Evidence for Owlet Smart Sock
The Owlet Smart Sock shows reasonable accuracy for heart rate monitoring in preterm and term infants but has significant limitations for oxygen saturation detection, particularly for clinically important hypoxemia events, and should not replace medical-grade monitoring in clinical decision-making.
Accuracy Data
Heart Rate Monitoring
The device demonstrates strong performance for heart rate measurement:
- In preterm infants <2.5 kg, the Owlet Smart Sock 3 showed excellent correlation with medical-grade Masimo pulse oximetry (r = 0.98, p < 0.001), with a mean difference of only -1.3 bpm and narrow limits of agreement (-6.3 to 3.4 bpm) 1
- A separate inpatient study of 50 neonates confirmed high correlation for heart rate (r = 0.926) with excellent interclass correlation (ICC = 0.961) 2
Oxygen Saturation Monitoring
The device has substantial limitations for SpO2 measurement:
- While showing positive correlation with medical-grade devices (r = 0.71, p < 0.001), the precision decreased significantly at lower oxygen saturations 1
- The estimated ARMS (average root mean square) was 2.3% for SpO2 in the 70-100% range, but accuracy worsened below this range 1
- In term and preterm neonates, SpO2 correlation was poor (r = 0.242 and 0.521, respectively) with only moderate interclass correlation (ICC = 0.719) 2
- Most critically, a 2025 study found the sensitivity for detecting SpO2 <80% for ≥3 seconds was only 14% with smoothed data and 74% with raw data, though specificity remained >96% 3
- For bradycardia events (HR <50 bpm ≥3 seconds), sensitivity was extremely low at 6% for smoothed data and 39% for raw data 3
Clinical Context and Limitations
Not a Substitute for Medical Monitoring
The American Academy of Sleep Medicine position statement, while addressing home sleep apnea testing broadly, emphasizes that home monitoring devices measuring fewer physiologic variables than polysomnography may lead to underestimation of disease presence or severity 4. This principle applies to consumer devices like the Owlet sock.
Key clinical pitfalls:
- The device has high specificity (low false alarm rate) but poor sensitivity, meaning it will miss many clinically significant events 3
- Sensitivity improves only with longer event durations and higher threshold values, meaning brief but potentially dangerous events are frequently missed 3
- Motion artifacts significantly affect data quality 1
- The device has not been validated against arterial blood gas measurements 1
Potential Clinical Applications
Where it may have utility:
- In infants with supraventricular tachycardia, Owlet use was associated with better-preserved left ventricular ejection fraction at ED presentation (p = 0.04) and 1.7 days shorter hospital stays (p = 0.02), though users also had more phone calls and ED visits 5
- For general parental reassurance in healthy term infants, with 94% of parents reporting better sleep quality 6
- Sleep pattern tracking in research settings for developmental studies 7
Where it should NOT be used:
- As a replacement for medical-grade monitoring in high-risk infants
- For clinical decision-making regarding oxygen therapy weaning (the ATS guideline emphasizes appropriate pediatric equipment and reference ranges for such decisions 8)
- For detection of brief but clinically significant cardiorespiratory events
- In situations where missing a hypoxemia or bradycardia event would have serious consequences
Practical Considerations
The device is primarily used by first-time parents in their 30s with college education, with 75% citing "peace of mind" as the reason for purchase 6. While 82% of users followed safe sleep guidelines 6, clinicians should counsel families that:
- The device cannot prevent SIDS or detect all dangerous events
- Medical-grade monitoring remains necessary for infants with cardiorespiratory conditions
- The high false-negative rate (low sensitivity) means absence of an alarm does not guarantee safety
- Parents should respond to clinical signs (color change, breathing difficulty, unresponsiveness) regardless of device readings