Chagas Disease Control Programme in Brazil: a study of the effectiveness of 13 years of intervention.

Chagas Disease Control Programme in Brazil: a study of the effectiveness of 13 years of intervention* F.C. Costa,1 R.W.A. Vitor,2 C.M.F. Antunes,3 & M. Carneiro2 Reported is an evaluation of 13 years of intervention by the Chagas Disease Control Programme in an endemic area (Montalvania) in the State of Minas Gerais, Brazil. The design used was an epidemiological panel study. The reduction ofTrypanosoma cruzi infection rates was estimated from data collected on three separate occasions: a serological survey in 1975-80, a quasi-experimental study in 1987, and the present investigation. A random sample of 156 households was selected and blood samples were collected from 653 inhabitants. The data routinely collected by the control programme were analysed to correlate the results with the incidence of T. cruzi. The overall prevalence of infection was 2.3%; however, no participant under 14 years of age was found to have a positive serological test. The total reduction in T. cruzi infection rates in this area from the start of the programme's activities was estimated to be 83.5%. Cross-sectional comparisons for the age groups 2-6 years and 7-14 years indicated a 100% reduction in T. cruzi incidence rates; but cohort comparisons showed that 100% reduction was achieved only for the 2-6-years age group. Introduction Public health services and programmes can be evalu- ated by different theoretical and methodological approaches which take into account the triad of structure-process-results (1). Since structure and process can rapidly be assessed, the evaluation con- clusions can provide important information to pro- gramme administrators. On the other hand, only the evaluation of results can identify changes in the mor- tality and morbidity of populations as a result of the programme's intervention (2, 3). Evaluation of the impact of the programme requires a considerable investment of time in order to observe, analyse and measure the changes in the health indicators being studied (2). Epidemiological studies are often used for evaluating the impact (results) of health pro- grammes, as well as estimating their efficacy and effectiveness, after controlling for extrinsic or con- founding factors and identifying other changes (4-6). The actions of the Chagas Disease Control Programme in Brazil are based on house spraying * From the Department of Parasitology, Institute of Biological Sciences, Federal University of Minas Gerais, Caixa Postal 486, 31270-901-Belo Horizonte, MG, Brazil. Requests for reprints should be sent to Dr M. Carneiro at this address. Doctoral student (Doutoranda). 2 Assistant Professor. 3 Professor. Reprint No. 5874 with insecticide, the objective being to control the Triatoma bug population (intermediate host of Trypanosoma cruzi, the etiological agent of Chagas disease). Until now, the assessment of the pro- gramme's impact has been mainly based on entomo- logical indicators, i.e. measuring the reduction of the household triatomine population. The results have shown success in controlling Triatoma infestans, the main vector of T. cruzi in Brazil. However, the re- placement of this species by secondary ones in household infestation, which could lead to a possible recrudescence in T. cruzi infection rates, has focused attention on this type of habitat in the transmission of Chagas disease (7, 8) and emphasized the need to evaluate the programme's activities in terms of a reduction in transmission rates. Hence, in addition to the entomological results, serological evaluations of selected cohorts have shown a reduction of T. cruzi infection among those who were born after the start of the programme (9-12). Seroprevalence of Chagas disease in this population is considered to be a sensi- tive indicator of triatomine household infestation rates, capable of estimating the risk of parasite trans- mission in the community (3). A quasi-experimental epidemiological study was conducted in Minas Gerais State in 1987, with the objective of evaluating the effectiveness of the programme. Areas with interventions lasting 5 and 10 years were compared to a control area with no intervention. A comparison of the infection rates found in the study areas, among those born after the Bulletin of the World Health Organization, 1998, 76 (4): 385-391 X) World Health Organization 1998 385 F.C. Costa et al. programme's implementation, with those published by Chagas Disease Serological Survey showed a de- crease in T. cruzi infection rates (cohort of 2-6-year- olds) in the area with 10 years of intervention. However, in the area with only 5 years of interven- tion (Montalvania), there was an increase in the T. cruzi infection rates in a similar cohort. The authors suggested that 5 years of programme activities were not sufficient to induce changes in the infection rates (13, 14). The present study was conducted in Montalvania, Minas Gerais State, in 1995 after 13 years of interventions under the Chagas Disease Control Programme. The objective was to investi- gate (1) the T. cruzi infection transmission pattern in a cohort born after the intervention, and (2) whether the duration of the programme was a determinant factor in inducing changes in the incidences. The panel study was the design chosen in order to evalu- ate the efficacy of the control programme, a series of cross-sectional studies being carried out in the same population at various time intervals (15). The results of this study were compared with two previous inves- tigations conducted in this area: the Chagas Disease National Serological Survey (1975-1980) (16) and the quasi-experimental study conducted in 1987 (13). Materials and methods Study population. This investigation was carried out in the rural area of Montalvania, situated in the north-east of Minas Gerais State, south-east Brazil. The necessary sample size (500) was estimated by assuming that a 50% reduction in T. cruzi infection rate would have occurred as a result of the pro- gramme, and accepting as a baseline the rates derived in 1987 (13), with a = 0.05 and 1 - i = 0.90 (17). The sampling unit was the household, identified through the control programme's records. A two- stage sampling scheme was used: a simple random sample of communities within the study counties; and a stratified random sample of houses within each selected community. The following criteria were used in order to obtain better representation of rural households: all houses in communities with less than 10 houses, 50% in communities with 11-30 houses, 30% in communities with 31-60 houses, and 10% in communities with >60 houses (13). All the inhabit- ants of the selected houses were eligible to be in- cluded in the study. Data collection and laboratory test. The study proce- dures were approved by the ethical committee of the Institute of Biological Sciences of the Federal University of Minas Gerais. Before data collection, the informed consent of all participants was ob- tained. The member responsible for the family was then interviewed using a pre-coded questionnaire developed specially for this study. The information collected included not only the respondent, but all family members participating in the survey. The following information was obtained: demographic and social characteristics, knowledge about Chagas disease, household characteristics including all an- nexes and buildings, any house improvements made, risk factors associated with T. cruzi infection, and factors related to the activities of the control pro- gramme. To guarantee reliability in the data collec- tion, all interviews were made by the same interviewer using an instruction manual. At the interview, a blood sample was collected from all the participants by fingerprick onto filter- paper, according to standard procedures defined by the Brazilian National Health Foundation (18). After drying at room temperature, the filter-papers were stored at 4°C before being sent for testing to the serology laboratory at the Department of Parasitology, Federal University of Minas Gerais. Eluates from the filter-papers were tested simultane- ously and blindly for anti-T cruzi antibodies using indirect immunofluorescence assay (IFA) and enzyme-linked immunosorbent assay (ELISA), as previously described (19). The indirect haemagglu- tination (IHA) test (20) was used when discordant results were obtained. A sample was considered positive when at least two serological tests were reac- tive: IFA 2 1: 40 and IHA 2 1: 40; the cut-off point for the ELISA test was the average of negative con- trol titres plus 3 standard deviations (21). Blood sam- ples were tested up to 60 days after collection. Data on the presence of triatomine bugs in this area were abstracted from the files of the Chagas Disease Control Programme. Data processing and analysis. The data were ana- lysed using Epi Info software (22) and the following were determined: frequency distribution of collected variables; estimate of the T. cruzi prevalence in the area; T. cruzi incidences, stratified by age groups; correlation between the results obtained in this study and those on triatomine infestation collected by the Chagas Disease Control Programme; the control programme's efficacy, by comparing the T. cruzi in- fection rates from this investigation with those ob- tained in the national survey conducted in 1975-1980 and those obtained by the programme evaluation conducted in 1987 (13). Using a relative index, the overall T. cruzi prevalences and incidences, by age group, were compared between the three studies. For this last analysis, both cross-sectional compari- WHO Bulletin OMS. Vol 76 1998386 Chagas Disease Control Programme in Brazil sons and cohort analysis were performed for the age groups 2-6 and 7-14 years. The age group compari- sons were carried out with the same age group struc- ture used in the National Serological Survey. Table 1: Numbers sampled and examined in Montalvania, Minas Gerais State, 1995 No. in sample No. in Montalvania Selected communities 16 150 (1 0.7)a Households interviewed 156 5623 (2.8) Study population 655 8033b (8.2) Blood samples examined 653 8033b (8.1) a Figures in parentheses are percentages. bRural population. Results Table 1 shows the distributions of the saml cedures; 156 households were selected a viewed, and blood samples were obtained participants out of 655 in the study (99.70/a The characteristics of the study popul be seen in Table 2. Most participants had i tion or incomplete elementary school edu Table 2: Distribution of the demographic studied in Montalvania, Minas Gerais State, Sex: Male Female Ethnic group: Non-White White Education: None Elementary (incomplete) Elementary (incomplete and cursory Elementary (complete) High school (incomplete) High school Birth place: Montalvania Another county (MG) Another state Blood transfusion pling pro- Lnd inter- Table 3: Distribution of housing characteristics in the study population, Montalvania, Minas Gerais State, 19958 Number/mean Number of houses studied: Duration of residence (mean, years) Persons per house (mean) Rooms per house (mean) Bedrooms per house (mean) Persons per bedroom (mean) Type of walls: Brick Mud Adobe Wood and straw Adobe and brick Type of roof: Tile Zinc Straw Others Type of floor: Tile or ceramic Cement Dirt Others from 653 Clothes storage: Box J-). Suspendedlation can Bags no educa- Objects on the walls: Ication. A Frame Calendars Supplies: Firewood variables Food storage 19958 Animals indoors: Dog Number Cat Chicken Type of annex: 325 (49.6)b Chicken house 330 (50.4) Pigsty For pile of wood 510 (77.9) For tiles 145 (22.1) For bricksFor waste 140 (26.0) 363 (65.2) 40 (7.2) 1 (0.2) 1 (0.2) 7 (1.5) 552 (84.3) 47 (7.2) 56 (8.5) 11 (1.7) 156 (2.8)b 11.9 5.2 5.3 2.9 2.1 43 (27.6) 56 (35.9) 44 (28.1) 2 (1.3) 6 (3.9) 112 (71.8) 2 (1.3) 7 (4.5) 35 (22.4) 3(1.9) 37 (23.7) 105 (67.3) 11 (7.1) 55 (35.3) 82 (52.6) 101 (64.7) 104 (66.7) 100 (64.1) 37 (23.7) 117 (75.0) 110 (70.5) 89 (57.1) 148 (94.9) 125 (80.1) 98 (62.8) 50 (32.1) 34 (21.8) 36 (23.1) 11 (7.1) a Data for refusals, "not done", and "don't know" are not shown. b Figures in parentheses are percentages. high proportion (84.3%) were born in Montalvania. Only 1.7% of the subjects reported ever receiving a blood transfusion. Distribution of the variables used to character- ize the houses and their occupants are shown in Table 3. The average time of residence was 11.9 years. Higher proportions of houses were found to have mud walls, dirt floors and tile roofs. Food stor- age areas, woodpiles, and pictures on the wall, as WHO Bulletin OMS. Vol 76 1998 a Data for refusals, "not done" and "don't know" are not shown. The mean age of study participants was 25.5 years. b Figures in parentheses are percentages. 387 F.C. Costa et al. Table 4: Proportion of positive serological tests for Trypanosoma cruzi in Montalvania, Minas Gerais State, 19958 Number Blood samples tested 653 (99.7)a IFA:b Positive 13 (2) Negative 640 (98) ELISA:b Positive 15 (2.3) Negative 638 (97.7) Positive in two tests 15 (2.3) a Figures in parentheses are percentages. b IFA = indirect immunofluorescence antibody test. ELISA enzyme-linked immunosorbent assay. The indirect haem- agglutination test (IHA) was performed for four discordant tests; three of them were positive. well as the presence of chickens, dogs and cats inside the dwellings were also reported for most of the houses. Table 4 shows the results of the serological tests. A total of 15 participants had two positive serolo- gical tests for T. cruzi infection and the overall prevalence was estimated as 2.3%. Among four dis- cordant results between IFA and ELISA, three were confirmed positive by IHA. The overall prevalences obtained in this study and reported previously from this area are shown in Table 5. The differences are expressed as relative reduction in the infection rates; the total reduction during the thirteen years of the control programme was estimated as 83.5%. Age-specific infection rates were determined using the same age groups as published in the National Survey of 1975-80. Table 6 compares the present results with those from the evaluation con- ducted in 1987 (prevalences were not available for the national survey). Reductions in the T cruzi posi- tive seroprevalences, in both cross-sectional and co- Table 5: Overall prevalence and relative reduction of Trypanosoma cruzi infection observed in Montalvania, Minas Gerais State, 1995 Prevalence Relative Total relative Studies (%) reduction (%) reduction (%) National survey 13.9 - - (1975-80) Evaluation in 1987 7.8 43.9 - Evaluation in 1995 2.3 70.5 83.5 Table 6: Trypanosoma cruzi prevalence, by age group, Montalvania, Minas Gerais State, 1987 and 1995 % prevalence in: Age group (years) 1987 1995 1-6 2.5 0 7-14 10.9 0 15-29 8.8 3.1 30-44 5.3 6.6 --45 3.9 2.5 hort analysis, for almost all age groups were ob- served. Prevalences were zero in 1995 for the 1-6- and 7-14-year age groups. Table 7 shows the percentage distribution of T. cruzi positive serology, by age group, between the three studies conducted in the area. Whereas an increase was observed in the 2- 6-year-old cohort between the National Survey and the 1987 evaluation, the same comparison between 1987 and present evaluations indicates an interrup- tion in parasite transmission. A marked reduction was also observed for the 7-14-year cohort. These differences and changes, which were used to estimate the effectiveness of the control pro- gramme, are shown in Table 8. The cross-sectional analysis suggested a reduction of 100% for both the 2-6 and 7-14-year age groups; the cohort analysis showed a 100% reduction only for the 2-6-year age group. The entomological indicators (data from the Control Programme) show that large numbers of T. infestans were captured at the start of the interven- tion; 76.3% of specimens were collected, 57.4% in the surrounding areas. The replacement of this species by T. sordida occurred during the second year of the programme's activities. While percent- ages of 3.9% and 1.5% were obtained, respectively, indoors and in the areas surrounding the houses at the start of the control programme, current reports Table 7: Trypanosoma cruzi infection rates, by age group, in Montalvania, Minas Gerais State, 1975-80, 1987 and 1995 Percentage infection rates Age group National survey Evaluation Evaluation (years) in 1975-80 in 1987 in 1995 2-6 1 5.5 5.7 0.0 7-14 26.6 54.7-___ 0.0 15-29 20.0 22.6 26.6 30-44 20.0 11.3 53.3 >45 17.7 5.7 20.0 WHO Bulletin OMS. Vol 76 1998388 Chagas Disease Control Programme in Brazil Table 8: Relative changes in Trypanosoma Montalvania, Minas Gerais State cruzi infection rates in Comparisons Age group % positive % change Cross-sectionat 2-6 years National Survey (1975-80) 15.5 Evaluation in 1987 5.7 63.2.1 Evaluation in 1995 0.0 100.01 Cross-sectionat 7-14 years National Survey (1975-80) 26.6 Evaluation in 1987 54.7 105.61 Evaluation in 1995 0.0 100.01. Cohort for 2-6-year-olds: National Survey (1975-80) 2-6 years 15.5 Evaluation in 1987 7-14 years 54.7 248.41 Evaluation in 1987 2-6 years 5.7 Evaluation in 1995 7-14 years 0.0 100.04 indicate 12.6% and 13.7%. T. infestans has not been observed in this area. Discussion The results of the present investigation, compared with those obtained in the previous studies in the same area, show that the control programme's effec- tiveness reached 100% for the age groups 2-6 and 7- 14 years. No participants in these age groups, who were born after the control programme started, were identified as infected by T cruzi. The reduction was confirmed by both cross-sectional analysis (age groups 2-6 and 7-14 years) and cohort analysis (age group 2-6 years). The total decrease observed in infection prevalence, estimated in terms of relative reduction, was 83.5%, when compared to the results of the first investigation. In addition, the 2.3% preva- lence found in this study reflects an infection already existing among those born before the intervention programme. It should be noted that although 2.5% of 2-6-year-olds were found to be positive in the 1987 evaluation, no infection was detected in the age group 7-14 years in the present study; a significant reduction in T. cruzi infection rates was also ob- served in the age group 15-29 years. The most plau- sible explanation for these findings is the high migration rate to urban centres among young adults searching for better living conditions (23). This investigation also demonstrated that, for this area, 13 years of control interventions were necessary to induce detectable changes in the trans- mission of T. cruzi infection. As an earlier evalua- tion in this area had concluded that 5 years of intervention were not enough to detect changes in the infection rates, an important question arises: for how long should the programme be active in order to induce changes in T. cruzi infection rates in areas similar to Montalvania? It should be empha- sized that this area had a very high proportion of houses infested with triatomine bugs at the start of the control programme. Even today, the proportion of houses with suitable conditions for triatomine colonization is still very high. The observed replace- ment of domiciled T. infestans by secondary vector species in household environs highlights the need for continuing surveillance to prevent recrudescence of T. cruzi transmission. The effectiveness of health interventions has traditionally been evaluated using the experimental or quasi-experimental design (2, 5, 6). The main ad- vantage of an experimental study is the absolute control exerted by the investigator on the study factor and the random allocation of subjects to guar- antee the comparability of the study groups, while ensuring the absence of confounding factors. The quasi-experimental design, which also permits con- trol of the study factor but without the random allo- cation of subjects, has been proposed as the method of choice to evaluate public health programmes, especially those affecting large populations (24). Only recently, observational studies have been sug- gested as an alternative method for evaluating health interventions; the major advantages are the reduced cost and the ease of execution (6, 25). Comparison between areas is crucial for impact evaluation; only by comparing areas with and without the intervention can the observed changes in health indicators be associated with the pro- gramme's actions. However, in some situations, a comparison area is not available. In 1995, for WHO Bulletin OMS. Vol 76 1998 389 F.C. Costa et al. example, all the endemic areas in the Minas Gerais State were under the Chagas Disease Control Pro- gramme's interventions so that external comparison could not be carried out; an evaluation was possible only by an internal comparison, i.e. studies con- ducted in the same area at different time intervals (6). In the present investigation, the epidemiological design of choice was the panel study which may be considered as a series of cross-sectional studies, car- ried out in the same population at various time inter- vals; unlike conventional cross-sectional studies, longitudinal measurements of individuals or samples are obtained. The design is not affected by absence of the disease at the start of the investigation, as in cohort studies (15, 26). Because time trends can be determinative, changes in the T. cruzi incidence can be attributed to the intervention programme. The data analysis was designed to identify any distortion that may have occurred during data collec- tion and in the execution of the diagnostic test used in identifying the infection. The data collection methodology was similar to that used in the 1987 evaluation (13). This approach assured high internal comparability. Besides, the interviews were con- ducted by the same interviewer, using a pre-coded questionnaire and an instruction manual specially developed for this study. The serological tests were carried out, blindly and simultaneously, using two different techniques. Any discordant result was con- firmed by a third method. The construction and internal organization of dwellings were considered to be the main deter- minant factors for triatomine colonization and consequently T. cruzi human infections (27, 28). This study identified a high proportion of mud houses ("cafuas"), with dirt floors and tile roofs. Food storage, woodpiles, pictures on the wall, and animals were also reported in the houses. A previous investigation has shown a risk of 2.5 for the presence of T. infestans associated with these characteristics (29). In the peridomestic area, the presence of a chicken house, pigsty and woodpile was reported for most of the dwellings; these charac- teristics are also associated with the presence of triatomines (30). The households and their surroundings in this investigation were similar to those previously re- ported from this area (13), which suggests that, during this period, no significant changes related to house improvement were made. Montalvania can still be considered as a high-risk area for triatomine colonization (28). After control of domiciled T infestans, the area was quickly re-infected by other triatomine species which are considered not to be important in T. cruzi transmission. This fact empha- sizes the need for active and continuing surveillance, considering both triatomine presence and T. cruzi infection. The above results demonstrate the adequacy of the panel design in evaluating the effectiveness of the Chagas Disease Control Programme to interrupt parasite transmission by vectors in this area and to reduce by 100% the incidence rate of infection. It was also shown that the duration of the programme's activities, to be effective, will depend on the baseline characteristics of the area. Acknowledgements This investigation was partly supported by Conselho Na- tional de Desenvolvimento Cientifico e Tecnol6gico (CNPq) and Fundagao de Amparo a Pesquisas do Estado de Minas Gerais (FAPEMIG). We are grateful to Fundaq&o Nacional de Sa6de-Coordenagao do Estado de Minas Gerais for their cooperation in the field work. Resume Programme de lutte contre la maladie de Chagas au Bresil: etude de l'efficacite des 13 annees d'intervention Cet article decrit l'6valuation des 13 ann6es d'intervention conduites par le programme de lutte contre la maladie de Chagas dans l'Etat de Minas Gerais au Bresil. L'6tude a consiste en une serie d'enquetes transversales r6petees sur la meme population. La baisse de fr6quence des infections a Trypanosoma cruzi a ete estim6e a partir des donn6es recueillies a trois reprises:une enquete s6rologique r6alis6e en 1975-1980, une 6tude quasi-exp6rimentale conduite en 1987 et la presente 6tude. Un echantillon al6atoire de 156 m6nages a ete selectionn6 et des pr6levements de sang ont ete recueillis chez 653 habitants. Les donnees enregistrees en routine par le programme de lutte ont ete analysees et les r6sultats correles a l'incidence de T. cruzi. La pr6valence globale de l'infection etait de 2,3%; aucun participant de moins de 14 ans n'6tait seropositif pour la maladie de Chagas. On estime que la frequence des infections a T. cruzi a diminue au total de 83,5% dans ce secteur depuis le lancement des activit6s du programme. La comparaison des enquetes transversales dans les classes d'age 2-6 ans et 7-14 ans montre une baisse de 100% des taux d'incidence de l'infection a T. cruzi; la comparaison des cohortes revele toutefois que la diminution de 100% n'est atteinte que dans la classe des 2-6 ans. WHO Bulletin OMS. Vol 76 1998390 Chagas Disease Control Programme in Brazil References 1. Donabedian A. Basic approaches to assessment: structure, process and outcome. In: Explorations in quality assessment and monitoring, Vol. I. Ann Arbor, Health Administration Press, 1980: 77-125. 2. Holland WW. Evaluation of health care. Oxford, Ox- ford University Press, 1984. 3. Health programme evaluation. Guiding principles for its application in the managerial process for national health development Geneva, World Health Organiza- tion, 1981 (Health for All Series, No. 6). 4. Ibrahim MA. Epidemiology: application to health services. The journal of health administration educa- tion, 1983, 1(1): 37-69. 5. Ibrahim MA. Epidemiology and health policy. Rockville, MD, Aspen Publications, 1985. 6. Omran AR. Health systems research: approaches and pitfalls. World health forum, 1990, 11: 286- 292. 7. Dias JCP, Loyola CCP, Brener Z. [Chagas disease in Minas Gerais. Current situation and perspective]. Revista Brasileira de Malariologia e Doen,as Tropicais, 1985, 37: 7-28 (in Portuguese). 8. Silveira AC, de Rezende DF. [Epidemiology and control of vector transmission of Chagas disease in Brazil]. Revista da Sociedade Brasileira de Medicina Tropical, 1994, 27 (suppl. 3): 11-22 (in Portuguese). 9. Coura JR et al. [Chagas disease morbidity 11 - Cross sectional studies in four areas of Campo no Brasil]. Mem6rias do Instituto Oswaldo Cruz, 1984, 79: 101- 124 (in Portuguese). 10. Mota EA et al. A nine-year prospective study of Chagas' disease in a defined rural population in north- east Brazil. American journal of tropical medicine and hygiene, 1990, 42: 429-440. 11. de Andrade ASSG et al. Surveillance of T. cruzi transmission by serological screening of school- children. Bulletin of the World Health Organization, 1992, 70: 625-629. 12. Diotaiuti L et al. Evaluation of the Chagas disease vector control program in Minas Gerais, Brazil, with special reference to Triatoma sordida. Bulletin of the Pan American Health Organization, 1994, 28: 211- 219. 13. Carneiro M, Antunes CM. A quasi-experimental model for evaluating public health programmes: effi- cacy of Chagas disease control programme in Brazil. Bulletin of the World Health Organization, 1994, 52: 721-728. 14. Carneiro M, Antunes CM. [Evaluation of the efficacy of the Chagas disease control program: methodologi- cal aspects.] Cadernos de saude publica, 1994, 10 (suppl. 2): 261-272 (in Portuguese). 15. Kelsey JL et al. Methods in observational epidemiol- ogy, 2nd edit. New York, Oxford University Press, 1996. 16. Camargo ME et al. [Serological investigation of the prevalence of Chagas' disease in Brazil, 1975-1980.] Revista do Instituto de Medicina Tropical de Sao Paulo, 1984, 26: 192-204 (in Portuguese). 17. Armitage P, Berry G. Statistical methods in medical research, 2nd ed. London, Cambridge University Press, 1987. 18. Ministry of Health - National Health Foundation. [Control of Chagas' disease: technical guidelines. 1 st ed.] Brasilia, Fundag5o Nacional de Saude, 1994 (in Portuguese). 19. Machado-Coelho GL et al. Validity of serology for American trypanosomiasis with eluates from filter paper. Mem6rias do Instituto Oswaldo Cruz, 1995, 90: 59-64. 20. Vitor RWA, Chiari E. [Determination of Trypanosoma cruzi antigens by an indirect haemagglutination reac- tion. Different antigenic extracts]. Revista do Instituto de Medicina Tropical de Sao Paulo, 1987, 29: 178- 182 (in Portuguese). 21. Richardson MD et al. Computer-assisted rapid enzyme-linked immunosorbent assay (ELISA) in the serological diagnosis of aspergillosis. Journal of im- munological methods, 1983, 56: 201. 22. Dean AG et al. Epi Info, Version 6. A word-processing database and statistics program for epidemiology on microcomputers. Atlanta, GA, Centers for Disease Control and Prevention, 1994. 23. Pereira MG. [Epidemiology: theory and practice.] Rio de Janeiro, Editora Guanabara Koogan, 1995: 235 (in Portuguese). 24. Kleinbaum DG et al. Epidemiologic research: princi- ples and quantitative methods. New York, van Nostrand Reinhold, 1982. 25. Briscoe J et al. Evaluating health impact: water supply, sanitation, and hygiene education. Ottawa, International Development Research Centre, 1986. 26. Last MJ. A dictionary of epidemiology. 3rd ed. New York, Oxford University Press, 1988. 27. Bricefio-Le6n R. La Casa Enferma - Sociologia de la Enfermedad de Chagas [The "sick" house - soci- ology of Chagas disease], 1 edit. Caracas, Fondo Editorial Acta Cientifica Venezolana - Cons6rcio de Ediciones Capriles CA., 1990 (in Spanish). 28. Control of Chagas disease. Report of a WHO Expert Committee. Geneva, World Health Organization, 1991 (WHO Technical Report Series No. 81 1). 29. de Andrade ASSG et al. Evaluation of risk factors for house infestation by Triatoma infestans in Brazil. American joumal of tropical medicine and hygiene, 1995, 53: 443-447. 30. Diotaiuti L et al. [Characterization of rural peridomestic areas of the municipality of Porteirinha, north of Minas Gerais, and its relation to infestation with and control of Triatoma sordida.] Revista da Sociedade Brasileira de Medicina Tropical, 1994, 27 (suppl. II): 70-72 (in Portuguese). WHO Bulletin OMS. Vol 76 1998 391