In the 1980s, weaning from long-term mechanical ventilation was an arduous process. Individualized plans for improving the patients’ overall physical state during the pre-weaning stage (ie, the patient is stable yet numerous clinical impediments to active weaning exist) were the norm, with little standardization. Determination of weaning readiness was often accomplished by testing the patient with various weaning predictors such as standard weaning parameters (ie, negative inspiratory pressure, positive expiratory pressure, spontaneous tidal volume, and vital capacity), the rapid-shallow breathing index (f_x/V_t), the wean index (WI), and occlusion pressure (P.01) to name just a few.^21–24 Despite early study outcomes that suggested the predictors reliably identified the ability of the patient to be successfully extubated, their use in practice and subsequent studies suggested that they were not reliable positive predictors, that is, they did not tell us when the patient could wean.^25,26 In addition, the predictors tested pulmonary factors to the exclusion of nonpulmonary factors for readiness to wean. Their use diminished except as a means of predicting weaning failure (ie, as negative predictors), and to that end, some have been used in current studies to do just that. One recent study used a threshold number for the f_x/V_t as a criterion for application of the “wake up and breathe” study protocol.⁷ The use of weaning predictors to date is rare except in examples such as noted in the Girard study.⁷ This is in large part due to the positive results of studies accomplished in the late 1990s that focused on weaning methods, specifically the use of weaning protocols and guidelines for sedation management.

The use of protocols for weaning trials was stimulated by the work of Brochard and colleagues and Esteban and colleagues, in RCTs testing the efficacy of different modes of ventilation applied by protocol to patients requiring prolonged ventilation.^27,28 The results of these two studies were similar in that they both demonstrated that weaning protocols, regardless of the ventilator mode, resulted in shorter ventilator duration than when weaning was accomplished via the existing “standard” (ie, individually tailored wean plan). In the Esteban study, patients (n = 546) who met study criteria were placed on a spontaneous breathing trial (SBT) for 2 hours.²⁸ If the patient did not tolerate the trial he or she was randomly assigned to a specific weaning protocol (ie, intermittent mandatory ventilation [IMV], pressure support ventilation [PSV], or SBT) or the control method (ie, standard weaning method). If the patient passed the 2-hour SBT he or she was extubated; 20% of these patients required reintubation.²⁸ Ely and colleagues tested the safety and efficacy of a SBT as described by Esteban.¹ The study used a weaning screen to identify stability (eg, a series of clinical markers such as breathing rate, oxygenation, and hemodynamic status) followed by a 2-hour SBT if the criteria were met. The study took place in medical and coronary care units. The process was found to be both safe and effective. Positive outcomes included shorter ventilator duration (4.5 vs 6 days, P = .003) and fewer complications such as reintubation and need for tracheostomy in the intervention versus control groups (20% vs 41% respectively, P = .001). These studies also demonstrated the efficacy of using health care providers such as nurses and respiratory therapists to apply the protocols.^{1,2,5,29–31} The positive outcomes, plus the ease of use and efficiency of the SBT protocols, were attractive to clinicians. After publication of these studies, many hospitals sought similar outcomes by applying the protocols to practice.

But, despite gains in shortening ventilator duration in some patients by standardizing the process of weaning trials, other outcomes such as ICU and hospital LOS (as noted earlier) were still not optimal.^1,2,5 A potential aspect of care, specifically the use of sedation infusions, was forwarded as a potential culprit of delayed weaning in a study by Kollef and colleagues.² Although the primary goal of the study was to test the efficacy of a weaning method using a nurse/therapist driven weaning protocol, the authors noted a relationship between delayed weaning and sedation infusion use.² Subsequently Brook and colleagues performed an RCT, testing the efficacy of a nurse-driven sedation management algorithm compared to standard management.³ Their findings noted that the algorithmic approach resulted in a significantly shorter ICU LOS (5.7 vs 7.5 days respectively, P = .013) and hospital LOS (14 vs 19 days respectively, P<.001).³ To explore this association further, Kress and colleagues conducted an RCT to compare daily sedation interruptions to standard sedation management and found that the interruption method decreased duration of ventilation (4.9 vs 7.3 days respectively, P = .004) and ICU LOS (6.4 vs 9.9 days respectively, P = .02).⁴ In another study using the same database, the investigators found that patients in the “interruption” group experienced fewer complications as compared to the conventional sedation management group (2.8% vs 6.2% complications, P = .04).⁶

But all were not comfortable with the process of withdrawing sedative agents. Concerns emerged related to the potential psychological harm imposed by the abrupt withdrawal of the drugs. To that end, the authors sought to determine the effect of the sedation interruptions on the patients who had been enrolled in the daily sedation interruption study.³² Using a battery of psychological tests, the authors found that those who experienced the daily interruptions experienced either a significant decrease in psychological stress or equivalent stress (“Total Impact of Events Score” 11.2 vs 27.3, P = .02; “Incidence of PTSD” 0 vs 3.2%, P = .06; and “Psychosocial Adjustment to Illness Score” 46.8 vs 54.3, P = .08), to those who were cared for in a traditional manner (ie, slower, gradual sedation withdrawal).³² Finally, in another RCT referred to as the “wake up and breathe” trial, Girard and colleagues paired a sedation interruption with a SBT and found that the paired intervention resulted in increased ventilator-free days (group: 14.7 vs 11.6, P = .02), shorter time to discharge from the ICU and hospital (ICU days: 9.1 vs 12.9, P = .01; hospital days: 14.9 vs 19.2, P = .04), and better 1-year mortality (44% vs 58%) than those in the control group.⁷

Although the association between weaning outcomes and sedation was noted and addressed in the described studies, an additional concern related to outcomes of ventilated patients surrounds the relationship of sedative use (especially benzodiazepines) to delirium and long-term cognitive dysfunction.^33–37 In a study of 224 ventilated coronary care and medical intensive care patients Ely and colleagues found that up to 81% of the mechanically ventilated patients experienced delirium.^35,36 Patients with delirium experienced higher 6-month mortality rates (34% vs 15%, P = .03) and spent 10 days longer in the hospital than those who never developed delirium (P<.001). Because sedation is commonly used in critical care to assist with stabilization and management of ventilated patients, and is associated with many complications (eg, delayed weaning, longer ICU and hospital LOSs, and the development of delirium), it is clear that sedation management is a key aspect of care for ventilated patients and must be part of what we do.

Unfortunately, the use of sedation is complex. Studies exploring the use of the protocols and guidelines (these terms are often used interchangeably) report that adherence to the described protocols is very poor.^9–19 Although most of the studies on adherence to sedation interruption have been done using survey methodology, the results are compelling and suggest multiple reasons for the lack of adherence. As described in a review of the literature by the author of this article, reasons include caregiver perceptions (eg, that amnesia is good for patients, that it is unsafe to remove sedatives abruptly), the complexity associated with using protocols and guidelines (eg, multiple steps and the inherent “take-it-or-leave it” nature of guideline recommendations), and the philosophy and care practices of the environment in which they are implemented.³⁸ Examples of the latter include a series of studies conducted in Australian ICUs, where the nurses’ practices incorporated their philosophy that sedatives must be used sparingly.^39–41 When these nurses tested outcomes associated with their accepted sedation management practices as compared to using sedation guidelines or other related tools such as sedation scales, they found the use of the tools to be redundant, ineffective, and potentially harmful.^39–41 For example, in the study by Elliott and colleagues, the use of a sedation algorithm resulted in slower sedation removal and an increase in ventilator duration (4.8 days preintervention vs 5.6 days postintervention [NS]) and LOS (7.1 days preintervention vs 8.2 days postintervention (P = .04).³⁹ These authors also found that patients assigned to the algorithm were more sedated than those not assigned (NS).³⁹ Williams and colleagues found that the addition of sedation scales delayed sedation withdrawal, and Bucknall and colleagues noted that use of a sedation protocol made no difference in duration of ventilation or ICU and hospital LOS.^40,41

It may be that our lack of progress in attaining desired weaning outcomes is due to poor translation of research to practice; that protocols are not designed well for clinical use. And, if we are not adhering to the protocols, we cannot expect to have the same desired weaning outcomes as described in the studies. So although the use of protocols and guidelines for weaning trials and sedation management has the potential for improving outcomes of mechanical ventilation, we have a way to go to ensure that what we “think” we are doing is actually what we are doing. Other aspects of care may also contribute to the outcomes of interest and should be considered. They include clinical factors that positively or negatively affect weaning potential and timing of tracheostomy placement.