September 3, 2019

photograph of a younger female nurse sitting with an older female patient as her vitals are read

Measuring health is important for many reasons. It can help doctors and scientists understand the risk of medical problems and develop prevention strategies that can improve patient care. Monitoring health status can also help economists understand financial outcomes and help policymakers identify the likelihood of people needing caregiver assistance or retiring early, life events that can have implications for programs such as Social Security, Medicare, and Medicaid.  Further, measuring health is essential for assessing the return on U.S. health care spending which is large—close to one fifth of U.S. gross domestic product—and growing.

In the United States, people usually take surveys that allow assessment of physical well-being. Self-assessments of health can help forecast life expectancy and functional ability, and whether a person may require medical care at some point in the future. However, in some cases, a better measure of health than self-assessments might be necessary.

Enter the frailty index
Federal Reserve Bank of Atlanta Research Economist and Associate Adviser Karen Kopecky
Karen Kopecky

In June 2019, the Atlanta Fed published a working paper cowritten by Karen Kopecky, a Federal Reserve Bank of Atlanta research economist and associate adviser. Kopecky and her coauthors discussed the frailty index, an alternative method of evaluating health. This measure, pioneered by researchers at Dalhousie University in Halifax, Nova Scotia, focuses on the total number of health ailments a person has and the nature of those problems.

Kopecky worked with researchers Roozbeh Hosseini, a visiting scholar at the Atlanta Fed who is also an assistant professor at the University of Georgia, and Kai Zhao, associate economics professor at the University of Connecticut, to create frailty indexes using three surveys of Americans that include a host of questions on various aspects of health conditions.

A key finding of the researchers' work was that the proportion of individuals in the U.S. population in good health decreases faster as people age when well-being is measured with the frailty index rather than with individual self-assessments. "For this reason the frailty index is an especially good measure for studying how health evolves with age," Kopecky said.

The architecture of the frailty index helps explain why it can be a better predictor of health during aging. The index combines information from a range of questions about an individual's specific health ailments to provide a summary of the person's overall well-being. Kopecky and her colleagues used 27 health variables to construct a frailty index for a sample of more than 18,500 Americans who responded to the Panel Study of Income Dynamics (PSID) from 2003 to 2015. The survey includes questions on specific medical conditions and activities of daily living. The variables the researchers looked at include difficulty with activities such as eating, dressing, walking, managing money, and getting in and out of bed, as well as the presence of conditions including cancer, diabetes, heart attack, stroke, and loss of memory.

The researchers derived the index by adding the total number of variables reported as ailments by an individual, then dividing that sum by the total amount of variables observed for that person overall in the year. The index captured expected variation in health: frailty was higher in older age groups compared with younger ones. Further, the sample showed that increases in frailty over time were three times more common than decreases.

Kopecky and her coauthors also compared the state of health over time using the frailty index with self-reported health status by making calculations based on the percentage of respondents in the PSID survey who self-reported their health as "excellent," "very good," "good," "fair," or "poor." Their analysis found that when health is measured by frailty, the proportion of individuals with excellent or very good health declines faster with age.

They set cutoff values for frailty based on the distribution of self-reported health of 25- to 29-year-olds. When the cutoff values and frailty were used to determine individuals' health categories as opposed to self-reported health, the researchers observed that health deteriorated much more rapidly with age. In other words, the analysis showed that the fraction of people with poor self-reported health status rose with age, but when they measured health by frailty, they observed a much faster rise than with the self-reports (see the charts). For example, only 17 percent of people aged 70 to 74 had a frailty index low enough to fall into the "excellent" or "very good" health category. That compares with 39 percent of 70- to 74-year-olds who self-reported their health as "excellent" or "very good."

"We interpret these patterns as evidence that self-reported health status underestimates the decline in observable health," the paper says. The researchers also found that the frailty index was a better predictor than self-reported health of mortality and the probability that a person would enter a nursing home or become dependent on Social Security Disability Insurance.

Individuals' self-assessments not always reliable

One reason frailty may be a better gauge of health than self-assessments has to do with the subjective nature of individuals' judgments of their well-being, Kopecky said. "People tend to compare themselves to others their age" in self-reporting their health condition rather than considering how their present medical status compares with their past state, she said. "People seem to be readjusting their self-reported health. So if you really want to map out how health evolves as people age, subjective measures don't work well."

That isn't to say that self-reported health information doesn't have value. It can play a role in helping researchers understand the variation of health within an age group, Kopecky said. She added that self-reported data can also help uncover private medical information that a frailty index would not easily discern, such as hereditary conditions that may put individuals at risk for certain diseases.

Frailty measures gaining traction

Frailty measures as a tool to gauge health are growing in use. Dalhousie University notes that they have been used in studies such as the American National Health and Nutrition Examination Survey, the Beijing Longitudinal Study of Aging, and the Survey of Health, Ageing and Retirement in Europe.

Kopecky said the frailty index model holds much potential in economics. It can provide insight into such matters as the effect of health on a person's earnings over time, a country's labor supply, and individual consumption patterns. "It's a step in the right direction in terms of improving our way of measuring health and as a result being able to understand how health interacts with economic variables and models," she said.

photo of Karen Jacobs
Karen Jacobs

Staff writer for Economy Matters