Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

An evaluation of extubation failure predictors in mechanically ventilated infants and children.

Farias JA, Alía I, Retta A, et al.

Intensive Care Med 2002; 28:752-757. [abstract]

Reviewed by Angela T. Wratney MD MHSc, Daniel Kelly Benjamin Jr. MD MPH PhD, and Ira M. Cheifetz MD FAARC FCCM, Duke University, Durham, NC

Review posted January 25, 2005

I. What is being studied?

Study objective:

1. Assess the accuracy of traditional weaning indices in predicting extubation failure.
2. Compare the accuracy of indices measured at the onset and at the end of a breathing trial (SBT) prior to extubation.

Study design

Prospective, cohort study.

Methods

Patients underwent a trial of spontaneous breathing (SBT) through a T-piece at the same FiO2 level used during mechanical ventilation. Within the first 5 minutes, respiratory therapists collected the following respiratory parameters: respiratory frequency (RR), exhaled minute volume and maximal inspiratory pressure (Pimax). Derived variables included: (1) tidal volume (VT) (indexed to body weight) and (2) frequency to tidal volume ratio (f/VT). The SBT continued for up to two hours with either pressure support ventilation of 10 cm H2O or a T-piece. Respiratory parameters were measured again at the end of the SBT. Those patients who tolerated the SBT were extubated. Those patients who failed to tolerate the SBT (trial failures) resumed mechanical ventilation support. Physicians caring for patients were blinded to the results of the respiratory measurements.

Outcome Assessed

Tolerance of SBT and the need for reintubation within 48 hours of extubation.

II. Are the results of the study valid?

Note: These questions follow from Randolph AG et al. Understanding articles describing clinical prediction tools. Crit Care Med 1998;26:1603-1612. [abstract]

1. Was a representative group of patients completely followed up?

Yes. All infants and children, mechanically ventilated ≥ 48 hours within the medical-surgical pediatric intensive care unit (PICU) of a tertiary care hospital and judged by the primary physician as ready to undergo a SBT prior to extubation were eligible for the study. Pediatric patients, 1 month to 15 years of age were enrolled from May, 1997 to December, 2000. Patients met the following inclusion criteria: (1) improvement or resolution of the underlying cause of acute respiratory failure, (2) PaO2 > 60 mmHg on FiO2 < 0.40 and positive end-expiratory pressure (PEEP) ≤ 5 cm H2O, (3) core temperature < 38.5°C, (4) alert mental status, (5) Hgb > 10 g/dl, and (6) no need for vasoactive agents. Patients with tracheostomy (n = 9), neuromuscular disease (n = 9) or audible air leak around the endotracheal tube (n = 16) were excluded.

A total of 418 patients were enrolled. 95 patients (23%; 95% CI 19-27%) failed the SBT and were not extubated. 323 pts were extubated, 275 patients successfully extubated, and 48 patients were reintubated (14%, 95% CI 11-19%). The primary reasons for the initiation of mechanical ventilation were: acute respiratory failure (n = 310), coma (n = 49) and acute on chronic pulmonary disease (n = 59). The most common causes of acute respiratory failure were pneumonia and bronchiolitis (n = 141) and postoperative state (n = 57). The 25th-75th percentile range for age was 3 to 42 months of age. The 25th to 75th percentile range for duration of mechanical ventilation prior to the SBT was 3-13 days. This cohort of mechanically ventilated pediatric intensive care patients from this tertiary care center in Argentina compares well with those patients identified by the Pediatric Acute Lung Injury and Sepsis Investigators Network among nine large PICUs in the United States and Canada (1). Importantly, Randolph et al demonstrated that a small, percentage (12-24%) of patients require mechanical ventilation ≥ 24 hours (1).

Extubation failure rates range from 2.7-22% depending upon the population studied and the way in which outcomes are defined (2-5). A large multi-center observational study in the US reported an extubation failure rate for the typical ICU patient population of 6.2% (95% CI 5.3-7.1) (2). Patients intubated > 48 hours had a higher extubation failure rate of 8% (2).

Was a follow up sufficiently long and complete?

Yes. All patients were followed for 48 hours after extubation to determine the need for reintubation.

2. Were all potential predictors included?

No. The authors included "traditional weaning indices" to predict extubation failure: RR, exhaled minute volume, VT, Pimax and f/VT . These predictors have demonstrated moderate to excellent discriminatory capacity in adult studies of extubation outcome. They also indexed these measures to body weight which had been shown to increase the predictive capacity of these measures in the pediatric population. Potential predictors of extubation outcome in pediatrics which are not assessed in this study include: the air leak test (7,8), dead space to tidal volume ratio (VD/VT) (4), CROP scores (5) (a derived index of compliance, respiratory frequency, oxygenation, and peak inspiratory pressure). Other predictors used in adult studies which are not included in this study: P0.1 (inspiratory pressure 100 msec into an airway occlusion) and a measure of secretion burden (9-11).

3. Did the investigators test the independent contribution of each predictor variable?

No. Variables were not tested in a regression analysis for independence. The authors tested each variable according to selected cutoff values to determine which parameters could statistically differentiate between those patients who would extubate successfully and those patients who would fail extubation. This is a form of bivariate testing and does not prove one variable is independent of the other(s).

4. Were outcome variables clearly and objectively defined?

Yes. There were three possible outcomes: (1) patients who failed the SBT trial (trial failure), (2) patients requiring reintubation (reintubation group), and (3) patients successfully extubated (successful extubation group). The outcome of SBT was determined by the primary physicians using objective standards for poor tolerance: (1) RR > 90% for age, (2) increased respiratory work (accessory muscle use, retractions, paradoxical breathing), (3) diaphoresis and anxiety, (4) HR > 90th percentile for age, (5) change in mental status, (6) BP < 3% for age, (7) O2 saturation < 90% by pulse oximetry, (8) PaCO2 > 50 mmHg or increased by > 10 mmHg, and (9) arterial pH < 7.30. The outcome of the extubation trial period was determined by the primary physicians based upon subjective evaluation of clinical examination, blood gas, or both. Reintubation or success was defined within 48 hours after extubation. The primary physicians were blinded to the results of the respiratory measurements but were not blinded to the assessment of the SBT.

For the statistical analyses the authors clearly defined success and failure as true positive (TP; index predicted extubation failure, patient reintubated); false positive (FP; index predicted extubation failure, patient successfully extubated); true negative (TN; index predicted extubation success, patient successfully extubated); and false negative (FN; index predicted extubation success, patient reintubated).

III. What are the results?

1. What is(are) the prediction tool(s)?

The ability of each predictor to discriminate between those patients who would extubate successfully and those patients who would fail extubation was tested using receiver operating characteristic curves (ROC) (12). The greater the area under the ROC curve (AUC), the better the discrimination power of the test. An AUC of 0.5 indicates the weaning index has no discriminatory value. From ROC curves, a cutoff value for the predictive index can be chosen which maximizes the accuracy of the test.

The AUCs for the respiratory parameters were not statistically different from one another, either at the beginning or end of the SBT. The AUC for all predictors fell between 0.49 and 0.61 (95% CI 0.40-0.70). The highest AUC corresponded to the f/VT measured at 120 minutes (0.61, 95% CI 0.51-0.70). The cutoff values chosen from the ROC curves were 4 ml/kg VT, 11 breaths/min/ml/kg for f/VT ratio, 45 breaths/min for RR, 20 cm H2O for Pimax, and 200 mmHg for PaO2/FiO2 ratio.

We can conclude:

1. The VT, RR, Pimax, and f/VT ratio are poor predictors of extubation failure in children who have passed a spontaneous breathing trial.
2. The accuracy of these weaning indices does not improve when they are remeasured immediately before extubation.

The authors did not test models using multiple parameters to predict extubation outcome. A model containing some combination of these weaning indices could provide a better predictor of extubation outcome.

2. How well does the model categorize patients into different levels of risk?

Not very well. The accuracy of these cutoff values was described by the sensitivity, specificity, positive and negative likelihood ratios for each predictor tested at the beginning and end of the SBT. The f/VT at the end of the SBT had a sensitivity of 27.1 (95% CI 14.5-39.7), specificity of 91.6 (95% CI 88.4-94.9), positive likelihood ratio of 3.24, and negative likelihood ratio of 0.8.

As the authors point out, any predictor of extubation failure that has 100% sensitivity (proportion of those who were predicted by the test parameter to fail who fail extubation) would be an absolute contraindication for extubation. In pediatric studies, the reported sensitivities for the parameters tested have ranged up to 78% (5,6,13). In this study, the sensitivity never reached above 50%. Therefore, using any one of these weaning parameters to exclude an extubation trial in a given patient, would keep 50% of patients intubated too long when they could have tolerated extubation.

Likelihood ratios (LR) describe the likelihood that a given test result would occur in a patient who will fail extubation compared to a patient who will successfully extubate (14,15). The LR was between 1.17 and 3.24 for all predictors studied at either the onset or after the SBT. Therefore, all predictors contribute a small but potentially important amount of information to help modify the probability of extubation failure.

The f/VT > 11 breaths/min/ml/kg at the end of a SBT had a positive LR of 3.24 indicating that this result is three times more likely to occur in a patient who will fail extubation than it is to occur in a patient who will successfully extubate. The LR is used to modify clinical judgment. If a particular patient is expected to have a 25% chance of extubation failure (pre-test probability), a f/VT ratio > 11 breaths/min/ml/kg measured at the end of the SBT will increase the chance of extubation failure to 52% (post-test probability).

3. How confident are you in the estimates of risk?

95% CI for the sensitivity and specificity of each predictor variable is provided. The f/VT ratio measured at the end of the SBT has a sensitivity of 27.1 (95% CI 14.5-39.7).

IV. Will the results help me in caring for my patients?

1. Does the tool maintain its prediction power in a new sample of patients?

Not assessed. These parameters were not tested in a separate population of similar patients. The authors note that the study should be validated particularly in populations of children weaned without performing a SBT.

2. Are your patients similar to those patients used in deriving and validating the tool(s)?

Potentially. Mechanically ventilated pediatric patients in most PICUs will match the characteristics of this study cohort (1,2,4,5). The age of PICU patients varies widely but can include premature infants to adolescents and young adults. It is not explicitly stated whether this cohort included patients with congenital heart disease, bone marrow transplantation, syndromic conditions, or other commonly identified subgroups of patients who receive mechanical ventilation in the PICU setting.

In this study, patients were enrolled if they met several clinical parameters and the primary physician judged them to be ready to undergo a SBT prior to extubation. By definition this is a selective group of patients. It has been shown that clinical criteria alone poorly correlates with a patients ability to wean successfully from mechanical ventilation (3,15). Farias et al demonstrated that the SBT is not commonly used (31%) prior to extubation attempts (16). Similarly, the weaning and extubation practices of 17 large PICUs revealed only two centers had established weaning protocols, five sites had extubation protocols, and only three reported pre-extubation respiratory mechanics measurements of any type were performed/assessed on ≥ 50% of patients (2). Therefore, the discriminatory power of these predictors will need to be validated in pediatric patients extubated without performing a SBT.

3. Will the results lead directly to selecting or avoiding therapy?

No. The VT, RR, Pimax, and f/VT measured either at the beginning or end of the SBT are poor predictors of extubation outcome in pediatric patients who have passed a SBT.

Patients who have passed a SBT have a low probability of reintubation between 5% and 20% and these weaning indices are unable to discriminate who will extubate successfully and who will fail.

Many physiologic indices do not demonstrate the ability to discriminate among patients who will likely fail extubation and those who will successfully extubate. This may be due to the fact that clinicians have already considered these clinical variables in determining patient distress during weaning attempts such as the SBT (17). Furthermore, physiologic indices do not assess the patency of the upper airway once the endotracheal tube is removed (subglottic edema). Subglottic edema is another cause for patients to require reintubation.

4. Are the results useful for reassuring or counseling patients?

No. These tests would not be useful to reassure or counsel patients as they do not help to modify the probability of extubation failure in the pediatric population which has passed a trial of spontaneous breathing.

References

1. Randolph A, Meert K, O'Neil M, et al. The Feasibility of Conducting Clinical Trials in Infants and Children with Acute Respiratory Failure. Am J Respir Crit Care Med 2003; 167:1334-1340. [abstract]
2. Kurachek SC, Newth CJ, Quasney MW, et al. Extubation failure in pediatric intensive care: a multiple-center study of risk factors and outcomes. Critical Care Medicine 2003; 31(11):2657-64. [abstract]
3. Epstein SK. Decision to extubate. Intensive Care Medicine 2002; 28(5):535-46. [abstract]
4. Hubble CL, Gentile MA, Tripp DS, Craig DM, Meliones JN, Cheifetz IM. Deadspace to tidal volume ration predicts successful extubation in infants and children. Critical Care Medicine 2000; 28(6):2034-2040. [abstract]
5. Baumeister BL, el-Khatib M, Smith PG, Blumer JL. Evaluation of predictors of weaning from mechanical ventilation in pediatric patients. Pediatr Pulmonol 1997; 24(5):344-352. [abstract]
6. Farias JA, Alia I, Esteban A, et al. Weaning from mechanical ventilation in pediatric intensive care patients. Intensive Care Medicine 1998; 24:1070-1075. [abstract]
7. Mhanna MJ, Yaacov BZ, Tichy CM, Super DM. The "air leak" test around the endotracheal tube, as a predictor of postextubation stridor, is age dependent in children. Critical Care Medicine 2002; 30:2639-2643. [abstract]
8. Seid AB, Godin MS, Pransky SM, Kearns DB, Peterson BM. The prognostic value of endotracheal tube-air leak following tracheal surgery in children. Archives of Otolaryngology--Head and Neck Surgery 1991; 117:880-882. [abstract]
9. Capdevila XJ. Perrigault PF. Perey PJ. Roustan JPA. d'Athis F. Occlusion pressure and its ratio to maximum inspiratory pressure are useful predictors for successful extubation following T-piece weaning trial. Chest 1995; 108(2):482-9. [abstract]
10. Coplin WM. Pierson DJ. Cooley KD. Newell DW. Rubenfeld GD.Implications of extubation delay in brain-injured patients meeting standard weaning criteria. American Journal of Respiratory and Critical Care Medicine. 2000; 161(5):1530-1536. [abstract]
11. Khamiees M. Raju P. DeGirolamo A. Amoateng-Adjepong Y. Manthous CA. Predictors of extubation outcome in patients who have successfully completed a spontaneous breathing trial. Chest 2001; 120(4):1262-1270. [abstract]
12. Randolph AG et al. Understanding articles describing clinical prediction tools. Crit Care Med 1998;26:1603-1612 [abstract]
13. Manczur T, Greenough A, Pryor D et al. Comparison of Predictors of Extubation from Mechanical Ventilation in Children. Crit Care Med 2000; 1:28-32. [abstract]
14. Likelihood ratios at the Center for Evidence Based Medicine http://www.cebm.net/likelihood_ratios.asp
15. MacIntyre NR, Cook DJ, Guyatt GH. Evidence-based guidelines for weaning and discontinuing ventilatory support: A collective task force facilitated by the American College of Chest Physicians; the American Association for Respiratory Care; and the American College of Critical Care Medicine. Chest 2001; 120(6):375S-395S. [abstract]
16. Farias J, Frutos F, Esteban A et al. What is the Daily Practice of Mechanical Ventilation in Pediatric Intensive Care Units? A Multicenter Study. Intensive Care Med 2004; 30:918-925. [abstract]
17. Meade M, Guyatt G, Cook D, et al. Predicting Success in Weaning from Mechanical Ventilation. Chest 2001; 120: 400-424. [abstract]

 

Comments

Back to the EB Journal Club Index