The risk assessment of an individual.that is determining a person's risk of a given disease has been described a peculiar,elusive,ambiguous process. The statistician,Richard Von Misses ( Ludwig's brother ) said that it is only possible to speak of probability in terms of a collective ( ie. a group or a set ) and that to say , for example that Mr. Jones has the probability of 0.3 in the next five years of , for example, having a heart attack is nonsense. Yet, physicians are encouraged ( and nowadays mandated in order to practice proper guideline driven medical practice) to do this in a variety of situations. The famous Framingham risk model is often used to assess the risk of a heart attack. For example,using an equation from the NCEP ( National Cholesterol Education Program) we find that a 67 year,non smoking man with a total cholesterol of 170 and an HDL of 75 would have a 10 years risk estimate of 9%. According to the Framingham data, there would be 9 deaths in a ten years period in a group of men with those characteristics ,but is it meaningful to say that a given member of that group has a 9% risk. Further, the risk number is not just something mentioned in the exam room but is often folded into guidelines from various organizations,for example the ACC .Someone with a risk of 10% ( or some committee derived cutpoint) might be advised to take a statin or aspirin.
What if we constitute another statistical collective since we can make up as many as our imagination and data allow.In a Framingham type study we could measure CRP,presence of absence of coronary calcification,results of a exercise stress test,blood sugar and whether or not the person takes daily vitamins. The person in the previous paragraph might now be determined to have a risk of 15%.So what is his " real " risk? Is there a real risk at all ?
Is an elusive ,ambiguous risk number sufficient reason to recommend lifetime medications? I don't think so now; but that happens regularly in doctor's offices and I used to do it on almost a routine basis and thought I knew what I was doing.
Frequently medical societies and committees will offer generic advice to generic patients by suggesting they complete a form or risk assessment tool but then consult with their physician /health care provider to discuss the issue and offer an "individualized" risk assessment. I wonder what sort of magical statistical tool the providers has at her disposal to achieve that result other than the same risk assessment tool the patient has already used.