Objective: The prognostic value of haemoglobin within normal references is seldom emphasized. The relationship between haemoglobin and mortality has been questioned because of the possible confounding of other risk factors. We investigated whether there was a curve linear relationship between haemoglobin and total mortality, and evaluated the possible modifying effects of smoking, body mass index, total cholesterol and systolic blood pressure.
Materials and methods: In all, 6541 men aged between 20 and 49 years were examined in 1974 in a prospective, population-based study from the municipality of Tromsø, Northern Norway. During 20 years of follow-up (127 120 person-years), 495 deaths were identified.
Results: We found a U-shaped relationship between quintiles of haemoglobin and total mortality. Among the 35-49 years group, the multiple adjusted hazard ratios (95% CI) were 1.83 (1.31-2.57) in quintile 1 and 1.72 (1.23-2.41) in quintile 5, compared to quintile 3 of haemoglobin. Compared to the age-adjusted hazard ratios, the multiple adjustments tended to non-significantly enhance the association in the lowest quintiles and non-significantly attenuate the association in the highest quintiles. The relationship was most pronounced in smokers in a dose-response manner, but also present in non-smokers.
Conclusions: High and low haemoglobin levels have an independent prognostic effect on mortality, although a possible effect of residual confounding cannot be ruled out. Smokers in quintile 1 and quintile 5 of haemoglobin were at increased risk of dying.