In the JournalsPerspective

Simple tool with ‘good accuracy’ predicts life expectancy in patients with dementia

Sara Garcia-Ptacek
Sara Garcia Ptacek

Using routinely collected patient characteristics, researchers said they were able to predict the life expectancy of patients with dementia with “good accuracy.”

Specifically, the tool is based on four characteristics obtained at diagnosis: the patient’s age, sex, comorbidity status and cognitive performance. It was designed with primary care physicians’ busy schedules in mind, according to Sara Garcia Ptacek, MD, PhD, of the Radboudumc Alzheimer Center in the Netherlands.

“We specifically designed a simpler version for PCPs because we wanted it to be applicable even in settings with time constraints,” she told Healio Primary Care. “Primary care is responsible for a substantial proportion of dementia diagnoses, so it was important to us to develop a tool useful in this setting.”

A second version of the tool, geared towards specialists, utilized the same factors as the tool for PCPs, but also factored in the specific subtype of dementia that the patient had.

The tools were validated in a cohort of 50,076 patients (mean age, 81.6 years; 59.4 women) from health centers in Sweden, including all memory clinics and about 75% of primary care clinics. Patients were followed starting in 2007 for a maximum of 9.7 years.

Researchers reported that by August 2016, 41.6% of the patients had died. Median survival time from dementia diagnosis was 5.1 years (interquartile range = 2.9-8) for women and 4.3 years (interquartile range = 2.3-7) for men. The tools yielded c indexes of 0.7 (95% CI, 0.69-0.71) to 0.72 (95% CI, 0.71-0.73) and showed good calibration, the researchers said. In addition, patients who were older, male, had an increased comorbidity burden and lower cognitive function at diagnosis, a diagnosis of non-Alzheimer’s dementia, lived alone and used more medications were more likely to die during the study period.

“It is important not to interpret results as absolutes,” Ptacek cautioned. “Survival probabilities are just that — probabilities — and patients with low survival probability may live long while patients initially classified as [having higher odds of survival] may die before 3 years.”

Researchers said the results are nearly identical to those of a similar study in the United Kingdom, but the tool needs to be validated in additional cohorts.

“The methods are simple enough, so we hope our colleagues develop their own tool for their own setting.” – by Janel Miller

Disclosures: The authors report no relevant financial disclosures.

Sara Garcia-Ptacek
Sara Garcia Ptacek

Using routinely collected patient characteristics, researchers said they were able to predict the life expectancy of patients with dementia with “good accuracy.”

Specifically, the tool is based on four characteristics obtained at diagnosis: the patient’s age, sex, comorbidity status and cognitive performance. It was designed with primary care physicians’ busy schedules in mind, according to Sara Garcia Ptacek, MD, PhD, of the Radboudumc Alzheimer Center in the Netherlands.

“We specifically designed a simpler version for PCPs because we wanted it to be applicable even in settings with time constraints,” she told Healio Primary Care. “Primary care is responsible for a substantial proportion of dementia diagnoses, so it was important to us to develop a tool useful in this setting.”

A second version of the tool, geared towards specialists, utilized the same factors as the tool for PCPs, but also factored in the specific subtype of dementia that the patient had.

The tools were validated in a cohort of 50,076 patients (mean age, 81.6 years; 59.4 women) from health centers in Sweden, including all memory clinics and about 75% of primary care clinics. Patients were followed starting in 2007 for a maximum of 9.7 years.

Researchers reported that by August 2016, 41.6% of the patients had died. Median survival time from dementia diagnosis was 5.1 years (interquartile range = 2.9-8) for women and 4.3 years (interquartile range = 2.3-7) for men. The tools yielded c indexes of 0.7 (95% CI, 0.69-0.71) to 0.72 (95% CI, 0.71-0.73) and showed good calibration, the researchers said. In addition, patients who were older, male, had an increased comorbidity burden and lower cognitive function at diagnosis, a diagnosis of non-Alzheimer’s dementia, lived alone and used more medications were more likely to die during the study period.

“It is important not to interpret results as absolutes,” Ptacek cautioned. “Survival probabilities are just that — probabilities — and patients with low survival probability may live long while patients initially classified as [having higher odds of survival] may die before 3 years.”

Researchers said the results are nearly identical to those of a similar study in the United Kingdom, but the tool needs to be validated in additional cohorts.

“The methods are simple enough, so we hope our colleagues develop their own tool for their own setting.” – by Janel Miller

Disclosures: The authors report no relevant financial disclosures.

    Perspective

    I want to start by commending Haaksma and colleagues for developing another survival prediction tool to use in patients with dementia. Recently, some of the major creators in this area have been licensing their tools and making it more expensive and difficult to use them clinically, so I’m pleased we will have more options to use in this area. For this reason alone, Haaksma and colleagues’ tool is not as difficult to use as, say, the Montreal Cognitive Assessment (MoCA). I'm also pleased that these researchers focused on creating a tool that primary care physicians can easily use, since that is where the vast majority of patients with dementia are seen.

    However, I'm also a little cautious. Any time you try to apply something clinically that's just emerging from research circles, applying it into daily practice can be challenging for several reasons. For example, your patient population can be different than the ones that were studied to develop the tool. One strategy to overcome this challenge is to apply the tool slowly in your practice, in a smaller way, so that you get each aspect of implementation right, and then slowly scale it into your entire practice. Another strategy is to find an office champion who agrees to try the entire tool, perfect its use and then introduce it across the broader practice.

    When PCPs discern an older patient’s survival risks using the Haaksma tool (or any similar tool) that indicates a limited prognosis, they may want to do what I do: focus on the health-related items that are the highest on the long list of health concerns. And if we can take care of those highest priorities, we'll keep moving down the list of priorities. Having said that, you also have to be consistently talking to patients about their priorities, since something further down your list of health risks may be very high on theirs. An example of this might be arthritis, which is very rarely life-limiting, but affects quality of life — consider someone with hand arthritis who plays piano or works on the computer. That might be really, really important to them, although not life-threatening. And then tell the patient, well, let's move that one up the list for you and we'll move something else down the list. And if I am able to improve the function and pain, you can try to decrease smoking. By knowing what you and the patient consider highest priorities, you are both working on important things and, at the same time, it helps build the trust and rapport with your patient, something called “shared decision-making.”


     

    • William Dale, MD, PhD
    • Director, Center for Cancer and Aging Research
      Arthur M. Coppola Family Chair, department of supportive care medicine
      City of Hope, Los Angeles

    Disclosures: Dale reports no relevant financial disclosures.