Stress Testing the Respiratory System: Too Little or Too Much?

Nearly 27% of hospitalizations in the United States require stays in the intensive care unit (ICU), and they account for 48% of hospital charges (1). A majority of such ICU use was by patients with respiratory conditions requiring mechanical ventilation support for less than 96 hours (1). As a consequence, the annual burden imposed by patients requiring mechanical ventilation in the United States amounts to a 2005 estimate of 800,000 hospitalizations that cost nearly $27 billion (2). Since 2005, the number of critical care beds in the United States, driven by demand, has grown by 15% (3). Such a burgeoning demand for critical care services in patients requiring mechanical ventilation is further compounded by their length of stay. In patients receiving mechanical ventilation, 40–50% of the total time spent in the ICU is devoted to the weaning process, which, in turn, is rate-limited by the intensivist being able to predict the patient’s ability to sustain spontaneous breathing postextubation (4, 5).

Assessment of readiness to breathe spontaneously can be performed before extubation by trying to mimic the postextubation conditions by providing low levels of pressure support (5–7 cm H₂O) with positive end-expiratory pressure (PEEP) of 5 cm H₂O, continuous positive airway pressure (CPAP), or T-piece. The rationale for providing the low levels of inspiratory assistance in the form of pressure support is to counter the airway resistance imposed by resistance of the endotracheal tube and concretions that may have accumulated since the time the tube was first inserted, and therefore to best simulate the postextubation physiological state (6). However, others have shown that the resistive load of the endotracheal tube is supplanted by that imposed by edematous upper airways after extubation, and that the work of breathing is identical both before and after extubation after a T-piece trial (7). Moreover, there is the issue of “occult” elastic load presented by intrinsic PEEP in patients with chronic obstructive pulmonary disease (COPD) that is effectively “masked” by application of CPAP or PEEP. In patients with COPD, CPAP set close to the intrinsic PEEP levels (∼8 cm H₂O) reduces the pressure time product of the diaphragm (a measure of respiratory effort) by 45% when compared with T-piece (8). CPAP therefore effectively masks the elastic load imposed by occult intrinsic PEEP in patients with COPD.

Occult is a term that could be used because, despite a significant body of scientific literature on the effect of intrinsic PEEP and dynamic hyperinflation on breathing effort in critically ill patients with COPD, it is infeasible to perform accurate measures of such physiological variables outside the research setting (9, 10). Potentially, in such patients who seemingly appear to breathe normally when receiving CPAP or pressure support combined with PEEP, withdrawal of CPAP/PEEP after extubation could lead to marked increase in elastic load, which, combined with respiratory muscle weakness in critically ill patients, could potentially lead to rapid respiratory compromise, warranting reintubation (5, 11, 12).

It can be argued that the aforementioned logic cites individual physiological studies, with very small sample size and high variability in the subject population. It is for this reason that the pooled analysis by the rigorous metaanalysis in this issue of the Journal by Sklar and colleagues (pp. 1477–1485) is most welcome (13). Sklar and colleagues reviewed 4,138 citations to yield 16 studies involving 239 patients (13). They demonstrate a systematic underestimation of the work of breathing in the postextubation state when breathing trials are performed, using pressure support and CPAP as opposed to spontaneous breathing while receiving T-piece, which better simulates the physiological conditions postextubation (13). The implications of these findings are that when we assess patients’ ability to breathe spontaneously before extubation, we may be providing more respiratory assistance than we would like to if the goal were to mimic the postextubation state before discontinuation of mechanical ventilation. This is in interesting contrast to the current American Thoracic Society and American College of Chest Physicians guideline that for acutely hospitalized patients ventilated more than 24 hours, there is a recommendation that initial spontaneous breathing trial be conducted with inspiratory pressure augmentation (5–8 cm H₂O), rather than without such support (T-piece or CPAP) (conditional recommendation; moderate-quality evidence [14]). Conceivably, after extubation, patients would be able to increase their respiratory effort and surmount such an increase in respiratory load and avoid needing reintubation. However, in certain instances such as the elastic load imposed by intrinsic PEEP in a patient with COPD who was weakened by critical illness, this may lead to rapid respiratory decline after extubation. Although such rapid declines may be few and far between, and could conceivably be tackled by noninvasive ventilation or reintubation, the original purpose of simulating the postextubation state is not being accomplished and may subject a few to grave risk.

Considering the aforementioned public health burden imposed by critically ill patients receiving mechanical ventilation, the benefits of providing the inspiratory assistance with pressure support or CPAP for breathing trials would be that we could expeditiously liberate patients from mechanical ventilation and reduce time spent on the mechanical ventilator, which incrementally increases their risk for nosocomial infections. Conversely, we may be increasing the risk for rare but catastrophic events. Perhaps a tailored approach could be undertaken in patients with COPD by adopting a two-stepped “respiratory stress test” of 30 minutes each of pressure support or CPAP followed by T-piece, and tested in an adequately powered large, pragmatic study. This would not be dissimilar to how the modified Bruce protocol for cardiac stress testing evolved.

We have nearly a million patients receiving mechanical ventilation in the United States alone. There are a few among them who are at risk for catastrophic events if we were not to accurately simulate their postextubation state. In a sense, we are balancing the betterment of many against the risk for a few. In the era of precision medicine, big data, large pragmatic trials, and integrated healthcare systems, this should not be a philosophical issue but a scientific problem.