To the best of our knowledge, this is the first thorough description of outcomes of such a large cohort of consequent patients who had unplanned mechanical ventilation during their PACU stay. This study provides data from 2 centers for a population of 698 patients undergoing surgery and requiring unplanned postoperative ventilatory support. Nearly 20% of included patients died before hospital discharge; mortality rate of 36% was found in those who needed protracted mechanical ventilation (beyond PACU stay) and 9% of patient with short need for mechanical support.

It appears that in most reports, roughly 1–2% of surgical patients unexpectedly require postoperative ventilatory support. The LAS-VEGAS observational study reported an overall incidence of unexpected prolonged need of postoperative mechanical ventilation of about 1%, reaching 2.3% among patients at increased risk of postoperative pulmonary complications [Schultz et al. 2017]. Another observational study of 1200 patients undergoing major surgery also reported that 1.7% required prolonged postoperative mechanical ventilation/re-intubation [Fernandez-Bustamante et al. 2017]. This estimation is in agreement with our finding of need for mechanical ventilation at surgery conclusion in 1–2% of the total surgical volume during the relevant study period (N approximately 61,000, data not shown). Our current study adds to the available knowledge by reporting the outcomes of these specific 1–2% of surgical patients.

The overall in-hospital mortality of adult patients recovering from surgery is highly variable, depending on the populations evaluated. The International Surgical Outcomes Study (ISOS) group reported postoperative in-hospital mortality of 0.5% when considering only elective inpatient noncardiac surgical cases [Pearse et al. 2016]. The European Surgical Outcomes Study (EuSOS) group reported a much higher mortality rate of 4% when urgent and emergency cases were also included in the analysis [Pearse et al. 2012]. In another report, mortality among patients undergoing major non-cardiac and cardiovascular surgery was as high as 8% [Ghaferi et al. 2009]. A mortality risk stratification model found that patients with severe comorbidities, having urgent high-risk surgery, can have a mortality rate exceeding 10% [Glance et al. 2012]. The present study reports a mortality rate twice as high for the overall population of patients admitted to the PACU with unplanned need for ventilatory support and more than three times higher for patients in need of mechanical ventilation beyond PACU stay. As anticipated, substantial differences in crude- and risk-adjusted mortality rates were identified between patients extubated during their PACU stay and those not extubated. The latter were significantly sicker, most of them had undergone urgent or emergent surgery, often during nighttime. Yet, the alarming 9% in-hospital mortality rate among patients who were extubated shortly after admission to PACU (compared with mortality rate of 1% among our patients who did not require postoperative ventilatory support) needs further investigations to explore preventable interventions.

Postoperative complications are frequent and often severe [Bartels et al. 2013]. The incidence of complications after noncardiac surgeries is as high as 35% [Belcher et al. 2017]. Pulmonary complications are the most common. Even mild pulmonary complications are associated with longer duration of hospital and ICU stay, greater risk of ICU admission, and death [Fernandez-Bustamante et al. 2017]. The ARISCAT observational study reported an overall postoperative pulmonary complication rate of about 10% [Canet et al. 2010]. Specifically, respiratory infections occurred in 1.6% of patients. The LAS-VEGAS observational study [Schultz et al. 2017] also reported an overall 10% incidence of respiratory complications. Nonetheless, patients at increased preoperative risk for postoperative respiratory complications had nearly double that incidence (19%). Pneumonia was reported in 0.4% of patients, but incidence increased to 1.1% among high-risk patients. In the present study, significantly much higher incidence of respiratory infections (22%) among patients who remained intubated beyond PACU stay reflect the preoperative status of the patients (for example, almost 30% were admitted with active infection), the complicated perioperative course, and the need for prolonged intubation. Yet, 14% respiratory infection among patients extubated already in the PACU needs further investigation to comprehend potential causes and interventions that might assist in reducing the need for short postoperative ventilatory support. Similarly, whereas 8% of surgical inpatients recovering from noncardiac surgery across Europe [Pearse et al. 2012] were admitted to the ICU (the EuSOS study), three times as many patients already extubated in there PACU were transferred to the ICU. Although this might reflect the unique risk that caregivers attributed to the complicated perioperative course of these patients, the specific cause needs to be sought in further studies.

Overall, our results reflect the high risk for postoperative morbidity and mortality of this unique patient population that unexpectedly requires postoperative respiratory support. Future investigations should be aimed at identifying these patients before surgery, stratifying their perioperative risk, and eventually making a better-calculated decision on acceptable risk-benefit balance.

Our study suffers several limitations. First, like most observational studies, it is retrospective in nature and is therefore limited in terms of data quality. It should be noted, though, that despite the large cohort size, data were collected by manual chart review of each case by trained anesthesiologists and are therefore presumably as accurate as patients’ medical records. Second, we did not collect data about a control group of patients with normal clinical course and extubation at surgery conclusion. Nevertheless, comparing mortality alone, the reported 30-day mortality among adults undergoing noncardiac surgery is 1–2% [Devereaux et al. 2012, 2017] comparable with a 1% in-hospital mortality rate among all adult surgical patients in our institutions.