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

Stress test and gastric-arterial PCO2 measurement improve prediction of successful extubation

Uusaro A, Chittock DR, Russell JA, Walley KR.

Crit Care Med. 2000;28(7):2313-9. [abstract]

Reviewed by Nancy Hoover MD and Atul Vats MD, Department of Critical Care Medicine, Children's Healthcare of Atlanta Atlanta, Georgia

Review posted September 20, 2002

I. What is being studied?

Study objective:

To determine whether a "stress test" improves prediction of extubation outcome and to determine the most predictive variables. The hypothesis is that repeated measurements during a period of increased respiratory muscle workload would predict success of extubation more accurately than single time point measurements.

Study design

Observational study that included 68 mechanically ventilated adults who were recovering from acute respiratory failure. Inclusion criteria were that the patients were judged ready for extubation by the attending physicians not involved in the study and the patients already had a NG tube in place. All patients in the study were extubated and successful extubation was defined as the ability to sustain spontaneous breathing without reintubation for > 24 hours.

All measurements on the patients were made 2 times. The first set of measurements was done after a 60 minute period of being on pressure support (PS) of +5 and CPAP of 5 without any other ventilator changes. The second set of measurements was done after being on PS of zero and CPAP of 5 for 60 minutes. All measurements were done the same way and during the last 5 minutes of each 60 minute trial period. All patients were then extubated.

Measurements included respiratory frequency for 1 minute, tidal volume (TV), frequency to TV ratio, airway pressure after 100 milliseconds against an occluded airway (P0.1), gastric PCO2, and ABG. This was used to determine the difference between gastric juice PCO2 and arterial PCO2 (DPg-aCO2).

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? Was follow-up sufficiently long and complete? Was the sample representative of the larger population they were trying to sample from?

All patients were extubated and were completely followed up for the first 24 hours after extubation. This follow-up is sufficient for this patient population recovering from acute respiratory failure whose endpoint was success of extubation. The patients represented the primary diagnostic categories of an adult ICU and included pulmonary, cardiac, gastrointestinal, trauma, neurologic and miscellaneous categories.

2. Were all potential predictors included?

The investigators used published threshold values for all variables to test the hypothesis that a stress test improves predictive ability of these published values. They did not include the other well-established criteria for extubation. In adults, objective measures have been reviewed and are as follows: Oxygenation criteria include (1) PaO2 ≥ 60 with FiO2 < 35% (2) AaDO2 < 350 (3) PaO2/FiO2 ratio > 200 and (4) QS/QT 10-20%. Ventilatory pump criteria are (1) VC > 10-15 ml/kg (2) maximum negative inspiratory pressure > -30 cm H2O (3) maximum voluntary ventilation ≥ twice the resting minute ventilation (4) thoracic compliance > 25 ml/ cm H2O (5) TV > 300 ml (6) dead space/TV ratio less than 0.6.

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

Receiver operating characteristic curves were determined for different variables to assess each variable's ability to discriminate between two groups of patients (success and failure of extubation) independent of threshold values. Differences between receiver operating characteristic curves were identified by using nonparametric comparisons of the areas under the curves. A Mann-Whitney test was used for between-group comparison of those who failed and succeeded extubation.

4. Were outcome variables clearly and objectively defined?

The results revealed 17/68 patients failed (25%) and 51/68 patients (75%) succeeded in remaining extubated for > 24 hours. The outcome of extubation for 24 hours was clearly defined. If a patient was reintubated for any reason, they were included in the failure group. Objectivity was maintained and evidenced by the fact that the authors did not evaluate the reasons for failed extubation. Had they not included patients who were electively reintubated for surgery or a procedure, they could not have been objective about the criteria used to evaluate failure or success.

III. What are the results?

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

The prediction tools chosen were easily obtained at the bedside in a couple of minutes if the patient had a NG tube already place and an arterial gas was simultaneously obtained. The authors determined that a DPg-aco2 of less than 12 after a 1-hour stress test was the best predictor of extubation success. There was a sensitivity of 0.70, a specificity of 0.94, a NPV of 0.55 and a PPV of 0.97 which all were significant improvements from before the stress test.

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

The authors included two receiver operating characteristic (ROC) curves: frequency to tidal volume ratio and the difference between gastric juice and arterial PCO2. The greater the area under the ROC curve, the better the discriminating power of the model. In each case, the 1-hour stress test increased the area under the receiver operating characteristic curves.

The following table is a small sample of Table 2 from the article, demonstrating the values of predicting extubation success for the different variables measured. The TV, respiratory rate, respiratory rate/tidal volume and the minute ventilation had similar sensitivity (SN), specificity (SP), negative predictive value (NPV), positive predictive value (PPV) and likelihood ratios (LR) before and after the stress test so all were not reproduced here. We have added the likelihood ratios that were not provided in the article.

Variable/condition	Sensitivity	Specificity	NPV	PPV	LR
TV (>325 cc)   PS5	0.92	0.24	0.50	0.78	1.2
Stress test	0.92	0.29	0.55	0.79	1.3
P0.1 (< 5.5 mmHg)   PS5	0.91	0.36	0.56	0.82	1.4
Stress test	0.74	0.56	0.43	0.83	1.7
DPg-aco2 (< 12 mm Hg)    PS5	0.69	0.55	0.29	0.87	1.5
Stress test	0.70	0.94	0.55	0.97	11.7

Table: Predictors of extubation success

The NPV represents the percentage of people with a negative test (e.g., a TV < 325) who do not have the disease being tested Ð the "disease" here representing extubation success. A PPV represents the percentage of people with a positive test (e.g., a TV > 325) who actually have the disease, i.e., are successfully extubated. A LR is the likelihood that a given test would be expected in a patient with a disease compared to the likelihood that the same result would be expected in a patient without that disease. So patients who are successfully extubated are 11.7 times as likely to have a DPg-aco2 < 12 that those who fail extubation. Note that the LR of greater than 10 is strongly supportive of a meaningful and useful test.

After the stress test of being on pressure support of zero, gastric PCO2 was 1.8 times higher and DPg-aCO2 was 14 times higher in the failure group. TV was 15% lower in the failure group. P0.1 was 41% greater in the failure group. DPg-aCO2 had the highest positive predictive values on pressure support of 5, which increased further after the stress test.

The main finding of the study was that a 1-hour stress test off pressure support, improved one's ability to predict successful extubation as evidenced by the high likelihood ratio of the change in gastric to arterial pCO2. The respiratory mechanics were generally sensitive but not specific predictors and resulted in lower positive predictive values.

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

The authors did not provide confidence intervals in the analysis of their data so it is difficult to comment on the precision of these estimates.

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 prediction tool was not tested again on a different, independent sample of patients.

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

A lot of the disease states are the same, but the patient population in a Pediatric Critical Care Unit is much more diverse. The age range and patient size is much more broad leading to a wide variety of ET tube sizes. This may alter the ability to do a 60-minute stress test on a small child with a small ET tube.

Another concern is that the authors do not provide any reasons for extubation failure. This test only predicts those that failed from respiratory insufficiency with hypercarbia and would not necessarily apply to those reintubated for mechanical failure such as extrathoracic airway edema and stridor.

3. Does the tool improve your clinical decisions?

Until a similar study is done in children, I would not use the gastric PCO2 as a clinical prediction tool. It might prove useful in older children that could tolerate a stress test off pressure support.

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

Not at this time.

References

1. Meade MO, Guyatt GH, Cook DJ. Weaning from mechanical ventilation: the evidence from clinical research. Respir Care 2001; 46(12):1408-15. [abstract]

 

Comments

Back to the EB Journal Club Index