Make your own free website on

Basis for Disease Prevention
and Health Promotion

Multicausality and Webs of Causation

Meet the Author
Epidemiologic Approach
Classic Model
Natality & Mortality
Disease & Disability

Germ theory suffered the weakness of being a single cause theory. All diseases were presumed to be caused by germs and by germs alone -- one species of germ per disease. When infection with, for instance, the cholera germ occurred, then the disease cholera should surely follow. The repeated proof that this was not true has not ended the tendency of most non-epidemiologist to seek an understanding of disease in terms of a single cause.

Thinking in terms of a single cause is highly detrimental to understanding and controlling disease. Evidence that smoking causes lung cancer is countered with the tale of Uncle Charlie, who smoked two packs a day and died at 97 without a sign of cancer. Even more compelling is the case of Cousin Sue, who died of lung cancer despite the fact that she never smoked a cigarette in her life. In each case, smoking is being assessed as a single and certain cause of cancer, when it is, in fact, only one of a set of interrelated factors that together can cause cancer.

The classic epidemiologic triad of host, agent, and environment provided a better model for understanding the complex realities of disease causation. The classic model, however, developed almost entirely as a model for the understanding of infectious diseases. The centrality of the concept of agent is in many ways a carryover from germ theory. In the triad model, the single element of the agent is represented as if it were equal in importance to the variety of relevant factors in the host and the multitude of environmental influences. The actual findings of epidemi¬ology in the study of a vast array of diseases have not supported this exaggerated weight given to the agent as a. cause at disease.

During the present century, medical and social progress have reduced the impact of infectious diseases on society and have increased public health concern with non-infectious chronic diseases. Difficulties have been encountered in applying the concept of agent to many of the chronic diseases. In fact, it often has seemed more like an intellectual game than a useful pursuit, to try to identify a single factor without which the particular disease under study could not occur.

The nature of this problem may be illustrated by reference to an attempt to develop a causal model of child abuse in terms of host-agent-environment-vector (Justice & Duncan, 1975, 1977; Justice & Justice, 1976). Viewing child abuse as a disorder within the parent dyad or parent (host), what element must always be present (or always be deficient) in order for child abuse to occur? The answer given by this model is that the child is the agent of child abuse. Obviously, child abuse cannot occur without a child -- but this is clearly an extension of the term agent be¬yond its original usage.

In seeking a model that better expressed the complex reality of multicausality, some epidemiologists began thinking in terms of chains of causation. In a chain of causation we looked at a causal event, then at the antecedents of that event, then at the antecedents of the antecedent, and so on. For instance, we note that a person developed a febrile illness known as leptospirosis after becoming infected with Leptospira interrogans; this organism entered through a cut finger while the person was emptying a cat litter box; this was possible because the litter was contaminated with Leptospira; this was because the cat had leptospirosis—a common disease of cats—and was shedding the infectious organism in its urine. The chain of causation is thus: Infected cat -> contaminated urine -> contaminated litter —> infected cut -»leptospirosis in host.

Such chains, however, left much out. For instance, this chain of events would not have occurred if the host had not had a cut finger. It would not have occurred if the host had worn rubber gloves to change the litter box. It might not have occurred if there had been a bandage on the cut. It probably would not have occurred if the host had promptly washed after changing the litter box. It would not have occurred if the host had immunity due to a prior case of the same type of leptospirosis or vaccination against that type of leptospirosis. It would not have occurred if the cat had been placed in isolation at a veterinary clinic until it was cured. It would not have occurred if the cat had been immunized against leptospirosis. And so forth.

Another example of a chain of causation is the "diet-heart hypo¬thesis" (DHH) as described by Sherwin (1978). As hypothesized in this chain, a diet high in saturated fat and cholesterol leads to high blood lipids, which lead to atherosclerosis (coronary artery disease), which leads to coronary heart disease and the clinical event of a myocardial infarct (heart attack). This simple chain, however, leaves out many important elements. Sherwin notes, for instance, that the link between diet and elevated blood lipids is also influenced by genetic factors and emotional stress. The link between high blood lipids and atheroscle¬rosis is also promoted by genetic factors, aging, hypertension, smoking, stress, and low blood levels of the protective high-density-lipoprotein cholesterol (HDL). Genetics, smoking, and stress again contribute to the link between atherosclerosis and coronary heart disease, as does the presence of free-floating blood clots (emboli), which can block an artery already constricted by atherosclerosis. Adequate representation of all of these additional elements requires a more complex model than the simple chain of causation.

MacMahon, Pugh, and Ipsen (1960) recognized that chains of causa¬tion suffered from "the defect of oversimplification" and proposed a new model that they called a "web of causation." Their first attempt at such a chain or web was to explain the occurrence of icterus (or jaundice) among some patients being treated for syphilis. Of this expanded chain of causation they state: When it is considered that only a few of the major components are shown, that these are indicated as broad classes of events rather than as the multiple minor events comprising each class, that each component shown is itself the result of a complex genealogy of antecedents, and that the myriad of effects of these components other than those contributing to the development of icterus are not shown, then the "chains" of causation become difficult to elucidate and the whole genealogy might be thought of more appropriately as a web, which in its complexity and origins lies quite beyond our understanding. Fortunately, to effect preventive measures, it is not necessary to understand causal mechanisms in their entirety, (p. 18)

In an attempt to more fully represent the "genealogy" of the disease and its causes, in a web of causation not only is the disease seen as having multiple causes, but each of these causes is seen as being an effect that resulted from multiple causes, each of which is an effect resultant to multiple causes, and so on. Each arrow on a web of causation means that the element at the source of the arrow is, to some degree, a cause of the element at the point of the arrow. In the web of causation proposed by MacMahon, Pugh, and Ipsen, injection of hepatitis virus is one of the direct causes of jaundice in syphilis patients. This comes about through the injection of foreign (that is, from another person) serum, the presence of epidemic hepatitis in the community, and the biological characteristics of the hepatitis virus that allow it to survive and be transmitted in this fashion. The injection of foreign serum occurs due to intravenous injection as a treatment choice for syphilis, prior use of the same syringe for intravenous injection, serum remaining in the sy¬ringe from that use, and poor syringe hygiene that fails to eliminate the contamination.

A later and more sophisticated, but rather generalized, web of causation was developed by Stallones (1966) to describe some of the interrelationships among the three major types of cardiovascular disease. While this web does not attempt to provide a step-by-step description of the mechanisms by which the causal factors operate, it does give a clear view of how different factors may work together to produce one form or another of cardiovascular disease. His web of causation, for example, shows that hereditary tendencies, stress, and lack of physical activity contribute to hypertension (high blood pressure). This is also probably enhanced by atherosclerosis—the buildup of plaque on the inside walls of the arteries. Salt or sodium in the diet is also shown as possibly playing a role in causing hypertension, although the evidence for the generality of this factor is far less certain than the public and most health professionals seem to believe. Hypertension, then, in combination with atherosclerosis and the breakdown (lysis) of blood clots (resulting in a free-floating clot within the blood supply—an embolus) may cause a myocardial infarction (heart attack) when the clot lodges in one of the arteries constricted by atherosclerosis and blocks the blood flow to part of the heart muscle.

A more detailed web of causation for myocardial infarction (the classic "heart attack") was developed by Friedman (1980). Complex though this web may seem, it is really only a begin¬ning at mapping what is known about the etiology of heart disease.

A web of causation can be of great value in identifying possible points for preventive interventions. By examining Friedman's web of causa¬tion for myocardial infarction, for instance, we recognize that little can be done about the immediate causes of coronary occlusion (the artery is blocked by an embolus), atherosclerosis, myocardial susceptibility, or a deficiency of collateral circulation. By looking at the next level above this, however, we find hypertension, which is a possible point for intervention. We can treat hypertension with an array of highly effective drugs. A low-fat diet should, in most cases, help to reduce hyperlipidemia (excessive fats in the blood); we also have a growing ability to treat this condition with drugs. With these two interventions the risk of significant coronary atherosclerosis would be greatly reduced and that, in turn, would reduce the risk of heart disease.

Related Links

Webs of Causation of Drug Use and Abuse

Recommended Readings

Duncan, D. F., & Petosa, R. (1997). Social and community factors associated with drug use and abuse among adolescents. In: T. P. Gullotta, G. R. Adams, & R. Montemayor (Eds.) Substance Abuse in Adolescence (pp. 56-91). Thousand Oaks, CA: Sage.

Hawkins, W. E., Duncan, D. F., & Hovet, R. (1992). Teaching multi-dimensional views of disease causation: The web of causation in epidemiology. Journal of Health Education, 23, 301-303

MacMahon, B. Pugh, T. F., & Ipsen, J. (1960). Epidemiologic Methods. Boston: Little, Brown.

Rothman, K. J. (1976). Causes. American Journal of Epidemiology, 104, 587-592.

Rothman, K. J., Greenland, S., & Walker, A. M. (1980). Concepts of interaction. American Journal of Epidemiology, 112, 465-466.

Stallones, R. A. (1966). Prospective epidemiologic studies of cerebrovascular disease. Public Health Monograph No. 76. Washington, DC: U.S. Government Printing Office

Susser, M. (1973). Causal Thinking in the Health Sciences: Concepts and Strategies of Epidemiology. New York: Oxford University Press.

(c) David F. Duncan, 2007
originally published by Macmillan Publishing Co., 1988

Powered by WebRing.