Using community health workers for malaria control: experience in Zaire.

Using community health workers for malaria control: experience in Zaire C. Delacollette,1 P. Van der Stuyft,2 & K. Molima3 The potential for using community health workers (CHW) for administering timely and effective treatment for presumptive malaria attacks was evaluated in the Katana health zone in Zaire. In each of the 12 villages of an intervention area (area A) with 13000 inhabitants, a CHW was trained in the use of a simple fever management algorithm. The CHWs performed their services under the supervision of the nurse in charge of the area's health centre (HC). Malaria morbidity and mortality trends were monitored during 2 years in area A and in an ecologically comparable control area (area B), where malaria treatment continued to be available at the HC only. Health care behaviour changed dramatically in the intervention area, and by the end of the observation period 65% of malaria episodes were treated at the community level. Malaria morbidity declined 50% in area A but remained stable in the control area. Parasitological indices showed similar trends. Malaria- specific mortality rates remained, however, at essentially the same levels in both areas. The non-compre- hensiveness of the CHWs' care and their ambiguous position in the health care system created problems that compromise the sustainability of the intervention. Introduction Early diagnosis and prompt treatment of malaria are essential to avert severe morbidity and mortality in non-immune individuals (1). Together with the im- plementation of sustainable preventive measures and the strengthening of local research capacities, the early detection and containment of epidemics constitute the basic technical elements of the Global Strategy for Malaria Control (2). Timely profes- sional treatment of malaria episodes is, however, hard to organize in regions with scattered popu- lations who have to travel long distances to the nearest health care facility. It has been suggested that community health workers (CHWs) could pro- vide correct case management at the community level and thereby increase the access to good quality care (3, 4). This study in the Katana health zone, Zaire, aimed to evaluate further the potential of CHWs to reduce malaria morbidity and mortality. Subjects and methods Study area and population The Katana health zone is situated in the eastern I Lutte contre les Maladies transmissibles et carentielles (LMTC Project), Malaria Branch, Bujumbura, Burundi. 2 Epidemiology Unit, Department of Community Health, Institute of Tropical Medicine, Nationalestraat 155, B-2000 Antwerp, Belgium. Requests for reprints should be addressed to this author. 3 FOMULAC (Fondation medicale de lUniversit4 de Louvain en Afrique centrale), Bukavu, Zaire. Reprint No. 5720 part of Zaire, on the western shore of Lake Kivu (Fig. 1). This mountainous region (altitude O1500m) has a temperate climate with two rainy seasons sepa- rated by a short (January) and a prolonged (May- September) dry season. The health zone has an area of approximately 1200km2. It is exclusively rural, with a homogeneous population of about 210000 persons who belong to the Shi tribe. Less than 1% of households have an income from salaried jobs or trading, and subsistence farming is virtually the only economic activity (5). The level of educational at- tainment is very low, particularly in girls, of whom almost 90% do not complete a single year of formal schooling (6). Primary health care (PHC) is deliv- ered through a network of 17 health centres (HC) and a well-equipped 660-bed hospital. The priority health problems are related to infectious and para- sitic diseases (7), and the infant mortality rate and child mortality quotient attain respectively 13% and 18%. Malaria epidemiology and control efforts Malaria ranks third among causes of mortality, and fever attributed to malaria is the most frequent diag- nosis at the HCs. The predominant malarial parasite is Plasmodium falciparum: found alone (86%), sometimes associated with P. malariae (10%), but rarely with P. ovale (<2%). In 1986, less than 5% of the isolated P. falciparum strains showed an RII resistance to chloroquine (8). Malaria is meso- endemic, and transmission is continuous with sea- sonal fluctuations (9). The spleen and parasite rates (the proportion, respectively, of healthy individuals with enlarged spleens, and with thick blood films Bulletin of the World Health Organization, 1996, 74 (4): 423-430 © Word Health Organization 1996 423 C. Delacollette et al. Fig. 1. Location of the Katana health zone and study areas. lected for monitoring malaria morbidity and mor- tality under routine health care protocols (i.e. treat- ment at the nearest HC only). The two areas, both peninsulas, are situated 22-km south and 10-km north of the hospital. They were closely comparable socioeconomically, and their populations were each covered by HCs that supplied curative and preven- tive services representative for the Katana health zone. The areas had the same malarial ecology, and malariometric indices were similar in each (9). The 12 villages composing area A had an average population of 1200 (range, 500-1500). These communities received, before the intervention, edu- cational messages on malaria. They approved the plans for the malaria control project, and in each village the inhabitants chose a literate volunteer as malaria CHW. The selected CHW received 2 weeks of in-service training in the area's HC in the use of a simple treatment algorithm for fever and the early recognition and local management of malaria pa- tients (Fig. 2). They were also instructed in the use of Fig. 2. Flow chart for management of fever (i.e. possi- ble malaria) patients by community health workers (CHW) in Katana, Zaire, 1986. HC = health centre. positive for Plasmodium parasites) among the popu- lation <5 years of age range from 5% to 18% and from 25% to 44%, respectively, and both attain their maximum during the long dry season. The age pattern for acquisition of antibodies against P. falciparum is characteristic for unstable malaria transmission (9). The health services' routine ma- laria control activities concentrate on the treatment of clinical cases presenting at the HC or hospital. Chemoprophylaxis is not promoted, either for pregnant women or children. Impregnated bednets are used, on the initiative of the individual, by an insignificant minority of better-off inhabitants. The use of mosquito repellents is somewhat more widespread. The intervention The zone's PHC development plan set up a pilot community trial for providing malaria treatment at the "grass-roots" level. In all villages of the interven- tion area (area A) villagers were identified who would quickly provide chloroquine phosphate treat- ment for isolated episodes of fever (i.e. presumed malaria attacks). In order to measure the impact of this intervention, a control area (area B) was se- WHO Bulletin OMS. Vol 74 1996 WHO 96263 424 Using community health workers for malaria control in Zaire Table 1: Population in the study areas, Katana, Zaire (1 September 1986) Area A Area B Age (years) Males Females Total Males Females Total <1 259 237 496 (3.8)a 311 318 629 (4.2) 1-4 1 038 1 071 2 109 (16.1) 1 166 1 204 2 370 (15.8) 5-9 994 1 005 1 999 (15.3) 1 137 1 147 2 284 (15.2) 10-14 795 806 1 601 (12.2) 906 897 1 803 (12.0) 15-24 1 080 1 188 2 268 (17.3) 1 307 1 395 2 702 (18.0) 25-34 837 865 1 702 (13.0) 947 936 1 883 (12.6) 35-44 578 581 1 159 (8.9) 728 635 1 363 (9.1) 45-54 312 366 678 (5.2) 406 364 770 (5-1) 55-64 298 268 566 (4.3) 303 362 665 (4.4) ¢65 307 199 506 (3.9) 305 225 530 (3.6) Total 6498 6586 13 084 (100) 7516 7483 14 999 (100) a Figures in parentheses are percentages. a contact register that had sex, age, major signs and symptoms, and referral as entries. The treatment of presumptive cases of malaria consisted of oral ad- ministration of chloroquine phosphate (25 mg/kg for 3 days), which was, at the time, the recommended "first line" drug in the region (10). It was provided at cost (US$ 0.008 per tablet containing 100 mg chloroquine base), and no consultation fees were charged by the CHW, making treatment three times cheaper than at the HC. After training, the CHWs started their activities in area A communities. Since CHWs were also local farmers, they were, in principle, always accessible to the villagers, who had been motivated through health education to consult the CHW for any fever episodes. They worked under the close supervision of the nurses in charge of the HCs and also attended monthly meetings chaired by the medical coordina- tor of the project (CD). They received only a sym- bolic monetary reward, as well as the standing gained in the community. Nevertheless, no CHW dropped out of the project. In area B, no specific malaria control effort was undertaken. Ill persons identified during the passage of the morbidity and mortality survey team were, however, referred to the area's HC. Such patients were exempted from pay- ing the normal consultation fees. Data collection, management, and analysis The survey methodology (6) and details of the ques- tionnaires and diagnostic criteria have been de- scribed previously (7).a In short, information on the mortality and morbidity in the population of both a Van der Stuyft P. [An epidemiology unit in a modern academic environment: redundant anachronism or vital quintessence?J (in Dutch). Antwerp, Universitaire Instelling Antwerpen, 1991 (Ph.D. thesis). areas was collected in a multi-round survey. The first round started on 1 August 1985 and was followed by three consecutive rounds at roughly 6-month inter- vals. Enumeration was on a de jure basis (including all declared residents, present or not), and recruit- ment of neonates and migrants continued until the end of the survey. In each round, all households were visited twice, with a 1-week interval between visits. At each visit, 7-day morbidity recall was elicited. The person-time morbidity sampled thus varied each 7- day interval. Because of their gradual recruitment and release, not all households were followed over the same calendar period, but between 1 March 1986 and 28 February 1987 the complete population of the two areas was under mortality surveillance. Mortal- ity recall for the previous 6 months was only elicited during the last three survey rounds. The last survey round ended 31 July 1987 (see Table 1). Surveys were carried out by two field-teams consisting each of one supervisor and six interviewers, the latter of whom were permanent residents in the area and had completed at least 6 years of primary schooling. They each visited 40-50 households per week. Supervisors were graduate nurses, who inspected all completed survey forms, reviewed doubtful or abnormal results with the surveyors, and made, when necessary, control visits. They revisited a random sample of 10% of households each week. The basic household questionnaire, written in the Shi language, consisted of four parts: administra- tive information, information on relatively fixed household characteristics, specific information on each household member, and information to be re- corded at each subsequent visit, among other things the household's total number of deaths, births, and migrations, the total number of ill persons on the day of the visit (prevalence), and the number of new illness episodes starting during the previous 7 days WHO Bulletin OMS. Vol 74 1996 425 C. Delacollette et al. (incidence). For 96% of the reported events the mother was the interviewee. For each death or episode of illness information on the subject concerned was collected on an indi- vidual mortality or morbidity questionnaire. The date and circumstances of the event, the occurrence and duration of standardized symptoms, the health care behaviour and, possibly, the treatment given were recorded. These questionnaires were adapted to the local situation from those proposed in earlier studies (11).b The morbidity questionnaire at- tempted to distinguish between seven syndromes: malaria, diarrhoea, acute upper respiratory infec- tion, acute lower respiratory infection, skin rash in children, trauma, and "others". A diagnosis of ma- laria was made if the presence of fever, possibly with chills or vomiting, but without further complaints indicative of other diseases, was reported. On the mortality questionnaire, in addition to the absence or presence of standardized signs and symptoms and their duration, any relevant observation or comment of the caregiver was recorded. These forms were interpreted by two doctors who independently ex- tracted the probable cause of death. A death was attributed to malaria when the verbal autopsy re- vealed isolated fever of acute onset, with or without convulsions, possibly accompanied by vomiting and headache, but without any further signs or symptoms (thereby excluding patients with, among other things, concomitant acute respiratory infection, mea- sles, or diarrhoea), leading to death within 7 days. Agreement between the doctors was high (K = 95%) and any disagreement was discussed in order to find a consensus. A cross-sectional malarial survey was con- ducted before (in February 1985) and during the intervention period (in February 1987). Thick blood films were collected from a sample of healthy sub- jects living in both areas. The blood films were Giemsa stained and 200 microscopic fields were ex- amined under oil immersion (objective X100, eye- piece x10). The crude parasitological index and the high parasitaemia index, i.e. the percentage of slides with more than 2000 asexual forms of P. falciparum per mm3 of blood were determined (12). All data were entered in dBase III and a print- out of the critical items was systematically compared with the survey forms; data consistency was checked with a custom-made verification programme. Data analysis was performed with SPSS/PC software. The bMacCormack CP. Estimation of malaria mortality and morbidity in different ecological areas of tropical Africa, such as tropical rainforest, forest, savannah and Sahel. Ad hoc consultation on applied field research in malaria in Africa. Nairobi, 1983 (TDR- FIELDMAL, Proposal No. 26). difference between proportions or rates was tested with a two-tailed X2 test, and the reported 95% con- fidence intervals are test based. Results Table 2 summarizes by area the evolution of the crude malaria morbidity and mortality rates before and during the CHW intervention. The prevalences and incidences showed a significant 50% decline in the intervention area but remained stable in the con- trol area. The specific mortality rates fluctuated in both areas, but pre- and post-intervention levels re- mained essentially the same. The parasitological in- dices reflected the malaria morbidity trend (Table 3): the crude parasitological index and high para- sitaemia index respectively showed a five- and six- fold reduction in the intervention area, whereas the decline was far more modest (a two-fold reduction) in the control area. During the intervention important changes took place in the health care behaviour of the population in area A (Table 4). The number of malaria episodes that remained untreated decreased significantly, more cases were treated at home (+16%) and by the CHW (+ 16%), and the utilization of the health sec- tor (-9%) and of the informal private sector (-23%) sharply decreased. This resulted in more than 65% of the episodes being treated at the com- munity level by the end of the observation period. The health care utilization profile in the control area showed little change, but the proportion of cases receiving treatment decreased by 8%. At the same time there was, in the intervention area but not in the control area, an increase in the use of chloroquine as the first-line treatment and a concurrent drop in the use of quinine (from 15% to 7%). The study permitted the observation of the in- stitutional, social and financial problems related to the use of CHWs for malaria control activities. The main findings are summarized qualitatively below. * Problems concerning the relation between the CHWs and the health care system, as follows: - CHWs desired more than a token financial re- ward for their activities; - CHWs were eager to receive further training and to broaden the scope of their therapeutic activities; - CHWs wished to be established formally in the hierarchy of the health care system and to have prospects for career development; and - project management and supervision of CHWs unacceptably increased the workload of HC staff. 426 WHO Bulletin OMS. Vol 74 1996 Using community health workers for malaria control in Zaire Table 2: Malaria morbidity and mortality, by period and area, Katana, Zaire, 1985-87 Mean incidence per 10000 Mortality per 10000 person- Mean prevalence person-weeks (No. of person- months (No. of person-months (per 10 000) weeks of observation) of observation) Period Area A Area B Area A Area B Area A Area B Aug. 85-March 86a 143 (21 455)b 121 (22630)b 218 (21 455) 205 (22630) 17 (102410) 27 (116541) April 86-July 86c 133 (17139) 91 (25458) 176 (17139) 115 (25458) 21 (51887) 35 (59490) Aug. 86-March 87d 74 (33285) 109 (31 235) 104 (33285) 145 (31 235) 14 (103704) 27 (120879) April 87-July 87d 75 (18870) 104 (25702) 99 (18870) 134 (25702) 32 (21 944) 22 (36530) Rate ratio, 1.9; 1.6-2.1' 1.0; 0.9-1.1 1.9; 1.7-2.2 1.1; 1.0-1.2 1.1; 0.6-1.9 1.1; 0.8-1.7 a Baseline period (no intervention). b Figures in parentheses are the number of observations made in the study population during the corresponding period. c Initial period of intervention in area A. d Programme period in area A (fully operational). (rate Aug. 85-July 86)/(rate Aug. 86-July 87) = (baseline + initial period)/(programme period). Figures in italics are the 95% confidence interval. Table 3: Crude parasitological index (PI) and high parasitaemia index (HPI) for Plasmodium falciparum by area and period, Katana, Zaire Area A Area B Period n Pi Hpla n Pi HPIa Feb. 85 255 34.1 13.3 254 37.8 17.3 Feb. 87 229 7.0 2.2 217 19.3 9.2 Rate ratiob 4.9 6.0 2.0 1.9 (3.0-8.1)c (2.4-15.3) (1.4-2.7) (1-1-3.1) a % of slides with >2000 asexual forms of P. falciparum per mm3 of blood. b (rate during Feb. 85)/(rate during Feb. 87). c Figures in parentheses are the 95% confidence interval. Table 4: Health care behaviour during a malaria outbreak, Katana, Zaire, 1985-87 Area A Area B Aug. 85-March 86a April 86-July 87b Difference Aug. 85-March 86a April 86-July 87b Difference No. of malaria episodes 467 835 - 464 1 089 - Episodes without treatment 31% 24% -7% 20% 28% +8% Mean number of treatment 1.2 1.2 0.0 1.3 1.2 -0.1 decisions per patientc Treatment decided on:d At home by family member 33% 49% +16% 48% 57% +9% By CHW - 16% +16% - - - By health care provider in 50% 27% -23% 39% 30% -9% the private sectore By health system personnel 17% 8% -9% 13% 13% 0% a No community health worker (CHW). b CHW present (fully operational from Aug. 86). c For treated patients. d Denominator = total number of treatments. e Predominantly practitioners using Western drugs (3% traditional healers). WHO Bulletin OMS. Vol 74 1996 427 C. Delacollette et al. * Problems concerning the relation between the CHWs and the community, as follows: - the community expected CHWs to deliver com- prehensive and continuous care and became pro- gressively disappointed by their limited services; - the community was not inclined to compensate CHWs for their efforts, financially or otherwise; - CHWs tended to elude community (and HC) control of their activities; and - CHWs did not catalyse genuine community par- ticipation in malaria control or in health care in general. Discussion The introduction of a dedicated (for malaria only) CHW in an area where access to the health care system is problematic because of a scattered popula- tion should result in increased access to timely and appropriate treatment. The presence of such CHWs does not, however, automatically lead to the utiliza- tion of their services. The concurrent delivery of health information and education enhances com- munity perception of the malaria problem. In fact, behavioural changes are critical to the effective utili- zation and success of a control programme based on the deployment of CHWs. Failure to secure sufficient behavioural changes in the population studied may, to some extent, ex- plain why CHWs treated only 16% of malaria cases in the intervention area. That their services were non-comprehensive could be another partial expla- nation for this, as could the existence of a financial barrier: although no consultation fees were charged, patients still had to pay for the drugs they received. Nevertheless, health care behaviour regarding ma- laria has dramatically changed in the intervention area but not in the control area. The number of malaria cases receiving no treatment has been sub- stantially reduced: 65% of all malaria episodes in the intervention area are now managed at the commu- nity level, and the frequency of visits to the private health sector has dropped sharply. There is also a strong indication that home treatment of malaria has become more appropriate. Extensive cinchona plantations in the Katana region make quinine easily available through the informal private sector and drug vendors. Although in the control area quinine is inappropriately relied upon, in the intervention area it is being replaced by chloroquine as the first-choice drug for self-care. While self-medication for malaria is widespread in Africa (13, 14), the correct use of drugs is the excep- tion (15). The presence, example, and educational efforts of the CHWs undoubtedly played a major role in the favourable evolution of drug use in the intervention area, as well as in the reduction of treat- ments obtained from the informal private sector. This result is even more striking when the low educa- tional level and 89% female illiteracy rate in the region (6) are taken into account, as these are factors that may strongly constrain the effectiveness of edu- cational efforts and of health interventions in gen- eral (16). A reduction in the proportion of patients who remain untreated, shorter delays between the first symptoms of malaria and the start of treatment, and increased adequacy of therapy all contribute to a decrease in the average duration of a malaria epi- sode. This must be partially responsible for the two- fold decline in malaria morbidity in the intervention area and the six-fold drop in the high parasitaemia index. The malaria mortality rates fluctuated in both areas, but pre- and post-intervention levels remained essentially the same in all age groups. In a hyper- endemic region in Kenya, Spencer et al. (17) also failed to demonstrate an impact on overall and malaria-specific mortality after the introduction of CHWs who delivered malaria treatment. Systematic chemoprophylaxis provided by CHWs succeeded, in contrast, in reducing malaria morbidity and mortal- ity rates in Farafenni, a holo-endemic region in the Gambia with highly seasonal transmission (18). The small size of the study population (approxi- mately 15 000 persons in each area), however, would hardly permit the demonstration of an effect on ma- laria mortality rates: to have an 80% chance of de- tecting a significant 20% reduction in mortality among the population <5 years of age would re- quire, as standard formulae (19) indicate, a sample five times as big as this. The validity of verbal autop- sies in diagnosing malaria-specific mortality should also be considered. Although the technique is quite specific, it has a relatively low sensitivity (20, 21), which could bias the mortality ratios in this study and lead to an underestimation of the effect of the intervention. At any rate, it has been suggested that malaria- specific mortality may not provide the best or most meaningful indicator for assessing community-based malaria control programmes (17, 22). Changes in intermediate outcome indicators- such as morbid- ity rates (12), or parasite and spleen rates (23) may provide sufficient evidence, and so-called pro- cess indicators - such as access to services, the pro' portion of cases receiving treatment and the quality and adequacy of care - would seem equally impor- tant. In this respect the CHWs' interventions made a difference in the Katana health zone and have con- WHO Bulletin OMS. Vol 74 1996428 Using community health workers for malaria control in Zaire tributed towards improving well-being by reducing morbidity due to malaria. Pilot project effectiveness does not, however, imply success in a programme context, and the prob- lems observed after the start of the present interven- tion should raise questions about its sustainability. The major problems were related to the long-term commitment of the volunteer health workers and to the failure to secure genuine community participa- tion, which were accompanied by a progressive loss of enthusiasm. These are comparable to the drawbacks that have hampered the implementation of CHW schemes of widely differing scope and nature (24-26). In essence, all such problems relate to the ambiguity of the CHWs' position as a makeshift liaison between the health care system and the community (27). In conclusion, this study indicates that the intro- duction of dedicated CHWs can lead to improved access to and utilization of health care for malaria and result in a decline of malaria morbidity. Never- theless, a substantial part of these results can prob- ably be attributed to the increase of geographical and financial accessibility to an essential drug (chloroquine), regardless of provider, that satisfied a perceived need in the community. At the same time, this suggests that deployment of dedicated CHWs may not be sustainable, since it leads to a series of problems that can be overcome only by training them for more comprehensive patient management and integrated care and offering them career prospects. A restricted approach seems meaningful only in specific epidemiological circumstances such as out- breaks or, perhaps, massive population movements of non-immune individuals into endemic areas. Acknowledgements This work received financial support from the UNDP/World Bank/WHO Special Programme for Research and Train- ing in Tropical Diseases (No. 850100) and from the Belgian Administration for Development Cooperation. Pro- fessor M. Wery's helpful suggestions for improving the manuscript are gratefully acknowledged. We are indebted to the nurses and technicians who assisted with the project and gratefully acknowledge the participation of the local communities, the health centre staff, and the admin- istrative authorities. Resume Utilisation des agents de sante communautaires pour la lutte antipaludique: une experience au Zaire La possibilite d'utiliser les agents de sante communautaires pour administrer en temps utile un traitement efficace en cas de presomption d'acces palustre a ete 6valu6e dans la zone de sante de Katana au Zaire. Dans chacun des 12 villages d'un secteur d'intervention (secteur A) comptant 13 000 habitants, un agent de sante communautaire a ete form6 a l'utilisation d'un algorithme simple de prise en charge des cas f6briles. L'agent de sante tra- vaillait dans la communaute, sous la supervision de l'infirmiere responsable du centre de sante du sec- teur. Une enquete comportant plusieurs passages a permis de surveiller pendant deux ans les ten- dances de la morbidite et de la mortalite palustres dans le secteur A et dans un secteur temoin 6co- logiquement comparable (secteur B), ou le traite- ment antipaludique restait uniquement disponible au centre de sant6. Les taux de morbidite ont baiss6 de 50% dans le secteur d'intervention et sont rest6s stables dans le secteur t6moin. Les taux de mor- talite palustre ont varie dans les deux secteurs, mais les valeurs avant et apres intervention sont restees sensiblement 6gales. Les indices parasito- logiques suivaient les tendances de la morbidite: l'indice parasitologique et l'indice de forte para- sit6mie ont ete divises respectivement par 5 et par 6 dans le secteur d'intervention alors que leur baisse 6tait beaucoup plus modeste dans le secteur t6moin. Au cours de l'intervention, des modifications importantes du comportement de la population en matiere de soins de sante ont ete enregistrees dans le secteur A: le nombre d'6pisodes palustres non trait6s a diminue de fa,on significative, un plus grand nombre de cas ont 6t6 trait6s a domicile (+16%) et par I'agent de sante communautaire (+16%), et le recours au secteur priv6 informel a fortement diminue (-23%). Le profil d'utilisation des services de sant6 dans le secteur temoin a peu vari6, mais la proportion de cas de paludisme ayant re,u un traitement a diminu6 de 8%. Dans le sec- teur d'intervention, on a observ6 une augmentation de l'utilisation de la chloroquine comme traitement de premiere intention et une baisse de l'utilisation de la quinine (de 15% a 7%). Cette derniere restait largement utilis6e dans le secteur t6moin. Cette 6tude a permis d'identifier certains problemes institu- tionnels, sociaux et financiers susceptibles de com- promettre la viabilite de l'utilisation des agents de sante communautaires pour les activit6s de lutte antipaludique. Ces problemes tiennent au fait que les agents de sant6 n'assurent pas la totalit6 des soins et qu'ils occupent une position ambigue, a mi- chemin entre le systeme de sant6 et la communaut6. References 1. Malaria control through primary health care in Africa. Bulletin of the World Health Organization, 1984, 62: 411-413. WHO Bulletin OMS. Vol 74 1996 429 C. Delacollette et al. 2. Implementation of the global malaria control strategy. Report of a WHO Study Group on the Implementation of the Global Plan ofAction for Malaria Control: 1993- 2000. Geneva, World Health Organization, 1993 (WHO Technical Report Series, No. 839). 3. WHO Expert Committee on Malaria. Seventeenth re- port. Geneva, World Health Organization, 1979 (WHO Technical Report Series, No. 640). 4. Malaria control as a part ofprimary health care: report of a WHO study group. Geneva, World Health Organi- zation, 1984 (WHO Technical Report Series, No. 712). 5. Wils W et al. Le Kivu montagneux. Bruxelles, Academie royale des Sciences d'Outre-Mer. Classe des Sciences naturelles et medicales. (Memoires, nouvelles series, tome XXI, fascicule 3, 1986). 6. Delacollette C et al. Demographic findings relevant for health care planning and evaluation collected through a malaria control project in the Kivu moun- tains, Zaire. Acta tropica, 1992, 52: 189-199. 7. Delacollette C et al. Etude de la mortalite globale et de la mortalite li6e au paludisme dans le Kivu montagneux, Zaire. Revue d'6pid6miologie et sant6 publique, 1989, 37: 161-166. 8. Wery M et al. Evolution de la sensibilit6 de P. falciparum a la chloroquine, a la quinine et a la m6floquine entre 1983 et 1985 au Zaire. Annales de la Societe belge de M6decine tropicale, 1986, 66: 309-324. 9. Delacollette C et al. Indices paludometriques selon l'age et selon les saisons dans la zone de sante de Katana, Kivu montagneux, Zaire. Annales de la Societe belge de M6decine tropicale, 1990, 70: 263- 268. 10. Delacollette C et al. Compared response to chloroquine, Fansidar, quinine and quinidine slow re- lease of infections with Plasmodium falciparum in the Kivu region of Zaire. Acta Leidensia, 1987, 55: 159- 162. 11. Greenwood BM et al. Mortality and morbidity from malaria among children in a rural area of The Gambia, West Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1987, 81: 478-486. 12. Delacollette C, Van der Stuyft P. Developpement d'une methode simple et fiable pour estimer la morbidite palustre a partir du modble de Meunch modifi6. Revue d'6pidemiologie et sant6 publique, 1993, 41: 416-421. 13. Dabis F et al. Monitoring selective components of primary health care: methodology and community as- sessment of vaccination, diarrhoea, and malaria prac- tices in Conakry, Guinea. Bulletin of the World Health Organization, 1989, 67: 675-684. 14. Mulumba MP et al. Le paludisme de l'enfant a Kin- shasa (Zaire). Influence des saisons, de l'age, de 1'environnement et du standing familial. M6decine tropicale, 1990, 50: 53-64. 15. Deming M et al. Home treatment of febrile children with antimalarial drugs in Togo. Bulletin of the World Health Organization, 1989, 67: 695-700. 16. Lovel H. Targeted interventions and infant mortality. Transactions of the Royal Society of Tropical Medi- cine and Hygiene, 1989, 83: 10-18. 17. Spencer HC et al. Impact on mortality and fertility of a community based malaria control programme in Saradidi, Kenya. Annals of tropical medicine and parasitology, 1987, 81 (suppl. 1): 36-45. 18. Greenwood BM et al. Comparison of two strategies for control of malaria within a Primary Health Care Programme in the Gambia. Lancet, 1988, i: 1121- 1127. 19. Smiths PG, Morrow RH, eds. A toolbox for field trials of interventions against tropical diseases. New York, Oxford University Press, 1991. 20. Alonso PL et al. The accuracy of clinical histories given by mothers of seriously ill African children. An- nals of tropical paediatrics, 1987, 7: 187-189. 21. Snow RW et al. Childhood deaths in Africa: uses and limitations of verbal autopsies. Lancet, 1992, 340: 351-355. 22. Chen LC. Primary health care in developing coun- tries: overcoming operational, technical and social barriers. Lancet, 1986, 2: 1260-1265. 23. Haegeman F et al. Malaria control by Village Health Workers in the province of Dosso, Niger. Part 1. Op- erational analysis. Annales de la Societ6 belge de M6decine tropicale, 1985, 65: 137-144. 24. MacCormack CP, Lwihula G. Failure to participate in a malaria chemosuppression programme, Nord Mara, Tanzania. Joumal of tropical medicine and hygiene, 1983, 86: 99-107. 25. Jancloes MF. Towards primary health self-care edu- cation without using village health workers. Acad6mie royale des Sciences d'Outre-Mer, 1987, 9: 83-91. 26. Walt G ed. Community health workers in national programmes: just another pair of hands? Bucking- hamshire, England, Open University Press, 1990. 27. Mercenier P. L'agent de sant6 communautaire. Ant- werp, Institute for Tropical Medicine, 1984 (ITG Work- ing Paper, No. 9). 430 WHO Bulletin OMS. Vol 74 1996