When “Self-Reported” Research Is Reported By Itself

Here’s the problem with most studies that assess the effects of diet, exercise, or sleep over time: They’re largely self-reported—and people, it seems, aren’t very good at remembering how much they ate, or what they ate, or how far they walked or slept, or for how long, or whether they were eating while they were walking or sleeping.

There are some of us, to be sure, who are inveterate chroniclers—natural-born Homers (the Greek, not the Simpson) who can relate from memory how many stairs were climbed or paces were paced in any given hour or day. But most of us, well, aren’t. We’re imperfect scribes with imperfect memories.

Which is why this report from JAMA Cardiology last week was so significant—more significant, I would say, than the amount of attention it received. In the study, Stanford researcher Euan Ashley and colleagues set out to determine if they could automatically measure “objective, real-world physical activity patterns, fitness, sleep, and cardiovascular health” from consenting volunteers who downloaded a Stanford Medicine-developed iPhone app called MyHeart Counts (which is based on Apple’s open-source framework, ResearchKit).

Lots of people downloaded the free app after it was made available in March 2015. No shocker there. But what was surprising, at least to me, was how many of those app users consented to participate in the study: 48,968 of them—which included individuals in all 50 states and the District of Columbia. And more surprising still is that nearly 82% of those folks, or 40,017 volunteers in all, actually uploaded data to the research team.

Though only 9% or so of consenters (4,552 people) uploaded a full week of reports on their physical activity, that’s still a big sample size for an objectively measured, data-driven research study. The “always-on, low-power motion chips” in the smartphones captured what no human memory ever could.

While one aspect of Stanford study involved self-reporting (some participating volunteers filled out a questionnaire on their perceptions of their own heart health), the most significant finding of the study didn’t come from that.

Rather, in the words of the researchers, it’s that both “mobile consent” and “real-time gathering of sensor and survey data from a large ambulatory population” are eminently feasible. And in a smartphone-driven world, that has changed public health research overnight.