Changes in network composition among the very old living in inner London.

This article examines changes over time in social network composition among a sample of people aged 85 and over at baseline interview in 1987, who were followed-up two and a half to three years later. Almost half of respondents at follow-up had smaller networks than in 1987, with 19% having fewer relatives in their networks, 30% having fewer friends, and 26% also having fewer confidants. However, for most (84%) there was no change in whether they could name a main helper. Network density (integration) had declined for 30%, a substantial minority. For the remainder there were either increases or no changes. Most of those who needed help with tasks of daily living were given help, and while much of the help was given by relatives, by 1990 professionals were also providing a major part of the help. There were no associations with network change and health or functional ability, or with use of services. Those whose network size had increased were more likely to report at follow-up interview (in 1990) a need for (more) help. Multivariate analysis confirmed that while the greater part of the change in network size was accounted for by changes in numbers of relatives, it was closely followed as an explanatory variable by changes in numbers of friends, indicating that both variables (relatives and friends) account almost equally for changes in network size over time among very elderly people. The research presented here reports a considerable amount of change in network size and structure.