Caring letters were not associated with lower suicide mortality in veterans who called a crisis line

Design

Prespecified secondary analysis of a randomised trial (clinician vs peer signatory) with observational historical control comparison

Population

227,502 US veterans who contacted the Veterans Crisis Line; 87,926 received caring letters

Primary outcome

Suicide mortality in the 12 months after the index contact

Effect

0.29% with letters vs 0.27% historical controls (P = .31); peer 0.21% vs clinician 0.22% (P = .95)

Caring letters, short supportive messages sent over time after a crisis contact, are among the few suicide prevention interventions reported to have lowered suicide mortality in a randomised trial, but that result had not been replicated. This analysis looked at the US Veterans Crisis Line programme: veterans who called between June 2020 and June 2021 were randomised to nine letters over a year signed by a clinician or a peer veteran, and compared with a historical control group who called in the preceding years and received nothing.

Of 227,502 patients, 87,926 received letters. Suicide mortality at 12 months was 0.29% with letters and 0.27% in the historical controls (P = .31). Within the letters group, peer-signed and clinician-signed letters gave 0.21% and 0.22% (P = .95).

The comparison with no letters relies on historical controls, so it is observational, and the trial itself randomised only the signatory. The result does not argue against contact after a crisis, which has other supporting evidence, but it does suggest that letters alone should not be counted on to lower deaths. The authors point to health care utilisation as the more promising effect to explore.

• Caring letters were not linked to lower 12-month suicide mortality in this large veteran cohort.
• Peer-signed and clinician-signed letters gave similar outcomes.
• Treat letters or brief contact as an addition to safety planning and follow-up, not a replacement.
• Keep follow-up after any crisis contact: a call, visit or appointment within days, per local protocol.
• The earlier trial finding has not been replicated here; avoid overstating either result.

Why it matters

It tempers enthusiasm for a low-cost intervention that had looked promising and had not been replicated.

Don't overread it

The letters-versus-no-letters comparison used historical controls, so it cannot show cause and effect in either direction.

The statistics, in plain English

Suicide is rare even in this high-risk group, about 3 in 1,000 over the year, so very large samples are needed to see small effects. The P value of .31 means a difference of this size could easily arise by chance. A non-significant result in a historical comparison does not prove the letters did nothing, only that no effect on mortality was seen here.

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.