Ch. 3, 4. Susser: Causal thinking in the health Sciences. pp. 26-47.
Ch. 2. MacMahon and Pugh. Epidemiology, principles and methods. pp. 17-27.
Ch. 2 (pp. l4-17)
Ch.10 (pp. 168-171) Text
Ch. 13(pp. 202-206)
An important concept in epidemiology is the premise that multiple determinants affect diseases in populations. A determinant as described in an earlier lecture could be a specific agent; a host factor, or an environmental factor which directly or indirectly increases the frequency of occurrence of a disease. It follows then that if specific determinants of a disease could be identified and the relative importance of each such determinant is known, preferably quantitatively, then, directing efforts at the elimination or reduction of the causal factors would lead to a decline of the disease in the population. Therefore, the prevention and control of a disease depends on the proper determination and analysis of its determinants.
Infectious agents have received much attention as determinants of disease and the concept of one disease - one causal agent is still strong in medical circles. Koch's postulates for establishing causal relationships of an infectious agent and a disease should be reviewed. (p. 168 Text).
The emphasis laid on the search for single causes has been detrimental to the concept of multiple causation. Such a limited view of causality has now been recognized as being counter productive in the prevention and control of diseases in populations. The web of causation that determines the patterns of disease occurrence to include interactions between host, agent and environment is central in today's epidemiological studies.
II. Methods of identifying determinants of a disease:
Basically, this requires a detailed understanding of the epidemiology of the specific disease of interest.
1. Identify and list all factors of causal importance considering the host species, agent and environmental factors. Items like the life cycles of the agent characteristics of the agent, incubation period, host factors (age, sex, breed, race, immune status) and environmental determinants (climate, season, housing, nutrition, management, etc.) should be considered.
2. Prepare a diagram to assist you in conceptualizing and visualizing the web of causation (and the determinants listed) that eventually determines the epidemiology of the disease of interest.
III C. Analysis
Once the variables of significance are listed and a causal diagram prepared, an appropriate analytic methodology should be selected to provide predictive and/or descriptive values to establish the relative importance of each variable. Based on such analysis, appropriate plans could be devised to recommend and implement preventive or disease control measures.
Epidemiologic dynamics involves a diverse array of multicausal factors of host, agent and environmental attributes directly or indirectly affect the states of health/ill health. The magnitude of such influences varies by time and place. Thus an understanding of multicausality is an essential aspect of epidemiologic dynamics.
From such a framework a more holistic picture evolves and one then tries to decipher, explain,or understand the patterns of occurrence and distributions of health/ill health in populations. Note that this is a systems concept; holistic, systematic, analytic and purposeful.
Epidemiology as referred to in Chapter 1, is conceptually goal oriented and purposeful. As such, it focuses on three distinct objectives, viz. therapeutic, preventive, health maintenance and promotion.These goals of epidemiology ultimately provide us with the framework for evaluating various disease control alternatives. The diagrammatic models facilitate the quantitative and mathematical analytic steps.
Since we are focussing on epidemiologic modelling at this stage, two major classes of epidemiologic models could be identified.
a) static and
b) dynamic models.
Static models are cross-sectional and invariant (still picture) representations of health processes. Two types of static models have potential use in health programming. The first ones are those of multivariate nature whose mathematical basis are consistent with the concept of the multifactorial nature of epidemiologic problems. Such models are pertinent in causal analysis and in the selection of the most significant variables which would have impact on decreasing prevalence or promoting the healthy state.
The second type of static models are optimization techniques, more specifically linear programming; problem solving tools that enable one to identify optimal solutions to disease control tasks from a number of alternatives in the presence of constraints on available resources, e.g. economics, expertise, time, etc.
Since we are dealing with multiple determinants of disease, one of the most useful analytic tools used in static models are multivariate analytic techniques. The statistical method is quite complex but a brief reference to its concepts and its application in biomedical studies will be presented.
1) Multiple regression - One of the most useful methods among multivariate techniques of wide applicability in biomedicine is regression analysis. It is a statistical tool that seeks to explain causal dependencies between a set of independent or predictor variables (referred to as Xi variables = the determinants of a disease) and the dependent variable (referred to as Y = the incidence or prevalence of the disease in a population). Via regression analysis, the strength and magnitude of causal associations among variables could be estimated.