Page 17
a spontaneous breathing trial (SBT), others have used permanent discontinuation of the ventilator, and others have combined successful discontinuation and extubation. This latter approach is not optimal, given the differences in the pathophysiology of discontinuation versus extubation failure (see below).102,106 In addition, different studies use different durations of ventilator discontinuation or extubation to define success or failure. Although 24–48 hours of unassisted breathing often is considered to define the successful discontinuation of ventilator support, many studies use shorter time periods to indicate success and often do not report subsequent reintubation rates or the need to reinstitute mechanical ventilatory support. Fourth, a number of ways have been used to express predictor performance, and many can be confusing or misleading. Traditional indexes of diagnostic test power include sensitivity/specificity and positive/negative predictive values. These indexes are limited, however, in that they rely on a single cut point or threshold and that they do not provide an easy way to go from pretest likelihood or probability, through testing, to a posttest probability. The McMaster AHCPR report4 recommends the use of likelihood ratios (LRs), and these will be used in this report to describe predictor performance. The LR is an expression of the odds that a given test result will be present in a patient with a given condition compared to a patient without the condition. An LR 1 indicates that the probability of success increases, while values 1 indicate that the probability of failure increases. LRs between 0.5 and 2 indicate that a weaning parameter is associated with only small, clinically unimportant changes in the posttest probability of success or failure. In contrast LRs from 2 to 5 and from 0.3 to 0.5 correlate with small but potentially important changes in probability, while ratios of 5 to 10 or 0.1 to 0.3 correlate with more clinically important changes EVIDENCE-BASED GUIDELINES FOR WEANING AND DISCONTINUING VENTILATORY SUPPORT RESPIRATORY CARE • JANUARY 2002 VOL 47 NO 1 73 in probability. Ratios of 10 or 0.1 correlate with very large changes in probability.115 Finally, because some investigators report data as continuous values (eg, means) rather than providing defined threshold values, combining studies using meta-analytic techniques often cannot be done. Recommendation 2. Patients receiving mechanical ventilation for respiratory failure should undergo a formal assessment of discontinuation potential if the following criteria are satisfied: 1. Evidence for some reversal of the underlying cause of respiratory failure; 2. Adequate oxygenation (eg, PaO2 /FIO2 150–200; requiring positive end-expiratory pressure [PEEP] 5–8 cm H2O; FIO2 0.4–0.5) and pH (eg, 7.25); 3. Hemodynamic stability as defined by the absence of active myocardial ischemia and the absence of clinically important hypotension (ie, a condition requiring no vasopressor therapy or therapy with only low-dose vasopressors such as dopamine or dobutamine 5 g/kg/min); and 4. The capability to initiate an inspiratory effort. The decision to use these criteria must be individualized. Some patients not satisfying all of the above the criteria (eg, patients with chronic hypoxemia below the thresholds cited) may be ready for attempts at discontinuation of mechanical ventilation. Rationale and Evidence (Grade B) While some investigators argue that the process of discontinuation starts as soon as the patient is intubated, it would seem reasonable that an appropriate level of ventilatory support should be maintained until the underlying cause of acute respiratory failure and any complicating issues have shown some sign of reversal. Indeed, patients with unresolving respiratory failure who require high levels of ventilatory support are probably at high risk for respiratory muscle fatigue (and the consequent prolongation of the need for mechanical ventilation) if aggressive reductions in support are undertaken.50,52,116–118 The criteria used by clinicians to define disease “reversal,” however, have been neither defined nor prospectively evaluated in a randomized controlled trial. Rather, various