The ABCDEF bundle (Assess, prevent, and manage pain and Delirium; Both spontaneous awakening and breathing trials; Choice of analgesia/sedation; Early mobility; and Family engagement) improves intensive care unit outcomes, but adoption into practice is poor.
To assess the effect of quality improvement collaborative participation on ABCDEF bundle performance.
This interrupted time series analysis included 20 months of bundle performance data from 15 226 adults admitted to 68 US intensive care units. Segmented regression models were used to quantify complete and individual bundle element performance changes over time and compare performance patterns before (6 months) and after (14 months) collaborative initiation.
Complete bundle performance rates were very low at baseline (<4%) but increased to 12% by the end. Complete bundle performance increased by 2 percentage points (SE, 0.9; P = .06) immediately after collaborative initiation. Each subsequent month was associated with an increase of 0.6 percentage points (SE, 0.2; P = .04). Performance rates increased significantly immediately after initiation for pain assessment (7.6% [SE, 2.0%], P = .002), sedation assessment (9.1% [SE, 3.7%], P = .02), and family engagement (7.8% [SE, 3%], P = .02) and then increased monthly at the same speed as the trend in the baseline period. Performance rates were lowest for spontaneous awakening/breathing trials and early mobility.
Quality improvement collaborative participation resulted in clinically meaningful, but small and variable, improvements in bundle performance. Opportunities remain to improve adoption of sedation, mechanical ventilation, and early mobility practices.
Notice to CE enrollees
This article has been designated for CE contact hour(s). The evaluation demonstrates your knowledge of the following objectives:
Evaluate the effects of quality improvement collaborative participation in ABCDEF bundle performance.
Examine monthly individual ABCDEF bundle element performance rates before and after quality improvement collaborative initiation.
Identify which bundle elements were least likely to improve with quality improvement collaborative participation.
To complete the evaluation for CE contact hour(s) for this article #A223113, visit www.ajcconline.org and click the “CE Articles” button. No CE evaluation fee for AACN members. This expires on January 1, 2024.
The American Association of Critical-Care Nurses is accredited as a provider of nursing continuing professional development by the American Nurses Credentialing Center’s Commission on Accreditation, ANCC Provider Number 0012. AACN has been approved as a provider of continuing education in nursing by the California Board of Registered Nursing (CA BRN), CA Provider Number CEP1036, for 1.0 contact hour.
Millions of survivors of critical illness worldwide experience profound physical, mental, and cognitive health impairments.1,2 These short- and long-term impairments are often caused or exacerbated by severity of illness, preexisting comorbidities, and conditions commonly experienced during the course of an intensive care unit (ICU) stay.3,4 For example, ICU-acquired pain, deep sedation, delirium, and weakness are associated with numerous adverse health outcomes, including increased risk of death, prolonged mechanical ventilation, depression, functional decline, and severe neurocognitive dysfunction.5,6
Although the results of numerous well-designed clinical trials suggest that a number of safe and effective ICU symptom management, mechanical ventilation liberation, and mobility interventions exist,4,6,7 analyses continue to show that many are underused in everyday clinical practice.8,9 This research-to-practice gap most likely contributes to the high morbidity, mortality, and cost associated with critical care.
The ABCDEF bundle is an integrated, interprofessional approach to optimizing ICU team performance and patient- and family-centered outcomes.3,4,10 The bundle consists of the following individual elements: (A) assess, prevent, and manage pain; (B) both spontaneous awakening trials (SATs) and spontaneous breathing trials (SBTs); (C) choice of analgesia and sedation; (D) delirium: assess, prevent, and manage; (E) early mobility and exercise; and (F) family engagement and empowerment. Previous investigations demonstrated the safety and effectiveness of incorporating earlier versions of the ABCDEF bundle into everyday care.11–13 Prior publications also have documented the many challenges that ICU providers experience when trying to deliver the interventions contained in the bundle in a consistent, collective, and coordinated manner.14–17 Implementation science methods may help address this problem by developing and testing strategies to overcome the known barriers to ABCDEF bundle delivery.
Quality improvement collaboratives (QICs) are used in a variety of health care settings to help facilitate adoption of evidence-based practices. A QIC is an organized and multifaceted implementation strategy that includes multiple health care teams who come together to learn, share improvement tactics, compare benchmark data, and support the dissemination and implementation of clinical evidence or effective models of care.18–21 Although previous QICs have improved clinical outcomes and/or adherence to targeted care practices for a number of ICU conditions (eg, central catheter–associated bloodstream infections, sepsis, and ventilator-associated pneumonia),22,23 until recently, none have specifically focused on common ICU syndromes such as pain, delirium, and weakness. Moreover, evidence is currently insufficient to draw conclusions about the overall effectiveness of QICs or their ability to evoke meaningful and sustained change.18,19,24–26
The Society of Critical Care Medicine (SCCM) ICU Liberation Collaborative aimed to support the successful implementation of the ABCDEF bundle in a large and diverse group of US ICUs.10 In a cohort of patients admitted to ICUs that participated in this QIC, ABCDEF bundle performance was associated with significant improvements in clinical outcomes, after relevant covariates were controlled for.27 However, the effect of QIC participation on important process measures such as overall bundle adoption and delivery of the separate evidence-based interventions contained in the bundle remains unclear. The range of impact the QIC had at the individual ICU level is also unknown. These are important knowledge gaps considering the limited empirical data describing current national ABCDEF bundle performance rates, the uncertainty surrounding how long it takes to effectively implement the bundle, and the paucity of detailed descriptions of whether QIC participation effects bundle adoption in some ICUs more than others. The purposes of this study were to evaluate the effect of ICU Liberation Collaborative participation on ABCDEF bundle performance and explore whether bundle performance differed among participating ICUs at the end of the QIC.
The purpose of this study was to evaluate the effect participating in a quality improvement collaborative had on ABCDEF bundle adoption.
Study Design and Participants
This implementation study used an interrupted time series model to analyze 20 months of ABCDEF bundle performance data in 15 226 critically ill adults admitted to the 68 academic, community, and Veterans Affairs ICUs participating in the SCCM ICU Liberation Collaborative. The Vanderbilt University Medical Center institutional review board (IRB) served as the coordinating center IRB and approved this quality improvement project. All QIC participants acquired site-specific IRB evaluation and approval.
Data Sources and Collection
The study database contained deidentified demographic, clinical, and bundle performance data for critically ill adults with a variety of diagnoses, receiving or not receiving mechanical ventilation, admitted to an ICU participating in the QIC. Excluded from data collection were patients who died, were discharged from a participating ICU within 24 hours of ICU admission, or were undergoing active life support withdrawal and/or receiving only comfort care within 24 hours of ICU admission.
To enhance reliability of data collection, operational definitions for all study variables were created before QIC initiation. These definitions and a step-by-step guide to data collection were communicated to participating sites via a standard operating procedures manual and online webinars. Local staff members, who received formal data collection training and as-needed support from SCCM personnel, manually abstracted data from eligible patients’ medical records (either electronic or paper) at their individual institution. The data were then entered into a Research Electronic Data Capture database (grant support UL1 TR000445 from the National Center for Advancing Translational Sciences, National Institutes of Health), a secure, web-based application for validated data entry, transmission, and storage.
We collected 20 months of ABCDEF bundle performance data, including 6 months of baseline (preimplementation) data from January 2015 through June 2015 and 14 months of data collected prospectively during the QIC from January 2016 through February 2017. No changes in usual care were reported during the baseline period. During the 14-month implementation period, sites were encouraged to use the bundle. Data were collected for the first 5 patients (baseline period) or first 15 patients (implementation period) consecutively admitted to the ICU each month. Performance data were collected for each qualifying patient for a maximum of 7 ICU days or until the patient was transferred out of the ICU, was designated as having non-ICU status, or died.
The implementation outcome examined in this analysis was ABCDEF bundle adoption (performance). The ABCDEF bundle consists of 7 discrete, evidence-based interventions (ie, 7 bundle elements). Eligibility criteria for receipt of bundle elements, definitions of each element, and bundle element performance criteria are provided in the Supplement (available online only at www.ajcconline.org). Consistent with prior work,27 ABCDEF bundle performance was defined in 2 ways: complete bundle performance and individual element performance. Complete bundle performance was a patient-day in which every eligible bundle element was performed (100% of the bundle was performed). Individual element performance was a patient-day in which an eligible patient received a particular bundle element (eg, a patient receiving mechanical ventilation had an SBT).
Complete and individual bundle element performance was assessed daily for each included patient and aggregated across all patients for each month before and after bundle implementation during the 20-month study period. Performance was measured only on the days the patient was in the ICU for a full 24 hours. The 7 bundle elements, eligibility criteria, and bundle element performance requirements did not change over the course of the QIC.
The history of the SCCM ICU Liberation Collaborative, methods used to recruit QIC sites, requirements for participation, site selection process, and strategies used to foster evidence adoption are detailed elsewhere10,28 and summarized in the Supplement (available online only). Briefly, 69 ICUs were officially invited to participate in the QIC after a formal review process. One site declined, leaving 68 adult ICUs from 29 states and Puerto Rico as QIC participants. All 68 sites contributed to the QIC database and completed the entire course of the collaborative. An interprofessional team composed of SCCM staff and experts in critical care, quality improvement, and implementation science led the collaborative. The collaborative consisted of in-person and virtual components in which a variety of improvement methods and implementation strategies were taught and used. All QIC sites were invited to participate in 4 in-person meetings (fall 2015, spring 2016, fall 2016, and spring 2017), monthly combined learning calls, database training sessions, a digital community, selected in-person site visits, and as-needed expert consultation and support. The key components, educational topics, and implementation strategies taught during the QIC are presented in Supplemental Table 1 (available online only).
We assessed characteristics of the overall patient cohort before and during QIC participation. We then calculated the aggregate monthly rates of complete bundle performance and element-specific performance per eligible ICU patient-days across all study sites. Supplemental Table 2 (available online only) provides the definitions of the performance measures and the numerators and denominators for the calculations. The time series data for monthly rates were plotted to illustrate the temporal trends, which were “interrupted” by the adoption of the QIC at the 7th month. Segmented regression analysis for interrupted time series data was used to model the linear trend of rates as a function of time for each study period (months 1-6 for the baseline period and months 7-20 for the implementation period).
From the segmented linear regression models, we derived estimates of level change and slope change for each performance measure. Level change is an estimate of the change in rate from the end of the baseline period to immediately after initiation of the implementation. Slope change is an estimate of the change in trend in the implementation period compared with the trend in the baseline period. A positive impact of QIC participation is indicated by either a significant level increase or a significant slope increase. We accounted for first-order autocorrelation, if indicated by the Durbin-Watson statistic, with the Yule-Walker method for segmented regression models. We examined the goodness of fit of the model by using the coefficient of determination (R2). In addition, we calculated unit-specific rates of complete bundle performance and element-specific performance for the last month of the QIC. We used descriptive statistics and box plots to illustrate the variability of performance rates across units. All statistical tests were 2-sided with a significance level of .05. Statistical analyses were performed with SAS statistical software, version 9.4 (SAS Institute Inc).
During the 20-month data collection period, 17 228 patients were enrolled. Excluded were 2002 patients with ICU stays of less than 24 hours. The overall study cohort included 15 226 critically ill adults (1713 in the baseline period and 13 513 in the implementation period) who spent 49 018 full days in an ICU. Most patients in the overall cohort were White (72%), male (58%), and admitted to teaching hospitals (63%); the most common admitting diagnosis was sepsis/septic shock or acute respiratory distress syndrome (22%) (Table 1). More than one-third of participants were aged 70 years or older, and more than half (54%) required mechanical ventilation. No clinically meaningful differences between the baseline and implementation cohorts were noted.
Complete bundle performance rates increased from 4% in the baseline period to 12% by the end of the collaborative.
Complete Bundle Performance
Figure 1 presents the monthly percentages of complete bundle performance before (baseline) and after QIC initiation. As illustrated in Figure 1, complete bundle performance rates were very low in the baseline period (<4%). Complete bundle performance increased by 2 percentage points (SE, 0.9 percentage points; P = .06) immediately after the start of the QIC. Each month of participation in the QIC was associated with a significantly greater upward trend in complete bundle performance rates (monthly increase, 0.6 percentage points; SE, 0.2 percentage points; P = .04), as compared with the relatively flat performance trend in the baseline period. By the end of the final month of the QIC, the complete bundle performance rate was 12%. Table 2 provides the estimates of level and slope changes from our segmented regression analysis of complete bundle performance in the implementation versus baseline periods.
Individual Bundle Elements
Figure 2 presents the monthly individual bundle element performance rates before and after initiation of the QIC. Performance rates for elements A (pain assessment), C (sedation assessment), and F (family engagement) increased significantly immediately after the QIC was begun and then increased monthly at the same rate at which they had increased during the baseline period (Table 2). The only bundle element with a significant change in slope from the baseline period was element B1 (SAT performance), in which the rate of change actually decreased. All remaining bundle elements had continued performance rate increases that were unchanged from the baseline period.
Figure 3 demonstrates variation among ICUs in complete and individual element bundle performance rates during the last month of the QIC. Bundle performance varied substantially among ICUs and across performance measures by the end of the QIC. Interquartile ranges were greater than 20% for all performance measures. The 3 elements with the highest variability were elements D (delirium assessment) (IQR, 37.5%-88.6%), B1 (SAT performance) (IQR, 22.2%-68.4%), and F (family engagement) (IQR, 59.3%-100%). By the end of the QIC, elements A (pain assessment), C (sedation assessment), and F were performed most frequently; more than half of the ICUs had performance rates of at least 70% for these elements. For element E (early mobility), most ICUs had a performance rate of less than 40% at the end of the QIC. Although complete performance rates were generally low (median, 7.5%; 75th percentile, 19.0%), a few ICUs reached rates of 53.7% to 71.4% at the end of the QIC.
Our study has 4 key findings. First, our analysis suggests that QICs are an effective, yet limited, way of increasing the adoption of a variety of evidence-based ICU practices. In this large national sample, QIC participation was associated with an 8% absolute increase in complete ABCDEF bundle performance. Second, although measurable process improvements occurred over the course of the QIC, opportunities remain to further enhance ABCDEF bundle adoption. At the end of this national QIC, complete ABCDEF bundle performance rates reached a mere 12%. Third, the low complete bundle performance rates appear to be explained by the bundle elements SAT, SBT, and early mobility (elements B1, B2, and E), which were performed in less than 40% of eligible patients by the end of the QIC. Fourth, we found substantial variability in ABCDEF bundle performance among ICUs. This variability presents the opportunity to learn the characteristics of high and low performers to enhance ABCDEF bundle performance and ultimately improve the short- and long-term physical, cognitive, and psychological outcomes of critically ill adults.
Despite the widespread use of QICs, questions remain regarding their effectiveness for improving health care quality and safety. These questions persist because much of the QIC literature to date consists of case studies, single-site evaluations, and/or qualitative descriptions of implementation success.18–20 Our study addresses some of these important concerns. We used a rigorous evaluation technique that is less vulnerable than other techniques to secular trends and can examine both the immediate and sustained effects of the QIC intervention. Moreover, the QIC we evaluated involved a large number of ICUs from diverse academic, community, and Veterans Affairs hospitals. These sites collected thousands of days of documented, rather than perceived, bundle performance data. These conditions allow us to conclude, with a reasonable amount of certainty, that participation in a QIC was an effective but limited way of increasing ABCDEF bundle performance. This finding is important considering the amount of practice change, interprofessional communication, and team-work needed to deliver this intervention.
Although we observed a significant increase in complete bundle performance during the relatively short data collection period, this improvement was slow (2 percentage points monthly) and far from complete (12% by QIC end). This low complete bundle performance is most likely explained by the limited delivery of daily SATs, SBTs, and early mobility interventions. Complete bundle performance depends on the delivery of all of the individual elements that a patient is eligible to receive, even the most challenging elements. In addition, some bundle elements are more interdependent than others. For example, early mobility requires that a patient be at least somewhat cognitively alert and physically engaged. Therefore, performing bundle element E would be more difficult in patients who have not undergone an SAT (element B1) or who remain physically tethered to mechanical ventilation because of nondelivery of an SBT (element B2). Complete bundle performance also requires the greatest amount of resources and care coordination among doctors, nurses, pharmacists, and respiratory, physical, and occupational therapists. Therefore, more time is likely required to reach larger gains in complete bundle performance because successful implementation depends on the interaction of multiple disciplines. It is also possible that elements B1 (SATs), B2 (SBTs), and E (early mobility) are less responsive to QIC participation.
Suboptimal complete bundle performance findings may also be explained by how the performance of certain bundle elements was defined. Normally, SATs, SBTs, and early mobility episodes are guided by safety screens that help clinicians determine if performing a particular intervention is prudent. In an effort to promote local adoption, each QIC site was allowed to develop its own SAT, SBT, and mobility safety screen criteria. The diversity in (and in some cases the lack of) safety screen criteria precluded determination of when an SAT, SBT, or mobility episode was appropriately not performed. For this reason, performance in this analysis was defined with a simple yes or no, meaning that either the patient did or did not have a SAT (if receiving continuously infused or intermittently scheduled sedatives), SBT (if receiving mechanical ventilation), or mobility episode (if in the ICU). Given this operational definition, achieving 100% complete bundle performance would indeed be difficult because of the high likelihood that some patients would be in too unstable a condition to receive certain bundle elements. Although a “dose-response” relationship between higher bundle performance and improvements in patient outcomes was previously demonstrated,12,27 the ideal complete bundle performance rate remains to be determined.
We found impressive and relatively quick improvements in the documented performance of pain and agitation/sedation assessments and family engagement (bundle elements A, C, and F). Because pain and agitation/sedation assessments are most often the responsibility of a single ICU profession (nursing), it is reasonable that these particular elements would be the first to improve and would be performed more often than the other bundle elements. Less clear is how QIC participation led to an improvement in family engagement. It is possible that QIC activities improved health professionals’ knowledge of the importance of engaging family members as active participants in ICU care or that interaction across QIC teams generated normative pressure and an opportunity to make changes in this area.
We found that the slope of change actually decreased for SAT performance (element B1) in the intervention phase compared with baseline. With newly competing processes as part of the entire bundle, the prior SAT momentum might have been attenuated by QIC participation. In addition to element complexity and competing bundle elements, acceptance or understanding of the benefits of this particular intervention might have been lower. For example, some debate remains about the benefit of performing SATs when patients are maintained at sedation levels reflecting both arousability and comfort while receiving continuously infused medications,29 and results from the initial trial showing the benefits of early mobility30 have yet to be replicated in other randomized trials.31,32 Continued improvement in ABCDEF bundle performance will require a much greater understanding of the factors associated with effective bundle adoption.
Although QIC participation may advance ABCDEF bundle adoption, QIC participation alone is most likely not a sufficient approach to implementation. For example, early mobility often depends on the presence of a trained physical therapist or additional personnel who may not be readily available in many ICUs. Staffing has been reported as a key barrier to early mobility in the ICU.14,33 Performance of SATs and SBTs similarly depends on nurse and respiratory therapist involvement. Unfortunately, SATs and SBTs are felt to increase workload in both of these professions.33 In addition to staffing and workload, culture and leadership are believed to be critically important to adoption of ABCDEF practices.15,16 For example, safety culture, staff receptivity to change, and prior QIC involvement were found to be associated with the use of SATs in a study of 386 US hospitals.34 Future cost-effectiveness research is required to better understand the financial implications of ABCDEF implementation to potentially support the necessary staffing levels to enhance implementation.
Participation in a quality improvement collaborative resulted in small, but meaningful improvements in ABCDEF bundle adoption.
Applying a positive deviance approach to efforts to implement the ABCDEF bundle may help facilitate the complex changes that are necessary to establish a culture of evidence-based practice. Instead of concentrating on what goes wrong, why errors occur, and the underlying cause of a problem, the positive deviance approach focuses on the behaviors, processes, and systems that contribute to safe and high-quality health care practices.35 Our data suggest that an opportunity exists to use such an approach in the ICU setting with QIC sites that demonstrated either high or low bundle performance rates. As shown in Figure 3, many ICU Liberation Collaborative sites met these criteria. By the end of the QIC, consistent variability in complete and individual bundle element performance among ICUs and across measures was observed. Although most sites had disappointingly low complete bundle performance rates at the end of the QIC, a few outliers achieved complete bundle performance rates of up to 71% in this period. Unfortunately, our current data cannot show the source of this variability. Future research should investigate the role that variables at the patient, clinician, and organizational levels play in ABCDEF bundle performance.
Our study had several limitations. As with all observational studies, residual confounding cannot be excluded as an explanation for the observed changes in bundle performance, although the length of follow-up and the heterogeneity of hospitals reduces the likelihood of residual confounding. In addition, although ICUs demonstrated improvements during a 20-month period, we cannot make any conclusions about longer-term sustainability. Because study participation was voluntary, our findings may have had volunteer bias, resulting in higher performance than would occur at nonparticipating sites. This possibility would mean that an even greater opportunity exists for improvement outside of the collaborative. Finally, the factors that contribute to the observed variability in unit-level bundle performance at the end of the QIC remain unclear.
Successful treatment of an acute illness is only one of the elements that leads to positive outcomes for ICU patients. Cognitive, functional, and mental health outcomes require attention toward common practices related to ICU symptom management, mechanical ventilation, and early mobility. Our findings and prior research suggest that adoption of the ABCDEF bundle is one way to improve these practices and that participation in a QIC may yield modest but clinically meaningful improvements in complete bundle performance. Nevertheless, substantial opportunities to improve the delivery of contemporary evidence-based ICU interventions remain. Examining units with the greatest and least gains in performance may provide a unique opportunity to understand key facilitators of and barriers to effective QIC implementation and to learn strategies for effective ABCDEF bundle adoption.
The authors sincerely thank and acknowledge Lori Harmon, RRT, MBA, CPHQ, director of quality at the SCCM, for her contribution to the creation of, and her immeasurable guidance provided to, the ICU Liberation Collaborative project from its inception and throughout the development of this publication. We would also like to acknowledge the invaluable contribution of time, passion, and subject matter expertise provided by the multidisciplinary collaborative faculty, SCCM staff partners, and collaborative participants without whom this project could not have reached fruition.
Research reported in this publication was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under award number R01 HL146781.
For more about ABCDEF bundle implementation, visit the Critical Care Nurse website, www.ccnonline.org, and read the article by Stollings et al, “Implementing the ABCDEF Bundle: Top 8 Questions Asked During the ICU Liberation ABCDEF Bundle Improvement Collaborative.” (February 2019).
To purchase electronic or print reprints, contact American Association of Critical-Care Nurses, 27071 Aliso Creek Road, Aliso Viejo, CA 92656. Phone, (800) 899-1712 or (949) 362-2050 (ext 532); fax, (949) 362-2049; email, firstname.lastname@example.org.