March 29, 2014

WASHINGTON, DC — Despite fears that it could extend statin therapy to vast swaths of the US population that might not need it, the risk equations recently introduced in the 2013 ACC/AHA Guideline on the Assessment of Cardiovascular Risk may be a more accurate predictor of atherosclerotic CV-disease events than its critics claim[1].

When the document's "pooled cohort" risk equations are applied to a contemporary cohort of people not on statins and without standard statin indications, its risk predictions are reasonably similar to observed event rates, at least over five years, according to an analysis published today in the Journal of the American Medical Association.

"Calibration [of the risk-prediction tool] in this group is particularly important because it represents the population for whom high predicted risk is intended to trigger a discussion about statin initiation," explain the authors, led by Dr Paul Muntner (University of Alabama at Birmingham). Muntner is also slated today to present the study here at the American College of Cardiology 2014 Scientific Sessions .

The analysis focused on a subgroup of about 11 000 white and black participants from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) prospective observational study, those who were not on statins, not diabetic, and had an LDL-cholesterol level 70 to 189 mg/dL. Another REGARDS cohort examined in the study consisted of about 3300 who could be cross-referenced with Medicare vascular-disease-events claims data.

Debate: Relevance of the Pooled-Cohort Equations

Their analysis, Muntner et al note, was meant to address concerns that application of the new risk equations to the larger overall REGARDS population, a validation cohort for the new risk-assessment tool, "appeared to overestimate atherosclerotic cardiovascular-disease risk." It also seemed to overestimate risk in another validation cohort from the Multiethnic Study of Atherosclerosis (MESA).

As covered extensively by heartwire , those inconsistencies fueled a vociferous back-and-forth among experts over the relevance of the pooled cohort risk equations to contemporary patients, including a report by two luminaries in the field that the equations overestimated risk by 75% to 150% in the primary-prevention cohorts of the Women's Health Study , the Physicians' Health Study , and the Women's Health Initiative Observational Study .

A month later, in a letter to Circulation, Muntner and some other authors of the current analysis published their thoughts and theories about why the equations overestimated risk in REGARDS[2]. For example, they proposed, REGARDS and MESA may have appeared to overestimate risk due to ascertainment bias because they did not include rigorous surveillance of CV events or use the medical records of all patients.

It is in part those ideas that they explored in the current analysis, which used Medicare claims as a proxy for surveillance and added two coauthors: Dr David C Goff Jr (Colorado School of Public Health, Aurora) and Dr Donald M Lloyd-Jones (Northwestern University, Chicago, IL), cochairs of the working group behind the 2013 guidance that introduced the pooled-cohort risk equations.

And its findings, according to the authors, "support the validity of the pooled-cohort risk equations to inform clinical management decisions." The caveats, they note, include shorter-term risk projections than the 10-year risk the pooled cohort risk equations were designed to predict; the ongoing REGARDS cohort study doesn't as yet have a 10-year follow-up, so the projections were limited to five years.

Another Study, Another Conclusion

Accompanying the published REGARDS analysis in JAMA but not its live presentation at the ACC sessions is another assessment of pooled-cohort equations from based on 4854 persons who entered the prospective, observational Rotterdam Study from 1997 to 2001 at age 55 or older[3].

"We found that nearly all men and more than 65% of women were recommended for drug treatment based on the recent ACC/AHA guideline," state the authors, led by Dr Maryam Kavousi (Erasmus University Medical Center, Rotterdam, the Netherlands).

Similar analyses based on the Adult Treatment Panel III (ATP-III) and the European Society of Cardiology (ESC) guidelines, applied to the Rotterdam cohort, also overestimated the risk of atherosclerotic CV events calling for statin therapy as determined by the new AHA/ACC guideline, although using slightly different clinical end points.

"Given the modest discrimination and poor calibration of the ACC/AHA risk prediction model, the choice of treatment threshold becomes central," they write. The new model controversially recommended a 7.5% 10-year risk as the threshold for considering statins.

In an editorial accompanying both the Rotterdam and REGARDS analyses[4], Dr Harlan M Krumholz (Yale University, New Haven, CT) writes, "Remarkably, the most radical aspect of the cholesterol guidelines—the decision to abandon targets for treatment—has drawn relatively little reaction. The guideline is mainly embroiled in a controversy about the calculator it endorses to assess patient risk."

The Rotterdam study, according to Krumholz, "raises substantial concerns about the generalizability of the risk equations and also raises the question of whether a threshold for treatment, which is admittedly arbitrary and is imbued with values about what level of risk is worth treating, is relevant to all countries."

As for the REGARDS analysis, he calls "disquieting" the conclusions "that the lack of active surveillance in observational studies may lead to many missing end points, which may explain the appearance of overestimation of risk in some cohorts."

Dr Paul Ridker (Brigham and Women's Hospital, Boston, MA), one of the authors Muntner et al were responding to with the current REGARDS analysis, told heartwire in an email: "In the new Rotterdam data and in the reanalysis of the REGARDS data as a whole, the ACC/AHA calculator overestimated risk and calibrated poorly, quite consistent with prior studies."

In the current subgroup analysis of REGARDS, he continued, "the risk calculator performed better, which is a step in the right direction. My colleagues and I look forward to continued work with the ACC and AHA to best align clinical recommendations with trial data."


Observed vs Predicted Risk in REGARDS

REGARDS currently includes about 30 000 adults who were at least aged 45 when they entered from 2003 to 2007 and who have been followed primarily for self-reported stroke events but also other types of vascular disease.

In the main study subgroup of 10 997 participants, there were 338 adjudicated instances of atherosclerotic CV events, defined—as in the new guideline—as nonfatal MI, CHD death, or fatal or nonfatal stroke.

For participants with different ranges of 10-year risk as predicted by the new guideline, the five-year incidence of atherosclerotic CV events per 1000 person-years, observed vs predicted, respectively, was:

  • 1.9 and 1.9 for <5% 10-year predicted risk.

  • 4.8 and 4.8 for 5% to <7.5% risk.

  • 6.1 and 6.9 for 7.5% to <10% risk.

  • 12.0 and 15.1 for >10% risk.

"The observed and predicted atherosclerotic cardiovascular-disease risks were much more similar when evaluated in participants with Medicare insurance coverage, [using] the atherosclerotic cardiovascular-disease events identified in Medicare claims," the group noted.

In that set of 3333 participants, there were 234 atherosclerotic CV events. For those with similarly different ranges of 10-year risk by the new guideline, the five-year observed vs predicted incidences per 1000 person years was:

  • 5.3 and 4.0 for <7.5% risk.

  • 7.9 and 6.4 for 7.5% to <10% risk.

  • 17.4 and 16.4 for >10% risk.

The consistent similarity between observed and predicted rates in the analysis, the group concluded, indicates that the pooled-cohort risk equations "were well calibrated in the population for which they were designed to be used and demonstrated moderate to good discrimination."

This research is supported by a cooperative agreement from the National Institute of Neurological Disorders and Stroke, National Institutes of Health, and Department of Health and Human Services. Additional support was provided by grants from the National Heart, Lung, and Blood Institute. Muntner discloses receiving grants and fees from Amgen for work not part of the current study. Disclosures for the coauthors are listed in the paper. Kavousi reports research grants from AXA Research fund (outside the submitted work). Krumholz reports that he is the recipient of research grants from Medtronic and from Johnson & Johnson, through Yale University, to develop methods of clinical-trial data sharing and is chair of a cardiac scientific advisory board for UnitedHealth. Ridker receives research grant support from Novartis, Pfizer, Amgen, and the National Heart, Lung, and Blood Institute; and is listed as a coinventor on patents held by the Brigham and Women's Hospital that relate to inflammation and vascular disease that have been licensed to AstraZeneca and Seimens.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.