The relationship between living arrangements and higher use of hospital care at middle and older ages: to what extent do observed and unobserved individual characteristics explain this association?

Background Previous research has shown that certain living arrangements, such as living alone, are associated with worse health at older ages. We assessed the association between living arrangements and hospital care use among middle-aged and older adults, and investigated to what extent observed and unobserved individual characteristics explain this association. Methods Longitudinal Finnish registry data for men and women aged 50–89 years were used for the period 1987–2007. The relationship between living arrangements (based on whether an individual lived with a partner, other adults or alone, and whether they lived with minor/adult children) and heavy hospital care use (i.e., having been in hospital for 8 or more days in a year) was studied. First, we applied logistic regression models and linear probability models controlling for observed time-invariant factors (socioeconomic status measured by education, labour force status, and household income; and marital status), and then individual linear probability models with fixed-effects to further account for unobserved time-invariant individual characteristics in the measurement period. Analyses were done separately for 10 year age-groups. Results In the logistic regression models, men and women who lived alone had higher crude odds of heavy hospital care use than those living only with their partner. These odds ratios were highest for men and women in the youngest age category (50–59 years, 1.72 and 1.36 respectively) and decreased with age. Adjusting for observed time-invariant socioeconomic status attenuated these odds by 14–40%, but adjusting for marital status did not affect the results. Lower odds were observed among adults aged 50–59 years who lived with their partner and (minor or adult) children. But odds were higher for individuals aged 60–79 years who co-resided with their adult children, regardless of whether they lived with a partner. Adjusting for observed time-invariant factors generally did not change these results. After further adjusting for unobserved time-invariant individual characteristics in the individual fixed-effects models, most of these associations largely attenuated or disappeared, particularly for ages 80–89 years. Conclusions The association between living arrangements and higher use of hospital care at middle and older ages is largely explained by socioeconomic disadvantage and unobserved time-invariant individual characteristics