Abstract

Objectives

Postoperative reintubation after planned extubation (RAP) following general anesthesia is a major anesthetic morbidity. A previous study on RAP identified the various risk factors for RAP, including chronic obstructive pulmonary disease (COPD), pneumonia, systemic inflammatory response syndrome (SIRS), and airway surgery. However, the prognosis and predictive risk index of RAP were not investigated.

Methods

Data on surgical patients who were reintubated after planned extubation at the end of surgery between January 1, 2005 and December 31, 2009 were retrospectively sorted out from the quality assurance database of the Department of Anesthesiology, Chang Gung Memorial Hospital. Risk factors and prognosis of RAP cases were compared with the control group (successful planned extubation) using descriptive statistics and logistic regression. The RAP predictive risk index was developed from multivariate logistic regression and the predictive accuracy was evaluated by goodness-of-fit test.

Results

Of the 227,876 patients who were subjected to endotracheal intubation for general anesthesia, 130 (0.06%) sustained postoperative RAP. The control group consisted of 390 patients who were randomly selected from those who underwent endotracheal intubation without RAP. A total of 30 variables, including demographic, operative, anesthetic data, and prognosis were analyzed. We found that significant risk factors for RAP included COPD (odds ratio: 4.30), pneumonia (odds ratio: 6.60), ascites (odds ratio: 4.86), SIRS (odds ratio: 7.52), hypothermia (body temperature <35°C; odds ratio: 2.45), rocuronium as muscle relaxant (odds ratio: 1.90), inexperienced anesthetic service (odds ratio: 3.44), and airway surgery (odds ratio: 4.34). An RAP predictive risk index was developed and the predictive accuracy was confirmed by goodness-of-fit test as excellent discrimination (c statistic: 0.873). RAP significantly increased postoperative stay in hospital (odds ratio: 2.46) and intensive care unit, as well as tracheostomy and mortality (odds ratio: 58.52).

Conclusion

The RAP predictive risk index included higher American Society of Anesthesiologists classification, conscious disturbance, COPD, pneumonia, SIRS, room air SpO₂ <95%, hypothermia, airway surgery, and head and neck surgeries. The RAP predictive risk index provides us an opportunity to take preventive measures including renewal of risk-reduction protocols for high-risk patients.

Keywords

airway extubation; hypothermia; nomograms; prognosis; quality assurance; health care;

1. Introduction

Extubation after the completion of surgery is generally considered to be safe and practicable. However, extubation failure leading to reintubation in the operating room (OR) or postanesthesia care unit (PACU) happens at times. Grave consequences of failed extubation after operation are pneumonia, prolonged stay in the hospital and care in the intensive care unit (ICU), tracheostomy, and even mortality as end result.1, 2 The most weaning strategy to evaluate critically ill patients with mechanical ventilation in the ICU is the use of spontaneous breathing trial (SBT). However, usual weaning indexes are poor predictors for extubation outcome in the overall simple-to-wean and ICU population.³ The reason might be that SBT could only evaluate a patient’s respiratory status, and is incapable of evaluating the patient’s mental status and ability to clear secretion after extubation. Therefore, to extubate patients in the OR, standardized weaning protocols and expertise of anesthesiologists in charge are applied, such as ability to obey orders, head lift >5 seconds, adequate cough. Inappropriate decision making and lack of supervision of experienced anesthesiologists were often criticized as contributing factors to failed extubation after general anesthesia.

Reintubation after planned extubation (RAP) in the OR may be due to hypoxia, respiratory muscle fatigue, hypercapnia, residual muscular blocking, upper airway obstruction, or unstable hemodynamics.⁴ Sometimes, urgent reoperation due to emergent surgical indication after planned extubation also accounts for RAP. The need for reintubation is a marker of increased severity of clinical illness and independent predictor of death.⁵ Delay of reintubation might increase the likelihood of death by prolonged hypoxia, myocardial ischemia, respiratory muscle fatigue, and aspiration pneumonitis. A previous study on RAP by Ting and co-workers identified the various risk factors for RAP, including chronic obstructive pulmonary disease (COPD), pneumonia, systemic inflammatory response syndrome (SIRS), and airway surgery.⁶ However, the prognosis and predictive risk index of RAP were not investigated. The present retrospective, case-controlled study further expanded the case numbers to amplify statistic power and establish an RAP predictive risk index model.

2. Methods

After obtaining approval from the Review Board of the Chang Gung Memorial Hospital (CGMH), we retrospectively analyzed all cases who underwent perioperative reintubation between January 1, 2005 and December 31, 2009, the data of which were documented in the quality assurance (QA) anesthesiology database of CGMH. All major complications were reviewed and analyzed by the anesthesiology department’s QA group each month. After monthly analysis, details were presented in a morbidity and mortality conference, where all personnel made certain the outcomes of quality assessment to initiate a positive impact on clinical practice.

In the present study, RAP was defined as reintubation after removal of the endotracheal tube for the initial intubation for general anesthesia in the OR up until the entire stay in the PACU. Criteria for appropriateness of tracheal extubation in the OR and timing of transferring to the PACU were at the discretion of the anesthesiologist in charge. Cisatracurium or rocuronium was elective choice of muscle relaxant for induction and maintenance. Neostigmine in proper dose was routinely administered before extubation to reverse muscle paralysis. The patients who were successfully extubated served as control. Preoperative variables included age, gender, body mass index (BMI), the American Society of Anesthesiologists (ASA) classification, preoperative oxygen saturation while breathing room air (room air SpO₂), and specific comorbidities such as central nervous system (CNS) disorder, conscious disturbance (Glasgow Coma Scale score <8), congestive heart failure (CHF) (2001 American College of Cardiology/American Heart Association guidelines for heart failure stages C and D), history of angina or myocardial ischemia, COPD (forced expiratory volume in 1 second/forced vital capacity <70% in pulmonary function test), preoperative pneumonia, other lung disease (pleural effusion, pulmonary edema, etc.), chronic renal failure (glomerular filtration rate <15 mL/minute/1.73 m²), ascites, SIRS, obesity (BMI ≧30 kg/m²). The criteria for SIRS were as follows: body temperature >38°C or <36°C, heart rate >90/minute, respiratory rate >20/minute, or arterial CO₂ <32 mmHg, and white blood cells >12,000 mm³ or <4000 mm³ or >10% band form. Anesthetic data included choice of muscle relaxant (cisatracurium or rocuronium, excluded discordant initial loading, and maintained choices), last opioids and last muscle relaxants given within 30 minutes before extubation, fluid balance, hypothermia (body temperature <35°C during extubation), and seniority of the anesthesiologist responsible. Fluid overload was defined as the periodic total input (crystalloids, synthetic colloids, and blood products) minus total output (basic maintenance, insensible loss, and urine and blood loss) larger than 1000 mL from the time of endotracheal intubation to the end of surgery. Inexperienced anesthetic service was defined as anesthetic practice less than 6 years (4-year residency plus 2-year fellowship) on the part of responsible anesthesiologist. Operative data included type of surgery [orthopedic, head and neck (oral cavity, thyroid, etc.), airway (oropharynx, hypopharynx to larynx), neurosurgery (brain, cervical spine, etc.), abdominal (liver, stomach, etc.), others], and surgical duration. Postoperative outcomes consist of duration of postoperative stay in hospital, postoperative ICU stay, tracheostomy, and mortality happened in that postoperative period.

2.1. Statistical analysis

All data were statistically analyzed using SPSS version 17.0 (SPSS Inc., Troy, NY). We used means and standard deviation to show the differences in the distribution of preoperative, operative, and anesthetic-related variables between RAP cases and controls for continuous variables, and frequency tables for categorical variables. Odds ratios and 95% confidence intervals (CIs) were assessed using univariate logistic regression for both continuous variables and categorical variables.

To estimate the simultaneous effects of multiple variables on reintubation, we used multivariate logistic regression for these significant variables in the univariate logistic regression. RAP predictive risk nomogram was then developed with significant variables in the multivariate logistic regression. By amplifying and rounding off the beta coefficient of those variables in the multivariate logistic regression to the nearest integer (e.g., a beta value of 0.12 would add 1 to the final risk score), the final RAP risk index was developed. Adding those total points for each patient, the specific predictive RAP risk scores were calculated. We divided the patients into three discrete RAP risk stratification (low, medium, and high) with the mean value of predictive RAP risk scores of the RAP group and the control group. The predictive RAP risk model was then compared with the observed data using the Hosmer–Lemeshow goodness-of-fit test. Discrimination of logistic regression was determined by the c-statistic test.

3. Results

Of the 227,876 patients who underwent general anesthesia with endotracheal intubation between January 1, 2005 and December 31, 2009, 130 (0.06%) sustained RAP. The control group consisted of 390 patients (randomly selected), who were successfully extubated without RAP. Among the RAP cases, 102 (78.5%) had the reintubation performed in the OR, whereas 28 (21.5%) had it in the PACU. The reasons for RAP included uneasy breathing (53 patients, 40.8%), airway obstruction (40 patients, 30.8%), surgical complications (27 patients, 20.8%), and unstable hemodynamics (10 patients, 7.7%). RAP due to vomiting postextubation with subsequent aspiration pneumonia (eight patients, 6.2%) was classified under airway obstruction. Most RAP cases had their reintubation performed within 30 minutes after planned extubation in spite of rescue measures such as corticosteroids to treat laryngeal edema, nasal airway to keep airway patency, or incremental doses of reversal agents as well as correction of metabolic acidosis. The mean duration of PACU stay for those patients who sustained reintubation in the PACU was 241.5 ± 12.7 minutes. Overall, the characteristic factors that are responsible for RAP could be categorically classified into three groups, namely, respiratory, cardiovascular, and surgical, among which postoperative respiratory problems stood out as the most common problem (71.6%).

The univariate analysis revealed that patients who required RAP were older, of male gender, higher in the ASA physical classification, obese (WHO classification defines BMI ≧30 kg/m² as obesity), and those with lower preoperative room air saturation (SpO₂ <95%) (Table 1). Higher risk comorbidities for RAP included neurologic diseases (CNS disorder, conscious disturbance), heart diseases (history of myocardial infarction, CHF), pulmonary diseases (COPD/asthma, pneumonia, pleural effusion, pulmonary edema), chronic renal failure, ascites, and SIRS. The choice of muscle relaxant such as rocuronium and hypothermia during extubation could also predispose patients to RAP. Inexperienced anesthetic service by green anesthesiologist was found to have significant correlation with RAP compared with senior anesthesiologists. With regard to surgical type, RAP most often occurred after airway surgery and neurosurgery in the univariate study and as far as opioids or muscle relaxants are concerned, the statistic analysis showed no significant difference between the two groups in whom the opioids or relaxants were administered less than 30 minutes before extubation (Table 2).

Table 1. Patient's demographics in the RAP and control groups.

Download full-size image

Table 2. Comorbidities and perioperative factors in the RAP and control groups.

Download full-size image

After adjusting multiple confounding factors for multivariate logistic regression (Table 3), the results showed that perioperative comorbidities with conscious disturbance (odds ratio: 3.51), COPD (odds ratio: 4.30), pneumonia (odds ratio: 6.60), ascites (odds ratio: 4.86), and SIRS (odds ratio: 7.52) were more likely to render RAP. Other perioperative factors such as ASA class III (odds ratio: 3.99), preoperative room air SpO₂ <95% (odds ratio: 4.35), hypothermia (odds ratio: 2.45), muscle relaxant such as rocuronium (odds ratio: 1.90), and experience of anesthesiologist <6 years (odds ratio: 3.44) also significantly increased the risk of RAP. With regard to surgical type, airway surgery (odds ratio: 4.34) remained significant, but neurosurgery abdicated its superior significance in favor of head and neck surgeries (odds ratio: 3.28).

Table 3. Multivariate logistic regression and RAP risk scorea.

Download full-size image

These significant variables (p < 0.05) in multivariate regression except seniority of the anesthesiologist responsible were used to build the RAP predictive risk nomogram (Fig. 1). Human factor concerning the seniority of anesthesiologist responsible was excluded from the RAP predictive risk nomogram because it could be corrected without issues by intervention of the consulting senior anesthesiologist, and thus it lacked a predictive value. Because the surgical types must be put into consideration altogether, neurosurgery and abdominal surgery were also included in the predictive risk nomogram in spite of insignificant results in multivariate study. The beta coefficient of each variable was amplified as clinical RAP risk score, and then the RAP risk points for each patient were calculated. The mean RAP risk score for the RAP group and the control group was 42 (RAP risk: 57%) and 21 (RAP risk: 14%), respectively. The RAP risk stratification was divided into three discrete ranges: low (RAP scores <21, probability <14%), medium (RAP scores 21–42, probability 14–57%), and high (RAP scores >42, probability >57%); with the clinical suggestion they were represented by regular extubation, extubation with caution, and delayed extubation, respectively (Table 4). The RAP predictive risk nomogram had excellent predictive accuracy (i.e., discrimination) with c statistic (i.e., area under the receiver operating characteristic curve) of 0.873 (95% CI: 0.835–0.912, p < 0.001) (Fig. 2). Results of the Hosmer–Lemeshow Chi-square test showed that the predictive model was not different from our observed data (p = 0.546). Therefore, the goodness-of-fit test showed that the RAP predictive risk model was in good agreement with the observed data.

Download full-size image

Fig. 1. Nomogram for RAP predictive risk index. ASA = American Society of Anesthesiologist; COPD = chronic obstructive pulmonary disease; RAP = reintubation after planned extubation; SIRS = systemic inflammatory response syndrome.

Table 4. Clinical RAP predictive risk index stratificationa.

Download full-size image

Download full-size image

Fig. 2. ROC curve for RAP predictive risk stratification. CI = confidence interval; RAP = reintubation after planned extubation; ROC curve = receiver operating characteristic curve.

For prognosis of RAP, the univariate regression revealed that RAP significantly increased postoperative stay in hospital (odds ratio: 2.46), ICU stay, incidence of tracheostomy (odds ratio: 88.02), and mortality (odds ratio: 58.52) (Table 5). The only case in the control group sustaining grave prognosis after successful extubation was a 76-year-old patient with subdural hemorrhage who had sudden intracranial hemorrhage 3 days later after ventriculostomy. He was then admitted to the ICU, and underwent tracheostomy 2 weeks later, but expired 1 month postoperatively due to severe sepsis.

Table 5. Postoperative prognosis between the RAP and control groups.

Download full-size image

4. Discussion

In this retrospective, QA data, and chart-reviewed case-controlled study, we investigated RAP risk factors and prognosis in surgical patients. The reintubation incidence in this study was 0.06%, far less than the reintubation rates (0.17–22.8%) in the literature we reviewed.1, 3, 7 The possible explanations were that we included surgical patients receiving general anesthesia with planned extubation other than critically ill patients in the ICU. Besides, we defined the act of postoperative reintubation be carried out exclusively in the OR and PACU (no longer than 6 hours after extubation), whereas the other reports defined them to be done within 24–48 hours after the initial extubation.

The present study found that the incidence and risk factors of RAP were similar to those in the previous study by Ting and co-workers⁶ who evaluated cases between 2005 and 2007. The previously identified significant risk factors for RAP such as COPD, pneumonia, ascites, SIRS, airway surgery, and head–neck surgery were still sound and valid as revealed in our present study results. The new variables such as obesity, muscle relaxant such as rocuronium, hypothermia, and inexperienced anesthetic service showed significances in the univariate analysis; hypothermia, rocuronium, and inexperienced anesthetic service were still significant risk factors as revealed by multivariate analysis. The final RAP predictive risk nomogram had included significant risk factors such as higher ASA classification, conscious disturbance, COPD, pneumonia, ascites, SIRS, room air SpO₂ <95%, hypothermia, rocuronium, airway surgery, and head–neck surgery. Human factors such as inexperience on the part of anesthesiologist responsible were found to be a significant contributing factor to RAP though it lacks predictive value. The predictive accuracy of the RAP predictive risk model is excellent in goodness-of-fit test.

4.1. Obesity

Obesity with BMI >30 kg/m² accounted for higher possibility of reintubation in the univariate study (odds ratio: 2.19; p = 0.005) but was not significant in the multivariate study (odds ratio: 1.66, p = 0.206). However, if morbid obesity (BMI >35 kg/m²) was put into comparison, it is significant in both analyses (odds ratio: 3.12, p = 0.02 in multivariate analysis). Obese people have more comorbidities such as obstructive sleep apnea syndrome, cardiovascular, respiratory, and CNS diseases. Reduced functional residual capacity, lower maximum voluntary ventilation, ease of upper airway obstruction, and prolonged emergence from general anesthesia might explain the higher possibility for RAP. The detrimental effects of obesity on respiration is even worse in morbidly obese patients than in common obese ones, thus it may explain why “obesity” was not a significant risk factor for RAP but “morbid obesity” was in multivariate analysis. In the retrospective review, Bamgbade and co-workers⁸ found that the morbidly obese patients had a higher prevalence of tracheal reintubation (p = 0.009) in comparison with nonobese patients.

4.2. Rocuronium

The choice of rocuronium as a muscle relaxant could result in higher RAP risk in both the univariate and multivariate analyses. We often choose rocuronium to facilitate rapid-sequence intubation or to avoid asthmatic attack. Clinically, patients with inadequate fasting or bronchial asthma have higher possibility of postextubation respiratory complications, which might result in selection bias in the analysis of rocuronium. However, rocuronium does render higher possibility to produce postoperative residual curarization, which impairs respiratory function in the postoperative period. Residual paralysis decreases upper esophageal tone, coordination of swallowing, and hypoxic ventilation drive. Longer acting duration and more variable metabolism of rocuronium in comparison with cisatracurium might partially explain the residual paralysis in elderly people, in patients with impaired liver function or renal insufficiency. In the randomized-controlled study by Maybauer and co-workers,⁹ the longer duration of action and variability in duration of action were apparent with rocuronium in comparison with cisatracurium. These pharmacodynamic differences might translate into a higher incidence of RAP, but experienced clinicians could compensate this shortcoming by enfeebling the neuromuscular blockade earlier. Adequate reversal of muscle blockade with sugammadex or neostigmine¹⁰ is absolutely required for successful extubation after the use of rocuronium. Train-of-four (TOF)-based evaluation of residual paralysis is advisable to determine the appropriate timing (suggested TOF >0.9) for reversal of rocuronium in elderly people and difficult-to-weaning patients.¹¹

4.3. Hypothermia during extubation

Hypothermia with body temperature lower than 35°C during extubation also accounted for higher possibility of reintubation in the univariate and multivariate analyses. General anesthesia impairs thermoregulation by decreasing the threshold (triggering core temperature) for vasoconstriction and shivering by 2–3°C. Hypothermia with postoperative shivering increases metabolic rate and risk of myocardial ischemia and ventricular arrhythmia.¹² Besides, intraoperative hypothermia increased the incidence of residual paralysis from muscle relaxants.¹³ If the body temperature falls below 35°C, the tone of adductor pollicis would decrease by 10–16% per degree Celsius.¹⁴ Maintaining perioperative normothermia with fluid warming or forced air-warming system can effectively reduce the elongated extubation time and postoperative shivering,¹⁵ and thus it might be able to reduce the probability of reintubation in the wake of the planned extubation.

4.4. Anesthetic service

In the Thai Anesthesia Incident Monitoring Study, the postanesthetic reintubation investigation conducted by Chinachoti and co-workers⁴ stated that human factors such as lack of experience, lack of vigilance, inappropriate decision making were the contributing factors that predisposed patients to reintubation. Most human factors were considered preventable and rapidly correctable. In our study, the inexperienced anesthesiologist responsible was found to be a significant contributing factor in univariate and multivariate analyses. This issue does not mean that green anesthesiologists should not perform extubation, but they should be more vigilant in caring for patients with high RAP risk scores and compliant to the supervision of senior anesthesiologists for better decision making before extubation. By sharing experiences to improve decision making, we might be able to achieve better perioperative patient care.

The sufficient communication between surgeons and anesthesiologists is also an important contributing factor to prevent extubation failure after general anesthesia. During our retrospective analysis, we found that the incidence of reintubation after airway surgery had been slightly decreasing in the recent years (13% in 2005 vs. 7% in 2009). The rational explanation might be the consensus of opinion of anesthesiologists and surgeons in delaying extubation in patients with deep neck infection, airway swelling, or undergoing airway reconstruction surgery. Delay of extubation in these patients might decrease the incidence of reintubation, and thus improves the postoperative prognosis.

4.5. RAP predictive risk index

Predictors of extubation failure in ICU patients have been reported in the literature, such as reduction of central venous saturation >4.5% during SBT,¹⁶ low serum anion gap <5.2 mEq/L and low PaO₂/FiO₂ <178,¹⁷ lower dead space fraction (V_d/V_t),¹⁸ Glascow Coma Scale score <8,¹⁹ insufficient self-confidence during SBT,²⁰ or chaotic spontaneous breathing pattern.²¹ Our RAP predictive risk model is based on the preoperative characteristics with excellent discrimination (c statistic = 0.873), and therefore, it is suitable for clinical prediction. For patients with medium RAP risk scores (RAP scores: 21–42; probability 14–57%), following strict standardized weaning protocols or delay of extubation in the PACU or ICU is suggested. As far as high-risk patients with high RAP risk scores are concerned (RAP scores >42; probability >57%), delay of extubation in the PACU or ICU could reduce the need for reintubation, the possibly higher hospital cost, incidence of tracheostomy and above all, reduction of mortality. For example, if a 50-year-old alcoholic male, with cirrhosis-related ascites, room air SpO₂ of 93%, fever with SIRS and ASA class 3 physical status, was scheduled to receive emergent fasciotomy for necrotizing fasciitis of the left arm, his RAP risk score would be 70 (18 for ASA class 3, 15 for ascites, 19 for SIRS, 18 for room air SpO₂ <95%), which suggests high risk for RAP (95% in the RAP predictive risk nomogram), and therefore delayed extubation in the ICU is usually suggested to prevent further morbidity associated with RAP.

Risk factors identified in this study are consistent with those reported in previous studies that had examined extubation failure or postoperative respiratory failure. In the study of postoperative respiratory failure, Johnson and co-workers²² stated that patients with higher ASA classification, undergoing emergent operations or complex operations, sustain sepsis and elevation of serum creatinine were more prone to see postoperative respiratory failure (3%). The cohort study by Ramachandran and co-workers²³ illustrated that independent predictors of unanticipated early postoperative tracheal intubation within 3 days postoperatively (incidence rate: 0.83–0.9%) after non-emergent, noncardiac surgery were smoking, COPD, dyspnea, acute CHF, impaired liver function, cancer, BMI <18.5 kg/m² or >40 kg/m², high-risk surgery, and sepsis. Unanticipated postoperative intubation was found to be an independent predictor of 30-day mortality with an adjusted odds ratio of 9.2. Comparing those results with ours in the present study, it can be inferred that immediate postoperative reintubation in the OR and the PACU after planned extubation is not exclusively the result of postoperative respiratory failure and has lower incidence but much higher mortality. Therefore, it requires more collaborate teamwork between anesthesiologists and surgeons to take preventive measures and comprehensive postoperative care for patients with high RAP risk to improve the clinical outcome.

4.6. Prognosis

Our study revealed that RAP was associated with increased complications, medical expenditure, and higher mortality. Besides, difficult airway might predispose patients to be subjected to inferior mask ventilation, prolonged hypoxia, failure of reintubation, which ultimately results in higher rate of emergent tracheostomy and mortality.

4.7. Study limitations

The present study has ample sample size and significant results in statistics. The RAP predictive risk index rendered here is clinically usable and has excellent predictive accuracy. However, it is a retrospective single-center study in Taiwan; therefore, further multicenter prospective application of the RAP predictive risk index model to assess the clinical events is required.

5. Conclusion

In summary, a meticulous extubation strategy should be thoroughly planned before administration of general anesthesia, especially for patients with high-risk RAP. Through this retrospective, case-controlled study, risk factors, prognosis, and predictive risk index of RAP have been demonstrated. We have found that patients with higher ASA classification, conscious disturbance, COPD, pneumonia, ascites, SIRS, room air SpO₂ <95%, hypothermia during extubation, receiving airway surgery are indicators of a higher incidence of RAP following general anesthesia. Inexperienced anesthesiologists should ask for supervision by senior anesthesiologists while extubating patients with high RAP risk scores. The RAP predictive risk index gives us an opportunity to re-evaluate the extubation strategy to achieve better perioperative patient management.

Financial support

No financial support.