lancet with huge study (1.8M) of 97 prospective cohorts from around the world, looking at the independent role (ie controlling for blood pressure, cholesterol and glucose) of BMI, overwt, and obesity on coronary heart dz (CHD) and stroke (see doi.org/10.1016/S0140-6736(13)61836-X). studies done from 1948-2005, with 2.7-57.5 yrs of follow-up. results:
–each 5 kg/m2 inc BMI was assoc with HR 1.15 (1.12-1.18) for CHD and HR 1.04 (1.01-1.08) for stroke, after adjusting for bp, chol, and gluc.
–these HRs suggest that 46% of excess risk of BMI for CHD and 76% of excess risk for stroke were mediated by these 3 risk factors (and 54% of BMI-attributable risk for CHD was not explained by them/24% for stroke).
–of the 3 risk factors, blood pressure was the major mediator of excess risk, accounting for 31% of the excess risk for CHD and 65% for stroke.
–no significant difference in % excess risk mediated by these risk factors when look at Asian vs western cohorts.
–compared to BMI 20-25, BMI 25-30 (overwt) had 50% of excess risk of CHD assoc with these 3 risk factors, and BMI>30 (obesity) had 44%. for stroke, it was 98% for overwt and 69% for obesity
–subgroup analysis for the 16 studies that also measured waist circumference found that the percentage of excess risk was 1-8% points higher
so, a few observations.
–BMI is a rather gross (?) instrument, looking at ratio of weight to height (in kg/m2 ), though we know that visceral fat (abdominal fat, and better reflected by waist circumference) is much more metabolically active, associated with both increased abnormalities of the traditional risk factors — htn, lipids, metabolic syndrome/diabetes — as well as other CAD markers, such as inflammatory markers. this study confirmed on subgroup analysis that waist circumference is more reflective of atherosclerotic risk (and it is waist circumference that is used in the assessment of metabolic syndrome. in the US, one criterion for metabolic syndrome risk is waist circ of 40″ men and 35″ women. note though that the WHO came out with international guidelines, with europeans being 37″ men and 31.5″ women, south asians 35.5″ men 31.5″ women….). most often if BMI>30, there is significant central obesity (and this is reflected, i think, in the finding above that obesity (vs overwt) has more excess risk not explained by the 3 traditional risk factors)
–adds further to the critique of the new AHA lipid guidelines (as well as the Framingham Risk Score, FRS), which only looks at the traditional risk factors and does not incorporate what seems to be other important cardiac risk factors (eg, waist circumference, physical fitness, psychosocial stress). there are some data suggesting that adding BMI to the FRS does improve risk score accuracy (see doi: 10.1161/01.CIR.0000161956.75255.7B), though it was never added. there have been assessments of adding some other risk factors to the FRS score, such as CRP, that have not found that they added much to the accuracy of the risk assessment. this huge, multinational study impressively suggests otherwise — there are very important non-traditional risk factors. and this supports the approach outlined in my critique of the AHA guidelines, suggesting the “gestalt” approach: running the FRS but moderating the result by your sense of other nonquantified risk factors, such as waist circumference/BMI, physical fitness, exercise, quality of diet, stress etc (ie, for those patients near the cutpoint of starting a statin by evaluation of the traditional risk factors, deciding whether to use a statin depending on these other factors).
geoff