Factors associated with mortality in older people.

A random sample of older people from Edinburgh (215 men and 272 women aged 62--90 years) was examined clinically and by a questionnaire. Various measurements were made. Five years later, mean values of measurements were compared in those who had died and in survivors. Where significant differences occurred, regression techniques were used to separate age and mortality effects. Variables in which death was the predominant independent variable in the regressions were body weight, bi-iliac diameter, FEV1.0, transverse chest diameter, index of kyphosis, leucocyte ascorbic acid and some nutrient intakes in men plus transverse cardiac diameter and leucocyte ascorbic acid in women. Apart from index of kyphosis in men and cardiac diameter in women, mean values were significantly larger in survivors. Dichotomous variables from questionnaire and examination significantly related to mortality were 'possible' ischaemic heart disease in women, diastolic hypertension in men, persistent cough in men and dyspnoea worse than grade 2 in men and women. Cigarette smoking had no mortality effects in this study.