RSS Feeds
Probability at the Bedside: The Knowing of Chances or the Chances of Knowing?
In this issue, Steiner (1) discusses the impact on the patient of presenting risk reduction estimates in two ways: as a statement of the patient's relative risk reduction and as the number needed to treat. Steiner calls the relative risk reduction the “language of individuals” and the number needed to treat the “language of populations.” This distinction is part of an ongoing discussion in the literature of problems with bringing evidence-based medicine to the bedside (2-6). What is less appreciated is that it is also part of a centuries-old, unresolved philosophical debate about the meaning of an individual patient's risk. The relevance of this debate today is manifest in the difficulty that many physicians have in discussing probabilities with patients.
We use the term risk as though its meaning is clear, but we rarely stop to think about what an extraordinarily peculiar medical characteristic it is, particularly when applied to an individual patient. We do not measure it directly in our patients, as we would any other clinical variable, but rather in others who are in some way “like” our patients. And it is not measured in another individual patient; we instead take the property of a group—for example, the percentage of patients that dies—and call that our patient's risk. But when we transfer a number from a group to an individual patient, its meaning can qualitatively change.
One of the earliest treatises on what we now call probability, Liber de ludo aleae (“Book on Games of Chance”), was written by Girolamo Cardano, a 16th-century physician (7). At that time and for centuries afterward, the notions of probability and chance were separate. Chance was a property of clearly random mechanisms, such as a roulette wheel, a coin flip, or a roll of the dice. This conception of probability …
