Predictive Models Flag Need for Nursing Home Care in Dementia Patients
New models to predict need for nursing home level of care in community-dwelling older adults with dementia may help enhance care management.
Source: Thinkstock
By - Predictive models using self-report and proxy data can accurately forecast need for nursing home level of care (NHLOC) in community-dwelling older adults with dementia, according to a study published this week in JAMA Internal Medicine.
The researchers noted that many older adults living with dementia eventually need NHLOC, as the condition is a leading cause of nursing home placement. The Alzheimer's Association estimates that as of March 2023, approximately 50 percent of nursing home residents have a dementia diagnosis.
However, predicting which dementia patients may need to enter a nursing home can be challenging, as the decision relies heavily on a combination of factors, including levels of functional impairment, sociocultural considerations, and environment.
The research team underscored that the complexity of the decision makes accurately estimating when a dementia patient might need NHLOC extremely valuable.
To improve this process, the researchers developed two models designed to predict need for NHLOC among older adults with probable dementia using self-reported and proxy data.
The research team defined need for NHLOC as “(1) [three] or more activities of daily living (ADL) dependencies, (2) [two] or more ADL dependencies and presence of wandering/need for supervision, or (3) needing help with eating.”
The models were developed using data collected from 3,327 participants with probable dementia between 1998 and 2016 in the Health and Retirement Study. Each model was validated using data from 1,712 participants from 2011 to 2019 in the National Health and Aging Trends Study. All participants were community-dwelling adults with probable dementia aged 65 or older.
Predictors for both models included demographic data, chronic conditions, behavioral and health factors, and functional measures. Model performance was evaluated in terms of discrimination and calibration, which were measured using integrated area under the receiver operating characteristic curve (iAUC) and calibration plots.
Just over 63 percent of participants from the Health and Retirement Study were classified as needing NHLOC at the end of follow-up.
For both the proxy model and the self-respondent model, predictors included driving status, baseline ADL and instrumental ADL dependencies, and age. The proxy model also utilized body mass index and falls history, while the self-respondent model leveraged date recall, female sex, and incontinence to make its predictions.
Both models achieved high performance.
Optimism-corrected iAUC following internal validation was 0.72 in the proxy model and 0.64 in the self-respondent model. On external validation, using data from participants in the National Health and Aging Trends Study, iAUCs in the proxy model and self-respondent models were 0.66 and 0.64.
These findings led the researchers to conclude that relatively simple predictive models using proxy or self-reported data can accurately forecast need for NHLOC in this patient population, which may help guide care planning discussions.
This research is just one of the efforts being made to improve dementia care using data analytics.
In a March interview with HealthITAnalytics, Jared Brosch, MD, a neurologist at Indiana University (IU) Health and assistant professor of clinical neurology at IU School of Medicine, and Phyllis Ferrell, global head of external engagement for Alzheimer’s disease at Eli Lilly & Company and director of the DAC Healthcare System Preparedness initiative, detailed how a new pilot program at the health system is leveraging artificial intelligence (AI) and digital screening tools to drive the early detection of cognitive impairment in primary care settings.
