A Quasi-Experimental Epidemiological Model for Evaluating Public Health Programmes: Efficacy of a Chagas Disease Control Programme in Brazil
Carneiro, M., Antunes, C. M. F., Bulletin of the World Health Organization
Public health programmes are traditionally evaluated in terms of their structures and processes since these can be assessed rapidly and can quickly provide answers to programme managers. However, only the evaluation of results can identify any changes in the morbidity and/or mortality rates in the population covered by the programme and. associate programme actions with the observed changes (1-3).
The evaluation of results requires the use of epidemiological methods and, of the various designs, quasi-experimental studies are the method of choice for health impact evaluation (3, 4). Such studies have the following major advantages:
-- they can be carried out by the health agency using data collected by the health programme;
-- they can be used to conduct large studies at reasonable cost;
-- they are ethical since they can be implemented together with the health programme; and
-- occasionally they may be the only available alternative.
On the other hand, the lack of randomization that characterizes a quasi-experimental study means that investigators have less control over the influence of possible confounding factors (4).
We have developed a quasi-experimental model to evaluate the impact of the actions of the Brazilian Chagas Disease Control Programme on the incidence of this condition. The activities of the programme --house spraying with insecticide--are aimed at controlling the triatomine bug, intermediate host of Trypanosoma cruzi, the etiological agent of Chagas disease. Evaluation of the programme has been based on entomological indicators, i.e., the elimination of triatomine populations. However, in addition to entomological indicators, vector control programmes should include serological and clinical evaluations to determine the effectiveness of their activities and to measure the duration of their effect and hence gauge the frequency of application of specific measures (5).
In order of evaluate the efficacy of the control programme, we compared the incidence of Chagas disease estimated using serological tests with the incidence of the disease among children born after the start of the intervention activities in areas with the same or different intervention times; and compared the results observed in the present study with those of the Brazilian Chagas Disease Serological Survey conducted 10 years previously.
The evaluation model attempted to use data collected by the health programme whenever possible; provide prompt answers; reduce the programme's operational costs, and be feasible to carry out by the health agency, so that it could be incorporated into the programme routine.
Materials and methods
The investigation was carried out in the rural northern part of Minas Gerais State, in southern Brazil. The three selected counties presented similar economic, social, and cultural characteristics: Triatoma infestans was identified by the control programme as the most frequent domiciliated vector of T. cruzi. These counties were selected because they had been exposed to the Chagas Disease Control Programme activities for different periods of time: 10 years (intervention-10 area); 5 years (intervention-5 area); and without intervention (control area), but with baseline data already collected.
The sample size (n=500 in each county) was estimated assuming that there had been a 50% reduction in T. cruzi infection in the areas under intervention, compared with the results obtained in the national survey (6) for the age group 2-14 years, accepting the following errors: [alpha] = 0.5 and 1-[beta] = 0.90 (7).
The sampling unit was the household, identified through the control programme records. A two-stage sampling scheme was used: (1) a simple random sample of communities within the study counties; and (2) a stratified random sample of houses within each selected community. …