Caring letters after a crisis call: no difference in suicide deaths

Design

Prespecified secondary analysis of a randomised trial (signatory) with historical observational control

Population

227,502 US veterans who contacted the Veterans Crisis Line

Primary outcome

Suicide mortality within 12 months of crisis line contact

Effect

0.29% with letters vs 0.27% without (P = .31); peer vs clinician 0.21% vs 0.22%

Caring letters — brief, non-demanding messages sent after a crisis contact — are one of the few interventions that has reduced suicide deaths in a randomised trial, but that result had not been replicated. This prespecified analysis in JAMA Psychiatry looked at the US Veterans Crisis Line programme, in which veterans received nine letters over a year, signed by either a clinician or a peer veteran assigned at random.

Among 227,502 veterans who called, 625 died by suicide within 12 months. Mortality was 0.29% in the 87,926 who received letters and 0.27% in 139,576 historical controls who did not (P = .31). Peer and clinician letters did no differently (0.21% and 0.22%).

Only the signatory was randomised; the comparison of letters with no letters used an earlier group of callers, so differences between periods could hide an effect. The parent trial had reported more use of healthcare after letters. Letters remain cheap and acceptable, but they should not be relied on as a service's main protection after a crisis.

• Do not rely on caring letters alone as the follow-up for a person after a suicidal crisis.
• Keep direct follow-up — a call or appointment within days of discharge or crisis contact — as the core of aftercare.
• If a service already sends caring letters, they may still support engagement; there is no signal of harm.
• Combine safety planning, means restriction counselling and early review rather than any single contact strategy.

Why it matters

One of the few interventions thought to reduce suicide deaths showed no mortality signal at scale.

Don't overread it

Letters versus no letters was an observational comparison with historical controls, not a randomised one.

The statistics, in plain English

Suicide death rates of 0.29% and 0.27% are nearly identical, and a P value of .31 means the small difference could easily be chance. Because suicide is rare, even a very large study can miss a modest benefit, and the comparison group came from an earlier period.

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.