Background An association between hematocrit (Hct) and coronary heart disea
se (CHD) mortality has been previously observed. However, the relationship
of Hct and CHID independent of other cardiovascular disease (CVD) risk fact
ors and differences between men and women remain unclear.
Methods We examined the association between Hct and CHD mortality with Cox
regression analyses of data from 8896 adults, aged 30-75 years, in the Seco
nd National Health and Nutrition Examination Survey (NHANES II) Mortality S
tudy (1976-1992). Covariates included age, sex, race, education, smoking st
atus, hypertensive status, total serum cholesterol, body mass index, white
blood cell count, and history of CVD and diabetes. Hct was categorized by u
se of sex-specific tertiles, and all analyses were stratified by sex.
Results During 16.8 years of follow-up, there were 545 (men 343, women 202)
deaths from CHD (international Classification of Diseases, 9th revision [I
CD-9] 410-414), 778 (men 426, women 279) deaths from diseases of the heart
(ICD-9 390-398, 402, 404, 410-414, 415-417, 420-429), and 2046 (men 1216, w
omen 830) all-cause deaths. Among men, the crude CHD mortality rate per 10,
000 population was 42.6, 31.9, and 46.3 among those with Hct in the lower,
middle, and upper tertiles, respectively. The corresponding crude CHD morta
lity rates among women were 12.6, 18.6, and 27.7. After adjustment for age
and other CVD risk factors, there was no association between Hct in the upp
er tertile compared with the lower tertile and mortality from either CHD, d
iseases of the heart, or all causes among men. Women with Hct in the upper
tertile were 1.3 times (95% Cl 0.9-1.9) more likely to die from CHD than we
re women with Hct in the lowest tertile, after multivariate adjustment. The
effect of high Hct on CHD mortality among women younger than 65 years of a
ge was slightly stronger (relative risk 2.2, 95% Cl 1.0-4.6).
Conclusions These results suggest that the association between Hct and mort
ality from CHD and all causes is complex, differing both by sex and age. Fu
rther research is needed to gain a better understanding of these age and se
x differences.