Identification of those at greatest risk of harm among cognitively impaired people who live alone

Our purpose was to determine whether we could predict, from among seniors w ith cognitive impairment who lived alone, those at highest risk of experien cing harm or needing emergency services. Specifically, our focus was on the problems that occurred as a result of negligence due to cognitive impairme nt. We assessed 139 people, 65 years of age and older, who lived alone in a large urban centre and who showed cognitive impairment on a screening test . On average, participants were 83-year-olds, had 11 years of education, an d a Mini-Mental State Score of 23. A total of 39% met diagnostic criteria f or dementia. All underwent assessment of their neuropsychological abilities , behavioural characteristics and social supports. A study informant was id entified for each participant who, together with the family physician, prov ided detailed information regarding the participants' use of emergency medi cal, fire or police services, their experience of harm or loss of property over the last 12 months. We determined that negligence due to cognitive imp airment caused harm or the need for emergency services based on informants' and primary care physicians' reports and by three independent raters. Logi stic regression analyses indicated that the best predictors of harm or use of emergency services were male sex, self-care deficits, and poor performan ce on the Trail Making Test, Part B. The sensitivity of this three-variable regression model was 81% and the specificity was 66%. The likelihood ratio of both the positive and negative tests for harm represented a small but m eaningful change in pretest to posttest probability of harm. These results indicate that we can estimate an individual's probability of harm based on performance on these three variables. The next step is to confirm the findi ngs with prospectively collected data.