Automatic Control of Pressure Support for Ventilator Weaning in Surgical Intensive Care Patients

Transcription

1 Automatic Control of Pressure Support for Ventilator Weaning in Surgical Intensive Care Patients Dirk Schädler 1, Christoph Engel 2, Gunnar Elke 1, Sven Pulletz 1, Nils Haake 3, Inéz Frerichs 1, Günther Zick 1, Jens Scholz 1, and Norbert Weiler 1 1 Department of Anesthesiology and Intensive Care Medicine and 3 Department of Cardiovascular Surgery, University Medical Center Schleswig- Holstein, Campus Kiel, Kiel, Germany; and 2 Institute for Medical Informatics, Statistics and Epidemiology, University of Leipzig, Leipzig, Germany Rationale: Despite its ability to reduce overall ventilation time, protocol-guided weaning from mechanical ventilation is not routinely used in daily clinical practice. Clinical implementation of weaning protocols could be facilitated by integration of knowledge-based, closedloop controlled protocols into respirators. Objectives: To determine whether automatedweaning decreases overall ventilation time compared with weaning based on a standardized written protocol in an unselected surgical patient population. Methods: In this prospective controlled trial patients ventilated for longer than 9 hours were randomly allocated to receive either weaning with automatic control of pressure support ventilation (automatedweaning group) or weaning based on a standardized written protocol (control group) using the same ventilation mode. The primary end point of the study was overall ventilation time. Measurements and Main Results: Overall ventilation time (median [25th and 75th percentile]) did not significantly differ between the automated-weaning (31 [19 101] h; n ¼ 150) and control groups (39 [20 118] h; n ¼ 150; P ¼ 0.178). Patients who underwent cardiac surgery (n ¼ 132) exhibited significantly shorter overall ventilation times in the automated-weaning (24 [18 57] h) than in the control group (35 [20 93] h; P ¼ 0.035). The automated-weaning group exhibited shorter ventilation times until the first spontaneous breathing trial (1 [0 15] vs. 9 [1 51] h; P ¼ 0.001) and a trend toward fewer tracheostomies (17 vs. 28; P ¼ 0.075). Conclusions: Overall ventilation times did not significantly differ between weaning using automatic control of pressure support ventilation and weaning based on a standardized written protocol. Patients after cardiac surgery may benefit from automated weaning. Implementation of additional control variables besides the level of pressure support may further improve automated-weaning systems. Clinical trial registered with (NCT ). Keywords: positive pressure ventilation; respiratory therapy; weaning protocols; closed-loop control of mechanical ventilation; knowledgebased system (Received in original form June 27, 2011; accepted in final form January 8, 2012) Supported by a restricted research grant (70,000 ) from Dräger Medical GmbH, Lübeck, Germany. Author Contributions: D.S., I.F., and N.W. contributed to the conception and design of the study; D.S., N.H., G.Z., J.S., and N.W. contributed to the implementation and supervision of the study; D.S., G.E., and S.P. recruited patients and contributed to the acquisition of data; C.E. and D.S. performed the statistical analysis and verified its accuracy. All authors had full access to the data. All authors participated in analyzing and interpreting the data, and drafting and critically revising the manuscript. All authors have approved the final version of the manuscript. Correspondence and requests for reprints should be addressed to Dirk Schädler, M.D., Department of Anesthesiology and Intensive Care Medicine, University Medical Center Schleswig-Holstein, Campus Kiel, Arnold-Heller-Strasse 3, Haus 12, Kiel, Germany. schaedler@anaesthesie.uni-kiel.de This article has an online supplement, which is accessible from this issue s table of contents at Am J Respir Crit Care Med Vol 185, Iss. 6, pp , Mar 15, 2012 Copyright ª 2012 by the American Thoracic Society Originally Published in Press as DOI: /rccm OC on January 20, 2012 Internet address: AT A GLANCE COMMENTARY Scientific Knowledge on the Subject Protocols for ventilator weaning decrease ventilation time. We sought to determine whether the use of an automated pressure support weaning system could reduce total ventilation time in an unselected surgical patient population. What This Study Adds to the Field Overall ventilation time did not significantly differ between an automated-weaning system and a protocol-driven conventional weaning approach. Shorter overall ventilation times in patients who underwent cardiac surgery imply a benefit from automated weaning in this patient group. Initiation of mechanical ventilation is a life-saving measure in critical clinical situations when patients gas exchange needs to be secured. However, patients requiring prolonged mechanical ventilation exhibit higher morbidity, including ventilator-associated pneumonia (1 3), ventilator-induced lung injury (4), and ventilator-associated diaphragmatic dysfunction (5). Therefore, all weaning strategies aim at reducing the duration of mechanical ventilation. Protocolguided weaning has been shown to decrease overall ventilation time (6 16) and its use is recommended (17, 18). Transfer of protocols into routine clinical practice is not easy (19, 20). It may be facilitated by implementation of protocols into medical devices. This approach has already been realized for ventilator weaning by embedding an automated, knowledge-based weaning protocol into a commercially available ventilator (SmartCare/PS; Dräger Medical, Lübeck, Germany). This automated weaning system continuously adjusts the pressure support (PS) level depending on the patient s needs. The goal is to transfer and keep the patient in a zone of respiratory comfort (21). In this zone, spontaneous breathing rate is in the range of breaths per minute, tidal volume is more than 300 ml, and partial pressure of carbon dioxide in the expired air is less than 55 mm Hg. When a specific target PS level is reached (depending on the type of artificial airway and the way of humidification) and only minimum positive endexpiratory pressure (PEEP) is needed, a spontaneous breathing trial is automatically conducted. The readiness for extubation is indicated when the patient passes this trial successfully. A multicenter randomized controlled trial compared this automated-weaning system with usual care in 144 patients and revealed shorter overall ventilation time and length of intensive care unit (ICU) stay in the automatically weaned patients (22). The drawbacks of this study were the low enrollment rate (14% of eligible patients) and the heterogeneity of the control group because four out of five study sites used locally different weaning protocols and one site did not use any protocol. A more recent randomized controlled trial in 102 patients with an enrollment rate of 10% found comparable overall ventilation

2 638 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL times during automated weaning and usual care (23). This finding may be related to the high intensivist and ICU nurse staffing and to the high levels of PEEP used during weaning. In contrast to these previous studies, we intended to compare automated, knowledge-based weaning with weaning using a standardized written protocol in a large, unselected, postsurgical study population. Some of the results of the study have been previously reported in the form of abstracts (24, 25). METHODS This randomized, open-label study was conducted in three ICUs of the University Medical Center Schleswig-Holstein, Campus Kiel, Kiel, Germany. The study was approved by the local ethics committee and registered at clinicaltrials.gov (NCT ). Written informed consent was obtained from the patients or their legal representatives before or during the first 36 hours after study inclusion and randomization using the procedure of deferred consent. Study Protocol A detailed description of the study protocol is provided in the online supplement. Briefly, all patients who, at 9:00 AM, were mechanically ventilated for longer than 9 hours after their admission in the ICU were eligible for study inclusion. The following exclusion criteria were used: cerebral surgery or trauma, age less than 18 years, do-not-resuscitate order, duration of mechanical ventilation greater than 24 hours, and patient already participating in the study. All patients were ventilated with Evita XL respirators (Dräger Medical, Lübeck, Germany) equipped with the automated weaning system SmartCare/PS version 1.1. The night rest option was switched to off. Heat and moisture exchange filters were always used. Analgesia was maintained by a continuous infusion of sufentanile within the range of mg/kg/h according to the patient s requirements. Sedation was achieved by a continuous infusion of propofol (maximum, 4 mg/kg/h) during 24 hours after inclusion. Thereafter, boluses of midazolam were administered to reach the target Ramsay Score of R2 (26). Patients were randomly assigned to the automated-weaning group or the control group. Randomization with an allocation ratio of 1:1 was concealed, not stratified, and performed by an electronically generated randomization plan. Assignment to one of the study arms was disclosed to the investigator by the patient number. The following subgroups were a priori defined: cardiac surgery (patients who underwent open-heart surgery); sepsis (patients meeting criteria for severe sepsis or septic shock [27]); and chronic obstructive pulmonary disease (COPD; patients with established COPD). Primary end point of the study was overall ventilation time during ICU stay considering any time a patient needed invasive or noninvasive ventilatory support during the 28-day study period. Secondary end points were time in the zone of respiratory comfort, numbers of ventilator manipulations and alarms during invasive mechanical ventilation, length of ICU and hospital stay, and 28- and 90-day mortality. Statistical Analysis Based on the data by Lellouche and coworkers (22) who found a median difference of 2 days in the time until the first successful extubation between the automated-weaning and control groups, a sample size of 133 patients in each group had a 90% power to detect a difference between means of ventilation time of 2 days with a significance level of 0.05 (twotailed) and an expected standard deviation of 5 days. We estimated 15 drop-outs per group and decided to include 150 patients per group. Data analysis was performed on an intention-to-treat basis. Results are presented as medians with interquartile ranges or as means with standard deviations where appropriate. The Mann-Whitney U test was used to determine whether the overall ventilation time was significantly different between the two study arms. The median difference of overall ventilation time between the two study arms was calculated using the Hodges- Lehmann estimator (28). This method provides a robust nonparametric Figure 1. Flowchart of patients in the Automatic Control of Pressure Support Ventilation in Surgical Intensive Care Units (ASOPI) trial.

3 Schädler, Engel, Elke, et al.: Automatic Control of Pressure Support Ventilation 639 estimator for the difference between two independent groups, in which the distributions of values are highly skewed. Proportions were compared using chi-square test or Fisher exact test when required. The cumulative probability of remaining on mechanical ventilation was assessed using the Kaplan-Meier method and compared between study arms using the log-rank test. To analyze the independent association of study arm, Acute Physiology and Chronic Health Evaluation (APACHE) II score, and subgroups on overall ventilation time, a multivariate Cox proportional hazards regression analysis was conducted. Statistical significance was accepted at P values less than or equal to All statistical analyses were performed with PASW Statistics 18 (SPSS, Munich, Germany). RESULTS Patients A total of 501 patients fulfilled the inclusion criterion and were considered eligible for enrollment between September 2006 and May The trial ended by plan after enrollment of 317 patients; 17 patients did not consent to participate. The remaining 300 patients (150 patients in each group) entered the analysis. A detailed flowchart of the study is given in Figure 1. Baseline characteristics of the included patients are shown in Tables 1 and 2 (see Tables E1 and E2 in the online supplement). There were no significant differences between the groups at inclusion except for a higher peak inspiratory pressure in the control group. regarding overall ventilation time (31 [19 101] vs. 39 [20 118] h; P ¼ 0.178). The estimated median difference of overall ventilation time between groups was 23.5 h (95% confidence interval [CI], to 1.6). The proportion of time in the comfort zone (23% vs. 25%; P ¼ 0.971), number of all ventilator manipulations per hour (6.3 [4.1 10] vs. 5.6 [ ]; P ¼ 0.141), number of alarms per hour (2.7 [ ] vs. 2.3 [1 7.1]; P ¼ 0.392), length of ICU or hospital stay, and 28- and 90-day mortality did not differ significantly between the two groups. In the subgroup cardiac surgery, overall ventilation time was significantly shorter in the automated-weaning than in the control group (24 [18 57] vs. 35 [20 93] h; P ¼ 0.035). In this subgroup, the estimated median difference of the overall ventilation time between groups was 26.4 h (95% CI, to 20.5). No significant differences were found in the subgroups sepsis (101 [47 234] vs. 144 [28 504] h; P ¼ 0.296) and COPD (55 [24 167] vs. 28 [20 69] h; P ¼ 0.095). The subgroup cardiac surgery was an independent significant predictor of shorter ventilation time (hazard ratio, 1.36; 95% CI, ), whereas the subgroup sepsis (hazard ratio, 0.65; 95% CI, ) and APACHE II score (hazard ratio, 0.97; 95% CI, ) were independent significant predictors of longer ventilation time (Figure 3). Study group (hazard ratio, 0.83; 95% CI, ) and subgroup COPD (hazard ratio, 0.986; 95% CI, ) were not significantly associated with overall ventilation time. End Points The main outcome results are given in Tables 3 and 4 (see Tables E3 and E4). Kaplan-Meier curves of overall ventilation time are shown in Figure 2. We found no statistically significant differences between the automated-weaning and the control groups Complications Complications associated with mechanical ventilation are shown in Table 5. We found a trend toward fewer tracheostomies in the automated-weaning group (n ¼ 17; 11.3%) than in the control TABLE 1. BASELINE CHARACTERISTICS OF THE STUDY PATIENTS Variable Automated Weaning (n ¼ 150) Control (n ¼ 150) P Value Age, yr 67 (55 75) 69 (60 74) Sex, male/female 103/47 106/ APACHE II score* 16 (12 19) 16 (11 19) Arterial partial pressure of oxygen, mm Hg 115 (97 140) 116 (91 136) Arterial partial pressure of carbon dioxide, mm Hg 41 (37 47) 41 (37 47) Ramsay score 3 (2 4) 3 (2 4) Preventilation time, h 15 (11 18) 16 (12 18) Comorbidities, number (%) Acute lung injury 14 (9.4) 11 (7.3) Acute respiratory distress syndrome 7 (4.7) 12 (8) ICU, number (%) Interdisciplinary ICU 52 (35.7) 47 (31.3) Surgical ICU 27 (18) 27 (18) Cardiovascular ICU 71 (47.3) 76 (50.7) Admission type, number (%) Postsurgical after elective surgery 65 (43.3) 65 (43.3) Postsurgical after emergency surgery 55 (36.7) 54 (36) Emergency, presurgical 16 (10.7) 14 (9.3) Planned, presurgical 1 (0.7) 2 (1.3) Emergency, postsurgical 9 (6) 12 (8) Nonsurgery 4 (2.7) 3 (2) Ventilator settings Tidal volume, ml/kg predicted body weight 9.6 ( ) 9.6 (8 11) Positive end-expiratory pressure, cm H 2 O 8 (5 10) 8 (6 10) Peak inspiratory pressure, cm H 2 O 18 (15 24) 21 (17 25) Inspired fraction of oxygen 0.41 ( ) 0.41 ( ) Ventilator mode at study inclusion, number (%) Pressure support ventilation 92 (61.3) 84 (56) Biphasic positive airway pressure 57 (38) 62 (41.3) Stand-by 1 (0) 4 (2.7) Definition of abbreviations: APACHE ¼ Acute Physiology and Chronic Health Evaluation; ICU ¼ intensive care unit. *Missing subscores on APACHE II score were counted as 0. y Stand-by was used in patients who were temporarily not ventilated with their study ventilators for transportation reasons.

4 640 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 2. BASELINE CHARACTERISTICS OF THE STUDY PATIENTS IN THE SUBGROUP CARDIAC SURGERY Variable Automated Weaning (n ¼ 62) Control (n ¼ 70) P Value Age, yr 69 (63 77) 71 (66 76) Sex, male/female 49/13 53/ APACHE II score* 16 (12 18) 15 (11 18) Arterial partial pressure of oxygen, mm Hg 124 ( ) 121 ( ) Arterial partial pressure of carbon dioxide, mm Hg 41 (39 46) 41 (36 47) Ramsay score 2 (2 3) 3 (2 4) Preventilation time, h 15 (12 18) 16 (14 18) Comorbidities, number (%) Acute lung injury 1 (1.6) 3 (4.3) Acute respiratory distress syndrome 0 (0) 2 (2.9) ICU, number (%) Interdisciplinary ICU 2 (3.2) 2 (2.9) Surgical ICU 0 (0) 0 (0) Cardiovascular ICU 60 (96.8) 68 (97.1) Admission type, number (%) Postsurgical after elective surgery 45 (72.6) 45 (64.3) Postsurgical after emergency surgery 14 (22.6) 15 (21.4) Emergency, presurgical 3 (4.8) 7 (10.0) Planned, presurgical 0 (0) 0 (0) Emergency, postsurgical 0 (0) 3 (4.3) Nonsurgery 0 (0) 0 (0) Ventilator settings Tidal volume, ml/kg predicted body weight 10.1 ( ) 9.8 (8.6 11) Positive end-expiratory pressure, cm H 2 O 9 (5 10) 10 (7 10) Peak inspiratory pressure, cm H 2 O 18 (15 23) 22 (18 25) Inspired fraction of oxygen 0.40 ( ) 0.41 ( ) Ventilator mode at study inclusion, number (%) Pressure support ventilation 44 (71) 34 (48.6) Biphasic positive airway pressure 18 (29) 34 (48.6) Stand-by 0 (0) 2 (2.9) Definition of abbreviations: APACHE ¼ Acute Physiology and Chronic Health Evaluation; ICU ¼ intensive care unit. *Missing subscores on APACHE II score were counted as 0. y Stand-by was used in patients who were temporarily not ventilated with their study ventilators for transportation reasons. group (n ¼ 28; 18.7%; P ¼ 0.075). The number of all other complications (need for noninvasive ventilation, self-extubation, number and reasons for reintubation, and new onsets of sepsis and acute respiratory distress syndrome during the study) did not significantly differ between the groups. Analgesics and Sedatives Both groups exhibited similar levels of sedation (Ramsay score automated-weaning group, 2.5 [2 3]; Ramsay score control group, 2.6 [2 3]; P ¼ 0.150). The required doses of analgesic and sedative agents did not differ significantly between the groups (see Table E5). Mechanical Ventilation Data A detailed analysis is presented in Table E6. Changes in the PS level per day occurred significantly more often in the automatedweaning than in the control group (41 [17 96] vs. 5 [0 12]; P, 0.001). The major cause of not being in the respiratory comfort zone was a spontaneous breathing frequency less than 15 breaths per minute in both groups. Overall and in the subgroup cardiac surgery, we found a significantly shorter time period TABLE 3. COMPARISON OF OUTCOME DATA BETWEEN THE AUTOMATED-WEANING AND CONTROL GROUPS Variable Automated Weaning (n ¼ 150) Control (n ¼ 150) P Value Ventilation time until first 1 (0 15) 9 (1 51) spontaneous breathing trial, h* Ventilation time until first successful 3 (1 34) 9 (1 51) spontaneous breathing trial, h* Ventilation time until first 8 (3 29) 10 (3 49) extubation, h* Ventilation time until 10 (3 60) 17 (3 85) successful extubation, h* Overall ventilation time, h 31 (19 101) 39 (20 118) Intensive care length of stay, d 3.8 ( ) 3.4 ( ) Hospital length of stay, d 18.5 ( ) 19.6 ( ) d mortality, number (%) 29 (19.3) 24 (16) d mortality, number (%) 36 (24) 32 (21.3) Values are given as median (interquartile range) unless otherwise noted. *Ventilation times were calculated from the time point of study inclusion until the defined event. y Overall ventilation time was the sum of preventilation time and ventilation time in the intensive care unit during the 28-day study period.

5 Schädler, Engel, Elke, et al.: Automatic Control of Pressure Support Ventilation 641 TABLE 4. COMPARISON OF OUTCOME DATA BETWEEN THE AUTOMATED-WEANING AND CONTROL GROUPS IN THE SUBGROUP CARDIAC SURGERY Variable Automated Weaning (n ¼ 62) Control (n ¼ 70) P Value Ventilation time until first 0.5 (0 3) 7 (1 27) spontaneous breathing trial, h* Ventilation time until first 2 (1 9) 7 (1 27) successful spontaneous breathing trial, h* Ventilation time until 6 (3 24) 10 (4 25) first extubation, h* Ventilation time until 7 (3 28) 13 (4 29) successful extubation, h* Overall ventilation time, h 24 (18 57) 35 (20 93) Intensive care length of stay, d 2.1 ( ) 2.4 ( ) Hospital length of stay, d 12.6 ( ) 14.3 ( ) d mortality, number (%) 6 (9.7) 6 (8.6) d mortality, number (%) 7 (11.3) 7 (10) Values are given as median (interquartile range) unless otherwise noted. *Ventilation times were calculated from the time point of study inclusion until the defined event. y Overall ventilation time was the sum of preventilation time and ventilation time in the intensive care unit during the 28-day study period. before the initiation of the first spontaneous breathing trial in the automated-weaning group. Protocol Violations Protocol violations for the whole study population are summarized in Table E7 and for the subgroup cardiac surgery in Table E8. Automated weaning was used 75% of the time it had to be used according to the study protocol. DISCUSSION In this study, overall ventilation time did not significantly differ between an automated-weaning system and a protocol-driven conventional weaning approach. Automated weaning significantly decreased overall ventilation time in patients who underwent cardiac surgery. A trend toward a lower number of tracheostomies was noted in the automated-weaning group. The PS level was changed more often in the automated-weaning group; however, Figure 2. Kaplan-Meier curves for overall ventilation time. Comparison of the overall ventilation time between the automated-weaning and the control groups in the (A) whole study population and in the subgroups of (B) cardiac surgery, (C) sepsis, and (D) chronic obstructive pulmonary disease (COPD). The respective population and subgroup sizes and P values are provided in each panel.

6 642 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL Figure 3. Hazard ratios for overall ventilation time. Hazard ratios (squares) with 95% confidence intervals (solid lines) of five predictor variables for overall ventilation time were estimated using the Cox proportional hazards model. A hazard ratio of more than 1 is predictive for shorter overall ventilation time and a hazard ratio of less than 1 is predictive for longer overall ventilation time. APACHE ¼ Acute Physiology and Chronic Health Evaluation; COPD ¼ chronic obstructive pulmonary disease. this did not increase the time proportion in the respiratory comfort zone. Our study is the largest randomized trial investigating automated weaning in postsurgical patients. Our enrollment rate (63%) was much higher than in the studies of Lellouche and coworkers (22) (14%) and Rose and coworkers (23) (10%), which may be in part explained by a different type of study consent. We could use deferred consent, whereas the other two studies had to obtain informed consent from the next of kin before including patients into the study. Our inclusion and exclusion criteria differed from those used previously because we wanted to test the performance of automated weaning in a large, unselected patient population. Therefore, we excluded (1) the typical postsurgical easy-to-wean patients in whom the weaning period is short and the probable effect of automated weaning small, (2) patients who were ventilated for longer than 24 hours because the probable early effect of automated weaning would not be revealed, and (3) patientswith neurosurgical trauma or surgery who could have confounded the results. Thanks to this relatively simple study design we could map our clinical routine and achieve the high enrollment rate. This study was designed as a superiority study. We hypothesized that automated weaning could decrease overall ventilation time (primary outcome variable) compared with protocol-guided weaning in a large postsurgical patient population. In the whole study population, overall ventilation times did not differ significantly between the automated-weaning and the control groups. Although this study was not designed to show equivalence of the two weaning strategies, the estimated median difference of overall ventilation time and its narrow CI suggest that very likely no clinically relevant differences existed. However, the equivalence remains to be confirmed in a study specifically designed to show this. The ICU staffing and use of weaning protocols influence the outcome of weaning trials. For instance, the quality of nursing is directly associated with the weaning duration (29). These factors have seemingly played a role in the two previous studies using automated weaning. In the study by Rose and coworkers (23) the ICU was well staffed (1:1 nurse-to-patient ratio, nine intensivists, and twice-daily structured ICU rounds), and this may in part explain the comparable overall ventilation times in patients weaned using the automated-weaning system and those weaned by usual care despite an absent formal weaning protocol in the control group. Lellouche and coworkers (22) did not comment on the ICU staffing but the longer overall ventilation times found in the control group could have been affected by the local care differing among the five study centers. In our study, the ICU staffing was lower than in the study by Rose and coworkers (23) and a standardized written weaning protocol was consistently used in the control group. In the subgroup COPD we found a trend to a higher overall ventilation time (P ¼ 0.095). This finding has to be interpreted with caution because of the small number of patients in this subgroup and a high 75th percentile in the automated-weaning group. Therefore, we consider these results to be potentially affected by only a small number of patients. The automated-weaning system used in our trial aimed at transferring and keeping the patients stable in the comfort zone by automatic adjustment of the PS level. A small, crossover study on 10 patients using an older version of the system revealed a higher time proportion with acceptable ventilation during TABLE 5. COMPLICATIONS OF MECHANICAL VENTILATION Complication Automated Weaning (n ¼ 150) Control (n ¼ 150) P Value Need for noninvasive ventilation 16 (10.7) 18 (12) Noninvasive ventilation time, h* Self-extubation 4 (2.7) 1 (0.7) Tracheostomy (one or more) 17 (11.3) 28 (18.7) New onset of sepsis during study 3 (2) 5 (3.3) New onset of ARDS during study 3 (2) 6 (4) Reintubation (76.7) 110 (73.3) 1 23 (15.3) 21 (14) 2 or more 12 (8) 19 (12.7) Reason for reintubation Surgery 33 (51.6) 36 (48.6) Respiratory failure 24 (37.5) 29 (39.2) Inability to protect airway 0 (0) 3 (4.1) Other 3 (4.7) 4 (5.4) Hemodynamic instability 4 (6.3) 2 (2.7) Prolonged mechanical ventilation Ventilation time.96 h 39 (26) 46 (30.7) Ventilation time.14 d 13 (8.7) 17 (11.3) Ventilation time.21 d 7 (4.7) 8 (5.3) Definition of abbreviation: ARDS ¼ acute respiratory distress syndrome. Values are given as number (percent). *Mean values 6 standard deviation.

7 Schädler, Engel, Elke, et al.: Automatic Control of Pressure Support Ventilation 643 automated weaning than during conventional PS ventilation (93 6 8% vs %; P ¼ 0.003) (30). This was accomplished with considerably more frequent changes in the PS level ( vs ). Although we also found a higher number of changes in the PS level in the automated-weaning group, the time proportions our patients spent in the comfort zone were not different between the groups. This may be related to the slightly different criteria for the definition of the comfort zone between the two studies. Another range of spontaneous breathing rates (12 28 breaths per min) was used in the older than in the newer version of the automated-weaning system (15 30 breaths per min). The current lower limit of this range may be too high for postsurgical patients. Further explanations could be that Dojat and coworkers (30) studied patients ventilated long-term in their study (mean ventilation time, d). The automated-weaning system used controlled only the PS level; PEEP, fraction of inspired oxygen, and the pressure slope were set manually, which may have affected the weaning duration. The message indicating readiness for extubation was displayed for the first time after the patient passed the spontaneous breathing trial at a PEEP level of less than 6 cm H 2 O. Apparently, the weaning time could be artificially prolonged by using too high PEEP or by decreasing PEEP too slowly. This could in part explain the negative results in the study by Rose and coworkers (23) because the median maximum PEEP setting during the weaning process was 10 cm H 2 O in their automatedweaning group. We tried to minimize this effect by attempted PEEP reduction three times a day in both groups. Our data show that the number of PEEP changes per study day did not differ significantly between the two studied groups. This study has limitations. First, this was not a multicenter trial, although the study was performed in three ICUs served by different departments. Second, masking was not feasible because of different interactions needed between ventilator and user in the two study groups. Third, nine patients in the automatedweaning and eight patients in the control group dropped out of the study. Fourth, the study was conducted in three European ICUs with physicians and nurses managing ventilation therapy. Thus, the results may not be transferable to ICUs where respiratory therapists manage ventilation therapy. Conclusions In this study, overall ventilation times did not significantly differ between weaning using automatic control of PS ventilation and weaning based on a standardized written protocol. Patients who underwent cardiac surgery may benefit from this type of automated weaning. This finding has to be investigated in further studies. Further development of automated, knowledge-based weaning (e.g., by implementation of additional control variables) may optimize its performance. Author disclosures are available with the text of this article at Acknowledgment: The authors are indebted to all physicians and nurses from the three study intensive care units for their help and support during the entire study period. References 1. Cook DJ, Walter SD, Cook RJ, Griffith LE, Guyatt GH, Leasa D, Jaeschke RZ, Brun-Buisson C. Incidence of and risk factors for ventilator-associated pneumonia in critically ill patients. AnnInternMed1998;129: Papazian L, Bregeon F, Thirion X, Gregoire R, Saux P, Denis JP, Perin G, Charrel J, Dumon JF, Affray JP, et al. Effect of ventilatorassociated pneumonia on mortality and morbidity. Am J Respir Crit Care Med 1996;154: Vincent JL, Bihari DJ, Suter PM, Bruining HA, White J, Nicolas- Chanoin MH, Wolff M, Spencer RC, Hemmer M. The prevalence of nosocomial infection in intensive care units in Europe. Results of the European Prevalence of Infection in Intensive Care (EPIC) study. Epic International Advisory Committee. JAMA 1995;274: Gattinoni L, Protti A, Caironi P, Carlesso E. Ventilator-induced lung injury: the anatomical and physiological framework. Crit Care Med 2010;38:S539 S Levine S, Nguyen T, Taylor N, Friscia ME, Budak MT, Rothenberg P, Zhu J, Sachdeva R, Sonnad S, Kaiser LR, et al. Rapid disuse atrophy of diaphragm fibers in mechanically ventilated humans. N Engl J Med 2008;358: Ely EW, Baker AM, Dunagan DP, Burke HL, Smith AC, Kelly PT, Johnson MM, Browder RW, Bowton DL, Haponik EF. Effect on the duration of mechanical ventilation of identifying patients capable of breathing spontaneously. N Engl J Med 1996;335: GrapMJ,StricklandD,TormeyL,KeaneK,LubinS,EmersonJ,Winfield S, Dalby P, Townes R, Sessler CN. Collaborative practice: development, implementation, and evaluation of a weaning protocol for patients receiving mechanical ventilation. Am J Crit Care 2003;12: Henneman E, Dracup K, Ganz T, Molayeme O, Cooper C. Effect of a collaborative weaning plan on patient outcome in the critical care setting. Crit Care Med 2001;29: Horst HM, Mouro D, Hall-Jenssens RA, Pamukov N. Decrease in ventilation time with a standardized weaning process. Arch Surg 1998; 133: , discussion Kollef MH, Shapiro SD, Silver P, St John RE, Prentice D, Sauer S, Ahrens TS, Shannon W, Baker-Clinkscale D. A randomized, controlled trial of protocol-directed versus physician-directed weaning from mechanical ventilation. Crit Care Med 1997;25: Marelich GP, Murin S, Battistella F, Inciardi J, Vierra T, Roby M. Protocol weaning of mechanical ventilation in medical and surgical patients by respiratory care practitioners and nurses: effect on weaning time and incidence of ventilator-associated pneumonia. Chest 2000;118: Saura P, Blanch L, Mestre J, Valles J, Artigas A, Fernandez R. Clinical consequences of the implementation of a weaning protocol. Intensive Care Med 1996;22: Scheinhorn DJ, Chao DC, Stearn-Hassenpflug M, Wallace WA. Outcomes in post-icu mechanical ventilation: a therapist-implemented weaning protocol. Chest 2001;119: Smyrnios NA, Connolly A, Wilson MM, Curley FJ, French CT, Heard SO, Irwin RS. Effects of a multifaceted, multidisciplinary, hospitalwide quality improvement program on weaning from mechanical ventilation. Crit Care Med 2002;30: Tonnelier JM, Prat G, Le Gal G, Gut-Gobert C, Renault A, Boles JM, L Her E. Impact of a nurses protocol-directed weaning procedure on outcomes in patients undergoing mechanical ventilation for longer than 48 hours: a prospective cohort study with a matched historical control group. Crit Care 2005;9:R83 R Vitacca M, Vianello A, Colombo D, Clini E, Porta R, Bianchi L, Arcaro G, Vitale G, Guffanti E, Lo Coco A, et al. Comparison of two methods for weaning patients with chronic obstructive pulmonary disease requiring mechanical ventilation for more than 15 days. Am J Respir Crit Care Med 2001;164: Boles JM, Bion J, Connors A, Herridge M, Marsh B, Melot C, Pearl R, Silverman H, Stanchina M, Vieillard-Baron A, et al. Weaning from mechanical ventilation. Eur Respir J 2007;29: MacIntyre NR, Cook DJ, Ely EW Jr, Epstein SK, Fink JB, Heffner JE, Hess D, Hubmayer RD, Scheinhorn DJ. Evidence-based guidelines for weaning and discontinuing ventilatory support: a collective task force facilitated by the American College of Chest Physicians; the American Association for Respiratory Care; and the American College of Critical Care Medicine. Chest 2001;120:375S 395S. 19. Pronovost PJ, Rinke ML, Emery K, Dennison C, Blackledge C, Berenholtz SM. Interventions to reduce mortality among patients treated in intensive care units. J Crit Care 2004;19: Weinert CR, Gross CR, Marinelli WA. Impact of randomized trial results on acute lung injury ventilator therapy in teaching hospitals. Am J Respir Crit Care Med 2003;167: Dojat M, Brochard L, Lemaire F, Harf A. A knowledge-based system for assisted ventilation of patients in intensive care units. Int J Clin Monit Comput 1992;9: Lellouche F, Mancebo J, Jolliet P, Roeseler J, Schortgen F, Dojat M, Cabello B, Bouadma L, Rodriguez P, Maggiore S, et al. Amulticenter

8 644 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL randomized trial of computer-driven protocolized weaning from mechanical ventilation. Am J Respir Crit Care Med 2006;174: Rose L, Presneill JJ, Johnston L, Cade JF. A randomised, controlled trial of conventional versus automated weaning from mechanical ventilation using SmartCare/PS. Intensive Care Med 2008;34: Schädler D, Elke G, Pulletz S, Haake N, Frerichs I, Scholz J, Zick G, Weiler N. The effect of automatic weaning with SmartCare/PS on ventilation time in postsurgical patients: a randomized controlled trial. Am J Respir Crit Care Med 2009;179:A Schädler D, Elke G, Pulletz S, Haake N, Frerichs I, Zick G, Scholz J, Weiler N. The influence of an automatic control of pressure support ventilation on total ventilation time: preliminary results of a randomized controlled trial. Intensive Care Med 2008;34:S Ramsay MA, Savege TM, Simpson BR, Goodwin R. Controlled sedation with alphaxalone-alphadolone. BMJ 1974;2: American College of Chest Physicians/Society of Critical Care Medicine consensus conference: definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992;20: Hodges JL, Lehmann EL. Estimates of location based on rank rest. Ann Math Stat 1963;34: Thorens JB, Kaelin RM, Jolliet P, Chevrolet JC. Influence of the quality of nursing on the duration of weaning from mechanical ventilation in patients with chronic obstructive pulmonary disease. Crit Care Med 1995;23: Dojat M, Harf A, Touchard D, Lemaire F, Brochard L. Clinical evaluation of a computer-controlled pressure support mode. Am J Respir Crit Care Med 2000;161: