Perioperative hypoxemia outcomes
The POH in the current study was found to be 30.0%. Ehrenfeld et al. demonstrated an intra-operative hypoxemia rate of 6.8% , while several other studies have documented PACU hypoxemia rates ranging from 17% to 50% [47–52]. Several investigators have found substantial POH during the first few days following abdominal surgery [53–55] or hip fracture surgery . Lampe et al. published the only study that monitored pulse oximetry following discharge from the PACU in a group of patients undergoing diverse operative procedures . During the first 24 hours following surgery, POH occurred in 60% of patients, with oxygen saturation improving on post-operative day two. Similar to our study, the literature corroborates the observation that post-operative POH is a frequent entity.
It is likely that the POH rate in the current study would have been higher had we not excluded hypoxemic events occurring during the first two hours following surgery. Investigators have described high rates of POH during the first post-operative hour in the PACU; however, they radically decrease over the subsequent one-hour [47, 52]. For the patients with POH, the number of days from surgery until hospital discharge was greater, when compared to those without POH. The two-day increased hospitalization for the 150 hypoxemic patients represents 300 days for the two months, which extrapolates to 1,800 additional hospital days for the year. Of importance, POH had an independent correlation with post-operative length of stay.
The rate of POH was substantial in virtually all of the 12 operative procedure categories. Although the primary operative body position was supine or lithotomy, the standard anesthesia practice was to maintain horizontal recumbency in all patients, except for the few patients in the sitting position. POH was associated with age, abdominal hypertension, weight, BMI, cranial procedures, decubitus position, ASA level, duration of surgery, and inability to perform extubation in the OR. Perioperative hypoxemic patients were older; however, the average remained less than 65, indicating that they were not elderly. According to the literature, PACU POH has been associated with the following similar conditions: increasing age , obesity [49, 50], ASA level [48, 49], and duration of surgery [48, 49]. The association of abdominal hypertension with POH in the current study may represent a mechanical effect, similar to weight, BMI, and obesity. The reasons for increased POH with the decubitus position and cranial procedures are uncertain. Conditions independently associated with POH in the current study were acute trauma, BMI, cranial procedures, ASA level, and duration of surgery. Lampe et al. found that post-operative oxygen saturation values were lower with older patients; however, age did not significantly increase the rate of POH in the post-operative period .
Perioperative hypoxia mechanism
To try to understand the potential mechanistic foundation for POH in the current study is intriguing. The analysis indicates that intra-operative fluid excess, elderly-age, and pre-existing lung disease were not POH risk factors. However, POH was associated with older age, abdominal hypertension, acute trauma, weight, BMI, cranial procedures, decubitus position, ASA level, duration of surgery, and glycopyrrolate administration. These observations suggest that conditions other than pulmonary edema or obstructive-restrictive lung disease were principals.
We found that glycopyrrolate administration was an independent predictor of POH. Parenteral glycopyrrolate has been shown to decrease oral, tracheobronchial, and gastric secretions [57–60]. Although the precise reasons for administering intravenous glycopyrrolate in the current study are unclear, administration is a discretionary decision  and is typically considered when it is important to decrease secretory production or prevent bradycardia . The lower POH rate with glycopyrrolate is mechanistically consistent with the notion that pulmonary aspiration may have been a factor in patients developing POH. The lower POH rate with glycopyrrolate establishes an additional link, along with duration of surgery, decubitus positioning, and cranial procedures, between POH and events that transpired during the operative procedure. Further, the multiple intra-operative conditions associated with POH (duration of surgery, glycopyrrolate administration, cranial procedures, and decubitus position) and the increased rate of inability to extubate POH patients in the operating room suggests that POH pulmonary injury was related to intra-operative events.
Some of the conditions associated with POH in the current study have also been linked to POPA or regurgitation and include the following: increased age [4, 9, 22], acute trauma [24, 31], obesity [9, 22, 24, 30], increased ASA level [11, 22, 30], and increased duration of surgery [6, 30]. In the current study, the rate of POH for open laparotomy was 49% and abdominal hypertension was found to have an association with POH. Some experts have found evidence that abdominal pathology and procedures increase the risk for POPA .
Just as POH was found to be a ubiquitous event in the current study, Blitt et al. found compelling evidence, in a prospective study, that regurgitation occurred in all surgical body positions . Other researchers have also found pervasive presence of POPA among the multiple types of surgery that were investigated in each of four studies [4, 8, 9, 11]. The current study findings and literature documentation are consistent with the notion that POH, in part, may be a manifestation of occult- or micro-pulmonary aspiration during horizontal recumbency.
Perioperative pulmonary aspiration outcomes
Published POPA rates are higher (1.4% to 2.9%) for investigations from voluntary claims reporting databases [5, 12–14], when compared to studies emanating from comprehensive database reviews (0.01% to 0.9%) [4, 6–11]. The nearly 5% POPA rate in the current study is higher than any published rate, a finding especially noteworthy when considering that our investigation is functionally a comprehensive database review. The seven historic comprehensive database reviews include the intra-operative and early post-operative periods in three studies [8, 10, 11], the intra-operative period only in three investigations [6, 7, 9], and an unspecified time period in one study . Data emanates from an anesthesia database in five of the investigations [7–11], a prospective database in one , and a statewide surgical database in another .
In the seven comprehensive database studies, the traits for determining POPA included non-respiratory secretions in four investigations [7, 8, 10, 11] and post-operative chest x-ray infiltrates in two studies [10, 11]. The investigation by Kozlow et al. required a discharge diagnosis of aspiration pneumonia within a statewide surgical database . The study by Olsson et al. did not specify the characteristics for POPA, only that it be documented in the anesthesia database . The Blitt et al. research was prospective and was an active search for regurgitation and aspiration .
The higher POPA rate in the current study is likely related to our reliance on POH monitoring as a signal for potential POPA and extending the period of observation to the first 48 post-operative hours. Ideally, all patients would have had a pre-operative and post-operative chest x-ray to detect a new perioperative infiltrate. This might have revealed a similar, higher, or lower POPA rate compared to the current study results. A requirement for pre-operative and post-operative radiographs in all patients would create operational complexity, e.g., funding for the investigation. Although one may quibble with our methodology, the fact that POPA patients had a higher mortality and substantially long hospitalization following surgery provides credibility.
Mortality was greater in the patients with POPA, when compared to the patients without POPA. Historic data documented in five publications provides evidence that POPA mortality rates have ranged from 1.5% to 15.6% [5, 9, 11, 14, 63]. Further, Kozlow et al. showed that POPA mortality was increased with an odds ratio of 7.6, when compared to patient mortality without POPA .
In the current study, the number of days from surgery until hospital discharge had nearly a four-fold increase in POPA patients, when compared to those without POPA. Importantly, POPA was independently associated with post-operative length of stay, along with duration of surgery and an acute traumatic condition. The study by Kozlow et al. demonstrated that surgical patients with aspiration pneumonia had a total hospital stay of nine days longer, in comparison to the non-POPA group . Of relevance, investigators have demonstrated that admission to an ICU has been warranted in 27% to 57% of patients with POPA [10, 11, 14].
In the current study, POPA had associations with cranial procedure, decubitus positioning, ASA level, duration of surgery, failure to extubate in the OR, and prolonged post-operative intubation. Of relevance, the proactive investigation by Blitt et al., demonstrated that nine percent of patients under general anesthesia were demonstrated to have regurgitated  and Kluger et al. showed that 55% of patients with vomiting or regurgitation had pulmonary aspiration . The Blitt study also proved that regurgitation was significantly more likely when the duration of the operative procedure was > two hours . The Blitt investigation further showed that regurgitation occurred in 8% with decubitus positioning and 17% of neurosurgical procedures . The increased rates of inability to extubate POPA patients in the operating room and prolonged post-operative intubation, in the current study, suggest that the pulmonary inflammatory process was related to the surgical procedure. Increased ASA levels have also been documented in the literature to be associated with higher rates of pulmonary complications  and POPA .
Substantial evidence from the literature indicates that horizontal recumbency during mechanical ventilation creates a risk for pulmonary aspiration with lung injury [22, 31] or ventilator-associated pneumonia [17, 18, 32–37]. The supine, lithotomy, prone, decubitus, and sitting positions are considered to be the most common anatomic postures utilized during surgical procedures [6, 39, 40]. In the current study, the primary operative body position was supine or lithotomy, a finding analogous to that of Blitt et al. . In the current study, standard anesthesia practice was to maintain horizontal recumbency, except for the few patients in the sitting position. Horizontal recumbency, for the typical operative body positions, is promulgated within the operative nursing literature and teaching circles, as common practice [39–41]. Specifically, horizontal positioning is disseminated by the use of specific narrative description statements [39, 40] and inclusion of illustrations and photographs [39, 41, 64] that portray horizontal recumbency.
We evaluated four review publications, related to POPA, for comments regarding body positioning. The most current review includes only a single comment regarding body positioning  and another makes no mention of body positioning . Ng et al. indicate that the Trendelenburg position is a risk for POPA and lithotomy positioning may be a risk . The greatest attention to body position, as a risk for POPA, was in a review publication by Kalinowski et al. in 2004 . Relevant statements in the manuscript indicate that aspiration is common in patients with impaired consciousness in the supine position and with successful tracheal intubation, pulmonary aspiration appears to be less frequent if the head is elevated 45 degrees .
In the current study, POH was a common occurrence among the various intra-operative body position postures and the multiple surgical procedural categories. Because POH and horizontal recumbency were pervasive in the current study, it is compelling to consider that these two conditions may be linked. We believe the multiple findings in the current study and the literature link horizontal recumbency to POPA and POH.
Routine pre-operative and post-operative radiographic chest imaging would have been ideal. Clear lung fields on the pre-operative film would have provided greater evidence that each patient had pre-surgical pulmonary stability. However, the pre-operative SpO2 and respiratory rate values are convincing. Routine post-operative chest imaging would have provided a more accurate determination for pulmonary inflammation in patients with or without POH. Thus, the rate of POPA would have been more precise. However, the POPA rate would have only increased, because we did not categorize any patient with POPA, unless a concomitant chest radiographic image demonstrated a pulmonary infiltrate.