Deployment of artemether-lumefantrine with rapid diagnostic tests at community level, Raya Valley, Tigray, Ethiopia: project report, April 2005 – June 2007

Deployment of artemether- lumefantrine with rapid diagnostic tests at community level, Raya Valley, Tigray, Ethiopia GLOBAL MALARIA PROGRAMME, JUNE 2009

Deployment at community level of artemether-lumefantrine and rapid diagnostic tests Raya Valley, Tigray, Ethiopia G L O B A L M A L A R I A P R O G R A M M E Project rePort, APril 2005–june 2007 MAlAriA And other Vector-borne diseAses control dePArtMent, tigrAy heAlth bureAu, ethioPiA Who library cataloguing-in-Publication data deployment of artemether-lumefantrine with rapid diagnostic tests at community level, raya Valley, tigray, ethiopia. 1.Malaria - drug therapy. 2.Antimalarials - economics. 3.drug therapy, combination. 4.Artemisinins - supply and distribution. 4.Artemisinins - therapeutic use. 5.ethanolamines - therapeutic use. 6.Fluorenes - therapeutic use. 7.ethiopia. i.World health organization. global Malaria Programme isbn 978 92 4 159837 8 (nlM classification: QV 256) © World health organization, 2009. All rights reserved. the designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World health organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or bounda- ries. dotted lines on maps represent approximate border lines for which there may not yet be full agreement. the mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World health organization in preference to others of a similar nature that are not mentioned. errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters. All reasonable precautions have been taken by the World health organization to verify the information contained in this publication. however, the published material is being distributed without warranty of any kind, either expressed or implied. the responsibility for the interpretation and use of the material lies with the reader. in no event shall the World health organization be liable for damages arising from its use. Photograph of community health worker on the front cover: Courtesy of Angela Bianchi, Novartis Italy For Further inForMAtion, PleAse contAct: dr Andrea bosman, global Malaria Programme World health organization 20, avenue Appia – ch-1211 geneva 27 infogmp@who.int www.who.int/malaria Contents iii Contents Acknowledgements ....................................................................................................................................................................................................................................................................v Preface ..............................................................................................................................................................................................................................................................................................................vii Executive summary ..........................................................................................................................................................................................................................................................viii 1. Introduction ............................................................................................................................................................................................................................................................................................1 2. Study area .....................................................................................................................................................................................................................................................................................................4 2.1 Geography and demography ..........................................................................................................................................................................................................4 2.2 Potential effect on vector control of malaria epidemic during the study...................................5 2.2.1 Meteorological conditions .............................................................................................................................................................................................5 2.2.2 Concurrent vector control.............................................................................................................................................................................................8 2.3 Health service infrastructure in Tigray ..........................................................................................................................................................10 3. Methods ..........................................................................................................................................................................................................................................................................................................11 3.1 Study design .................................................................................................................................................................................................................................................................11 3.2 Data collection .........................................................................................................................................................................................................................................................12 3.2.1 Institution-based morbidity, slide positivity, malaria admissions, case fatality and proportionate mortality rates ..........................................................................................................................13 3.2.2 Institution-based survey of adherence and study of performance of rapid diagnostic tests in the intervention district ....................................................................................................13 3.2.3 Community-based malaria morbidity reports in the intervention district .......................................................................................................................................................................................................................................................................................14 3.2.4 Community-based surveys.......................................................................................................................................................................................14 3.2.5 Pharmacovigilance ....................................................................................................................................................................................................................18 3.2.6 Capacity-building and database management ...........................................................................................................19 4. Results and discussion ..............................................................................................................................................................................................................................................22 4.1 Profile of community health workers and training outcomes ......................................................................22 4.1.1 Characteristics of community health workers ............................................................................................................22 4.1.2 Training outcomes .....................................................................................................................................................................................................................22 4.2 Community treatment services ..........................................................................................................................................................................................23 4.2.1 Numbers of cases treated ............................................................................................................................................................................................23 4.2.2 Pattern of service by age and gender ............................................................................................................................................23 4.2.3 Prescription for artemether-lumefantrine ..........................................................................................................................27 4.3 Serial surveys of seasonal prevalence of malaria and of knowledge, attitude and practices.............................................................................................................................................................................................................................28 4.3.1 Seasonal prevalence of malaria ....................................................................................................................................................................28 4.3.2 Knowledge, attitude and practices ......................................................................................................................................................30 4.4 Performance of rapid diagnostic tests ...............................................................................................................................................................32 4.4.1 Health institution level .................................................................................................................................................................................................32 4.4.2 Community health worker level ..................................................................................................................................................................33 iv Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 4.5 Adherence of patients to the 6-dose artemether-lumefantrine regimen ..................................35 4.6 Pharmacovigilance ..........................................................................................................................................................................................................................................39 4.7 Health impact of use of rapid diagnostic tests and artemether- lumefantrine ...............................................................................................................................................................................................................................................................40 4.7.1 Morbidity trends at clinics and health posts ................................................................................................................40 4.7.2 Morbidity trends at health centres and the hospital ...................................................................................41 4.8 Mortality surveys..............................................................................................................................................................................................................................................44 4.9 Challenges to assessing impact ..........................................................................................................................................................................................49 4.10 Satisfaction of beneficiaries of the project ...........................................................................................................................................50 4.11 Cost of diagnosis clinically and with the rapid diagnostic test at community level ........................................................................................................................................................................................................................................52 5. Lessons learnt .................................................................................................................................................................................................................................................................................53 5.1 Study design .................................................................................................................................................................................................................................................................53 5.2 Project personnel ................................................................................................................................................................................................................................................53 5.3 Database management ...........................................................................................................................................................................................................................53 5.4 Logistics..................................................................................................................................................................................................................................................................................53 5.5 Project management ..................................................................................................................................................................................................................................54 5.6 Capacity-building ..............................................................................................................................................................................................................................................54 5.7 Effect of low incidence of malaria in Phase II of the project .......................................................................54 6. Project budget .....................................................................................................................................................................................................................................................................................55 6.1 Project budget input and expenditure ..............................................................................................................................................................55 6.2 Financing and supplies .........................................................................................................................................................................................................................56 6.3 Operational costs ................................................................................................................................................................................................................................................57 7. Monitoring and evaluation ............................................................................................................................................................................................................................58 8. Conclusions ...........................................................................................................................................................................................................................................................................................60 9. Recommendations ................................................................................................................................................................................................................................................................63 Appendix A. Maps ..................................................................................................................................................................................................................................................................................66 Appendix B. Summary tables .....................................................................................................................................................................................................................................68 Appendix C. Project timetable ..................................................................................................................................................................................................................................73 Acknowledgements v Acknowledgements The Tigray Health Bureau, the World Health Organization (WHO) through its Global Malaria Programme, the Italian Ministry of Labour, Health and Social Policies – Health Sector, Novartis Farma SPA (Italy), the Department of Preventive Medicine in Migration, Tourism and Tropical Dermatology of the San Gallicano Hospital (IRCCS) and the Italian Dermatological Centre of Mekelle are grateful to all the institutions and individuals that contributed their time, energy and funding to make this project successful. Special thanks are due to Dr Tedros Adhanom, Minister of Health of Ethiopia, Dr Gebre Ab Barnabas, Head of the Tigray Health Bureau, Mr Hailemariam Lemma, Mr Alem Desta and Mr Asefaw Getachew of the Malaria Control Department of the Tigray Health Bureau, and Dr Andrea Bosman and Dr Wilson Were of the WHO Global Malaria Programme, for their dedicated efforts throughout the project. Monitoring and follow-up of the project would have been impossible without the involvement of Dr Gianfranco Costanzo of the Italian Ministry of Labour, Health and Social Policies – Health Sector Department of Innovation, Ms Angela Bianchi (Novartis Farma SPA, Italy), Dr Anne-Claire Marrast (Novartis Pharma AG, Basel) and Professor Aldo Morrone and Dr Luigi Toma (Department of Preventive Medicine in Migration, Tourism and Tropical Dermatology of the San Gallicano Hospital in Rome), Dr Nathan Mulure (Novartis Pharma, Nairobi, Kenya), Mr Roberto Ferrara (Novartis Farma SPA, Italy), Mr Ambachew Medhin (WHO Ethiopia) and Dr Giacomo Stefanoni (intern, Department of Preventive Medicine in Migration, Tourism and Tropical Dermatology, San Gallicano, Rome). These individuals also provided administrative and technical guidance, comments to strengthen the project and continuous monitoring of activities through regular teleconferences. Grateful thanks are also due to Mr Hailemariam Lemma, Mr Alem Desta, Mr Goitom Mehari and Mr Abrha Kahsay for their personal commitment in preparing operational manuals for the different project tasks, training participants and regularly tracking project activities, including data management, and to Mr Berhane Hailesilassie for designing, conducting and summarizing the focus group discussions. Professor Peter Byass and Dr Edward Fottrell (Umeå University, Sweden) were responsible for the design, implementation and analysis of the vi Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia mortality survey; and Ms Appia Augustina Appiah-Danquah (University of Ghana Medical School, Accra) contributed to the mid-term evaluation of the pharma- covigilance activities. The contributions of the Italian Ministry of Labour, Health and Social Policies – Health Sector and of Novartis Farma SPA (Italy), which provided financial support to this project are gratefully acknowledged. Above all, however, the key elements in the success of this project were the community health workers, who generously volunteered their and their families’ valuable time, resources and efforts to best serve people affected by malaria in remote rural areas. Last but not least, this project would not have been possible without the full involvement and continuous support of the communities, the health authorities and the health workers in both districts throughout the project. Preface vii Preface Since 2004, WHO has promoted strategies for improving access to treatment through home management of malaria, in order to ensure prompt diagnosis and effective treatment for people living in areas with limited access to general health services. Such situations are common in remote rural areas, where the burden of malaria is highest. Experience in many countries during the 1990s showed that this strategy, implemented with chloroquine or sulfadoxine-pyrimethamine provided by community health workers, community drug distributors, accredited medicine sellers and even mothers as coordinators, can affect mortality and severe morbidity due to malaria and considerably reduce the burden on health facilities. The introduction of artemisinin-based combination therapies, which cost at least 20 times more than conventional antimalarial agents, has reduced home-based management of malaria in recent years, as these highly effective medicines are administered mainly by the general health services. A few countries in Asia (e.g. Cambodia and Viet Nam) have used these therapies in home-based management of malaria, but no African countries have done so. In order to evaluate the feasibility and effect of implementing this strategy in Africa, a community-based observational study was designed for application in Tigray, northern Ethiopia. In line with international best practice, an important component of this project was use of rapid diagnostic tests to confirm a diagnosis of malaria before treatment with artemisinin-based combination therapies as part of home-based management of malaria. This region1 was chosen because a large-scale, community-based malaria control programme in 1994–2002 involving treatment in the community by over 700 health workers had been completed successfully. Both projects were rooted in a strong spirit of community involvement and participation, which pervades all aspects of the life of the people of Tigray. This report describes the approaches and the results obtained, including the evidence base and a detailed description of the project, to provide guidance for extension of this experience to other regions of Ethiopia and other countries of Africa. The communities in the project area have benefited; they appreciated the service provided by the community health workers and the availability of free diagnosis and effective antimalarial treatment near their homes. The challenge now will be to sustain this approach to malaria control, which has proven to be feasible and effective, with multiple sources of funding, thus helping the people to best take care of themselves. 1. The term region in this report refers to a national regional state. viii Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Executive summary In July 2004, following a survey at 11 sentinel sites in Ethiopia in 20032, which showed high failure rates with sulfadoxine-pyrimethamine (up to 71.8% on day 28 of follow-up), the Ethiopian Ministry of Health issued new guidelines for the treatment of malaria. In the new policy, sulfadoxine-pyrimethamine was replaced by artemether-lumefantrine as first-line therapy. This new drug is more expensive than sulfadoxine-pyrimethamine; thus, so that it would be used effectively, the policy stated that a diagnosis of malaria must be confirmed by parasitological examination, either by microscopy or by rapid diagnostic testing. This posed a challenge in Tigray in northern Ethiopia, where management of malaria close to patients’ homes in remote rural areas was provided by well-established community- based treatment by volunteer health workers. Additional concern was raised about widespread, possibly irrational drug use, which can lead to the development of resistance to artemether-lumefantrine. Because of the high cost of the new drug, the risk for greater drug pressure, potential adverse drug reactions in communities with no adequate safety monitoring system and exposure of uninfected patients, it was decided to conduct operational research to evaluate the role of the existing community health worker service in providing access to artemisinin-based therapies to widely dispersed, poor, predomi- nately rural populations with poor access to health-care services. A project was therefore designed to improve the skills of community health workers, teaching them to use simple rapid diagnostic tests, to administer artemether-lumefantrine safely and to report regularly to staff in the general health services. If the perform- ance of community workers with rapid diagnostic tests was as good as that in peripheral health institutions, the approach would ensure early diagnosis and treatment of malaria in outlying rural areas, decreasing the burden of malaria and promoting community participation in malaria control. To validate the feasibility of this approach, a pilot operational research project was launched in May 2005 in Tigray, involving provision of artemether-lumefan- trine after a simple rapid diagnostic test (ParaCheck-Pf® cassette) at community level by volunteer health workers. Two approximately homogeneous districts3 were selected: Alamata, the intervention district, and Raya Azebo, the control district (see Appendix A, Map 1). 2. Federal Ministry of Health. National malaria treatment guidelines 2004. Addis Ababa, 2004. 3. Districts are locally known as wereda, which are clusters of approximately 15 tabias, with a local government and an average population of about 110 000. Executive summary ix The intervention was initiated at the end of April 2005. During the first year (Phase I), 33 community health workers provided prompt treatment with artemether- lumefantrine on the basis of clinical diagnosis alone. During the second year of the project (Phase II), 50% of the health workers were equipped with and trained to use rapid diagnostic tests, while the remainder continued to make a diagnosis solely on the basis of clinical evidence. In the control district (Raya Azebo), many health workers were gradually demobilized from the service, in line with the new national treatment policy. The project was multifaceted. It included improving the knowledge and skills of community health workers in the diagnosis and case management of malaria, providing them with a rapid diagnostic test and artemether-lumefantrine and building the capacity of malaria managers through short- and long-term post- graduate training courses. In addition, a number of surveys were undertaken to evaluate the prevalence of malaria infection, adherence to treatment regimens, performance of the rapid diagnostic test, morbidity and mortality from malaria and pharmacovigilance. “Focus group” discussions were used to measure community satisfaction and perceptions. In the intervention district, the community health workers treated approxi- mately 58% of all suspected and confirmed cases of malaria, representing, during the 26 months of the project, 75 654 patients. This resulted in a lower patient load at health institutions in the intervention district (54 774) than in the control district (101 535), reducing the burden on health services in the intervention district. During Phase II of the project, the health workers who were equipped with rapid diagnostic tests screened 5123 patients and identified 526 cases of P. falciparum malaria (10.3% positivity). The positive cases were treated with artemether- lumefantrine, and negative cases were given chloroquine for possible P. vivax infection, as per the new treatment policy. During the same period, the remaining community health workers in the intervention district treated 10 475 patients for malaria on the basis of clinical diagnosis alone. If the positivity rate among patients treated solely on the basis of clinical diagnosis is assumed to be the same as that after diagnosis with the rapid diagnostic test (10.3%), up to 9397 patients might have been inappropriately treated with artemether-lumefantrine. A comparison of diagnosis with the rapid diagnostic test and by expert micros- copy showed that the test performed very well, with 96.4% sensitivity, 76% specifi- city, a positive predictive value of 59.5% and a negative predictive value of 98.2%. The theoretical saving made by equipping community health workers with rapid diagnostic tests was calculated to be US$ 1.41 per patient examined (based on the public sector price of Coartem® [20 mg artemether-129 mg lumefantrine tablets, Novartis Pharma AG] in 2001–2006). Moreover, blood safety procedures and the diagnostic performance of community health workers using Paracheck-Pf® were good and quantitatively similar to those of personnel in health institutions. x Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Patients treated by the community health workers had similar adherence to the treatment regimen as those treated at health facilities, although the figures were low. During an initial survey, adherence to the six-dose regimen by patients treated by community health workers was poorer than those treated in health facilities (19.4% versus 8.4% definitely not adherent). The project was initiated coincidentally at the time of a malaria epidemic, which made it possible to evaluate the effectiveness of the intervention. Although the two districts had similar eco-epidemiological conditions at the start of the epidemic and an overall reduction in malaria transmission during the post-epidemic year (second year of the project), the intensity of transmission and the severity of the epidemic were lower in the intervention district. During the first year of the project, the crude parasite rate was threefold lower in the intervention than in the control district. The adjusted incidence rate ratio for mortality from malaria at the end of the two years, assessed in a household mortality survey, was 0.60 (p = 0.013; 95% confidence interval : 0.4–0.9), indicating that 40% fewer deaths due to malaria occurred in the intervention district than in the control district. In conclusion, distribution of artemether-lumefantrine by community health workers using simple rapid diagnostic tests at community level for a widely dispersed, poor, primarily rural, hard-to-reach population is feasible if the health workers are appropriately trained, equipped with simple tests and supported by frequent supervision. The findings of this study indicate that the community health worker service should be upgraded by giving them artemisinin-based combination therapies, such as artemether-lumefantrine, and rapid diagnostic tests. Such interventions in areas of low-to-moderate malaria transmission can reduce morbidity, mortality and transmission of malaria as well as the case loads of the general health services. The use of rapid diagnostic tests can therefore generate cost savings, and programmes in which both artemisinin-based combination therapies and rapid diagnostic tests are used are less expensive than those based only on clinical approaches for the management of malaria. Chapter 1. Introduction 1 Chapter 1. Introduction An estimated 52 million (68%) of the Ethiopian population (77 million as per projection of the 1994 census to July 2007) live in areas where they are at risk for malaria. Each year, health facilities report 5–6 million clinical cases of malaria and 1 million confirmed cases. In 2005, malaria accounted for 15.5% of outpatient visits, 20.4% of total admissions and 27% of total hospital deaths in the country.4 With the exception of a few areas, the malaria situation in Ethiopia is unstable, and most people have little or no immunity. In areas of unstable transmission, influenced by local weather conditions and seasonal and interannual climate variations, malaria epidemics are a major public health problem. Reports on earlier malaria epidemics by Italian investigators date from the 1930s; official unpublished ministry of Health reports and published data indicate that nationwide malaria epidemics occurred in 1953, 1958 (also reported by Fontaine et al., 19615), 1965–1966, 1972–1973, 1980–1981 (all documented by Gebre-Mariam6) and 1987–1988 (also documented by Teklehaimanot7). These epidemics were characterized by high malaria transmission, with high mortality and morbidity rates over 1–3 months. One of the more recent epidemics (1998) was unusually severe, protracted and widespread. The epidemics of 1953, 1972, 1987 and 1998 coincided with documented El Niño years.8 In 2002, severe focal outbreaks occurred in the Amhara and Southern Nations and Nationalities People’s regions of Ethiopia, resulting in large-scale epidemics in 2003. During that year, 3689 villages9 in approximately 211 districts were affected, with more than 2 million reported cases and 3000 deaths in the main regions of the country (Oromiya, Amhara, Southern Nations and Nationalities People’s and Tigray regions). The low immunity of the population and the failing therapeutic efficacy of sulfadoxine-pyrimethamine contributed to these recent epidemics.10 4. Ministry of Health. Health and health-related indicators, 2004/05. Addis Ababa, 2005. 5. Fontaine RE, Najjar AE, Prince JS. The 1958 malaria epidemic in Ethiopia. American Journal of Tropical Medicine and Hygiene, 1991; 10:795–803. 6. Gebre-Mariam N, Abdulahi Y, Mebrate A: Malaria. In Ecology of Health and Disease in Ethiopia. Edited by: Zein ZA, Kloos H. Ministry of Health, Addis Ababa, 1988;136-150. 7. Kisewski A, Teklehaimanot A. A review of the Clinical and Epidemiological Burden of Epidemic Malaria. American Journal of Tropical Medicine and Hygiene, 2004; 71:128–135 8. Ropelewski C. The great El Niño of 1997 and 1998: impacts on precipitation and temperature. Consequences, 1999; 5:17–25. 9. Locally known as kushet, with an average of 500 households and a population of 2500 10. Negash K et al. Malaria epidemics in the highlands of Ethiopia. East African Medical Journal, 2005; 82:186–192. 2 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia In July 2004, following a nationwide, multicentre study of the efficacy of sulfadoxine-pyrimethamine, and artemether-lumefantrine, Ethiopia changed its antimalarial drug policy. Artemether-lumefantrine was introduced as the first-line drug after about one year, in April 2005. Additionally, during the period 2005–2007, 17.7 million long-lasting insecticidal nets out of a planned 20 million were distrib- uted throughout the country, covering about 8.8 million households in malarious localities, at an average of two nets per household.11 Ethiopia continues indoor residual spraying in epidemic-prone areas every year. Tigray is the northernmost regional state of Ethiopia, located between latitude 12° and 15° North. The region has 47 districts (35 rural and 12 urban) in five zones.12 The lowest elevation is at Erob (170 m) and the highest mountain peak at Alaje (3923 m). The region is divided into northwestern and southern lowlands and central highlands and covers 80 000 km2. The projected population (from the 1994 census) of the region was 4.5 million in July 2007, with an estimated growth rate of 3% per year. The average temperature ranges from about 22 °C in areas below 2400 m to less than 16 °C at higher altitudes; the average temperature in most of the lowland areas where malaria is endemic is about 27 °C. The temperature and rainfall patterns of selected stations are presented in Appendix A (Maps 2 and 3). Major population settlement movements are usually from the highlands to the lowlands, in a search for unexploited, agriculturally rich land. Migration also has a temporal pattern, highlanders being attracted to the agriculturally rich lowlands in search of seasonal employment, such as weeding and harvesting; harvesting overlaps with the major malaria transmission season of September to November. Almost 75% of Tigray is malarious, and about the same proportion of the population resides in these areas and is therefore at risk for malaria. Malaria transmission is seasonal and depends on both altitude and rainfall. Transmission varies widely with the complex topography, which ranges from high-altitude plateaux and mountainous terrain to deeply incised river valleys and canyons. At altitudes above 2000 m, the temperature is generally too low to support the development of the parasite in the mosquito vector. As in the rest of the country, Anopheles arabiensis is the major malaria vector in the region. It breeds in small sun-exposed pools, mainly during the rainy seasons. Development of the malaria parasite and vector longevity depend both on tempera- ture and relative humidity. Favourable conditions for increased transmission prevail after the June–August rainfall, i.e. in September, October and November, and these months represent the major transmission season. In eastern and southern Tigray, there is a second transmission season, in May and June, following the minor rains 11. Federal Democratic Republic of Ethiopia, Ministry of Health (2006). National five-year strategic plan for malaria prevention and control in Ethiopia, 2006–2010. 12. A political administrative unit composed of several geographically contiguous districts Chapter 1. Introduction 3 of March and April. As in other areas of Ethiopia, malaria is unstable in Tigray: transmission is hypoendemic, with crude parasite rates in cross-sectional surveys ranging from 3–10% during high-transmission months and 0–3% during low-trans- mission months.13 The unstable nature of malaria makes the region prone to epidemics with high morbidity and mortality in all age groups, making this disease a major public health problem. Furthermore, as malaria transmission strikes during the planting and harvesting season, it adversely affects food security and impover- ishes and isolates affected communities. At the end of 1991, traditional malaria control strategies (institution-based diagnosis and treatment, selective indoor residual spraying and larviciding) and epidemic control were re-introduced into Tigray after many years of interruption due to the civil war. During the 1990s, many new interventions were introduced in the region, including insecticide-treated nets and pilot use of rapid diagnostic tests. Most febrile patients (70%) were previously treated by community health workers at village level, where the first-line treatment was chloroquine and, later, sulfadoxine-pyrimethamine. According to the health profile of the region in 2005,14 malaria is a leading cause of outpatient visits, admissions and deaths in health facilities. 13. WHO. The community-based malaria control programme in Tigray, Ethiopia. A review of programme set-up, activities, outcomes, and impact. Tigray, Malaria Control Department, Regional Health Bureau; Geneva, Technical Support and Capacity Development, Roll Back Malaria, Communicable Diseases (WHO/CDS/RBM/99.12. WHO/MAL/99.1090). 14. Tigray Regional Health Bureau. Annual health profile for 2001/002. Unpublished. Mekelle, Tigray, 2005. 4 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Chapter 2. Study area 2.1 Geography and demography The study area covers two districts located in the southern part of Tigray, collectively called the Raya Valley. The intervention was conducted in the Alamata district, while the Raya Azebo district served as a control. Urban and rural settle- ments and health institutions in both districts have been mapped (see Appendix A, Map 1). Alamata, the intervention district, is located in the southern part of the region. It has an average population of 89 377, living in 12 subdistricts (locally referred to as tabias15) in 64 villages (known as kushets). The altitude of the villages ranges from 1438 m to 2571 m. Most of the population (81%) lives at altitudes of less than 2000 m. At the start of the project, the potential health service coverage in the district was 24%, defined as a target of one health post per 5000 people. There are 33 volunteer community health workers, who make early diagnoses, provide prompt treatment, promote proper use of insecticide-treated nets and mobilize residents to participate in environmental management. Raya Azebo, the control district, is located in the same zone and has an average population of 118 693 living in 13 subdistricts and 58 villages. The altitude of the villages ranges from 1459 m to 1908 m. Most of the population (98.5%) lives at altitudes 1500–1908 m above sea level. The potential health service coverage was 36% at the start of the project. The 56 volunteer community health workers, who formerly provided early diagnosis and prompt treatment with sulfadoxine-pyrimeth- amine, were gradually reoriented after implementation of the new treatment policy with artemether-lumefantrine to promote preventive measures, such as use of insecticide-treated nets and community mobilization for environmental management. 15. A subdistrict or locality that is the most peripheral administrative unit, comprising a cluster of up to five kushets (average, four), with an average population of 10 000. Chapter 2. Study area 5 2.2 Potential effect on vector control of malaria epidemic during the study 2.2.1 Meteorological conditions In areas of unstable malaria transmission, such as Tigray, transmission is influenced by the local microclimate and by seasonal, interannual and periodic climate variations. Several authors have offered explanations for the associations between resurgence and malaria epidemics and climate in Ethiopia. It was therefore considered important to investigate the meteorological conditions prevailing during the epidemic in the study area in June, July and August 2005. Examination of the trend in morbidity during the study period (Phases I and II) and comparison with mean 3-year data for May–July and September–November showed that a malaria epidemic occurred during Phase I of the project (Fig. 1). The meteorological data for this period were obtained from the Alamata district station located at 39°68 E, 12°51 N, at an altitude of 1700 m, on the border of the project districts. The data on rainfall from this station are partly applicable to the study area, as there are no major obstacles between the two districts. The temperatures in some relatively high villages (e.g. 1900 m above sea level) might be a few degrees lower than those recorded at the station. Number of patients treated (thousands) 1.1 1.2 1.7 1.7 2.1 1.3 0 1 2 3 4 5 6 7 8 9 May Jun. Jul. Aug. Sept. Oct. Nov. Dec. Jan. Feb. Mar. Apr. Patients treated by community health worker in Phase I (after clinical diagnosis) Patients treated by community health worker in Phase II (after clinical or RDT diagnosis) Monthly average of patients treated by community health worker in 1996–1998 Figure 1 Numbers of patients treated per month by community health workers during Phases I and II of the project and during the period 1996-1998 6 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Meteorological data for 2005 were compared with mean values for the period 1999–2006, excluding the epidemic year 2005 in order to diminish the bias produced by the abnormal rainfall. It was not possible to analyse a longer period because of interruptions due to the civil war and inauguration date of the station. The mean monthly temperatures were calculated as the average of the mean maximal and minimal monthly temperatures, giving a generally acceptable approximation. In an average year, there is bimodal rainfall: a minor rainy season in March–April and a major one in July–August (Fig. 2). The temperatures form a perfect curve, peaking in June and dropping in December and January, the coldest months of an average year (Fig. 3). Location of the meteorological station 0 300 250 150 100 50 MayApr.Mar.Feb.Jan. Jun. Jul. Aug. Sept. Oct. Nov. Dec. 200 30 0 30 60 km N Meteorological station location Alamata Raya Azebo Rainfall (mm) Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sept. Oct. Nov. Dec. Mean minimum Mean maximum Mean monthly 1999–2006 Mean monthly 2005 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Degree Celsius Figure 2 Average annual rainfall, 1999–2006, and location of the meteorological station Figure 3 Average monthly temperature, 1999-2006, at Alamata meteorological station Chapter 2. Study area 7 In an average year in the study area, the temperatures are sufficient for the development of sporozoites (P. falciparum) throughout the year; however, transmis- sion is lower between December and January, when the mean temperature is between 18 and 19 °C. In 2005, the temperatures were above average in almost every month of the year, especially during the first quarter (Fig. 4). A difference of more than 1 °C was observed between January–March and September–November. It is significant that the temperatures were well above the average during the cool seasons of 2005, so that transmission was not inhibited during that time. The rainfall at the end of the cool season (March–June) in 2005 was excessive, but it was well below the average thereafter (July–December) (Fig. 5). Figure 4 Temperature deviation in 2005 compared to average values for 1999–2006 (difference of monthly mean temperatures) Alamata station Figure 5 Rainfall deviation in 2005 compared to average values for 1999–2006 (difference of monthly rainfall) Alamata station Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sept. Oct. Nov. Dec. 0 0.5 1.0 1.5 2.0 2.5 – 0.5 – 1.0 – 1.5 – 2.0 Degree Celsius Difference in monthly mean temperature 100 0 –100 –50 50 –150 Monthly rainfall (mm) Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sept. Oct. Nov. Dec. Difference in monthly mean rainfall 8 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia The period around the minor rains of 2005 was therefore particularly favour- able for more intense malaria transmission in the study area, because of relatively high temperatures during the cool season at the beginning of the year and above- normal rainfall in March, April and May. The higher-than-average temperatures throughout most of the year might have prolonged the epidemic during the below- normal rainy season in July–August 2005. 2.2.2 Concurrent vector control Malaria transmission varies not only with the amount and pattern of rainfall and with temperature but also with the effectiveness of curative and preventive activities, including residual insecticide spraying and use of insecticide-treated nets for vector control. In the study, the coverage and timing of vector control were controlled to the maximum extent possible. Before the project, the vector control in both districts was applied as one- or two-cycle residual insecticide house spraying with malathion, depending on the forecast for the minor rainfall season. As the minor rains had been insignificant in the valley for several years, second-cycle spraying was not done during that time, including 2005. This decision was reinforced by the introduction of insecticide-treated nets on a large scale through a cost-sharing scheme in 2004, although initially this had limited coverage and limited public health benefit. Although the intervention was promoted in all the subdistricts of the region, coverage at household level was very low, as the nets were distributed only to those who were willing and could buy them. Moreover, the distribution overlapped with a period of drought, when people had other priorities than paying for insecticide-treated nets. Therefore, there was no significant vector control during the 2004 transmission seasons, except for re-impregnation of existing nets. The above-normal rainfall in March, April and May 2005 was not foreseen; however, this would have made little or no difference, as there was no stockpile of malathion in the region or in the country. Control of the epidemic thus depended largely on case management in the first few months, until the insecticide was made available, in August 2005. As malathion was still not available in the country, a plan for vector control was implemented with DDT (dichlorodiphenyltrichloroethane) as a 75% water- dispersible powder. There was blanket coverage in both the intervention and the control districts, with 100% of the planned villages and households being sprayed. After the impact of the major transmission season had been curbed, vector control coverage of both districts was strengthened from November 2005 onwards by free distribution of insecticide-treated nets from the Global Fund. Districts authorities covered the villages that they considered priorities, and 12 208 and 16 810 households in the intervention and control districts, respectively, had at least one net. Details of the vector control activities in the two districts are given in Table 1. Chapter 2. Study area 9 Main vector control activities Alamata (intervention district) Raya Azebo (control district) Planned Achieved Percentage Planned Achieved Percentage July 2004–June 2005 (Gregorian calendar); 1997 (Ethiopian calendar) regular residual insecticide house spraying (june–july 2004) interrupted due to failure of malathion production in the country Villages covered unit structures sprayed Population protected insecticide used (75% ddt) house-to-house re-impregnation of old nets distributed through a cost-sharing scheme; poor coverage per household because of expectation of free distribution Villages covered 34 34 100 58 58 100 total nets treated 10 217 10 274 100.5 17 601 28 911 164.0 households covered 13 168 13 247 100.5 18 974 14 702 77.5 Population protected 54 597 34 578 63.4 76 990 46 515 60.0 K-o tabs* used – 10 274 – – 14 702 – distribution of new nets (Permanet®) Villages covered households with at least one insecticide-treated net Population protected July 2005–June 2006 (Gregorian calendar); 1998 (Ethiopian calendar) regular residual insecticide house spraying after the june 2005 epidemic to prevent aggravation of the epidemic after the main rainy season, june, july and August 2005 Villages covered 33 36 100 49 49 100 unit structures sprayed 18 896 21 976 116.0 25 416 31 873 125.4 Population protected 61 105 71 091 116.0 85 591 106 570 124.5 insecticide used (75% ddt) – 5 830 – – 8 928 – supplementary house-to-house re-impregnation of old nets Villages covered 32 32 100 58 52 89.7 total nets treated 10 670 9 276 86.9 9 827 9 554 97.2 households covered 12 706 11 435 89.9 28 911 14 331 49.6 Population protected 57 179 37 586 65.7 44 016 19 993 45.4 K-o tabs* used – 9 276 – – 9 554 – distribution of new nets (Permanet®) Villages covered 37 36 97.3 41 40 97.6 households with at least one insecticide-treated net 20 008 12 208 61.0 24 572 16 810 68.0 Population protected 75 139 45 847 61.0 121 150 63 037 52.0 * K-O tab is a water dispersible tablet, containing deltamethrin, which is mixed with water to prepare insecticide to treat mosquito-nets. Table 1. Main vector control activities in the study area 10 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 2.3 Health service infrastructure in Tigray In 1991, upgrading of general health services began, including rehabilitation or construction of health facilities, training and deployment of health personnel and expansion of maternal and child health immunization services. The consequent increase in the number of institutions was a step towards improving the access of the rural population to health care. In 2006 in Tigray, there were 13 hospitals, 42 health centres, 116 nucleus health centres16, 86 clinics and 424 health posts. The physician to patient ratio was 1:111 000, which was still far below the projected target of 1:10 000. At that time, there were 37 physicians, 70 health officers, 1379 nurses, 1200 health extension workers, 3779 traditional birth attendants and 2114 community health workers. The distribution of health services and health personnel in the Alamata and Raya Azebo districts in 2006 is given in Table 2 and Table 3. Table 2. Health services in Alamata and Raya Azebo districts, 2006 District No. of hospitals No. of health centres No. of nucleus health centres No. of clinics No. of health posts Alamata 1 1 4 1 9 raya Azebo 0 1 3 5 9 Table 3. Numbers of professionals working at health facilities in Alamata and Raya Azebo, 2006 District No. of physicians No. of health officers No. of nurses No. of health extension workers No. of traditional birth attendants No. of community health workers Alamata 1 3 49 23 62 33 raya Azebo 0 4 42 25 74 – 16. Clinic or health post that will be upgraded to health centre. Chapter 3. Methods 11 Chapter 3. Methods 3.1 Study design It was not possible to conduct a randomized, controlled trial with the smallest population groups – villages – as the study units because of ethical considerations and potential contamination of study units allocated to different interventions. It was therefore decided to select entire districts, i.e. large geographical areas with similar malaria eco-epidemiological conditions and operational convenience of access. The project was implemented as an observational study with random assign- ment of the intervention to one district, the other serving as control. Two convenient, comparable districts in the Raya valley of southern Tigray were selected and were randomly allocated to the intervention arm (health service plus community distribution of artemether-lumefantrine combined with rapid diagnostic test: Alamata district) and the control arm (distribution of artemether- lumefantrine in the health services only, according to the national treatment policy: Raya Azebo district). There was a risk for cross-contamination of the control area, in the form of patients seeking better access, especially in areas bordering the districts. While such contamination could not be prevented, a mechanism for analysing patients according to their address was devised. A schematic presenta- tion of the study design is given in Table 4. All other antimalarial interventions were implemented with high levels of coverage, over similar periods in both districts. These included both indoor residual spraying and widescale distribution of insecti- cide-treated nets (Table 1). Table 4. Study design Arm Year Health-care level Health institution Community Intervention: Alamata district 1 Artemether-lumefantrine + rapid diagnostic test (as per new drug policy) distribution of artemether-lumefantrine at village level; 100% of community health workers (n = 33) 2 Artemether-lumefantrine + rapid diagnostic test (as per new drug policy) distribution of arte- mether-lumefantrine; 50% of community health workers (n = 17) (treatment based on clinical diagnosis) distribution of artemether-lume- fantrine + rapid diagnostic test; 50% of community health workers (n = 16) (treatment based on diagnosis confirmed by rapid diagnostic test) Control: raya Azebo district 1 Artemether-lumefantrine + rapid diagnostic test (as per new drug policy) no treatment services at community level (as per national treatment policy) 2 Artemether-lumefantrine + rapid diagnostic test (as per new drug policy) no treatment services at community level (as per national treatment policy) 12 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 3.2 Data collection As part of the standard operating manual and curriculum for training project staff at various levels, a series of pre-coded case recording forms, a standard WHO survey questionnaire and guidelines (survey, assessment, treatment and referral) were developed or adapted, pre-tested, standardized and used. The main indicators selected, data sources and end-points are shown in Table 5. Table 5. Framework for data collection on indicators No. Indicator Source of data End-point measured 1 community-based mortality rates Mortality census in both districts impact on malaria mortality 2 Monthly malaria morbidity rates Monthly slide positivity rate Proportions of P. falciparum and P. vivax Monthly malaria admission rate Annual hospital malaria case fatality rate Annual health institution-based proportional mortality rate routine data collected longitudinally impact on malaria morbidity and mortality impact on malaria transmission impact on falciparum transmission 3 community-based malaria prevalence rates cross-sectional malaria prevalence surveys impact on malaria transmission 4 Proportions of malaria cases reported and treated by community health workers and in health facilities over time use of community health worker services (by gender and age) community health workers’ drug prescription practices data reported routinely by community health workers number of malaria patients treated with artemether-lumefantrine by community health workers and in health facilities over time in both intervention and control districts expected versus actual number of tablets prescribed 5 Proportional reduction of cost of management by community health workers of uncomplicated malaria with artemether- lumefantrine + rapid diagnostic test and with artemether-lumefantrine given on clinical basis routinely reported data from community health workers after introduction of rapid diagnostic tests in second year of study cost savings on disease management with use of rapid diagnostic test vs. clinical diagnosis by community health workers 6 sensitivity, specificity, positive and negative predictive value of rapid diagnostic tests performed by health workers and community health workers compared with expert microscopy rapid diagnostic test performance survey in health institutions and among community health workers diagnostic performance of rapid diagnostic tests at different levels of service provision 7 Adherence of patients to artemether- lumefantrine treatment by level of service provider survey of adherence to treatment Adherence of patients to six-dose artemether-lumefantrine regimen in intervention district 8 index of satisfaction Focal group discussion beneficiaries’ satisfaction 9 rapid diagnostic test performance by community health workers survey of rapid diagnostic test performance by community health workers skill of community health workers in performing rapid diagnostic test Chapter 3. Methods 13 3.2.1 Institution-based morbidity, slide positivity, malaria admissions, case fatality and proportionate mortality rates During the project, data were collected from routine reports at health institu- tions, compiled by district health personnel and relayed to the regional health bureau. The data included records from outpatient departments, slide positivity rates, the “parasite formula” (relative proportions of P. falciparum and P. vivax infections), records from inpatient departments and deaths in institutions. Data were reported monthly to allow monitoring of trends and to serve as indicators of project achievements. Mechanisms for tracing missing notifications, verifications and corrections of data were instituted in a database management scheme established in the Tigray Health Bureau. 3.2.2 Institution-based survey of adherence to treatment and study of performance of rapid diagnostic tests in the intervention district The survey on adherence to treatment and evaluation of the performance of the rapid diagnostic test were conducted simultaneously by different service providers in Phases I and II. The sample size was determined for two levels of health-care service (hospital + health centres and health posts + clinics, respectively) to estimate adherence to treatment in three age groups (< 5, 5–14 and > 15 years), assuming 80% adherence to treatment in all age groups and confidence limits of 10%. In order to account for an expected 10% loss to follow-up, the final sample sizes were calculated to be 68 and 204, respectively, for each age group at the two levels of health-care service. Questionnaire and survey guides adapted from the WHO Global Malaria Programme were translated into the local language. The survey personnel were health professionals serving in outpatient departments, with proper training in conducting the performance study and the adherence survey. Patients with uncompli- cated P. falciparum malaria confirmed by either the rapid diagnostic test or micros- copy and to whom artemether-lumefantrine had been prescribed were enrolled and then visited at home on the day they were expected to finish the drug course or the next day by an interviewer. Patients who were absent were revisited on the same day or up to 2 days after the initial attempt. Patients who could not be found on these occasions were described as lost to follow-up. Patients were defined as “probably adherent” if they reported taking all doses on the correct day at the correct time and in the correct amount (replacing any dose if vomiting occurred within 30 min), and with information on all doses of artemether-lumefantrine in missing or empty blister packs. Patients were defined as “probably not adherent” if they did not report taking all doses on the correct day, time and amount or with no replacement in the event of vomiting within 30 min (providing doubt that the patient had followed the instructions correctly) 14 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia and with either missing or empty blister packs. Patients were defined as “definitely not adherent” if they had leftover tablet(s) in the blister pack. The first dose of artemether-lumefantrine was given under direct supervision on day 0. During the home visit, patients were asked about their adherence to the remaining five doses. Patients were not informed about the follow-up home visit on day 0; consent was obtained during the visit. Corresponding thick and thin blood films were collected from the same patient and the same finger-prick sample and labelled with an identification code identical to that of their corresponding rapid diagnostic test. A thin smear of blood film was fixed upon preparation. Blood films were stained within 3 days of collection with 3% Giemsa for 30 min. Thick films were examined in district laboratories, and all slides were cross-checked at regional level by senior technicians. Any discordant readings were assigned a final result by a third expert microscopist at regional level. 3.2.3 Community-based malaria morbidity reports in the intervention district During the project, data on the community-based treatment services were recorded by the community health workers using standard patient record forms, then collated and compiled by district health personnel and relayed to the Tigray Health Bureau. The data included: • patient’s age, gender, address and travel history; • treatment services provided, on the basis of either clinical examination or the rapid diagnostic test result (when applicable); • type of medicine administered, and stocks of antimalarial drugs and rapid diagnostic tests available; and • referrals to health institutions. These data were reported weekly in order to monitor trends and to serve as indicators of project achievements. Mechanisms for tracing missing data, verifica- tion and correction by database management were instituted at the Tigray Health Bureau. 3.2.4 Community-based surveys Serial malaria prevalence survey and survey of knowledge, attitude and practice in control and intervention districts The sample size for the malaria prevalence survey was estimated for an observational study from tables for estimating a proportion with specified absolute precision. Previous community-based studies in Ethiopia in areas of intense seasonal Chapter 3. Methods 15 malaria transmission have shown a malaria parasite prevalence of 5% in the high-transmission season.17,18 For an anticipated infected population of 5%, a confidence level of 95% and an absolute precision of 2 percentage points (3% to 7%), the required sample size was 456.19 To correct for the effect of cluster sampling, a design effect of 3 was assumed, giving a required sample size of 1368 in each district. With a mean family size of 4.6, it was estimated that 300 households in 30 clusters from each district were required; 10% was added as a safety margin. The numbers of households required for the blood survey were also adequate for the knowledge, attitude and practices survey. The same kushets were visited for all prevalence surveys in order to monitor temporal variations. The kushets comprising the primary sampling units were chosen by popula- tion proportional to size sampling. Within each kushet, the household was the ultimate sampling unit. Households were chosen from a random starting point: in the centre of each kushet, a pen was spun, all the houses in a line between the pen point and the kushet’s boundary were numbered, and the first household was chosen randomly. Subsequent households were chosen by serial proximity. If no household members were present at the time of the survey visit, the closest household was chosen. The surveys were undertaken twice each year in both districts. The initial survey was conducted in April 2005 and the second in September 2005. Both studies were carried out in the low- and high-transmission seasons by 10 trained surveyors (five per district). The same surveys were repeated in April 2006 and September 2006 (Phase II), with the same methods and the same study teams. During blood film collection, information about recent febrile illness and use of community health workers or health facilities was recorded by interviewers for each household member on a standard questionnaire, with information on knowledge about malaria and practices in the household. Community acceptance of the teams during the survey was ensured by preliminary meetings with district health officials and local governing bodies before the survey was conducted, in order to maximize blood collection and responses. Blood films were collected from all household members in the early morning before the family members left home for their daily activities, and the knowledge, attitude and practices interview was conducted later the same day. During the interview, blood films were collected from people who had been absent during the morning session. If the woman of the house was present, she 17. Nigatu W et al. Some aspects of malaria prevalence, vector infectivity and DDT resistance studies in Gambella Region, South Western Ethiopia. Ethiopian Journal of Health Development, 1994; 8:1–10. 18. Malaria Eradication Service. Guidelines for parasitological studies in indicator areas. Addis Ababa, Ministry of Public Health, 1968. 19. Lwanga SK, Lemeshow S. Sample size determination in health studies. A practical manual. Geneva, World Health Organization; 1991:25. 16 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia was interviewed for the knowledge, attitude and practices survey; if not, the male household head was questioned. The questions asked during both Knowledge, Attitude and Practice (KAP) surveys are listed in Table B1 (Appendix B). All individuals who reported fever within the previous 24 h were referred to the community health worker or the health post for evaluation after the blood film had been taken. Blood films were stained within 3 days of collection with 3% Giemsa for 30 min, according to the WHO standard protocol. Films were examined independ- ently by two experienced laboratory technicians. For the thick film examination, parasites were counted according to WHO method against 200–500 white blood cells.20 A film was declared negative if no parasites were observed in 100 fields. Discordant results were checked and confirmed by a senior technician. Owing to the high illiteracy rates in Tigray, verbal informed consent was routinely obtained from the participants. Survey of adherence and performance of rapid diagnostic tests by community health workers in the intervention district An adherence survey at community health worker level was conducted in both Phases with the same protocol as used in the institution-based survey described above. In the first Phase, the enrolled patients were clinically diagnosed cases of malaria, while in the second Phase P. falciparum positivity had been confirmed with the rapid diagnostic test. In Phase I, 10 community health workers who had prescribed artemether- lumefantrine presumptively were selected for the adherence survey. The follow-up home visits were conducted by health extension workers. In Phase II, studies of rapid diagnostic test performance and adherence surveys were conducted simultaneously in the catchment areas of all 16 community health workers who treated on the basis of the rapid diagnostic tests, with the same questionnaire and survey guide as used in the health institutions. Patients with a diagnosis of P. falciparum infection confirmed by the rapid diagnostic test and who were given artemether-lumefantrine were enrolled in the adherence survey and then visited at home on the day they were expected to finish the drug or the next day by an interviewer. Patients who were absent were revisited on the same day or up to 2 days after the initial attempt. Patients who could not be found on these occasions were described as lost to follow-up. During Phase II of the project, 16 high-school graduates were recruited from the intervention district as enumerators, to serve as counterparts to the community health workers. They received a 3-day training in the procedures of the adherence 20. WHO. Bench aids for the diagnosis of malaria, 2nd ed. Geneva, World Health Organization, 2000. Chapter 3. Methods 17 survey and blood film preparation. Additional training was given for rapid diagnostic test preparation and safety procedures, to enable them to observe and record the procedures carried out by the community health workers. Rapid diagnostic tests prepared by the health workers were read 15 min later, at the time the last buffer was dropped into the well. Corresponding thick and thin blood films were collected from the same patient and the same finger-prick sample and labelled with an identi- fication code identical to that of their corresponding rapid diagnostic test. A thin smear of blood film was fixed on preparation. Blood films were stained within 3 days of collection with 3% Giemsa for 30 min. Thick films were examined in the district laboratories, and all slides were cross-checked at regional level by senior technicians. Any discordant readings were assigned a final result by a third microscopist at regional level. The enumerators had the additional task of observing and recording the performance and safety procedures conducted by community health workers in patient care and rapid diagnostic test preparation. A questionnaire with 30 variables was prepared and completed by assigned enumerators without the knowledge of the health workers, on the basis of direct observation of routine procedures. To ensure that the health workers would behave as usual, they were told that the enumera- tor’s mission was to help them collect corresponding blood films for each rapid diagnostic test undertaken, for the study of test performance. Mortality survey with verbal autopsy techniques in both districts Verbal autopsy interpretation based on InterVA model21 was used to document the mortality rate in Phase I of the project, covering the period May 2005 to April 2006. All deaths during the specified period were captured in a census of 109 villages in both districts carried out by 109 high-school graduates (58 and 51 temporary enumerators in the control and intervention districts, respectively). The questionnaire was adapted and translated into the local language, Tigrigna. Before the enumerators were trained, the local language version was pre-tested in a rural village and then corrected and edited for training. It was then used to collect data on the identity of deceased individuals, documenting the sequence of events leading to death (primarily the signs and symptoms of the illness preceding death). In the analysis of results obtained from the InterVA model, one or more specific cause of death was attributed for each death. A second survey was conducted with the same survey tools and techniques, the same numbers of enumerators and in the same villages one year later to evaluate the mortality rate during Phase II of the project, covering the period May 2006 to April 2007. 21. Fantahun M et al. Assessing a new approach to verbal autopsy interpretation in a rural Ethiopian community: the InterVA model. Bulletin of the World Health Organization, 2006; 84:204–210. 18 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Focus group discussions in the intervention district at the end of the project A qualitative study based on focus group discussions was conducted among beneficiaries to assess their satisfaction with the use of rapid diagnostic tests and distribution of artemether-lumefantrine at community level. All people who were permanent residents in the intervention district, ≥ 18 years old and had been treated at least once by community health workers during the project were eligible to take part in the discussions. The discussions were held in villages that had community health worker treatment services with and without the rapid diagnostic test, in separate male and female groups. The group size was restricted to eight people to ensure individual participation and to allow sufficient time for each speaker, as well as to reduce dominance within groups or separate conversations among participants. The issues addressed included the extent of the malaria problem before and after the project, the courtesy and respect shown by the community health workers, the services provided and factors that adversely affected the project. Participation in the discussions was voluntary. Respondents were told that their views were important in order to improve the malaria control programme. Permission was obtained from the relevant bodies of the Alamata District Administration and the district health office. Before participating in the groups, people were asked to give consent. A malaria expert and two assistants from the district health office recruited participants, took notes and recorded the discus- sions while respecting participants’ privacy, thus encouraging them to be as honest as possible during the discussions. 3.2.5 Pharmacovigilance During Phase I of the project, a programme of pharmacovigilance was introduced, which was based on spontaneous reporting. The WHO draft protocol for pharmacovigilance of artemisinin-based combination therapies (version 2003)22 was adopted, and training was given to a limited number of health workers in Alamata Hospital and Alamata Health Centre, introducing a standard case reporting form to report serious adverse events suspected to be related to artemether-lumefantrine. The flow of spontaneous reports of suspected adverse drug reactions was planned as follows: bi-weekly collection of case reporting forms filled in by trained health workers for submission to the Malaria Department in Alamata; monthly submissions from the Malaria Department to the Tigray Health Bureau; and, every two months, transfer of data from the Tigray Health Bureau to the Italian Dermatological Centre in Mekelle for entry into a safety database. Subsequently, a more extensive training programme was organized for 18 health workers, comprising a physician, four 22. WHO. Draft protocol for pharmacovigilance of artemisinin-based combination therapies. Unpublished document. Geneva, World Health Organization, 2003. Chapter 3. Methods 19 nurses, eight pharmacy technicians and five health officers. Both training programmes focused on the importance of pharmacovigilance and procedures for spontaneous reporting of suspected adverse drug reactions after treatment with artemether-lumefantrine. The adverse drug reaction reporting form was revised at the end of the first year of the project, printed and distributed to all health facili- ties in the intervention district (Alamata). A “district investigating team” was established to coordinate adverse drug reaction activities in the district and to submit reports on such reactions to the Tigray Health Bureau, and a system for reporting was put in place. As no adverse drug reaction was reported through the spontaneous reporting system established in Phase I, an active pharmacovigilance system was introduced during Phase II of the project in order to collect information prospectively on all adverse events occurring after treatment with artemether-lumefantrine in selected health facilities in the intervention district. Specific reporting forms were designed and field-tested with patients and health workers, and the protocol was finalized in February 2007. Fourteen health workers were selected and trained to conduct the study at the Alamata Hospital, the Alamata Health Centre and the Tumuga Health Centre. All the patients participating in the active pharmacovigilance study were enrolled on the basis of uncomplicated P. falciparum malaria confirmed by rapid diagnostic test or microscopy, with fever or a history of fever within the previous 24 h. For each patient, the first dose was given under direct observation, and the patient was retained for 30 min to see whether the medicine was vomited. The patients were asked to return to the health facility for a follow-up visit 3 days after the start of treatment. If they did not do so, health workers made a home visit 4– 6 days after the start of treatment. At the follow-up visit, the health worker made a full clinical evaluation, collected a blood specimen when there was clinical deteriora- tion or lack of improvement, and completed the adverse drug reaction report. The planned sample size for the study was 3500 malaria patients with complete follow-up, in order to detect specific adverse events that occurred at an incidence of 1 per 1000 patients with 95% probability. The active pharmacovigilance study was conducted in the period June–September 2007. 3.2.6 Capacity-building and database management Health professionals In April 2005, 81 health workers in eight health institutions in the interven- tion district received training on malaria diagnosis and treatment, with emphasis on the use of rapid diagnostic tests and administration of artemether-lumefantrine. The health workers’ professions ranged from health extension workers to health officers. Training was accompanied by practical work on rapid diagnostic test 20 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia preparation and interpretation and demonstration of drug packs. After training was completed, artemether-lumefantrine was distributed. Health workers in both districts were given serial training and were involved in surveys of malaria prevalence, knowledge, attitude and practices about malaria, and mortality surveys during the project. In addition, health workers in the intervention district were trained and involved in surveys of adherence to treatment and evaluation of rapid diagnostic test performance. Community health workers In April 2005, 33 former community health workers in the intervention district received a three-day training in malaria diagnosis, treatment with artemether- lumefantrine, patient counselling and referral of cases of severe illness or signs and symptoms probably not due to malaria. The training also covered proper storage and use of the drug, recording and reporting forms and mechanisms for ensuring re-supply of the drug. Each community health worker was given a book containing reference and treatment charts with age-specific regimens, as well as posters for patient education. Refresher training was conducted in April 2006 for the health workers randomly selected to treat patients on the basis of the rapid diagnostic test, this time including safe use of the test for diagnosing malaria. Enumerators for mortality surveys The mortality survey was implemented in both the intervention and the control districts in May 2006 and 2007. Each year, 117 interviewers (all high-school graduates) and 10 supervisors in both districts participated in a three-day course on the survey questionnaire and data collection techniques. On the first day, the questionnaire was introduced, and interviewing practice was covered. This helped to ensure that the enumerators asked the questions correctly. The second day was devoted to a rehearsal in a nearby village and a plenary feedback session on its findings. On the third day, feedback was given on performance during the field test, and discussions were held. Post-graduate training Three project managers received post-graduate training: two participated in short training courses in epidemiological field research methods and the other was enrolled into a Master of Public Health training course at Umeå University, Sweden. The objective of these courses was to strengthen the capacity of the malaria control department of the Tigray Health Bureau in scientific management of the project. Data management: data entry, cleaning and analysis All data generated by the community health workers and health institutions were reported to the routine health management information system, compiled by Chapter 3. Methods 21 the malaria control office in each district, and then relayed to the regional malaria control department in Mekelle. Data were sent weekly by community health workers via the health institutions to the district health office (malaria control unit), then relayed to the regional malaria control department for entry, cleaning and analysis. Data relay included collating, tallying and compilation, especially at district level. Data obtained through surveys were either collected directly by the project focal person or relayed via the malaria control unit to the Tigray Health Bureau. The database for entry, cleaning and analysis was established by a senior data manager under the guidance of the project coordinator in Tigray, on the basis of experience in data management from the community-based malaria control programme.23 While routine data from the health institutions and community health workers were entered by temporary data technicians, large-scale survey data (from the mortality survey, the study of rapid diagnostic test performance, the adherence survey, the prevalence study and the knowledge, attitude and practices household survey) were entered by data technicians recruited on a contractual basis. Cleaning was done routinely by the data manager and project supervisors, who cross-checked the data entered with the original case record forms. Except for the epidemic period, during which all routine data reporting was constrained and data flow slackened, most data management activities were undertaken with no significant delay. Data cleaning and analysis of the results of the prevalence and mortality surveys were the most complicated. For the prevalence surveys, discordant slide results from two independent microscopists had to be validated. The data from the mortality survey were difficult to verify against the original case recording forms and to clean. The qualitative data from the focus group discussions were transcribed manually from audio-taped records, and the results were analysed by summarizing the ideas put forward by the group members. EPI version 6 software was the main tool used for questionnaires, entry and analysis. 23. Tigray Regional Health Bureau. Annual health profile for 2001/2002. Unpublished. Mekelle, Tigray, 2005. 22 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Chapter 4. Results and discussion 4.1 Profile of community health workers and training outcomes 4.1.1 Characteristics of community health workers The 33 community health workers involved in the project were a subset of the hundreds of community volunteers who were treating patients close to home before the national treatment policy changed to artemether-lumefantrine. The project built on the established system, with minor modifications to satisfy the requirements of the study design. The survey of community health workers showed that they were all male, with a mean age of 56 (± 6 SD) years, all married and all farmers, and that 93% of them had a primary-school education. Of the 33 community health workers, 32 provided services in their own home. Their mean duration of service before the project was 5.9 (± 4 SD) years. All of them volunteered to continue the service. The community health workers were in the catchment areas of eight clinics or health posts, and the mean walking distance from their treatment post to the nearest health institution was 64 min (± 34 SD). 4.1.2 Training outcomes The first level of training was for health workers from various health institu- tions, ranging from health extension workers to health officers, 81 of whom received basic training with a focus on malaria diagnosis and treatment. The second level of training was for all 33 community health workers, who received a three-day basic course supplemented with periodic refresher training sessions. Training was also given in preparation for the surveys, with periodic refresher or basic re-training depending the type of personnel involved and the frequency of each survey. Three project supervisors underwent high-level training. The different training courses are described in the Methods (section 3.2.6). Chapter 4. Results and discussion 23 4.2 Community treatment services 4.2.1 Numbers of cases treated The number of patients treated by community health workers varied according to the season. During the first year, the health workers in the intervention district treated an average of 12 533 patients per quarter, but with wide variation between seasons (SD ± 7921). Unusually, most cases occurred in the period June, July and August 2005, during the major epidemic, which contributed 41% of the total case load, followed by September, October and November 2005, which contributed 36% of the case load. The remaining cases were spread almost equally over the period March, April and May (the minor rainy season in the study area) and December, January and February, corresponding to the coolest, driest season (see Fig. 1). The seasonal variation in the second year was similar to that in the first year, but the average case load during the first two quarters was threefold lower. In the first two quarters, community health workers treated 6426 patients per quarter on average, with relatively little variation by season (SD ± 680). The cases were more uniformly distributed than in the first year: the period after the minor rainy season (June, July and August 2006) contributed 54% of cases in the first two quarters of the year. The patient load of the two groups of community health workers in the intervention district was expected to be similar, as they were randomly allocated to use or non-use of rapid diagnostic tests, and the number of health workers was approximately equal in the two groups. When rapid diagnostic tests were first introduced in June 2006, although no systematic investigation has been conducted, it appears that clients shifted from health workers equipped with the tests to those who continued to treat on the basis of a clinical diagnosis. Figure 6 and Table 6 show that the patient load of the group without the tests was consistently twofold higher than that of community health workers with the tests. 4.2.2 Pattern of service by age and gender In Phase I, the age and gender distribution of patients was assessed to see if there were disparities in the treatment service provided by community health workers, especially with regard to service use by women and children. Although all age groups in the region are vulnerable, because of the unstable nature of malaria transmission, the access to community health workers by different population groups was assessed from data disaggregated by age and gender. 24 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Table 6. Quarterly community health worker case load in the intervention district and rates of positivity with rapid diagnostic test (Phases I and II) Community health workers 2005–2006 2006–2007 2007 Total May1 jjA son djF MAM jjA son djF MAM june All 33 diagnosing clinically 4 065 20 497 18 181 6 074 5 379 5 8602 60 056 50% (n = 17) diagnosing clinically 7743 69.2% 4 044 68.0% 2 416 68.4% 2 500 66.2% 791 61.1% 10 475 67.2% 50% (n = 16) using rapid diagnostic test 323 30.9% 1 902 32.0% 1 118 31.6% 1 277 33.8% 503 38.9% 5 123 32.8% % rapid diagnostic tests positive 12.4% 12.0% 8.9% 6.6% 14.5% 10.3% jjA, june, july, August; son, september, october, november; djF, december, january, February; MAM, March, April, May 1. As the project was launched at the end of April, the figure for this quarter represents only May. 2. First week of August 3. Weeks 2–4 of August 100% CHWs: treated on clinical basis RDT positive (percentage) MAM, March, April, May; JJA, June, July, August; SON, September, October, December; DJF, December, January, February; CHW, community health worker 0 MAM JJA SON DJF MAM JJA SON DJF MAM 22 100 90 80 70 60 50 40 30 20 10 0 20 18 16 14 12 10 8 6 4 2 Number of patients (thousands) 50% CHWs: treated on clinical basis 50% CHWs: examined with RDTs RDT positive (percentage) 2005 2006 2007 Figure 6 Quarterly case load of community health workers in the intervention district (Phases I and II) Chapter 4. Results and discussion 25 Figure 7 shows the distribution of treated patients by age group during Phase I: 69.4% of patients were over 10 years of age, with 11.2% aged 8–10 years, 14.8% aged 3–7 years and the remaining 4.6% under 2 years. Figure 8 shows the proportion of treated patients in each age group in the general population. The proportion of community health worker-treated patients under 10 years of age was 23% higher than in the total population, while those over 10 years were treated 8% less, on the assumption of uniform malaria risk across age groups. Number of febrile patients treated by community health workers (thousands) 0 6 5 4 3 2 1 May Jun. Jul. Aug. Sept. 2005 Age group (years): 2006 May Jun. Jul. Aug.Oct. Nov. Dec. Jan. Feb. Mar. Apr. < 2 3–7 8–10 > 10 Figure 7 Number of patients treated per month by community health workers by age group (Phase I) Figure 8 Proportions of patients treted by community health workers by age group and age distribution in the general population (according to census data in year 1) Age group (years) Percentage Patients treated by community health workers (percentage) General population 100 80 60 40 20 0 20 40 60 80 100 > 10 < 10 24.930.7 75.169.3 26 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Figure 9 illustrates the distribution of treated patients by gender during Phase I: 41.3% of the patients were female and 58.7% male. Figure 10 shows the distribution of treated female patients by age category during the initial Phase of the project: only 39.2% of patients over 10 years and 47.2% of patients under 10 years were female. Only 78% of females over 10 years were treated as often as expected, i.e. 22% less than their proportion in the general population, while female patients under 10 years were treated more or less proportionally to the population, on the assumption of uniform malaria risk by gender and age group. Number of patients treated (thousands) 0 1 2 3 4 5 6 7 8 9 10 May Jun. Jul. Aug. Sept. 2005 2006 Oct. Nov. Dec. Jan. Feb. Mar. Apr. Total patients Male patients Female patients Figure 9 Patients treated by community health workers by gender (Phase I) Figure 10 Age distribution of female patients treated by community health workers and age distribution of female in the general population (according to census data in year 1) Age group (years) Percentage Female patients treated by community health workers (percentage) Female population 100 80 60 40 20 0 20 40 60 80 100 > 10 < 10 48.847.2 50.439.2 Chapter 4. Results and discussion 27 4.2.3 Prescription of artemether-lumefantrine The distribution of artemether-lumefantrine was assessed by comparing the amount of drug actually dispensed with the expected amount, given the reported age distribution of treated patients. During Phase I, the ratio was approximately 1 (Table 7). Therefore, prescription by community health workers on the basis of the recorded age of the patients was correct. This ratio does not, however, reflect compliance with the six-dose drug regimen, as the health workers supervised only the initial dose of treatment given and recorded the total amount dispensed. Table 7. Quarterly community health worker case load in the intervention district and rates of positivity with rapid diagnostic test (Phases I and II) May 2005 JJA 2005 SON 2005 DJF 2005–2006 MAM 2006 JJA 2006 Total total number of treated patients 4 065 20 497 18 181 6 074 5 379 5 860 60 056 tablets distributed 85 972 426 385 382 478 124 582 111 428 nA 1 130 845 tablets expected1 85 974 427 524 380 196 124 116 111 162 nA 1 128 972 ratio (distributed: expected) 1.000 0.997 1.006 1.004 1.002 nA 1.002 1 expected = [(# < 2 years x 6) + (# 3–7 years x 12) + (# 8–9 years x 18) + (# ≥ 10 years x 24)] tablets jjA, june, july, August; son, september, october, november; djF, december, january, February; MAM, March, April, May; nA, not available A total of 1.1 million tablets of artemether-lumefantrine were dispensed during the period May 2005 to May 2006 to treat 54 498 patients with a clinical diagnosis, and the overall ratio of dispensed versus expected number of dispensed tablets was 1.002 (Figure 11). Figure 11 Drug distribution by community health workers: dispensed versus expected (Phase I) 0 20 40 60 80 100 120 140 160 180 May Jun. Jul. Aug. Sept. 2005 2006 Oct. Nov. Dec. Jan. Feb. Mar. Apr. Total patients treated Number of patients (thousands) Coartem® tablets actually distributed Coartem® tablets expected to be distributed based on the age distribution of patients treated 28 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia The planned use of blister packs as the basic unit to monitor the adequacy of drug distribution in relation to patient age groups was not possible, as there was always an imbalance between what was required and what was available in terms of blister packs. In practice, blister packs were shared among different age groups in order to overcome shortages. Despite this, all patients received the appropriate number of tablets. 4.3 Serial surveys of seasonal prevalence of malaria and of knowledge, attitude and practices 4.3.1 Seasonal prevalence of malaria We conducted four seasonal surveys of malaria prevalence in order to observe the trends in infection, temporally and spatially, in the study districts. Blood films were prepared from an average of 1377 patients in each survey, which was slightly more than the 1368 planned. The age and gender distribution did not differ signifi- cantly between the control and intervention groups. In the intervention district, the crude rate of parasite infection (all stages and all species) during the low-transmission season decreased from 4.9% in April 2005 to 0.4% in April 2006, while the rate in the high-transmission season decreased from 7.4% in September 2005 to 1.2% in September 2006. The control district showed a similar overall downward trend across the seasons (Table 8). Table 8. Prevalence of malaria before and after the intervention, by transmission season Study area Parasite rate April 2005 (low transmission) September 2005 (high transmission) April 2006 (low transmission) September 2006 (high transmission) control (raya Azebo) All stages and species (number) 3.4 (1430) 20.8 (1370) 0.4 (1381) 0.8 (1329) P. falciparum slide positivity rate 2.4 14.2 0.1 0.6 P. falciparum gametocyte rate 0.8 7.0 0.0 0.2 intervention (Alamata) All stages and species (number) 4.9 (1416) 7.4 (1382) 0.4 (1406) 1.2 (1303) slide positivity rate 4.3 4.9 0.2 0.5 gametocyte rate 1.5 1.4 0.1 0.2 Chapter 4. Results and discussion 29 The main difference between the two districts can be seen by comparing the initial status and the aftermath of the epidemic in June, July and August 2005. At the start of the project, the parasite rate in the two districts was broadly similar, with slightly higher rates in the intervention district in all measurements. In the high-transmission season (September 2005), after the major epidemic of June, July and August 2005, however, the control district had a threefold higher crude parasite rate (all species and stages), a twofold higher crude asexual parasite rate (P. vivax and P. falciparum) and an approximately threefold higher P. falciparum rate (both asexual and gametocyte stages) than the intervention district. This indicates that the intervention had an effect, as all other vector control activities were comparable, with even better coverage in the control district (see Section 2.2.2). These findings are consistent with the trends in morbidity observed in the health institutions. Generally, there was higher malaria transmission in the control district, by all indicators, as shown in figure 12 and Table 9. Table 9. Rate of increase in parasite rates: control versus intervention district, 2005 Parasite rate Prevalence survey Low-transmission season (April 2005) High-transmission season (September 2005) Rate of increase (%) crude parasite rate (all parasite stages and species) control 3.4 20.8 511.8 intervention 4.9 7.4 51.0 P. falciparum slide positivity rate control 2.4 14.2 491.7 intervention 4.3 4.9 14.0 0 25 20 15 10 5 Apr. low Sept. high Apr. low Sept. high 2005 Crude parasite rate (all stages & species) P. falciparum parasite rate Percentage Crude asexual parasite rate INTERVENTIONCONTROL 2006 Apr. low Sept. high Apr. low Sept. high 2005 2006 Figure 12 Malaria prevalence surveys (2005–2006) 30 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia The serial surveys of parasite prevalence indicate that malaria transmission was lowered in the intervention district, with a lower crude parasite rate, a lower P. falciparum prevalence rate and a lower P. falciparum gametocyte rate. When the P. falciparum gametocyte rate before the start of the study (April 2005), which was double in the intervention district, was compared with that measured during the high-transmission season of September 2005, a 7% decrease was observed in the intervention district and a 10-fold increase in the control district (Table 9). While there may be variations in transmission in the Raya valley due to local factors, a general downward trend in parasite prevalence was observed during Phase II in both the intervention and control districts, as would be expected during a post-epidemic period. During Phase I, however, a large effect on malaria prevalence was observed only in the intervention district, where community distribution of artemether-lumefantrine reduced the intensity of malaria transmission. This was largely due to the gametocytocidal effect of the drug, which reduces malaria transmission. 4.3.2 Knowledge, attitude and practices Two surveys of knowledge, attitude and practices were undertaken. The responses recorded during both surveys are given in Appendix B, Table B1. The first survey was performed during Phase I of the project (Fig. 13). 50 5025 25075 75100 100 CONTROL Percentage of “yes” INTERVENTION Cure: modern medicine? ITNs prevent malaria? Untreated malaria cause death? Ever used ITNs? A problem in this village? IRS prevent malaria? Preventable? Ever own net(s)? Heard of malaria? Cause of malaria? Mosquito bite? Environmental management helpful? Participated in environmental management this month? Own radio? Cure: traditional healer? Coartem® stated as medicine? Figure 13 Knowledge, attitude and practice with regard to malaria in intervention and control districts, April 2005 Chapter 4. Results and discussion 31 In April 2005, knowledge about malaria was moderate to high, and use of preventive and curative measures was adequate, except for low rates of community participation in environmental management for vector control and a low rate of awareness about the new first-line drug (artemether-lumefantrine). More than 93% of interviewees knew that malaria could be treated with modern medicines; more than 78% knew that malaria is preventable; more than 72% knew that residual insecticide spraying could prevent malaria, and more than 74% believed that there was no traditional cure for malaria. Although people in both areas considered that environmental management could be helpful in disease control, less than 36% of households had a member who had participated in such activities during the preceding month. Six months later, in October 2005, another survey was undertaken (Fig. 14). The main finding was that some preventive and curative aspects were well understood; however, more improvement since the first survey was found in the control district than in the intervention district. For example, the name of the new drug was better known in the control district. There are a number of possible explanations, including: • the severity of the epidemic in the control district, which, coupled with limited access to artemether-lumefantrine, might have resulted in panic and hence lack of intake of the health education provided by community health workers; • the period between the two surveys, which was too short to permit a signifi- cant change; or • the fact that health education was a weaker component of the intervention than actual distribution of the drug. Figure 14 Knowledge, attitude and practice with regard to malaria in intervention and control districts, October 2005 50 5025 25075 75100 100 CONTROL Percentage of “yes” INTERVENTION Untreated malaria cause death? Nets prevent malaria? IRS prevent malaria? Cure: modern medicine? A problem in this village? Can malaria be prevented? Ever used ITNs? Ever own net(s)? Heard of malaria? Cause of malaria? Mosquito? Environmental management helpful? Participated in environmental management this month? Own radio? Coartem® stated as medicine? Cure: traditional healer? 32 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 4.4 Performance of rapid diagnostic tests 4.4.1 Health institution level The results for 387 patients were evaluated between June and July 2005, during the epidemic. As this period is a high-transmission season, 7.6% of patients had received an antimalarial drug during the week before presentation. At the initial diagnosis with the rapid diagnostic test, 38% of tests were positive for P. falciparum. In addition to fever, most patients evaluated with the rapid diagnostic test in the health facilities had headache (80.7%), back and joint pain (58.4%), chills (56.3%), anorexia (53.5%) and sweating (52.4%) (see Appendix B, Table B2). Vomiting was more common in children (56–68.6%) than adults (30.4%). Final readings were available from 384 blood films for comparison with the Paracheck-Pf® readings. These showed an overall slide positivity rate (all species, all forms) of 47.7%, comprising 65.5% P. falciparum and 34.5% P. vivax. The positivity rate was 38% with Paracheck-Pf® and 31% with microscopy (Table 10). Table 10. Results obtained by microscopy and with the rapid diagnostic test at health institution level Diagnostic method Test result Number (%) Microscopy Positive for P. falciparum (± gametocytes) 120 31.3% Positive for P. vivax 63 16.4% negative 201 52.3% total 384 100% rapid diagnostic test Positive 147 38.3% negative 237 61.7% total 384 100% P. vivax cases are not detectable with the rapid diagnostic test and are therefore diagnosed as negative. With microscopy results as the standard and excluding P. vivax infection (which is not detectable with this test), the diagnostic performance of the test in detecting P. falciparum at health institutions was 94.2% sensitivity and 87% specif- icity, with 76.9% positive predictive value and 97.0% negative predictive value. The high negative predictive value ensures that most patients who do not have malaria are not treated unnecessarily (Table 11). Chapter 4. Results and discussion 33 Table 11. Performance of rapid diagnostic tests at health institution level Microscopy result Total Positive for P. falciparum (± gametocytes) negative rapid diagnostic test result Positive 113 34 147 76.9% positive predictive value negative 7 230 237 97.0% negative predictive value total 120 264 384 94.2% sensitivity 87% specificity 4.4.2 Community health worker level The results of rapid diagnostic tests for 507 patients aged from less than 1 to 90 years (mean, 20.2 ± 16) were evaluated during September, October and November 2007, a high-transmission season. At the initial diagnosis with the tests, 44% were positive for P. falciparum and the rest were declared to be negative. Final readings from 507 blood films were available for comparison with the readings. Microscopic examination showed an overall slide positivity rate (all species, all forms) of 40%, comprising 67% P. falciparum and 33% P. vivax. The positivity rate for P. falciparum was 44% with the rapid diagnostic test and 27% by microscopy (Table 12). Table 12. Results obtained by microscopy and with the rapid diagnostic test at community health worker level Diagnostic method Test result Number (%) Microscopy Positive for P. falciparum (± gametocytes) 137 27.0% Positive for P. vivax 68 13.4% negative 302 59.6% total 507 100% rapid diagnostic test Positive 221 43.6% negative 286 56.4% total 507 100% P. vivax cases are not detectable with the rapid diagnostic test and are therefore diagnosed as negative. With microscopy results as the standard and excluding infection with P. vivax, the diagnostic performance of the test in detecting P. falciparum infections by community health workers was 96.4% sensitivity, 76% specificity, with 59.7% positive predictive value and 98.3% negative predictive value (Table 13). 34 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Table 13. Performance of rapid diagnostic test at community health worker level Microscopy result Total Positive for P. falciparum (± gametocytes) negative rapid diagnostic test result Positive 132 89 221 59.7% positive predictive value negative 5 281 286 98.3% negative predictive value total 137 370 507 96.4% sensitivity 76% specificity Ideally, the comparison of the performance of the test at health institutions and by community health workers would have been performed in the same season with the same batch numbers of the test. This was not possible, however, mainly because of the study design, which introduced use of the tests by community health workers only in Phase II. During this Phase, the low malaria transmission resulted in insufficient samples for conducting a study at both levels. With this limitation in mind and in view of the observed similarity in positivity rates at the two levels, the rapid diagnostic test at health institution level was relatively more specific, with a higher positive predictive value. A masked assessment by direct observation was undertaken as part of the survey of rapid diagnostic test performance and adherence at community level, to establish how safely the health workers were processing the tests. It was found that they complied with 72% of the steps in use of the tests and achieved a high level of competence in most of the critical procedures. They required help in recording results on the device, reading results in a timely manner, using a new glove for each patient, observing the colour of the desiccant and registering time correctly (Table 14). The evidence from this pilot study indicates that community health workers are capable of providing treatment based on the rapid diagnostic test correctly and that they can handle these simple kits safely. Generally, the safety and diagnostic performance of the test in the hands of community health workers were not inferior to those at the level of formal health institutions. Therefore, the traditional, community-based treatment service can be implemented throughout the region, with provision of artemisinin-based combination therapy after confirmation of the diagnosis with the rapid test. Chapter 4. Results and discussion 35 Table 14. Direct observation of community health workers’ skill in safe processing of rapid diagnostic tests No. Action Complied with standard (%) 1 records result on device 42.9 2 reads result after 15 min 50.0 3 uses new glove for test on each patient 50.0 4 checks colour of desiccant 64.3 5 registers time of application of last drop of buffer solution 85.7 6 Appropriately transfers blood to test well 92.9 7 registers time on device for reading test result (ending time) 92.9 8 Wears gloves in processing test 92.9 9 reads result correctly 92.9 10 holds buffer bottle perpendicular to cassette 100 11 takes blood immediately before clotting 100 12 reads result in well-lit area 100 13 registers starting time on device for development of result 100 14 no clotting of blood in transferring specimen 100 15 Places test cassette on a level surface 100 16 disposes of used test kits in a safe disposal box 100 17 selects the correct hand and finger for pricking 100 18 uses a sterile lancet for each patient 100 19 Applies the correct number (six) of drops of buffer solution 100 4.5 Adherence of patients to the 6-dose artemether-lumefantrine regimen Surveys of patient adherence in clinics and in the community setting were undertaken by the Alamata Hospital Health Centre in both Phases of the study. The first survey was conducted during May-July 2005 in five health institutions. Male and female patients were studied in three rural health stations, two urban health facilities and 10 randomly selected villages or other localities where health workers were distributing artemether-lumefantrine in the community. Of the 650 patients, 34% were treated by community health workers, 32% at health posts and clinics and 34% at health centres and hospitals. All ages were represented, with 29% under 5 years, 38% aged 5–14 years and 34% aged 15 years or more. For the purpose of this study, three categories of adherence were defined, as described in section 3.2.2 and the footnote to Table 15. Those who were definitely not adherent represented 12% of the sample; 41% were probably not adherent and 36 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 47% were probably adherent. The proportion of patients who were definitely not adherent was highest among those treated by community health workers (19%), with 10% of those treated at health posts and clinics and 9.6% of those treated at health centres and hospitals. Patients under 5 years were less frequently non- adherent (8%), while 13.5% of those aged 5–14 years and 14% of those over 15 years were definitely not adherent. Irrespective of date and time interval, 87% of the patients reported that they took the correct number of tablets each time. Table 15. Main outcomes of first adherence survey Main variables measured Details Community health workers (10) Health posts or clinics (3) Alamata Health Centre and Hospital (2) Sample % of sample Patients enrolled by age group (years) < 5 53 57 76 186 28.6 5–14 93 69 83 245 37.7 ≥ 15 76 84 59 219 33.7 All 222 210 218 650 100.0 Adherencea to treatment Probably adherent 91 (41%) 107 (51%) 107 (49%) 305 46.92 Probably not adherent 88 (40%) 82 (39%) 96 (44%) 266 40.92 definitely not adherent 43 (19%) 21 (10%) 15 (9.6%) 79 12.15 No. of doses missed by consultation site 0 190 175 168 533 82.0 1 17 12 20 49 7.5 2 10 16 9 35 5.4 3 4 5 5 14 2.2 4 1 1 12 14 2.2 5 0 1 4 5 0.8 No. of doses missed by age group no. of doses < 2 years 3–7 years 8–10 years >10 years % of doses missed 0 62 159 70 259 84.0 1 6 16 7 21 7.7 2 4 7 2 14 4.2 3 1 2 1 12 2.5 4 0 2 1 2 0.8 5 0 1 0 1 0.3 Total 73 187 81 309 100 a. Probably adherent: blister packs either missing or empty; patient reports taking all doses on the correct day, at the correct time and in the correct amount, replacing them if any vomiting within 30 min; information available for all doses Probably not adherent: blister packs either missing or empty; patient does not report taking all doses on the correct day, at the cor- rect time and in the correct amount, with no replacement in case of vomiting within 30 min; but information available for all six doses (i.e. there may be doubt that the patient has followed the instructions correctly) Definitely not adherent: patient has leftover tablet(s) in blister pack. Chapter 4. Results and discussion 37 The second adherence survey was conducted with the same method but with a slightly smaller sample and disproportionate distribution of the different scales, mainly because of a lower incidence of malaria, especially at the higher-level institu- tions. In contrast to the first survey, most (54.3%) of the patients assessed were treated by community health workers; 39% were treated at clinics and health posts and 6.6% at hospitals and health centres. The 545 male and female patients were assessed at three rural clinics and health posts, two urban health facilities and 10 randomly selected villages or localities where community health workers distrib- uted artemether-lumefantrine tablets. All ages were represented, with 18.5% under 5 years, 29% aged 5–14 years and 52.5% aged 15 years or more. A total of 87 patients were definitely not adherent. We considered it inappro- priate to compare the rate of non-adherence by type of provider, as most of the cases (54.3%) had been treated by community health workers. In this Phase, however, 13% of those aged under 5 and 5–14 years were probably not adherent and almost 20% were definitely not adherent. Overall, 16% of the patients were definitely not adherent, 46% were probably not adherent and 38% were probably adherent (Table 16). Table 16. Main outcomes of second adherence survey Main variables measured Details Community health workers (10) Health posts and clinics (3) Alamata Health Centre and Hospital (2) Sample % of sample Patients enrolled by age group (year) < 5 32 65 4 101 18.5 5–14 100 52 6 158 29.0 ≥ 15 164 96 26 286 52.5 All 296 213 36 545 100.0 Lost to follow-up After a 3-day attempt 18 6 0 24 4.4 Adherencea to treatment Probably adherent 150 (72%) 53 (25%) 5 (2%) 208 38.2 Probably not adherent 174 (70%) 45(18%) 31 (12%) 250 45.9 definitely not adherent 71 (82%) 16 (18%) 0 (0%) 87 16.0 No. of doses missed by consultation site 0 209 152 28 389 71.4 1 30 29 4 63 11.6 2 30 15 2 47 8.6 3 13 10 2 25 4.6 4 7 4 0 11 2.0 5 7 3 0 10 1.8 No. of doses missed by age group no. of doses < 2 years 3–7 years 8–10 years > 10 years % of doses missed 0 36 85 42 304 85.7% 1 4 11 5 30 9.2% 2 2 4 1 9 2.9% 3 1 0 2 4 1.3% 4 1 1 0 1 0.6% 5 0 0 1 1 0.4% Total 44 101 51 349 100% a. See previous page 38 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia We found that the definitions of adherence were too prescriptive and limiting and did not take into account the practical situation in the field. Many of the patients did not have a watch and therefore had to estimate the time at which the drug was to be taken. Moreover, the enumerators had to calculate the approximate time of intake from natural events or phenomena, e.g. when cows left the shed or at sunrise. This situation led to both recall and enumerator bias, as interpretation of adherence could have been subjective. The definition of “probably not adherent” had too many variables, which might have led to the exclusion of adherent patients. The results must therefore be interpreted with caution, taking into account factors that might influence adherence. For practical purposes, the proportion of patients who completed the 3-day course of artemether-lumefantrine was 84% in Phase I and 86% in Phase II. In Phase II, 95% of patients completed five doses of treatment. Figure 15 shows the outcomes of the two surveys. 0 20 10 30 40 50 Probably adherent Probably not adherent Definitely not adherent Probably adherent Probably not adherent Definitely not adherent Adherence survey 2005 Adherence survey 2007 PercentageFigure 15 Adherence to six-dose regimen of artemether- lumefantrine (all levels) Chapter 4. Results and discussion 39 4.6 Pharmacovigilance Throughout the project, not a single suspected adverse drug reaction was reported spontaneously. The reason is not clear, but multiple factors might have played a role, including the known difficulty of pharmacovigilance in poor settings, the limited training and lack of supervision of health-care workers, and cultural factors that affect people’s perception of illness and of the beneficial effects of medicine. During the second year of the project, the introduction of the active pharma- covigilance study coincided with a period of extremely low malaria transmission. This resulted in very low patient enrolment (108 patients). When the study was extended throughout the malaria transmission season, there was no additional enrolment, and, as a result, it was decided to terminate the study after 2 months. The results of the pharmacovigilance study, before termination, are shown in Table 17. No meaningful conclusions could be drawn owing to lack of causality assessment, the absence of a comparator and the limited number of patients enrolled and followed-up before its discontinuation. Most of the signs and symptoms recalled may be attributable to malaria itself. Table 17. Adverse events reported during the active pharmacovigilance study Adverse event No. of patients % of patients Any adverse event 20 18.5 gastrointestinal symptoms (nausea, vomiting, abdominal cramps) 8 7.4 Fever 3 2.8 headache 3 2.8 sweating 3 2.8 loss of consciousness 2 1.8 swelling of lower limbs 1 0.9 itching 1 0.9 none 88 81.5 20 patients reported adverse events; one patient reported two adverse events 40 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 4.7 Health impact of use of rapid diagnostic tests and artemether-lumefantrine 4.7.1 Morbidity trends at clinics and health posts In the control district, 125 011 patients were treated during the study period (April 2005–June 2007) at rural health facilities. Of these, 51% were treated for malaria on the basis of rapid diagnostic test results or clinical diagnosis as per the national treatment policy. In the intervention district, 34 825 patients were examined at clinics and health posts, of whom 38% had malaria confirmed by rapid diagnostic tests or clinical diagnosis. Figure 16 shows the numbers of outpatients and malaria patients at health facilities in each district. Although the initial conditions in the two districts appeared similar, there was a clear increase in the number of malaria cases in both districts during June 2005 due to the epidemic; however, the patient load in the control district was approximately four- to fivefold higher than in the intervention district. One of the most likely explanations for this difference in patient load is the presence of community health worker treatment services in the intervention district, serving approximately 58% of all malaria patients near their homes during the project (Fig. 17), offering indirect evidence of the availability of prompt diagnosis and treatment in the intervention area. INTERVENTIONCONTROL 5 0 10 15 20 25 30 100 90 80 70 60 50 40 30 20 10 0 MA M JJA SON DJ F MA M JJA SON DJ F MA M JJA 2005 2006 2007 MA M JJA SON DJ F MA M JJA SON DJ F MA M JJA 2005 2006 2007 Number of outpatients (thousands) Malaria outpatients (percentage) Malaria outpatients (percentage) Malaria outpatients Total outpatients (clinics and health posts) Figure 16 Quarterly numbers of outpatients at clinics and health posts and percentages with malaria jjA, june, july, August; son, september, october, november; djF, december, january, February; MAM, March, April, May Chapter 4. Results and discussion 41 4.7.2 Morbidity trends at health centres and the hospital The results should be analysed and interpreted cautiously, as the situation in the control and intervention districts with regard to high-level health facilities differed. The intervention district has a district hospital (the Alamata Hospital and Health Centre), while the control district shares a zonal hospital with another district 17 km away. During the study period (April 2005–June 2007), out of the 72 037 outpatients in the control district health centre, 51% were diagnosed with malaria (clinical or confirmed) and were treated as per the national policy. The outpatient load in the intervention district during the second quarter (June, July and August 2005) was as high as that in the control district, but the proportionate morbidity due to malaria remained low. Of all 121 676 outpatient consultations at the Alamata Hospital and Health Centre, 41 708 (34%) had a diagnosis of malaria (clinical or confirmed). Figure 18 shows the proportions of malaria outpatients at higher health institutions in the intervention and control districts. Figure 18 Quarterly numbers of outpatients at health centres and hospital and percentages with malaria Figure 17. Proportions of malaria patients by treatment site in the intervention district Health centres and hospital (32.3%) Clinics and health posts (10.1%) Community health workers (57.6%) INTERVENTIONCONTROL 4 0 8 12 16 20 100 90 80 70 60 50 40 30 20 10 0 MA M JJA SON DJ F MA M JJA SON DJ F MA M JJA 2005 2006 2007 MA M JJA SON DJ F MA M JJA SON DJ F MA M JJA 2005 2006 2007 Number of outpatients (thousands) Malaria outpatients (percentage) Malaria outpatients (percentage) Malaria outpatients Total outpatients (health centres and hospital) jjA, june, july, August; son, september, october, november; djF, december, january, February; MAM, March, April, May 42 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia It should be noted that the data for the intervention district represent outpatients at both health centres and the hospital, while those for the control district represent outpatients only at health centres. Despite the higher total number of cases in the intervention district, the proportion of malaria outpatients was lower than in the control district. Confirmed malaria cases and proportions of P. falciparum and P. vivax infections in the two districts The proportions of P. falciparum and P. vivax malaria cases changed consider- ably during the project. Although a reduction in the proportion of P. falciparum cases was seen in most parts of the region, the scale of the reduction was different in the two districts. For instance, the ratio of P. falciparum to P. vivax in June, July and August 2005 (epidemic season) was 91%:9% in the control district and 72%:28% in the intervention district, which changed to 2%:98% and 11%:89% during the same period in 2006. The total proportion of cases was twice as high in the control district as in the intervention district, but cases of P. vivax were half as frequent. Throughout the project, the proportion of cases that tested positive for P. falciparum on microscopy was lower in the intervention district (48%, 4778) than in the control district (64%, 6778). These variations could be due to the wide use of artemether- lumefantrine in the 2005 epidemic at community level in the intervention district, with an effect on gametocyte rate (see Table 8) affecting falciparum transmission. Figure 19 shows the seasonal trend in the proportions of microscopy- confirmed cases of P. falciparum and P. vivax malaria, in heath facilities in which a microscope was available. INTERVENTIONCONTROL 4 6 May June 2 0 8 12 10 14 16 4.5 4 3.5 2.5 3 2 1.5 1 0.5 0 MA MJJA SON DJ F MA MJJA SON DJ F May June MA MJJA SON DJ F MA MJJA SON DJ F 2005 2006 2007 2005 2006 2007 P. falciparum and P. vivax cases (thousands)Total and confirmed malaria cases (thousands) Confirmed malaria casesTotal malaria cases P. vivaxP. falciparum Figure 19 Trends of P. falciparum and P. vivax cases in hospital and health centres (Phases I and II) jjA, june, july, August; son, september, october, november; djF, december, january, February; MAM, March, April, May Chapter 4. Results and discussion 43 Inpatient morbidity and mortality During the same study period (April 2005–June 2007), 4371 inpatients were seen in the health centre of the control district, of which 79% were found to have malaria. The inpatient load at the Alamata Hospital and Health Centre in the intervention district during this period was higher (6943), but the proportion found to have malaria (42%, 2930) was lower than in the control district (Fig. 20). Again, in contrast to the data from the intervention district, those for the control district were only from the health centre; therefore, different scales are being compared. Both the number and the proportion of malaria admissions in the intervention district were lower than in the control district. Data are available from the zonal hospital, but records with the patient’s address remain to be examined. INTERVENTION CONTROL 3 2.5 2 1.5 1 0.5 0 100 90 80 70 60 50 40 30 20 10 0 MA M JJA SON DJ F MA M JJA SON DJ F MA M 2005 2006 2007 MA M JJA JJASON DJ F MA M JJA SON DJ F MA M 2005 2006 2007 Number of inpatients (thousands) Malaria inpatients (percentage) Malaria inpatients (percentage)Malaria inpatients Total inpatients (health centres and hospital) JJA Although previous surveys in Tigray indicate that most deaths (> 90%) occur at home,24 inpatient deaths were compared for the sake of completeness. During the same study period (April 2005–June 2007), 38 deaths occurred among inpatients at the health centre in the control district, of which 90% were attributed to malaria. The inpatient facilities at the Alamata Hospital and Health Centre in the interven- tion district had a similar number of deaths (35), of which only 34% were attributed to malaria – approximately 2.6-fold lower than in the control district. The epidemic in June, July and August 2005 claimed many lives, especially in the control district (Fig. 21). Figure 20 Quarterly numbers of inpatients at health centres and hospital and percentages with malaria 24. Ghebreyesus TA et al. Community-based malaria control in Tigray, Northern Ethiopia. Parasitologia, 1999; 41:367–371. jjA, june, july, August; son, september, october, november; djF, december, january, February; MAM, March, April, May 44 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia INTERVENTIONCONTROL 0 20 16 12 8 4 MA M JJA SON DJ F MA M JJA SON DJ F MA M JJA 2005 2006 2007 2005 2006 2007 Number of inpatients deaths MA M JJA SON DJ F MA M JJA SON DJ F MA M JJA Deaths attributed to malariaTotal deaths 4.8 Mortality surveys The entire population over the age of 1 year in the study areas of Alamata and Raya Azebo was surveyed twice for mortality by verbal autopsy, in 2006 and 2007 (Fig. 22). The average study population was 202 670, with 192 713 in 2006 and 223 427 in 2007. The basic characteristics of the populations in the two districts in the two survey periods are shown in Table 18. The age distributions of the populations in the two districts were similar (Fig. 23). Figure 21 Quarterly numbers of deaths among inpatients at health centres and hospital and percentages with malaria Figure 22 Surveyed households and health facilities in Raya Azebo (control district) and Alamata (intervention district) Chapter 4. Results and discussion 45 Table 18. Background characteristics of the populations (> 1 year) surveyed for deaths from malaria Characteristic Alamata Raya Azebo 2005–2006 no. (%) 2006–2007 no. (%) 2005–2006 no. (%) 2006–2007 no. (%) total population 84 184 (100) 94 570 (100) 108 529 (100) 128 857 (100) Age (years) and gender 1–4 8 056 (9.6) 10 152 (10.7) 12 012 (11.1) 16 296 (12.6) 5–14 26 045 (30.9) 27 374 (28.9) 35 030 (32.3) 40 547 (31.5) 15–49 male 18 294 (21.7) 21 674 (22.9) 23 717 (21.9) 27 688 (21.5) 15–49 female 21 255 (25.2) 23 788 (25.2) 26 313 (24.2) 30 970 (24.0) 50–64 6 694 (8.0) 7 400 (7.8) 7 540 (6.9) 8 894 (6.9) ≥ 65 3 840 (4.6) 4 182 (4.4) 3 917 (3.6) 4 462 (3.5) Area urban 30 823 (36.6) 34 292 (36.3) 28 380 (26.1) 32 218 (25.0) rural 53 361 (63.4) 60 278 (63.7) 80 149 (73.9) 96 639 (75.0) owns a bed net no 28 767 (34.2) 15 239 (16.1) 28 761 (26.5) 13 753 (10.7) yes 55 417 (65.8) 79 331 (83.9) 79 768 (73.5) 115 104 (89.3) Altitude (m) < 1600 65 131 (77.4) 73 063 (77.3) 19 342 (17.8) 23 214 (18.0) ≥ 1600 19 053 (22.6) 21 507 (22.7) 89 187 (82.2) 105 643 (82.0) distance to health facility (km) < 2 33 964 (40.3) 37 491 (39.6) 34 583 (31.9) 43 345 (33.6) 2–5 35 721 (42.4) 41 456 (43.8) 46 477 (42.8) 55 610 (43.2) ≥ 5 14 499 (17.2) 15 623 (16.5) 27 469 (25.3) 29 902 (23.2) Age group (years) Population (percentage) Raya Azebo Alamata 20 15 10 5 0 5 10 15 20 1–4 10–14 20–24 30–34 40–44 50–54 60–64 70–74 80–84 90+ Figure 23 Age distributions of populations in the two study districts 46 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Among the people surveyed, 1179 deaths were reported in the year before the 2006 interview and 918 were reported in the year before the 2007 interview, representing overall crude mortality rates of 6.1 and 4.1 per 1000 in 2006 and 2007, respectively. Of the deaths in the year before the 2006 survey, 659 out of a popula- tion of 108 529 were in the Raya Azebo district (crude mortality rate, 6.1 per 1000) and 520 out of 84 184 were in the Alamata District (crude mortality rate, 6.2 per 1000) (Table 19). Of the deaths in the year before the second survey in 2007, 447 out of 128 857 were in Raya Azebo (crude mortality rate, 3.5 per 1000) and 471 out of 94 570 were in Alamata (crude mortality rate, 5.0 per 1000). Table 19. Mortality from all causes by district, year and background factors Characteristic Alamata Raya Azebo 2005–2006 deaths (per 1000) 2006–2007 deaths (per 1000) 2005–2006 deaths (per 1000) 2006–2007 deaths (per 1000) Population > 1 year of age 520 (6.2) 471 (5.0) 659 (6.1) 447 (3.5) Age (years) and gender 1–4 46 (5.7) 40 (3.9) 61 (5.1) 39 (2.4) 5–14 37 (1.4) 31 (1.1) 68 (1.9) 18 (0.4) 15–49 male 110 (6.0) 110 (5.1) 191 (8.1) 103 (3.7) 15–49 female 158 (7.4) 130 (5.5) 174 (6.6) 116 (3.7) 50–64 62 (9.3) 59 (8.0) 66 (8.8) 61 (6.9) ≥ 65 107 (27.9) 101 (24.2) 99 (25.3) 110 (24.7) Area urban 237 (7.7) 220 (6.4) 173 (6.1) 118 (3.7) rural 283 (5.3) 251 (4.2) 486 (6.1) 329 (3.4) owns a bed net no 216 (7.5) 81 (5.3) 210 (7.3) 57 (4.1) yes 304 (5.5) 390 (4.9) 449 (5.6) 390 (3.4) Altitude (m) < 1600 429 (6.6) 350 (4.8) 129 (6.7) 73 (3.1) ≥ 1600 91 (4.8) 121 (5.6) 530 (5.9) 374 (3.5) distance to health facility (km) < 2 239 (7.0) 223 (5.9) 230 (6.7) 161 (3.7) 2–5 212 (5.9) 194 (4.7) 259 (5.6) 187 (3.4) ≥ 5 69 (4.8) 54 (3.5) 170 (6.2) 99 (3.3) Verbal autopsy interviews yielded sufficient information to determine the cause of death for 2003 deaths (95.5%). The data from the interviews were interpreted with the InterVA model25,26 in order to determine up to three likely causes of death in each case. The cause-specific mortality fractions, weighted by the likelihood of specific causes from the InterVA model, are shown by district and year in Table 20. 25. Fantahun M et al. Assessing a new approach to verbal autopsy interpretation in a rural Ethiopian community: the InterVA model. Bulletin of the World Health Organization, 2006; 84:204–210. 26. Umeå Centre for Global Health Research (www.interva.net). Accessed 20 January 2009. Chapter 4. Results and discussion 47 Table 20. Cause-specific mortality fractions by district and year Cause of death Alamata 2005–2006 (%) 2006–2007 (%) Raya Azebo 2005–2006 (%) 2006–2007 (%) Accidental drowning 0.54 0.24 2.32 2.00 Accidental poisoning 0.14 0.28 0.17 0.30 Acute cardiac accident 0.20 0.71 0.09 0.48 Acute respiratory disease 0.29 0.27 0.14 0.18 bloody diarrhoea 0.83 0.94 1.43 0.10 chronic cardiac disease 0.58 1.10 1.06 1.25 chronic respiratory disease 0.71 0.00 0.51 1.03 congenital malformation 0.12 0.39 0.07 0.04 diabetes 8.49 9.45 5.63 8.71 disease of nervous system 0.11 0.11 0.00 0.13 hiV/Aids-related 20.99 21.61 24.64 22.30 haemoglobinopathy 0.04 0.21 0.27 0.00 homicide 0.24 2.26 1.45 0.25 undetermined 5.24 5.13 4.29 3.28 Kidney or urinary disease 6.37 4.79 2.41 3.10 liver disease 11.28 12.76 11.01 12.76 Malaria 2.65 2.28 6.11 2.78 Malignancy 3.35 4.14 3.13 3.37 Malnutrition 0.30 0.13 0.03 0.03 Maternity-related 2.58 1.98 1.06 0.95 Measles 0.10 0.23 0.00 0.00 Meningitis 3.49 4.29 5.33 2.66 non-bloody diarrhoea 0.24 0.34 0.34 0.55 other chronic infection 0.03 0.11 0.08 0.27 other fatal accident 0.29 0.08 0.15 0.19 Pneumonia/sepsis 2.47 2.65 4.34 4.61 stroke 1.61 1.64 1.34 2.31 suicide 0.15 0.65 0.78 0.27 tetanus 0.17 0.42 0.57 0.19 transport-related accident 2.96 2.32 2.41 6.22 tuberculosis (pulmonary) 23.43 18.52 18.83 19.70 When the survey data for 2005–2007 are combined, there were 991 deaths in Alamata – average population of 89 337 (crude mortality rate, 11.09 per 1000), and 1106 deaths in Raya Azebo – average population of 118 693 (crude mortality rate, 9.32 per 1000). In Alamata, 24 of the deaths were determined to be probable cases of malaria (malaria mortality: 0.27 per 1000), with 53 probable malaria deaths in the control district of Raya Azebo (malaria mortality rate: 0.45 per 1000) (Table 21). Table 21. Mortality from all causes and from malaria between May 2005 and April 2007 in the study districts, on the basis of repeated mortality surveys and InterVA Alamata Raya Azebo Total Average population 89 377 118 693 208 070 Number of deaths 991 1 106 2 097 Deaths probably due to malaria 24 53 77 Mortality due to malaria / 1000 0.27 0.45 0.37 48 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Poisson multivariate regression models were constructed to determine the incidence rate ratios for mortality from all causes and from malaria in the data for 2006 and 2007 combined. The variables included age and gender group, altitude, urban or rural residence, use of bed nets and distance to the nearest health facility. Possible clustering effects at the village (kushet) level were built into the models. The results are shown in Table 22. Table 22. Poisson regression models of incidence rate ratios for mortality from all causes and from malaria, allowing for clustering effects at village level All causes Malaria Adjusted irr 95% ci p Adjusted irr 95% ci p district Alamata 1.03 0.87–1.21 0.751 0.60 0.40–0.90 0.013 raya Azebo reference reference Age (years) and gender 1–4 3.37 2.66–4.25 0.000 4.35 2.35–8.05 0.000 5–14 reference reference 15–49 male 4.63 3.87–5.54 0.000 1.48 0.74–2.94 0.267 15–49 female 4.64 3.79–5.69 0.000 3.00 1.63–5.51 0.000 50–64 6.72 5.39–8.38 0.000 2.24 0.97–5.20 0.06 ≥ 65 20.67 16.90–25.28 0.000 2.57 0.79–8.37 0.118 Area urban 1.11 0.97–1.28 0.119 1.14 0.76–1.71 0.521 rural reference reference owns a bed net no reference reference yes 0.83 0.73–0.94 0.004 0.54 0.35–0.84 0.007 Altitude (m) < 1600 reference reference ≥ 1600 0.93 0.79–1.08 0.342 0.94 0.63–1.38 0.736 distance to health facility (km) < 2 reference reference 2–5 0.93 0.82–1.05 0.237 1.42 0.89–2.27 0.142 ≥ 5 0.91 0.77–1.08 0.279 1.28 0.72–2.26 0.396 irr, incidence rate ratio; ci, confidence interval The main finding in terms of differences in malaria control between the two districts was that the adjusted malaria-specific mortality rate in the control district of Alamata, where artemether-lumefantrine was in use at community and health facility level and where (since 2006) half of all the health workers were trained to use the rapid diagnostic test, was 0.6 that of the rate in Raya Azebo district, where artemether-lumefantrine was available only at health facilities. This approximate 40% reduction in the risk for death from malaria in the intervention district was achieved even though there was no appreciable difference in mortality from all causes between the two districts, suggesting that the difference observed is quite specific. This is a robust, statistically signif- icant finding, in which account was taken of age, gender, distance to a health facility, urban or rural area, altitude and bed-net ownership and allows for possible clustering of mortality at the village level. The design of the survey does not, however, permit a strict causal association to be made. Chapter 4. Results and discussion 49 Use of the InterVA model to interpret verbal autopsy data for this kind of assessment of cause-specific mortality fraction is relatively new. It could be argued that it is a less subtle and nuanced approach than that offered by well-trained physicians. It is, however, much easier and quicker and has the considerable advantages of complete consistency and repeatability. Thus, despite its potential shortcomings, the model could not have treated the verbal autopsy material from the two districts differently. Furthermore, the malaria-specific mortality rates shown in Table 21 and the age- and gender-specific incidence rate ratios in Table 22 indicate that the deaths attributed to malaria in the model follow the variations by age and gender that might reasonably be expected. Relatively high rates of HIV-related deaths were found. This should be seen in the context of a population seropositivity rate of around 5% and, in this area at the time of survey, almost no access to antiretroviral treatment. Thus, the survey probably reflects a period of considerable accumulation of HIV-positive adults and their children, who were progressing to AIDS and death. 4.9 Challenges to assessing impact Evaluation of the impact of community distribution of artemether-lumefan- trine after rapid diagnostic testing was limited by a number of factors. • Initiation of the project coincided with a major epidemic of malaria in the Raya valley. The study design and procedures were affected by the emergency situation, in which the health system had to respond in order to mitigate the impact of the epidemic. • Although the control and intervention districts were homogeneous in many respects, the health infrastructure differed, with no district hospital in the control area. • Balancing concurrent interventions (e.g. indoor residual spraying and distri- bution of long-lasting insecticidal nets) in the control and intervention districts was not easy, especially in the epidemic situation. • One major assumption made was that the microclimate of the two districts was more or less the same during the study period. 50 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 4.10 Satisfaction of beneficiaries of the project Almost all the men and women consulted agreed that the problem of malaria was noticeably worse before distribution of artemether-lumefantrine at community level. One male participant said, “For farmers, artemether-lumefantrine is the drug of choice to treat malaria, as we feel at ease as soon as we take it.” One woman commented, “After the rainy season, it was normal for people to remain in bed with malaria at home, but now it is completely reversed and malaria has ceased to be a public health problem.” All those involved in the discussions were pleased that the diagnosis and treatment of malaria could be provided near their homes, so that transport expenses, time spent seeking treatment and disease severity had diminished dramatically. The participants, noting that the community health workers are subsistence farmers, welcomed their commitment, as they provided diagnosis and treatment not only at morning and evening sessions but whenever malaria cases presented, thus sacrificing their own work. The community health workers were seen to provide the service with respect and courtesy. One woman said, “Health workers in Government health facilities don’t have courtesy and respect and they neglect us, so we are not encour- aged to visit them.” One man said “It feels as if they are giving care to their relatives and undertake the work with passion.” They used their own water and lamps to treat cases in the evening, even though these supplies were scarce and expensive. The group found that community health workers provided timely diagnosis and treatment, by checking clinical symptoms against the results of the rapid diagnostic tests, whereas there were long queues at health facilities and sometimes no treatment was given even queuing for a whole day. One man said, “Although these people are poor farmers, they give sound judgement to us and our families.” One man said, “I went for malaria treatment to the community health worker and after he checked me with the rapid diagnostic test he diagnosed that I had P. vivax which wasn’t identified with the test. He gave me chloroquine and I become more vigorous. I was very surprised how knowledgeable our community health workers are.” Another man said, “I took my child to the community health worker, who gave him a referral paper to the higher health facility since his disease was not malaria. Finally, my son was diagnosed with TB in the Alamata Hospital. So we trust them when they tell us we are negative for malaria.” A few respondents said that they were disappointed when community health workers told them they did not have malaria, because they did not have enough money for transport and treatments prescribed by health facilities. One woman said, “One day I was sick and asked the community health worker to provide me with artemether-lumefantrine, but he rejected my request and told me to be examined for the presence of the disease.” Chapter 4. Results and discussion 51 Artemether-lumefantrine is highly effective, even after the first dose, and the group welcomed its introduction. One woman said, “I was infected with malaria and took the first regimen at 8:00 am. I got relief after about 4 hours and was able to perform my household duties.” The group was aware that artemether-lumefan- trine is harmless unless taken without an indication or used inappropriately. All those questioned agreed that, without the project, they would be unable to obtain prompt diagnosis and treatment of malaria and might develop complica- tions. Because of the project, they had saved time, money and, indeed, life. They said that community health workers should be given further training, to increase their skills and capacity. “When they get training, we get better services from them.” As they were sometimes absent from their homes, they were expected to teach family members to provide the service. Almost all the participants said that community health workers should have a monthly salary like other health workers. One woman said, “They should be supported by the Government in maintaining their household income, and any agricultural activities should also be carried out by the community living in the same area.” The community health workers were, however, willing to continue without payment (see section 4.1.1) The participants agreed that it would be disastrous if the service ceased. They said that the Government should maintain widespread distribution of insecticide- treated nets and distribute artemether-lumefantrine through community health workers. Most of those questioned also agreed with use of indoor residual house spraying, and one man suggested larviciding at community level. The participants agreed to conduct environmental activities, such as filling, clearing and draining. They mentioned appropriate adherence to prescribed drugs, particularly artemeth- er-lumefantrine. Overall, therefore, the participants said that: • they were very satisfied with the project for distribution of artemether-lume- fantrine by community health workers and considered that it should be sustained; • the project had markedly decreased the burden of malaria and should be extended to involve mothers treating malaria at home; and • the project should be extended to nearby districts as such promising results had been obtained. 52 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 4.11 Cost of diagnosis clinically and with the rapid diagnostic test at community level Of the 5122 patients examined and treated by community health workers on the basis of the results of Paracheck-Pf® during Phase II, 527 (10.3%) were infected with P. falciparum. Treatment of these 527 patients with artemether-lumefantrine and of all patients testing negative for P. falciparum with chloroquine resulted in an expenditure of US$ 0.93 per patient examined. In contrast, the cost per patient if treated on the basis of clinical diagnosis alone by community health workers would have been US$ 2.33. Therefore, the potential saving in unnecessary antimalarial treatment is US$ 1.41 per patient examined (on the basis of the public sector price of Coartem® [20 mg artemether-120 mg lumefantrine tablets, Novartis Pharma AG] in 2001–2006) (Table 23 and Table 24). Table 23. Costs of treatment by community health workers after clinical versus rapid diagnostic test diagnosis (all costs in US$) Cost of treatment on the basis of rapid diagnostic tests (Paracheck-Pf®) Cost if same patients were treated without rapid diagnostic test Age (years) no. examined P. falciparum positives cost of glove cost of Paracheck- Pf® cost of artemether- lumefantrine cost of quality control for Paracheck-Pf® negatives total cost Artemether- lumefantrine > 10 3611 369 231.1 1641.0 974.2 226.9 3442.2 9 533.0 8–10 479 61 30.7 217.7 127.5 29.3 466.2 1 000.1 3–7 715 81 45.8 324.9 124.7 44.4 620.8 1 101.1 < 2 317 16 20.3 144.1 6.2 21.1 207.7 313.8 All 5122 527 327.8 2327.7 1232.6 321.7 4736.9 11 949.1 Table 24. Cost of community health workers’ treatment services on the basis of clinical diagnosis (all costs in US$) Age group (years) No. clinically treated Cost of glove Cost of Paracheck-Pf® Cost of artemether- lumefantrine Cost of quality control for Paracheck-Pf® negatives > 10 7601 0 0 20 066.64 0 8–10 881 0 0 1 841.29 0 3–7 1379 0 0 2 123.66 0 < 2 614 0 0 239.46 0 All 10475 0 0 24 271.05 0 Use of better rapid diagnostic tests (at comparable cost), which would allow identification of both P. falciparum and P. vivax, could maximize savings, obviate unnecessary treatment of uninfected patients and improve prompt referral of patients with other diseases. Chapter 5. Lessons learnt 53 Chapter 5. Lessons learnt 5.1 Study design The ideal study design would have been a randomized, controlled community trial of use of artemisinin-based combination therapies and rapid diagnostic tests at community level. In order to obtain a sufficient number of units of randomiza- tion, however, the study units would have to have been relatively small geograph- ical areas (such as villages). For ethical reasons and to avoid cross-contamination between the control and intervention groups, we decided to conduct the study as an observational study at district level. 5.2 Project personnel The project would ideally have been conducted by a minimum of two full-time, dedicated professionals, for project management and coordination. In fact, no single full-time expert was involved. The assumption that people will meet their profes- sional and moral obligations is not always true, and an incentive package might have been more productive. The fast turnover of responsible staff, due mainly to training, was not antici- pated and hence was not considered in the planning Phase. Management of this type of project requires a full-time senior staff member who is responsible for scientific and operational aspects. 5.3 Database management The use of different calendars (the Gregorian calendar for the community health worker database and the Ethiopian calendar for the health institution data) complicated database management. A single part-time database manager was insufficient for this large, compli- cated project. 5.4 Logistics In the initial plan for the mortality survey, mapping houses to assess risk factors was not included, and lack of global positioning system units compromised such mapping. 54 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia A shortage of artemether-lumefantrine tablets of the planned blister types for all age groups created some inconvenience, but a mechanism was found to correct this deficiency. Nevertheless, the dispensing of the drug as equivalent tablets might have affected the compliance of patients with the correct regimen. During community prevalence and mortality surveys, supervisors sometimes had difficulty in locating the surveyor. A mechanism for tracking surveyors and arrangements for meeting supervisors frequently should have been planned before the survey. 5.5 Project management Establishment of a technical or administrative group that met monthly in teleconferences was instrumental for establishing common ground, accommodating the interests of all partners and ensuring thorough follow-up of the project. Throughout the project, all the partners were kept up to date. 5.6 Capacity-building In order for a project to be successful, the capacity of the personnel involved must be built, both as an incentive for study team members and to allow continuity of the project to completion. 5.7 Effect of low incidence of malaria in Phase II of the project As researchers and operation managers, we had mixed feelings about the decrease in malaria incidence during the second Phase of the project. Most of the studies that depended on a certain number of malaria patients were made possible by extending the number of survey days or maximizing the surveyors. For example, the adherence study during the second Phase took several months to complete. Despite our hypothesis that the intervention would reduce the incidence of malaria, the implications of a low malaria incidence on the sample size requirements of studies carried out in the second year of the project (second mortality survey, rapid diagnostic test performance survey, active pharmacovigilance study) were not taken into consideration in the planning and design of the project. Chapter 6. Project budget 55 Chapter 6. Project budget 6.1 Project budget input and expenditure The project was supported by the volunteer work of community members and by funds from the Regional Government, Novartis Farma SPA (Italy) and the Italian Ministry of Health, through WHO. The Regional Government budget included depart- mental staff salaries, purchase of insecticide, operational costs for insecticide spraying and other recurrent costs, including drugs used by community health workers other than artemether-lumefantrine. A total of US$ 569 900 was used to meet the costs of supplies, local infrastructure, international capacity-building and procurement. Details of the budget and supply are presented in Table 25, and operational costs are reported in Table 26. The associated project timetable is given in Appendix C. 56 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 6.2 Financing and supplies Table 25. Commodities with their prices and funds transferred by Novartis and WHO to the Tigray Health Bureau Channel Total external contributions to the project (in US$) Main category Specific items Year 1 Year 2 Novartis commodities coartem® (Artemether- lumefantrine – 20 mg/120 mg tablets – novartis Pharma Ag) 133 501.66 99 562.50 Paracheck-Pf® kits 29 602.50 27 372.84 local operating costs1 see details of activities in table 26 85 061.38 0 capacity-building training abroad 0 14 000.00 technical support scientific writer2 0 9 050.00 dissemination workshop 0 20 124.14 subtotal 1. contribution from novartis 248 165.54 170 109.48 WHO total contribution from the italian government for the project commodities 1 toyota hi lux 4x4 double cabine 18 429.90 0 4 suzuki 125-cc motorcycles 9 374.40 0 4 olympus microscopes 6 388.45 0 1 canon photocopier and 2 notebooks 7 357.45 0 20 gPs equipment 3 455.40 0 capacity-building short training course in epidemiology (umeå, sweden) 1 357.40 0 local operating costs3 see details of activities in table 26 0 61 187.00 Who programme support costs (13%) 0 14 013.00 subtotal 2. contribution from the italian government 46 363.00 75 200.00 Who contribution for pharmacovigilance3 technical support for pharmacovigilance Agreement for performance of work 3 895.00 3 895.00 Who temporary adviser 2 273.35 0 technical support for mortality survey technical service agreement 10 000.00 10 000.00 subtotal 3. contribution from Who 16 168.35 13 895.00 total 310 696.89 259 204.48 grand total 569 901.37 1. direct transfer to the tigray health bureau 2. Additional input from novartis 3. transfer of funds through Who representative in Addis Ababa Chapter 6. Project budget 57 6.3 Operational costs Table 26. Project operational funds transferred to the Tigray Health Bureau Expenditure Novartis directly to Tigray Health Bureau (amount in US$) WHO via Representative to Tigray Health Bureau (amount in US$) Total transferred to Tigray Health Bureau Total (US$) First installment 30 000.00 61 187.00 91 187.00 second installment 55 061.38 – 55 061.38 third installment 20 124.14 – 20 124.14 total 105 185.52 61 187.00 166 372.52 Exchange rate, Ethiopian birr in US$ 8.64 8.77 – Total Ethiopian birr 908 510.81 536 500.00 1 445 010.81 1 express transit service 2 004.13 0 2 004.13 2 stationery1 5 334.43 2 170.99 7 505.42 3 Per diem2 40 117.54 37 774.25 77 891.78 4 Fuel and lubricants3 10 816.44 7 186.52 18 002.96 5 contract salaries4 4 781.86 845.10 5 626.96 6 Vehicle and motor maintenance 2 430.33 1 695.88 4 126.21 7 Printing5 6 829.57 – 6 829.57 8 refreshment6 1 456.39 990.52 2 446.91 9 Medicine7 4 136.32 3 421.45 7 557.78 10 computer maintenance8 243.13 – 243.13 11 transport costs 854.95 4.22 859.17 12 telephone 1 917.64 122.43 2 040.07 13 Warehouse rent 338.57 27.37 365.94 14 labour cost 156.36 26.23 182.59 15 community health worker work table 1 338.39 – 1 338.39 16 Mobile telephone 130.25 – 130.25 17 Fax machine – 872.47 872.47 18 other (miscellaneous) 529.90 124.65 654.55 19 total 83 416.21 55 262.09 138 678.30 20 Amount remaining in tigray health bureau 1 645.17 5 924.91 7 570.08 21 dissemination of results, workshop, printing in local language 20 124.14 22 total of 19, 20 and 21 166 372.52 these expenditures were used for the activities listed in Appendix c (project timetable). 1. duplicating paper, duplicating ink, photocopies, pens, pencils, stencils, bags, memory sticks, cds, etc. 2. For community health workers, health workers, supervisors (regional and district), drivers and enumerators during training and data collection and laboratory technicians 3. Vehicles and motorcycles 4. data entry technician, health workers and others 5. guidelines, posters, reporting pads, job aids, etc 6. coffee and tea during training and meetings 7. chloroquine, slides, gloves and reagents 8. Antivirus and software 58 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Chapter 7. Monitoring and evaluation Regional and district focal persons, district malaria unit staff and seasonally mobilized and trained health staff undertook regular direct and indirect supervi- sion of community health workers and health institutions, including surveys. Supervision also included technical assistance in use of the rapid diagnostic test for P. falciparum, checking dosing schedules and treatment and consultation with patients. Regional and district supervisors were given terms of reference and a checklist for all the surveys conducted. Supervisors monitored the surveys continu- ally and provided immediate feedback. In line with the recommendations of the last joint supervision by WHO and the Tigray Health Bureau, supportive supervision was given in the project area during December 2005 in all health facilities, in order to reinstitute stock manage- ment mechanisms, such as bin and stock control cards. In addition, a short orienta- tion was given by the project focal person and the district pharmacy expert during a visit to the pharmaceutical store of each institution. A total of 33 community health workers and six of eight health institutions were supervised (two institutions in highland subdistricts were excluded). Most technical problems were resolved during the supervision. Besides regular regional and district supervision, partners from the Department of Preventive Medicine in Migration, Tourism and Tropical Dermatology of the San Gallicano Hospital in Rome (Dr Luigi Toma) and WHO collaborators (Dr Wilson Were, Mr Ambachew Medhin, Dr Andrea Bosman and Dr Josephine Namboze) made supporting visits to the project. A joint mission by representatives from the Tigray Health Bureau, the Italian Ministry of Health, the San Gallicano Dermatological Institute, Novartis Farma SPA (Italy), the WHO Office in Ethiopia and the WHO Regional Office for Africa visited the area for a 5-day mid-term evaluation, in May 2006. A visit by journalists was organized by Novartis Farma SPA (Italy). The main recommendations after the mid-term evaluation were that WHO should: • consider hiring an epidemiologist to work with the team to refine data collec- tion and analysis for use in policy-making; • assist in ensuring proper pharmacovigilance, based not only on dermatolog- ical manifestations, as other important reactions may be missed; and • work with the Tigray Health Bureau to design proper field monitoring of rapid diagnostic tests that would be used for improving programmes. Chapter 7. Monitoring and evaluation 59 The system in place for spontaneous reporting of suspected adverse drug reactions (see section 3.2.5) was monitored and evaluated by an external consultant, Ms Augustina Appiah-Danquah from the National Centre of Pharmacovigilance in Ghana, with terms of reference from WHO Geneva. Her aim was to review the system of pharmacovigilance used in the project area from November 2005 onwards. Using the experience of the first six months, she updated the reporting system and the investigation and data review mechanisms at Alamata Hospital. She also conducted a training course for health professionals from the hospital, health centres and clinics on the reporting of suspected adverse drug reactions; and she established a link between the responsible person at Alamata Hospital and relevant project staff. A new active pharmacovigilance system was established, based on cohort event monitoring, and a new adverse drug reaction reporting form was designed by the project team and finalized after field testing by experts from the Tigray Health Bureau and San Gallicano. A copy of the form can be accessed at: http://apps.who. int/malaria/docs/diagnosisandtreatment/Malaria-PharmaVigil.pdf 60 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Chapter 8. Conclusions In areas of unstable malaria transmission, such as Ethiopia, epidemics are frequent, and early identification and treatment of infected patients is life-saving. The epidemic observed during the main rainy season of June, July and August 2005 was unusual but typical of a malaria epidemic in the study area. This epidemic coincided with distribution of the new first-line drug, artemether-lumefantrine, in the study area, providing an opportunity to save hundreds of lives. Many clinical malaria cases (approximately 58% of all cases) were diagnosed and treated by community health workers. The cases occurred during both the malaria transmission season and the unusual epidemic, and their case load was similar to that of the formal health institutions. The high proportion of treatment given by community health workers in the intervention district allowed peripheral health institutions, including health extension workers, to focus on other pressing public health problems and preventive interventions. The diagnostic performance of the community health workers with the rapid test in the intervention district matched that of personnel in the peripheral health institutions. There was relative underuse of community health workers by girls and women over 10 years, if the disease is assumed to be distributed equally by gender. This group should therefore be a focus for health extension workers, who are exclusively female and who might be expected to manage the health problems of women more effectively than male community health workers. The serial parasite prevalence surveys indicated that the epidemic was better managed in the intervention than the control district, as seen by the relatively rapid clearance of gametocytes. At the start of the project, in April 2005, the proportion of carriers of P. falciparum gametocytes was higher in the intervention district, but during the high-transmission season of September 2005 it was reduced by 7% in the intervention district and increased by 10-fold in the control district. Moreover, after the major epidemic of June, July and August 2005, in September 2005, the intervention district had a threefold lower crude parasite rate (all species and stages), a twofold lower crude asexual parasite rate (P. falciparum and P. vivax) and a threefold lower P. falciparum rate (both stages) when compared with the control district, indicating the efficacy of community-based distribution of artemether-lumefantrine, as the major vector control interventions were similar in the two districts. Chapter 8. Conclusions 61 Changes in transmission due to development, including water harvesting for food security, climatic factors, organization for the prevention and control of malaria and vector control coverage, are considered to have been broadly similar in the two districts, due to socioeconomic, cultural and geographical homogeneity. The study design was, however, unable to control for these possible confounders. A randomized controlled trial would be necessary in order to draw causal inferences. Despite the presence of community health workers, the survey of knowledge, attitude and practices indicated poorer awareness about malaria in the interven- tion than in the control district. This might be due partly to the severity of the epidemic, the resulting panic and the aggressive response of the health system in the control district, such that information, education and communication activities in that district were better than those in the calmer environment of the interven- tion district. It would be potentially misleading to compare the performance of rapid diagnostic tests applied by community health workers with that of personnel in formal health institutions, as the surveys were not conducted concurrently or with the same batch of tests. The low positive predictive value observed is due mainly to the fact that patients with positive results in the test and negative slides might have persistent antigenaemia after effective treatment or persistent but low (submicroscopic) parasitaemia due to noncompliance with the six-dose artemether- lumefantrine regimen, as indicated by the adherence survey. Evaluation of the skill of the community health workers in applying the rapid diagnostic test showed, however, that it was used safely. This simple, rapid test made it possible for appropriate antimalarial therapy to be administered to infected patients, permitted the early identification and referral of patients with diseases other than malaria, decreased the frequency of adverse drug reactions in uninfected patients and reduced expenditure on unnecessary drugs. When patients were treated by community health workers equipped with the rapid diagnostic test, the cost saving was as high as US$ 1.41 per patient examined when compared with the cost of diagnosing the same number of clients clinically. Generally, the test evaluated in this study met most of the requirements in this setting, except for its lower positive predictive value and its ability to identify only P. falciparum. In the initial adherence survey, the proportion of community health worker- treated patients who adhered to the six-dose artemether-lumefantrine regimen was lower than that of patients treated at health institutions (19% versus 10.7% definitely not adherent, respectively). In Phase II, a similar trend was observed, indicating that more work is required to promote adherence to treatment at community level. As the project was initiated at the same time as a malaria epidemic, the situation was perfect for testing the effectiveness of the intervention. Moreover, the parasite rates and parasite composition in the study area were monitored at the 62 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia start of the study in serial prevalence surveys, during the transition from phasing out of sulfadoxine-pyrimethamine to introduction of the new antimalarial drug, artemether-lumefantrine, and subsequently throughout the project. Although the two districts had similar initial parasite rates and showed overall downward trends over time, the transmission and the severity of the epidemic were extremely high in the control district, as demonstrated by a 10-fold lower crude parasitic cure rate in the intervention district. The mortality rates measured at the end of the 2-year project revealed an approximate 40% reduction in malaria-specific mortality. Generally, communities in the intervention district were satisfied with the intervention, as reflected in the focus group discussions, and expressed willingness to contribute to its sustainability. They appreciated its role in reducing morbidity and mortality from malaria, especially in light of the recent epidemic. The communi- ties clearly understood that halting the project would be detrimental and would undermine the success achieved to date. The project has shown that this interven- tion can reduce the burden of malaria. It should now be further integrated into the “health extension package” of the region. Main successes A significant proportion of the population at risk for malaria in the intervention district was treated at village level by volunteer community health workers. During the project, about 58% of all patients treated for malaria were managed by community health workers, close to home, achieving a major global strategy – prompt diagnosis and treatment. The project activities are highly likely to have contributed to the lower intensity and severity of the epidemic in the intervention district, as demonstrated by the mortality rates, prevalence rates, trends in morbidity and mortality rates in health institutions in both districts and morbidity rates in communities in the intervention district during Phase I. The project demonstrated that rapid diagnostic tests can be integrated into the practice of community health workers and that use of these tests, as opposed to clinical diagnosis, saved as much as US$ 1.41 per patient examined. As both P. falciparum and P. vivax are found in the project area and as the latter responds better to chloroquine, introduction of simple rapid diagnostic tests into community health worker treatment services maximized savings on the drug artemether-lumefantrine. Chapter 9. Recommendations 63 Chapter 9. Recommendations 1■ The pilot project in the study area was built on a tradition of volunteer partic- ipation in community activities. The new approach should therefore be sustained and replicated by full integration into the health extension package programme of the region, to contribute to the prevention and control of this debilitating disease. 2■ Beneficiaries and community health workers are still enthusiastic enough to maintain the established tradition of volunteer health services, and the health system and its partners should support integration of this volunteer service into existing community-based services. 3■ Community health workers are indispensable in malaria prevention, and they could not be replaced by health extension workers, with their current workload, vast catchment areas and multiple responsibilities. Therefore, any health extension worker outreach services for distributing artemether-lumefantrine with rapid diagnostic tests would only be a reasonable compromise, not a substitute for, community health workers. 4■ Strengthening of peripheral health services with the health extension package programme should be based on existing community health worker services, so that they are integrated into one, seamless system. 5■ Evidence from this pilot study shows that the community health worker volunteer service can upgrade to a treatment service based on rapid diagnostic tests. Continuous supportive supervision, provision of appropriate supplies, evaluation and encouragement must be maintained to ensure the sustaina- bility of the system. 6■ This upgrading is a constructive solution to problems such as cost, drug pressure and adverse drug reactions in uninfected patients. In view of the observed skills and reliability of community health workers and their good performance of rapid diagnostic tests, removal of these workers from the treatment service would delay prompt diagnosis and treatment, lower their morale and commitment and consequently increase mortality. 7■ The role of community health workers and their relationship to the health extension programme should be planned on the basis of these findings and in accordance with the regional strategic plan. 64 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia 8■ The treatment service delivered by community health workers should be strengthened and progressively extended to cover other districts of the region, taking into account the gaps identified in this study. 9■ Community health worker consultation guidelines should be updated on the basis of the findings of the surveys of e.g. knowledge, attitude and practices and adherence. 10■ Although previous community-based studies partly explain the dispropor- tionate service provided to women in terms of their heavy household workload, the root cause should be investigated further, and the possibility of dispro- portionate infection in adults by gender due to occupation or division of labour should be ruled out. 11■ Substantial effort is required to continue health education in the community to increase knowledge and influence care-seeking behaviour, adherence to prescribed doses of treatment and appropriate use of prevention. The focus should be on women, who manage family affairs, and on schools, where health education could be a means of reaching the whole family. 12■ The rapid diagnostic test evaluated in this study meets many of the require- ments in this setting and is recommended for extension of the service. The search for further simple, rapid diagnostic tests for peripheral use should, however, continue, to obtain good diagnostic performance, better tolerance to storage conditions, capacity to identify several species (P. falciparum and P. vivax) and capacity to distinguish persistent antigenaemia from low-level parasitaemia in a single kit that is safe and affordable, i.e. within the amount that could be diverted from increasing positive predictive value and expendi- ture resulting from late referral of patients with non-malarial illness. 13■ Community health workers, who are native insiders, cannot be bettered in giving a prompt diagnosis and treatment service for a widely dispersed, poor, predominately rural, inaccessible population, provided they are appropriately trained, well equipped with simple rapid diagnostic tests and effective antima- larial drugs and supported through frequent supervision. 14■ The factors underlying poor adherence should be assessed. Ensuring adherence to treatment should be strengthened, especially at community level, to maximize the impact of home-based management. 15■ Measures of quality for health workers and community health workers should be assessed so as to improve health service delivery. 16■ The long-term commitment of community health workers in Tigray to the service is exceptional. In many parts of the world, the drop-out rate of volunteers is almost always high. The factors that motivate community health workers in Tigray should be explored and replicated elsewhere. Appendices 65 Appendices 66 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Appendix A. Maps MAP 1. Study area with subdistricts ALAMATA RAYA AZEBO Appendix A. Maps 67 MAP 2. Rainfall in Tigray, estimated from satellite measurements, 1995–2004 MAP 3. Average temperatures in selected districts of Tigray 68 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Question Survey 1 Survey 2 control intervention control intervention 15–24/4/2005 13–20/4/2005 19–24/10/2005 19–24/10/2005 n % n % n % n % 1 Is there a radio in the household? 341 334 321 324 yes 63 18.5 66 19.8 57 17.8 82 25.3 no 278 81.5 268 80.2 264 82.2 242 74.7 2 In the rainy season,, where does the family sleep most often? 341 334 321 324 inside 336 98.5 334 100.0 312 97.2 318 98.1 outside 5 1.5 0 0.0 9 2.8 6 1.9 3 In the dry season, where does the family sleep most often? 341 334 321 324 inside 303 88.9 294 88.0 294 91.6 254 78.4 outside 38 11.1 40 12.0 27 8.4 70 21.6 4 Is there a community health worker in this village? 341 334 321 324 yes 290 85.0 168 50.3 264 82.2 187 57.7 no 46 13.5 82 24.6 35 10.9 32 9.9 don’t know 5 1.5 84 25.1 22 6.9 105 32.4 5 What is the main work of the community health worker in this village? 290 168 264 187 drug dispensing 224 77.2 124 37.1 223 84.5 154 82.4 education 11 3.8 1 0.3 6 2.3 0 0.0 environmental management 44 15.2 3 0.9 7 2.7 1 0.5 Farming 11 3.8 40 12.0 26 9.8 32 17.1 Mobilization 0 0.0 0 0.0 2 0.76 0 0.0 6 Is malaria a problem in this village? 341 334 321 324 yes 278 81.5 278 81.5 302 94.1 283 87.3 no 51 15.0 27 7.9 8 2.5 14 4.3 don’t know 12 3.5 29 8.5 11 3.4 27 8.3 7 Have you heard malaria discussed at meetings? 341 334 321 324 yes 214 62.8 180 52.8 242 75.4 172 53.1 no 109 32.0 96 28.2 57 17.8 112 34.6 don’t know 18 5.3 58 17.0 22 6.9 40 12.3 8 How does a person get malaria? 341 334 321 324 Mosquito bite 235 68.9 144 42.2 244 76.0 161 49.7 hunger 2 0.6 12 3.5 3 0.9 13 4.0 sleeping outside 9 2.6 10 2.9 5 1.6 21 6.5 small dam near the house 45 13.2 67 19.6 36 11.2 46 14.2 don’t know 50 14.7 101 29.6 33 10.3 83 25.6 9 Can untreated malaria cause death? 341 334 321 324 yes 246 72.1 319 93.5 283 88.2 313 96.6 no 87 25.5 4 1.2 27 8.4 3 0.9 don’t know 8 2.3 11 3.2 11 3.4 8 2.5 Appendix B. Summary tables Table B1. Summary of responses in two surveys of knowledge, attitude and practices Appendix B. Summary tables 69 Question Survey 1 Survey 2 control intervention control intervention 15–24/4/2005 13–20/4/2005 19–24/10/2005 19–24/10/2005 n % n % n % n % 10 Can malaria be treated with modern medicine? 341 334 321 324 yes 317 93.0 334 97.9 287 89.4 299 92.3 no 20 5.9 5 1.5 10 3.1 5 1.5 don’t know 4 1.2 18 5.3 24 7.5 20 6.2 11 If yes, where can you get medicine in your village? At a health institution? 317 311 287 299 yes 205 64.7 271 87.1 245 85.4 237 79.3 no 112 54.6 38 14.0 42 14.6 62 20.7 don’t know 0 0.0 2 5.3 0 0 0 0.0 From a community health worker? 317 3 287 299 yes 249 78.5 113 35.6 233 81.2 181 60.5 no 68 21.5 198 62.5 54 18.8 118 39.5 don’t know 0 0.0 0 0.0 0 0 0 0.0 12 Can you name the medicine? Chloroquine 317 311 287 299 yes 65 20.5 47 15.1 136 47.4 40 13.4 no 252 79.5 251 80.7 151 52.6 259 86.6 don’t know 0 0.0 13 5.2 0 0 0 0.0 Fansidar 317 334 287 299 yes 213 67.2 136 40.7 251 87.5 107 35.8 no 104 32.8 162 48.5 36 12.5 192 64.2 don’t know 0 0.0 13 3.9 0 0 0 0.0 Artemether-lumefantrine 317 334 287 299 yes 0 0.0 2 0.6 192 66.9 56 18.7 no 317 100.0 296 88.6 95 33.1 243 81.3 don’t know 0 0.0 13 4.4 0 0 0 0.0 others: Quinine, primaquine 317 334 287 299 yes 0 0.0 3 0.9 7 2.4 1 0.3 no 317 100.0 295 88.3 280 97.6 298 99.7 don’t know 0 0.0 13 4.4 0 0 0 0.0 13 Can malaria be treated by a traditional healer? 341 334 321 324 yes 24 7.0 22 6.6 32 10.0 12 3.7 no 261 76.5 247 74.0 268 83.5 269 83.0 don’t know 56 16.4 65 19.5 21 6.5 43 13.3 If yes, what is the treatment? 24 22 32 12 herbs 0 0.0 6 25.0 0 0 0 0.0 holy water 1 4.2 2 9.1 2 6.3 1 8.3 smoke 7 29.2 13 59.1 15 46.9 5 41.7 others 7 29.2 0 0.0 12 37.5 6 50.0 don’t know 9 37.5 1 4.5 3 9.4 0 0.0 14 Can malaria be prevented? 341 334 321 324 yes 311 91.2 259 77.5 296 92.2 271 83.6 no 27 7.9 16 4.8 12 3.7 13 4.0 don’t know 3 0.9 59 17.7 13 4.1 40 12.3 70 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Question Survey 1 Survey 2 control intervention control intervention 15–24/4/2005 13–20/4/2005 19–24/10/2005 19–24/10/2005 n % n % n % n % 15 If yes, in what way? Insecticide-treated nets? 341 334 296 271 yes 59 17.3 101 30.2 140 47.3 114 42.1 no 252 73.9 259 77.5 156 52.7 157 57.9 don’t know 0 0.0 0 0.0 0 0 0 0.0 House spray? 341 334 296 271 yes 17 5.0 61 18.3 136 46.0 111 41.0 no 294 86.2 194 58.1 160 54.1 160 59.0 don’t know 0 0.0 4 1.2 0 0 0 0.0 Prophylaxis? 341 334 296 271 yes 32 9.4 27 8.1 35 11.8 36 13.3 no 279 81.8 228 68.3 261 88.2 235 86.7 don’t know 0 0.0 4 1.2 0 0 0 0.0 Environmental management? 341 334 296 271 yes 214 62.8 115 34.4 174 58.8 127 46.9 no 97 28.4 140 41.9 122 41.2 144 53.1 don’t know 0 0.0 4 1.2 0 0 0 0.0 16 Has anyone in your family helped the community in environmental management for vector control this month? 341 334 296 271 yes 123 36.1 89 26.6 118 39.9 98 36.2 no 171 50.1 145 43.4 173 58.5 157 57.9 don’t know 17 5.0 25 7.5 5 1.7 16 5.9 17 Does indoor house spraying help prevent malaria? 341 334 321 324 yes 245 71.8 263 78.7 292 91.0 300 92.6 no 83 24.3 25 7.5 21 6.5 6 1.9 don’t know 13 3.8 46 13.8 8 2.5 18 5.6 18 Do you have your own insecticide- treated nets? 341 334 321 324 yes 196 57.5 182 54.5 204 63.6 183 56.5 no 145 42.5 152 45.5 117 36.5 141 43.5 19 Can insecticide-treated nets prevent malaria? 196 182 204 183 yes 184 93.9 173 95.1 197 96.6 176 96.2 no 10 5.1 5 2.7 7 3.4 2 1.1 don’t know 2 1.0 4 2.2 0 0 5 2.7 20 Has your family ever used insecticide- treated nets? 196 182 204 183 yes 116 59.2 170 93.4 177 86.8 149 81.4 no 80 40.8 12 6.6 26 12.8 30 16.4 don’t know 0 0.0 0 0.0 1 0.5 4 2.2 21 Is there a difference between impregnated and non-impregnated nets? 196 182 204 183 yes 142 72.4 139 76.4 184 90.2 169 92.3 no 45 23.0 4 2.2 13 6.4 4 2.2 don’t know 9 6.3 39 28.1 7 3.4 10 5.5 22 Did you impregnate your mosquito net? 196 182 204 183 yes 162 82.7 179 98.4 194 95.1 176 96.2 no 34 17.3 3 1.6 10 4.9 7 3.8 Appendix B. Summary tables 71 Question Survey 1 Survey 2 control intervention control intervention 15–24/4/2005 13–20/4/2005 19–24/10/2005 19–24/10/2005 n % n % n % n % 23 If yes, when did you last impregnate it? 196 182 204 183 6 months or less ago? 161 82.1 147 80.8 157 77.0 172 94.0 1 year or less ago? 35 17.9 35 19.2 47 23.0 11 6.0 24 How many insecticide-treated nets do you have in your house now? 215 210 25 If you have insecticide-treated nets, where did you get them? 196 182 204 183 commercially 1 0.5 8 4.4 3 1.5 7 3.8 government 194 99.0 169 92.9 200 98.0 176 96.2 other 1 0.5 5 2.7 1 0.5 0 0.0 26 How long have you had the insecticide- treated nets? 196 182 1–6 months 7 3.6 11 6.0 7–12 months 91 46.4 44 24.2 More than a year 98 50.0 127 69.8 27 Who slept under the insecticide-treated nets this week? 196 182 204 183 Adults 90 45.9 72 39.6 77 37.745 46 25.1 children 0 0.0 0 0.0 4 1.9608 11 6.0 Pregnant women and children 47 24.0 69 37.9 105 51.471 67 36.6 none 59 30.1 41 22.5 18 8.8235 59 32.2 28 What problems have you had with insecticide-treated nets? 196 182 204 183 can’t use it outside 1 0.5 7 3.8 1 0.49 3 1.6 dislike colour 22 11.2 4 2.2 0 0 1 0.5 holes 22 11.2 18 9.9 23 11.3 49 26.8 hot 2 1.0 4 2.2 1 0.5 0 0.0 not enough 47 24.0 27 14.8 123 60.3 49 26.8 small 0 0.0 5 2.7 2 1.0 3 1.6 no problem 102 52.0 117 64.3 54 26.5 78 42.6 29 Bed net inspection 196 196 204 183 Hanging over sleeping places? 196 196 204 183 yes 196 100.0 182 92.9 204 100 183 100.0 no 0 0.0 14 7.1 0 0 0 0.0 Holes? 196 182 204 183 yes 60 30.6 44 24.2 88 43.1 106 57.9 no 136 69.4 138 75.8 116 56.9 77 42.1 30 Do you need additional insecticide- treated nets? 196 182 204 183 yes 82 41.8 90 49.5 168 82.4 128 69.9 no 114 139 92 50.5 36 17.7 35 19.1 don’t know 0 0.0 0 0.0 0 0 20 10.9 72 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Symptom Age category (years) Gender 0.25–2 3–7 8–10 ≥ 10 total Male Female total headache no. 17 54 28 215 314 197 117 314 % 33.3 72.0 84.8 93.5 80.7 81.1 80.1 80.7 % ? 5.9 2.7 6.1 3.0 3.6 4.9 1.4 3.6 shivering no. 9 32 17 82 140 88 52 140 % 17.6 42.7 51.5 35.7 36.0 36.2 35.6 36.0 % ? 7.8 14.7 18.2 33.9 25.4 28.8 19.9 25.4 Vomiting no. 35 42 21 70 168 100 68 168 % 68.6 56.0 63.6 30.4 43.2 41.2 46.6 43.2 % ? 3.9 13.3 24.2 37.4 27.2 31.7 19.9 27.2 chills no. 18 37 19 145 219 141 78 219 % 35.3 49.3 57.6 63.0 56.3 58.0 53.4 56.3 % ? 5.9 8.0 12.1 11.3 10.0 11.5 7.5 10.0 Arthralgia no. 7 30 16 174 227 144 83 227 % 13.7 40.0 48.5 75.7 58.4 59.3 56.8 58.4 % ? 13.7 13.3 21.2 12.6 13.6 14.8 11.6 13.6 sweating no. 30 47 15 112 204 115 89 204 % 58.8 62.7 45.5 48.7 52.4 47.3 61.0 52.4 % ? 3.9 5.3 27.3 23.9 18.0 21.8 11.6 18.0 Anorexia no. 28 47 15 118 208 126 82 208 % 54.9 62.7 45.5 51.3 53.5 51.9 56.2 53.5 % ? 2.0 14.7 27.3 24.8 20.1 21.8 17.1 20.1 All patients 51 75 33 230 389 243 146 389 %?: percentage questionable Table B2. Main symptoms by age and gender of persons enrolled for rapid diagnostic test Appendix C. Project timetable 73 Appendix C. Project timetable (Phase I and Phase II) 74 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Ph as e I N o. M ai n ac ti vi ti es De ta ils o f pl an ne d ac ti vi ti es Ta rg et di st ri ct To ta l ta rg et (s am pl e) Re sp on si bl e in st it ut io n Pr ep ar at or y Ph as e, 20 05 20 05 20 06 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 1 W ri ti ng pr op os al , si gn in g ag re em en t an d pr ep ar in g gu id el in es W ri ti ng p ro je ct p ro po sa l, pr ep ar in g do cu m en ts an d si gn in g ag re em en t Pr ep ar at io n of o pe ra ti on al m an ua ls , in cl ud in g da ta c ol le ct io n fo rm s, t ra in in g m an ua ls a nd in fo rm at io n, c om m un ic at io n an d ed uc at io n m at er ia ls 2 Tr ai ni ng tr ai ni ng h ea lt h w or ke rs in m al ar ia d ia gn os is , tr ea tm en t an d re fe rr al w it h us e of r ap id di ag no st ic t es t an d ad m in is tr at io n of ar te m et he r- lu m ef an tr in e (fi rs t cy cl e) Al am at a 90 ti gr ay h ea lt h bu re au tr ai ni ng c om m un it y he al th w or ke rs in m al ar ia di ag no si s an d tr ea tm en t w it h ar te m et he r- lu m ef an tr in e Al am at a 33 ti gr ay h ea lt h bu re au tr ai ni ng o n m al ar ia p re va le nc e an d kn ow le dg e, at ti tu de a nd p ra ct ic es s ur ve ys f or d at a en um er at or s (f or fi rs t cy cl e) Al am at a & ra ya A ze bo 10 (5 e ac h) ti gr ay h ea lt h bu re au tr ai ni ng h ea lt h w or ke rs f or a dh er en ce s ur ve ys Al am at a 30 ti gr ay h ea lt h bu re au tr ai ni ng h ea lt h w or ke rs f or r ap id d ia gn os ti c te st p er fo rm an ce s ur ve ys Al am at a 10 ti gr ay h ea lt h bu re au tr ai ni ng h ig he r- le ve l h ea lt h pr ac ti ti on er s in ph ar m ac ov ig ila nc e fo r ar te m et he r- lu m ef an tr in e Al am at a 5 ir cc s1 re fr es he r tr ai ni ng o n su rv ey s fo r m al ar ia pr ev al en ce a nd k no w le dg e, a tt it ud e an d pr ac ti ce s fo r he al th w or ke rs ( se co nd c yc le ) Al am at a & ra ya A ze bo 10 (5 e ac h) ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng o f he al th w or ke rs in m al ar ia di ag no si s (r ap id d ia gn os ti c te st s) , tr ea tm en t an d re fe rr al in cl ud in g ph ar m ac ov ig ila nc e on ar te m et he r- lu m ef an tr in e Al am at a 81 he al th w or ke rs ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng o f co m m un it y he al th w or ke rs on m al ar ia d ia gn os is a nd t re at m en t w it h em ph as is o n tr ea tm en t w it h ar te m et he r- lu m ef an tr in e Al am at a 33 ti gr ay h ea lt h bu re au Appendix C. Project timetable 75 Ph as e I N o. M ai n ac ti vi ti es De ta ils o f pl an ne d ac ti vi ti es Ta rg et di st ri ct To ta l ta rg et (s am pl e) Re sp on si bl e in st it ut io n Pr ep ar at or y Ph as e, 20 05 20 05 20 06 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 3 Su rv ey s M al ar ia p ar as it e pr ev al en ce a nd k no w le dg e, at ti tu de a nd p ra ct ic es s ur ve ys p er d is tr ic t (fi rs t ro un d) Al am at a & ra ya A ze bo 30 cl us te rs , ea ch w it h 13 68 sa m pl e ti gr ay h ea lt h bu re au Ad he re nc e of p at ie nt s to a rt em et he r- lu m ef an tr in e at c om m un it y an d he al th in st it ut io ns Al am at a 5 he al th fa ci lit ie s an d 10 co m m un it y he al th w or ke rs (6 75 ) ti gr ay h ea lt h bu re au su rv ey o f ra pi d di ag no st ic t es t pe rf or m an ce at 5 h ea lt h in st it ut io ns ( ur ba n an d ru ra l) Al am at a (4 92 ) 5 he al th fa ci lit ie s di st ric t he al th o ffi ce , he al th f ac ili ti es , ti gr ay h ea lt h bu re au M al ar ia p ar as it e pr ev al en ce s ur ve ys in e ac h di st ric t (s ec on d ro un d) Al am at a & ra ya A ze bo (1 36 8) ea ch 30 cl us te rs ea ch di st ric t he al th o ffi ce , he al th f ac ili ti es , ti gr ay h ea lt h bu re au Ph ar m ac ov ig ila nc e of a rt em et he r- lu m ef an tr in e Al am at a 3 he al th fa ci lit ie s di st ric t he al th o ffi ce , he al th f ac ili ti es , ti gr ay h ea lt h bu re au 76 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Ph as e I N o. M ai n ac ti vi ti es De ta ils o f pl an ne d ac ti vi ti es Ta rg et di st ri ct To ta l ta rg et (s am pl e) Re sp on si bl e in st it ut io n Pr ep ar at or y Ph as e, 20 05 20 05 20 06 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 Su pe rv is io n di st ric t h ea lt h of fic e an d ti gr ay h ea lt h bu re au Al am at a & ra ya A ze bo ro ut in e Al l h ea lt h fa ci lit ie s an d co m m un it y he al th w or ke rs jo in t di st ric t he al th o ffi ce , t ig ra y he al th b ur ea u an d W h o he ad qu ar te rs Al am at a & ra ya A ze bo Ad h oc sa m pl e fr om h ea lt h fa ci lit ie s an d co m m un it y he al th w or ke rs jo in t ti gr ay h ea lt h bu re au a nd W h o (e th io pi a & h ea dq ua rt er s) M ek el le Ad h oc sa m pl e fr om h ea lt h fa ci lit ie s an d co m m un it y he al th w or ke rs 5 Da ta m an ag em en t da ta c ol le ct io n, c om pi la ti on a nd d oc um en ta ti on Al am at a & ra ya A ze bo ro ut in e ti gr ay h ea lt h bu re au de si gn in g da ta ba se ( en tr y qu es ti on na ire , c he ck fil es a nd a na ly si s pr og ra m m e on e Pi i nf o) Al am at a & ra ya A ze bo ro ut in e ti gr ay h ea lt h bu re au da ta e nt ry a nd c le an in g Al am at a & ra ya A ze bo ro ut in e ti gr ay h ea lt h bu re au 6 Su pp ly m an ag em en t Pr oc ur em en t of d on at ed p ha rm ac eu ti ca ls : co ar te m ® (a rt em et he r- lu m ef an tr in e, 2 0 m g/ 12 0 m g ta bl et s, n ov ar ti s Ph ar m a Ag ) an d Pa ra ch ec k- Pf ® ti gr ay Pr oj ec t An nu al no va rt is ( it al y) , ti gr ay h ea lt h bu re au re qu is it io n fo r Ph as e ii ( co ar te m ® an d Pa ra ch ec k- Pf ®) ti gr ay Pr oj ec t An nu al ti gr ay h ea lt h bu re au , no va rt is ( it al y) Pr oc ur em en t of n on -p ha rm ac eu ti ca l s up pl ie s ti gr ay Pr oj ec t bi an nu al ti gr ay h ea lt h bu re au Pr oc ur em en t of m ed ic al a nd n on -m ed ic al su pp lie s (v eh ic le , la pt op c om pu te rs , m ot or cy cl es , m ic ro sc op e, p ho to co pi er s) ti gr ay Pr oj ec t W h o de liv er in g ar te m et he r- lu m ef an tr in e, Pa ra ch ec k- Pf ® an d gl ov es t o th e pr oj ec t di st ric t Al am at a Qu ar te rly h ea lt h fa ci lit ie s, co m m un it y he al th w or ke rs Appendix C. Project timetable 77 Ph as e I N o. M ai n ac ti vi ti es De ta ils o f pl an ne d ac ti vi ti es Ta rg et di st ri ct To ta l ta rg et (s am pl e) Re sp on si bl e in st it ut io n Pr ep ar at or y Ph as e, 20 05 20 05 20 06 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 7 Re po rt in g Pr og re ss r ep or t Al am at a & ra ya A ze bo Qu ar te rly ti gr ay h ea lt h bu re au M id -y ea r pr og re ss r ep or t Al am at a & ra ya A ze bo An nu al ti gr ay h ea lt h bu re au M id -t er m e va lu at io n Al am at a & ra ya A ze bo An nu al ti gr ay h ea lt h bu re au , W h o, F ed er al M in is tr y of h ea lt h, ir cc s1 , no va rt is ( it al y) 1. d ep ar tm en t of P re ve nt iv e M ed ic in e in M ig ra ti on , to ur is m a nd t ro pi ca l d er m at ol og y, s an g al lic an o h os pi ta l, ro m e 78 Deployment at community level of artemether-lumefantrine and rapid diagnostic tests – Raya Valley, Tigray, Ethiopia Ph as e II N o. M ai n ac ti vi ti es De ta ils o f pl an ne d ac ti vi ti es Ta rg et di st ri ct To ta l t ar ge t (s am pl e) Re sp on si bl e in st it ut io n 20 06 20 07 4 5 6 7 8 9 10 11 12 1 2 3 1 Tr ai ni ng tr ai ni ng e nu m er at or s fo r m or ta lit y ce ns us Al am at a & ra ya A ze bo 11 0 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng f or h ea lt h w or ke rs in m al ar ia di ag no si s, t re at m en t an d re fe rr al w it h ra pi d di ag no st ic t es t an d ad m in is tr at io n of a rt em et he r- lu m ef an tr in e Al am at a 90 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng f or c om m un it y he al th w or ke rs in m al ar ia d ia gn os is a nd t re at m en t w it h ar te m et he r- lu m ef an tr in e Al am at a 33 ti gr ay h ea lt h bu re au tr ai ni ng c om m un it y he al th w or ke rs in m al ar ia di ag no si s, t re at m en t an d re fe rr al , in cl ud in g us e of r ap id d ia gn os ti c te st s fo r 50 % Al am at a 16 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng in m al ar ia p re va le nc e fo r da ta e nu m er at or s (fi rs t ro un d) Al am at a & ra ya A ze bo 10 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng f or c om m un it y he al th w or ke rs a nd t em po ra ry e nu m er at or s in a dh er en ce su rv ey s Al am at a 31 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng f or h ea lt h w or ke rs in a dh er en ce su rv ey s Al am at a 16 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng f or c om m un it y he al th w or ke rs a nd t em po ra ry e nu m er at or s in r ap id di ag no st ic t es t pe rf or m an ce s ur ve ys Al am at a 16 ti gr ay h ea lt h bu re au re fr es he r tr ai ni ng in m al ar ia p re va le nc e su rv ey s fo r en um er at or s (s ec on d ro un d) Al am at a & ra ya A ze bo ti gr ay h ea lt h bu re au tr ai ni ng in p ha rm ac ov ig ila nc e fo r he al th w or ke rs at t hr ee h ea lt h ce nt re s an d on e ho sp it al ti gr ay h ea lt h bu re au Appendix C. Project timetable 79 Ph as e II N o. M ai n ac ti vi ti es De ta ils o f pl an ne d ac ti vi ti es Ta rg et di st ri ct To ta l t ar ge t (s am pl e) Re sp on si bl e in st it ut io n 20 06 20 07 4 5 6 7 8 9 10 11 12 1 2 3 2 Su rv ey s M or ta lit y ce ns us Al am at a & ra ya A ze bo > 50 h ou se ho ld s ti gr ay h ea lt h bu re au M al ar ia p ar as it e pr ev al en ce a nd k no w le dg e, at ti tu de a nd p ra ct ic es s ur ve ys in e ac h di st ric t (fi rs t ro un d) Al am at a & ra ya A ze bo ti gr ay h ea lt h bu re au M al ar ia p ar as it e pr ev al en ce a nd k no w le dg e, at ti tu de a nd p ra ct ic es s ur ve ys in e ac h di st ric t (s ec on d ro un d) Al am at a & ra ya A ze bo ti gr ay h ea lt h bu re au Ad he re nc e su rv ey o f pa ti en ts t o ar te m et he r- lu m ef an tr in e at c om m un it y le ve l a nd h ea lt h in st it ut io ns (6 75 ) ti gr ay h ea lt h bu re au ra pi d di ag no st ic t es t pe rf or m an ce s ur ve y at 1 6 co m m un it y si te s du ri ng m ai n tr an sm is si on p er io d Al am at a 30 0 ti gr ay h ea lt h bu re au Ph ar m ac ov ig ila nc e of a rt em et he r- lu m ef an tr in e Al am at a & ra ya A ze bo Al l o ut pa ti en t de pa rt m en t ir cc s Fo cu s gr ou p di sc us si on s (c om m un it y) Al am at a tw o gr ou ps ti gr ay h ea lt h bu re au 3 Da ta m an ag em en t da ta c ol le ct io n, c om pi la ti on a nd d oc um en ta ti on Al am at a ro ut in e Pr oj ec t co or di na to r da ta e nt ry a nd c le an in g Al am at a ro ut in e da ta t ec hn ic ia ns da ta a na ly si s an d su m m ar y Al am at a ro ut in e da ta t ec hn ic ia ns 4 Pr oc ur em en t an d de liv er y of s up pl ie s Pr oc ur em en t of d on at ed p ha rm ac eu ti ca ls ti gr ay Pr oj ec t An nu al no va rt is ( it al y) , W h o, ti gr ay h ea lt h bu re au Pr oc ur em en t of n on -p ha rm ac eu ti ca ls Al am at a An nu al ti gr ay h ea lt h bu re au de liv er y of a rt em et he r- lu m ef an tr in e, c hl or oq ui ne , Pa ra ch ec k- Pf ®, g lo ve s, e tc . to p ro je ct d is tr ic t Al am at a Qu ar te rly ti gr ay h ea lt h bu re au 5 Su pe rv is io n di st ric t he al th o ffi ce a nd t ig ra y h ea lt h bu re au Al am at a ro ut in e di st ric t he al th o ffi ce , di st ric t M al ar ia co nt ro l u ni t di st ric t he al th o ffi ce a nd t ig ra y h ea lt h bu re au Al am at a & ra ya A ze bo An nu al di st ric t he al th o ffi ce , ti gr ay h ea lt h bu re au 6 Re po rt in g Pr og re ss r ep or t Al am at a & ra ya A ze bo Qu ar te rly ti gr ay h ea lt h bu re au Pr oj ec t re po rt Al am at a 2 ye ar s ti gr ay h ea lt h bu re au 1 de pa rt m en t of P re ve nt iv e M ed ic in e in M ig ra ti on , to ur is m a nd t ro pi ca l d er m at ol og y, s an g al lic an o h os pi ta l, ro m e

Deployment at community level of artemether-lumefantrine and rapid diagnostic tests Raya Valley, Tigray, Ethiopia GLOBAL MALARIA PROGRAMME