Apple, Google, and other tech companies interested in digital health learned of some strong competition this week.
It turns out predicting health is not so complicated. Nor is it digital at all.
For persons of middle age (40 to 70 years), self-reported overall health and walking speed were the best predictors of death in the next 5 years, according to a study published this week in the Lancet.
In an analysis of nearly 500,000 UK citizens followed for 5 years, these two simple questions outperformed 655 measurements of demographics, health, and lifestyle. Is your health excellent, good, average, or poor? Is your walking pace slow, average, or brisk? Along with smoking, those two basic questions, inquiries that hardly require a digital device, were the best predictors of staying alive in the next 5 years.
Pause for a moment here and ponder the beauty of that top-line result. Half a million people followed for 5 years; 655 measures of health, including heart rate, blood pressure, and lab tests, and the best predictors were that simple.
The UK BioBank project is a remarkable source of health-related big data. Baseline health information was obtained on 500,000 UK citizens between 2007 and 2010. Subjects were followed for 5 years. Mortality data, including cause of death, were recorded. BioBank data is made available to researchers by application.
In the first major publication of BioBank data, Dr Erik Ingelsson (Uppsala University, Sweden) and Dr Andrea Ganna (Karolinska Institutet, Stockholm, Sweden) compiled not just an analysis of predictors of death but also an interactive website allowing people to assess their own health age compared with the overall UK population.
When you study associations between 655 measures in almost 500,000 people over 5 years, you use complicated statistics. The most important thing to say about the methods of this study was set out by Dr Ingelsson in the press briefing. He said their methods do not allow one to assign cause. He used the gray-hair example: We know gray hair associates with older age, but gray hair does not cause one to be old.
With that as a caveat, the researchers used the concordance index (C-index) to assess how well each measure predicted mortality. The C-index can range from 0.5 (no discrimination or random chance) to 1.0 (perfect correlation.)
In the 5 years of follow-up, 8532 (1.7%) subjects died. Overall, cancer was the most common cause of death (53% in men; 69% in women). The most common cancer-related cause of death was lung cancer in men (n=546) and breast cancer in women (n=489). Cardiovascular disease was the second leading cause of death (26% in men; 33% in women).
There were gender differences in predictors of death. Self-reported health was the strongest predictor of death in men (C index 0.74). In women, a previous cancer diagnosis was the strongest predictor (C index 0.73).
Self-reported walking pace was a strong predictor of death in both men and women (C index 0.72 and 0.69, respectively.) For example, a man aged 40 to 52 years who reported a slow walking pace was 3.7 times more likely to die than a similarly aged man who reported a steady walking pace.
In a large subset of subjects with no reported health conditions, smoking was the best predictor of mortality.
The final, and perhaps niftiest, aspect of this study was that researchers developed an 11 to 13 question risk prediction score, which they then put on an interactive website. Anyone can answer these simple questions and get their health-related age relative to the UK population. The researchers call this age the UK Longevity Explorer (UbbLE) age.
The two authors of this study deserve great credit. This was a huge project, and in the age of multiauthored papers, it's shocking that this paper had only two authors. Also commendable was the time the authors spent in the discussion section explaining the data and the limits of what they say and do not say. They explicitly stated their goal was to generate hypothesis for further study. Good on them.
That said, I will use this study in my practice starting tomorrow.
I am drawn to these findings because they emphasize something that is increasingly lost on both doctors and patients. True health is not complicated. And the big picture is still useful.
Any experienced clinician will testify that patients know when they are well and when they are not. The finding that self-reported health predicts death urges clinicians, generalists and specialists alike, to ask our patients how they feel about their health. I often ask my patients to give their health a grade; I use letter grades A, B, C, D. When a patient tells me they are an A, I feel better. I go into harm-avoidance mode. When patients grade their health lower, I ask why. This line of questioning reliably leads me to the major issue.
Then there is the matter of self-reported walking pace. How easy it is to be distracted by digital data. We walk into the exam room to see our patient. He is still. We look at him. We poke and listen to his body. We assess his ECG and other measures. Soon we will review his smartphone metrics and DNA data. Yet we tend to forget the obvious: to move is to be healthy. Drs Ganna and Ingelsson teach us that to move briskly may be healthier.
Distraction is a growing problem of the digital age. I see it every day. Nurses, doctors, and patients are diverted from the basics. We study the numbers, the rhythms, and the images, yet the best predictors of outcomes may be the simple things that our patients can tell us.
These results take us right back to the core of doctoring—the person before us. How does she feel? How well does she move? Does she smoke? Details are important, yes, surely they are, but so is the big picture.
And there's a bonus: I like to post studies in my exam rooms. Could there be a better nudge for exercise and mobility than this study?
© 2015 WebMD, LLC
Cite this: Health Is Not Complicated—Just Ask the Patient - Medscape - Jun 05, 2015.