INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients who require reintubation (extubation failure) have a poor prognosis, with hospital mortality rates exceeding 30 to 40% (1). The need for reintubation is an important independent predictor of mortality even after controlling for the presence of comorbid conditions and generalized severity of illness (4). It remains unclear why extubation failure is associated with such a high mortality rate, though several hypotheses have been suggested. Specifically, the act of reintubation itself may result in life-threatening complications. Alternatively, failed extubation may not directly cause a poor outcome but rather serve as an additional independent marker of severity of illness. Lastly, significant clinical deterioration may occur between the time of extubation and eventual reintubation. If either of the first two hypotheses is correct then outcome should be independent of the cause for extubation failure and the timing of reintubation. Conversely, if the last hypothesis is correct then outcome should worsen with increasing delay between extubation and reintubation. Therefore, we studied the effect of cause for extubation failure and timing of reintubation on hospital mortality.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

We conducted a post hoc analysis of prospectively collected data on 745 consecutive medical ICU patients who were mechanically ventilated for a minimum of 6 h. The study population consisted of the 74 patients (representing 14% of all patients who underwent a trial of extubation) requiring reintubation within 72 h of extubation. Data on the first 37 patients have been previously reported (4). Of the study patients, 68 were extubated after successfully completing a weaning trial (see below). Six others had unplanned extubation during a weaning trial that had been successful up to the point of extubation. Patients were excluded if: reintubation occurred > 72 h after extubation; reintubation was absolutely required within 72 h but the patient refused; reintubation occurred after unplanned extubation not occurring during weaning trials; reintubation occurred solely to replace a defective endotracheal tube.

Extubation and Reintubation Criteria

Patients were extubated after successful completion of a weaning trial. Weaning was accomplished by a reduction in the intermittent mandatory ventilation (IMV) rate and/or pressure support ventilation (PSV) level (for patients ventilated with IMV plus PSV) or pressure support alone (for patients receiving PSV). Patients were extubated only if they tolerated at least 0.5 to 2 h of minimal ventilatory support, defined as PSV  10 cm H₂O, IMV rate of 0, and positive end-expiratory pressure level  5 cm H₂O), and passed the weaning trial by satisfying the following criteria: arterial oxygenation saturation  90% or Pa_O2  60 mm Hg while receiving FI_O2  0.40 to 0.50; need for infrequent suctioning of airway secretions; alert mental status; presence of cough and adequate airway protective mechanisms; stable hemodynamic profile; stable cardiac rhythm; increase in Pa_CO2 < 10 mm Hg; and decrease in pH < 0.10. In general, isolated criteria such as tachypnea, tachycardia, diaphoresis, agitation, or anxiety were considered inadequate for deeming the patient a weaning failure. Criteria for considering reintubation included an increase in Pa_CO2  10 mm Hg and decrease in pH 0.10; Pa_O2 < 60 mm Hg or Sa_O2 < 90% while receiving FI_O2 > 0.50 to 1.0; increased signs of respiratory work (tachypnea, use of accessory respiratory muscles, thoracoabdominal paradox); inability to protect the airway because of upper airway obstruction or excess pulmonary secretions.

Data Collection and Definitions

For all patients, clinical, physiologic, and laboratory data were collected 6 h after the onset of mechanical ventilation, at the onset of weaning trials, and prior to extubation. Etiologies for respiratory failure necessitating mechanical ventilation were assigned and categorized as follows. Pulmonary (COPD, asthma, pneumonia, ARDS, other noncardiogenic pulmonary edema, lung collapse, pulmonary hemorrhage, interstitial lung disease, pulmonary vasculitis, bronchiolitis, aspiration, respiratory muscle weakness, upper airway obstruction, pleural disease, and hypoventilation syndrome). Cardiac (acute myocardial infarction, congestive heart failure, arrhythmia, pericardial disease, and endocarditis). Other (acute renal failure, acute hepatic failure, sepsis, sepsis syndrome, gastrointestinal tract bleeding, noncardiogenic shock, overdose, alcohol withdrawal, encephalopathy, seizures, stroke, intracranial bleeding, or infection). Generalized severity of illness was determined using the APACHE II score measured at the onset of mechanical ventilation and at the onset of weaning. Chronic comorbid conditions were recorded, including the presence of active malignancy, organ transplantation, HIV positivity, cirrhosis, and chronic, dialysis-dependent, renal failure. The presence of acute renal failure (necessitating dialysis) and acute hepatic failure were noted.

The time to reintubation was measured in hours from extubation, rounded off to the nearest hour. Causes for extubation failure were assigned using the following definitions.

Airway etiologies. Upper airway obstruction (stridor with resolution upon reintubation) and aspiration or excess pulmonary secretions (witnessed aspiration or an inability to maintain airway patency because of pulmonary secretions).

Nonairway etiologies. Congestive heart failure (cardiogenic pulmonary edema confirmed by physical examination [presence of abnormal jugular venous distension, rales, third heart sound], chest radiography, or pulmonary artery catheterization data and response to diuresis and/or inotropic agents); respiratory failure (signs of increased work of breathing [tachypnea, use of accessory respiratory muscles, thoracoabdominal paradox, wheezing] and either hypoxemia [Pa_O2 < 60 mm Hg or Sa_O2 < 90% while receiving FI_O2 > 0.5 to 1.0] or hypercapnia [increase in Pa_CO2  10 mm Hg and decrease in pH  0.10] not solely resulting from one of the other defined etiologies); encephalopathy (decreased consciousness leading to hypoventilation); other (reintubation resulting from sepsis without pneumonia or upper gastrointestinal bleeding). When more than one airway or nonairway etiology was responsible, categorization was according to the principal cause determined by the MICU team members blinded to the purpose of the study. When both an airway and a nonairway etiology were identified, the principal cause was considered to be the latter. For example, in a patient failing with both excess airway secretions and pneumonia, the cause for extubation failure was considered to be respiratory failure. Complications related to reintubation were recorded. Pneumonia within 72 h after reintubation was determined by a retrospective review of the chart, flow sheets, chest radiographs, and microbiologic data. Criteria for pneumonia included a new or increasing infiltrate on chest radiograph, fever (> 38.3° C and 1° C increase), leukocytosis (25% increase in white blood cell count and absolute count > 10,000/mm³), purulent tracheal aspirate, and positive tracheal aspirate culture (6).

Statistical Analysis

Continuous variables were analyzed using an independent Student's t test, and categorical values were analyzed using the Chi square test (Pearson's test or Fisher's two-tailed exact test). Reintubated patients surviving the hospitalization (Survivors) were compared with reintubated patients who died during the hospitalization (Nonsurvivors). To assess the independent effect of the cause of extubation failure and the timing of reintubation while controlling for other important factors, we constructed a multiple logistic regression analysis model using hospital death as the dependent variable and the following independent variables: age > 65 yr, APACHE II score at onset of mechanical ventilation and weaning, presence of a chronic comorbid condition, acute renal failure requiring dialysis, time to reintubation (as either a continuous variable or dichotomized using a 12-h cutoff) and cause for extubation failure (dichotomized by airway versus nonairway etiology). All statistical analysis was performed using SPSS Version 6.1 (SPSS, Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Demographics

The 74 patients consisted of 47 men (63.5%) and 27 women (36.5%) with a mean age of 64 ± 2 yr. Causes for respiratory failure were categorized as pulmonary (74.4%), cardiac (40.5%), and other (28.4%), with many patients having multiple etiologies. The APACHE II score at onset of mechanical ventilation was 16 ± 1 and it was 12 ± 1 at the onset of weaning. The duration of mechanical ventilation prior to extubation was 139 ± 19 hr. After reintubation, patients spent 305 ± 32 additional hours receiving mechanical ventilation. Complications related to reintubation occurred in 21 (28%) patients (Table 1). Fifteen (20%) patients had a tracheostomy performed after reintubation. Thirty-one of the 74 patients (41.9%) died during their hospitalization. The most common causes of death were sepsis and multiple organ failure (Table 2).

Causes of Extubation Failure

Causes for extubation failure are listed in Table 3. The most common reasons for reintubation were respiratory failure, congestive heart failure, aspiration or excess pulmonary secretions, and upper airway obstruction. Mortality was significantly higher for patients failing because of nonairway etiologies compared with those who failed principally because of an airway problem (27/51, 52.9% versus 4/23, 17.4%; p < 0.01). When further comparing patients on the basis of airway versus nonairway causes for extubation failure no differences were found in age, APACHE II scores at onset of mechanical ventilation and weaning, presence of comorbid condition, need for acute dialysis, acute hepatic failure, or cause for mechanical ventilation.

Timing of Reintubation

The median time to reintubation was 21 h (interquartile range, 9 to 46 h), with the distribution of times shown in Table 4. Patients with upper airway obstruction or aspiration/excess pulmonary secretions tended to be reintubated earlier than those reintubated for other reasons (21 ± 4 versus 31 ± 3 h, p = 0.07). The former group was also more likely to be reintubated within 12 h of extubation (11/23 versus 14/51; p = 0.11). There were no differences in age, APACHE II scores, acute renal or hepatic failure, or the prevalence of chronic comorbid conditions when patients with earlier reintubation were compared with those with later reintubation. As in previous studies, the development of pneumonia after reintubation was unrelated to time to reintubation (pneumonia, 32 ± 6 h versus no pneumonia, 27 ± 3 h, p = NS) (7). The mortality rate increased with increasing duration of time from extubation to reintubation and was significantly lower for patients reintubated within either 12 h (24% versus 51%, p < 0.05) or 24 h (30.2% versus 58.1%, p < 0.05) of extubation when compared with those reintubated after longer periods of time. One of 11 (9.1%) failing for airway reasons and requiring reintubation within 12 h died compared with three of 12 (25%) reintubated later than 12 h after extubation. For nonairway etiologies, time to reintubation had an impact on survival for patients with respiratory failure (mortality: 1/6,  12 h versus 11/15, > 12 h; p < 0.05).

Comparison of Survivors and Nonsurvivors

Patients compared on the basis of hospital survival are shown in Table 5. Nonsurvivors had higher APACHE II scores at onset of mechanical ventilation and weaning and were more likely to have a chronic comorbid condition. There were no differences in the incidence of emergent reintubation or complications related to reintubation when nonsurvivors were compared with survivors. There was no difference in hospital length of stay after reintubation (survivors, 26 ± 3 d versus nonsurvivors, 20 ± 3 d, p = NS). Although 24/31 (77%) nonsurvivors had a do not resuscitate order written, this occurred an average of 11 ± 2 d after reintubation. Seven of 31 (23%) nonsurvivors had a tracheostomy performed sometime after reintubation.

Survivors were more likely to require reintubation for an airway problem, whereas nonsurvivors were more likely to require reintubation for respiratory failure. Survivors were reintubated an average of 15 h earlier than nonsurvivors and were more likely to be reintubated within either 12 or 24 h after extubation. Using multiple logistic regression analysis, age > 65 yr (adjusted odds ratio, 4.2; 95% CI, 2.1 to 8.4), APACHE II score at onset of mechanical ventilation (adjusted odds ratio, 1.14; 95% CI, 1.07 to 1.22), the presence of a chronic comorbid condition (adjusted odds ratio, 5.7; 95% CI, 2.8 to 11.7), nonairway cause for extubation failure (adjusted odds ratio, 11.2; 95% CI, 5.1 to 24.4), and time to reintubation of > 12 h (adjusted odds ratio, 6.0; 95% CI 2.9 to 12.3) were all independently associated with hospital mortality. Increased time to reintubation continued to be an independent predictor of hospital mortality when it was entered as a continuous rather than as a dichotomized variable. To examine airway etiologies more closely we did a post hoc analysis in which upper airway obstruction and aspiration/excess pulmonary secretions were each considered as a separate category. For both univariate and multivariate analyses, both etiologies continued to be independently associated with outcome when compared with patients reintubated for nonairway causes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The principal findings of this study are that both cause for extubation failure and time to reintubation are important independent predictors of hospital mortality for patients failing extubation. Previous studies have demonstrated that patients requiring reintubation are at increased risk for death, with mortality rates ranging as high as 43% compared with less than 12% in patients successfully extubated (1). The explanation for this excess mortality remains unclear, although a number of hypotheses have been put forth.

One hypothesis is that the increased mortality associated with extubation failure may simply reflect a sicker cohort of patients. This appears unlikely because extubation failure exerts a strong independent effect on mortality even after controlling for generalized severity of illness at weaning onset, the presence of chronic comorbid conditions, age, and the need for acute dialysis (4, 5). Nevertheless, it is possible that extubation failure is an additional marker of severity of illness, detecting a tendency not otherwise identifiable using conventional methods.

There are several reasons why extubation failure or the need for reintubation may independently correlate with an adverse outcome. Reintubation is an invasive procedure and may be associated with numerous life-threatening complications such as cardiac arrest, esophageal intubation, right mainstem intubation, gastric aspiration, cardiac arrhythmias, aspiration, pneumonia, and pneumothorax (5, 8). Although complications are frequent during emergency intubation in hospitalized patients, mortality occurs principally in those hemodynamically unstable prior to intubation (8). A recent study noted that reintubation increased the risk of pneumonia resulting in increased mortality (7). In contrast, using a retrospective design, Daley and colleagues (9) found no increase in mortality for the more than 50% of reintubated patients suffering a proximate complication. Esteban and colleagues (5, 10), using a prospective design, noted complication rates of 15 and 18% with reintubation, with no further increase in mortality when compared with patients not experiencing a complication. Similarly, we found that patients suffering a complication related to reintubation had a hospital mortality rate comparable to patients not experiencing a complication. If the act of reintubation is the primary explanation, mortality should not vary for different causes for extubation failure. In contrast, Daley and colleagues (9) found that trauma patients with extubation failure, the majority of whom were reintubated because of stridor, inability to clear secretions, or declining sensorium, were not at increased risk for death (two of 24 or 8%) when compared with patients successfully extubated. Similarly, Esteban and colleagues (5) found a lower mortality for patients reintubated for upper airway obstruction (one of 14 or 7%) compared with other etiologies. Our study extends these observations by finding that the lower mortality for patients reintubated for airway reasons is maintained after controlling for other important factors that impact on outcome. In addition, mortality increased with duration of time between extubation and reintubation, another variable that should not impact on the nature of reintubation. Taken together the above findings strongly suggest that the act of reintubation itself is an insufficient explanation for the high mortality rate seen with extubation failure.

Another leading hypothesis is that clinical deterioration occurs during the time of unsupported ventilation, allowing for the development of new organ dysfunction that leads to increased mortality. This may explain in part the relationship between cause for extubation failure and outcome. Reintubation for airway causes serves to rapidly correct respiratory dysfunction, whereas the organ dysfunction related to other causes of extubation failure may not be readily reversible with the reinstitution of ventilatory support. This clinical deterioration hypothesis is further supported by finding increasing mortality with increasing duration of time between extubation and reintubation, independent of etiology of extubation failure. For example, mortality for respiratory failure increased fourfold when reintubation occurred more than 12 h after extubation. Another possible explanation is that reintubation may have been purposely delayed in some as the physician sought to avoid reinstituting "invasive" life support in patients with a poor prognosis. This is unlikely because no differences in severity of illness, age, or comorbidity were identified when comparing patients reintubated early with those reintubated later.

Clinical Implications

Unlike weaning outcome, the capacity to accurately predict extubation outcome is lacking. Standard respiratory functional parameters determined prior to weaning and during the weaning trial do not accurately predict patients who will fail extubation (3, 11). This is likely related in part to the distinct pathophysiologic mechanisms for extubation failure that differ from those seen with weaning failure. A sizeable fraction of extubation failures relate to airway problems that are difficult to recognize when the endotracheal tube is still in place. Our findings and those of other investigators suggest that, from a mortality perspective, it may not be as important to accurately identify such patients because their outcome is similar to those successfully extubated (5, 9). Conversely, mortality for patients failing from other etiologies is exceedingly high, reinforcing the importance of developing instruments for identifying at-risk patients. The association between time to reintubation and eventual outcome suggests that clinical deterioration might be reduced by the early reinstitution of mechanical ventilatory support, especially for patients with respiratory failure. Studies designed to examine both the evolution of respiratory and other physiologic functional parameters and the earliest onset of new organ dysfunction after extubation may be necessary to improve identification of patients at risk for death from extubation failure. Although predictors for early reintubation are desirable, accuracy would need to be high to avoid needless invasive reintubation. Conversely, test accuracy may be less important if noninvasive ventilation proves to be an effective therapy for extubation failure.

In conclusion, we found that etiology of extubation failure and time to reintubation are both independent predictors of outcome for medical ICU patients failing extubation. The disproportionately higher mortality rate for patients requiring reintubation for nonairway problems such as congestive heart failure and respiratory failure indicates that increased efforts should be directed at identifying patients at risk for these etiologies. The association of time to reintubation and outcome suggests that early identification of patients after extubation and the early reinstitution of ventilatory support has the potential to reduce the increased mortality associated with extubation failure from respiratory failure.