Orion users experienced better sleep against placebo

The study ran in two phases, each seven nights long. In the first phase the device was running a placebo mode: it did not perform its regular functionality of both temperature regulation and simultaneous water circulation. It still powered on, ran its fans, and simulated an active state, but it did no actual heating or cooling. The placebo design was intended to prevent participants from knowing what thermal regulation schedule was being applied to their mattress topper. Nothing changed about the placement or location of the device or any of its parts in the participant's home. The sleep environment remained exactly the same, with the only exception being the placebo state, which was applied programmatically and remotely to the device.

In the active phase participants had full control over their devices and were not held to any fixed protocol. They engaged in the typical Orion user experience, each setting their own overnight temperature schedule.

Participants included 8 men and 6 women, with a mean age of 37.3 years (range 26 to 50). All were healthy sleepers with no reported medication use or diagnosed sleep disorders.

Sleep was measured across both phases with a short self-report survey that participants were asked to complete within two hours of waking each morning, and with wearable devices worn overnight, including the Oura Ring. Each person's sleep was then compared against their own results from the other phase, which removes the wide person-to-person variation in sleep and is what makes a sample this size informative.

For each survey measure (n = 14) we took every participant's average across the placebo phase and their average across the active phase, computed the difference for that person, and then averaged those per-person differences across the group. The Oura data (n = 13, with one individual excluded for too many missing nights) were handled the same way, using only main overnight sleeps and excluding naps.

Every self-reported measure, thermal comfort, mental sharpness, mood, sleep quality, and restedness, trended positively from phase one to phase two. Thermal comfort, mental sharpness, and mood reached statistically significant increases, and thermal comfort was higher in 13 of the 14 participants. On average, thermal comfort rose 24.8% from phase one to phase two, mental sharpness increased 14.0%, and mood improved 11.4%.

The objective measures most plausibly linked to a conditioned sleep surface, how quickly people fell asleep, how much they woke through the night, overall sleep efficiency, and their overnight cardiac profile, all drifted favorably rather than randomly. Sleep onset latency fell 20.1%, wake after sleep onset dropped 12.2%, and sleep efficiency rose 1.1% from phase one to phase two.

On the morning survey (0 to 100 scales, higher is better; n = 14), the active phase was rated higher than placebo on every affective and comfort measure. The from-to figures and point changes are group averages; “up to” marks the single strongest individual response. Thermal comfort rose from 66.6 to 83.1 (+16.5 points; p = 0.001; up to +215%), mental sharpness from 68.5 to 78.1 (+9.6 points; p = 0.009; up to +84%), and mood from 70.1 to 78.1 (+8.0 points; p = 0.028; up to +116%), all three reaching significance. Restedness (71.5 to 78.2; +6.7 points; p = 0.065; up to +124%) and perceived sleep quality (72.9 to 79.4; +6.5 points; p = 0.073; up to +79%) moved up by a similar margin, and self-reported sleep period was essentially unchanged (7.9 to 7.8 h; -0.03 h; p = 0.887; n = 13).

On the Oura Ring (n = 13), the three largest favorable shifts all fell in the sleep-continuity cluster. Sleep onset latency fell from 21.7 to 17.4 min (-4.4 min, roughly 20% faster; p = 0.068; up to -49%), wake after sleep onset fell from 69.9 to 61.4 min (-8.5 min, -12.2%; p = 0.096; up to -40%), and sleep efficiency rose from 86.4 to 87.4% (+0.97 points, +1.1%; p = 0.166; up to +6%). Each moved in the favorable direction without reaching significance.