Delirium in the intensive care unit (ICU) affects an estimated 60% to 80% of all mechanically ventilated patients and 20% to 50% of all non–mechanically ventilated patients.1-3 Delirium is defined by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition as a disturbance in consciousness, attention, and perception that develops over a short time and can fluctuate during the course of the day.4 Patients who develop delirium in the ICU are at a greater risk of complications such as prolonged mechanical ventilation and length of stay (LOS), catheter self-removal, self-extubation, increased restraint use, and increased mortality.5,6 Annually, the cost of providing care to mechanically ventilated patients with delirium in the United States is estimated at $6.5 to $20.4 billion.7 The cost of treating patients with delirium is 2.5 times higher per day than in those without delirium, and increased consumption of health care resources may last up to 9 months after discharge from the acute care setting.8
One emerging evidence-based approach to managing delirium is the ABCDE bundle. This targeted process is the culmination of years of research conducted in critical care medicine and has the potential to decrease the incidence and duration of delirium and to positively affect other patient outcomes such as mortality, ICU and hospital LOS, and physical function.9 The ABCDE bundle incorporates awakening, breathing, coordination (or choice of sedative), delirium monitoring and management, and early mobility on a daily basis to minimize potential deleterious effects from prolonged hospitalization, including the development of delirium.
The ABCDE bundle is a multidisciplinary approach to patient care that utilizes nurses, respiratory care practitioners (RCPs), physicians, and rehabilitation specialists to reduce the incidence and duration of delirium.7,10-12 In non–mechanically ventilated ICU patients, the implementation of the ABCDE bundle is associated with a lower prevalence of delirium and a reduction in the number of delirium days.10,12 Likewise, in mechanically ventilated patients, the ABCDE bundle may also reduce the likelihood of developing delirium and shorten the duration of delirium and may lead to fewer days on the ventilator.10 Implementation of the bundle contributes to early mobility, which leads to less deconditioning and the potential to return more readily to baseline physical functioning.10,12
A coordinated effort by nurses and RCPs to pair awakening trials and spontaneous breathing trials are safe and effective mechanisms for assessing a patient’s readiness to wean from the mechanical ventilator.3 Excessive sedation can prohibit or prolong successful weaning from the mechanical ventilator. Awakening trials alone have demonstrated a decrease in the total number of patients days sedated, where breathing trials have demonstrated decreased length of time on the ventilator.13 Collaboration among the disciplines to pair awakening and spontaneous breathing trials has demonstrated that patients who receive coordinated trials spend fewer ICU days breathing with ventilator assistance, have shortened ICU and hospital LOSs and a reduced all-cause mortality within 1 year after discharge.3
Delirium has been linked to many complications in the ICU including increased length of ICU and overall hospital stays; increased ICU, hospital, and 6-month mortality rates; prolonged mechanical ventilation; prolonged continuous sedation; increased use of physical restraints; post-ICU cognitive impairment; and increased health care costs.2,5,14 Physically, patients who experience delirium during a critical illness are more likely to experience disability in their activities of daily living for up to 12 months after discharge.15 It is estimated that approximately 11% to 25% of patients present to the hospital with delirium, and reports of delirium among ICU patients vary from 20% to 80%.9,16 For every day a patient is in a delirious state in the ICU, the risk of death increases by 10%.13
Current recommendations for detecting delirium include the routine use of validated delirium screening tools such as the Confusion Assessment Method for the ICU or the Intensive Care Delirium Screening Checklist (ICDSC).13,14,16,17 Even though these bedside measurements of delirium were found to be reliable,17 1 study found that only 59% of surveyed health care practitioners routinely screen for delirium, and of those, only 33% used a validated tool.18
Once detected through the use of a validated screening tool, delirium may be managed through the use of pharmacological and nonpharmacological interventions. Pharmacological management of delirium is directed at the prevention of delirium through the use of evidence-based guidelines for the treatment of pain, agitation, and sedation. Recommendations for pain management include the preemptive administration of analgesics prior to invasive and potentially painful procedures, with intravenous opioids being the first-line treatment. Agitation and sedation guidelines recommend sedation at the lightest level possible for patient comfort, utilizing valid and reliable sedation assessment tools, such as the Richmond Agitation-Sedation Scale or the Sedation-Agitation Scale.14 In addition, administration of nonbenzodiazepine sedatives is preferred over sedation with benzodiazepines as this class of drug has been proven to be a strong predictor for the development of delirium.9,14
Nurses play a crucial role in preventing and reducing the duration of delirium. Nursing interventions include promoting a regular sleep-wake cycle, performing frequent reorientation, and optimizing the patients’ environment. This can be accomplished through the use of personal sensory-assistive devices, such as glasses and hearing aids; implementing strategies to control light and noise; and clustering activities to minimize disruption to the patient.14 The most protective, nonpharmacological, modifiable risk factor for the development of delirium is the receipt of early mobility.9
Prolonged immobility in critically ill individuals leads to physical weakness and neuropsychiatric dysfunction and complications with the overall course of illness.19,20 Despite evidence to support early physical mobility in ICU patients, ambulation and physical therapy are often not initiated until after a patient is discharged from the ICU. “Early” mobility is defined as the period beginning immediately upon return to physiological stability.19 Furthermore, pairing early mobility with interruption of sedation, physical therapy, and occupational therapy has been linked to improved functional outcomes at discharge, shorter duration of delirium, and a reduction in ventilator days versus standard care alone.10,19,20
Experts are calling for ABCDE bundle implementation as a standard of care for every ICU patient.12 Implementing daily multidisciplinary rounds for all ICU patients, engaging key leaders in unit initiatives, providing sustained and multimodal staff educational efforts, promoting high-quality consistent implementation of the interventions within the bundle, and thoroughly documenting all actions to promote compliance have all been cited as effective measures to successfully implement the ABCDE bundle.11,21 Despite evidence to support implementing the ABCDE bundle to minimize the incidence of complications among critically ill adults, many organizations have not yet implemented such strategies. Barriers to bundle implementation include intervention-related issues, communication and care coordination challenges, knowledge deficits, workload concerns, and documentation burden.21 Additional organizational barriers noted include excessive staff turnover and subsequent use of registry staff, low staff morale, and lack of respect among disciplines.11 Health care professionals who utilize the ABCDE bundle feel as though patients benefit from its implementation; they also report feeling valued and more autonomous.21
The purpose of this article is to describe an evidence-based practice (EBP) project to implement the ABCDE bundle in a rural community hospital ICU. The outcomes related to this project were to assess pre– and post–ABCDE bundle implementation patient outcomes related to hospital and ICU LOSs, ventilator days, and postbundle baseline delirium prevalence. The PICOT question for this EBP project was: Does the implementation of the ABCDE bundle care, versus the usual care (absence of the ABCDE bundle components), reduce the incidence of delirium, decrease patient LOS in the ICU, decrease patient total hospital LOS, and decrease length of mechanical ventilation of patients, thus decreasing costs in the ICU?
Sample and Setting
In 2013, prior to a 2-hospital system merger, the ABCDE bundle was successfully implemented in the ICUs of 2 hospitals in a rural community health care system. Following the affiliation, efforts to regionalize all hospital system policies and procedures led to the decision to implement the ABCDE bundle in the newly affiliated third rural community hospital’s 6-bed general adult ICU.
Evidence-Based Practice Project Process
REVIEW OF THE EVIDENCE
The process of implementing this EBP practice change included several key steps in order to ensure successful implementation. Prior to planning and implementing the ABCDE bundle, a review of the literature was conducted to assess the current body of evidence on the ABCDE bundle. The databases CINAHL and PubMed were searched utilizing the terms ABCDE and delirium. The search was limited to peer-reviewed articles published since 2007 that examined adult intensive care patients. This literature was used to examine the current body of knowledge regarding the evidence of the individual bundle components, as well as barriers and facilitators to implementation. The Johns Hopkins Nursing Evidence-Based Practice Model was used to rate and grade the current body of evidence as well as facilitate in project implementation.22
Prior to the implementation of several complementary practice changes, the current practices of the ICU in regard to the components of the ABCDE bundle were assessed. This was completed utilizing the American Association of Critical Care Nurses’ Unit Gap Analysis-ABCDE Bundle with the assistance of the clinical specialist.23 In addition, in order to identify which practice changes were required, meetings were held with the direct care providers and department leadership for in-depth information regarding current practices in patient sedation management, ventilator weaning, and current levels of collaboration between nursing and RCP staff.
Key stakeholder support was garnered from the senior leadership team. Approval was obtained from the Nurse Executive Committee, the chief medical officer, and subsequently from physicians with ICU admitting privileges. Education was provided to the physicians via e-mail and organizational interoffice mail, which included a letter of explanation for the proposed policy, a copy of the proposed policy, a 1-page fact sheet on the ABCDE bundle with additional references, and contact information for the project leader seeking implementation to provide the opportunity for questions or clarification. Individual face-to-face and telephone meetings were held with 2 physicians who requested the opportunity to further discuss the policy. The policy was approved by the Medical Executive Committee in September 2014.
Multiple educational sessions open to nursing, respiratory therapy, and rehabilitation staff were conducted utilizing various instructional methods. A live-contact-hour presentation that highlighted the evidence to support implementation of the ABCDE bundle was presented on several occasions. A second, nursing-specific, live-contact-hour program was presented to the nursing staff on the proper administration of the ICDSC and the effects of delirium on the hospitalized patient. The ICDSC was originally selected as the delirium screening tool of choice by the system, after nurses participated in a pilot study in 2012 comparing the ICDSC and the Confusion Assessment Method for the ICU. An asynchronous, online learning module with information on the new administrative policy was created and made available to nursing, respiratory, and rehabilitation staff on the organization’s learning management system. This method was chosen to increase staff accessibility to the information. The online module was modified into a paper format and placed in a binder housed in the ICU for staff to access. Staff education and support were of the utmost importance for successful implementation.
The ABCDE bundle was implemented as a routine standard of care for all adult patients admitted to the ICU on or after the implementation date of October 15, 2014. Prior to receiving awakening or breathing trials, or mobility exercises, a patient safety screen was performed (Table 1). This consisted of reviewing contraindications that would preclude patient participation in a particular bundle component. Patients were ineligible if they did not successfully pass the safety screening or did not receive clearance from their attending surgeon. Physicians could write an order to opt out of any component of the ABCDE bundle they believed was not in the best interest of the patient.
To capture the patient data relating to each of the components of the ABCDE bundle, documentation prompts were built in to the electronic medical record (EMR) by the nurse informaticist. A retrospective chart review via the EMR was used to obtain prebundle and postbundle data. The prebundle data implementation period was from October 15, 2013, to January 15, 2014. The post–bundle implementation period was from October 15, 2014, to January 15, 2015. The daily census of patients in the ICU was used to identify which medical records to access. Patients were assigned an observation number, free of identifying demographic information such as age, gender, or admitting diagnosis.
Outcomes of interest included the following:
- rate of compliance by direct care providers for each bundle element as documented in the EMR;
- changes in ICU LOS from prebundle to postbundle;
- changes in total hospital LOS from prebundle to postbundle;
- changes in number of ventilator days from prebundle to postbundle; and
- prevalence of postbundle delirium, as evidenced by a positive ICDSC score of 4 or greater.
Prior to the initiation of the project, institutional review board approval was obtained from the health care organization and from the educational institution in which the project leader was enrolled as a candidate for doctor of nursing practice.
Both pre–bundle and post–bundle implementation data were evaluated for effects on the outcomes of interest. Microsoft Excel was used for data analyses. Evaluation of the rate of compliance with bundle components was conducted immediately upon bundle implementation and continued throughout the postimplementation time frame. The purpose of this type of formative evaluation was to detect any process barriers to full bundle implementation that may not have been addressed during the planning phase. Summative evaluation was performed to identify and report changes in the length of ICU stay (reported in days), changes in overall hospital LOS (reported in days), and changes in length of time the patient receives mechanical ventilation (reported in days) from preimplementation to postimplementation. Summative evaluation is an essential method for establishing process outcomes. A 2-tailed t test was run to assess for significant changes in LOSs and ventilator days. As no prebundle delirium data existed, the prevalence rate of delirium after implementation was calculated and compared with the literature. Finally, the cost savings related to positive patient outcomes was calculated based on reductions in LOS and reduction in ventilator days.
Data were collected from a total of 47 patients in the prebundle group and 36 patients in the postbundle group (Table 2). Patients were not included if there was missing documentation regarding any component of the ABCDE protocol, as was the case with 6 patients in the postbundle group. The average daily census for the prebundle group was 3.3 patients and 1.8 patients in the postbundle group. No patients were excluded from the bundle due to physician order.
Lengths of stay were calculated for both the ICU LOS and overall hospital LOS. There was a slight decrease in the mean ICU LOS from 3.04 days in the prebundle group to 2.99 days in the postbundle group (P = .66). However, when examining overall hospital LOS, there was a 26% decrease in length of hospital admission from 6.9 days to 5.09 days (P = .06). Implementation of the ABCDE bundle focused on increasing multidisciplinary collaboration between nursing and respiratory care staff to coordinate awakening and breathing trials. Postbundle data revealed compliance with awakening and breathing documentation 100% of the time, either as having been performed or by checking an appropriate contraindication. As a result, the number of ventilator days per patient, from pre– to post–bundle implementation, decreased 29% from 3.3 to 2.3 (P = .33). The goal for ICDSC completion, twice a day on patients in the postbundle group, was met 92% of the time. Seven of the 36 patients (19%) had a positive delirium screening. Nonpharmacological interventions for delirium prevention and management were documented 89% of the time.
The EMRs were reviewed to determine the level of compliance staff nurses had on performance of early mobility activities. There were 104 documented opportunities for early mobility. Of these, the majority (55%) were assisting the patient with ambulation, to a chair, or to a seated position. Thirty-seven percent of patients received passive range-of-motion exercises. One opportunity for early mobility was missed because of a medical contraindication. Four opportunities for early mobility were missed because of staff either not performing or not documenting any type of early mobility for that shift. Indicators of early mobility were not assessed in the prebundle group because of lack of documentation on these activities in the prebundle period.
The costs to implement the ABCDE bundle in this facility included only the time of the multidisciplinary team members. No additional equipment or supplies were required. As a result of the decrease in ventilator days per patient, the organization had the potential to save approximately $700 per patient per ventilator-free day. This cost savings reflects charges incurred from respiratory care services, daily radiological studies to confirm endotracheal tube placement, and pharmacy charges related to continuous sedation and oral-care topical solutions required for mechanically ventilated patients. As a result of a decrease in overall hospital stay by 1.8 days, the organization had the potential to save an average of $2156 per patient, based on current organizational charges for a medical-surgical bed.
The positive patient outcomes experienced based on ABCDE bundle implementation in the ICU at a small, rural community hospital are consistent with the findings in the literature based on studies performed at large, tertiary-care medical centers. The successful implementation of this multidisciplinary process would not have been possible without organizational support and the dedicated members of the multidisciplinary team. There are facilitators and barriers to the implementation process within the context of the organizational structure that one should consider when planning an EBP change.
As compared with prebundle data, postbundle ICU and hospital LOS decreased, a finding that is supported in the literature that the ABCDE bundle may reduce the length of both overall and ICU LOSs.3 Documentation of the awakening and breathing was completed all of the time, and there were lower ventilator days in the postbundle group. These findings are consistent with other studies that also noticed a decrease in number of ventilator days per patient after implementation of coordinated awakening and breathing trials.3,10,13 While a statistically significant difference in ventilator days was not found, the patients in this study had a 1-day decrease in ventilator days, which would be clinically significant for ICU patients on the ventilator and may have significant implications for ventilator-associated events.
The delirium screening tool was completed the majority of the time, and the overall prevalence of delirium in this rural hospital was slightly lower than the rates in the literature of 20% to 80%.9,16 The lower prevalence of delirium may be related to the high rates of nonpharmacological delirium interventions reported by staff. Postbundle opportunities for early mobility were identified, and opportunities for missed early mobility were found. These data can be used to incorporate into future staff continuing education related to the ABCDE bundle and to develop an action plan for future quality improvement projects. Cost savings to the organization highlighted the financial importance of making EBP practice changes.
The greatest facilitator in this project implementation was the expertise of the multidisciplinary team. Teams play a critical role in improving the performance of health care systems. It was only through the knowledge of various team members that the project leader could fully assess the current organizational climate and make necessary adjustments to existing policies and procedures. A second facilitator was organizational support from key stakeholders. A third facilitator was the business case for implementation of the ABCDE bundle. A business case for a health care improvement intervention exists if the entity that invests in the intervention realizes a financial return on its investment in a reasonable time frame.24 Data from 3 months after ABCDE bundle implementation at the other 2 ICUs within the health care system demonstrated positive patient outcomes that could be translated directly into dollars.
A final facilitator that enhanced implementation was the ability to educate the staff nurses using a multimodal approach. The educational sessions were invaluable for project implementation. The project leader was able to connect with the direct care staff who carried out the daily activities of the policy and established rapport. During these sessions, the staff realized the necessity of changes to the current standard of care and saw the project as a means to improve patient outcomes. However, despite these identified facilitators, barriers were encountered from project conception through the actual implementation date.
The first barrier was timing, which affected several aspects of the planning and implementation process. Institutional review board approval was obtained in June 2014. However, because it was summer, it was very difficult to coordinate with the various members of the multidisciplinary team because of scheduled vacations. This barrier was overcome through numerous telephone conferences and electronic written communications. The second issue of timing was a delay in approval for the final ABCDE administrative policy. Approval of the policy by the Medical Executive Committee was delayed from July until September. Thus, implementation, which was originally intended to occur on August 15, 2014, could not occur until at least October 1, 2014. Another barrier was a major upgrade to the EMR. This created a stressful environment for the physicians and staff nurses and created a strain on the nursing informatics staff, a crucial part of the multidisciplinary team. It was suggested that ABCDE implementation be delayed until at least October 15, 2014, at which time the nurse informatacist could have the clinical documentation ready for go-live. This new timeline was agreed upon by all members of the team. Although an employee of the hospital system, the project leader was not an on-site employee of the hospital in which the bundle was implemented. Strategies to overcome such barriers include having an on-site project leader or liaison and scanning the horizon to determine other large-scale projects for conflicting implementation timelines.
Based on the current evidence available and the positive patient and financial outcomes realized from its implementation, the ABCDE bundle is a cost-effective method to enhance multiple patient outcomes in the ICU. Because of the synergistic effects of the bundle components, organizations should adopt all of the components of the bundle to optimize therapeutic outcomes. Proper multidisciplinary collaboration and comprehensive staff education using multiple educational delivery methods are essential for sustained culture change. Direct care staff nurses are a crucial component of successful implementation of EBPs at the bedside, such as the ABCDE bundle. As coordinators of care, they are the leader of the multidisciplinary team and active participants in each of the bundle’s components to ensure safe and effective care is delivered. Open communication and cooperation among the multidisciplinary team can assist the nurse in minimizing potential barriers that may be faced during the implementation process and beyond.
In conclusion, in a rural community hospital, the implementation of the ABCDE bundle is an evidence-based, cost-effective approach to improving clinical outcomes in the adult ICU. Full implementation requires a coordinated multidisciplinary approach, including nurses, physicians, RCPs and rehabilitation specialists. A strong leader, comprehensive educational tactics, and transparency in the implementation process are needed to facilitate adherence by direct care staff. When all elements of the bundle are implemented simultaneously, the incidence and duration of delirium decline, patient length of ICU stay and length of hospital stay decline, the amount of time a patient may spend sedated and on a mechanical ventilator decreases, and overall health care costs are reduced.
The authors acknowledge the dedication and the expertise of the following individuals: Steve Eisemann, BS, RRT; Michael Kingan, DNP, AGPCNP-BC, CWONC; Kitty Neff, BSN, RN; and Mickey Roderick, MSN, RN, CCRN.
1. Brummel NE, Girard TD. Preventing delirium
in the intensive care unit. Crit Care Clin
. 2013; 29( 1): 51–65.
2. Ely EW, Shintani A, Truman B, et al. Delirium
as a predictor of mortality in mechanically-ventilated patients in the intensive care unit. JAMA
. 2004; 291( 14): 1753–1762.
3. 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 randomized controlled trial. Lancet
. 2008; 371( 9607): 126–134.
4. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders
. 5th ed. Arlington, VA: American Psychiatric Publishing; 2013.
5. Ouimet S, Kavanagh BP, Gottfried SB, Skrobik Y. Incidence risk factors and consequences of ICU delirium
. Intensive Care Med
. 2007; 33( 1): 66–73.
6. Shehabi Y, Riker RR, Bokesch PM, Wisemandle W, Shintani A, Ely EW; SEDCOM (Safety and Efficacy of Dexmedetomidine Compared With Midazolam) Study Group. Delirium
duration and mortality in lightly sedated, mechanically-ventilated intensive care patients. Crit Care Med
. 2010; 38( 12): 2311–2318.
7. Balas MC, Vasilevskis EE, Burke WJ, et al. Critical care nurses’ role in implementing the ABCDE bundle
into practice. Crit Care Nurse
. 2012; 32( 2): 35–38–47; 40–47.
8. Leslie DL, Marcantonio ER, Zhang Y, Leo-Summers L, Inouye SK. One-year health care costs associated with delirium
in the elderly population. Arch Intern Med
. 2008; 168( 1): 27–32.
9. Vasilevskis EE, Han JH, Hughes CG, Ely EW. Epidemiology and risk factors for delirium
across hospital settings. Best Pract Res Clin Anaesthesiol
. 2012; 26( 3): 277–287.
10. 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( 5): 1024–36.
11. Carrothers KM, Barr J, Spurlock B, Ridgely MS, Damberg CL, Ely EW. 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( 9 suppl 1): S128–S135.
12. Olsen KT, Burke WJ, Peitz G, et al. The ABCDE bundle
reduces the incidence of delirium
in non–mechanically-ventilated patients [abstract]. Crit Care Med
. 2012; 40( 12): S1.
13. Vasilevskis EE, Pandharipande PP, Girard TD, Ely EW. A screening, prevention, and restoration model for saving the injured brain. Crit Care Med
. 2012; 38( 10 suppl): S683–S691.
14. 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: executive summary. Am J Health Syst Pharm
. 2013; 70( 1): 53–58.
15. 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( 2): 369–377.
16. Morandi A, Jackson JC, Ely EW. Delirium
in the intensive care unit. Int Rev Psychiatry
. 2009; 21( 1): 43–58.
17. Vasilevskis EE, Morandi A, Boehm L, et al. Delirium
and sedation recognition using validated instruments: reliability of bedside intensive care unit nursing assessments from 2007-2010. J Am Geriatr Soc
. 2011; 59( suppl 2): S249–S255.
18. Patel RP, Gambrell M, Speroff T, et al. Delirium
and sedation in the intensive care unit (ICU): survey of behaviors and attitudes of 1384 healthcare professionals. Crit Care Med
. 2009; 37( 3): 825–832.
19. Bailey P, Thomsen GE, Spuhler VJ, et al. Early activity is feasible and safe in respiratory failure patients. Crit Care Med
. 2007; 35( 1): 139–145.
20. Schweickert WD, Pohlman MC, Pohlman AS, et al. Early physical and occupational therapy in mechanically-ventilated, critically-ill patients: a randomized controlled trial. The Lancet
. 2009; 373( 9678): 1874–1882.
21. Balas MC, Burke WJ, Gannon D, et al. Implementing the awakening and breathing coordination, delirium
monitoring/management, and early exercise/mobility bundle into everyday care: opportunities, challenges, and lessons learned for implementing the ICU pain, agitation, and delirium
guidelines. Crit Care Med
. 2013; 41( 9 suppl 1): S116–S127.
22. Dearholt D, Dang D, eds. Johns Hopkins Nursing Evidence-Based Practice: Model and Guidelines
. Indianapolis, IN: Sigma Theta Tau International; 2012.
24. Leatherman S, Berwick D, Iles D, et al. The business case for quality: case studies and an analysis. Health Aff
. 2003; 22( 2): 17–30.