# Can Sleep and Exercise Work Better Together?

One Paper a Day. Today: a 2026 randomized clinical trial about sleep, exercise, and what happens when two plausible interventions are delivered as one program.

This is independent educational commentary, not medical, sleep, or exercise advice. The exercise in the study was high intensity, supervised in a laboratory, and tested in a carefully selected group. That framing holds for the whole episode, so I will state it once here rather than repeat it at every turn.

## 1. The question and the bottom line

Here is the bottom line before any background, so you have it even if you stop after this minute.

A four-arm randomized trial tested whether combining supervised exercise with a structured sleep-health program improves sleep more than either one alone. The headline finding: the combined program produced modest but real gains on some objective measures of sleep continuity. People in the combined group spent less of the night awake and slept more efficiently than people who got only one program.

The biggest caveat, in the same breath: the combined program did not increase the objective total amount people slept, and it was not clearly better than the single programs on how people rated their own sleep. So this is a story about sleeping more continuously, not sleeping more.

Who should care? Anyone weighing a bundled wellness program, and anyone who wants to read health headlines honestly. If you were hoping this proves a sleep app fixes insomnia, it does not, and I will show you exactly why. The paper is most useful as a clean example of a promising group result with clear limits.

The word people reach for is synergy: the idea that a combination does more than its parts. The responsible question is narrower. Which outcomes favored the combination, by how much, with what uncertainty, and for whom?

## 2. Why sleep and performance matter

Sleep is often treated as downtime stolen from productive hours. In reality it is part of the system that supports attention, mood, learning, and recovery. Poor sleep makes the same task cost more effort.

But note a boundary that will matter later: this trial measured sleep. It did not measure work output, exam grades, or athletic performance. Any claim that the program made people more productive travels beyond the evidence.

The commercial relevance is still real. Wellness businesses routinely bundle movement, coaching, sleep education, and an app. A bundle can make sense because activity and sleep influence each other, and because one service reinforces engagement with another. Research has to separate those plausible stories from measured effects. A program can feel coherent and still fail to move the outcome that matters.

## 3. The field so far

To judge whether this trial is a breakthrough or a confirmation, you have to know where the field already stood. Three prior results frame it, and they do not all point the same way.

First, the foundation. A 2015 meta-analysis by Kredlow and colleagues, pooling sixty-six studies, found that exercise has only small beneficial effects on sleep, including sleep efficiency and quality. Helpful, but modest. That sets a realistic expectation: exercise nudges sleep, it does not transform it.

Second, and more pointed, a 2018 meta-analysis by Banno and colleagues in the journal PeerJ found that exercise improved how people rated their sleep, but did not significantly improve the objective measures: sleep-onset latency, total sleep time, wake after sleep onset, and sleep efficiency. Hold onto that, because today's trial reports objective gains in exactly the measures this larger body of work has struggled to move. That is a genuine tension, not a footnote.

Third, the benchmark. A 2020 network meta-analysis by Baglioni and colleagues compared non-drug options for insomnia and concluded that none of them, exercise included, showed evidence superior to cognitive behavioral therapy for insomnia, known as CBT-I, which remains the recommended first-line treatment. Today's trial did not compare its program against CBT-I at all. So whatever it shows, it does not show that exercise plus sleep coaching beats the current standard.

That is the backdrop: exercise reliably shifts how sleep feels, less reliably shifts what sensors measure, and has not overtaken CBT-I. Now the new trial.

## 4. The four-arm trial and its programs

The study ran for eight weeks between July and September 2024. It enrolled 112 women of Chinese nationality, aged 18 to 30, with an average age of about twenty-three and a half. The population was deliberately narrow: participants were sedentary, defined as at least eight hours of sedentary behavior a day by an activity monitor, and had poor sleep health, defined by a score above five on the Pittsburgh Sleep Quality Index. People with a range of psychiatric, cardiovascular, respiratory, neurological, musculoskeletal, metabolic, and endocrine conditions were excluded.

The 112 participants were randomized in equal numbers, twenty-eight each, to four groups: the combined program, exercise alone, sleep support alone, or a usual-lifestyle control. A researcher outside the study handled assignment with sealed opaque envelopes. Participants knew which program they received, which is hard to avoid in exercise and coaching research. One hundred and four completed both assessments, and all randomized participants were analyzed by intention-to-treat. No serious adverse events were reported.

The two programs were more substantial than their labels suggest. The exercise was high-intensity circuit training, performed in a laboratory three times a week, roughly forty to sixty minutes per session, with warm-up, body-weight circuits, rest intervals, and cool-down. The sleep-health program began with an individualized, face-to-face counseling session with a trained researcher, followed by seven weekly follow-ups with workshops and sleep-education modules, supported by material from a smartphone cognitive-behavioral-therapy app called Resleep. So the sleep arm was a human-supported program that used digital content, not a self-service app. Adherence was reasonably high: about eighty-six to eighty-nine percent across the exercise and sleep programs.

Randomization strengthens the comparison among these four assigned programs. It does not make the sample representative of everyone. The result belongs first to sedentary young women with poor sleep health who met the safety criteria and took part in a supervised study.

## 5. How sleep was measured

Participants were assessed before the program and again forty-eight hours after the last session, with two kinds of measures.

The subjective measure was the Pittsburgh Sleep Quality Index, a questionnaire where lower scores mean healthier sleep. The objective measures came from actigraphy: a wrist accelerometer whose algorithm estimates when you are asleep or awake, yielding total sleep time, sleep efficiency, and wake after sleep onset. Sleep efficiency is the share of time in bed spent asleep. Wake after sleep onset is time spent awake after first falling asleep. A person can sleep the same total time while sleeping more continuously, with less waking in the middle.

One distinction matters for reading the results: actigraphy is not polysomnography. It does not directly record the brain, eye, muscle, and breathing signals of a full clinical sleep study, so it is better at continuity than at fine sleep structure. The analysis used linear mixed models adjusted for baseline, reported ninety-five percent confidence intervals, applied a Bonferroni correction to the main comparisons for multiple testing, and ran a sensitivity analysis with multiple imputation. Those are signs of a serious attempt to avoid mistaking noise for a result.

## 6. The results and what is genuinely new

The clearest advantage for the combined program appeared in sleep continuity.

Compared with sleep support alone, the combined program improved sleep efficiency by an additional 2.75 percentage points, with a confidence interval from 0.65 to 4.85. For wake after sleep onset, the combined group improved by an additional 14.51 minutes compared with exercise alone, interval 3.76 to 25.27, and by 16.26 minutes compared with sleep support alone, interval 5.50 to 27.01. Descriptively, the combined group moved from an average sleep efficiency near eighty-eight percent to about ninety-three percent, and its average wake after sleep onset fell from roughly fifty-four minutes to twenty-eight. The randomized between-group differences are the stronger evidence, because they help account for change that might have happened anyway.

So what is genuinely new here, read against the field? Recall that the larger 2018 meta-analysis found exercise did not reliably move objective sleep continuity. This trial reports that a combined, supervised program did move objective continuity, in a randomized design with corrected comparisons. That is the real contribution: not that combining works in general, but that in this population a combined program produced measurable objective continuity gains where the broader literature has mostly found subjective-only effects.

It also fits where the field is heading. The newest reviews are shifting from single-tactic studies to network meta-analyses that compare and combine modalities. A 2026 review by Zhou and colleagues is now ranking combinations of exercise with other approaches for insomnia, and a 2025 network meta-analysis by Bu and colleagues, pooling twenty-two trials, compares specific exercise types rather than treating exercise as one thing. This trial is one carefully run data point feeding that larger, still-unsettled synthesis.

The honest framing is not "the combination fixed sleep." It is that the combined assigned program beat its own components on some objective continuity measures, in a literature where such objective gains have been hard to demonstrate.

## 7. The nulls and secondary outcomes

Now the result a promotional summary would bury: actigraphy found no change in total sleep time. Participants did not objectively gain more minutes of sleep. Some spent less of the night awake and slept more efficiently, which can still matter, but continuity and duration are different outcomes.

The subjective results add a layer. Compared with control, all three active programs improved the total Pittsburgh Sleep Quality Index score. On that self-reported total, the combined program was not uniquely superior to the single programs. That disagreement between sensor and questionnaire is not a defect to hide. People can perceive sleep differently from how a movement sensor estimates it, and this trial shows both signals at once, which is exactly the subjective-versus-objective split the 2018 meta-analysis flagged.

The researchers also explored waist measures, blood lipids, and inflammation-related markers. They reported some favorable changes in the exercise groups but described the overall effects as limited, and any benefit could reflect exercise rather than a special combined effect. These were secondary and exploratory, in a short study of mostly not-overweight participants, so they should not be read as headline health outcomes.

## 8. What you can take away

Five takeaways you could repeat to a friend, each labeled by how settled the evidence is.

One. In this trial, a combined exercise-plus-sleep program improved objective sleep continuity, less waking and higher efficiency, more than either program alone. Well supported here, within this population.

Two. It did not increase the objective total time people slept, and it did not clearly win on self-rated sleep. Well supported here, and easy to forget.

Three. That objective-continuity gain is notable because the broader literature has mostly found exercise improving self-rated sleep but not objective measures. Suggestive: it is one randomized trial against a larger body of mixed results.

Four. This is not evidence that a stand-alone app improves sleep, and not evidence that the program beats CBT-I, the first-line treatment, which was never compared here. Well supported, by what the study did and did not include.

Five. Where the field goes next is combining and ranking modalities, and testing them with polysomnography, longer follow-up, and broader populations. Suggestive, based on the direction of current reviews.

Notice what these takeaways deliberately separate: what the paper showed, what it left unknown, and what would change the conclusion. A longer trial with a clinical sleep study and a CBT-I comparison could confirm or shrink this result.

## 9. Personal and commercial interpretation

What changes for an individual listener is mostly a mental model. Sleep support and exercise need not be treated as isolated boxes; a program can improve the continuity of sleep without increasing its total duration; and a result that is promising for a group is not an instruction for a specific person. If you have persistent sleep problems or a medical condition, a qualified professional, and the first-line option of CBT-I, matter more than any single trial.

For a wellness company the lesson is sharper. Do not call this an app trial: the sleep intervention included individualized counseling, a trained researcher, and seven weekly follow-ups, and removing that human layer could change adherence and outcomes. Do not market productivity gains: the trial measured sleep, not output. And do not imply superiority over standard care: there was no comparison with CBT-I. A responsible product test would name the whole service, predefine the customer outcome, measure adherence and adverse events, compare against a credible alternative, and follow people long enough to see whether effects last. The most interesting commercial idea here is not "sell exercise plus sleep." It is "design and evaluate the support system as a system."

## 10. Skeptical checklist and exact source card

Six quick questions before repeating any headline. Who was studied? Sedentary women of Chinese nationality, eighteen to thirty, with poor sleep health. What was compared? Two supervised programs, alone and together, against usual lifestyle, but not against CBT-I. What improved? Some objective measures of sleep continuity favored the combination. What did not? Objective total sleep time, and the combined program's edge on self-rated sleep. How long do we know? Eight weeks, no follow-up. What is the product boundary? Human guidance was part of the sleep program, so the study does not isolate an app-only effect.

The academic source is "High-Intensity Circuit Training Plus Sleep Health Intervention for Sleep Improvement: A Randomized Clinical Trial," by Borui Zhang and colleagues, published in JAMA Network Open on February 16, 2026. The digital object identifier is 10.1001 slash jamanetworkopen dot 2025 dot 56927. The PubMed identifier is 41697702, the PubMed Central identifier is PMC12910389, and the reported trial registration is ChiCTR2400086853. The article is available under a Creative Commons Attribution, NonCommercial, NoDerivatives license. The field-context studies named in this episode are Kredlow and colleagues 2015, Banno and colleagues 2018, Baglioni and colleagues 2020, Zhou and colleagues 2026, and Bu and colleagues 2025, each cited from its own published record. This episode is newly written commentary and does not reproduce the papers' prose, figures, or tables.

The final verdict: a strong, current trial with a useful positive signal on sleep continuity, a clear null on sleep duration, and narrow boundaries. Against a field where objective gains have been hard to find and CBT-I still sets the bar, the combination may improve parts of sleep continuity for people like those studied. The next step is broader, longer replication with a clinical sleep study and a head-to-head comparison, not a universal prescription.