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

Validation of predictors of extubation success and failure in mechanically ventilated infants and children

Venkataraman ST, Khan N, Brown A.

Crit Care Med. 2000;28(8):2991-6. [abstract]

Reviewed by David Epstein, MD and Robert Bart III, MD, Childrens Hospital of Los Angeles, Keck School of Medicine, University of Southern California, Los Angeles, CA

Review posted December 21, 2000

I. What is being studied?

Study objective:

The purpose of this study was to validate predictors of extubation success and failure in mechanically ventilated infants and children by using bedside measures of respiratory function. The original predictors of extubation success and failure used in this study were previously defined and analyzed by the same authors (1).

Study design

This was a prospective, descriptive study of infants and children mechanically ventilated for greater than 24 hours in a neonatal or pediatric intensive care unit in a large urban, academic children's hospital. It was designed as a validation study of variables identified and analyzed to predict successful extubation in a prior study by these authors. The predictors analyzed in the prior study included: respiratory rate standardized to age, spontaneous inspiratory pressure, maximum negative inspiratory pressure of a spontaneous breath, inspiratory pressure/maximal negative inspiratory pressure ratio, spontaneous tidal volume indexed to body weight, mean airway pressure, oxygenation index, fraction of total minute ventilation provided by the ventilator, peak ventilatory inspiratory pressure, dynamic compliance, mean inspiratory flow, respiratory rate to tidal volume ratio indexed to body weight, and CROP index (dynamic compliance, respiratory rate, oxygenation, and maximum negative inspiratory pressure). The data collected were similar to the prior study, so a comparison could be made between the demographic and ventilator variables.

The threshold values for prediction of a low ( 10%) and high ( 25%) extubation failure rate were established in the prior study. These values were discovered to be statistically significant in the prior study. They were used in the validation study to categorize patients into different levels of risk for the same study variables in this new group of 312 patients.

Variable	Low-Risk Value	High Risk Value	Odds Ratio	95% CI of OR
Vtspont	6.5 cc/kg	3.5 cc/kg	3.2	1.1 to 9.6
FiO2	0.30	> 0.40	3.6	1.2 to 11.1
Paw	< 5 cm H2O	> 8.5 cm H2O	4.9	0.6 to 38.4
OI	1.4	> 4.5	6.1	1.2 to 30.4
FrVe	20%	30%	4.3	1.6 to 11.4
PIP	25 cm H2O	30 cm H2O	4.5	1.8 to 11.7
Cdyn	0.9 cc/kg/cm H2O	< 0.4 cc/kg/cm H2O	3.3	0.6 to 18.7
Vt/Vi	14 cc/kg/sec	8 cc/kg/sec	3.8	1.4 to 10.7

Patients excluded were those with neuromuscular disease, neonates < 37 weeks gestation, and patients reintubated for upper airway obstruction.

All patients were weaned using synchronized intermittent mandatory ventilation (SIMV), pressure support ventilation, or a combination of SIMV with pressure support. SIMV breaths were volume regulated and time cycled ventilation with no control or limitation of peak inspiratory pressure. The decision to extubate was made by the primary physician and was based on clinical examination, blood gases, and available ventilatory parameters. Extubation failure was defined as reintubation within 48 hours.

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?

It was difficult to determine if a representative group of patients was selected because underlying disease processes were not mentioned or classified. The incidence of reintubation was similar (16.0% vs. 16.3%) between the two studies; however, there were no extubation failures due to upper airway obstruction. This is unusual, since the reported reintubation rates for stridor/airway obstruction are about 5%. The prior study had an incidence of 2.3% extubation failure due to upper airway obstruction.

There were no significant differences among the selected characteristics of the validation (current study) and the prediction (prior study) groups. The preextubation data (age, arterial blood gas variables, duration of intubation, and ventilator rate) were similar between the two groups. The reasons for reintubation were similar as well and classified with regards to disease process. The groups were both completely followed up within the time period of the study. It is assumed that the duration of the patient's involvement is limited to their hospitalization in the intensive care unit and need for mechanical ventilatory support.

2. Were all potential predictors included?

Yes. The authors were thorough in their evaluation of potential predictors of failure of extubation. The potential predictors included were the aforementioned variables in the study design. Air leak around the endotracheal tube was not a study variable.

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

Yes. Using a regression test for linear trends in proportion with preselected ranges and the corresponding failure rates, each predictor was analyzed. The ranges for each variable and index were selected based on the ranges established in the prior study. A Z statistic of 1.96 corresponding to p 0.05 was considered statistically significant. The results of the Z statistic and p value for each variable were compared between the prediction (prior study) and validation (current study) groups.

4. Were outcome variables clearly and objectively defined?

Yes. The outcome assessed was clearly defined as the successful extubation versus the need for reintubation at 48-hours. This was not objectively defined, as the decision to extubate or reintubate a patient was based on clinical examination, arterial blood gas values, or mechanical ventilation parameters interpreted by individual physicians blinded to the results of the study. It is expected that variation exists between different physicians and it was mentioned that there was a change in clinical practice since the prior study. While an objective protocol for extubation or reintubation was not defined, the study represents actual clinical practice and not theoretical, rigid, protocol driven practice.

III. What are the results?

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

The study identified possible univariate and multivariate predictors for extubation success or failure in mechanically ventilated infants and children. The prediction tools established in the prior study proved to be significant in this follow up study of similar patients. The spontaneous tidal volume (without pressure support) indexed to body weight, fraction of inspired oxygen, oxygenation index, fraction of total minute ventilation provided by the ventilator, peak ventilatory pressure, and mean inspiratory flow were statistically significant in establishing extubation success or failure in this study. They showed equivalent Z statistics and p values to the prior study population. Mean airway pressure was found to be statistically significant according to the Z statistics and p value criteria, but in the prior study the confidence interval lower limit was less than one. This suggests that it was not statistically significant because when using confidence interval with odds ratios, a value less than one indicates there is no intergroup difference. Dynamic compliance was included, even though it did not reach statistical significance, due to the fact that as dynamic compliance decreased, there was an increased trend for failure of extubation.

While the authors tested a multivariable model, the CROP index, it was not found to be statistically significant, even though the four variables used to calculate the CROP index were statistically significant. No other multivariable models were tested.

The authors concluded: "infants and children fail extubation because of poor effort (a decreased spontaneous tidal volume), increased load on the respiratory muscles (a high PIP and a low dynamic compliance), or a decreased inspiratory drive (decreased Vt/Ti). Additionally, an increased preextubation level of ventilator support (a high Paw, OI, or FrVe) is associated with increased failure rates."

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

It categorizes patients into different levels of risk to some extent. The prior study established the cutoff variables of high risk ( 25%) and low risk ( 10%) for the various predictors of failure of extubation. Odds ratios were established in the prior study. In this study, the definitions and results of the predictive variables and their relative risks were consistent with the prior study.

While the low risk of extubation failure ( 10%) seems like a reasonable cutoff, some would argue that the high risk of extubation failure ( 25%) is too low. It could be said that 75% of the high-risk group patients had a successful extubation in this case. Also, there was no intermediate group identified to cover the risk of extubation failure of 10 to 25%.

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

Somewhat confident. While this study mentions that confidence intervals were included in the analysis, they do not appear in the paper. One could assume that the prior study's confidence intervals are similar because of the similar patient population (with regards to the measured variables), larger number of patients, and similar statistically significant results.

The 95% confidence intervals for the odds ratios, detailed in the prior study, appear quite wide and there is some overlap in the variables to a confidence interval of a value close to one or below. While the odds ratios for the Paw and the Cdyn show a strong trend, they are not statistically significant because the 95% confidence intervals cross the value of one.

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?

Yes. This study was a new sample of patients that validated the prediction tools established in the prior study. The results were very similar and consistent with the results provided in the prior study. However, it may be better to validate the prediction tools in a similar patient population from a different institution and with different physicians. This may have decreased the bias introduced by the fact that the prior study was completed and the results presented before the validation study was started.

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

Possibly. While our hospital is an urban, tertiary care children's hospital with a demographically similar patient population, it is difficult to compare the patient groups with certainty without knowing the underlying disease processes of the patients in the study.

3. Does the tool improve your clinical decisions?

Possibly. While no single prediction tool should affect the clinical decision to extubate a patient, the combination of variables presented by the authors may improve the clinical decision to extubate a patient. The parameters aid a clinician if the patient is classified as low risk for extubation failure, but they hinder the clinician's decision in patients that are classified as high risk for extubation failure. If extubation is delayed in patients because they are classified as high risk, 75% of these patients could possibly be intubated longer than necessary. The results support the idea that poor effort, increased load on the respiratory muscles, decreased inspiratory drive, and increased preextubation level of ventilatory support affect extubation outcome adversely.

References

1. 1. Khan NK, Brown A, Venkataraman ST: Predictors of extubation success and failure in mechanically ventilated infants and children. Crit Care Med 1996; 24(9): 1568-1579. [abstract]

 

Comments

Back to the EB Journal Club Index