ICU-acquired delirium and weakness can lead to devastating cognitive and physical impairments and psychiatric symptoms in ICU survivors, also known as “post-intensive care syndrome” (1–6). The (A)wakening and (B)reathing, (C)oordination, (D)elirium monitoring and management, and (E)arly mobilization (ABCDE) bundle (7, 8) is an interdisciplinary patient-centered evidence-based strategy endorsed by critical care societies and national quality improvement agencies to prevent and reduce ICU delirium and weakness, and operationalize the Society of Critical Care Medicine’s Pain, Agitation, and Delirium clinical practice guidelines (9–13).
Individual components of the ABCDE bundle are associated with substantial benefits in research settings (14–20). Although studies in clinical practice settings suggest that implementation of the full ABCDE bundle is associated with clinical benefits, its uptake has been limited and implementation often-incomplete (21–28). Sequential implementation of bundle components may improve overall execution by allowing providers to: 1) maximize efficacy of implementation by focusing on individual components, 2) assess process improvement by performing stepwise evaluation of components, and 3) make practice adjustments before moving to the next component. In addition, studies suggest that the efficacy of early mobilization can be maximized if programs to reduce unnecessary sedation and delirium are already in place (25, 29).
Accordingly, we sought to determine the impact of adding EC to B-AD in the context of staged implementation of the ABCDE bundle in mechanically ventilated (MV) patients. We hypothesized that implementation of early mobilization on a foundation of targeted sedation practices and routine delirium monitoring would improve clinical outcomes and reduce hospital cost. Preliminary results have been presented in abstract form (30, 31).
MATERIALS AND METHODS
See Supplemental Appendix (Supplemental Digital Content 1, http://links.lww.com/CCM/E506) for a more detailed description of study procedures.
Study Design and Setting
This prospective study took place in two academic medical ICUs at Montefiore Medical Center (Bronx, NY). ICUs had the same size (14 beds) and staffing (two patients per nurse, 24 hr onsite intensivist coverage), except the full bundle ICU was staffed by medical residents and the partial bundle ICU by physician assistants. The Institutional Review Board (IRB) approved a waiver of informed consent (IRB number 2014–3466).
Our primary cohort consisted of all MV adults (≥ 18 yr) admitted to the ICUs for greater than or equal to 24 hours between July 1, 2011, and June 30, 2014 (Fig. 1). This cohort was used for analyses of clinical outcomes; alternative cohorts were used for process of care and cost outcomes (Fig. S1 and text in Supplemental Appendix, Supplemental Digital Content 1, http://links.lww.com/CCM/E506).
Interdisciplinary teams of critical care nursing, physician, pharmacy, respiratory therapy, and rehabilitation leadership and champions developed and implemented bundle components.
(A)wakening and (D)elirium Monitoring/Management (AD) (Both ICUs). At baseline, both
ICUs used MV order sets that included daily sedation vacations and spontaneous (B)reathing trials (B) (Fig. 1); however, no guidance was given on performance or coordination of these bundles. Beginning in January 2012, the (A)wakening from sedation and (D)elirium monitoring/management (AD) bundles were implemented in both ICUs; this included physician-directed targeted sedation using the Richmond Agitation and Sedation Scale (32, 33), twice daily3233 delirium assessments using the Confusion Assessment Method-ICU by nurses (32, 34), and suggestions for nonpharmacologic delirium reduction methods. To account for time to adopt these changes, AD bundles were considered fully implemented by July 1, 2012.
(E)arly Mobilization and (C)oordination of Components (EC): (Full Bundle ICU Only).
(E)arly mobilization (E) consisted of evaluation by physical therapy (PT) and occupational therapy (OT) at ICU admission, and daily rehabilitation by PT and/or OT according to a staged protocol in which patients advanced from passive range of motion to independent ambulation with respiratory therapy and nursing assistance as needed (17, 35) (Fig. S3, Supplemental Digital Content 1, http://links.lww.com/CCM/E506). As part of this bundle, daily structured interdisciplinary rounds were established for ICU nurses, respiratory therapists, and rehabilitation staff to (C)oordinate bundle components (C), diagnostic tests and procedures. On July 1, 2013, EC were implemented in the full bundle ICU only because of resource and staffing limitations.
Clinical data were extracted from electronic medical records using healthcare surveillance software (Clinical Looking Glass; Emerging Health Information Technology, Yonkers, NY ). To determine if practices changed after ICU-wide implementation of bundle components, we also examined process of care data (Fig. 1; and Fig. S1 and text in Supplemental Appendix, Supplemental Digital Content 1, http://links.lww.com/CCM/E506).
The primary outcome of interest was the hospital length of stay (LOS) after the index ICU admission (i.e., ICU LOS + post-ICU LOS). Secondary outcomes included ICU LOS, duration of MV, hospital mortality, and discharge location.
Total hospital and ICU cost and average daily ICU cost (i.e., total cost divided by ICU LOS) were determined using cost-to-charge ratios at Montefiore Medical Center. Because cost-to-charge ratios differ by calendar year, the cohort in the cost analyses was limited to patients with hospitalizations that ended between January 1, 2012, and December 31, 2013. Costs were calculated as the sum of daily direct variable costs from cost centers related to inpatient, nonoperative care (e.g., respiratory support, room and board, laboratory, medications) as previously described (37).
Clinical Quality Outcomes.
Clinical quality metrics that may be affected by implementation of the ABCDE bundle (e.g., ICU restraint use, prevalence of ICU-acquired pressure ulcers) were obtained from aggregate hospital-reported data for Centers for Medicare and Medicaid Services quality indicators in both full and partial bundle ICUs. Data were only available for periods 1 and 2.
Patient characteristics and unadjusted clinical outcomes were compared across ICUs and time periods using standard descriptive statistics. Nonparametric tests were used for skewed continuous measures.
To evaluate the impact of EC on clinical and cost outcomes, we compared trends in these outcomes in the full versus partial bundle ICUs before and after EC implementation using a multivariable difference-in-differences (DiD) approach (38, 39). This methodology uses a multivariable regression model that includes an interaction term for “time period” (e.g., period 1 vs 2) and “ICU” (full vs partial bundle) that measures the magnitude of the effect of EC (Fig. 1). In contrast to standard before-after studies, DiD controls for temporal trends in patient characteristics (e.g., increasing severity of illness) that might impact outcomes. DiD analyses are based on four assumptions to ensure validity of the model, the most important of which is the parallel trend assumption (i.e., prior to interventions, temporal changes in outcomes for both ICUs are similar) (38). To test this assumption, separate regression models were constructed for each outcome in the baseline period (for B vs B-AD analysis) and period 1 (for B-AD vs B-AD-EC analysis); models included interaction terms for ICU admission date and ICU. We also performed sensitivity analyses to determine if including patients with hospital LOS greater than 90 would alter our estimates. Although AD was implemented in both units, we used DiD (baseline vs period 1) to evaluate for differential impact of AD implementation on clinical outcomes between ICUs. All models were adjusted for patient-level characteristics that differed between ICUs (univariable p ≤ 0.2). Because Acute Physiology and Chronic Health Evaluation (APACHE) IV scores were missing in 10% of patients, we used dummy variable adjustment (40).
All tests were two-tailed and p value of less than 0.05 defined statistical significance. Analyses were performed with STATA/MP 13 (Statacorp, College Station, TX).
Between July 1, 2011, and June 30, 2014, 1,855 MV patients were admitted to the full (1,036, 56%) and partial bundle (819, 44%) ICUs. The full bundle ICU had younger patients and more minorities (Table 1). Patients in the full bundle ICU also had more comorbidities, higher severity of illness, and fewer lived at home prior to hospitalization. Severity of illness (APACHE IV) increased across periods in both ICUs (p ≤ 0.001).
Process of Care Evaluation (Full Bundle ICU Only)
Sedative Use and Delirium Prevalence(Fig. S2, Supplemental Digital Content 1, http://links.lww.com/CCM/E506). In the rull bundle ICU, the proportion of patients receiving continuous sedation decreased across all three periods (p < 0.001 for midazolam and fentanyl; p = 0.06 for propofol) (Fig. S3A, Supplemental Digital Content 1, http://links.lww.com/CCM/E506). The proportion of patients with ICU delirium and/or coma also decreased across all three periods (p ≤ 0.02) and similar to sedative use, the largest decrease occurred after AD was implemented (Fig. S3B, Supplemental Digital Content 1, http://links.lww.com/CCM/E506).
After EC was implemented in the full bundle ICU (period 1 vs 2), the proportion of patients evaluated by the rehabilitation team (i.e., either PT and/or OT) increased from 19% to 90% and the proportion of patient days spent passively lying or sitting in bed decreased from 95% to 37%. Patients received rehabilitation therapy within 1 day of ICU admission (median ICU day 1; interquartile range [IQR], 0–1) for a median of 60% of all ICU days (IQR, 50–80%); 77% of patients dangled at the bed’s edge, 65% stood, and 54% walked at least once during their ICU stay. No serious complications occurred during the 1,345 rehabilitation treatments. The main reasons why patients did not receive rehabilitation therapy were lack of staff and clinical instability (61% and 29% of patient days with no rehabilitation, respectively).
Clinical Quality Outcomes.
The proportion of patients with ICU-acquired pressure ulcers decreased (39% to 23%; p < 0.001) and the proportion of ICU patient days in restraints decreased (30% to 26%; < 0.001) after implementation of EC in the full bundle ICU (period 1 vs 2; Fig. 2A). In contrast, the prevalence of ICU-acquired pressure ulcers increased (18% to 23% of patients; p = 0.04) and proportion of ICU days in restraints increased (50% to 54%; p = 0.001) in the partial bundle ICU during the same periods of time (Fig. 2B).
Duration of MV and ICU LOS significantly changed in the full bundle ICU but not in the partial bundle ICU across three periods (Table 2). The duration of MV was significantly shorter in period 2 in the full versus partial bundle ICU, and ICU LOS was significantly shorter across all three periods in the full versus partial bundle ICU (p < 0.001). Hospital LOS and hospital mortality did not differ across all periods in both ICUs.
In our DiD analyses, implementation of AD in both full bundle and partial bundle ICUs was associated with no significant changes in clinical outcomes, except for increased hospital LOS in the full versus partial bundle ICU (5.9%; 95% CI, 4.6–7.2%; p = 0.011) (Table 3). Implementation of EC in the full bundle ICU after AD was associated with a 22.3% decrease in duration of MV (95% CI, –22.5% to –22.0%; p < 0.001), a 10.3% decrease in ICU LOS (95% CI, –15.6% to –4.7%; p = 0.028), and a 7.8% decrease in hospital LOS (95% CI, –8.7% to –6.9%; p = 0.006) compared with the partial bundle ICU (Table 3). The parallel trend assumption was met for all outcomes except for hospital LOS in period 1, where hospital LOS increased more in the full versus partial bundle ICU (0.17% change per calendar day; 95% CI, 0.10–0.24%; p = 0.022) (Table S2, Supplemental Digital Content 1, http://links.lww.com/CCM/E506). Sensitivity analyses including patients with hospital LOS greater than or equal to 90 days (n = 28, who had been excluded from our primary cohort) revealed similar results (Table S3, Supplemental Digital Content 1, http://links.lww.com/CCM/E506).
In DiD analyses, implementation of AD in both full and partial bundle ICUs was associated with no significant changes in cost between the two units (Table 3). Implementation of EC in only the full bundle ICU was associated with a 24.2% reduction in total ICU cost (95% CI, –41.4% to –2.0%; p = 0.034; Table 3) and a 30.2% reduction in total hospital cost (95% CI, –46.1% to –9.5%; p = 0.007) in the full versus partial bundle ICU; there was no reduction in average daily ICU cost (4.4%; 95% CI, –4.5% to 14.1%; p = 0.342). The parallel trend assumption was met for all cost outcomes (Table S2, Supplemental Digital Content 1, http://links.lww.com/CCM/E506).
This is the first large-scale prospective quality improvement study demonstrating the value of staged implementation of a bundle of evidence-based interventions aimed at reducing ICU associated weakness and delirium. We showed that the addition of (E)arly mobilization and structured interdisciplinary (C)oordination of bundle components to a spontaneous (B)reathing trial, (A)wakening from sedation, and (D)elirium monitoring/management program (B-AD + EC), is feasible, associated with improvements in quality of care, and is independently associated with substantial reductions in MV duration, ICU LOS, hospital LOS, and cost savings after adjusting for secular trends and patient-level confounders.
Our findings complement the growing literature demonstrating the clinical benefit of ABCDE bundle (25, 41). Simultaneous implementation of ABCDE/F bundle components has been associated with increased hospital survival and delirium and coma-free days, and reduced duration of MV (21–23).
Studies suggest that ABCDE, and early mobilization in particular, can be challenging to implement in routine practice (24–26, 42). Over 100 unique barriers have been identified in recent literature reviews (43). Dubb et al (44) classified these barriers into four categories: patient-related (e.g., deep sedation, delirium, new immobility/weakness), structural (e.g., lack of mobility protocol, limited staff and equipment, inadequate training), process related (e.g., lack of coordination), and cultural (e.g., lack of ICU mobility culture, staff buy-in, expertise). The positive outcomes in our study may be explained by our use of strategies specifically targeting these barriers, including: 1) reducing sedative use and delirium (B-AD, period 1) “before” implementation of EC (period 2) so patients were more awake and could actively engage in mobilization; 2) mobilization of patients within 1 day of ICU admission to prevent the development of new immobility/weakness; 3) developing an interdisciplinary mobility protocol with prespecified roles and responsibilities before EC implementation; 4) obtaining administrative buy-in to finance dedicated rehabilitation staff and rehabilitation equipment; 5) interdisciplinary simulation training of mobilization scenarios to enhance skills, improve interdisciplinary communication, and increase buy-in; 6) daily interdisciplinary coordination of staff and bundle components; and 7) including local nursing, respiratory, rehabilitation champions in protocol development, training, and dissemination.
Our large effect size may also be explained by our use of DiD analysis which mitigates against secular trends that can confound pre-post study designs (21, 23). In addition, prior studies implemented bundle components all at once, which may reduce overall bundle compliance and offset clinical benefit if components are not fully adopted (25, 45). Barnes-Daly et al (22) showed that for every 10% increase in ABCDEF bundle compliance, odds for hospital survival increased by 7%. Finally, our study excluded non-MV patients from analysis since only a fraction of the bundle (i.e., D, E) applies to them. Their inclusion in prior studies may have diminished any effect seen (21, 23).
This is the first report on the financial impact of the entire ABCDE bundle. Prior analyses on the awakening/delirium bundle components suggested cost savings, but studies on early mobilization have reported conflicting results (35, 46, 47). Using patient-level data, we found that adding EC to B-AD led to substantial cost savings which appear to be primarily explained by reductions in LOS (as indicated by decreased overall costs but unchanged average daily ICU cost before and after EC implementation).
Our finding that EC implementation was associated with shorter ICU and hospital LOS is consistent with prior randomized controlled trials (RCTs) and quality improvement studies (17, 35, 48). However, two recent RCTs on early mobilization found no effect on hospital LOS. In Morris et al (49), a sedation protocol was not used, which may have limited the efficacy of spontaneous breathing trials and early mobilization. In Moss et al (50), mobilization was initiated 8 days after ICU admission (vs 1 d in this study). Given the rapid degradation of muscle of critically ill patients, mobilization may be less effective if initiated after muscle loss has occurred (4).
Our study highlights several areas for future research. These include assessment of patient-centered outcomes such as short and long-term disability and readmission rate, determination of return on investment, cost analyses accounting for payer status, and evaluation of bundle dissemination and sustainability. The ABCDE bundle has been reframed since our 2014 study to include assessment, management and prevention of pain, and (F)amily empowerment and engagement (“F” in ABCDEF) (28). Future studies will need to reconcile our findings with the updated components.
This study has several strengths. Our DiD approach allowed us to adjust for secular trends which could have confounded prior historically-controlled studies. We also fulfilled a majority of the rigorous assumptions required for internal validity of the DiD estimates. Our cost data were generated from costs attributed to individual patients rather than assumptions based on average published costs. Finally, our study evaluated one of the largest cohorts to date.
This study has some limitations. Despite adjusting for patient characteristics, unmeasured differences and/or changes in cohort composition could have impacted our results. We also did not include discharge location in our model. Our study was conducted in a single medical center, which may limit generalizability. For example, the bundle’s impact on quality metrics (e.g., pressure ulcers) may be greater in ICUs with higher rates at baseline than sites that have already achieved low rates. There was potential for cross-contamination of practices between the two ICUs. However, cross-contamination would have biased the estimates toward the null. Because cost-to-charge ratios change across calendar years, we were unable to compare costs between the same seasonal periods and needed to use a smaller cohort for the cost analyses. Although changes in processes of care were demonstrated in the full bundle ICU, data were not collected in the partial bundle ICU for comparison. Finally, we were unable to fulfill the parallel trend assumption for hospital LOS as it increased in the full bundle ICU relative to the partial bundle ICU in period 1. However, this would bias our findings “toward the null” making it more difficult to demonstrate subsequent decreased hospital LOS after EC implementation in period 2. Because hospital LOS decreased despite this bias, our results may underestimate the full impact of ABCDE bundle implementation.
This study demonstrates that the complex ABCDE bundle can be successfully implemented into routine care. We showed that the addition of early mobilization and bundle coordination to an established targeted sedation and delirium management program led to substantial reductions in MV duration, LOS, and hospital cost, liberated patients from restraints, and reduced iatrogenic complications. These data underscore the value of the ABCDE bundle and support the concept that the entire bundle is truly greater than the individual parts.
We acknowledge the data collection contributions of Stephanie Coehlo, PT, Glenmore Wiggan, OT, and the Department of Rehabilitation Medicine staff for this project.
1. Hopkins RO, Weaver LK, Collingridge D, et al. Two-year cognitive, emotional, and quality-of-life outcomes in acute respiratory distress syndrome. Am J Respir Crit Care Med 2005; 171:340–347
2. Brummel NE, Jackson JC, Pandharipande PP, et al. Delirium
in the ICU and subsequent long-term disability among survivors of mechanical ventilation
. Crit Care Med 2014; 42:369–377
3. Pandharipande PP, Girard TD, Jackson JC, et al.; BRAIN-ICU Study Investigators: Long-term cognitive impairment after critical illness. N Engl J Med 2013; 369:1306–1316
4. Puthucheary ZA, Rawal J, McPhail M, et al. Acute skeletal muscle wasting in critical illness. JAMA 2013; 310:1591–1600
5. Jackson JC, Pandharipande PP, Girard TD, et al.; Bringing to light the Risk Factors And Incidence of Neuropsychological dysfunction in ICU survivors (BRAIN-ICU) study investigators: Depression, post-traumatic stress disorder, and functional disability in survivors of critical illness in the BRAIN-ICU study: A longitudinal cohort study. Lancet Respir Med 2014; 2:369–379
6. Needham DM, Davidson J, Cohen H, et al. Improving long-term outcomes after discharge from intensive care unit: Report from a stakeholders’ conference. Crit Care Med 2012; 40:502–509
7. Vasilevskis EE, Pandharipande PP, Girard TD, et al. A screening, prevention, and restoration model for saving the injured brain in intensive care unit survivors. Crit Care Med 2010; 38:S683–S691
8. Morandi A, Brummel NE, Ely EW. Sedation, delirium
and mechanical ventilation
: The ‘ABCDE’ approach. Curr Opin Crit Care 2011; 17:43–49
9. Bassett R, Adams KM, Danesh V, et al. Rethinking critical care
: Decreasing sedation, increasing delirium
monitoring, and increasing patient mobility. Jt Comm J Qual Patient Saf 2015; 41:62–74
10. Barr J, Fraser GL, Puntillo K, et al. Clinical practice guidelines for the management of pain, agitation, and delirium
in adult patients in the intensive care unit. Crit Care Med 2013; 41:278–280
14. Ely EW, Baker AM, Dunagan DP, et al. Effect on the duration of mechanical ventilation
of identifying patients capable of breathing spontaneously. N Engl J Med 1996; 335:1864–1869
15. Kress JP, Pohlman AS, O’Connor MF, et al. Daily interruption of sedative infusions in critically ill patients undergoing mechanical ventilation
. N Engl J Med 2000; 342:1471–1477
16. Girard TD, Kress JP, Fuchs BD, et al. Efficacy and safety of a paired sedation and ventilator weaning protocol for mechanically ventilated patients in intensive care (Awakening and Breathing Controlled trial): A randomised controlled trial. Lancet 2008; 371:126–134
17. Schweickert WD, Pohlman MC, Pohlman AS, et al. Early physical and occupational therapy in mechanically ventilated, critically ill patients: A randomised controlled trial. Lancet 2009; 373:1874–1882
18. Needham DM, Korupolu R. Rehabilitation quality improvement in an intensive care unit setting: Implementation
of a quality improvement model. Top Stroke Rehabil 2010; 17:271–281
19. Strøm T, Martinussen T, Toft P. A protocol of no sedation for critically ill patients receiving mechanical ventilation
: A randomised trial. Lancet 2010; 375:475–480
20. Klompas M, Anderson D, Trick W, et al.; CDC Prevention Epicenters: The preventability of ventilator-associated events. The CDC Prevention Epicenters Wake Up and Breathe Collaborative. Am J Respir Crit Care Med 2015; 191:292–301
21. Balas MC, Vasilevskis EE, Olsen KM, et al. Effectiveness and safety of the awakening and breathing coordination, delirium
monitoring/management, and early exercise/mobility bundle. Crit Care Med 2014; 42:1024–1036
22. Barnes-Daly MA, Phillips G, Ely EW. Improving hospital survival and reducing brain dysfunction at seven California community hospitals: Implementing PAD guidelines via the ABCDEF bundle in 6,064 patients. Crit Care Med 2017; 45:171–178
23. Kram SL, DiBartolo MC, Hinderer K, et al. Implementation
of the ABCDE bundle to improve patient outcomes in the intensive care unit in a rural community hospital. Dimens Crit Care Nurs 2015; 34:250–258
24. Carrothers KM, Barr J, Spurlock B, et al. Contextual issues influencing implementation
and outcomes associated with an integrated approach to managing pain, agitation, and delirium
in adult ICUs. Crit Care Med 2013; 41:S128–S135
25. Miller MA, Govindan S, Watson SR, et al. ABCDE, but in that order? A cross-sectional survey of Michigan intensive care unit sedation, delirium
, and early mobility practices. Ann Am Thorac Soc 2015; 12:1066–1071
26. Boehm LM, Dietrich MS, Vasilevskis EE, et al. Perceptions of workload burden and adherence to ABCDE bundle among intensive care providers. Am J Crit Care 2017; 26:e38–e47
27. Morandi A, Piva S, Ely EW, et al. Worldwide survey of the “assessing pain, both spontaneous awakening and breathing trials, choice of drugs, delirium
monitoring/management, early exercise/mobility, and family empowerment” (ABCDEF) bundle. Crit Care Med 2017; 45:e1111–e1122
28. Balas MC, Devlin JW, Verceles AC, et al. Adapting the ABCDEF bundle to meet the needs of patients requiring prolonged mechanical ventilation
in the long-term acute care hospital setting: Historical perspectives and practical implications. Semin Respir Crit Care Med 2016; 37:119–135
29. Jolley SE, Regan-Baggs J, Dickson RP, et al. Medical intensive care unit clinician attitudes and perceived barriers towards early mobilization
of critically ill patients: A cross-sectional survey study. BMC Anesthesiol 2014; 14:84
30. Hsieh SJ, Levi D, Prince D, et al. Staged implementation
of the ABCDE bundle improves ICU patient outcomes. Am J Respir Crit Care Med 2014; 189:A2540
31. Otusanya O, Hsieh SJ, Gong MN, et al. Awakening and Breathing Coordination, Delirium
Monitoring/Management and Early Mobilization
(ABCDE) bundle reduces hospital costs. Am J Respir Crit Care Med 2016; 193:A4357
32. Ely EW, Truman B, Shintani A, et al. Monitoring sedation status over time in ICU patients: Reliability and validity of the Richmond Agitation-Sedation Scale (RASS). JAMA 2003; 289:2983–2991
33. Sessler CN, Gosnell MS, Grap MJ, et al. The Richmond Agitation-Sedation Scale: Validity and reliability in adult intensive care unit patients. Am J Respir Crit Care Med 2002; 166:1338–1344
34. Ely EW, Inouye SK, Bernard GR, et al. Delirium
in mechanically ventilated patients: Validity and reliability of the confusion assessment method for the intensive care unit (CAM-ICU). JAMA 2001; 286:2703–2710
35. Morris PE, Goad A, Thompson C, et al. Early intensive care unit mobility therapy in the treatment of acute respiratory failure. Crit Care Med 2008; 36:2238–2243
36. Bellin E, Fletcher DD, Geberer N, et al. Democratizing information creation from health care data for quality improvement, research, and education-the Montefiore Medical Center Experience. Acad Med 2010; 85:1362–1368
37. Gershengorn HB, Garland A, Gong MN. Patterns of daily costs differ for medical and surgical intensive care unit patients. Ann Am Thorac Soc 2015; 12:1831–1836
38. Abadie A. Semiparametric difference-in-differences estimators. Rev Econ Stud 2005; 72:1–19
39. Dimick JB, Ryan AM. Methods for evaluating changes in health care policy: The difference-in-differences approach. JAMA 2014; 312:2401–2402
40. Puma MJ, Olsen RP, Bell SH, et al. What to Do When Data Are Missing in Group Randomized Controlled Trials (NCEE 2009-0049). 2009Washington, DC, National Center for Education Evaluation and Regional Assistance, Institute of Education Sciences, U.S. Department of Education.
41. Ely EW. The ABCDEF bundle: Science and philosophy of how ICU liberation serves patients and families. Crit Care Med 2017; 45:321–330
42. Honiden S, Connors GR. Barriers and challenges to the successful implementation
of an intensive care unit mobility program: Understanding systems and human factors in search for practical solutions. Clin Chest Med 2015; 36:431–440
43. Costa DK, White MR, Ginier E, et al. Identifying barriers to delivering the awakening and breathing coordination, delirium
, and early exercise/mobility bundle to minimize adverse outcomes for mechanically ventilated patients: A systematic review. Chest 2017; 152:304–311
44. Dubb R, Nydahl P, Hermes C, et al. Barriers and strategies for early mobilization
of patients in intensive care units. Ann Am Thorac Soc 2016; 13:724–730
45. Trogrlić Z, van der Jagt M, Bakker J, et al. A systematic review of implementation
strategies for assessment, prevention, and management of ICU delirium
and their effect on clinical outcomes. Crit Care 2015; 19:157
46. Lord RK, Mayhew CR, Korupolu R, et al. ICU early physical rehabilitation programs: Financial modeling of cost savings. Crit Care Med 2013; 41:717–724
47. Awissi DK, Bégin C, Moisan J, et al. I-SAVE study: Impact of sedation, analgesia, and delirium
protocols evaluated in the intensive care unit: An economic evaluation. Ann Pharmacother 2012; 46:21–28
48. Needham DM, Korupolu R, Zanni JM, et al. Early physical medicine and rehabilitation for patients with acute respiratory failure: A quality improvement project. Arch Phys Med Rehabil 2010; 91:536–542
49. Morris PE, Berry MJ, Files DC, et al. Standardized rehabilitation and hospital length of stay among patients with acute respiratory failure: A randomized clinical trial. JAMA 2016; 315:2694–2702
50. Moss M, Nordon-Craft A, Malone D, et al. A randomized trial of an intensive physical therapy program for patients with acute respiratory failure. Am J Respir Crit Care Med 2016; 193:1101–1110