Continuous Postoperative Monitoring Reduces Hypoxemia, Not Other Thresholds

Key Takeaways

• Continuous monitoring was associated with shorter time spent with oxygen saturation below 90% during early ward recovery.
• Durations of hypotension and tachycardia were not significantly different, and recorded interventions and short-term clinical events were broadly similar.
• The trial was pragmatic, and implementation disruptions plus incomplete bedside documentation complicated interpretation.

Continuous unblinded monitoring was associated with less time with oxygen saturation below 90% than clinician-blinded monitoring with intermittent 4-hour checks in higher-risk adults after noncardiac surgery. The comparison came from a cluster randomized crossover JAMA Network Open trial on postoperative wards. Median hypoxemia duration over the first 48 hours was 70.8 minutes with continuous monitoring and 103.5 minutes with intermittent checks. The same pattern was not seen for hypotension or tachycardia. The findings were concentrated in hypoxemia rather than across monitored abnormalities.

The multiple crossover cluster trial was conducted on 2 postoperative wards at Atrium Health Wake Forest Baptist Medical Center from October 7, 2020, to October 7, 2021. Final analysis included 798 adults recovering from noncardiac surgery, with 404 assigned to continuous monitoring and 394 to intermittent monitoring. Per protocol, the analytic cohort was limited to adults aged 65 years or older, or 45 years or older with at least 1 cardiovascular risk factor. Median age was 70.7 years, 55% were female, and 70% had ASA class III. The primary observation window covered the first 48 ward hours or continued until the third postoperative morning.

At Atrium Health Wake Forest Baptist Medical Center, continuous monitoring was unblinded, and live values were visible to clinicians during intervention periods. Comparator periods used blinded continuous data plus routine intermittent vital-sign checks every 4 hours. Prespecified alert thresholds were mean arterial pressure below 65 mm Hg, oxygen saturation below 90%, and heart rate above 110 per minute. The ViSi Mobile system captured oxygen saturation, heart rate, respiratory rate, 5-lead ECG, and noninvasive blood pressure estimated from pulse transit time, while blinded periods retained fail-safe extreme alerts. The prespecified primary outcomes were the durations of threshold breaches for hypotension, hypoxemia, and tachycardia during the defined monitoring window.

Adjusted analyses showed lower hypoxemia duration with continuous monitoring, with a GMR of 0.86, a 95% CI of 0.78-0.95, and P=.002. Mean arterial pressure below 65 mm Hg did not differ significantly, with a GMR of 0.94, a 95% CI of 0.87-1.01, and P=.11. Heart rate above 110 per minute also was not significantly reduced, with a GMR of 0.96, a 95% CI of 0.87-1.05, and P=.36. Exploratory composite outcomes A and B were lower with continuous monitoring, although individual components other than oxygen saturation were not independently significant. The measurable difference remained concentrated in hypoxemia rather than across all monitored thresholds.

Recorded clinical interventions were broadly similar between groups, and new oxygen therapy was the most common recorded intervention. Oxygen therapy was started in 224 patients with continuous monitoring and 214 with intermittent monitoring. ICU transfer occurred in 21 versus 31 patients, while rapid response calls occurred in 33 versus 44. Naloxone use was 37 versus 55, and in-hospital mortality was 4 versus 7; exploratory myocardial injury after noncardiac surgery was 60, or 14.9%, versus 65, or 16.5%. These exploratory and short-term clinical results were broadly similar across monitoring strategies.

Administrative and technical problems shortened the planned alternating 4-week exposure blocks to 26 to 37 days and limited protocol implementation. Bedside nursing responses were not adequately documented in real time, and signal quality or time away from monitoring were not captured for individual patients. Alarm fatigue could not be prospectively assessed, and the device's blood pressure estimates had not been externally validated. Generalizability was limited because the trial was conducted at a single center on only 2 wards, and staff knew the study was underway.

The investigators concluded that continuous ward monitoring appeared to reduce time spent with hypoxemia more than time spent beyond other thresholds or short-term clinical outcomes.