Cross-sectional study.
NHs (175) located in France.
A total of 5684 subjects who were living in the NH for at least 1 year.
Information on NHs' characteristics and on NH residents' health was recorded by NH staff. Participants were divided in 3 groups according to their dementia status: diagnosed dementia, potential dementia without formal diagnosis, and nondementia. The outcome measure was a binary variable: ED visits in the last 12 months (yes vs no). A mixed-effects logistic regression was performed on ED visits accounting for the random effects of living in a particular NH.
From the 5684 participants, 1036 had been seen in the ED. Adjusted odds ratio (AOR) showed that having a potential dementia without formal diagnosis, compared with a diagnosed dementia, was associated with an increased probability of ED visits (AOR聽= 1.25, 95% confidence interval: 0.99-1.59, P聽= .061); however, when a random NH effect was entered into the model, the association between potential dementia without formal diagnosis and ED visits disappeared (AOR聽= 1.22, 95% confidence interval: 0.95-1.57, P聽= .11).
The association of potential dementia without formal diagnosis with ED visits varies across NHs. This intra-NH aspect (eg, organization and care habits) should be taken into account when examining the rates of hospitalization and possibly the use of health care services in general among NH residents.