ASSESSMENT OF PREVALENCE OF LEFT-VENTRICULAR HYPERTROPHY IN HYPERTENSION

The reported prevalence of left ventricular hypertrophy (LVH) in human hypertension is much lower than that among animals with experimental hypertension. With current methods of determining left ventricular mas s by M-mode echocardiography, the standard error of a single estimate is high and consequently so is the SD of the population distribution. This accounts for the large overlap in individual values of left ventr icular mass index (LVMI) between hypertensive and normotensive groups. The high SD is due to the use of the cube algorithm for relating meas urements made in a single plane to the whole left ventricle, and to th e difference between actual and assumed left ventricular geometries, T hese are not problems with nuclear magnetic resonance imaging, which p rovides information about the entire left ventricle without assumption s about geometry. M-mode echocardiography is well suited for estimatin g differences between mean LVMI values for groups of subjects but it u nderestimates the prevalence of LVH, In most series only about 30% of hypertensives have been reported to have LVH, The estimated prevalence of structural remodelling is increased to 50-60% of the same group of subjects when 'low-SD' measurements such as wall thickness and the wa ll thickness:internal radius ratio are employed. The estimated prevale nce of LVH and remodelling is still greater with multivariate discrimi nant function analysis, with which it is found in about 70% of hyperte nsives. Overall, the data suggest that prevalence of LVH in establishe d hypertension is high. The 30% of subjects reported to have LVH on th e basis of LVMI measurements that are beyond the limits of the control group probably have the most severe changes. The inability to detect lesser grades of left ventricular remodelling reliably is due to the w ay LVMI is derived by echocardiography, rather than to intrinsic inacc uracies. It suggests that existing approaches should be supplemented b y greater use of 'low-SD' variables and discriminant functions. Detect ing the full spectrum of left ventricular structural changes in indivi duals with hypertension is needed for risk assessment and, increasingl y, for management aimed at minimizing irreversible myocardial damage. Nuclear magnetic resonance imaging provides 'global' and more accurate information about left chamber structure than does M-mode echocardiog raphy but its cost at present is much greater. Nevertheless, the infor mation provided by echocardiography may be adequate for the above appl ications, but the high SD of LVMI is a weakness. Greater use of 'low-S D' variables and multivariate discriminant functions may help overcome this problem. (C) 1998 Lippincott-Raven Publishers.