Predicting Nursing Home Care for Dementia Patients Is Possible With Models 

With dementia being one of the leading indications for nursing home placement, developing and perfecting models to predict level of care can be crucial when patients and families – and nursing homes – dive into future planning.

Simple self-reporting or proxy responses by patients on models are good ways to estimate if and when a possible dementia patient needs a nursing home level of care (NHLOC), according to a study published in JAMA Network.

“This prognostic study showed that relatively simple models using self-report or proxy responses can predict the need for NHLOC in community-dwelling older adults with probable dementia with moderate discrimination and excellent calibration,” the researchers concluded.


Predictors for needing care were based on models that factored in age, baseline activities of daily living (ADL) and instrumental ADL dependencies, along with driving status.

The proxy model, or someone answering for the dementia patient, also included questions on body mass index and falls history. The self-response model added a question on if the patient was female, had incontinence and good data recall, according to the study.

The study has wide implications. Typically, about 50% of nursing home residents have a dementia diagnosis, and for those diagnosed with dementia at age 70, nursing home admission would be expected for roughly 75% within a decade, researchers found.


And, high nursing home rates are related to functional impairments, medical comorbidities and behavioral symptoms like wandering or hallucinations, which makes home care more challenging.

“In clinical practice, those with early-stage dementia will start out as the primary source of medical information. Eventually, persons with dementia will progress to where a surrogate becomes the primary source of information,” researchers said. “Use of these models should mirror this clinical reality, with the self-respondent model used in early-stage dementia and proxy model used in later stages.”

When self-respondent and proxy information is available, researchers turned to proxy only when the patient was unable or unwilling to provide information.

“While the final predictors in these models are slightly different, they have considerable overlap, with age and baseline ADL/IADL dependencies being prominent prognostic factors,” researchers noted.

And, such models could provide a standardized metric applicable across cultural and socioeconomic backgrounds, and be used for families that want to estimate when care should be provided at home, according to the study.

“Deciding when to enter [a nursing home] is complicated and often influenced by sociocultural and systems/environmental factors,” researchers wrote. “Individuals with similar levels of functional impairment may have dramatically different [nursing home] admission times based on cultural beliefs, financial considerations, presence of family and caregiver support, occurrence of crisis events, and … differential access to in-home services and federal/state benefits.”

Essentially, there’s complexity surrounding such a decision, and providing an estimate of when an individual with dementia should be expected to need nursing home level of care based on functional impairments along with behavioral issues can guide families toward the right care setting.

Within a population, predictors for each model could identify early on those patients that could most benefit from in-home and community-based services as well.

For developing and validating the models, researchers leveraged two nationally representative prospective cohorts, the Health and Retirement Study (HRS) and the National Health and Aging Trends Study (NHATS), to develop and validate models to predict need for nursing home level of care among older adults with probable dementia.

Companies featured in this article: