This thesis investigates health trends among very old people and the allocation of public home-help services. A further aim is to examine methodological issues in mortality analysis. Three data sources are used: (1) The Tierp study of community-dwelling persons (n=421, ages 75+), (2) the SWEOLD nationally representative samples (n=537 in 1992 and n=561 in 2002, ages 77+), and 3) SNAC-K comprised of home-help recipients in a district of Stockholm (n=1108, ages 65+).

Study I suggests that the length of the follow-up period may explain some of the differences found in predictor strength when comparing mortality studies. Predictors that can change rapidly (e.g., health) were found to be strongest for the short term, with a lower average mortality risk for longer follow-ups. Stable variables (e.g., gender) were less affected by length of follow-up.

Studies II and III present a measure of complex health problems based on serious problems in at least two of three health domains. These were diseases/symptoms, mobility, and cognition/communication. Prevalence of complex health problems increased significantly between 1992 and 2002. Older age, female gender, and lower education increased the odds of having complex problems. Complex problems strongly predicted 4-year mortality. Controlled for age, gender, health, and education, mortality decreased by 20% between 1992 and 2002. Men with complex problems accounted for this decrease. Thus, in 2002 the gender difference in mortality risk was almost eliminated among the most vulnerable adults.

Study IV revealed that physical and cognitive limitations, higher age, and living alone were significantly related to home-help allocation, with physical and cognitive limitations dominating. Psychiatric symptoms did not affect the assessment.

The increased prevalence of complex health problems and increased survival among people with complex needs have important implications concerning the need for collaboration among service providers.