Original research article:
A group of 814 men, 40 to 59 years old, were placed on a diet rich in polyunsaturated fatty acids and underwent intensive nutritional counseling (experimental group). This experimental group was compared to a control group of 463 men, who received no dietary modification.
The findings of this study were celebrated by the authors as “significantly reducing the incidence of obesity and hypertension”; producing a “statistically significant” difference in “risk factors”; and “a statistically significant difference between the two groups in morbidity from new coronary heart disease.”
However: the death rate of the experimental group was more than twice that of the control group, albeit, not as far as I have been able to see through my own statistical tests, statistically significant.
To herald the “prudent diet” as a success (as the authors appear to), and then to focus on “risk factors” as the desired outcome, strikes me as just a little bit shifty.
What, after all, is the point of medicine? To reduce risk factors or to prevent death?
I hope the answer to this question is obvious.
50 years later, this methodology is still prevalent. Reporting on one measure, or one cause of mortality, and not reporting or de-emphasizing the importance of the most relevant changes (say, in mortality, or a more direct measurement of the phenomena), remains a not uncommon practice in biomedical science.
Targeting narrow measures of health certainly increases the ability of a researcher to detect a change; it also increases the risk that such a measure is meaningless; perhaps most dubiously, it gives free reign to promote an idea or product that may quite well be useless–and potentially dangerous.
Because this practice is so prevalent, and has proven problematic in some high profile cases in supporting overdiagnosis and overuse of medical resources, we want to give this practice a name. For now, we will call it mortality partitioning or variable bias. Or perhaps risk factor bias, in honor of the title of this paper. We give this a name, in particular, because we will be returning to this concept repeatedly in the course of this series on PUFAs on this blog; we want to be able to use a single term to track the incidence of this phenomenon.