Two or three mornings of accelerometry were enough to predict five-year risk

Design

Secondary analysis of a national cohort accelerometry substudy with mixed effects location scale and two-stage regression modelling

Population

584 community-dwelling older adults in the National Social Life, Health and Aging Project, wrist accelerometry for 72 hours in 2010-2011; 104 died within 5 years

Primary outcome

Five-year mortality and adapted frailty phenotype score among survivors

Effect

Mean morning hourly activity OR 0.64 per SD for mortality (p=0.04); within-subject morning variance OR 0.67 per SD for better frailty score (p=0.01)

Accelerometry tracks frailty well, but the standard five to seven day wear protocol is exactly what older adults are least likely to complete — so the measure that works is the measure that does not get collected. This analysis asked how short the recording could be.

It used the 2010-2011 accelerometry substudy and 2015-2016 follow-up of the National Social Life, Health and Aging Project: 584 community-dwelling older adults who wore wrist accelerometers for 72 hours, with an adapted frailty phenotype measured at both timepoints; 104 died within five years. A mixed effects location scale model derived two measures from the morning window of 7:00 to 11:59 — mean hourly activity and within-subject variance across mornings — each adjusted for day of week and month of wear.

After adjustment for baseline frailty, demographics, comorbidities, BMI and cognition, higher mean morning activity predicted lower five-year mortality (OR 0.64 per standard deviation increase, p=0.04). Higher within-subject morning variance predicted better frailty scores among survivors (OR 0.67 per SD, p=0.01).

The variance result is the counterintuitive one and probably the more interesting. Day-to-day variation in morning activity is not instability — it is the signature of someone whose mornings differ because their life has different things in them. A flat, identical morning every day is what a constrained life looks like on an accelerometer.

• Seventy-two hours of wear may be enough — that changes what is feasible in an older population
• The morning window, 7:00 to 11:59, carried the signal
• Day-to-day variability was protective, not a warning sign
• This is a research finding; no validated clinical threshold exists to act on
• Wrist-worn consumer devices are not interchangeable with research accelerometers

Why it matters

The obstacle to accelerometry as a frailty screen has been wear adherence, and this suggests the protocol can shrink.

Don't overread it

A secondary analysis of one cohort with 104 deaths and borderline significance for the mortality outcome; no clinical threshold has been validated.

The statistics, in plain English

An odds ratio of 0.64 per standard deviation means each standard-deviation increase in morning activity was associated with about a third lower odds of dying within five years — in a sample with 104 deaths, which is modest. The p value of 0.04 for mortality is only just inside conventional significance, so this needs replication rather than adoption. The variance finding predicted frailty scores only among survivors, which introduces a selection step worth remembering.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.