A few studies examine the time evolution of delirium in long-term care (LTC) settings. In this work, we analyze the multivariate Delirium Index (DI) time evolution in LTC settings.

The multivariate DI was measured weekly for six months in seven LTC facilities, located in Montreal and Quebec City. Data were analyzed using a hidden Markov chain/latent class model (HMC/LC).

The analysis sample included 276 LTC residents. Four ordered latent classes were identified: fairly healthy (low “disorientation” and “memory impairment,” negligible other DI symptoms), moderately ill (low “inattention” and “disorientation,” medium “memory impairment”), clearly sick (low “disorganized thinking” and “altered level of consciousness,” medium “inattention,” “disorientation,” “memory impairment” and “hypoactivity”), and very sick (low “hypoactivity,” medium “altered level of consciousness,” high “inattention,” “disorganized thinking,” “disorientation” and “memory impairment”). Four course types were also identified: stable, improvement, worsening, and non-monotone. Class order was associated with increasing cognitive impairment, frequency of both prevalent/incident delirium and dementia, mortality rate, and decreasing performance in ADL.

Four ordered latent classes and four course types were found in LTC residents. These results are similar to those reported previously in acute care (AC); however, the proportion of very sick residents at enrolment was larger in LTC residents than in AC patients. In clinical settings, these findings could help identify participants with a chronic clinical disorder. Our HMC/LC approach may help understand coexistent disorders, e.g. delirium and dementia.

The delirium index (DI) is a valid measure of delirium severity. We proposed to describe longitudinal patterns of severity scores in older long-term care (LTC) residents.

A prospective cohort study of 280 residents in seven LTC facilities in Montreal and Quebec City, Canada, was conducted. DI, Barthel Index, Mini-Mental State Examination, Charlson Comorbidity Index, Cornell Scale for Depression in Dementia, dementia assessment by an MD, and prevalent or incident probable delirium defined according to the Confusion Assessment Method were completed at baseline. The DI was also assessed weekly for 6 months. Demographic characteristics were abstracted from resident charts. Cluster analysis for longitudinal data was used to describe longitudinal patterns of DI scores.

During the 24 weeks following enrolment, 28 (10.0%) of 280 residents who had prevalent delirium and 76 (27.1%) who had incident delirium were included in our analysis. Average observation period was 18.3 weeks. Four basic types of time evolution patterns were discovered: Improvement, Worsening, Fluctuating, and Steady, including 22%, 18%, 25%, and 35%, of the residents, respectively. With the exception of the Worsening pattern, the average trajectory was stabilized at the 4th week or earlier. Poor baseline cognitive and physical function and greater severity of delirium predicted worse trajectories over 24 weeks.

The longitudinal patterns of DI scores found in LTC residents resemble those found in an earlier study of delirium in acute care (AC) settings. However, compared to AC patients, LTC residents have a smaller DI variability over time, a less frequent Improvement pattern, and more frequent Worsening and Fluctuating patterns.