Postoperative pulmonary complications (PPC) often lead to the death of patients after surgery [1]. Of the several entities summarized as CPP, postoperative aspiration pneumonia (PAP) is the most precarious condition, with mortality rates of up to 38.5% after general and visceral surgery [2,3]. Considering an incidence of 0.8% to 1.9%, the overall mortality associated with NAP in surgical patients is expected to be approximately 0.5% [3,4].
NAP leads to increased length of hospital stay, increased need for intensive care, and increased costs [5,6]. Incidence and mortality rates have not improved in recent decades despite advances in anesthesia, intensive care, and surgical techniques [2.4-7].
Known patient-related risk factors for NAP are advanced age, low preoperative arterial oxygen saturation (SpO2), lung infection within one month before surgery, and preoperative anemia [2].
Procedure-related risk factors include type of surgery (thoracic, upper gastrointestinal [GI], abdominal aortic aneurysm repair), duration of surgery, emergency procedures, and type of anesthesia (general, regional) [2, 8].
Current practice guidelines from the American Society of Anesthesiologists (ASA) recommend fasting for solid foods greater than 6-8 hours, and for clear liquids for 2-4 hours before elective operations [9]. Following these recommendations in an elective setting, patients undergoing anesthesia are assumed to be at low risk for perioperative aspiration.
In emergency situations, however, gastric emptying is impaired; therefore, even after prolonged periods of fasting, patients cannot be assumed to have an empty stomach [10].
There is limited evidence on how to avoid NAP. There are few preventive measures that have proven to be effective. Preoperative training of the inspiratory muscles has been shown to be effective in preventing PAP, but not in reducing 30-day mortality [11]. Routine placement of a nasogastric tube after abdominal surgery is associated with an increase in NAP [12].
This exact-matched, weighted case-control study identifies risk factors for NAP in patients undergoing general and visceral surgery.
| Methods |
> Patients
This case-control study was conducted at a tertiary referral center. All patients undergoing general surgery at the Department of General and Visceral Surgery, from January 2012 to December 2018, were eligible for this study.
Inclusion criteria were : age ≥ 18 years and written authorization for future use of medical data in database analysis.
Exclusion criteria were : aspiration pneumonia occurring preoperatively up to 24 hours postoperatively, patients with preexisting pneumonia, patients with regurgitation during anesthesia, and patients undergoing thoracic or vascular surgery.
The cases were all patients with NAP documented in medical records, and definitive data recorded used by cost holders. The diagnosis of NAP was usually confirmed using the following information: recent clinical signs of pneumonia, plus one of the following: fever, presence of typical infiltrates on chest x-rays or CT scans (e.g., infiltrates predominantly affecting the right lower lobe ), or an unexplained increase in C-reactive protein or white blood cell count.
All patients without NAP who met the inclusion criteria were considered controls.
> Data collection and processing
Data were extracted from the electronic database of the surgery department. That database is fed prospectively and semi-automatically, capturing all accessible electronic sources of patient data, throughout the hospital’s database system.
Physical and electronic records of all identified cases were reviewed by 1 investigator (MS) to confirm consistency of diagnosis and complete missing data.
All surgical procedures were subdivided into predefined procedures according to the Swiss catalog of operations (CHOP codes). Annual changes in the catalog were considered in the statistical analysis, grouping the CHOP codes according to the reference catalog issued by the Swiss Federal Health Bureau (BFS reference CHOP codes) [13].
The following data were included: date and duration of surgery, category in relation to the emergency (elective vs emergency surgery; further divided into patients fasting > 6 hours and those fasting ≤ 6 hours), extra-abdominal/open surgery /laparoscopic, number of procedures per surgery, and blood loss.
When appropriate, continuous variables were categorized into clinically relevant groups, such as body mass index (BMI), which was divided into underweight, normal weight including moderately overweight subjects (< 18 kg/m2 vs ≥ 18 – ≤ 35 kg/m2 and > 35 kg/m2). For blood loss, the data were dichotomized into 2 different groups, establishing the cut-off value at the 90th percentile of blood loss, being 100 mL.
All patients were given anesthesia following the institution’s standardized guidelines. Patients who were classified as emergency and patients with a BMI > 35 kg/m2, pre-existing immobility of the cervical spine, facial hair, anatomical abnormalities of the face and neck, or poor dental status, received a rapid induction sequence, or They underwent awake fiberoptic nasotracheal intubation.
The study was approved by the local ethics committee (No. 2020-01506).
> Statistics
Statistical analyzes were performed using the R program (www.r-project.org). Continuous data are expressed as means ± standard deviation. To compare proportions, c2 analysis was used; for continuous variables, the Student t test or the Mann-Whitney U test was used , as appropriate. For multivariable logistic regression analyses, P values were estimated using likelihood ratio tests, and confidence intervals (CI) were estimated with the Wald method.
After the descriptive analysis, the impact of the different procedures was evaluated based on their quantity. For each group, the odds associated with NAP were estimated as univariate odds ratio (OR), and the P value was calculated using an unbiased median estimate (mid- P ).
To adjust for potential collinearity between different surgeries, a multivariate logistic regression model was used, applying the Firth correction to the penalized maximum probability (penalized MP), to avoid separation. The CIs were based on the penalized MPs [14,15].
Surgeries containing 1 or more procedures, with a univariate OR > 2 for NAP, were considered “high-risk” surgeries. The impact of including variables for high-risk procedures, other procedures, patient data, and Charlson score on aspiration risk was evaluated with Firth’s univariate and multivariate logistic regression analyses. Missing data in the control group were imputed using the randomized forest survival method [16].
The effect of age on the odds of NAP was modeled by logistic regression analysis, using B-splines, considering grouping factorial covariates [17].
A weighting and exact agreement analysis was performed using the “ matching ” package of the R program [18,19]. Cases and controls with identical procedures per surgery were grouped into subgroups for further analyses. Univariate and multivariate conditional logistic regression analyzes were performed to evaluate the effect of patients’ baseline data and selected Charlson score variables on the OR for NAP.
Robust sandwich variance estimates with weighted stratification between subgroups were performed to evaluate the effect of overall patient data on NAP, regardless of surgery type.
Therefore, potential collinearity between surgery-related data (e.g., “procedures containing high risk,” “procedures by surgery,” “elective vs. emergency surgery,” and “type of surgery”) was avoided. Exact matching and weighting was used to better explain the strong effect of some OP parts.
The impact of patient and operation data on 90-day mortality was evaluated using Firth univariate and multivariate logistic regression analyses, with an additional backward variable selection process, based on the Akaike information criterion.
| Results |
> Patient characteristics
A total of 23,647 patients underwent 33,088 surgeries. A total of 187 patients did not meet the inclusion criteria.
Therefore, 23,460 patients, who had 32,901 surgeries, consisting of 51,013 surgical procedures classified in CHOP, were included in the analysis. A total of 144 cases of NAP were documented, resulting in an incidence of 0.44% (95% CI: 0.37%-0.52%). Forty patients (27.8%) with NAP died within 90 days.
> Elective and emergency surgery
The 90-day mortality associated with NAP was not associated with emergency surgery ( P = 0.581).
About two-thirds ( n = 88; 61.1%) of patients with NAP underwent emergency surgery and fasted >6 hours.
After the exact weighting and matching procedure, the OR for NAP in those patients was 3.25 (95% CI: 1.46-7.26), compared with patients undergoing elective surgeries ( P < 0.001). In patients with elective surgeries, NAP occurred in 1.14% ( n = 54 of 4723 matched cases), compared with a rate of 4.23% ( n = 88 of 2082 matched cases).
> ASA Score
The ASA score was not a predictor of 90-day mortality associated with NAP ( P = 0.85).
Patients with NAP had a higher ASA score than patients in the control group. However, after exact weighting and matching, despite rates of 5.34% and 12.08% in patients with ASA scores III and IV/V, this increase did not reach statistical significance.
> Underweight and obesity
Patients with NAP had a lower mean BMI (25.8 ± 5.2 kg/m2), than those in the control group (27.2 ± 5.9 kg/m2; P = 0.008).
In patients with a BMI < 18 kg/m2, the OR for NAP was 2.53 (95% CI: 1.04-6.11: P = 0.029), while obese patients with a BMI > 35 kg/m2, did not have a significant increase in the probability for NAP. Patients with NAP who died within 90 days had a significantly lower mean BMI than survivors (24.3 ± 4.9 kg/m2 vs. 26.4 ± 5.2 kg/m2; P = 0.033).
> Age
Patients with NAP who died within 90 days were older than survivors, with a mean age of 77.4 ± 12.2 years, compared with 73.1 ± 11.1 years ( P = 0.017).
Older age was associated with an increased risk of NAP. While in this study, 33.6% of surgeries were performed in patients aged 65 years or older, 82.6% ( n = 119) of all NAP cases were observed in that age group.
After exact weighting and matching, advanced age was associated with increased ORs of 5.23 (95% CI: 2.18-12.51) in the elderly (65 to 80 years), and 13.72 (95 % CI: 4.94-38.09) in octogenarians and above ( P < 0.001).
> Impact of the type of surgery and specific procedures
Open abdominal surgery was associated with the highest rate of NAP, being 1.6% ( n = 102), compared with rates of 0.1% and 0.2% in extra-abdominal and laparoscopic abdominal surgery. respectively. While open surgery was performed in 19.8% of all surgeries analyzed ( n = 6506), 70.8% of all NAP cases were observed after open surgery.
The highest odds for NAP occurred in intra-abdominal surgeries, especially colorectal operations, and upper GI tract operations. Operations associated with a significantly reduced probability for NAP were removal of subcutaneous tumors, sacrococcygeal cysts, thyroidectomies, proctologic procedures, and venous access surgery (ie, ports, dialysis catheter implantation). In all the operations mentioned, there was no record of NAP cases.
The risk of NAP was higher in older patients undergoing high-risk procedures.
In the 1522 bariatric surgeries performed in this study, there was only 1 case of documented NAP, after a sleeve gastric resection. Therefore, bariatric procedures carry one of the lowest risks among visceral procedures.
> Blood loss
In univariate analysis, blood loss > 100 mL was associated with a significantly increased risk of CAP (OR 4.43; 95% CI: 3.08-6.26; P < 0.001). However, in the multivariate analysis, this effect was no longer present.
After exact weighting and matching, there was even a negative association of high blood loss on the risk of CAP (OR 0.50; 95% CI: 0.25-0.99; P = 0.005). Blood loss was not associated with a high 90-day mortality rate, nor was it a predictor of mortality in the backward variable selection analysis.
> Sex
Women had a significantly lower odds for NAP (OR 0.4; 95% CI: 0.23-0.69; P < 0.001). The 90-day mortality associated with NAP was similar between male and female patients.
> Comorbidities
After exact matching, congestive heart failure was found to be the most relevant isolated risk factor for NAP, with an OR of 4.92 (95% CI: 1.42-17.03; P = 0.002). Patients who were classified as having a difficult airway had an OR of 2.64 (95% CI: 1.13-6.18; P = 0.013).
Patients with known postoperative nausea and vomiting, asthma, and thyroid disease had the lowest risk of NAP, with OR of 0.01 (95% CI: 0.00-0.23, P < 0.001), 0.05 (95% CI: 0.00-4.65; P = 0.003), and 0.27 (95% CI: 0.06-1.19; P = 0.007), respectively.
> Alcohol and nicotine
Alcohol consumption was associated with an increased OR for NAP (OR 2.33; 95% CI: 1.15-4.7; P = 0.008). Alcohol use was twice as common in the NAP group as in the control group (16.7% vs 7.3%; P < 0.001). Smoking, defined as sustained nicotine use, was not associated with an elevated OR for NAP.
In descriptive analysis, neither alcohol nor nicotine use was associated with increased 90-day mortality in patients with NAP. Interestingly, in the backward variable selection analysis, sustained smoking was associated with a much smaller risk of 90-day mortality (OR 0.29; 95% CI: 0.09-0.79; P = 0.015).
| Discussion |
This study identified patient- and procedure-related risk factors for NAP. Patient-related risk factors included advanced age, male gender, ASA score ≥ III, congestive heart failure, and cachexia.
Procedure-related factors were open abdominal surgery, prolonged preoperative fasting in emergency surgeries, a duration of surgery greater than 2 hours, and specific high-risk procedures.
Many risk factors for NAP have been established in the literature. Emergency operations, prolonged operative time, and upper abdominal or thoracic procedures have been shown to be procedure-related risk factors for NAP. Patient-related factors include older age, chronic obstructive pulmonary disease, alcohol use, and elevated ASA scores, as well as cigarette use, congestive heart failure, and functional dependence [3,8,20].
In this study, emergency surgery was confirmed as one of the most important risk factors for NAP. Canet et al. found an OR of 2.2 for PPC in emergency surgeries, reflecting the results of a systematic review by Smetana et al., which showed a pooled OR of 2.21 [1,2].
Interestingly, in the present study population, emergency surgeries were not per se associated with higher odds for NAP. In this analysis, emergency surgery was differentiated into 2 distinct groups, including patients who were fasting for more or less than 6 hours preoperatively.
Only in emergency surgeries with preoperative fasting ≥ 6 hours was there an increase in the observed rate of PAP. The NAP rate for emergency surgery without preoperative fasting was similar to that for elective surgery.
One possible explanation is microaspiration during intubation. In an emergency setting, due to physiological changes in gastrointestinal motility and gastric emptying, patients still have significant amounts of gastric contents even after prolonged periods of fasting, with a possible increase in gastric volume over time [21 ].
A second possible explanation for the increased rates of NAP in patients with prolonged preoperative fasting is that impaired gastric emptying manifested preoperatively, and new-onset intestinal paralysis in the early postoperative phase, increase the risk of NAP.
It can be hypothesized that patients who were able to eat until shortly before surgery might have better preoperative bowel function than those with prolonged fasting. However, it can also be assumed that differences in anesthetic techniques contribute to an increased risk for PAP.
The finding that prolonged preoperative fasting , before emergency operations, leads to an increased risk of NAP, adds to the ongoing discussion about the possible detrimental effects of extended preoperative fasting [22,23].
In conclusion, no emergency patient can be considered to have an empty stomach, regardless of their fasting time. Standard anesthesia for any emergency operation should be rapid sequence induction, or awake fiberoptic nasotracheal intubation [24,25].
At the authors’ institution, patients scheduled for non-elective operations, or those considered at high risk for regurgitation, undergo rapid sequence induction, or awake fiberoptic nasotracheal intubation, following institution guidelines.
Regurgitation events during anesthesia were not included in this study because they have a fundamentally different entity from aspiration pneumonia. First described by Mendelsohn in 1946 in 66 obstetric patients, regurgitation and aspiration of large volumes of gastric contents during anesthesia are rare but serious complications, with mortality rates of up to 50% [26,27]. However, chemical pneumonitis associated with regurgitation of large volumes of gastric contents differs so much from NAP that the authors excluded those events.
Consequently, current strategies where emergency surgeries are frequently postponed due to a short preoperative fasting interval probably need to be reviewed. Measures to prevent CAP, such as incentive spirometry, or dedicated prehabilitation programs, should focus on high-risk patients [28,29].
In this study, the rate of NAP increased in underweight patients. It is well known that obese patients are not at high risk for PPCs [30-33].
Data showing the detrimental influence of low BMI on the risk of NAP are scarce. A study by Mullen et al., analyzed more than 2200 patients from the National Surgical Quality Improvement Program Patient Safety in Surgery Study , and found a dramatically increased 30-day mortality in patients with a BMI < 18 kg/m2. However, the number of cases was relatively low, with 55 underweight patients, and there have been no other studies to date that have specifically evaluated this topic [33].
Low weight is associated with malnutrition and sarcopenia, and leads to metabolic stress and catabolism, which lead to altered immune response mechanisms. Due to small numbers in some of the subgroups evaluated, the authors were unable to demonstrate a significant association for any other potential risk factors involved.
Age is consistently reported as the most important patient-related risk factor for NAP. The authors confirmed this finding, especially in patients over 80 years of age, with a 14-fold increased risk in that group. Other studies have confirmed that age is the most important risk factor for NAP [1-3,20].
Prolonged operative time is an independent risk factor for an increased risk of NAP [1,20]. The authors of this work confirmed this finding. However, after the exact matching procedure for CHOP-classified surgical procedures, the impact of duration of surgery no longer reached statistical significance.
In the literature, there are multiple different definitions of a “prolonged” operative time, ranging from > 1 hour to > 3 hours, making it difficult to compare these findings [34-37]. The data from this work show that patient-related factors, such as age, comorbidities, emergency status, and specific surgical procedures, have much greater effects on NAP than operative time alone.
Searching the literature, smoking is associated with an elevated risk for pulmonary complications [37]. The authors also found a high OR for NAP in the univariate and multivariate logistic regression analyses, which did not reach statistical significance. However, when looking at the observed mortality rates in their data, and the associated risk of dying from NAP in the backward variable selection analysis, smoking was much less prevalent in patients who died from NAP than in those who survived. .
Smoking was also associated with a statistically significant lower OR for mortality. This is consistent with the findings of other studies, where active smokers did not have a statistically increased risk of pulmonary complications [2].
A Cochrane review showed a beneficial effect of smoking cessation in relation to overall complications, but no significant effect on cardiopulmonary complications [38].
In the present population study, male patients had a higher risk of developing NAP than females, but did not show increased mortality. This confirms previous findings in the studies by Canet et al., and Nobili et al. In the study by Studer et al., there was a trend toward an increased risk of mortality in male patients with NAP, which did not reach significance, probably due to the small sample size [2,3,20].
Several large cohort studies on CPP and NAP have been conducted with data from veterans, who are predominantly men; therefore, their findings may be difficult to generalize to other populations [8,39,40].
Others have focused on special groups of patients, such as those undergoing liver or esophageal resections. Furthermore, the use of exact weighting and matching allowed us to evaluate risk factors for NAP, regardless of the high impact that specific surgical procedures have on the odds of NAP.
A difficulty in interpreting the results of the present study, compared to the existing literature, is the lack of standardized definitions for NAP and its differentiation from other CPPs [1,2,6,42,43].
A recent publication of a systematic review and consensus definition found that many existing definitions of CPPs are “imprecise or difficult to apply.” Consequently, to date, NAP still lacks a validated standardized definition [44].
The authors of this study believe that the 0.44% NAP rate they observed reflects the clinical reality in many surgical departments, where NAP is treated more as a clinical syndrome than as an exactly defined disease.
For further studies, standardized definitions of NAP are necessary. Future studies should evaluate preventive measures in populations that are at increased risk for NAP.
>Limitations
This study is limited by its retrospective design. Although most of the data in the departmental database are collected semi-automatically on a prospective basis, it was unavoidable to manually complete some of the records, potentially reducing data quality.
The retrospective design and lack of a clear definition of NAP potentially lead to an underestimation of the true incidence of NAP. There were substantial differences in the coding of aspiration pneumonia across a large sample of more than 4200 hospitals in the US, with a rate of 4% to 26%. Hospitals reporting the highest rates had a lower risk-adjusted mortality rate [45].
Although this study is one of the largest single-center population studies of NAP cases to date, the number of cases remains relatively low, limiting statistical power. This leads to a potential limitation of the validation of the statistical model, although attempts were made to add subgroups whenever appropriate. The use of random imputation of the forest missing value introduced a potential source of bias, as the random missing assumption could not be tested.
In conclusion, in this study, risk factors for NAP were analyzed in a large data set, with a higher risk in elderly or underweight patients, and after prolonged preoperative fasting periods. The relevance of these findings is that, by focusing on the populations at highest risk, the burden of postoperative pneumonia can be reduced.
