The American Society of Anesthesiologists’ (ASA) classification of physical health (from I to V) is universally applied for assessing preoperative health of patients requiring any surgical, therapeutic, or diagnostic procedure. Although ASA > II is associated with increased risk of postoperative morbidity and mortality, large interobserver variability and poor specificity preclude accurate estimation for individual patient risk [2].

The Charlson Comorbidity Index (CCI), composed of 19 weighted medical diagnoses, is a valid predictor of 1-year mortality in medical patient population, score greater than 5 being associated with 1-year mortality greater than 50 % [3]. In patients undergoing noncardiac surgery, CCI score ≥ 3 was associated with a 16-fold increase of death at one year [4]. Likewise, among lung cancer patients undergoing curative resection, a CCI score ≥3 was associated with a tenfold greater incidence of major complications [5].

The National Surgical Quality Improvement Program (NSQIP) was jointly established by the American College of Surgeons (ACS) and the Department of Veterans Affairs (VA) to compare risk-adjusted 30-day mortality between different VA hospitals [6]. Based on 21 variables, the NSQIP is currently available with web-based surgical risk calculator. Derived from the NSQIP database, the Surgical Mortality Probability Model (S-MPM) includes three relevant components – the ASA physical status, surgical risk class, and emergency status – to predict all-cause postoperative mortality at 30 days [7] (Table 2.1).

More specific for thoracic surgery, the Cardiopulmonary Risk Index (CPRI) was developed in 1993 and consists in a combination of a cardiac risk index (congestive heart failure, myocardial infarction during the previous 6 months, greater than five premature ventricular contractions, arrhythmias, age >70 years, important valvular aortic stenosis, poor general medical condition) and a pulmonary risk index (BMI ≥27 kg/m², smoking within 8 weeks of surgery, productive cough within 5 days of surgery, diffuse wheezing or rhonchi within 5 days of surgery, FEV1/FVC <70 %, PaCO₂ >45 mmHg). Later on, Ferguson et al. have validated a simple scoring system (EVAD) that utilizes pulmonary function test data (forced expiratory volume in one second [FEV1], diffusion capacity of the lung for carbon monoxide [DLCO]) and patient age to predict the likelihood of complications after major lung resection [8].

More recently, the Thoracoscore derived from the French national thoracic database EPITHOR has incorporated eight independent risk factors (age, sex, ASA physical status, performance status, dyspnea, priority of surgery, extent of resection, carcinoma) to predict in-hospital mortality [9].

The Revised Cardiac Risk Index (RCRI) was developed for prediction of major cardiac complications in non-emergent, noncardiac surgery [10] (Table 2.2). Major cardiac complications include myocardial infarction, pulmonary edema, ventricular fibrillation or primary cardiac arrest, and complete heart block. The RCRI is composed of six variables of approximately equal prognostic importance: high-risk surgery (including intrathoracic surgery), history of ischemic heart disease, history of congestive cardiac failure, history of cerebrovascular disease, insulin therapy for diabetes, and preoperative serum creatinine >177 μmol/L. A RCRI ≥3 is associated with a risk of major postoperative cardiac complications for more than 11 % of patients and may be considered as a cutoff to delineate high-risk patients. Derived from the original RCRI, a thoracic risk score (ThRCRI) for lung resections was established [11] (Table 2.3). The predictive power of both of these scores in patients undergoing lung resections is controversial.

The Myocardial Infarction and Cardiac Arrest (MICA) risk calculator [12] was developed with the intent to improve predictive power for major cardiac adverse events as compared to RCRI. The model was based on analysis of the National Surgical Quality Improvement Program (NSQIP) database with more than 200,000 patients. Five predictors of perioperative risk of MICA at 30 days were identified: type of surgery, age, functional dependency, creatinine >133 umol/L, and ASA class. The MICA risk calculator resulted in a more accurate cardiac risk prediction than RCRI, although no data is available specifically for thoracic surgical patients. The MICA risk calculator is available on the web.

Postoperative pulmonary complications (PPCs) include respiratory failure, reintubation within 48 h, weaning failure, pneumonia, atelectasis, bronchospasm, exacerbation of chronic obstructive pulmonary disease (COPD), pneumothorax, pleural effusion, and various forms of upper airway obstruction. They are a major cause of postoperative morbidity and mortality, possibly accounting for a higher mortality than cardiovascular complications.

The ARISCAT study established a risk score for the development of PPCs in a mixed cohort of surgical patients [13]. Seven independent risk factors emerged: low preoperative SpO2, preoperative anemia, age, lung infection in the previous month, duration of surgery >2 h, upper abdominal or intrathoracic surgery, and emergent surgery (Table 2.4). Both the patient-related and the procedure-related risk factors contributed roughly 50 % to total risk. The score was prospectively and externally validated across many European countries, with a satisfactory predictive power especially for Western European countries [14].

The degree of dyspnea is correlated with the risk of postoperative mortality [15]. Standardized symptom-limited stair climbing is a simple cost-effective test to objectively determine cardiorespiratory reserve and may have superior predictive ability than traditional spirometry values [1]. The test involves climbing three flights of stairs without interruption, equivalent to 12 m ascent that corresponds to metabolic equivalents (METs) greater than 4. The inability to climb more than 12 m warrants further lung functional testing. A patient able to climb at least 22 m (5–6 flights of stairs) has a low risk of postoperative complication, regardless of lung function test results [16].

FEV1 is a reliable predictor of perioperative complications in thoracic surgery for patients with FEV1 <70 % [17]. According to the guidelines of European Respiratory Society (ERS) and the European Society of Thoracic Surgery (ESTS) on fitness for lung resection in cancer patients, a predicted postoperative (ppo)-FEV1 <30 % separates patients into normal and high-risk groups. It should be remembered that the calculated ppo-FEV1 may overestimate the actual FEV1 on the first postoperative day by about 30 % and that measured FEV1 on postoperative day one may provide more accurate prediction of cardiopulmonary risk [18–20]. On the other hand, patients with a moderate to severe obstructive pulmonary syndrome may have improved respiratory dynamics after lung resection [21]. The ppo-diffusion capacity of the lung for carbon monoxide (DLCO) is another powerful predictor of perioperative complications. According to the ERS/ESTS guidelines, a ppo-DLCO <30 % delineates a high surgical risk [1].

Peak VO₂ allows further refinement of perioperative risk prediction. Patients with values of peak VO₂ >20 mL/kg/min qualify for resection up to pneumonectomy, whereas values <10 mL/kg/min indicate a high risk for any type of lung resection [22]. A value of ppo-peak VO₂ <10 ml/kg/min is associated with a mortality rate exceeding 50 % [23].

Given age-related decline in organ function and impairment in physiological reserve, aging is considered a major risk factor for perioperative morbidity and mortality. Sarcopenia affects not only limb skeletal muscles but also respiratory muscles and those controlling the upper airways. Accordingly, obstructive sleep apnea and occult aspiration occur more frequently particularly in the context of underlying neurological disorders (e.g., previous stroke, dementia, Parkinson disease) [24]. The risk of postoperative hypoxia and hypercapnia is increased because of altered chemosensitivity, respiratory muscle weakness, and increased pulmonary shunting. Impaired thermogenesis favors the occurrence of wound infection, bleeding, and cardiac ischemia events, resulting in prolonged postoperative recovery [25]. The risk of postoperative cognitive disorder (POCD) is increased, especially with benzodiazepine premedication [26].

Frailty is a composite measure of geriatric conditions. It includes measures of cognition, strength, energy, nutrition, physical mobility, mobility, and mood. Patient assessment for frailty may be a valuable aid in determination of operability and planning of postoperative care. A multidimensional frailty score was elaborated for prediction of 1-year postoperative mortality [27]. It represents an adaptation of the comprehensive geriatric assessment (CGA) and comprises a total of nine items, with a maximal score of 15. The authors used a cutoff of a score of 5, to distinguish between a high and a low risk of postoperative mortality (mortality >10 %). Although superior to the ASA score for prediction of 1-year mortality, its computation is complicated and time-consuming and must be performed by a medical consultant familiar with the score.

The literature on the risk of thoracic surgery primarily focuses on lung resections, particularly in the context of cancer surgery. Broadly, the more extensive the lung resection, the higher is the risk of developing postoperative complications.

The highest risk of postoperative morbidity and mortality is associated with extended pneumonectomy [28]. A study based on the Society for Thoracic Surgeons (STS) General Thoracic Surgery Database (GTSD) examined major morbidity and mortality after pneumonectomy in 1267 patients. The risk factors independently associated with major adverse outcomes were age >65 years, congestive heart failure, FEV1 <60 %, underlying benign lung disease, and extended pneumonectomy. Overall mortality was 5.6 % and the incidence of major morbidity was 30.4 %. A study based on data of the French national database for thoracic surgery (EPITHOR) on 4498 patients with lung cancer reports an overall mortality of 7.8 % for pneumonectomy, with risk factors for mortality identified as age >65 years, ASA physiologic status ≥3, underweight, right-sided pneumonectomy, and extended pneumonectomy [29].

A large study based on the STS GTSD, with 18,800 lung cancer resections performed at 111 participating centers revealed an overall perioperative mortality of 2.2 %. Independent predictors of mortality were pneumonectomy, bilobectomy, ASA rating, functional status, renal dysfunction, induction chemoradiation therapy, steroids, age, urgent procedures, male gender, FEV1, and body mass index [30]. According to an analysis based on data of the American National Cancer Database (NCDB) on almost 120,000 patients, 30-day mortality of lung resections for non-small cell lung carcinoma (NSCLC) was 3.4 % overall, with a mortality of 8.5 % for pneumonectomies, 4 % for extended lobectomies and bilobectomies, and 2.6 % for lobectomies and bilobectomies. Mortality for wedge resections was 4.2 % and slightly higher than for lobectomies, which may be explained by a higher rate of tumor recurrences, and a lower functional preoperative status, indicating a more conservative surgical approach.

Overall, a right-sided lung resection carries a higher risk of complications than a left-sided resection owing to greater propensity to bronchopleural fistula formation, a greater increase in right ventricular afterload, and potential alteration in cardiac sympatho-vagal balance [29, 31].

Thoracic surgical interventions, which require one-lung ventilation (OLV) and a thoracotomy, can be considered high-risk procedures. Similar to lung resections, they expose patients to the risk of cardiovascular complications as well as atelectasis, pneumonia, and ventilator-induced lung injuries (VILI) leading to acute lung injury (ALI) or acute respiratory distress syndrome (ARDS).

For patients undergoing esophagectomies, a nomogram has been developed to predict the occurrence and severity of postoperative complications [32]. Independent risk factors are increasing age, a history of cerebrovascular accident (CVA) or transient ischemic accident (TIA), a history of myocardial infarction, a reduced forced expiratory volume in one second (FEV1), electrocardiographic (ECG) changes, and extensive surgery. The nomogram was validated and proved useful for risk prediction in high-volume hospitals [33].

Lung or pleural biopsies and simple bullectomy with or without pleurodesis under video-assisted thoracic surgery (VATS) are usually short-lasting and minor procedures that require short-term admission in a PACU for monitoring anesthesia emergence, titration of analgesic intravenous regimen, and detection of residual air leakage, lung re-expansion, and atelectasis. Mediastinoscopies can generally be monitored in PACU, with special attention to the risk of occult postoperative hemorrhage.

Uni- or bilateral lung volume reduction surgeries in patients with severe emphysema are considered high-risk procedures given preexisting severe airflow limitations and major impairments in gas exchange. These patients require cautious titration of analgesics (preferably epidural or paravertebral block) and are preferably admitted in ICU or HDU given the risk of life-threatening deterioration in pulmonary function (e.g., bronchopleural fistula, opiate-induced hypercapnic acidosis).

Little evidence supports the use of a muscle-sparing thoracotomy as opposed to a posterolateral thoracotomy, but incision length may be proportionally related to post-thoracotomy complications [34]. Given limited tissue trauma and consequent reduced neuroendocrine and inflammatory responses, VATS is associated with lower rates of overall perioperative mortality, morbidity (e.g., pneumonia and atrial arrhythmia), as well as length of stay [31]. In the absence of other major risk factors for postoperative complications, patients with a VATS lung resection do not require neuraxial analgesic techniques and are commonly managed in PACU for vital monitoring and anesthesia emergence.

Operative mortality may be lower if board-certified thoracic surgeons perform a minimal case load of procedures [35]. Differences in postoperative mortality rates between hospitals may also be explained by a different quality of postoperative patient management [36]. As a consequence, local experience should be included in the process of postoperative patient triage.

Surgery performed on an emergent basis has repeatedly been associated with worse postoperative outcomes. Various pre- and postoperative scores integrate this factor into risk stratification.

Finally, the occurrence of major intraoperative complications may require a higher level of postoperative monitoring and treatment, than initially planned. Myocardial ischemia, hemodynamically significant arrhythmias, refractory hypotension or hypoxemia, bronchial aspiration, and major bleeding are considered major complications that justify admission in HDU or ICU (Table 2.6).