Assessing health worker performance in malaria ... - Wiley Online Library

Tropical Medicine and International Health

doi:10.1111/j.1365-3156.2006.01753.x

volume 12 no 1 pp 52–61 january 2007

Assessing health worker performance in malaria case management of underfives at health facilities in a rural Tanzanian district J. Eriksen1,2, G. Tomson2, P. Mujinja3, M. Y. Warsame2, A. Jahn4 and L. L. Gustafsson1 1 2 3 4

Division of Clinical Pharmacology, Karolinska Institute, Stockholm, Sweden Division of International Health, Karolinska Institute, Stockholm, Sweden School of Public Health and Social Sciences, Muhimbili University College of Health Sciences, Dar es Salaam, Tanzania Department of Tropical Hygiene and Public Health, Ruprecht-Karls-Universita¨t Heidelberg, Heidelberg, Germany

Summary

objective To study the quality of malaria case management of underfives at health facilities in a rural district, 2 years after the Tanzanian malaria treatment policy change in 2001. methods Consultations of 117 sick underfives by 12 health workers at 8 health facilities in Mkuranga District, Tanzania were observed using checklists for history taking, counselling and prescription. Diagnoses and treatment were recorded. Exit interviews were performed with all mothers/guardians and blood samples taken from the underfives for the detection of malaria parasites and antimalarial drugs. Quality of care was measured using indicators adopted from the integrated management of childhood illnesses multi-country evaluation. results Quality of care measured by indicator scores averaged 31% of what was considered optimal. The poorest results were for history taking. Nevertheless, 89% of febrile children were treated with antimalarials, in line with national guidelines for fever treatment. Of these, 61% had a parasitaemia ‡2000/ll. There was no difference in treatment given to those with parasitological malaria compared with those without parasites. Pre-treatment levels of chloroquine and sulphadoxine/pyrimethamine were low and detected in 2% and 13%, respectively. conclusion Although most febrile children were given antimalarial treatment, quality of care in terms of history taking and counselling was sub-optimal. Despite this, the study community had changed behaviour from self-treatment to seeking care at health facilities. This is encouraging for introduction of artemisinin-based combination therapies policies as one could focus resources into improving care at health facilities and still reach out with treatment to most febrile children. keywords health worker performance, Tanzania, antimalarials, policy change, quality indicators, selftreatment

Background Every year about 1.1 million people die from malaria worldwide, more than 75% of them are underfives living in sub-Saharan Africa (Black et al. 2003; Bryce et al. 2005). In East Africa, the proportion of deaths attributable to malaria has increased from 18% to 37% from 1982–1989 to 1990–1998, and in Tanzania, about 30% of all deaths in children are directly attributable to malaria (Korenromp et al. 2003). Effective treatment of malaria episodes is a fundamental pillar of the malaria control strategy. As a result of the development of chloroquine (CQ) resistance, Tanzania changed its first-line malaria treatment from CQ to sulphadoxine/pyrimethamine (SP) in August 2001. During 52

the introduction of the SP policy, health workers were trained in the new treatment guidelines, partly based on the integrated management of childhood illnesses (IMCI) manual. The IMCI (WHO 1997) is a symptom-based strategy for resource-limited settings for classifying and treating the most common childhood illnesses, including malaria. The Tanzanian policy is based on the IMCI strategy, and thus recommends that all fevers should be treated as malaria in facilities where no microscopy is available. Health personnel such as prescribing nurses and clinical officers working in peripheral dispensaries and health centres provide the first contact with formal health care services. Studies from Tanzania and also other countries like Bangladesh and Burkina Faso have shown poor health

ª 2006 Blackwell Publishing Ltd



J. Eriksen et al. Health worker performance in rural Tanzania

worker performance in terms of history taking, physical examination and consultation time at primary health care facilities (Krause et al. 1998; Nsimba et al. 2002; Arifeen et al. 2005). Inadequate clinical assessment could cause incorrect diagnosis and inappropriate treatment of malaria episodes and other conditions. Quality of care has been assessed using different methods. Most involve some kind of scoring system based on quality indicators. The WHO has developed a tool for evaluating IMCI in different settings, the multi-country evaluation (MCE) (WHO 2001). This has been used by several other researchers (Armstrong Schellenberg et al. 2004; Arifeen et al. 2005) and has been suggested as a standard for easier comparison across studies (Rowe et al. 2005). In a study of malaria case management performed before the national treatment policy change in Kibaha District, Tanzania (Nsimba et al. 2002), the main findings were poor performance in terms of short consultation time, infrequent physical examination and limited history taking. CQ was prescribed for cases without a confirmed malaria diagnosis and without asking about prior drug intake. There was a high rate of self-medication with CQ (first-line antimalarial at the time) causing high drug pressure in the population and potentially toxic blood levels in children prescribed the drug. The current study assesses home management of febrile disease and the quality of malaria case management in underfives at health facilities in a rural district of Tanzania, 2 years after the change of the national malaria treatment policy and anticipating a future change in the malaria treatment policy to artemisinin-based combination therapy (ACT). This study is part of the MAMOP (improving the management of childhood malaria: an experiment to bridge the gap between mothers and health care providers) project, a controlled malaria community intervention with a pre-post design conducted in rural Burkina Faso and Tanzania in 2002–2004. The overall objective of the MAMOP study was to evaluate the feasibility and effectiveness of an intervention aimed at improving case management of malaria in underfive children through the primary caretakers in collaboration with local women groups and existing health centres.

Materials and methods Study sites The study was performed in Mkuranga District, Coast Region, Tanzania, selected for the MAMOP project. The district capital is located about 60 km south of Dar es Salaam. Mkuranga is one of the poorest districts of


Tanzania, it is holoendemic for malaria and has a population of 187 428 (Nbost 2003). The study was conducted during 6 weeks in June to July 2003, at the end of the peak malaria transmission. At the time of the study, IMCI had not been implemented in any part of the Mkuranga District. All medical staffs were supposed to have received training according to the new malaria guidelines in the fall of 2001, i.e. almost 2 years before the study was conducted. Administratively, the district is divided into divisions that in turn are divided into wards. The Mkuranga District consists of 31 wards, 11 of which were excluded from the study sampling because another intervention project was conducted in this area. From the remaining 20 wards, 10 were randomly selected and all health facilities (n ¼ 9) in these wards were included in the study. These included seven dispensaries (one government, one mission and five private), one government health centre and one government hospital. However, during our fieldwork, we realized that one of the (private) dispensaries did not treat any underfives at all. This unit was thus excluded, leaving us with eight health facilities. In Tanzania, dispensaries are the primary health care facilities providing basic curative care. Health centres are second-level facilities, have simple laboratory facilities and can usually admit patients. Hospitals are third-level facilities. Prescribers at all facilities should be at least an assistant clinical officer (ACO), with 3 years of professional training. At hospital level, there are also medical doctors, but this was not the case in Mkuranga at the time of our study. Study population Case management of sick children was the main focus of the study and we observed the interaction between health workers and sick underfives. Several studies have conducted observations of patient–provider interactions. Some observe care at facilities for a certain time period, with a wide range from 3 days (Arifeen et al. 2005) to 14 days (Krause et al. 1998). Others use a certain number of observations per facility, ranging from two observations per condition in each facility (Ehiri et al. 2005) to six observations per facility (Armstrong Schellenberg et al. 2004). None of the studies provide a rationale for the time spent for observations or the number of observations. We decided to observe case management of all underfive consultations during 3 days at each of our study facilities, as 3 days were considered adequate to capture the interindividual variations in the behaviour of health workers. The 3 days of observations were chosen at random and were not consecutive. The health workers did not know on which days the research team would arrive. 53




Mothers/guardians queuing to see the care providers were asked about the age of their children. The information was confirmed by the maternal and child health cards. After being informed about the study and its procedure, the mothers/guardians of the underfives and health care providers performing the consultations at the facilities were asked to participate. Following this, a total of 117 children were enrolled for the observations after their mothers consented to participate. After the consultation with the prescriber, exit interviews were performed in another location. After the completion of 3 days of observations at each facility, the health workers were interviewed about their level of education and the resources of the health facilities, including showing the available drugs. Data collection tools Direct observation. During the observation, a researcher (first author and/or one of three Tanzanian MDs) sat in the consultation room during the interaction and assessed the performance of the health workers using a detailed, pretested checklist (available upon request from first author) containing the information on history taking, physical examination and counselling. All observers were trained in the national guidelines and the first author sat in on the first observations performed by the three Tanzanian MDs to make sure they all used the checklist in the same way. Directly observed treatment was not practiced at the health facilities and drug dispensing was not observed by the research team. When developing the checklist, we used the questionnaire from our previous study (Nsimba et al. 2002) as a basis, modifying it to fit better with the IMCI indicators. The Tanzanian policy requires prescribers to follow an algorithm very similar to the IMCI when reaching their syndromic classification (Msengi et al. 2001). For easier comparison with other research (Rowe et al. 2005), we therefore chose to use the IMCI as a standard for our evaluation. The Tanzanian algorithm uses fever as the criteria for malaria treatment where no diagnostic tools are available. Where microscopy is available, this should be used as a basis for malaria diagnosis. In the previous study (Nsimba et al. 2002), the observer was also located outside the consultation room, just observing through a window. This was thought to interfere less with the patient–provider interaction, but we realized that keeping the observer in the same room would allow more detailed recording. To minimize interference with the health worker behaviour, the observer sat in on all patient consultations during the days of observations but only the observed underfive children were included in the data analysis. 54

Quality of care. The quality of care was assessed using a ‘scoring system’ we developed based on nine quality indicators. Eight indictors were taken from the multicountry evaluation of the IMCI (WHO 2001) and one was added by the research team for information on health worker education. Each indicator granted 1 point if completely fulfilled. If only part of the criteria for the indicator was fulfilled, the score was 0. Negative scores were not possible. The results in Table 2 are presented as percentages of the possible scores. We chose this strategy for our results to be easily comparable to other similar studies also using the MCE indicators for evaluating quality of care (Armstrong Schellenberg et al. 2004; Arifeen et al. 2005). The indicators involved a combination of factors in the case management extracted from direct observation of the patient–provider interaction, exit interviews with mothers/guardians and recording of diagnosis. Exit interviews with mothers/guardians. On exit, after each consultation, the mother/guardian of the underfive was taken aside and asked what she/he remembered from the information received about the diagnosis, the drugs prescribed, how to take the drug and other advice or information given by the health worker. Each person who attends a health facility in Tanzania has to bring a small notebook, called daftari. In the daftari, the prescriber records examination results, diagnosis and treatment. From this, we extracted and recorded the diagnosis made and prescribed drugs including formulations and dose regimens for each child. Blood sampling. After the exit interview, before advancing to the dispensing room, capillary blood samples were taken in each child. Through a single finger prick, two samples, of 100ll each, were taken for detection of blood levels of sulphadoxine and CQ and one drop was taken for a blood slide for malaria parasite density calculation. Information on health workers and facilities. At the end of the survey, the academic qualifications (education/training) of the prescriber and the time that she/he had worked at the study health facility were recorded. In addition, availability of thermometer, antimalarial and supportive drugs, number of malaria/fever episodes treated at health facilities per month and availability of national malaria treatment guidelines were recorded. Data and laboratory analysis The data from the interviews, recordings and observations were entered by the first author in a statistica file and





analysed using the statistica software (StatSoft Inc., Tulsa, OK, USA) package. The blood samples were analysed for drug levels using two different HPLC methods for CQ (Minzi et al. 2003) (performed at the Division of Clinical Pharmacology, Huddinge, Sweden) and sulphadoxine (Bergqvist et al. 1987) (performed at the Department of Clinical Pharmacology, MUCHS, Dar es Salaam, Tanzania). The number of asexual parasites was counted per 200 white blood cells (WBC) in thick films (performed by a laboratory technician and reread by one of the co-authors (MW) in Kilosa, Tanzania. In case of discrepancies in parasite count, both readers reread the slide and an average was used. In case of discrepancy in slide positivity rate, the slide would be reread and the two matching results were used). Parasite densities (asexual parasites/ll blood) were calculated assuming a leukocyte count of 8000 WBC/ll blood. A blood film was considered negative when examination of 500 WBC fields did not show the presence of asexual malaria parasites. Definition of malaria. We used a parasite density of 2000 asexual parasites/ll in combination with fever as criteria for clinical malaria. As we did not have body temperature measurements of the children, this definition of malaria is a modification of the enrolment criteria for uncomplicated malaria used in the WHO protocol for assessing efficacy of antimalarial drugs (WHO 2003). Ethical consideration The study was approved by the region, district and village administrative authorities. Informed consent was obtained from the mothers/guardians of underfives and the health workers. Ethical approval was obtained from the Human Ethics Committees of the National Institute for Medical Research (NIMR) in Dar es Salaam, Tanzania (ref. no. NIMR/HQ/R.8a/Vol.IX/253) and the Karolinska Institute in Stockholm, Sweden (D-nr: 03–206).

Results Background data We observed the performance of 12 health workers at 8 different facilities during 117 patient consultations. The number of observations per health worker ranged from 3 to 25. Almost all the parents/guardians were females (1% male) and half of them had no formal education. The children’s age was on average 21 months, 49.6% were girls.


Presenting conditions The most common presenting condition among the 117 underfives was fever or ‘malaria’ (81%), followed by diarrhoea/gastrointestinal problems (41%) and cough/ breathing difficulty (30%). Some guardians stated more than one condition. Treatment-seeking behaviour Of 117 caretakers, 68% stated to have given some kind of treatment at home prior to coming to the health facilities. In the majority of cases, this was stated to be antipyretics (81%) or antimalarials (Quinine and SP) in 4% only. Sulphadoxine blood concentrations were analysed for 99 patients (n ¼ 117). Of these, 87% had undetectable levels of the drug in their blood and only one patient had therapeutic levels (100–300 lmol/l). CQ blood concentrations were analysed for 110 patients. Of these, 2% had low levels (