A brief message saying “I’m thinking of you” seems almost too small to count as suicide prevention. That is precisely the appeal of Caring Contacts: the intervention is inexpensive, does not demand a reply, and can extend a thread of human connection long after an assessment or crisis visit. But a large new trial asks the harder question. When people already receive attentive, community-based support, do the messages add measurable protection?
The answer in this study was no. In a randomized clinical trial published November 2 in *JAMA Network Open*, culturally adapted Caring Contacts did not reduce suicidal thoughts, attempts, or related hospitalizations more than an enhanced-support comparison condition among American Indian and Alaska Native adults. That does not prove the messages are useless. It shows that a plausible, well-liked intervention did not outperform a substantial package of resources in this setting—and that distinction matters.
The trial tested connection against real support, not neglect
The CARE trial enrolled 709 adults with active suicidality from five American Indian and Alaska Native communities in Alaska, the Northern Plains, and the Southern Plains. Recruitment ran from 2018 to 2022, and follow-up ended in 2023. Participants were randomly assigned either to enhanced usual care alone or to enhanced usual care plus Caring Contacts.
The comparison condition was not “nothing.” Everyone received individualized information about behavioral-health services, crisis resources, traditional healing, and community support. Staff reviewed those options and could make warm handoffs to services. They remained available to help participants gain access, although they did not provide psychotherapy or ongoing clinical management.
The intervention group also received 25 brief messages over 12 months, beginning the day after the baseline interview. Messages arrived frequently at first, then less often, with additional birthday and winter-holiday contacts. They expressed care, support, and inclusion; staff could tailor language to local culture and respond personally if a participant wrote back.
That adaptation was not cosmetic. Before the trial, researchers used a community-based process involving 189 community members, health-system stakeholders, and leaders across four tribal communities. The resulting qualitative study shaped message timing, content, eligibility, and study measures. In other words, the trial did not simply import a generic text-message program and relabel it.
Both groups improved, but the messages did not separate them
At 12 months, suicidal-ideation scores had fallen sharply in both groups. Among participants with follow-up data, past-year suicide attempts were reported by 14.2% of the Caring Contacts group and 16.4% of the enhanced-care group. The adjusted odds ratio was 0.90, with a 95% confidence interval from 0.55 to 1.50. Because that interval spans both potentially meaningful benefit and possible harm, the estimate is uncertain; it is not evidence of equivalence.
Nor were there statistically significant differences in suicide-related hospitalizations or the proportion still classified as high risk. The central result is therefore comparative: adding the messages did not produce a detectable advantage over enhanced support. The large improvements within both groups cannot be credited to either condition. They could reflect the support package, the unusually compassionate baseline engagement, regression to the mean, other care received during the year, or a mixture of these factors.
This result sits inside an already mixed literature. A 2023 systematic review and meta-analysis combined six randomized trials with 6,218 participants. It found a protective effect for suicide attempts at one year, but estimates for suicide deaths and emergency or hospital use were compatible with increases as well as decreases in risk. The new trial adds an important population and a demanding comparison, rather than delivering a final verdict on every version of Caring Contacts.
A null result can still change the question
There are substantial limits. Only 64% of participants completed the 12-month follow-up, and those lost to follow-up differed on several social characteristics. Much of the study occurred during COVID-19, forcing phone-based data collection and disrupting tribal communities. Outcomes were assessed at one year, which may miss short-lived effects. An invited commentary also notes that the adult trial used an ideation scale validated in Native children and younger adolescents, raising questions about measurement, and that fewer attempts occurred than anticipated, which may have reduced statistical power.
At the same time, the study’s realism is a strength. Community members conducted the work, and eligibility included people often excluded from trials, including those experiencing homelessness, substance misuse, or severe mental illness. The design tested whether Caring Contacts added value under conditions that participants and researchers considered ethically acceptable—not whether a message was better than abandonment.
That reframes the takeaway. Caring messages may be welcome, culturally meaningful, and inexpensive without being sufficient to improve clinical outcomes beyond active support. Programs should not treat their warmth or popularity as proof of effectiveness, and this single trial should not be used to strip away contact or community resources. The more useful conclusion is narrower: connection deserves rigorous testing, and the quality of the care surrounding it may determine whether an extra message can make a measurable difference.
