yet another analysis that lipid treatment by new guidelines overdoes it

Cook and Ridker initially challenged the AHA guidelines on lipid management by looking at 3 datasets showing that they dramatically overpredicted cardiac events by 75-150% through the new AHA risk calculator. they just published a further analysis of the Women's Health Study (see chol AHA guideline womens health study JAMA intmed 2014 in dropbox, or doi:10.1001/jamainternmed.2014.5336). The Women's Health Study was an aspirin/vitamin E intervention trial of a nationwide cohort of 27,542 US women aged 45-79 (ave age of 54.2, 1.8% Black), initially free of  cardiovascular disease, cancer or other major illnesses (overall healthy cohort: 11.6% smokers, 2.5% diabetic, 13.5% on antihypertensives). these women had lipids and other risk factors assessed and were followed 10.2 years. results:

    --average cholesterol 212, LDL 124, HDL 54
    --632 women had a cardiovasc event over the course of the followup (including 250 MIs, 302 strokes)
    --this observed risk of 2.2% was at odds with the predicted risk of 3.6% anticipated by the risk calculator
    --for different risk groups, the ratio of predicted risk to observed values were (and adjusting for statin use and revascularization, which increased over time):
            --ratio of 1.90 in groups with 0-5.0% risk, with adjustment for statin use/revascularization: 1.84
            --ratio of 1.91 in groups with 5.0-7.5% risk, with adjustment for statin use/revascularization: 1.80
            --ratio of 1.43 in groups with 7.5-10.0% risk, with adjustment for statin use/revascularization: 1.33
            --ratio of 1.45 in groups with >10.0% risk, with adjustment for statin use/revascularization: 1.35

so, further validation of the nonvalidation of the AHA/ACC risk assessment tool... at least 6 prior validation cohorts have shown significant overassessment of predicted risk. this other large cohort adds to the list. from Cook and Ridker's prior analysis (see Lancet 11/30/13), this overassessment means that, by this new AHA calculator, an additional 46 million middle-aged Americans without known cardiovasc disease should get statins (including pretty much all males >60 years old, independent of their lipids).

for those of you who have been on the listserve for several years, you have been exposed to my rantings about the AHA/ACC guidelines. in brief, i think they 
    --dismiss using a targeted LDL goal by holding this question to the highest standard (ie, having a couple of really good randomized controlled trials with this as the primary outcome assessed, which just do not exist for this issue), despite impressive secondary analyses that the achieved LDL  does predict outcomes -- see 3rd point below. 
    --leave HDL completely out of the picture. again, no  studies done showing that if treat to HDL goal  that it alters the outcome: this is a difficult issue since statins have the best evidence of clinical efficacy by far but don't affect HDL much, it is hard to just treat HDL without affecting other lipids, and there are some pro-inflammatory, bad-acting HDLs which confuse things. for example, i discount the drugs which are CETP inhibitors: they do raise the HDL but do not have clinical benefit. maybe there are other effects of CETP inhibitors which are harmful??? however, there are old studies of infusing HDL itself, with significant changes in the size and lipid content of atherosclerotic plaques. so, i would argue that we just don't have great drugs yet that improve HDL, but that it has some role in determining how aggressive to be in treatment. see next point.
    -- in terms of both the targeted LDL and HDL points above,  the Treating to New Targets study comparing atorvastatin 10mg vs 80mg showed that the dose of atorvastatin did not matter, just that both the achieved LDL and HDL did matter. in fact an achieved LDL of >100 with an HDL >55 had a slightly lower 5-yr risk of major cardiovasc events than an LDL <70 with HDL <38; and within each HDL quintile, the achieved LDL correlated with cardiac events.
    --but, at the same time they are touting the importance of RCTs (which are useful...), they have many recommendations for which there are really no data: testing people as young as 20 (which i agree with, but no data on their clinical outcomes, vs starting therapy in the mid 40s), their risk assessment tool has not been fully validated (as noted by cook and ridker above), they suggest treating all diabetics over age 40 (ie, they would treat someone with baseline LDL of 70 and HDL of 70, which i do actually occasionally see. not that it is wrong to treat, but there are no RCTs on diabetics with really favorable lipid profiles on which to rigorously back their recommendation)...
    --so, seems to me, you can't have it both ways: demand rigorous RCTs for some recommendations and not even raise the issue on others.