696 adults tried a self-directed weight-loss programme for sleep apnea. The extra loss was 1.1 kg — and missed the trial's own statistical bar

Weight loss can improve obstructive sleep apnea, but a programme only helps if people can use it in ordinary life. The POWER trial, published on 30 October 2026, tested a low-contact approach across US Veterans Affairs sites. It randomly assigned 696 adults with sleep apnea and a BMI from 30 to 45 to usual care or a self-directed programme based on the Diabetes Prevention Program, delivered through videos, readings and regular coaching reminders.

After 12 months, average unadjusted weight had fallen by 2.1 kg in the programme group and 0.9 kg with usual care. The adjusted difference was 1.1 kg more loss with the programme, with a 95% confidence interval from 0.1 to 2.0 kg more. That may look positive at first glance, because the p value was 0.037 — below the familiar 0.05 line.

But the trial had two co-primary outcomes, so the researchers set a stricter significance threshold of p = 0.025 before analysing them. The weight result did not cross that line. This is not wordplay: choosing the rule in advance limits the chance of declaring a win simply because a study tested more than one main outcome. The authors therefore concluded that the intervention did not improve weight loss.

The other main outcome did cross the statistical threshold. Scores on the Functional Outcomes of Sleep Questionnaire, which asks how sleepiness affects daily activities, improved by 1.4 points with the programme and 0.9 with usual care. The adjusted advantage was 0.6 points, with a 95% confidence interval from 0.2 to 1.0 and p = 0.001. Even so, the researchers judged that small difference to be of limited clinical importance: statistical confidence does not automatically make a change large enough to notice or value.

The pragmatic design is useful because it asked how a scalable programme performs under routine conditions rather than under intensive supervision. It also narrows who the result represents. Participants averaged 56.2 years old, only 19.7% identified as female, and all were recruited through the VA. At 12 months, weight was available for about 82% of each group, while questionnaire completion differed more: 78.2% in the programme group versus 90.5% in usual care. Missing responses can matter if the people who disengage differ from those who remain.

The takeaway is not that lifestyle support cannot help sleep apnea. It is that this particular self-directed package produced, at most, a modest average effect and should not be confused with more intensive treatment. Sleep apnea still needs proper diagnosis and treatment; weight management can be one part of care, alongside options such as positive airway pressure and other clinician-recommended therapies. A programme's convenience, engagement and size of benefit all matter, not just whether one p value falls below 0.05.