The esteemed Royal Mathematician, Dr. (KN)J writes in:
Dear Dr. J.,
A recent missive of yours touched on a subject on which I have long been meaning to touch. Namely, the applicability, or lack thereof, of the “ounce of prevention, pound of cure” aphorism to the current Healthcare debate. Mathematically speaking, while an ounce of prevention may indeed be worth a pound of cure to an individual patient, systemwide it is only so if the ailment is sufficiently widespread. Specifically, an ounce of universal prevention is only worth a pound of universally-provided cure if the ailment afflicts at least 1/16 of the population (1/12 of the population for Troy ounces/pounds). I leave it to your able medical judgments as to whether each individual prevention/cure combination would be in that range.
More seriously, this is the fundamental problem of assessing the cost-effectiveness of preventive medicine. It may improve outcomes (hence the validity of the aphorism from the view of the patient) but deciding about cost-effectiveness requires accurate knowledge of disease incidence probabilities (both absolute and conditional) that are not nearly as tightly
known as the relative costs (though those may not always be known, either). It seems to this mathematician that the best solution here, as in many cases, may be a mixed-strategy, with distributed decision-making – i.e., leave it up to the individual patients and their doctors.
Royal Mathematician to the Gormogons
As always, you make an excellent point. Medicine requires an understanding of probability and statistics (albeit a largely qualitative one), because the reality is that a lot of what we do is manage risk. Surgeons get to fix things, but for the potion pushers
, we’re all about risk reduction.
For example, people who have had a heart attack, and as a consequence their cardiac function is halved, then they are at higher risk of sudden death due to an arrhythmia. There is a device called an implantable cardioverter defibrillator (ICD). In a clinical trial
of 1232 such patients who were randomly assigned to an ICD or medical therapy, it was found that at 2 year follow up, 84% of defibrillator patients were still alive and 78% of medical therapy patients were alive (NEJM 346:12, 877, March 21, 2002). So, to prevent one death over 2 years, 17 ICD’s had to be implanted. Despite the $50,000 price of an ICD, this study with a 2 year number-needed-to-treat (NNT) of 17 drove a significant increase in ICD placement in patients. Indeed, to stave off the cost, only subsets of patients who had the greatest benefits (those with a ‘wide’ EKG spike) were initially approved for devices. At 8 year follow up, the benefit of having an ICD grew such that the NNT dropped to 6, and the ubiquity of ICD placement grew with it.
You are also right in that patient care should be personalized. Just because someone meets the inclusion criteria for an ICD (low left ventricular ejection fraction and prior heart attack or congestive heart failure), doesn’t mean that they’re the best candidate for such a device. They may have a short life expectancy or other significant medical conditions that would render the benefit of the device moot. This decision is best made after a careful conversation between the doctor and the patient occurs. And if a strange gentleman
is standing outside the room offering a blue pill
as an alternative, just run away…fast.