Is the New ACC/AHA Risk Calculator Overestimating Risk in Low Risk Subjects?

[bohm2]

Just looking at the new guidelines, it seems that prescriptions for statins are going to increase above the already very high level:

The new guideline recommends moderate- or high-intensity statin therapy for these four groups:

•Patients who have cardiovascular disease;
•Patients with an LDL, or "bad" cholesterol level of 190 mg/dL or higher;
•Patients with Type 2 diabetes who are between 40 and 75 years of age; and
•Patients with an estimated 10-year risk of cardiovascular disease of 7.5 percent or higher who are between 40 and 75 years of age (the report provides formulas for calculating 10-year risk).

New Guideline for Management of Blood Cholesterol: Focuses On Lifestyle, Statin Therapy for Patients Who Most Benefit
http://www.sciencedaily.com/releases/2013/11/131112163210.htm

 

[bohm2]

A criticism of the calculator to estimate 10-year risk of cardiovascular disease: http://my.americanheart.org/professional/StatementsGuidelines/PreventionGuidelines/Prevention-Guidelines_UCM_457698_SubHomePage.jsp has just been published in Lancet by Drs. Ricker and Cook:

The new ACC/AHA risk prediction algorithm systematically overestimated observed risks by 75-150%, roughly doubling the actual observed risk.

Statins: new American guidelines for prevention of cardiovascular disease
http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(13)62388-0/fulltext

 

[paulmarin]

The big change seems to have come from the Cholesterol Treatment Trialists (CTT) study published in The Lancet last year. The issue with this study is the denominator – they looked at the effect on cardiovascular risk per unit LDL reduction. The risk of using this denominator is that there’s no way to calculate an NNT for statins – they potentially excluded those that are on a statin with minimal LDL reduction, but included those on a statin that may have lowered their LDL thru other means.

 

[SW VandeCarr]

The big change seems to have come from the Cholesterol Treatment Trialists (CTT) study published in The Lancet last year. The issue with this study is the denominator – they looked at the effect on cardiovascular risk per unit LDL reduction. The risk of using this denominator is that there’s no way to calculate an NNT for statins – they potentially excluded those that are on a statin with minimal LDL reduction, but included those on a statin that may have lowered their LDL thru other means.

Could you please provide a link for these assertions. The Lancet article was regarding a meta-analysis of 27 trials which showed actual reductions for cardiovascular events. More importantly, it showed equivalent reductions in both low risk and high risk categories. I saw no data on "number needed to treat" (NNT) but it could be inferred from the consistency of risk reduction across low to high risk categories. The reduction in LDL-C was another analysis.

I would be interested in any critique of these findings. The issue would be with the meta-analysis rather than the trials themselves since there are still some issues with the methodology of meta-analysis. I look forward to your link. The reduction of LDL-C is well established for statins as a class.

http://www.ncbi.nlm.nih.gov/pubmed/22607822

 

[bohm2]

I think the problem here is confounding factors. There are other cholesterol medications like Ezetimibe (Ezetrol) that reduce LDL (whether as monotherapy or in combination with a statin) but have no effect on overall or cardiovascular mortality. For example, have a look at the Enhance trial.

Simvastatin with or without Ezetimibe in Familial Hypercholesterolemia
http://www.nejm.org/doi/full/10.1056/NEJMoa0800742

In fact, there was a slight increase (not significant) in cardiovascular events and plaque formation (as measured by carotid artery intima-media thickness) when Ezetrol was combined with a statin despite a statistically greater reduction in LDL. To me, this implies that it may not be only the LDL reductions that give statins the benefits. It may the result of anti-inflammatory properties, etc. (as also acknowledged by the authors). So, on the one hand, I like the fact that there isn't this obsession achieving LDL targets, but at the same time I think the risk calculator is a problem. You can play around with it and some have and here's an interesting finding one physician noted:

I'm not an anti-statin guy at all, but I will not be using this risk calculator. Take someone not too unlike myself: a 50yo white man with all optimal risk factor levels per the calculator (TC 170, HDL 50, SBP 110, not on treatment for HTN, not diabetic, not smoker) - such a person works out to have a lifetime ASCVD risk of 5%. The same man with an SBP of 119 has the same risk - 5%. But when his SBP is 120, it is 36%. Wow! A 7-fold relative risk increase due to 1mmHg blood pressure difference! Forget statins, apparently I only need to reduce BP by 1mmHg to get a 31% absolute risk reduction and a 720% relative risk reduction!

 

[SW VandeCarr]

I think the problem here is confounding factors. ………………………………………………………………………………………………………………………………………………………………..

To me, this implies that it may not be only the LDL reductions that give statins the benefits. It may the result of anti-inflammatory properties, etc. (as also acknowledged by the authors). So, on the one hand, I like the fact that there isn't this obsession achieving LDL targets, but at the same time I think the risk calculator is a problem. You can play around with it and some have and here's an interesting finding one physician noted:

While I agree that inflammatory processes indicated by elevated levels of C-reactive protein are correlated with risk, the 27 studies in question were randomized trials which makes it unlikely that true confounding is present. As I said, if there is to be a challenge to the interpretation of the meta-analyisis, I think one has to look at the details of the meta-analysis itself in terms of the methodology used to combine data from different studies.

 

[bohm2]

As I said, if there is to be a challenge to the interpretation of the meta-analyisis, I think one has to look at the details of the meta-analysis itself in terms of the methodology used to combine data from different studies.

Criticism has been made that the risk calculator overestimates risk in some groups:

As reported last week by heartwire , the guideline committee also attempted to validate the risk score in the Multiethnic Study of Atherosclerosis (MESA) and Reasons for Geographic and Racial Differences in Stroke (REGARDS) studies and found that it significantly overestimated risk.

"The big picture is that they should have slowed the process down," Ridker told heartwire . "Seeing that the risk model overestimated risk in two cohorts [MESA and REGARDS], in retrospect, they should have asked colleagues to externally validate the model to determine whether it overestimated risk in other cohorts."

As it stands now, the risk algorithm significantly overestimates risk in at least five contemporary patient cohorts. For one cardiologist, it is still not too late to call a time-out.

Dr Steven Nissen (Cleveland Clinic, OH) said he has long been an advocate of using intensive doses of statins in high-risk patients, so the question for him is not about the efficacy of the drugs for reducing morbidity and mortality. "But we have to treat the right patients, and the problem is that the risk calculator in the guidelines has never been previously published and therefore could not be independently verified," he told heartwire .

How Good Is the New ACC/AHA Risk Calculator?
http://www.medscape.com/viewarticle/814579

While agree that inflammatory processes indicated by elevated levels of C-reactive protein are correlated with risk, the 27 studies in question were randomized trials which makes it unlikely that true confounding is present.

What is interesting and not consistent with the C-reactive protein/anti-inflammatory hypothesis is that in the Enhance trial, the ezetrol + statin group vs statin alone group there was a significantly greater reduction in C-reactive protein (-49% vs -29%) and yet there was no cardiovascular benefit.

 

[SW VandeCarr]

Criticism has been made that the risk calculator overestimates risk in some groups:

How Good Is the New ACC/AHA Risk Calculator?
http://www.medscape.com/viewarticle/814579

What is interesting and not consistent with the C-reactive protein/anti-inflammatory hypothesis is that in the Enhance trial, the ezetrol + statin group vs statin alone group there was a significantly greater reduction in C-reactive protein (-49% vs -29%) and yet there was no cardiovascular benefit.

This trial is controversial. In any case, it studied the potential benefit of this drug combination in cases of familial hypercholesterolemia. I'm not sure how you compare this result with those in subjects without this condition and taking only a statin.

http://www.medscape.com/viewarticle/568763

Regarding the accuracy of the risk calculator, I have no comment. I was only addressing the important point of risk reduction in low risk subjects. If the risk calculator is in fact overestimating risk, it only enhances the finding of efficacy in "low risk" subjects. If the calculator is over estimating risk (and the proposed calculator is the one used in the meta-analysis) than "low risk" subjects should have had even lower risk.