Caring letters did not cut suicide deaths among veterans in a crisis-line analysis

A long-running US programme that mails brief, non-demanding notes of goodwill to people who have contacted a suicide helpline did not lower suicide deaths among veterans, according to a prespecified analysis published in JAMA Psychiatry [s1]. The finding is a setback for one of suicide prevention's most durable and intuitively appealing ideas, and a reminder that warmth and good intentions do not automatically translate into lives saved.

So-called "caring letters" or "caring contacts" trace back to a 1970s study that reported fewer suicide deaths among psychiatric patients who received periodic letters after discharge. The approach is cheap, scalable and humane: a short message that simply says someone is thinking of you, sent on a schedule, with nothing asked in return. The US Department of Veterans Affairs built a version of the intervention for callers to the Veterans Crisis Line, the national helpline for current and former service members [s2]. The new analysis asked the hardest possible question of it — whether the letters changed how many people died.

What the analysis did

The study examined 227,502 Veterans Health Administration patients who contacted the Veterans Crisis Line [s1]. Their mean age was 53.6 years and 84.4% were men [s1]. Of this group, 87,926 veterans received caring letters and 139,576 formed a historical comparison group — people who had contacted the line before the letters programme was in place [s1]. Because the groups are drawn from different time periods rather than randomly assigned within the same period, this is an observational comparison, not a clean randomised experiment, and the authors frame it as a prespecified secondary analysis [s1].

The outcome was the one that matters most: death by suicide during the study period. In total, 625 suicides (0.27%) were recorded across the full cohort [s1].

The result

Suicide mortality was 0.29% among veterans who received caring letters and 0.27% among the historical controls, a difference that did not reach statistical significance (P=0.31) [s1]. In other words, the letters group died by suicide at a rate statistically indistinguishable from — and numerically slightly higher than — the comparison group.

The researchers also tested whether it mattered who signed the letters. Among recipients, the suicide rate was 0.21% for letters signed by peers and 0.22% for letters signed by clinicians (P=0.95) [s1]. No signatory advantage emerged. The authors concluded that caring letter receipt was not associated with suicide mortality and that there were no differences by who signed them [s1].

The scale of the comparison is worth dwelling on, because it cuts both ways. With 227,502 veterans in the analysis, a programme that genuinely reduced suicide by a meaningful margin would have had a good chance of showing it [s1]. That the two rates landed within two-hundredths of a percentage point of each other, with the letters group slightly higher, is not the picture of a strong effect hidden by small numbers. At the same time, suicide is a rare event even here, and the raw count — 625 deaths spread across an enormous cohort — leaves the estimate sensitive to differences between the two eras being compared [s1].

Why a null result deserves attention

It is tempting to dismiss a negative finding, but this one is informative precisely because the intervention is so widely admired and so widely deployed. Caring contacts are recommended in suicide prevention frameworks and have shown promise on softer outcomes — people often report feeling connected and valued. The question this analysis settles, at least for this population and design, is whether that goodwill shifts the hardest endpoint. On the evidence here, it did not.

Several caveats cut in both directions. The historical-control design means the two groups lived through different periods, with potentially different background suicide rates, pandemic effects and changes in VA care; such differences can mask or mimic a true effect. A helpline-contact population is also already receiving some level of attention and services, so an added letter may have less room to move the needle than it would among people with no other contact. And suicide, mercifully rare even in high-risk groups, is statistically hard to shift: detecting a small benefit would require enormous numbers, and a programme can be worthwhile for reasons other than mortality.

None of that rescues a positive claim the data do not support. The honest summary is that, in the largest look yet at this programme's effect on deaths, caring letters were not shown to reduce suicide mortality, and the point estimate did not even trend in the hoped-for direction [s1]. Whether the VA and other systems continue the letters on the grounds of connection, cost and acceptability is a reasonable policy judgement — but it should be made without overstating what the letters do. Readers and clinicians should treat claims that caring letters prevent suicide deaths as unproven for this setting, not established.

If you or someone you know is struggling with thoughts of suicide, help is available. In the US, call or text the 988 Suicide and Crisis Lifeline; in the UK and Ireland, call Samaritans on 116 123. Elsewhere, contact a local crisis line or emergency services.