Derek Brown, PhD, is an assistant professor at the George Warren Brown School of Social Work at Washington University in St. Louis, where he teaches applied linear modeling and health economics courses in the Master of Public Health curriculum. His research focuses, among other things, on health and quality of life, child maltreatment, and how to facilitate better public health decision-making.
ElderBranch interviewed Dr. Brown to discuss his recently published paper, “Associations Between Health-Related Quality of Life and Mortality in Older Adults,” which he wrote with Dr. William W. Thompson and Matthew M. Zach of the Centers for Disease Control and Prevention, Sarah E. Arnold of RTI International, and John P. Barile of the University of Hawaii at Manoa.
What led you to research health-related quality of life measures and mortality? Why is this important? How does your current research augment other work in this area?
As a health economist, I’ve always been interested in how we measure health outcomes. Much of the burden of disease from chronic conditions today comes through nonfatal reductions in health, known as morbidity, and one way to measure the impacts of morbidity is through “health-related quality of life” (HRQOL).
The Centers for Disease Control and Prevention (CDC), and to a lesser extent, the Centers for Medicare and Medicaid Services (CMS), have been monitoring HRQOL through population surveys for several years now.
To the extent that we learn about the association between HRQOL and other health outcomes, like mortality, we can use surveillance data to monitor how people’s lives and their quality of life trends over time. We hope to find ways of improving quality of life, as well as extending it.
Knowing more about what aspects these different survey measures capture in a population can help us to guide public health promotion, to see where the burden of health is concentrated and what drives it, and to make predictions about what may happen to populations over time.
Please describe your study. What methods did you use? What were your in-going hypotheses?
In this case, we suspected that reductions in HRQOL would be associated with increased mortality, but we were not sure how strongly.
From previous studies, we knew that a simple measure of general, self-rated health (“Would you say that in general your health is excellent, very good, good, fair, or poor?”) was associated with mortality. We expected that adding three additional measures – the CDC “Health Days” items, which ask about physical health, mental health, and activity limitations – would capture considerably more information.
What were your findings?
All four simple questions were predictive of mortality, but the single measure of general self-rated health remained by far the strongest predictor. The strength of the association surprised us a little bit.
The other measures are predictive on their own but are dominated by the general health measure if you use all four items. These are good measures of 30- and 90-day mortality, but not as good for 1- or 2.5-year mortality. That was not as surprising because these HRQOL measures capture current or recent health in the last month. Once you look ahead to more than a year, a lot of things can change with anyone’s health.
What are the implications of these findings?
For predicting mortality, there is relatively little loss of information from just asking the one general health item as opposed to all four items. Responses of “fair” or especially “poor” indicate a significantly increased risk of mortality over the time frames described above. However, predicting mortality is an ancillary application of HRQOL.
The primary use of HRQOL questions, including these four items, is to know more about the physical and mental health and the kind of life that someone is experiencing in a multi-dimensional way. As a provider or caregiver, we want to ensure not just longevity, but also maintaining the highest quality of life for people.
Would you make any specific recommendations to care providers as a result of your findings?
For the average caregiver who wants to monitor how a loved one is doing over time – or for a provider or facility that seeks to track how an elderly population in a nursing or assisted living facility is doing over time – you would want to continue to gather all four measures of HRQOL.
These are very short items, which are validated and straightforward to answer. Measuring quality of life gives us an important gauge beyond basic vitals.
Sharp declines for a given person in any of these items would definitely be a marker of concern that we should look closely at that person’s individual circumstances to try to understand what has caused a major shift in their health or quality of life.
What are the next steps to further your work in this area?
One limitation of our study is that, while it was a national sample from the mid-to-late 2000s, the respondents were all enrolled in Medicare Advantage populations (“Medicare Part C” managed care) and were non-institutional at the baseline when they answered the HRQOL questions.
We know that this population is slightly different from the entire Medicare population, so a logical follow-up study would be to replicate it on a broader sample of persons 65 and older. Also, conducting the study on an institutionalized population would be extremely interesting and important from the perspective of providers and facilities.
Is there anything else about your work that you think is important to share with our readers?
I’d like to remind readers that there is no single, right or wrong way to balance health and quality of life. Each family arrives at the decisions that are right for its particular circumstances.
For facilities and providers who have to care for a large number of people and don’t know individual circumstances as well as families might, the idea of measuring, maintaining, and improving health-related quality of life is a good, simple way to capture what is happening to a large number of people.
Also, always keep in mind that individual circumstances vary, and that statistical averages are exactly that—a mix of individual characteristics and outcomes—so our conclusions necessarily apply to populations as a whole.