Predicting coronary heart disease from body measurements of Framingham males

Abstract

Among 1625 Framingham (Mass.) Study males, aged 35–69 when measured, ratings of somatotype (body build) and gynandromorphy (morphological femininity in the male physique) were calculated from 7 body measurements. The 198 men who developed coronary heart disease (CHD) within 12 yr of entry into the Study were significantly more endomorphic, or fat, and less ectomorphic, or linear and fragile, than the 1427 men who did not. The observed effect was confined to the 55 men aged 35—49, and within this group, to the 26 men with angina pectoris, coronary insufficiency, or sudden death (CHD-NMI). These 26 men were significantly (0.01 > p > 0.001) more endomorphic, more mesomorphic (muscular, bony), and l than the 29 men with myocardial infarction (CHD-MI) or those without CHD (NCHD). Gynandromorphy did not distinguish among any of the age or disease categories. In the 35–49 yr age group, a 3-way discriminant analysis regressed body measurements directly on disease status (NCHD, CHD-MI, CHD-NMI), bypassing the intermediate somatotype ratings. The function discriminating CHD-MI from the other two groups was not significant; that discriminating CHD-NMI was highly significant ( p < 0.001). Men scoring above various cut-off points on this latter function had risks of developing CHD-NMI between 4:1 and 7:1 relative to those scoring below (“relative risk,” or likelihood ratio). The “risk ratio” (predictive value for CHD-NMI/predictive value for NCHD plus CHD-MI) was 2.6, higher than published risk ratios for serum cholesterol (2.1) or blood pressure levels (1.3) among all Framingham Study males. Of the 7 body measurements taken singly, weight, chest depth, arm girth, and subscapular skinfold, but not height, triceps skinfold, or hand grip strength distinguished among the 3 CHD-outcome groups at highly significant levels. In conjunction with other measurements, however, in the discriminant function which distinguished subjects with CHD-NMI, triceps skinfold and grip strength also became significant, with negative coefficients. This pattern—positive association of CHD-NMI with weight, chest depth, arm girth, and subscapular skinfold, but negative association with triceps skinfold—showed, as did somatotype, that mesomorphy as well as adiposity contributed to the risk of CHD other than myocardial infarction. As for prediction of new cases, 6 men aged 35–49 when measured developed CHD-NMI between the twelth and fourteenth year of observation. Their mean score on the appropriate discriminant function was in the expected direction and close to that for the previous 26 men with CHD-NMI. As in earlier reports, physique contributed to the risk of coronary heart disease independently of blood pressure and serum cholesterol. The relative risk, 3:1 for physique with normal levels of blood pressure and serum cholesterol, rose with elevated levels of these other risk factors. Comparison of the present findings with those in the literature disclosed a pattern which resolves many of the apparent contradictions. When studies were grouped according to sample characteristics—population or selected sample—age of subjects, and manifestation of CHD, the differences long observed clinically, epidemiologically, biochemically, and pathologically between CHD in early and late adult life, and between male survivors of MI and those with other manifestations of CHD, were confirmed from the distinct standpoint of physique.