Automated fitness-tracker nudges gave older adults at most a tiny cognitive edge

Programs that combine exercise, diet and other healthy habits are the main approach being tested to slow the decline of memory and thinking with age. They usually rely on coaches and group sessions, which are expensive and hard to offer to everyone. A University of Tokyo team tested a cheaper version: a wrist-worn activity tracker that sends automatic text prompts, with no coaching at all. After six months the result was small and uncertain, and the authors say so plainly.

The best-known example of such a program is the Finnish FINGER trial, which combined diet, exercise, brain training and checks of heart risk factors over two years. Later trials gave mixed results. Wearable devices are known to nudge people into walking more, but few trials have checked whether that translates into better thinking.

What they did

The researchers mailed invitations to over 2,000 members aged 65 or older of a community health centre in Kashiwa, near Tokyo. To join, people had to live independently and show at least one sign of physical vulnerability: low body weight for their age, frailty or an early stage of it, or signs of muscle loss such as a weak grip or slow walking. People who already used a wearable for their health were not included.

355 people aged 65 to 92 were randomly split into two groups:

• Prompting program (178 people): a Fitbit tracker worn day and night for six months. After a first month of plain recording, each person received automatic text messages built from their own data, with step targets rising in stages to 8,000 a day, exercise targets up to 150 minutes a week, and sleep targets of 6 to 9 hours a night. An optional food-diary app added prompts about protein intake, and those who opted in could collect free protein-rich food developed for the trial. There was no counselling or education, only targets.
• Assessment only (177 people): no prompts or advice; they wore the tracker only during the measurement periods at the start and end.

The trial had three main outcomes of equal rank: a widely used 30-point screening test of memory and thinking (the Montreal Cognitive Assessment, or MoCA), a timed connect-the-dots test of mental speed and flexibility (the Trail Making Test), and muscle mass.

What they found

All 355 people were included in the main analysis, with missing results statistically filled in.

• Thinking test (MoCA): the prompted group scored 0.58 points higher on the 30-point scale than the control group. On its own this passed the usual threshold for statistical significance, but after the correction the team applied for testing three main outcomes it did not (adjusted p = 0.12).
• Connect-the-dots test and muscle mass: no difference.
• Behaviour: among people who kept to the protocol, the prompted group's daily steps and sleep fell less than the control group's, by about 700 steps a day and 12 minutes of sleep a night, and their active time rose a little. Their protein intake was about 8 grams a day higher.
• Exploratory analyses, decided after the results were seen: those who kept wearing the device did better on the MoCA, and bigger gains in exercise went along with bigger MoCA gains. The authors call these results hypothesis-generating, not proof.

There were no serious side effects linked to the program; some people had mild skin irritation from the device or found daily food logging stressful.

What to keep in mind

The authors list the weaknesses themselves, unusually openly:

• Not blinded. Everyone knew their group, the staff who ran the program also did the final tests, and the person who assigned groups could see the sequence. With an effect this small, the authors note, even a little bias would matter.
• Small and uncertain effect. 0.58 points is below what is usually considered a meaningful change for one person. The authors read it as, at best, a small shift in the group average, "of uncertain clinical significance".
• A problem with one test. The connect-the-dots test was first given in a non-standard way and had to be repeated after people knew their group.
• Short and local. Six months, one centre in suburban Japan, people who owned a smartphone. Part of the behaviour difference may simply reflect that only the prompted group wore the tracker all the time.
• Industry links. The trial was partly funded by a food company and Google, which supplied the trackers; employees of both helped design the study.

Why it matters

A program that needs no coaches could reach many more people than one built around group classes. This trial shows that such a program can run with very little staff time and that people did change their habits somewhat while being prompted. It does not show that this protects memory or thinking: the one signal was small, did not survive a correction for multiple tests, and was not seen on the second thinking test. The authors call for larger, longer and blinded trials.