Exercise deepens sleep in sleep apnoea, but only past 12 weeks

Obstructive sleep apnoea repeatedly interrupts breathing during the night and fragments sleep, and exercise is already considered an effective intervention for the condition. What nobody had pinned down is whether exercise changes the objective structure of sleep itself: the balance between light, deep and REM stages that a sleep laboratory can measure. Subjective reports of better sleep are easy to collect but easy to bias. This meta-analysis pulled together the randomised trials that used polysomnography, the reference overnight recording, to see what exercise actually does to sleep architecture in these patients.

The authors searched for randomised controlled trials in adults over 18 with obstructive sleep apnoea, where sleep was measured by polysomnography after an exercise programme and the proportions of sleep stages were reported. Six trials with a combined 236 patients qualified. Risk of bias was checked with the revised Cochrane tool and funnel plots, and results were pooled with a random effects model. Outcomes included total sleep time, sleep efficiency, sleep onset latency and the share of the night spent in each stage, from light N1 and N2 through deep N3 to REM sleep.

Across all six trials together, exercise made no significant difference to total sleep time, sleep efficiency, sleep onset latency, stage N1 sleep or REM sleep. The interesting movement was in the architecture of the night: programmes lasting more than 12 weeks significantly cut the share of lighter stage N2 sleep and increased deep stage N3 sleep. That shift did not reach significance in the full analysis, but it did once trials involving patients with heart disease were excluded, suggesting that condition was muddying the average. The authors note the deeper sleep pattern may correspond to better subjective sleep quality.

Exercise already has a place in managing sleep apnoea, but its credentials have mostly rested on symptom and severity measures rather than what happens inside the sleeping brain. This analysis extends the case by showing an objective shift towards deep sleep, while complicating it with two conditions: the programme must run long enough, and heart disease appears to blur the signal. With only six trials and 236 patients, the finding needs more data before it hardens. Whether the extra deep sleep translates into feeling more rested or into better daytime function remains an open question.

If you have sleep apnoea and have started training partly for your sleep, this is a reason for patience rather than doubt. Nothing here suggests exercise adds hours of sleep or gets you to sleep faster; what it may do is change the quality of the hours you get, tilting the night towards the deep N3 stage. That shift only showed up in programmes running past 12 weeks, so a month of effort proving nothing would be entirely consistent with these results. It is information about timescale, not a promise.