Risk Predictions Can Guide Alzheimer's Treatment: Large Study of Cost and Mortality Suggests That More Emphasis Should Be Placed on Palliative Care

Article excerpt

It is possible to figure out which Alzheimer's patients are more likely to die within a year, according to a study by Robert J. Newcomer, Ph.D., and colleagues. And having that information might lead to more palliative care and less-aggressive treatment of illness.

"With fairly simple information, you can put people in groups that can identify them as being at relatively higher risk" of mortality, Dr. Newcomer, professor of social and behavioral science at the University of California, San Francisco, said in an interview. Physicians can consider placing Alzheimer's patients with higher risk of dying in the next year into palliative care sooner rather than later.

Dr. Newcomer and colleagues studied 2,255 patients with irreversible dementia. All patients were enrolled voluntarily and all had two or more limitations in activities of daily living (ADL).

The researchers developed a "mortality risk index" based on factors such as sex, age, number of previous inpatient stays in the last few months, number of ADL with which the patient needed maximum help, caregiver relationship, and presence of specific chronic conditions, including cancer, congestive heart failure, and diabetes.

The investigators stratified patients by the mortality risk and calculated their monthly costs of nonphysician medical care. 'At a mortality risk score high enough to correctly identify half those who died (236 of 471), the model correctly identified 75% of those who lived (1,363 of 1,598). The average monthly expenditure total for those predicted to die was $1,477, compared with $732 for those predicted to live, Dr. Newcomer said in an interview.

"Prediction of mortality became more accurate when we used a lower mortality-risk score, but the percentage of those correctly predicted to live declined. Differences in average monthly expenditures declined between those predicted to die, compared with those predicted to live, but they remained significant," he noted.

In their article, the investigators noted that "the findings provide some evidence that a predictive classification may be possible. Across all levels of mortality risk, patients predicted to be at high risk of death had average monthly Medicare expenditures that were significantly higher than for subjects predicted to be at low risk of death. …