 |
 |
.gif)
Brian A. Burt, B.D.S., Ph.D., M.P.H.:
Risk is the possibility that an event will occur. The word, of course, is used in everyday language with more or less that meaning, but it has more specific meanings in the worlds of insurance and epidemiology. In epidemiology it is related to probability and to causality, and it is most often used to express the degree of probability that a particular outcome will occur following a human being¹s exposure to a particular action or event. There are very few circumstances that constitute a sufficient cause in chronic or infectious disease (a sufficient cause being one where exposure to a specific action or event will probably result in a particular outcome). If there were, it would not be necessary to deal with risk, which essentially deals with varying degrees of necessary cause (a necessary cause being human exposure to an action or event that must always precede a particular outcome). The concept of risk in epidemiological study has also spread to include broader issues, such as risk assessment and risk-benefit analysis. This paper suggests definitions of risk and risk-related terms that can be used by the consensus panel for this conference.
There is general agreement that the term "risk factor" means an action or event that is statistically related in some way to an outcome‹smoking, for example, is a risk factor for periodontitis. But beyond that broad generality there is little agreement. There is uncertainty in the literature on whether a risk factor should be truly causal‹that is, a necessary link in the etiological chain‹or whether it can be only occasionally associated with an outcome.
There is also uncertainty about what strength of association is needed for an action or event to be called a risk factor for a disease, and just how directly it needs to be associated with the outcome. There is also disagreement over whether a risk factor must be immutable, like race or gender, or whether it is something that can be modified‹for example, a smoking habit. In the current studies to determine if periodontitis is a risk factor for cardiovascular disease, it is already clear that there is a measure of association between the two factors. However, it is also evident that periodontitis is neither a necessary nor sufficient cause of cardiovascular disease, and it remains to be demonstrated whether periodontitis interacts with other factors in leading to cardiovascular disease, or whether it is causal only in particular circumstances, or whether it is not causal at all but is a marker for other conditions that may be causal‹that is, people with periodontitis are likely to exhibit other factors which may be more directly linked with heart disease.
Any branch of science demands specific terminology, where words have precisely the same meaning among researchers who come from a variety of backgrounds, live and work in different parts of the world, and speak different languages. If we think about an enterprise like constructing the orbiting space station, for example, which involves multidisciplinary teams of scientists from different countries, it is clear that the project would quickly degenerate into chaos if there was not total uniformity in the meaning of many complex terms. Even in less demanding scientific projects, a failure to use precise terminology can result in frustration, inefficiency, and ultimately an inability to move our knowledge base forward.
Epidemiology is a relatively new science, and perhaps it is not surprising that there is uncertainty in our use of terms. The literature on measures of risk is replete with terms of uncertain definition, and supposedly standard terms are used in variable ways by different authors. Even the use of a supposedly standard term like "risk factor" is far from uniform. Rarely does an author define how the term is being used, and the evidence that leads to identification of a risk factor is often unclear. The term comes with a cluster of related terms like risk indicator, modifiable risk factor, risk marker, determinant, and demographic risk factor, which are often used more or less interchangeably in the literature. This sort of uncertainty means that the reader has to decide what the author has in mind.
If we turn to the standard dictionaries on epidemiology, we find they are not particularly helpful. In Last¹s Dictionary of Epidemiology (Last,1995), a risk factor (a term only in use since the 1960s) is defined as an aspect of personal behavior or lifestyle, exposure to an environmental event, or an inborn or inherited characteristic which on the basis of epidemiological evidence is known to be associated with health-related condition(s) whose prevention is considered important. That is a broad and rather loose definition that leaves unanswered questions about causal role, strength of association, and modifiability. The definition then goes on to list several different meanings that have been ascribed to the term "risk factor":
Last agrees that the term "risk factor" is rather loosely used, and I think we would agree that these definitions still leave important issues unanswered. In an effort to clarify the matter, Beck (1998) offered a definition that was adopted for the World Workshop on Periodontics in 1996:
This definition is longer than the one offered by Last, but in my view it is much clearer. The key contributions of this definition are (a) the emphasis on a temporal sequence of events preceding the outcome; (b) the unequivocal acceptance that a risk factor is part of a causal chain; and (c) the acceptance that risk factors are involved in the onset of disease but not necessarily in its progression or resolution.
Beck argues convincingly that it must be clearly established that the action or event occurred before the outcome, or before conditions exist that make the outcome likely. This in turn means that longitudinal studies are necessary to demonstrate risk factors. However, there are many situations in biomedicine, and certainly in dentistry, where this has not been done, and indeed where it is unlikely that it will ever be done. In these circumstances, exposure to an event that is associated with an outcome only in cross-sectional data is called a "risk indicator." A risk indicator may be a probable, or putative, risk factor, but the cross-sectional evidence upon which it is based is weaker than longitudinal data. This is because a temporal association usually cannot be specified from cross-sectional data.
If these definitions of the terms "risk factor" and "risk indicator" were used consistently, knowledge would most likely progress more quickly.
Burt BA. Risk factors, risk markers, and risk indicators... [editorial]. Community Dent Oral Epidemiol 1998;26:219.
Last JM, ed. A dictionary of epidemiology. 3rd edition. New York: Oxford University Press, 1995.
 
|
 |
The main page for this site is located at http://www.lib.umich.edu/dentlib/nihcdc/
and is maintained by the University of Michigan Dentistry Library (dentistry.library@umich.edu)
in partnership with the National Institutes of Health, National Institute of Dental and Craniofacial Research.
Please see also the Consensus Development Program pages, and in particular the NIH Consensus Development Conference on Caries.
Copyright © 2000 Regents of the University of Michigan. All rights reserved. |
 |
