Elderly health check-ups: a policy with a plausible effect it cannot prove

Design

sequential target trial emulation within a population-based cohort, inverse probability-weighted pooled logistic regression, three-month grace period

Population

3,096 community-dwelling adults aged 75-84 without functional disability in Sukagawa City, Japan, 2018-2021, contributing 21,484 person-trials

Primary outcome

composite of incident functional disability (long-term care certification level 3 or above) or death at five years

Effect

20.6% vs 26.7%; risk difference -6.1 percentage points (95% CI -11.6 to 0.9), risk ratio 0.77 (95% CI 0.57-1.03)

Japan runs a national Elderly Health Check-up programme with the explicit policy aim of keeping people out of long-term care. This sequential target trial emulation used the Sukagawa Study, a population cohort in a rural, ageing municipality in Fukushima, to estimate what first attendance actually achieves. It compared attendance within a three-month grace period against non-attendance in adults aged 75 to 84 without existing functional disability, generating 21,484 person-trials from 3,096 individuals, and estimated five-year risks with inverse probability weighting.

The composite of functional disability — long-term care certification at care-need level 3 or above — or death occurred in 20.6% of attenders and 26.7% of non-attenders: risk difference -6.1 percentage points (95% CI -11.6 to 0.9), risk ratio 0.77 (95% CI 0.57 to 1.03). Mortality alone gave -3.7 points (95% CI -10.4 to 3.9) and disability alone -5.4 points (95% CI -12.9 to 0.1). Every interval touches or crosses the null.

The authors' conclusion is the correct one and worth quoting in spirit: the evidence neither supports expanding the programme nor justifies stopping it. What makes the study valuable is the method — a target trial emulation with a defined grace period is a serious attempt to get a causal estimate out of observational attendance data, and it still could not resolve the question with 3,096 people. The people who attend a voluntary check-up are healthier and more organised than those who do not, in ways no weighting fully removes. For a clinician the message is narrower than the policy question: a check-up is not a substitute for identifying and acting on the individual risks in front of you.

• Do not present attendance at a screening check-up to a patient as proven to prevent disability
• Target case-finding at the people least likely to attend — attenders are the healthier group
• Use a check-up as an opportunity to act on identified risks, which is where any benefit must come from
• Note the design: target trial emulation is the strongest observational approach here and still returned an inconclusive answer
• Findings come from one rural Japanese municipality with a national long-term care certification system; they do not transfer directly

The statistics, in plain English

A risk difference of -6.1 percentage points with an interval from -11.6 to +0.9 means the data are compatible with a substantial benefit and with a small harm, so the honest summary is uncertainty rather than a positive result. The disability-only interval (-12.9 to 0.1) stops just short of excluding the null, which is a reminder that statistical significance is a threshold rather than a change in the evidence. Target trial emulation reduces some biases of observational comparison — notably immortal time bias — by defining eligibility and follow-up as a trial would, but it cannot remove confounding by the health and motivation that made someone attend.

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.