One of the most powerful patient education tools I've used in practice is the Framingham Heart Study's coronary risk calculator, which estimates the 10-year risk of having a heart attack using age, sex, smoking status, cholesterol and blood pressure levels. I've frequently used this calculator to illustrate that quitting smoking is far more effective at lowering one's heart attack risk than taking drugs to lower cholesterol and blood pressure to recommended levels.
It turns out, though, that not all risk calculators are created equal. A recent study published in the Journal of General Internal Medicine found that many online heart risk calculators, including the one on the American Heart Association's website, use a simplified algorithm (a point-based tool, rather than an equation) that ends up estimating an artifically high risk for 10% of adults and an artificially low risk for 5%. Worse, using the faulty calculator, 39% of patients with high cholesterol would have met criteria for more intensive therapy (either a higher drug dose or the addition of a new drug) than necessary. Since most doctors are unaware of differences between the calculators, it's possible that millions of patients might be misclassified and subjected to more treatment than they really need.
While the solution to this problem is fairly obvious - toss the point-based tool, which was created to help docs do pen-and-paper calculations in the days before widespread Internet access - the solution to another problem of estimating cardiac risk is much less clear. As it is with many other health conditions, socioeconomic status (i.e., being poor) is an independent risk factor for heart attacks that isn't counted by the Framingham calculator.
But it's clearly impractical (and ethically questionable) to practice primary prevention of heart disease on the basis of an individual patient's family income. So a group of researchers publishing in the current issue of Annals of Family Medicine suggested a somewhat more workable alternative: substituting individual incomes with residential income quartiles derived from census areas, since people of similar income levels tend to live in the same neighborhoods. They suggest that electronic health records might be programmed to automatically incorporate the patient's street address into the coronary risk calculation, or conversely, lower the risk thresholds for cholesterol-lowering treatments.
I was attracted to family medicine partly out of a desire to reduce health disparities, but I honestly hadn't imagined that doing so would potentially involve prescribing more cholesterol-lowering drugs to people just because they happened to live in poor neighborhoods. On the other hand, it seems unethical to expose poorer patients to higher heart attack risks just because I feel uneasy about differentiating patients based on socioeconomic status. I'd very much enjoy hearing other clinician and patient perspectives on this issue.