Measuring Resiliency in Older Adults
Karen Bandeen-Roche uses statistical methods to identify biological signatures of resilience and frailty.
Unobservables are Karen Bandeen-Roche’s specialty. While some problems associated with aging are readily apparent—walking slows, hearing and vision weaken—some can’t be observed directly, like processes underlying cognitive decline. That’s where Bandeen-Roche, PhD, MS, Hurley Dorrier Professor and Chair in Biostatistics, gets creative. Using surrogate measures—things we can observe, like muscle strength—she develops models that explain the underlying processes of problems such as dementia and frailty. Such models could help identify targets for effective treatment—or even better, prevention. The need is great for these insights: By 2050, the global population of people age 60 and older will number more than 2 billion, compared to 962 million in 2017.
How do you define resilience and frailty?
My colleagues and I consider the two as manifestations of the same underlying physiology, where the network of systems that can absorb shocks either is well-tuned or loses the ability to compensate. We think of physical resilience in older adults as the ability to recover from stressors such as a surgery, or perhaps a fall or an infection. Maybe [they] take a temporary hit but ultimately rebound to the same level of health they began with. That’s resilience. In frailty, on the other hand, a person becomes vulnerable to adverse outcomes following a stressor.
These aren’t easy to measure.
Correct. But there are, particularly for frailty, a few measures regarded as the leading approaches in the field. One of them was developed here at Johns Hopkins. My colleagues Linda Fried [now dean of the Columbia Mailman School of Public Health], Jeremy Walston and others developed what’s known as the physical phenotype of frailty, [which includes] several criteria including weakness, low activity, weight loss and slowness.
My contribution has been to validate that approaches such as this are succeeding at assessing what they intend.
Do any of these tools measure where people are before a stressor to see where they should be after? To create a baseline?
This is exactly the frontier of where we’re going with a resiliency study Dr. Walston and I, with others, now have in the field.
One pilot study we conducted almost 10 years ago in very old women looked at exactly the idea you just said: You bring people in, you subject them to a mini-stressor—an oral glucose tolerance test is a good example. You measure the insulin response over, say, two hours, so that you get a whole curve of response. You would hypothesize that resilient people should have an appropriate response and then quickly return to their baseline, whereas the frail people might have an exaggerated response and not bounce back nearly as quickly. And in fact, that’s what we see.
This is the frontier, we believe, of how to measure both resiliency and frailty before older adults experience a serious stressor, such as surgery. Hopefully we could then do something to either guide their clinical care or to fortify them.
What might that look like—physical therapy before a surgery?
That’s exactly what a short-term intervention might look like—physical therapy, or perhaps a nutritional boost. It might be pharmaceutical interventions. It’s a hot area of development, whether there are pharmaceutical interventions that, for example, could boost mitochondrial functioning so that energy production is restored more toward what a healthy individual would look like.
This is the frontier, we believe, of how to measure both resiliency and frailty before older adults experience a serious stressor such as surgery. Hopefully we could then do something either guide their clinical care or to fortify them.
What are the next big questions that need to be answered about frailty?
I think there are three big questions. The one I’m most interested in is, what causes frailty, and therefore what can we do to delay the onset of frailty or boost resilience?
There’s a second question in the clinical realm. Geriatricians believe it’s beneficial to screen for frailty, or to case-find for frailty. But nobody quite knows what to do once you’ve done that. Developing a randomized, controlled evidence base to answer how persons identified as “frail” should be clinically managed is a super important next step. And then the third question is, how do you identify individuals who are about to become frail? There are measures for “pre-frailty,” but a next generation of those measures would help us identify people who are on the cusp of frailty before they’re too far along to be brought back.
You’ve said that you’ve “evolved into a hybrid scientist, roughly equal parts statistician and gerontologist.” How did that happen?
By accident. My statistical interest is in learning how to measure things we are able to define conceptually but for which no accurate and precise measures yet exist; the best we can do is to use surrogate measures that together may allow us to infer the target that conceptually you’re after. A framework for evaluating how well the surrogate measures succeed is called latent variable models.
About a year after I got here, I met Linda Fried, who I mentioned before. Her aging study seemed interesting. I had no prior interest in aging, but the prospects for public health, reducing suffering and promoting vibrant years of life were so great for older adults. And at the same time, how to measure geriatric concepts like frailty was a good fit with my statistical interests.
Have you found anything along the way that particularly surprised you?
In work we have done to evaluate the epidemiology of frailty in the U.S., we observed massive disparities in frailty prevalence by race and ethnicity—a 60% to 80% increase in black and Hispanic older adults as compared to their white counterparts. I guess I should have expected it, but just the magnitude of the disparities was surprising to me. We also found disparities by income and geographic regions. There’s lots of targeted prevention work to be done. I don’t think there’s a one-size-fits-all sort of solution.