Last week, a new heart risk calculator was launched with great fanfare. This week, it’s under fire.
For patients long frustrated by the sometimes whiplash-inducing back and forth of health recommendations, it might seem another cause to doubt taking any study–or advice from doctors–seriously. And, as the Lancet prepares to publish its critique of the calculator tomorrow, no doubt there are some lessons in coordinating how information like this is released.
But no evaluator is perfect. Indeed, the controversy highlights the biggest problem in preventive medicine – finding the best way to identify the highest risk individuals and match them to the best treatments available to lower that risk of developing disease. For something as complex as heart disease, it’s not likely that a single formula, such as the risk calculator, can accurately incorporate the various contributions of everything from weight, diet, smoking, race, and new factors such as inflammation. In 2003, for example, the AHA and the Centers for Disease Control studied the available data on the role that inflammation, as measured by a protein called C-reactive protein in the blood, plays in heart attacks and determined that the data wasn’t strong enough to recommend routine screening of all adults for CRP, but did advise that doctors could take advantage of the test, at their discretion, when evaluating heart patients for risk of second heart attacks. The new model does expand criteria for treating heart disease beyond cholesterol levels, but it still does not include an assessment of CRP.
When scientists want to test a new way of predicting disease, they look back. Because the best way to evaluate how accurately a new formula or model or equation predicts who will and who won’t get sick is by starting with patients who are already affected, plugging in their information and seeing whether the new calculator would have correctly anticipated their disease.
That’s what Dr. Paul Ridker and Dr. Nancy Cook did with the new heart risk calculator from the American Heart Association and the American College of Cardiology. The model, intended to be used by doctors to predict their patients’ risk of developing heart disease in the next 10 years, included factors such as age, race, gender, smoking history and blood pressure, as well as cholesterol levels. When it was released on last week, some experts anticipated an avalanche of confusion and criticism over the calculator, since it wasn’t clear how the formula was developed.
Ricker, a cardiologist at Brigham and Women’s Hospital and Harvard Medical School, and Cook, also at Harvard, took data from three large heart studies where all of the participants’ health information was known – their ages, smoking habits, and cholesterol levels, for example – as well as their heart disease outcomes. When they entered these measurements into the new risk calculator, however, they found that it overpredicted heart problems by 75% to 150%. That could mean the difference between prescribing a statin and not prescribing one, the doctors said in their report on their analysis, published in the Lancet.
Ridker, a pioneer in the CRP field, helmed a large study in 2008 that showed that healthy men and women with low cholesterol levels but high CRP levels who took statins lowered their risk of heart events by 44% compared to those taking placebo. He shares a patent on a blood test for the protein, and has therefore been criticized for having a conflict of interest in supporting the role of inflammation in heart disease. But his colleagues in the field have stood by the data on the landmark trial, which do show that in some people, inflammation may play an important role in contributing to heart disease.
Dr. Sidney Smith, a heart expert who served on the guidelines committee for the AHA, said to the New York Times that Ridker’s and Cook’s criticism of the model “merit attention,” and the organization is reassessing the model to see if changes need to be made to the formula.