A prediction tool developed by U.S. researchers could help decide treatment and future care decisions in older adults living outside nursing homes with suspected dementia by predicting those who will die within the next decade.
The mortality prediction tool identified several easily available clinical predictors of death within this time, such as mobility and low body mass index (BMI), and was able to identify approximately three-quarters of community-dwelling individuals who subsequently died within 10 years.
Reporting in the journal JAMA Internal Medicine, investigators say the estimates could be used by clinicians, patients, and families to help frame discussions around the clinical management of medical comorbidities, preventive screening, and advance care planning.
“An estimate of an individual’s prognosis can be an important factor in financial planning for families, particularly as many people with dementia need increased support at home and are ultimately admitted to nursing homes,” explained researcher W. James Deardorff, a geriatrician at the University of California at San Francisco.
“Additionally, individuals with limited life expectancy may wish to focus on quality of life and being comfortable, rather than trying to live as long as possible. This may lead them to forego certain interventions, such as cardiopulmonary resuscitation in the event of a cardiac arrest.”
Several clinical guidelines related to cancer screening and chronic disease management now encourage the use of life expectancy when making treatment decisions.
To better determine this in the context of dementia, the team studied participants from two U.S. nationally representative cohorts aged 65 years and old who likely had the condition, based on a previously validated algorithm.
The study included a total of 6671 community-dwelling older adults overall, with the model created from 4267 participants from the 1998-2016 Health and Retirement Study and validated in 2404 Medicare beneficiaries from the 2011-2019 National Health and Aging Trends Study.
Participants were followed for a median of 3.9 years, with 81.2% dying by end of follow-up.
The model showed older age was strongly predictive of death between 1 and 10 years, with female sex a protective factor and prognosis often driven by comorbid disease.
The final model included age, sex, BMI, smoking status, the degree of dependency in activities of daily life, such as dressing, bathing or eating, and difficulty in instrumentally performing activities of daily life such as preparing meals and managing money, medications or grocery shopping.
It also included difficulty walking several blocks, participation in vigorous physical activity, and chronic conditions including cancer, heart disease, diabetes and lung disease.
On external validation, the area under the receiver operating characteristic curve (AUC) was 0.73 at 1 year and 0.74 at 5 years, with good calibration across the range of predicted risk from 1 to 10 years.
For example, a woman in her 80s with a BMI of 19.5 who never smoked, was dependent for three activities in daily life and had four difficulties in instrumental activities in daily life, who also had difficulty walking, had a median time to death of 3.8 years, with a predicted risk of death of 12% at one year, 64% at five years, and 94% at 10 years.
The researchers conclude: “For interventions that have immediate risks or burdens and delayed benefits, such as cancer screening and tight glycemic control in patients with diabetes, life expectancy can help differentiate between patients most likely to benefit vs most likely to be harmed by interventions.
“Therefore, accurate estimates of life expectancy in people with dementia are important for providing the foundational data for patients, family, and clinicians to help define patient-centered care goals in their remaining lifetime.”
