It may be beneficial to rethink the whole concept of causation and recognize that many chronic diseases emerge from a complex maze of interconnected factors that cannot be viewed separately.
Recent interest in the association between oral infections and nonoral diseases has drawn attention to the importance of establishing causation. The impact on health care delivery, if true cause and effect can be found, will be immense and has unfortunately led many researchers, in the interest of doing good, to claim causation by means of leaps of faith rather than by scientific rigor.
Pinpointing the etiology of chronic complex diseases, such as cardiovascular disease, diabetes and the like, is a slippery slope and, until now, has involved the use of traditional epidemiologic methods to imply causation. Unfortunately, these types of approaches are inadequate for this purpose.
Several obstacles need to be overcome to establish causality. Most importantly, current epidemiologic methodology needs to be supplanted by more dynamic modes designed to detect complex interactions in which risk factors have unpredictable effects.1 Multifaceted biological and sociological interactions seldom can be accounted for by empiric observations. Inductive approaches (moving from specific observations to broader generalizations and theories) typically are used to infer answers from collected observations, but these approaches are limited as future contributing variables cannot be induced, nor can an appropriate statistical or other framework be established that can distinguish between the clinical significance of different study outcomes.
Clinical trials today are the favored epidemiologic method used to establish a foundation for causation. But can these trials give us answers that explain why certain diseases and conditions occur? At most, they can tell us if one treatment is more efficacious than another treatment. What these trials cannot tell us is why the treatment was more effective in one group of people and less effective in another.
The reason for this, among many other things, is that randomized controlled trials are limited by using fairly strict inclusion and exclusion criteria to narrow down confounding factors. This will result in an inability to generalize study outcomes owing to the exclusion of factors that could have explained different results between groups. Unfortunately, epidemiologic investigations can, in the best of circumstances, only estimate the causal association between exposure and disease. It is not clear how appropriate it may be to use the results from these types of investigations to recommend individual interventions for patients.
For example, we know that perhaps up to 90 percent of patients with cardiovascular disease will have at least one major risk factor. Yet, almost the same percentage—up to 90 percent—of those with the same major risk factors will not develop cardiovascular disease.2 Not all smokers will develop lung cancer, oral cancer or cardiovascular disease. However, because of the dire consequences associated with these diseases, it is prudent, from a public health standpoint, to recommend smoking cessation to all patients, even though only a certain percentage of them will benefit from this type of intervention. Thus, risk factors can predict the development of disease among specific groups of patients, but they cannot answer questions about causation or susceptibility on an individual basis.
Elucidating causation serves several purposes. Most importantly, it helps define a point of intervention. With complex diseases, it will not be possible to identify a single cause, but rather several factors that may contribute to the development of disease must be recognized. Unifactorial thinking will thwart the recognition of complex etiologies and impede scientific progress. It is the responsibility of researchers to discern a hierarchy of magnitude of specific risk factors that can define their impact. This would enable further inquiry into biological and sociological pathways that underlie the development of complex diseases. This is not purely a biomedical issue; it also is an ethical issue.
Causation can be used to attribute blame, treatment choices and prevention. In the beginning of the HIV epidemic, the individual behaviors were blamed for the disease. Not until the actual virus and modes of transmission were established could effective public health and medical interventions be implemented.
The recent development of a vaccine against human papillomavirus to prevent cervical cancer is another example of how effective intervention can be implemented only after causation has been determined. When a microbial etiology is the driving force in the development of a disease, prevention and management can be addressed accordingly. However, if the major contribution to disease development is thought to be genetic susceptibility, can we use the same logical approach as when dealing with a microbial threat and justify genetic screening and manipulation? Choosing lifestyle interventions to combat disease can bring about specific health policies that may have political consequences. Unfortunately, we sometimes select specific factors as causative only because they can be monitored and controlled.3 It may be more beneficial to rethink the whole concept of causation and recognize that many chronic diseases emerge from a complex maze of interconnected factors that cannot be viewed separately.
If the major contribution to disease development is thought to be genetic susceptibility, can we use the same logical approach as when dealing with a microbial threat and justify genetic screening and manipulation?
Determining causation has scientific, biomedical, ethical, political and health policy implications. The impact of etiologic risk factors must serve to explain biological pathways and clinical outcomes. Medical plausibility is not enough to determine causality. Instead we need to approach new findings with caution and rely on sound scientific principles before acting on what may look like attractive explanations and easily achieved interventions.
As Karhausen4 noted: "No doubt, what clinicians and epidemiologists call the cause is quite different from the causa vera; it refers to the factor to which they wish to call attention or the most obvious one, bracketing off temporarily the other contributing factors or standing condition."
