YOU ARE HERE: LAT HomeCollections

Risk calculator for determining statin use needs fixing, doctors say

November 18, 2013|By Melissa Healy
  • The risk calculator unveiled last week to help doctors and patients figure out who should take statins has some serious flaws, critics say.
The risk calculator unveiled last week to help doctors and patients figure… (JB Reed / Bloomberg News )

Less than a week after the American Heart Assn. and the nation's cardiologists issued guidelines that would greatly expand the number of Americans taking a statin medication, the guidelines have been faulted for overestimating patients' risk of heart attack or stroke.

Few authors of the new recommendations had even returned to their clinical practices before learning that an influential Harvard cardiologist and his biostatistician collaborator had taken the guidelines to task, arguing they use unreliable data on Americans' health to calculate which patients would benefit from taking the medication.

The criticism come amid a collective sigh of exasperation from physicians who were expected immediately to start using the new guidelines, along with the "risk calculator" that helped identify those patients who are still free of heart disease but who should get a statin to ward off heart attack or stroke.

"If you put the data from hypothetical patients into the calculator, you can get results that are very implausible," said Cleveland Clinic cardiologist Steven Nissen, who urged the American College of Cardiology  and the American Heart Assn. to "take a deep breath and take another look at this" before putting the guidelines into widespread practice.

"It's concerning: We’re not talking about an error or miss of 20 or 30%, but of 75 to 150%" in assessing a patient's risk of heart attack or stroke, said Nissen. "This is not a tempest in a teapot."  

Dr. Paul Ridker and Dr. Nancy Cook, both professors at Harvard Medical School, estimate that between 13 and 16 million of the 33 million middle-aged adults targeted by the new guidelines for statin therapy do not have sufficiently high odds of having a heart attack or stroke over the next decade to warrant statins' use. The risk calculator might give an all-clear signal to other patients for whom statin therapy might reduce heart attack and stroke, the researchers also warned.

The critical look at the guidelines was to be published Tuesday in the British journal the Lancet. Cook is a biostatistician at Brigham & Women's Hospital and Ridker is a cardiologist at Brigham & Women's who has been an advocate for expanded statin use.

Cook and Ridker used the calculator to estimate the 10-year odds of heart attack or stroke in a population of study subjects different from those used by the guidelines' authors. To check the validity of the risk calculator, Cook and Ridker used the data on actual heart attacks and strokes tallied from the landmark Women's Health Initiative, the Women's Health Study and the Physicians Healthy Study patients.

But when they calculated the 10-year heart attack and stroke risks of subjects in those trials, they got estimates that "were almost twice as high as they actually were," said Nancy Cook.

"They saw it themselves," said Cook in an interview Monday, who said the guidelines' scientific advisors acknowledged in the report's supplement that the calculator's assessment of risk does not always agree when populations from other studies are used.

"They knew this," said Cook. "I think they should have done something about it, tried to figure out what was going on here."

The pair's analysis Monday prompted a stout defense of the guidelines' usefulness from their chief author, Dr. Neil Stone.

"We anticipated this," said Stone, who said a physician's use of the risk calculator was just the first step in a process of consultation in which both patient preference and a doctor's clinical judgment are to be discussed and considered.

"We erred on the conservative side of caution" in identifying patients who should have that conversation, said Stone. But, he added, the bases for the risk calculator "were reviewed on several occasions by multiple reviewers. They were internally and externally validated, and all the populations they checked it against were microcosms of the U.S. population."

By contrast, the study subjects that Ridker and Cook used for comparison were "in all fairness ... low-risk populations," said Stone. Because many of them were health professionals, they were less likely to suffer heart attack or stroke than today's adult Americans, Stone said. And groups whose risks tend to be particularly high, such as African Americans and older patients, were less well represented in the study populations Ridker and Cook used, he added.

Stone also suggested the Ridker-Cook critique does not materially change the recommendations most patients would get. He said a "significant number" of those who would get a higher stroke and heart attack risk assessment with the new calculator than Ridker and Cook would assign them are patients already at highest risk for those outcomes. Scaling back to the criteria used by Ridker and Cook "would not remove them from the treatment group," said Stone.

Los Angeles Times Articles