advised that ACT is the recommended antimalarial.

The health worker questionnaire asked about access to in-service training, guidelines, and recommended practices for treating uncomplicated malaria. The facility audit recorded data on diagnostic services and antimalarials available. The health worker and facility data were collected once the patient exit survey was complete. All health workers within each facility (including medicine retailers) responsible for prescribing or dispensing medicines and available at the time of the health worker and facility audit were included in the survey. Written consent was obtained from all individuals that participated in the study.

Statistical analysis

Data were double-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 (StataCorp 2009). Thus, the percentages, 95% confidence intervals 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.

Treatment outcomes by strata were estimated using the Rao and Scott chi-square correction (Rao & Scott 1981). Survey logistic regression was used to assess factors associated with being provided an ACT (which is defined as either being prescribed or receiving an ACT). Patient characteristics, health worker characteristics, details of the consultation and type of facility were investigated for their potential association with being provided an ACT (as listed in Table 4). Factors whose univariable association with being provided an ACT reached statistical significance at the 10 per cent level or which were strongly associated with the outcome (with an odds ratio <0.5 or >1.5) were included a multivariable model. The multivariable model was developed using stepwise regression, and all factors were retained if they remained significantly associated at the 10% level or with an adjusted odds ratio of <0.5 or >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 guidelines recommend alternative treatments.

Ethical approval

Ethical approval was obtained from the ethics committees of the London School of Hygiene and Tropical Medicine and Cameroon National Ethics Committee. Administrative clearance was obtained from the Directorate of Operational Research in Health from the Ministry of Public Health.

Results

Patient characteristics

In total, 964 eligible patients (or their caregivers) consented to participate in the exit survey. Of these, 16 pregnant women and 10 children under 6 months were excluded from the analysis. Thus, the analysis is based on exit data collected from 938 febrile patients attending 174 facilities (Figure 1). The characteristics of patients surveyed at each type of facility varied by age, socioeconomic status (SES) and level of education, with those attending public facilities of a slightly lower SES and education level (Table 1). In total 544 (59%) patients reported, it was the first time treatment was sought for this illness episode and 182 (19%) were seeking treatment on the same day or the day following onset of symptoms. Of those that had previously sought treatment, 154 of 386 (43%) recalled receiving an antimalarial at the last place they sought treatment and 62 of 386 (18%) recalled receiving an ACT.

Figure 1.

 Survey population in Cameroon.

Table 1.   Patient characteristics
 PublicPrivateMedicine retailerTotalP-value
N = 329%N = 183%N = 426%N = 938%
  1. ACT, artemisinin-based combination therapy; SES, socioeconomic status.

  2. *Missing nine observations (three from public, three from private and three from medicine retailers).

  3. †Principal components analysis was undertaken to generate a SES index based on household asset ownership (Filmer & Pritchett 2001) The SES index was disaggregated into quintiles.

  4. ‡Missing 15 observations (five from public, one from private and nine from medicine retailers).

  5. §Missing eight observations (two from public and six from medicine retailers).

Study site
 Bamenda22569.18339.222658.653457.50.148
 Yaoundé10430.910060.820041.540442.6
Gender*
 Male15746.28143.222852.746649.40.149
 Female16953.89956.819547.346350.6
Age
 >15 years (adult)19157.611159.129469.159664.60.090
 5–15 years5315.22816.3377.611811.0
 <5 years8527.24424.69523.322424.4
Socioeconomic status (by wealth quintile)†
 Poorest10431.43016.85516.718920.10.025
 Second7021.22411.99222.018720.0
 Third6319.64426.58018.518720.2
 Fourth5115.84828.48918.518819.7
 Richest4012.03716.511024.318720.0
Education (or caregiver education)‡
 None or primary education14342.26942.212935.534138.30.142
 Secondary education14446.78441.419042.941843.5
 Tertiary education3711.12916.49821.616418.2
Number of days since start of symptoms (n = 938)
 ≤1 day5415.03216.49620.718218.60.453
 2 days6018.32413.49221.017619.0
 3–5 days11635.87038.013833.23434.7
 6 days or longer9930.95732.210025.125627.8
Was first time sought treatment§
 Yes18453.59347.626765.454459.30.055
 No14346.59052.415334.738640.7
Recall of treatment received at last place sought treatmentN = 143%N = 90%N = 153%N = 386% 
 Any AM4530.63442.67449.815442.90.048
 ACT149.3109.43826.36217.70.002
 Antibiotic2214.797.31410.34510.80.450
 Antipyretic8761.55454.26341.220449.80.025

Health facility and health worker characteristics

Public and private facilities were well equipped, with microscopy testing available in 36 (91%) public facilities and 43 (100%) private facilities (Table 2). ACT was available in 121 (70%) facilities. Public facilities tended to stock artesunate-amodiaquine; though, other types of ACT including artemether-lumefantrine and dihydroartemisinin-piperaquine were available at private facilities and medicine retailers. Most facilities reported quinine and sulfadoxine-pyrimethamine were available, while artemisinin-monotherapy was available in 54 (40%) facilities.

Table 2.   Health facility and health worker characteristics
Health facilitiesPublicPrivateMedicine retailerTotalP-value
N = 43%N = 43%N = 88%N = 174%
  1. ACT, artemisinin-based combination therapy.

  2. *Missing one observation (from pharmacy).

  3. †Missing two observations (one from public & one from private).

  4. ‡The availability of ACTs was verified by the field staff.

Study site
 Bamenda3258.32028.65957.111151.50.005
 Yaoundé1141.72371.42943.06348.5 
Equipment and services available
 Weighing scale*4093.9431003949.012266.5<0.001
 Thermometer*3994.64295.45761.213873.4<0.001
 Microscopy services*3690.54310033.38236.7<0.001
 RDT in stock*00510.911.162.9<0.001
Availability of antimalarials
 Any antimalarial4298.24298.08810017299.30.023
  Artesunate monotherapy†932.21535.43042.75439.60.237
  Amodiaquine†39.7712.53635.04626.5<0.001
  Quinine†3891.83995.68496.216195.40.774
  Sulfadoxine-pyrimethamine (SP)†3586.23482.88092.514989.60.028
 Any ACT‡3482.82765.26071.812172.20.039
  Artemether lumefantrine59.91849.53747.86042.1<0.001
  Artesunate amodiaquine3377.01433.75062.69759.00.002
  Artesunate mefloquine28.5411.72132.92724.7<0.001
  Artesunate sulfadoxine-pyrimethamine0012.01825.91916.90.003
  Dihydroartemisinin-piperaquine314.3817.52334.93428.10.004
Health workersN = 134%N = 129%N = 184%N = 447%P-value
 HW has attended malaria training in past 3 years7449.24033.94829.516234.5<0.001
 HW has access to malaria treatment guidelines8854.85442.284.615024.1<0.001
 HW accurately reported ACTs are the recommended treatment for uncomplicated malaria10880.18360.49354.328460.9<0.001

Across all facilities, 284 (61%) health workers reported that ACT is the recommended treatment for uncomplicated malaria, 150 (24%) had access to guidelines and 162 (35%) had attended malaria training in the past 3 years. The knowledge of health workers at public facilities was higher, with 108 (80%) health workers aware that ACT is the recommended treatment compared with 83 (60%) at private facilities and 93 (54%) at medicine retailers.

Prescribed treatment for malaria

Patient-reported consultations differed by type of facility, with health workers at public and private facilities more likely to ask about symptoms, examine and test the patient than health workers at medicine retailers (Table 3). Patients were more likely to request antimalarials, including ACT, at medicine retailers, whilst those attending private facilities were more likely to be tested for malaria.

Table 3.   Recommended treatment of malaria
 Health facility type
PublicPrivateMedicine retailerTotalP-value
N = 329N = 183N = 426N = 938
% (95% CI)% (95% CI)% (95% CI)% (95% CI)
  1. ACT, artemisinin-based combination therapy.

  2. *61.4% (49.4–72.2%) of children under 5 years that were prescribed or received an ACT. By facility type the percentage of children under 5 years that were prescribed or received an ACT was 79.4% (69.0–87.0%) at public facilities; 50.6% (26.2–74.8%) at private facilities; and 56.6% (39.7–72.1%) at medicine retailers (P = 0.080).

  3. †Defined as dose that is consistent guidance on dosage by patient age (and excludes suspensions and syrups). Based on 306 observations (112 from public, 68 from private and 126 from medicine retailers).

  4. ‡Defined as patient reports treatment regimen is consistent with guidance on dosage by patient age (and excludes suspensions and syrups). Based on 283 observations (111 from public, 64 from private and 108 from medicine retailers).

Patient reported consultation
 Told HW about patient symptoms95.6 (90.8–97.9)97.6 (90.6–99.4)62.1 (52.5–70.8)76.6 (68.6–83.0)<0.001
 Told HW that had a fever66.9 (58.9–74.0)64.3 (49.2–77.0)49.5 (39.9–59.2)56.4 (49.1–63.4)0.024
 HW asked follow up questions about patient’s symptoms75.8 (65.8–83.6)66.9 (53.7–77.8)32.7 (24.0–42.7)49.1 (41.2–56.9)<0.001
 Patient was examined60.0 (48.8–70.3)75.3 (54.7–88.6)12.3 (7.9–18.7)35.2 (26.5–44.9)<0.001
 Patient had temperature taken67.9 (54.3–79.0)55.9 (26.1–82.0)9.0 (5.0–15.7)31.4 (23.6–40.4)<0.001
 Patient tested for malaria at this facility35.1 (25.7–45.7)44.4 (30.0–59.7)1.0 (0.2–6.1)17.0 (12.3–23.1)<0.001
% Of patients who requested
 Any type of medicine8.6 (5.0–14.6)2.2 (0.7–6.3)53.7 (42.2–64.8)33.8 (25.3–43.4)<0.001
 Antimalarial (any type)7.5 (4.1–13.4)1.5 (0.4–5.0)37.3 (26.8–49.2)23.8 (16.7–32.7)<0.001
  Any ACT3.7 (1.8–7.3)1.0 (0.3–3.8)25.2 (17.2–35.2)15.7 (10.6–22.6)<0.001
  Artemisinin monotherapy000.3 (0.0–1.5)0.1 (0.0–0.9)0.814
  Quinine3.3 (1.1–9.5)0.4 (0.1–2.0)6.7 (3.3–13.0)4.8 (2.6–8.7)0.033
  SP0.5 (0.0–6.2)04.4 (2.1–8.8)2.7 (1.3–5.5)0.084
% of all patients who were prescribed or received
 Antimalarial (any type)78.3 (71.9–83.6)84.5 (70.5–92.6)66.9 (55.5–76.7)72.8 (64.7–79.7)0.016
  ACT*64.8 (55.3–72.3)54.5 (39.5–68.8)44.9 (34.5–55.7)51.3 (44.0–58.6)0.023
  Artemisinin monotherapy1.1 (0.3–4.1)3.1 (0.3–24.4)0.1 (0.0–1.1)0.9 (0.2–3.5)0.031
  Quinine13.4 (7.6–22.4)21.1 (11.0–36.8)15.2 (8.7–25.2)15.9 (10.9–22.5)0.477
  SP0.2 (0.0–3.9)10.7 (4.5–23.2)4.5 (2.1–9.3)4.7 (2.6–8.2)0.008
Antibiotic41.9 (32.8–51.7)41.1 (33.3–49.2)12.0 (8.0–17.5)24.4 (19.6–29.9)<0.001
Antipyretic69.4 (60.8–76.8)71.4 (53.0–84.7)44.9 (35.0–55.3)55.6 (46.5–64.2)0.002
% Of patients who were not tested for malaria and were prescribed or received
 Antimalarial (any type)78.4 (71.1–84.2)85.1 (67.4–94.0)67.3 (56.0–76.9)71.5 (62.6–79.0)0.046
  ACT*67.6 (56.4–77.1)63.8 (50.6–75.1)45.0 (34.7–55.7)51.4 (43.5–59.3)0.005
  Artemisinin monotherapy0.8 (0.2–3.0)00.1 (0.0–1.1)0.2 (0.1–0.7)0.120
  Quinine10.7 (5.4–20.1)15.7 (8.5–27.1)15.3 (8.9–25.0)14.5 (9.6–21.4)0.597
  SP07.3 (2.2–21.3)4.6 (2.2–9.2)4.1 (2.1–7.6)0.203
 Antibiotic41.2 (31.4–51.6)45.3 (35.9–55.1)11.8 (8.0–16.9)21.3 (16.8–26.6)<0.001
 Antipyretic71.3 (62.3–78.9)72.1 (50.6–86.7)45.4 (35.8–55.3)53.4 (44.3–62.3)0.002
Dosage and advice given for ACT dispensed
 % Of ACTs dispensed that were an accurate dose†92.4 (84.0–96.6)98.5 (93.5–99.7)97.5 (92.3–99.2)96.5 (93.3–98.1)0.065
 % Patient that has accurate knowledge of treatment regimen‡83.9 (70.7–91.8)89.0 (74.3–95.7)85.8 (74.7–92.5)86.1 (79.8–90.6)0.781

Almost three-quarters (73%, 95% CI: 65–80%) of all patients were prescribed or received an antimalarial. Antibiotics were prescribed or received by 24% of febrile patients, and antipyretics were prescribed or received by 56% of febrile patients. Approximately half (51%, 95% CI: 44–59%) of all patients were prescribed or received an ACT, the recommended treatment for uncomplicated malaria; 65% (95% CI: 55–72%) of patients at public facilities, 55% (95% CI: 40–69%) at private facilities and 45% (95% CI: 35–56%) at medicine retailers (P = 0.023). This includes 17% (95% CI: 12–23%) of patients who reported they were tested for malaria during their consultation; though, the results are similar if the sample is restricted to those that were diagnosed based on symptoms alone: 51% (95% CI: 44–59%) of patients were presumptively prescribed or received an ACT.

The odds of a febrile patient being prescribed or receiving an ACT were significantly higher for patients who asked for an ACT (OR = 24.1, P < 0.001), were examined by the health worker (OR = 1.88, P = 0.021), had not previously sought an antimalarial for the illness (OR = 2.29, P = 0.001) and sought treatment at a public (OR = 3.55) or private facility (OR = 1.99, P = 0.003) (Table 4). There was no evidence that the treatment prescribed or received was significantly associated with the patient’s demographic characteristics or being tested for malaria. In the univariable analysis, the odds that febrile patients were prescribed or receiving an ACT were significantly associated with facilities that had one or more health workers who had i) attended malaria training and ii) knew ACT is recommended; though, these factors did not remain significant in the multivariable model.

Table 4.   Factors influencing whether a patient is prescribed or received an ACT
Variablen/N%Univariable analysisMultivariable analysis
OR (95% CI)P-valueOR (95% CI)P-value
  1. ACT, artemisinin-based combination therapy.

 Study siteBamenda194/35555.01.20 (0.69–2.10)0.502  
Yaoundé282/48950.51.0 
Patient characteristics
 GenderMale234/41852.91.04 (0.65–1.66)0.850  
Female242/42651.91.0 
 Age Group>15 years291/53949.61.00.173 
5–15 years55/10547.91.02 (0.45–1.90) 
<5 years130/20062.01.65 (0.95–2.89) 
 QuintileRichest97/15964.42.06 (0.98–4.34)0.371  
Fourth102/17255.41.41 (0.68–2.94) 
Third88/17150.11.14 (0.59–2.21) 
Second93/17346.40.99 (0.56–1.75) 
Poorest96/16946.71.0 
 Education level of patient (or caregiver)Tertiary90/14760.31.68 (0.93–3.02)0.221  
Secondary214/38753.41.27 (0.77–2.07) 
None/Primary172/31047.61.0 
 Time before treatment sought≤1 day98/16453.71.00.203  
2 days102/16261.61.38 (0.71–2.67) 
3–5 days143/28945.00.70 (0.39–1.26) 
6+ days133/22954.11.01 (0.53–1.93) 
 First time sought treatmentYes293/49554.71.26 (0.84–1.89)0.245  
No183/34948.91.0 
 Had not previously sought an antimalarialYes416/70455.01.81 (1.03–3.17)0.0392.29 (1.46–3.59)0.001
No60/14040.31.01.0
 Had not previously sought an ACTYes447/78652.91.34 (0.50–3.61)0.537  
No29/5845.51.0 
Patient-reported consultation
 Health worker (HW) is told patient has feverYes275/45852.10.97 (0.58–1.62)0.917  
No201/35952.71.0 
 HW asks follow up questionsYes280/47654.81.21 (0.77–1.91)0.390  
No196/36850.11.0 
 Patient is examined by HWYes222/35761.51.77 (1.11–2.83)0.0201.88 (1.11–3.18)0.021
No254/48747.41.01.0
 Patient has temperature takenYes207/33658.71.45 (0.85–2.49)0.162  
No269/50849.51.0 
 HW tests patient for malariaYes102/18451.10.94 (0.53–1.67)0.832  
No373/66052.61.0 
 Patient is tested and RDT negativeYes61/10759.11.35 (0.78–2.36)0.269  
No415/73751.71.0 
 Asked for ACTYes115/12990.111.11 (4.90–25.15)<0.00124.14 (9.40–62.04)<0.001
No361/71545.11.01.0
Health facility characteristics
 Type of facilityPublic193/29067.52.46 (1.35–4.48)0.0233.55 (1.87–6.74)0.003
Private96/17454.11.39 (0.72–2.71)1.99 (0.91–4.39)
Medicine retailer187/38045.81.01.0
 One or more HWs at facility attended malaria trainingYes346/56061.62.46 (1.44–4.49)0.002  
No130/28439.41.0 
 One or more HWs at facility have access to guidelinesYes273/42760.61.77 (0.89–3.55)0.100  
No203/41746.41.0 
 One or more HWs at facility know ACT is recommendedYes431/73156.02.44 (0.84–7.14)0.097  
No45/11334.31.0 

Quinine was prescribed or received by 16% (95% CI: 11–22%) of patients and a further 5% (95% CI: 3–9%) of patients were prescribed or received sulfadoxine-pyrimethamine (Table 3). Almost all ACTs dispensed were estimated to be dispensed in the correct dose for the patient’s age, and 86% (95% CI: 80–91%) of patients receiving an ACT accurately reported how the medicine should be taken.

Appropriate treatment for malaria

RDTs were used by the study team to test for malaria in 746 (79%) patients, and malaria was confirmed using RDTs in 29% (95% CI: 22–36%) of patients tested (Table 5). Based on these findings, 39% (95% CI: 33–45%) of patients received appropriate treatment; 43% (95% CI: 34–52%) in public facilities, 29% (95% CI: 21–39%) in private facilities and 41% (95% CI: 33–49%) in medicine retailers (P = 0.08). Of those who were positive for malaria, 59% (95% CI: 50–66%) received an ACT, while of those negative for malaria, 48% (95% CI: 40–57%) received an ACT. Almost three-quarters of patients (70%, 95% CI: 60–78%) received an antimalarial despite not having malaria.

Table 5.   Appropriate treatment of malaria
 NRDT result% Prescribed or received an antimalarial% Prescribed or received ACT% Received appropriate treatment
YesNoYesNo
% (95% CI)% (95% CI)% (95% CI)% (95% CI)% (95% CI)% (95% CI)
  1. ACT, artemisinin-based combination therapy.

All facilities746     38.7 (32.9–45.0)
 Malaria positive22228.6 (22.3–35.9)82.8 (73.4–89.3)17.3 (10.7–26.6)58.8 (50.7–66.4)41.2 (33.6–49.3) 
 Malaria negative52471.4 (64.1–77.7)69.5 (60.1–77.4)30.7 (22.7–40.0)48.4 (40.0–56.9)51.7 (43.2–60.1) 
Public facilities294     42.6 (33.9–51.6)
 Malaria positive10233.0 (22.9–45.0)89.5 (81.7–94.3)10.5 (5.7–18.3)75.6 (64.1–84.3)24.4 (15.7–35.9) 
 Malaria negative19267.0 (55.0–77.1)73.8 (65.8–80.6)26.2 (19.4–34.3)61.5 (49.7–72.1)38.9 (28.2–50.7) 
Private facilities162     29.0 (20.5–39.3)
 Malaria positive4733.6 (18.0–53.9)79.4 (64.1–89.3)20.6 (10.7–35.9)56.7 (39.5–72.4)43.3 (27.7–60.5) 
 Malaria negative11566.4 (46.1–82.0)85.0 (69.1–93.5)15.0 (6.5–30.9)57.4 (44.6–69.3)42.6 (30.8–55.5) 
Medicine retailers290     40.8 (33.3–48.8)
 Malaria positive7324.3 (17.9–32.0)79.7 (59.8–91.3)20.3 (8.7–40.2)47.8 (33.5–62.5)52.2 (37.5–66.5) 
 Malaria negative21775.7 (68.0–82.1)61.5 (50.0–71.8)38.5 (28.2–50.0)39.0 (28.5–50.7)61.0 (49.3–71.5) 

Health workers’ choice of treatment was investigated using patient-reported information on whether a test was conducted during the consultation, the test result (determined by the field team) and the treatment prescribed or received. The test result of the RDT conducted by the field team was used because patient-reported results were unreliable as many patients did not know their test result and routine test data were not accessed. Treatment does not significantly differ for patients who were tested during their consultation; 78% (95% CI: 66–86%) of patients who were tested positive, 82% (95% CI: 71–89%) of patients who were tested negative and 72% (95% CI: 62–79%) of patients not tested were prescribed or received an antimalarial (Table 6).

Table 6.   Treatment prescribed to patients tested/not tested during patient consultation
 NPrescribed or received an antimalarialPrescribed or received an ACTPrescribed or received quininePrescribed or received an antibiotic
% (95% CI)% (95% CI)% (95% CI)% (95% CI)
  1. ACT, artemisinin-based combination therapy.

  2. *Patient reported data on test results was considered unreliable as more than half of respondents reported that they did not know the result of the test that was conducted and routine test data was not available to the field team. The malaria test result used here is that from the RDT conducted by the field team.

Patient tested during consultation and malaria positive*5677.5 (66.1–85.9)52.3 (35.3–68.9)15.8 (61.2–34.8)38.9 (22.1–58.8)
Patient tested during consultation and malaria negative*12181.9 (71.2–89.2)56.3 (39.8–71.5)22.1 (11.2–33.0)37.9 (25.7–51.9)
Patient was not tested during consultation73071.5 (62.5–79.1)51.5 (43.4–59.5)14.5 (9.5–21.6)21.3 (16.7–26.8)
All patients93872.8 (64.7–79.7)51.3 (44.0–58.6)15.9 (10.9–22.5)24.3 (19.5–29.9)

Discussion

Almost three-quarters (73%) of all patients who reported seeking treatment for a fever were prescribed or received an antimalarial, and approximately half (51%) were prescribed or received an ACT. These estimates include patients who were tested for malaria; though, the prescribing patterns were similar for presumptive treatment (based only on a symptomatic diagnosis). Sixty-one per cent of children under 5 years with a fever were prescribed or received an ACT. Studies undertaken in East and Southern Africa reported similar results: 50% of febrile children under 5 years were prescribed an ACT in Zambia in 2006 and 66% in Uganda in 2007 (Zurovac et al. 2007, 2008a).

The results of this study show an improvement of the situation in Cameroon in 2005, when prescribing records indicated that less than 15% of antimalarials prescribed were an ACT (Sayang et al. 2009b,c). The use of quinine in non-severe cases has fallen substantially since 2005 (Sayang et al. 2009b,c); though, it remains a concern: 16% of patients with symptoms of uncomplicated malaria were prescribed or received quinine. These improvements over time may reflect efforts to disseminate the policy change, as observed in Kenya where the percentage of febrile children under 5 years that were prescribed an ACT at government facilities with AL in stock increased from 28% in 2006 to 69% in 2010 (Zurovac et al. 2008b; Juma & Zurovac 2011). In this study, attendance at malaria training and health worker knowledge of malaria guidelines were no longer significant in the multivariable model; though, the association between patients provided an ACT and health worker attributes may be an underestimate because these factors were investigated at the facility level.

The odds of a febrile patient being prescribed or receiving an ACT were significantly associated with patients who asked for an ACT, were examined, had not previously obtained an antimalarial and the type of facility at which treatment was sought. Treatment practices at medicine retailers were significantly worse than at public and private facilities, although it is encouraging that patients asking for an antimalarial most often requested an ACT. Moreover, the percentage of patients receiving an ACT (45%) was better than was observed at pharmacies and patent medicine stores in neighbouring Nigeria, where 23% of febrile patients received an ACT (Mangham et al. 2011).

Reliance on presumptive treatment has led to considerable over-diagnosis of malaria. In this survey, malaria was confirmed in less than a third of suspected cases using RDTs. The RDTs used have high specificity and sensitivity (WHO 2010b); though, it is a limitation that their results were not validated using gold standard microscopy. It was also beyond the scope of this study to investigate the cause of non-malarial febrile illness, and we are unable to comment on the suitability of other medicines received. Based on the RDT results, the majority of patients (61%) were prescribed or received antimalarials they did not need, and thus, many patients received ineffective medicines and incurred unnecessary costs obtaining treatment. Over-diagnosis also has adverse cost implications for the Cameroon government, which subsidises ACT at public facilities.

Microscopy was widely available in public and private facilities, but underused, with less than half of patients tested for malaria during their consultation. Similar findings were reported in an Angolan study, in which 40% of the patients were tested despite the widespread availability of microscopy and RDTs (Rowe et al. 2009). Moreover, we observed no significant differences in the treatment prescribed or received by patients between those that were tested positive, tested negative and not tested during their consultation with the health worker. Poor adherence to test results has been observed in Ghana and Tanzania, with 50% of patients with negative RDT test results being prescribed an antimalarial in Ghana (Ansah et al. 2010) and 54% in Tanzania (Reyburn et al. 2007). Inferences about health workers’ use of the test result in selecting treatment in this study are limited because there is some uncertainty whether the RDT result used for the analysis was consistent with the result of routine microscopy undertaken at the facility. Health worker perspectives on malaria testing were subsequently explored using qualitative methods (C. I. R. Chandler, L. Mangham, A. N. Njei, O. Achonduh, W. F. Mbacham, V. Wiseman, unpublished data).

Increasing the use of malaria testing has the potential to promote the rational use of ACT and appropriate treatment of non-malarial febrile illness (Sayang et al. 2009a). The government’s plans to introduce RDTs should increase the proportion of patients tested because RDTs are simple to use and provide quick results. The study also highlights the potential benefits of extending the availability of RDTs to private facilities and medicine retailers. However, the findings suggest that attention needs to be given to the role of testing within the therapeutic process to ensure uptake of RDTs and prescriptions that adhere to the test result.

On a related point, the lack of clear guidance in the malaria guidelines, which advises confirmation using microscopy but does not explain what actions to take if the test is negative, may lead to over-treatment of malaria in patients with negative test results. The uncertainty over how to manage febrile patients with a negative malaria test result is not confined to the Cameroon setting, and there is currently no consensus on how these cases should be managed (Björkman & Mårtensson 2010). Needless to say, if the introduction of RDTs is to be cost-effective, it will be important to revise clinical guidelines and provide health workers with advice on how to undertake diagnosis and provide treatment for patients presenting with a fever in situations when the test is positive and when the test is negative for malaria.

Conclusion

This study provides timely insight into the quality of malaria case management at health facilities and medicine retailers in Cameroon. ACT was prescribed or received by 51% of patients; though, quinine was also provided for uncomplicated malaria. Symptomatic diagnosis is inefficient because two-thirds of febrile patients do not have malaria. Government plans to extend malaria testing should promote rational use of ACT; though, the introduction of rapid diagnostic testing needs to be accompanied by updated clinical guidelines that provide clear guidance on the treatment of patients with negative test results. Based on these findings and with the support of the NMCP, the REACT Project has developed provider training interventions that should improve malaria case management in public and mission health facilities and these are currently being evaluated.

Acknowledgements

We are grateful to all health workers, patients and caregivers of sick children who participated in this study. We would also like to acknowledge useful discussions in with the REACT Cameroon Advisory Group and appreciate the support of the laboratory for Public Health Research Biotechnologies at the Biotechnology Centre, University of Yaoundé I. We are also grateful for the reviewers’ comments on an earlier draft of the manuscript. Any remaining errors are the authors own. The research was supported by the ACT Consortium, which is funded through a grant from the Bill and Melinda Gates Foundation to the London School of Hygiene and Tropical Medicine.

Ancillary