Pulmonary aspiration of gastric contents during the perioperative period is rare, but it can be associated with significant morbidity and mortality. Since the initial comprehensive review of the incidence of pulmonary aspiration by Mendelson in 1946 (1), several large retrospective observational studies have been published (2–6). Most of these reports concentrated on either the pediatric population (2,3) or mixed populations (4,5) and only one study focused on the adult surgical/obstetric population (6). In the past 10 yr, no large population-based studies describing the incidence and outcome of perioperative pulmonary aspiration (PPA) have been published. We therefore performed a 4-yr retrospective study (January 2001–December 2004) in an adult, nonobstetric surgical population at a tertiary university medical center to examine the recent incidence and the outcome of PPA. In our medical center, all the pediatric and obstetric procedures were performed in the individual specialty hospitals.
During the 4-yr study period (January 2001–December 2004), 99,441 anesthetics were administered at our institution to patients aged 18 yr or older. Of these 90,493 (91.0%) were elective and 8948 (9.0%) were emergent surgical or diagnostic procedures from various surgical specialties, with the exception of obstetrics.
With the approval of the Total Quality Council at our institution, we first identified suspected cases of PPA. We used two separate databases for the identification of the potential PPA cases: The Department of Anesthesiology Quality Improvement (QI) database and the hospital-based electronic medical archive system. The web-based Department of Anesthesiology QI database was accessed through a computer in each operating room and in the postanesthesia care unit (PACU). Throughout the study period, a policy was in place and enforced mandating that all anesthesia-related complications be entered by the anesthesia providers within 24 h. The entire QI database was reviewed for entries of either “aspiration” or “respiratory problems” during the 4-yr study period. At the same time, the hospital-wide medical archive recording system, which includes all the electronic dictation discharge records, operative records, as well as intensive care unit (ICU) records, was searched for the following key words: “pulmonary aspiration” and “perioperative” or “postoperative.”
After identification of the patients with suspected PPA, individual anesthetic and medical records were thoroughly reviewed to confirm the diagnosis. The diagnostic criteria that were used to confirm the diagnosis were based on the report by Warner et al. (6) and included either of the following: 1) the presence of bilious secretions or particulate matter within the tracheo-bronchial tree by direct suctioning or by fiberoptic bronchoscopy, or 2) after the episode of passive regurgitation, postoperative chest radiograph demonstrated a new infiltrate that did not exist in the preoperative chest radiograph or on physical examination and that developed postoperatively within 24 h. Therefore, episodes of passive regurgitation, which occurred while patients were unconscious were not automatically considered to cause pulmonary aspiration unless the above diagnostic criteria were met. The exclusion criteria for the study were the following: Any patient whose airway was established before arrival to the operating room or a patient who underwent some kind of airway instrumentation postoperatively by nonanesthesiology personnel. We recorded the following data: age, sex, height, body weight, body mass index, preoperative comorbidity, airway examination result, preoperative medication, ASA physical status (ASA-PS), anesthetic management, hospital course, and patient’s outcome.
Statistical analysis was performed using the χ2 test when appropriate. A P value <0.05 was considered to be statistically significant (Stata version 9.0; StataCorp LP, College Station, TX).
A total of 99,441 anesthetics were performed during the 4-yr period. The male/female ratio was 1.11 to 1 (52,357/47,085). 30.2% (30,066) of patients were ≥65 yr of age, while 6.1% (6025) were ≥80 yr of age. Emergency cases accounted for 9.0% of the total case load.
Thirty-five patients with suspected PPA were identified in the QI database during the 4-yr period. Among them, 13 fulfilled the diagnostic criteria and were confirmed with PPA. The reasons for the exclusion of the other 22 patients are detailed in Table 1. Similarly, 39 patients were identified with the key words “pulmonary aspiration” and “perioperative” or “postoperative” in the hospital-wide medical archive recording system. Among them, 14 were confirmed with PPA according to the diagnostic criteria. The other 25 did not fulfill the PPA criteria (Table 1). All 13 patients who had been identified by the QI database were detected in the medical archive system as well. One additional case was identified as a confirmed PPA by this secondary data search. This patient was a 36-yr-old female who underwent partial esophagogastrectomy with roux-en-Y esophagojejunostomy via thoracotomy for congenital tracheoesophageal fistula. The surgical note described the presence of some gastric content in the trachea, which was suctioned under direct fiberoptic bronchoscopy at the end of surgery after the exchange of a double-lumen tube to a single-lumen endotracheal tube (ETT). However, this incident was not reported with the QI form. Therefore, 14 patients were confirmed with PPA during the 4-yr study period. The incidence of PPA was 1 in 7103 anesthetics. The median age of these patients was 51 yr, with a range of 26–82 yr. Ten patients were male. Twelve patients were younger than 65 yr (the ratio to the total cases in this age group was 1:5781), while 2 patients were older than 80 yr (1:3013).
All 14 patients had one or more risk factors for PPA. These risk factors included esophageal pathology and/or previous esophageal surgery, concurrent opioid administration, gastrointestinal obstruction and/or dysfunction, recent oral intake, posttrauma, lack of coordination of swallowing and respiration, obesity, and depressed level of consciousness that was not due to head injury (6). Preoperative gastric acid reducers were administered in 8 patients, including proton-pump inhibitors in 4 patients and H2 blockers in the other 4 patients.
Preoperative airway examination revealed that one patient had a chronic tracheostomy, and two trauma patients had their cervical spine immobilized by a cervical collar as a precautionary measure. The Mallampati airway classification of all 14 patients was I or II.
Surgical/Diagnostic Procedures (Table 2)
Thoracic esophageal surgeries have the most frequent incidence of PPA (6 cases in 6993 thoracic-esophageal anesthetics), with an incidence of 1 in 1166, which was significantly more frequent than that of the nonthoracic esophageal surgeries (P = 0.00001). The two cases of PPA that occurred during orthopedic procedures were in posttrauma patients. In abdominal surgery, one case of PPA occurred during small bowel transplantation and another during repair of a perforated colon, both of which were emergent cases. One case of PPA was confirmed during each of the following procedures: esophago-gastro-duodenoscopy (EGD) (gastrointestinal medicine), placement of a vetriculo-peritoneal shunt (neurosurgery), breast reduction (plastic surgery), and bilateral blepharoplasty with forehead lift (plastic surgery).
Choice of Anesthesia and Intraoperative Management
The modes of anesthesia used during the entire study period were general anesthesia (GA), in 73,007 cases (73.4%), and monitored anesthesia care (MAC), in 26,434 cases (26.6%). PPA occurred during GA in 10 patients: eight with endotracheal intubation, one through a preexisting tracheostomy, and one who had been converted to GA using a laryngeal mask airway (LMA) after failed brachial plexus block. The other four PPA developed during MAC. Therefore, the incidence of PPA in GA was 0.014% and in MAC was 0.015% (P = 0.87). In GA patients, five PPA occurred after induction, two during laryngoscopy, two during the exchange of the airway devices (one during replacement of a tracheostomy tube and one during replacement of a double-lumen ETT by a single-lumen ETT), and one with LMA. In the MAC cases, all four incidents occurred during the surgical procedures, with the patients sedated. None of the PPA episodes occurred during emergence from anesthesia or during the recovery phase in the PACU.
Passive regurgitation was detected in 11 patients and active vomiting in 3. In all 14 cases, PPA occurred while the patients were unconscious under the effect of anesthetic medications. Of the 4 MAC cases, 3 were quickly converted to general endotracheal anesthesia (GETA) and in one, spontaneous breathing was maintained without any airway instrumentation. A fiberoptic bronchoscopic examination was performed in 8 of 14 patients, while the other 6 patients were treated with blind suctioning. The quantity of pulmonary aspirate was described as a “moderate to large amount” in 5 patients, 3 of whom developed pulmonary complications postoperatively. In 4 patients the volume of the pulmonary aspiration was documented to be a “small amount,” and none of them developed pulmonary complications postoperatively. In 5 cases there was no comment on the volume of aspirate. Only one of the documented aspirates contained particulate materials; these were described as a “coffee ground” material.
The documented signs and symptoms after PPA in our series included desaturation (six patients), wheezing (two patients), hypotension (one patient), and laryngospasm (one patient). The seven patients who did not show any signs or symptoms after the incident did not develop postoperative pulmonary complications.
After the recognition of PPA, patients were managed in the following ways: cancellation of the procedure (two cases), switching to GETA from MAC (three cases), switching to GETA from LMA (one case), IV administration of steroid (one case), and bronchial lavage with normal saline solution through fiberoptic bronchoscope (five cases) or through ETT (one case). Two patients were successfully tracheally extubated in the operating room and another two were extubated in the PACU. One patient remained with a tracheostomy mask and one patient with MAC remained without any airway instrumentation. Eight patients were kept intubated and were maintained on mechanical ventilation during the postoperative period.
Outcome After PPA
One patient was discharged home on the same day. Five patients were admitted to a regular floor for observation, and eight patients were admitted to the ICU.
Six patients developed respiratory complications after PPA. One patient had a full recovery on conservative treatment. Three patients required tracheostomies, and one patient recovered from respiratory failure and was discharged from the ICU (but died on the 25th postoperative day because of hemorrhagic shock, which was unrelated to PPA). One patient died after respiratory failure due to PPA. Therefore, pulmonary morbidity due to PPA in this series was 1 in 16,573 patients, with an overall mortality rate of 1 in 99,441. Table 3 shows the morbidity and mortality with respect to the ASA-PS of the cases. Pulmonary complications and death due to PPA occurred only in ASA-PS III and IV patients.
The sole death due to PPA occurred in an 82-yr-old man with end-stage liver disease and hepatic encephalopathy (ASA-PS IV), who underwent EGD for gastrointestinal bleeding under MAC. This patient developed passive regurgitation during the procedure and aspirated what was described in the anesthesia record as a “moderate amount of brownish coffee ground material.” During the incident, the patient became hypotensive and his oxygen saturation decreased quickly to below 60%. An ETT was placed immediately and IV infusions of vasopressers were initiated. The patient was transferred to the ICU. He developed adult respiratory distress syndrome and acute renal failure, and subsequently died on the 13th postoperative day.
Anesthetic Management Issues
Critical review of the anesthesia records of the 14 cases of PPA revealed that in 10 cases (6 GA and 4 MAC) the anesthetic management played a role in the outcome (Table 4). In four of the GA cases, cricoid pressure was not applied at the time of induction for intubation using a cuffed ETT. In the two MAC cases, GA with a cuffed ETT may have been a better choice because of the expected duration of surgery (one case) and the high preoperative risk for PPA (one case). In the other two MAC cases, the level of sedation was considered deep enough to compromise the patients’ ability to protect their airway.
This study confirmed that the incidence of PPA is rare, but that it is still associated with serious adverse events. Our review of anesthesia practice between 2001 and 2004 in our institution revealed that out of 99,441 anesthetics, 14 patients were confirmed with PPA, an incidence of 1 in 7103. Forty-three percentage of the PPA patients (6 of 14) developed pulmonary complications, a pulmonary morbidity of 1 in 16,573. Seven percentage of PPA cases (1 of 14) died, giving a mortality of 1 in 99,441.
Our report is the second study on the incidence and outcome of PPA that focuses on an adult surgical population. Warner et al. (6) from the Mayo Clinic reviewed the incidence of PPA for 215,488 general anesthetics delivered to an adult population between 1985 and 1991. The proportions of elderly patients (40% of the patients were ≥65 yr) and emergency cases (6% of the case load in their study) were comparable to our study. In the Mayo Clinic study, the same rigid criteria for the diagnosis of PPA were used as in our study. Their incidence of PPA was 1 in 3216, with a mortality rate of 1 in 71,829 and morbidity of 1 in 11,971 (6). They defined morbidity as any patient who required “intensive care or respiratory support or developed pulmonary complications.” In our study, the incidence of PPA was about half that of Warner et al. Although direct comparison of such epidemiological studies is often difficult and misleading, the single major difference in the incidence between the two studies can be found in the incidence of PPA in emergency cases: 15 in 13,427 anesthetics (1:895) for the Mayo study versus 2 in 8948 (1:4474) in our study. The reasons for the lower incidence of PPA in our emergency cases are not clear. They might be related to the regular allocation of senior hands-on providers in our hospital during the off-hour shifts, when the majority of the emergency cases were handled. Inclusion of the obstetric population in the Mayo study did not seem to be a factor in their higher incidence of PPA, because no pulmonary aspiration developed in the 645 cesarean deliveries in their study (6). Compared with other large studies on the practice reported decades ago, our study has the lowest incidence of PPA, but our patients still suffered significant pulmonary morbidity and mortality (Table 5).
Is there any way to prevent or to further decrease the incidence of PPA and its devastating morbidity and mortality? One of the most notable findings in our study is that 70% of the aspiration incidents were associated with anesthetic management that could have been improved with more aggressive reflux prophylaxis and/or airway control strategy. This suggests that most of the PPA might be related to the methods of anesthesia care, and therefore could have been prevented.
As described in Table 4, six of our patients who developed PPA (60%) in the GA group had anesthetic management issues. One patient who underwent esophageal intubation had achalasia, and another patient, in whom fiberoptic intubation was performed after induction without an apparent indication, had a history of esophagectomy. In both these cases the PPA may have been avoided. Staff failed to apply cricoid pressure while changing the airway device in two patients and before the induction/intubation in two other patients with full stomachs. “Proper” cricoid pressure application is often difficult in terms of the force needed to occlude the upper portion of the esophagus because of the anatomical alignment of the larynx and the esophagus (7). Also, the application of cricoid pressure can decrease the barrier pressure due to the reflex relaxation of the lower esophageal sphincter tone (8). However, this maneuver has been considered the standard of care in rapid sequence induction in order to prevent pulmonary aspiration. Unless the efficacy of this maneuver can be challenged in a prospective comparative study, its application should be mandatory in patients at high risk for regurgitation and PPA.
In the MAC group, all four patients’ anesthetic care could have been improved. The depth of sedation was a concern in two patients and the choice of MAC over GETA was a concern in the other two.
None of the large retrospective pulmonary aspiration studies published in the literature investigated or reviewed the anesthetic care of patients who suffered PPA. Interestingly, Kluger and Short (9) reported the factors contributing to PPA in their review of 133 cases from the Australian Anesthetic Incident Monitoring Study among the anonymous self-reporting database of 5000 incidents. They collected multiple factors related to the PPA, including error of judgment, improper technique, inadequate patient preparation, and others, and then they further postulated that the application of the simple antiaspiration guidelines might have prevented the incidents in 60% of all cases of aspiration.
All patients who experienced PAA in our study had one or more predisposing risk factors for pulmonary aspiration. Notable among the predisposing risk factors was esophageal pathology: six patients had preoperative esophageal conditions, including four with a history of esophagectomy. Accordingly, 7 of 14 cases (50%) of PPA occurred during esophageal procedures, including one diagnostic EGD. This finding is consistent with Olsson et al. (4), who reported 13 patients (16%) with esophageal disorders out of 83 who suffered PPA. Olsson et al. also reported 12 patients with EGD who had a history of esophageal pathology and developed PPA, an incidence of 1 in 188. Our finding further warrants the application of strict aspiration precautions in patients with esophageal pathology during esophageal-related procedures.
Warner et al. (6) demonstrated, in their landmark paper published in 1993, that pulmonary complications secondary to PPA were unlikely if the patients did not show signs and symptoms within 2 h after the incident of PPA. Their finding was reconfirmed in the current study, in which all patients who developed postoperative pulmonary complications showed signs and symptoms within 2 h of PPA. Five patients in our series did not show any signs and symptoms after the PPA, and none of them developed postoperative pulmonary complications. The development of postaspiration pulmonary complications and resultant mortality seems to relate to the preoperative condition of the patients. We confirmed the finding of previous studies (4,6) that none of the ASA-PS I patients developed PPA, while patients who were ASA-PS II suffered no serious or fatal consequences after PPA, and only the patients with ASA-PS III or higher suffered serious outcomes.
Data acquisition within any retrospective study is challenging. First, a self-reporting system of perioperative anesthetic complications using QI data, such as ours, potentially carries the risk of under-reporting. Although we implemented mandatory QI reporting in the anesthesiology department throughout the study period, there is a possibility of under-reporting of occult PPA, which did not catch the attention of the anesthesiology personnel during the perioperative period. Indeed, one patient who clearly suffered PPA was not reported in the QI database. The attending anesthesiologist for that particular case was a new visiting faculty member from another system. Therefore, the QI system might not have been implemented correctly. Second, relying on only one source of data to calculate the incidence could result in potential under-estimation of infrequent anesthetic complications such as PPA. Therefore, it is important to use a secondary data source to ensure complete and fair review of the incidents. In this study, we used the anesthetic QI database as the primary source and the hospital-wide medical archive recording system as the secondary database to report potential cases of PPA. Among the 14 patients who were confirmed with PPA in our study, one additional patient was detected by this secondary survey, as described earlier. The same strategy was used by Olsson et al. (4), who reviewed a computer-based registry at the Karolinska Hospital in Stockholm as the secondary source. Warner et al. (6) used a secondary database of master billing records, medical diagnoses, laboratory tests and radiologic interpretations.
Diagnostic criteria also should be established and specified to define the complication. We applied the same diagnostic criteria that were used by Warner et al. (6). These may have been more restrictive than the criteria that are commonly used by practicing anesthesiologists, who rely on clinical findings such as the presence of gastric contents in the pharynx/larynx/trachea and decrease in O2 saturation (5). After passive regurgitation was noted, the diagnosis of PPA was confirmed either by the identification of nonpulmonary secretions below the vocal cords or by the identification on a postoperative chest radiograph of a new infiltrate that did not exist in the preoperative chest radiograph or on physical examination.
Since this is an anesthesia QI study, we excluded any cases with PPA that occurred during airway management by nonanesthesiology personnel. The main purpose of this study was to review and investigate the etiology and incidence of PPA that occurred while the patients were still under the care of anesthesia personnel. A comparison of the incidence of PPA related to airway manipulation between nonanesthesiologists and anesthesiologists would be extremely interesting. It was not possible for us to collect data on the incidence of in-hospital and out-of-hospital PA related to airway manipulation by nonanesthesiologists, however, because of the absence of a reliable data collecting system like the anesthesiology QI reporting system in nonanesthesiology departments. This could lead to significant under-reporting of this complication. An informal estimate of the number of complications based on potentially incomplete data would be misleading. Since we could not provide data of similar quality to those we used in this study, we decided to refrain from commenting on the incidence of PPA during airway management by nonanesthesiologists.
Our report indicated that the incidence of PPA is very infrequent (1 in 7103 anesthetics), but that PPA is still associated with significant pulmonary morbidity (1 in 16,573) and mortality (1 in 99,441). Fifty percent of the PPA occurred during gastroesophageal procedures. Seventy percentage of the incidents were associated with anesthetic management that could have been improved with more aggressive reflux prophylaxis and/or airway control strategy, which suggests that the majority of PPA can be prevented.
We are grateful for the help of Shekhar Harshad Mehta, BS (MS candidate), with the statistical analysis.