PACU discharge is a handoff into the window where respiratory depression happens

Why it matters

The PACU discharge order is written by the person with the most continuous view of the patient, and it sends the patient to the place with the least. The sources for this day were designed to see what ward observation cannot. On the ward, with blinded continuous monitoring, 46% of patients on parenteral opioids had at least one respiratory depression episode khanna-2020. Nurses missed 90% of episodes in which saturation stayed below 90% for at least an hour sun-2015. The first episode arrived a median 8.8 h after surgery, and all episodes peaked between 02:00 and 06:00 driver-2021.

The consequence of getting this wrong is a discharge decision made as if the PACU’s observation continued after the patient left it.

Mechanism

Opioids depress the brainstem response to carbon dioxide and, with sedatives and residual anaesthetic, reduce upper airway tone and the arousal that would normally terminate an obstructed or hypopnoeic breath. The result is episodic rather than continuous: periods of slow or shallow breathing, or obstruction, interrupted by arousals, each episode ending before the next spot check and leaving nothing behind for an intermittent observer to see. Sleep deepens the depression, which is why the events cluster overnight. Continuous monitoring sees the episodes; a check every few hours samples the gaps between them.

Evidence

PRODIGY monitored 1335 patients on parenteral opioids with blinded continuous capnography and oximetry at 16 sites in the US, Europe and Asia, with vital signs otherwise taken intermittently per standard of care khanna-2020. A respiratory depression episode was a respiratory rate of 5 per minute or less, a saturation of 85% or less, an end-tidal carbon dioxide of 15 mmHg or less or 60 mmHg or more for at least 3 min, apnoea longer than 30 s, or any respiratory opioid-related adverse event; 46% of patients had at least one, and a risk score was derived from 46 candidate factors khanna-2020.

Driver 2021, a post hoc analysis of PRODIGY patients from two US sites, found 2539 episodes in 155 of 250 patients (62.0%), a median time to the first episode of 8.8 h (5.1 to 18.0), a peak of initial events between 14:00 and 20:00 and a peak of all episodes between 02:00 and 06:00 driver-2021. Monitoring began a median 4.3 h (3.2 to 6.2) after the end of surgery, well after PACU discharge driver-2021.

Sun 2015 started blinded continuous oximetry on arrival to the ward after PACU or ICU discharge in 833 patients sun-2015. Saturation was below 90% for at least 10 min per hour in 21% and at least 20 min per hour in 8%; 37% had at least one episode below 90% lasting an hour or more, 11% an episode of 6 h or more, and 3% were below 80% for 30 min or more sun-2015. Nursing records captured clinical hypoxaemia in only 5% of the subset examined, and nurses missed 90% of episodes below 90% lasting at least an hour sun-2015.

Taenzer 2010 introduced continuous oximetry surveillance with pager notification on one postoperative unit, 11 months before and 10 months after, with two concurrent unchanged comparison units: rescue events fell from 3.4 (1.89 to 4.85) to 1.2 (0.53 to 1.88) and ICU transfers from 5.6 (3.7 to 7.4) to 2.9 (1.4 to 4.3) per 1000 patient days taenzer-2010. The denominator was corrected by erratum ten years later, from patient discharges to patient days; the authors state the findings are unaltered taenzer-2010-erratum.

Inside the PACU, Epstein 2014 recorded saturation every 30 to 60 s in 137,757 patients and compared hypoxaemic episodes, saturation below 90% for at least 2 min, beginning before and after 30 min from admission; it carries the observation that desaturation is not front-loaded into the first minutes of the PACU stay epstein-2014.

What this does not show

Both of the day’s main ward sources exclude the PACU by design. Driver’s monitoring began a median 4.3 h after surgery and Sun’s began on arrival to the ward, so neither can support a claim that PACU observation is structurally blind driver-2021 sun-2015. PACU observation is continuous by staffing rather than by device, and the only PACU-specific source here is about the timing of desaturation within the stay, not about opioid-induced respiratory depression as such epstein-2014. That is why the day is framed around the discharge decision and the window it hands the patient into, not around the PACU itself.

Nothing here is a randomised trial. PRODIGY is prospective observational with blinded monitoring, Sun 2015 is blinded observational, Driver 2021 is a post hoc analysis, Epstein 2014 is a retrospective analysis of recorded data, and Taenzer 2010 is a before-and-after study with concurrent controls khanna-2020 sun-2015 driver-2021 epstein-2014 taenzer-2010. No trial on this page shows that continuous monitoring changes outcome, in the PACU or on the ward. Taenzer’s fall in rescue events is the closest thing to it and it is not randomised taenzer-2010.

The PRODIGY primary record carries a PubMed identifier ambiguity in one institutional record that has not been resolved khanna-2020.

At the bedside

When you write the PACU discharge order, name the window you are sending the patient into: the first episode is likeliest hours from now and the peak is overnight driver-2021. Ask what will be watching then, and how often.

Treat the intermittent ward check as what the sources show it to be, a sample that misses most prolonged desaturation sun-2015. Where continuous monitoring exists for a patient on parenteral opioids, the case for using it after PACU discharge is the case this day makes; where it does not, the discharge decision should be made knowing that. [PRACTICE VARIES: which patients receive continuous oximetry or capnography after PACU discharge, and on which units, is an institutional decision; find out what your wards can see.]

Do not read a quiet PACU stay as evidence that the ward will be quiet. The PACU is the part of the day when someone is looking.

Check yourself

Four items. Every option carries an explanation. Progress is not saved.