Study area
The study was conducted in Rufiji and Ifakara Health and Demographic Surveillance System (HDSS) sites. Rufiji HDSS is situated in Rufiji District, Coast Region, with a catchment population of approximately 85,000 people living in 16,000 households [14]. Ifakara HDSS is situated in and covers parts of Kilombero and Ulanga Districts in Morogoro Region. Ifakara HDSS site constitutes more than 99,000 people, living in 28,000 scattered rural households [15]. The two HDSS sites have higher malaria transmission during the major rainy season, usually occurring between March and June annually. Malaria transmission in these study areas is endemic with seasonal fluctuations. Malaria parasitaemia is also most prevalent during a period of long rain from March to June.
Health services in Tanzania are provided by the government and non-government organizations; over 70% of health facilities are in rural areas where the majority of the population lives [16]. Levels of health care delivery in Tanzania start with home/village/community primary health care post, pharmacy, and drug stores, including accredited drug dispensing outlets (ADDO), followed by dispensaries, health centres, district hospitals, regional hospitals and the highest level is referral/consultant hospital [16]. The Rufiji HDSS has 24 health facilities in their surveillance area (one non-government hospital, two health centres and 21 dispensaries) while in Ifakara HDSS there are 14 health facilities (two health centres and 12 dispensaries) [13]. A dispensary caters for between 6,000 to 10,000 people and supervises all the village health posts in its ward; a health centre is expected to cater for 50,000 people, which is approximately the population of one administrative division [17]. A dispensary has medical assistant/clinical officers, assistant clinical officers, and lower cadres (nurse or nurse midwife, rural health assistant and nurse aide) while a health centre has a medical assistant/assistant medical officer/clinical officer, an assistant clinical officer and lower cadres. The hospital has a good mix of qualified staff of different specialism and experiences, including graduate medical officers, medical assistant/assistant medical officer/clinical officers, assistant clinical officers and lower cadres [17].
Sample size and power calculations
Sample size was calculated based on the assumption of 75% of malaria patients are treated with ACT and design effect of 2. The target sample size was 720 patients per HDSS to estimate the population of those with uncomplicated malaria correctly treated with ACT with 10% precision, assuming 20% of all patients present with uncomplicated malaria. All government and non-government health facilities providing outpatient care during the survey within HDSS areas were included (17 in Rufiji and 14 in Ifakara) and 7 health facilities were not operating during the survey. Investigators visited each facility for two to three days and collected information on attending patients. All patients attending for initial illness at the health facility on the days of the survey were eligible. Health workers were following the national guidelines for diagnosis and treatment of malaria.
Study design and data collection
This was a two sets cluster survey were conducted: one in high malaria transmission season (March, 2010) and another in low malaria transmission season (November 2010), where a cluster was defined as all patient consultations performed in a health facility on the one day of survey during regular working hours. The survey was conducted in all outpatient health facilities licensed to prescribe ACT (artemether-lumefantrine, AL) within the Rufiji and Ifakara HDSS sites.
Twenty interviewers and two supervisors were trained together on survey procedures and blood slide collection in a classroom setting and then practiced these activities in test health facilities which were different from those surveyed in the study. At health facilities, all health workers performing patient consultation on the day of survey were given an identification number. All outpatients presenting for initial consultation on the day of survey, and who consented to participate in the survey, were interviewed prior to leaving the health facility. Patients eligible for the survey were asked to provide informed consent. All patients who consented were included in the survey and given a study identification card with their study identification number. The health workers seeing a consented patient noted their initials and health worker identification number on the patient’s study identification card as well the patient’s medical record (file) number. In the laboratory, the patient identification number on the study identification card was used to label any extra blood slides made for the patient.
When the patient was ready to leave the health facility after visiting the laboratory and pharmacy as needed, the surveyor interviewed the patient and collected a blood smear (if a blood smears had not been done in the laboratory). The interview was used to determine if patients had understood the information provided by the health worker regarding diagnosis, referral, treatment, follow-up, and home care. All information and prescribed medications were recorded on a standardized questionnaire.
At the end of the survey day all health workers who had performed patient consultations were interviewed to collect information on their demographics, cadre, pre-service and in-service training, work experience, access to printed copies of the national malaria treatment guidelines and wall charts, and exposure to supervision in the preceding six months. Finally, a health facility assessment was undertaken to record the availability of AL and other anti-malarials on the survey day and in the previous three months, malaria diagnostics and any displayed case-management wall charts.
Definition
Uncomplicated malaria in this analysis was defined as either: 1) history of fever in the previous 48 hours; or 2) presence of fever at the time of presentation, both confirmed by blood slide or RDT positive for malaria infection. For children aged two to 59 months, fever or history of fever in the previous 48 hours was also considered a clinical diagnosis of uncomplicated malaria according to integrated management of childhood illness classification, if a blood slide or RDT was unavailable or not done. Correct prescription of ACT was defined as prescription of AL for patients with a qualifying diagnosis of uncomplicated malaria as described above.
Data management and analysis
Data were double entered into EPIDATA version 3.1 [18] and validated by checking completeness and consistency. The analyses were performed using STATA Version 11.0 [19], using survey procedures that account for clustering and stratification. All percentages and odds ratios reported are population-average estimates which have been adjusted to take into account the clustering of the study design.
The outcome variable in this analysis was patients with uncomplicated malaria correctly prescribed an ACT and the determinant/explanatory variables were health worker factors: health worker cadre, in-service training, having three or more years’ working experience, supervision visit in previous six months, age of health worker, and availability of job aids. In-service training was classified into two groups: first, health workers trained on either malaria case management, integrated management of childhood illness (IMCI) or use of new anti-malarials; and, second, health workers’ who were not trained in any of the above three categories. Availability of job aids included possession of the printed national malaria treatment guideline and wall chart that described the current treatment recommendations. Availability of job aids was assessed by observation by the study team. Ages of health worker were classified into three groups: age below or 35, 36 to 60 and finally health worker with age above 60.
To identify health worker factors associated with correct prescribing practices, treatment practices were analysed at all health facilities with AL on the survey day. The outcome variable in logistic regression model was coded “1” if a patient with uncomplicated malaria was correctly prescribed an ACT and “0” if a patient with uncomplicated malaria had not received a correct prescription of ACT. Health worker factors suspected to be associated with correct prescription of ACT for uncomplicated malaria were identified in the multivariate model.
The analysis used data from both seasons and employed both descriptive and analytical statistics. Frequency distribution of responses by categories of each variable were calculated and presented. Further analysis was performed in multivariable fashion using logistic regression to assess health worker factors associated with correct prescribing of ACT for uncomplicated malaria. Explanatory variables were selected for inclusion in the multivariate logistic regression automatically regardless of the significance status in the univariate analysis because all explanatory variables are of interest. The model was checked for statistical interactions and adequacy before being approved as final. An alpha level of 0.05 was used for all tests of significance. The study did not consider children aged below two months or children weighing below five. Because the guideline for management of malaria indicates children weighing below five kilogram or age below 2 months are not recommended for AL [1].
Ethical approval
Ethical approval for this study was received from the Ifakara Health Institute Ethical Review Board (IHI/IRB/No.A67-2009) and national ethical clearance.