Eight phone calls cut loneliness in older adults — and the effect held a year later

Most loneliness interventions that get tested are built for people who are not the loneliest. They run on video calls, or in community groups, or through apps, all of which quietly require that a participant can already get online, get out of the house, or both.

A trial published in JAMA Network Open on February 4 deliberately inverted that. It recruited older adults in Hong Kong who were living alone, lonely, in financial hardship, and digitally excluded — and delivered the entire intervention over the telephone, by trained lay counsellors rather than clinicians [s1].

What was tested

HEAL-HOA — Helping Alleviate Loneliness in Hong Kong Older Adults — was a three-arm, assessor-blinded randomised clinical trial conducted from April 1, 2021 to April 28, 2024 [s1]. Participants were 65 or older and were recruited from public housing estates, community centres, academies for older adults, and word of mouth [s1].

Each arm received eight 30-minute telephone sessions over four weeks, delivered by trained lay counsellors [s1]. Two arms were active: telephone-delivered behavioural activation (Tele-BA), which works by rebuilding engagement in rewarding activity, and telephone-delivered mindfulness (Tele-MF). The third arm, telephone-delivered befriending (Tele-BF), was the attention control — the same number of calls, the same amount of human contact, without the specific technique.

That control arm matters more than it might appear. Without it, any improvement could be attributed to the technique when it was produced by the contact — by somebody phoning at all. Befriending holds the contact constant, so what separates the arms is attributable to the method.

The primary outcome was loneliness, measured on two instruments: the UCLA Loneliness Scale (UCLA-LS) and the De Jong Gierveld Loneliness Scale (DJGL) [s1]. Secondary outcomes included perceived stress, social support, sleep quality, life satisfaction, anxiety, depressive symptoms, and psychological well-being. Outcomes were assessed at baseline and at 1, 3, 6, and 12 months [s1].

What it found

Of 4,152 individuals screened, 1,151 were randomised — mean age 76.6 years (SD 7.8), 843 (73.2%) female, and 965 (83.8%) living with at least one chronic illness [s1]. The trial reports allocation of 335 (29.1%) to Tele-BA, 460 (40.0%) to Tele-MF, and 356 (30.9%) to Tele-BF, against a stated 1:1:1 randomisation [s1].

At 12 months, in intention-to-treat analyses using linear mixed-effects models, both active arms beat befriending on loneliness [s1]:

• Tele-BA: UCLA-LS mean difference −0.73 (95% CI, −1.29 to −0.16; P = .01); DJGL mean difference −0.13 (95% CI, −0.23 to −0.03; P = .01)
• Tele-MF: UCLA-LS mean difference −0.72 (95% CI, −1.24 to −0.20; P = .003)

Secondary outcomes including sleep quality, psychological well-being, and life satisfaction also improved significantly [s1].

The authors also tested a mechanism. Social isolation — the objective count of contacts, as distinct from the subjective feeling of loneliness — measured at 6 months partially mediated the 12-month loneliness effect, accounting for 13.5% of the total effect for Tele-BA and 18.0% for Tele-MF [s1].

How to read the size of it

These are small differences. A mean difference of −0.73 on the UCLA-LS is not the disappearance of loneliness; it is a shift. What makes the result interesting is less its magnitude than its shape: it was still measurable at 12 months, eleven months after the last call, against a control group that received the same number of calls.

The mediation finding is the more useful part scientifically. It suggests that at least some of what these brief interventions do is get people back into contact with other people, and that the contact then does its own work over the following months. But the mediated share is a minority of the effect — roughly one part in seven for behavioural activation, closer to one part in five for mindfulness [s1]. Most of the effect is running through something the trial did not measure.

The limits the authors state

The authors are direct about generalisability: they write that future research should investigate whether these interventions apply outside the Hong Kong context, and whether they are cost-effective [s1]. That is not a formality. The trial recruited a specific population — older, overwhelmingly female, chronically ill, financially strained, living alone, and digitally excluded — in a specific city, during a period that overlapped with pandemic conditions.

Two further caveats follow from the design. It was assessor-blinded, not participant-blinded; people knew which sort of calls they were receiving. And the reported arm sizes are uneven relative to the stated allocation ratio, which the reader should note when weighing between-arm comparisons [s1].

Why it is worth attention anyway

The delivery model is the finding that would matter most if it replicates. Nothing in the intervention required a psychologist, a psychiatrist, a clinic, a device, or a broadband connection. Eight half-hour phone calls, made by trained lay counsellors, produced measurable differences a year later in a population that most digital mental-health products cannot reach at all.

Whether that holds anywhere else is the open question, and this is a single trial.

What to watch

Replication outside Hong Kong; whether the 12-month effect persists further out; and whether the cost-effectiveness analysis the authors call for supports scaling a lay-counsellor telephone model in publicly funded systems.

This article describes a research finding and is not medical advice.

Sources

• [s1] Tang VFY, Jiang D, Kwok JYY, et al. Behavioral Activation and Mindfulness Interventions in Reducing Loneliness and Improving Well-Being in Older Adults: The HEAL-HOA Randomized Clinical Trial. JAMA Network Open, published online 2026-02-04. Trial registration ChiCTR2300072909.