Effect of Failed Extubation on the Outcome of Mechanical Ventilation

Effect of Failed Extubation on the Outcome of Mechanical Ventilation* Scott K. Epstein, MD, FCCP; Ronald L. Ciubotaru, MD; and John B. Wong , MD

Objective: To examine medical outcomes associated with reintubation for extubation failure after discontinuation of mechanical ventilation. Design: Prospective cohort study of consecutive intubated medical ICU patients who underwent a trial of extubation at a tertiary-care teaching hospital. The failed extubation group consisted of all patients reintubated within 72 h or within 7 days (if continuous ICU care had been required) of extubation. All others were considered to be successfully extubated. Study end points included hospital death vs survival, the number of days spent in the ICU and in the hospital after the onset of mechanical ventilation, the likelihood of requiring '2:.7 or '2:.14 days of ICU care after extubation, and the need for transfer to either a long-term care or rehabilitation facility among the survivors. Results: Of289 intubated patients, 247 (85%) were successfully extubated, and 42 (15%) required reintubation for failed extubation (time to reintubation 1.5±0.2 days). Reintubation for extubation failure resulted in 12 additional days of mechanical ventilation. When compared with successfully extubated patients, reintubated patients were more likely to die in the hospital (43% vs 12%; p<0.0001), spend more time in the ICU (21.2±2.8 days vs 4.5±0.6 days; p<0.001) and in the hospital (30.5±3.3 days vs 16.3±1.2 days; p<0.001) after extubation, and require transfer to a long-term care or rehabilitation facility (38% vs 21 %; p<0.05). Using multiple logistic regression, extubation failure was an independent predictor for death and the need for transfer to a long-term care facility. Compared with those successfully extubated, patients who failed extubation were seven times (p<0.0001) more likely to die, 31 times (p<0.0001) more likely to spend <::.14 days in the ICU after extubation, and six times (p<0.001) more likely to need transfer to a long-term care or rehabilitation facility if they survived. Conclusion: After adjusting for severity of illness and comorbid conditions, extubation failure had a significant independent association with increased risk for death, prolonged ICU stay, and transfer to a long-term care or rehabilitation facility. Extubation failure may serve as an additional independent marker of severity of illness. Alternatively, poor outcomes may be etiologically related to extubation failure. If the latter proves to be the case, identifying patients at risk for poor outcomes from extubation failure and instituting alternative care practices may reduce mortality, duration of ICU stay, and need for transfer to a long-term care facility. (CHEST 1997; 112:186-92) Key words: extubation; ICU outcome; mechanical ventilation; weaning Abbreviations: APACHE=acute physiology and chronic health evaluation; C I=confidence interval; Flo 2 =fraction of inspired oxygen; fNT=spontan eous rapid shallow breathing index; PPV=positive predictive value

The incidence of reintubation after discontinuation of mechanical ventilation ranges from 3 to 19%, but the eventual outcome for these patients has not been examined comprehensively. 1- 12 Patients *From the Pulmonary and Critical Care Division, Division of Clinical Decision Making, Department of Medicine, New England Medical Center, Tufts University School of Medicine, Boston. Manuscript received August 2, 1996; revision accepted D ecember 11. Reprint requests: Dr. Epstein, Pulmonary and Critical Care Division, New England Medical Center, Box 369, 750 Washington St, Boston, MA 02111 186

requiring reintubation have an associated crude mortality rate ranging from 10% in a bum-trauma unit,! to 40% in a general surgical ICU,1 35% in a mixed surgical-medical ICU,2 and 33% in a small series (52 patients ) of exclusively medical ICU patients. 3 The increased mortality rate associated with reintubation may reflect patients with greater underlying severity of illness or the presence of significant comorbid conditions. Alternatively, reintubation may independently contribute to an adverse outcome if (1) significant clinical deterioration occurs during the period of unsupported ventilation, (2) the act of Clinical Investigations in Critical Care

reintubation itself results in important complications, or (3) the total duration of mechanical ventilation is markedly prolonged. Therefore, to examine eventual outcomes for patients with extubation failure and to explore the effect of severity of illness measures and reintubation on those outcomes, we prospectively evaluated 289 extubated patients.

MATERIALS AND METHODS

Study Population As part of an ongoing prospective study of mechanical ventilation, 404 consecutive patients admitted to the medical ICU service of the New England Medical Center, intubated and mechanically ventilated for a minimum of 6 h , were studied during the 24-month period from May 1993 to May 1995. Of these, 115 (28%) were never extubated (104 died while receiving mechanical ventilation and ll were transferred to a long-te rm care facility while receiving mechanical ventilation). Among the remaining 289 patients (72%) who underwent a trial of extubation, successful extubation occurred in 247 (85%) patients, and 42 (15%) patients were classified as xe tubation failures. W e specifically sought to identify a fa iled extubation group whose need for reintubation would be more likely to be etiologically related (directly or indirectly) to the just completed course of mechanical ventilation. We wanted to reduce the likelihood that this group would include patients whose reintubation represented an unrelated and th erefore unpredictable e vent. Accordingly, the failed extubation group consisted of all 39 p atients who failed within 72 h of extubation and the three patients who required continuous ICU care (4, 5, and 6 days after extubation) and were reintubated within 7 days of extubation (mean time between extubation and reintubation, 1.5 ± 0.2 days) . To avoid biasing the impact of early failed extubation on outcome, two patients, who died within 72 h of extubation directly as a r esult of refu sing reintubation, we re excluded from further analysis. Of the 247 successfully extubated p atients, 232 were never r eintubated, while 15 had late reintubations occurring 4 days (one patient), 5 days (one patient ), and 7 or more days (13 patients ) after extubation (mean time between e xtubation and reintubation, 12±2 days). All 15 patients had been transferred in stable condition from the ICU prior to reintubation. Sixteen episodes of unplanned extubation, not occurring during weaning trials, followed b y immediate reintubation we re not considered to represent extubation failures. Only one of these patients died after subsequently failing a planned extubation. In four patients undergoing weaning trials, reintubation was not immediate but did occur within 72 h of unplanned extubation (two died). Because these patients were thought to be capable of successful spontaneous ventilation at the time of unplanned extubation, they were included as xe tubation f ailures.

Extubation and Reintubation Criteria During this observational, noninterventional study, weaning and extubation decisions, including onset, timing, pace, and mode of weaning, were made bythe medical ICU team which always included a board-certified critical care or pulmonary physician. Decisions were based on clinical ju dgment supplemented b y application of clinical, vital sign, arterial blood gas, and phys iologic crite ria, including results from w eaning and extubation indexes. General criteria for extubation included the follow-

ing: significant improvement from the etiology of r espiratory failure; appropriate respirat01y drive as indicated b yspontaneous e support was decreased; respirations as the level of vntilatory return of adequate airway protective mechanisms; need for infrequent suctioning of airway secretions; alert mental status; stable he modynamic profile; stable cardiac rhythm ; arterial oxygen s aturation 2:90% or Pa0 2 2:60 on fraction of inspired oxygen ( Fio 2 )~ 0. 40 to 0.50; positive end-expiratory pre s s ure~ 5 em H 2 0 ; and negative inspiratory force <- 20 em H 2 0. In addition, 174 of287 (61%) patients had weaning/extubation indexes determined (at the onset of the weaning trial l eading to extubation) with commonly used threshold values generally applied to weaning/extubation decision making: vital capacity (> 10 mUkg), minute ventilation (~ 15 Umin ), and the spontaneous rapid shallow breathing index (fNT < lOO, measured according to the method of Yang and Tobin 13 ) . Weaning was usually by rapid reduction in intermittent mechanical v entilation rate or pressure support level with patients extubated after successfully tolerating a minimum of0.5 to 2 h onminimal ventilatory support (pressure support ventilation ~ 10 em H 2 0 ; intermittent mechanical ventilation 0; positive end-expiratory pressure, ~ 5 em H 20 ). Customary criteria for passing a weaning trial were similar to those for assessing extubation but also included an increase in PaC0 2 < 10 mm Hg and decrease in pH < 0.10. In general, isolated criteria such as tachypnea, tachycardia, diaphoresis, agitation, or anxiety were considered to be insufficient for deeming the patient a weaning failure. Criteria for ocnsidering reintubation were similar to those used to evaluate weaning trials but also included inability to protect the airway or manage secretions and Pa0 2 < 60 mm Hg or arte rial oxygen saturation < 90% on Fio 2 > 0.50 to 1.0.

Data Collection and Definiti ons For each patient, clinical, physiologic, and laboratory data we re recorded 6 h fter a the onset of mechanical v entilation, a t the onset of weaning trials, and prior to extubation. One or more causes for respiratory failure necessitating mechanical ventilation was assigned for each patient, including the followin g: (1) pulmonary- COPD, asthma, pneumonia, ARDS, other noncardiogenic pulmonary edema, lobar collapse, pulmonary hemorrhage, interstitial lung disease, pulmonary vasculitis, bronchiolitis, pulmonary hemorrhage, aspiration, respiratory muscle weakness, upper airway obstruction, pleural disease, and hypoventilation syndrome; (2) cardiac-acute myocardial infarction, congestive heart failure, arrhythmia, pericardia! disease, and endocarditis; and (3) other-acute renal failure, acute hepatic failure, sepsis, sepsis syndrome, GI tract bleeding, noncardiogenic shock, ove rdose, alcohol withdrawal, e ncephalopathy, seizures, stroke, intracranial bleeding, or infection. To assess the severity of illness, the APACHE II (acute physiology and chroni c health evaluation) score was measured at the 6-h mark of mechanical ventilation (APACHE II, mv) and at the onset of weaning (APACHE II, wean). Factors predicted t o beassociated with outcome from mechani cal ventilation were recorded: comorbid conditions (presence of active malignancy, bone marrow transplantation, HIV positive, cirrhosis, and chronic, dialysisdependent renal failure), acute renalfailure requiring dialysis, or acute hepatic failure while receiving mechanical ventilation.

Outcomes The end points for this study included the following: (1) hospital death vs survival; (2) the number of days spent in the ICU and in the hospital after the onset of mechanical ventilation; (3) the likelihood of requiring 2: 7 or 2: 14 days of ICU care after CHEST / 112 / 1 / JULY, 1997

187

Table !-Comparison of Patients Who Failed Extubation With Patients Successfully Extubated*

Patient, No. Male, No. Age, yr APACHE II, rnv APACHE II, wean Acute hepatic failure Acute renal failure , dialys is Corrnorbiditi es Active cancer Bone marrow transplant Cirrh osis Chronic renal failure HIV positive Corrnorbidities :2:2 Etiology respiratory failure Pulmonary Cardiac Other Preextubation , d1

Successful

Failed Extubation

All

No Reintubation

Late Reintubation

p Value Failed vs Successful

40 26 (65 ) 64±2 16±1 12±1 0 (0) 3 (8) 14 (35) 10 (25) 2 (5) 2 (5) 1 (3) 1 (3) 2 (5)

247 144 (58) 55±1 15±1 10±1 7 (3) 17 (7) 86 (35) 37 (15) 8 (3) 29 (12) 6 (2) 13 (5) 12 (5)

232 134 (58) 55 ± 1 14±1 10±1 7 (3) 12 (5) 78 (34) 33 (14) 5 (2) 26 (ll ) 6 (3) 10 (4) 7 (3)

15 10 (67) 55±6 20±2 1 15 ± 1 1 0 (0) 5 (33) 8 (53) 4 (27) 3 (20) 3 (20) 0 (0) 3 (20) 5 (33) 1

NS < 0.01 NS < 0.05 NS NS NS NS NS NS NS NS NS

28 (70) 17 (43) 8 (20) 5.1±1.0

155 (63) 64 (26) 116 (47) 5.4±0.5

144 (62) 61 (26) 107 (46) 5.4±0.5

ll (73) 3 (20) 9 (60) 5.2±1.3

NS < 0.05 < 0.01 NS

*Number in parentheses is percentage of patients. NS=not sign ificant; MV=mechanical ventilation. 1 p< 0.05 failed extubation vs late reintubation. 1Days receiving mechanical ventilation prior to extubation.

extubation; and (4) the need for transfer t o either a long-term care or rehabilitation facility among the survivors.

Data Analysis Continuous variables were analyzed using an independent Student's t test and categorical values were analyzed using the l test (Pearsons test or Fisher's two-tailed Exact Test). To determine the independent effect of failed extubation on outcome, several multiple logistic regression models (forward stepwise) were constructed using hospital death , :2:7 or :2: 14 day ICU stay after extubation , and need for long-term care as th e dependent variables, respectively, and all univariately significant variables (p < 0.1 ) as the independent variables. All statistical analysis was performed using statistical software (SPSS version 6.1; Chicago).

RESULTS

Failed vs Successful Extubation When compared b y univariate analysis with those successfully extubated (Table 1), patients who failed extubation were older, had higher APACHE II scores at the onset of weaning trials, and were more likely to have a cardiac but less likely to have a noncardiopulmonary etiology for their respiratory failure. The groups were otherwise comparable in clinical features (Table 1) and preextubation arterial blood gas values and weaning and extubation index measurements made prior to extubation (Table 2). Extubation failure rates were similar among patients ventilated for <24 h (ll/55) compared with those ventilated for 2::24 h (29/232; p> 0.2) prior to extubation. 188

Table 2-Comparison of Preextubation Arterial Blood Gas Values and Extubation Indexes in Patients Who Failed Extubation (n=40) With Patients Successfully Extubated (n=247) * Failed Successful p Extubation Extubation 1 Value No. (%) of patients with blood gases Flo 2 Pa02> mm Hg PaO,!Fio 2 PaC0 2 , mm Hg pH a No. (%) of patients with indexes Tidal volume, VT, mL Respiratory rate, f, breaths/min f!VT, breaths/min/L No. (%), f!VT :2:100 breaths/ min/L Vital capacity (VC ), mL No. (%) VC < 10 mUkg Minute ventilation (VE ), Um in No. (%) VE > 15 U min IF, em H 2 0 No. (%) NIF> -20 em H 2 0

40 (100)

216 (88)

0.36 ±.01 104±5 289±15 40 ± 1 7.43±.01 27 (68 ) 460 ±29 25±2 65 ±8 4 (15)

0.37± .01 105±2 302±7 40± 1 7.43±.01 147 (60) 472± 13 25±1 61±3 14 (10)

NS NS NS NS NS NS NS NS NS NS

1078±89 4 (15) 10.9±0.6 4 (15) 50±2 0 (0)

1285±50 17 (12) 11.2±0.4 19 (13) 49± 1 0 (0)

NS NS NS NS NS NS

NS

*NIF=negative inspiratory force; NS=not significant. All patients had Flo 2 :50.50 at the time of extubation. 1 Seven late reintubation patients had blood gas values and 11 had weaning/extubation indexes measured. Because no differences were fo und between these and th e successfully extubated patients without late rein tubation, the results w ere combined. Clinical Investigations in Critical Care

Table 3-Hospital Outcome of Patients Who Failed Compared With Those Successfully Extubated Successful

No. of patients Died 1 ICU length of stay, d 1 Hospital length of stay, days 1 ICU days after initial extubation 27 ICU days after initial extubation 1 214 ICU days after initial extubation 1 Hospital days after initial extubation Long-term care facility 1 Discharged to home 1

Failed Extubation

All

No H.eintubation

Late Reintubation

p Value*

40 17 (43) 26.3::'::3.1 35.7::'::3.4 21.2::'::2.8 33 (83)

247 29 (12) 9.9::'::0.8 21.7::'::1.4 4.5::'::0.6 32 (13)

232 21 (9) 8.5::'::0.6 19.8::':: 1.3 3.1::'::0.3 20 (9)

15 8 (53) 32.3::'::7.2 51.5::+::7.3 27.1::+::7.0 12 (80)

<0.00001 <0.001 <0.001 <0.001 <0.00001

22 (55)

13 (5)

7 (3)

8 (53)

<0.00001

30.5::'::3.3 15 (38) 8 (20)

16.3::':: 1.2 53 (21) 165 (67)

14.4::'::1.1 48 (21) 163 (70)

46.3::'::7.0 5 (33) 2 (13)

<0.001 <0.0001 <0.00001

*p value is for failed vs all successful patients. 1 Number of patients (percentage of group). Other values are mean::'::SEM. 1Length of stay after initial intubation.

Of the 247 successfully extubated patients, 15 required reintubation late in the hospital course. When compared with failed extubation patients, late reintubation patients had significantly higher APACHE II scores at onset of ventilation and weaning and were more likely to have multiple comorbidities present (late reintubation, 33% vs failed

extubation, 5%; p<0.01) and to require dialysis for acute renal failure (33% vs 8%; p<0.05) (Table 1). Hospital Outcome

Failed extubation was associated with a markedly increased hospital mortality rate and transfer rate to

Table 4-Multiple Logistic Regression Analyses* Variable

Beta

p Value

OR (95% CI)

1.99 1.65 0.13 1.32 or <14 days 1

0.0000 0.0001 0.0014 0.033

7.3 (4.6-11.7) 5.2 (3.4-8.0) 1.14 (l.l-1.19) 1 3.8 (2.0-7.0)

3.53 3.43

0.0000 0.0000

34.1 (21.1-55.2) 30.7 (18.9-49.9)

0.079 0.131

0.0098 0.0009

1.08 ( 1.05-1.12) 11 1.14 (l.l-1.19)

1.33

0.0158

1.76 0.39 0.003

0.0005 0.0004 0.0037

5.8 (3.5-9.6) 1.04 (1.03-1.05)11 l. 003 ( l. 002-l. 004 )II

0.058

0.048

1.06 (1.03-1.09)11

1

Death vs Survival Failed extubation Co morbidity APACHE II, wean Acute renal failure, dialysis ICU days after extubation, 27 or 214 days vs <7 days Failed extubation (27 d) (214 d) APACHE II, mv (27 d) (214 d) Acute renal failure, dialysis (27 d) Discharge to long-term care/rehabilitation vs home§ Failed extubation Age Ventilator days before extubation APACHE II, mv

3.9 (2.2-6.6)

*OR=adjusted odds ratio; mv=mechanical ventilation. 1 Variables not in model: APACHE II, mv, and pulmonary etiology for respiratory failure. 1Variable not in model: APACHE, wean; age; ventilator days prior to extubation; pulmonary etiology for respiratory failure; comorbidities (27 day model). APACHE, wean; age; cardiac etiology for respiratory failure; comorbidities (214 day model). §Variables not in model: APACHE II, wean. 1 0R for each unit of change in APACHE II score, years of age, days receiving mechanical ventilatory assistance prior to extubation. Comorbidities is presence of 21 of the following: active malignancy, bone mruTow transplantation, cirrhosis, chronic renal failure, HIV positive. CHEST /112 I 1 1 JULY, 1997

189

a long-term care or rehabilitation facility which far exceeded that seen among those successfully extubated (Table 3). After their initial extubation, patients with extubation failure spent substantially more time in the ICU and in the hospital than those successfully extubated (Table 3). Failed extubation was associated with a high likelihood of requiring > 7 or 14 days of ICU care after extubation (Table 3). Patients who failed extubation spent an additional 12.2:±:2.3 days receiving mechanical ventilation. Multivariate Analysis: Using multiple logistic regression analysis, failed extubation, higher APACHE II scores at the onset of weaning, the presence of one or more comorbid conditions, and the need for dialysis were all independently correlated with death (Table 4). Based on the adjusted odds ratios, patients who failed extubation were approximately seven times more likely to die in the hospital when compared with all those successfully extubated (adjusted odds ratio, 7.3; 95% confidence interval [CI], 4.6 to 11.7; p<0.0001). Failed extubation and APACHE II score at onset of ventilation were independently associated with a postextubation duration ofiCU stay of, or exceeding, either 7 or 14 days (Table 4). The need for dialysis for acute renal failure was also an independent factor in the :::::.7 day model. Patients with extubation failure were 31 times more likely to require 2 additional weeks of ICU care after extubation than those successfully extubated (adjusted odds ratio, 30.7; 95% CI , 18.9 to 49.9; p < 0.0001 ). Similarly, older age, higher APACHE II scores at onset of ventilation, longer duration of mechanical ventilation prior to extubation, and extubation failure were independently associated with the need for transfer to a long-term care or rehabilitation facility (Table 4). Patients who failed extubation, and survived the hospitalization, were nearly six times more likely to need transfer to long-term care or rehabilitation than those successfully extubated (adjusted odds ratio, 5.8; 95% CI, 3.5 to 9.6; p < 0.001 ). Restricting the analysis to patients ventilated for :::::.24 h prior to extubation did not change the results of the multiple regression analyses except that acute renal failure requiring dialysis no longer achieved statistical significance at the p<0.05 level in the death vs survival model only.

DISCUSS ION

As with previous studies of surgical or mixed ICU patients, we found that medical ICU patients who fail extubation have a significantly elevated hospital mortality rate. After using a logistic regression model to control for comorbidities and physiologic markers 190

associated with increased risk for death , extubation failure still had a strong independent effect on mortality. Multiple regression analysis also showed that extubation failure was a strong independent predictor for a prolonged ICU stay and eventual transfer to a long-term care or rehabilitation facility among hospital survivors . Of patients who failed extubation, but survived hospitalization, 65% required transfer to a long-term care or rehabilitation facility, while only 35% were discharged home. There are several reasons why extubation failure or the need for reintubation might independently correlate with an adverse outcome. First, the act of reintubation itself has been suggested to be harmful by increasing the risk for pneumonia.2 Torres et al, 2 using case control methods in a mixed medicalsurgical ICU, found that compared with control subjects, patients requiring reintubation were significantly more likely to develop nosocomial pneumonia (47% vs 10%). Importantly, approximately one half of the crude mortality rate among their reintubated patients (35%) was attributable to the development of pneumonia. Because ventilator-associated pneumonia cannot be accurately diagnosed by clinical criteria and because protected specimen technology was not used in our medical ICU, we were unable to assess the association among reintubation, pneumonia, and consequent mortality. In addition, the reintubated p atients analyzed herein differ from those studied b y Torres et al, 2 in which only 30% of reintubated patients were extubation failures by our definition. Another possibility is that significant clinical deterioration may occur between the time of extubation and eventual reintubation and contribute to poor outcome by causing new or worsening organ dysfunction. Along these lines, the outcome for patients who failed extubation was similar to that of patients who had late reintubation, even though the latter group had a higher anticipated mmtality based on higher APACHE scores and the presence of comorbid conditions . Once reintubated, patients with failed extubation continued to receive mechanical ventilation for approximately 2 additional weeks. Although no relationship between duration of mechanical ventilation and mortality has been convincingly reported, presumably additional time receiving mechanical ventilatory assistance incrementally increases the likelihood that ventilator-related complications will occur. At least some of these complications, such as GI tract bleeding, 14 nosocomial pneumonia,2 and pulmonary embolism 15 have been shown to result in increased mortality. An increased duration of mechanical ventilation increases the likelihood that the patients will require long-term care or rehabilitation because of Clinical Investigations in Critical Care

Table 5-Studies Reporting the Incidence of Failed Extubation* Study (Year)

Patient Type

No. of Patients

% Reintubated

%Mortality

( Sahn et al 1973)5 Hilberman et al (1976)8 Tahvanainen et al (1983)6 DeHaven et al (1986)10 Demling et al (1988) 1

MICU/SICU Cardiac surgery MICU SICU/trauma General SICU Burn/trauma unit MICU/SICU MICU RICU MICU MICU/SICU MICU/SICU MICU/SICU MICU

100 124 47 48 400 300 269 40 29 52 109 170 530 289

17.0 17.7 19.0 6.3 5.5 3.3 10.4 12.5 14.3 17.0 11.0 23.5 15.7 14.5

NR NR 22.2 NR 40.0 10.0 NR NR NR 33.3 NR 35.0 NR 42.5

Krieger et al (1989)7 Sassoon et al (1993) 11 993)12 Mohsenifar et al (1 Lee et al (1994)3 9 ( Brochard et al 1994) Torres et al (1995)2 Esteban et al (1995)4 Current study

*MICU=medical ICU; SICU = surgical ICU; RICU=respiratory ICU; NR=not reported.

associated prolonged inactivity and consequent increase in debility and deconditioning. Lastly, it is possible that failed extubation is not a direct etiologic contributor to poor outcome but rather serves as an independent marker of severity of illness providing additional prognostic information to that derived from the APACHE II score and the presence of comorbidities. What are the clinical implications of the findings of the current study? To our knowledge, the incidence of extubation failure and the resulting outcome have not been reported in a large cohort of exclusively medical ICU patients. Therefore, these data provide additional prognostic survival information to that afforded with traditional measures. In addition, ICU staffing decisions and discharge planning may be optimized because, once reintubated, on average these patients will spend an additional 12 days receiving mechanical ventilation and approximately two thirds who survive reintubation will ultimately need transfer to a long-term care or rehabilitation facility. The overwhelming majority of patients in this study satisfied generally accepted criteria for proceeding with a trial of extubation. Among those who had weaning and extubation indexes measured, all satisfied at least two of four criteria thought predictive of weaning/extubation success (negative inspiratory force < -20 em H 2 0, vital capacity> 10 mUkg, minute ventilation <15 Umin, fNT < 100 breaths/L) . We found the positive predictive value (PPV) of the fNT (PPV= patients successfully extubated with fNT < 100 divided by all patients with fNT < 100, 133/ 156 or 85%) to be similar to that reported previously.13 In contrast, the negative predictive value (patients who failed with fNT 2':100 divided by all patients with fNT 2':100, 4/18 or 22%) was lower than that seen in studies defining failure as either

weaning or extubation failure 11 · 13 but similar to that reported for extubation failure alone. 3 Although 39% of extubated patients did not have indexes determined prior to extubation, the outcome for these patients was no different from those with index measurements. Accordingly, the observed frequency of extubation failure at 24 h (7%), 48 h (11 %), and 72 h (15%) was equivalent to or below that noted in many of the previous studies reporting reintubation rates (Table 5) . Although our findings indicate that a substantial percentage of patients who appear ready for extubation by standard criteria but who fail ultimately experience significant adverse outcomes, the study design did not allow us to identifY why these poor outcomes occur. Future studies should focus on whether these poor outcomes are etiologically related to extubation failure because it would indicate a need to reassess how extubation decisions are made. Under these circumstances, exclusively emphasizing the end point of extubation success or failure may not be the best outcome measure. Current threshold values for existing predictive indexes were selected to minimize both false-positives (predicted success, actual failure ) and negatives (predicts failure but could be successfully extubated), to yield the highest positive and negative predictive values . In contrast, a threshold value that preferentially reduces the number of false-positives may be desirable if extubation failure occurs at a high cost (mortality, increased duration of ICU stay, need for long-term care). Alternatively, new instruments that emphasize prediction of outcome after failed extubation, rather than the need for reintubation alone, might be sought. In summary, after adjusting for severity of illness and comorbid conditions, extubation failure had a significant independent association with increased CHEST/112/1 / JULY, 1997

191

risk for death, prolonged ICU stay, and need for transfer to a long-term care or rehabilitation facility. Extubation failure may not directly lead to a poor outcome but rather serve as an additional independent marker of severity of illness. Alternatively, poor outcomes may be etiologically related to extubation failure. If future studies prove the latter to be the case, identifYing patients at risk for poor outcomes after extubation failure and instituting alternative care practices could potentially reduce mortality, duration of ICU stay, and need for transfer to a long-term care facility.

7

8

9

lO

REFERENCES 1 Demling RH, Read T, Lind LJ, eta!. Incidence and morbidity of extubation failure in surgical intensive care patients. Crit Care Med 1988; 16:573-77 2 Torres A, Gatell JM , Aznar E, et al. Re-intubation increases the risk of nosocomial pneumonia in patients needing mechanical ventilation. Am J Respir Crit Care Med 1995; 152:137-41 3 Lee KH, Hui KP, Chan TB, et al. Rapid shallow breathing (frequency-tidal volume ratio) did not predict extubation outcome. Chest 1994; 105:540-43 4 Esteban A, Frutos F, Tobin MJ, e t a!. A comparison of four methods of weaning patients from mechanical ventilation. N Eng! J M ed 1995; 332:345-50 5 Sahn SA, Lakshminarayan S. Bedside criteria for discontinuation of mechanical ventilation. Chest 1973; 63:1002-05 6 Tahvanainen J, Salmenpera M, Nikki P. Extubation criteria

192

ll

12 13 14 15

after weaning from intermittent mandatory ventilation and continuous positive airway pressure. Crit Care Med 1983; 11:702-07 Krieger BP, Ershowsky PF, Becker DA, et al. Evaluation of conventional criteria for predicting successful weaning from mechanical ventilatory support in elderly patients. Crit Care Med 1989; 17:858-61 Hilberman M, Kamm B, Lamy M, et a!. An analysis of potential physiological predictors of respiratory adequacy following cardiac surgery. J Thorac Cardiovasc Surg 1976; 71:711-20 Brochard L, Rauss A, Benito S, et al. Comparison of three methods of gradual withdrawal from ventilatory support during weaning from mechanical ventilation. Am J Respir Crit Care Med 1994; 150:896-903 DeHaven CB, Hurst JM , Branson RD. Evaluation of two different extubation criteria: attributes contributing to success. Crit Care Med 1986; 14:92-4 Sassoon CSH, Mahutte CK. Airway occlusion pressure and breathing pattern as predictors of weaning outcome. Am Rev Respir Dis 1993; 148:860-66 Mohsenifar Z, Hay A, Hay J, et al. Gastric intramural pH as a predictor of success or failure in weaning patients from mechanical ventilation. Ann Intern Med 1993; 119:794-98 Yang KL, Tobin MJ. A prospective study of indexes predicting the outcome of trials of weaning from mechanical ventilation. N Engl J M ed 1991; 324:1445-50 Harris SK, Bone RC, Ruth WE. Gastrointestinal hemorrhage in patients in a respiratory intensive care unit. Chest 1977; 72:301-04 Papadakis MA, Mangrone CM, Lee KK. Treatable abdominal pathologic conditions and unsuspected neoplasms at autopsy in veterans who received mechanical ventilation. JAMA 1991; 265:885-89

Clinical Investigations in Critical Care