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

Weaning from mechanical ventilation in pediatric intensive care patients.

Farias JA, Alia I, Esteban A, Golubicki, Olazarri FA.

Intensive Care Med 1998; 24: 1070-1075. [abstract]

Reviewed by Patricia Jines, MD and Adranik Madikians, MD, University of California, Los Angeles

Review posted September 21, 2000; modified December 18, 2000

Editorial note: This review has been modified based on additional editorial input on December 18, 2000. Additional analysis is included, and the conclusions altered somewhat - that although these predictors may not predict extubation failure well, they do predict extubation success, and therefore could be valuable in the appropriate setting.
The original review is here.

I. What is being studied?

Study objective:

Evaluate the accuracy of some weaning indices in predicting weaning failure in mechanically pediatric ventilated patients.

Study design

Prospective interventional study.

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?

Patients were 84 consecutive children requiring ventilation for at least 48 hours, recruited from the medical/surgical intensive care unit of the Children's Hospital R Gutierrez, Buenos Aires, from March 1995 to April 1996. Inclusion criteria for weaning was judged by the primary physician and was based in the following criteria: improvement or resolution of the underlying cause of acute respiratory failure, pharmacological control of bronchoconstriction, absence of respiratory acidosis, a fractional inspired oxygen of (Fi02) 0.40, a positive end-expiratory pressure of PEEP 5 cm H20, a peak inspiratory pressure < 25 cm H20, body temperature < 38.5 C, no sedation, correction of electrolyte disorders, and absence of neuromuscular blockade 24 hours prior to the start of weaning. Excluded were patients with neuromuscular disease or tracheostomy. From the 84 consecutive patients enrolled in the study, 75 patients successfully underwent a trial of spontaneous breathing, were extubated (criteria noted in II.4 below) and followed during at least 48 hours after extubation. Of these 75 patients, 12 required reintubation within 48 hours.

2. Were all potential predictors included?

Several variables were included in order to generate predictors: age, body weight, duration of ventilation support, peak inspiratory pressure, tidal volume (Vt) frequency-to-tidal volume ratio (f/Vt ratio), respiratory rate, Fi02, heart rate, blood pressure and pulmonary gas exchange values. All of them were compared as median with 25th %ile and 75th %ile. The only variables considered as predictors were the ones that showed significant differences between the successful extubation, reintubation, and trial failure groups

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

The predictor variables were tested independently and then with multiple regression. Of note, the variables of tidal volume and f/Vt ratio both include tidal volume; therefore these two variables are not truly independent variables (1).

4. Were outcome variables clearly and objectively defined?

Two weaning outcome variables were clearly defined: 1) successful weaning, if extubation was performed after a 2 hour trial of spontaneous breathing and reintubation was not required within 48 hours of extubation. 2) failure to wean, defined as requiring reinstitution of mechanical ventilation at any time during the 2 hour trial of spontaneous breathing (trial failure group), or needing reintubation within 48 hours of extubation after a successful breathing trial (reintubation group). Criteria for termination of the spontaneous breathing trial and reintubation were: 1) respiratory rate of more than 62, 52, or 40 breaths/min in patients aged 12 months or less, 13 to 48 months, and 49 to 180 months, respectively; 2) mean blood pressure lower than 45, 55, or 65 in patients aged 12 months or less, 13 to 48 months, or 49 to 180 months, respectively; 3) increase in heart rate of more than 20% with respect to the heart rate on mechanical ventilation; 4) signs of increased respiratory work; 5) PaO2 < 60 mmHg or arterial oxygen saturation < 90%; 6) arterial pH 7.30; 7) partial pressure of arterial carbon dioxide > 50 mmHg or an increase 8 mmHg. The decision to reintubate was based on clinical examination, blood gases, or both.

III. What are the results?

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

Multivariate logistic regression analysis was used to determine which variables were useful to predict trial failure of spontaneous breathing and which were useful to predict intubation.

Of all the characteristics of the study population compared the only two measurements that were statistically significant between patients in the trial success vs trial failure and reintubation groups were tidal volume and f/Vt ratio. Reintubated patients were significantly younger than patients successfully extubated and were not different from patients failing the spontaneous breathing trial. Trial failure was found in significantly smaller patients and with longer duration of ventilatory support .

From Tables 2 and 3 we can say that according to the NPV (true negatives/true negatives + false negatives) for a f/Vt cutoff value of 11, the trial success is 0.92 and the extubation success is 0.90. That is, if the f/Vt < 11 there is a 92% chance of trial success or a 90% of extubation success as well. (True positives are defined here as when the test predicted failure and failure actually occurred, while a true negative is defined as when the test predicted weaning success and success actually occurred.)

We can also say that according to the NPV for a tidal volume cut off value of 4cc/kg the trial success is 0.94 and the extubation success is 0.88. This means that if the VT is > 4 cc/kg the child has a 94% chance of trial success or 88% chance of extubation success.

The PPV (true positives/true positives + false positives) for for trial failure for Vt is 0.38 and for f/Vt is 0.21. Due to these low values a low Vt or a high f/Vt does not help to predict failure very well.

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

This model categorized patients according to different levels of risk based upon tidal volume and f/Vt. They found the risk of failing a trial of spontaneous breathing increased 2.6 times as tidal volume decreased by 1 ml/kg (OR 2.60, 95% CI 1.40, 24.9, for each decrement of 1 cc/kg) and the risk of reintubation after discontinuation of mechanical ventilation increased 23% as f/Vt ratio indexed to body weight increases by 1 breath/min per ml per kg (OR 1.23, 95% CI 1.11, 1.36, for each increment in one point).

The likelihood ratio (LR) of a positive test is the true positive rate divided by the false positive rate or sensitivity divided by 1-specificity. Knowing the pretest probability of extubation failure (0.25 in their hands), using a nomogram, one can determine the "posttest probability" of failure. In this case, the LR of a positive test (say Vt = 4 cc/kg) is 0.55/0.11 or 5. This gives us a posttest probability of about 60% (they say .53 in their discussion for an f/Vt ratio of > 11, obviously using a real formula instead of a nomogram to do the calculation). So a Vt = 4 cc/kg "predicts" about half of patients will fail - which means half will succeed! If we used this test to avoid extubation, we'd be keeping half of our patients intubated too long!

ROC or receiver operator characteristic curves are graphical representations of the trade off between the true positive and false positive rates for every possible cut off. The plot shows the false positive rate (1- specificity) on the X axis and the true positive rate (sensitivity) on the Y axis for the different values of VT or f/Vt The area under the curve would provide a visual and quantitative means of assessing the strength of association of the test and perhaps better assist in picking a unique cut off time.

3. How precise are the estimates of likelihood?

The authors do not provide confidence intervals for the positive and negative predictive values, but with the data they provide, these are readily calculated. (See http://members.aol.com/johnp71/ctab2x2.html for an online interactive calculator.) As might be expected, the low positive predictive values overall have broad confidence intervals, while the stronger negative predictive values have narrower confidence intervals. This is in part due to the small numbers of failures. In other words, we can be more certain of predicting success than failure!

Failure of...	PPV	95% CI	NPV	95% CI
Trial Failure
f/Vt > 11	0.21	.086, 0.36	0.92	0.88, 0.97
Vt </= 4 cc/kg	0.39	0.17, 0.58	0.94	0.90, 0.98
Extubation Failure
f/Vt > 11	0.40	0.19, 0.60	0.90	0.85, 0.95
Vt </= 4 cc/kg	0.50	0.19, 0.81	0.88	0.84, 0.92
Overall Weaning Failure
f/Vt > 11	0.53	0.32, 0.71	0.83	0.77, 0.89
Vt </= 4 cc/kg	0.69	0.42, 0.89	0.83	0.78, 0.87

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?

The study was designed for the study population above described. It was not applied to any other group of patients in the same fashion.

One of the predictors (f/Vt) has shown similar results in other studies. In the adult population, as it is mentioned in this article, a f/Vt ratio < 100 breaths/min/liter is helpful to identify patients with a high probability of weaning success or extubation success. In the pediatric population, two studies have evaluated the Rapid Shallow Breathing Index (f/Vt) and CROP index (dynamic compliance x maximum negative inspiratory pressure x arterial-alveolar gradient divided by respiratory rate) and have found them potentially useful to predict extubation success (2,3).

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

The patient population in our PICU is similar to the one enrolled in the study but the weaning was performed differently because the use of sedation. Usually continuous sedation is withdrawn and intermittent doses are started 12 to 24 hours anticipating ventilator weaning in our unit.

3. Does the tool improve your clinical decisions?

The indexes considered predictors of weaning can be very useful. The larger the tidal volumes, the more you can be certain of weaning success. If the tidal volume is 4 ml/kg, the chance of extubation success and failure is about 50:50. Most institutions would consider a failure rate of 50% as unacceptable overall but in some patients, clinicians may have to take the chance. If the patient has a 50% predicted failure rate then the clinician may want to be ready to reintubate. Furthermore, the negative predictive values are strong, e.g., a tidal volume of more than 4 cc/kg predicts successful extubation almost 90% of the time.

References

1. Randolph AG et al. Understanding articles describing clinical prediction tools. Crit Care Med 1998;26:1603-1612. [abstract]
2. Thiagarajan RR. Bratton SL, Martin LD, Brogan TV, Taylor D. Predictors of Successful Extubation in Children. Am J Respir Crit Care Med Volume 1999; 160:1562-1566. [abstract] [full-text for subscribers]
3. 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):344-52. [abstract] [PedsCCM EBJC Review]

 

Comments

Back to the EB Journal Club Index