[A health survey of elderly people in Reykjavik]

In 1982 a multifactorial population health survey of people aged 80 years and older and living in the community, was carried out. The mortality registry was examined up to the end of year 1988. By this time 67% had died and the mortality was assessed by Cox's regression analysis on serum cholesterol as one of the main variables included in the survey. The original sample contained 170 persons and 148 were alive at the time of examination. Participation rate of sample is 72%. By end of 1988, 35 persons were alive and 71 dead. Excluded were 22 dead before entry. Serum cholesterol was found to be negatively predictive of mortality in both univariate (p < 0.01) and multivariate analysis when triglycerides, age and smoking were also included (p < 0.01). There was a 0.9% decrease in relative risk for each mg/dl increase of serum cholesterol. Ten persons had serum cholesterol of 160 mg/dl or less and they all died within 10 years from the examination, one of disseminated carcinoma and the rest of non-cancer causes. The strong inverse relationship between mortality and serum cholesterol indicates that this risk factor in old people may be of different nature than in other age groups.