Pilot of a Pathway to Improve the Care of Imminently Dying Oncology Inpatients in a Veterans Affairs Medical Center

544 Journal of Pain and Symptom Management Vol. 29 No. 6 June 2005 Original Article Pilot of a Pathway to Improve the Care of Imminently Dying Oncology Inpatients in a Veterans Affairs Medical Center Carol A. Luhrs, MD, Sheni Meghani, MD, Peter Homel, PhD, Mary Drayton, NP, Eileen O Toole, RN, Maria Paccione, MSW, Louisa Daratsos, CSW, David Wollner, MD, and Marilyn Bookbinder, PhD, RN VA-New York Harbor Healthcare System (C.A.L., S.M., M.D., E.O. T., M.P. L.D., D.W.), Brooklyn, New York; Bronx/New York Harbor Geriatric Research, Education and Clinical Center (C.A.L., D.W.), Bronx, New York; and Department of Pain Medicine and Palliative Care (P.H., M.B.), Beth Israel Medical Center, New York, New York, USA Abstract We report on the implementation of a previously developed clinical pathway for terminally ill patients, Palliative Care for Advanced Disease (PCAD), on a Veterans Administration (VA) acute care oncology unit, comparing processes of care and outcomes for patients on and off the pathway. The PCAD pathway is designed to identify imminently dying patients, review care goals, respect patients wishes, assess and manage symptoms, address spirituality, and support family members. Retrospective chart reviews from 15 patients who died on PCAD, 14 patients who died on general wards during the same time, and 10 oncology unit patients who died prior to PCAD revealed that PCAD patients were more likely to have documentation of care goals and plans of comfort care (P 0.0001), fewer interventions, and more symptoms assessed (P 0.004), and more symptoms managed according to PCAD guidelines (P 0.02). Implementation of PCAD improved care of dying inpatients by increasing documentation of goals and plans of care, improving symptom assessment and management, and decreasing interventions at the end of life. J Pain Symptom Manage 2005;29:544 551. 2005 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Clinical pathways, symptom control, goals of care, end-of-life care, supportive oncology care Introduction Optimal end-of-life (EOL) care for hospitalized patients should provide comfort, respect for patient preferences, use of interventions Address correspondence to: Carol A. Luhrs, MD, VISN 3, VA-New York Harbor Healthcare System, 800 Poly Place, Brooklyn, NY 11209, USA. Accepted for publication: February 9, 2005. 2005 U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. appropriate for the goals of care, and support for the family. The Palliative Care for Advanced Disease (PCAD) clinical pathway was developed by Bookbinder et al. at Beth Israel Medical Center in New York City (www.stoppain. org) to raise the standard of care for imminently dying inpatients. 1,2 This report describes a validation study of this clinical pathway, conducted on an acute care oncology unit in a Veterans Administration (VA) medical center. 0885-3924/05/$ see front matter doi:10.1016/j.jpainsymman.2005.02.010

Vol. 29 No. 6 June 2005 Improving Care of Dying Patients 545 Although 65 80% of people die in institutions, few hospitals employ standardized quality improvement (QI) methodologies to monitor and improve EOL care. Well-documented deficiencies in EOL care 3,4 have been the impetus for an Institute of Medicine report 5, which has called for the development of system-oriented approaches to address errors of omission and errors of commission. 6 QI methodologies provide a model to structure and support system change and are well-suited to the implementation of a system-based clinical pathway to improve care of the imminently dying. 7 9 We describe the translation of PCAD for use in a VA medical center and present the results of its implementation on an acute care oncology unit. Novel features of PCAD include computerized templates, interdisciplinary monitoring, a computerized reminder to the team to notify them of the patient s status, and feedback to improve staff performance. Methods Following Institutional Review Board approval, the pilot was conducted over an 18- month period (June 2002 through November 2003). Of the 119 patient deaths occurring on the 24-bed oncology unit, 77 (64.7%) died while receiving care according to the PCAD clinical pathway. Following Institutional Review Board approval, a medical record review was conducted of the following groups: 15 consecutive patients who died on the oncology unit on PCAD from March to May 2003 (designated PCAD patients ); 14 consecutive patients who died on general medicine wards from the same time period (designated non-pcad or control group) and 10 consecutive patient deaths from the oncology unit prior to the initiation of PCAD (designated pre-pcad or baseline group). Implementation of PCAD QI Team. An interdisciplinary QI team of ten core members from the oncology unit (two physicians, two nurses, two social workers, and one psychologist, pharmacist, dietitian, and chaplain) was assembled to review and modify the PCAD clinical pathway for use in the VA and its computerized medical record (CMR). The PCAD Clinical Pathway. The pathway s three structural components: (1) care path, (2) daily flowsheet, and (3) physician order sheet, and the instructional manual, including a flowchart of the PCAD process, were downloaded from the Web site, www.stoppain.org, integrated into sections of the documentation templates and used for in-servicing of staff. Modifications of the original pathway included a) the preparation of discipline-specific templates for the CMR, b) integration of a computer alert (reminder) to members of the clinical team when a patient was placed on the pathway, and c) development of a guideline for the management of common symptoms. Additionally, the QI team devised a family satisfaction survey and a feedback questionnaire for physicians. The discipline-specific templates for pharmacy, dietary, chaplaincy, and psychology were designed to promote more detailed assessments appropriate to the dying patient and to improve documentation in the medical record. The nursing template (Figure 1) was derived from the nurses daily PCAD flowsheet and the social work template was based on guidelines for psychosocial evaluation at the end of life. 10 Symptom Management Guidelines. Guidelines for the pharmacologic management of common symptoms were evidence based and derived from the original PCAD. These guidelines were posted in areas designated by the staff: the physicians office and nursing station. Modifications of these guidelines were based on availability and cost of medications in the VA and were felt by all clinicians on the QI Team to represent best practice in EOL care. Educating the Team and Clarifying Roles. All clinical staff were educated on the overall goals of the pathway, the computer reminder alert, the use of the templates and the symptom management guidelines. Weekly interdisciplinary unit meetings helped to clarify each discipline s role when a patient was placed on the pathway. The attending physician s role was to review goals of care with the patient and family and reevaluate the need for laboratory and imaging, intravenous and other therapies. In some cases, all therapies were discontinued and in others, treatment continued, based on physician, patient, or family preferences. Nurses completed

546 Luhrs et al. Vol. 29 No. 6 June 2005 Fig. 1. Portions of the computerized nursing PCAD template. the PCAD nursing templates, which included detailed symptom assessment and interventions to provide physical comfort for the patient and emotional support for the family as well as the patient. The psychologist and social worker provided support for patient and family, assessed bereavement needs and coping skills, discussed practical issues and participated in behavioral and cognitive therapies. The chaplain s role was to provide spiritual support for the patient and family. The pharmacist re-evaluated the patient s medication regimen and assisted the physician in identifying unneeded medications. The dietitian re-evaluated the diet and educated the patient and family about feeding issues at the end of life. Identification of PCAD Patients. As in the original PCAD pathway, the patient could be identified by any member of the team as likely to die during the hospitalization. Initially, patients were considered eligible for PCAD by an affirmative answer to the question Is this patient likely to die during this admission? For some patients, the members of the healthcare team found this question difficult to answer and another question was posed, Is this patient a candidate for comfort care? If the attending physician agreed that the patient was eligible for PCAD, the pathway was initiated by a physician order. The attending physician and, in most cases, another member of the treatment team met with the patient and family to discuss prognosis, clarify goals of care, explore the patient s and family s wishes and preferences, and generate a treatment plan to meet the goals of care. Consent. Patient or surrogate consent for PCAD was not required because initiation of the pathway did not alter medical care but enhanced supportive measures considered best practice for EOL care. In some cases, patients and families wanted all aggressive measures continued, including resuscitation, and their wishes were respected. Communication. When PCAD was initiated for a patient, a computer alert was sent to all members of the interdisciplinary team. Team members were also notified of the initiation of the pathway by e-mail, morning report, rounds, and in team meetings. PCAD patients were discussed in team meetings twice weekly. Nursing assessments were completed on each shift and the nurses notified physicians if a new symptom was noted or if a symptom was not controlled. Two weeks after a PCAD patient died, a bereavement card and family satisfaction survey were mailed to the patient s family by the social worker. Evaluation-Feedback Tools End-of-Life Chart Audit Tool. The Beth Israel End-of-Life Chart Abstraction Tool (BI EOL- CAT) evaluated processes of care and outcomes

Vol. 29 No. 6 June 2005 Improving Care of Dying Patients 547 on three patient groups: 15 consecutive patients who died on the oncology unit on PCAD from March to May 2003 (designated PCAD patients ); 14 consecutive patients who died on general medicine wards from the same time period (designated No PCAD ); and 10 consecutive patient deaths from the oncology unit prior to the initiation of PCAD (designated Pre PCAD ). The BI EOL-CAT elicits information in six areas: demographics, admission and transfer data, decision-making at the end of life, symptom assessment and management, death event, and healthcare resource utilization. To minimize inter-rater variability, one physician, who was not a member of the treatment team, reviewed all the charts. The Family Satisfaction Survey. A 15-item family survey, adapted from surveys used by community hospices, was designed by the QI Team to elicit responses from families about participation in decision-making, goals of care, satisfaction with care, and the control of pain and other symptoms. The surveys were mailed to families of PCAD patients six weeks after the patient s death. The results of these surveys were not tabulated due to the poor return, with only 16 surveys returned, representing 22% of the 77 PCAD patients. Physician Questionnaire. This questionnaire, completed when a patient died off PCAD on the oncology unit, elicited the attending physician s opinion on whether the patient was terminally ill, likely to die during the hospitalization and whether, in retrospect, the patient should have been on PCAD. Statistical Analysis Normally distributed data (e.g., age) were characterized as mean standard deviation, whereas skewed data (e.g., length of stay) were characterized as median (minimum, maximum). Categorical data (e.g., diagnosis) were characterized as frequency (percent). The student s t-test and one-way analysis of variance were used to compare patient groups with regard to normally distributed outcomes and the Mann-Whitney and Kruskal-Wallis tests were used for skewed outcomes. Fisher s exact and chi-square tests were used for comparisons involving categorical outcomes. The No PCAD and Pre PCAD groups tended to be very similar and so these groups were combined for comparison with the PCAD patients. The combined group is designated as Non-PCAD. Comparisons between the Pre PCAD and No PCAD groups are also provided where appropriate. All comparisons were done using a significance level of 0.05. SPSS 11.5 (SPSS Inc, Chicago) was used for all analyses. Results The characteristics of patients whose charts were reviewed using the BI EOL-CAT are summarized in Table 1. The patients in the three groups did not differ significantly in age, diagnosis, or number of previous hospital admissions. The primary diagnoses of the Pre PCAD and PCAD patients was cancer in 100% and 87%, respectively. However, there was a significant difference in the number of patients with a diagnosis of cancer for No PCAD patients from acute medicine, as expected. Median length of stay also tended to be longer for the PCAD patients (23 days vs. 10 for the other two groups; P 0.04). Decision Making at the End-of-Life The identification of the imminently dying patients and documentation of goals of care are seen in Table 2. Documentation of impending death was noted in 41.6% of Non-PCAD (No Pre PCAD) patients and 100% of patients on PCAD. A review of the goals of care was documented in 33.3% of Non-PCAD and 66.7% of PCAD patients and an explicit plan of comfort care in 4.2% of Non-PCAD and 100% of patients dying on PCAD (P 0.0001). Healthcare Resource Utilization Fewer interventions were noted in the last three days of life in PCAD patients compared to Non-PCAD patients. Cardiopulmonary resuscitation (CPR) was attempted in 29.2% of Non- PCAD patients and 6.7% of PCAD patients. Other interventions were also more common in the Non-PCAD patients, including intravenous fluids (75% Non-PCAD, 20% PCAD), intravenous antibiotics (58.3% Non-PCAD, 20% PCAD), and blood drawing (91.6 % Non-PCAD and 0% PCAD). Wrist restraints were documented in 16% of patients who died off PCAD and 7% of patients who died on PCAD within the last three days of life.

548 Luhrs et al. Vol. 29 No. 6 June 2005 Table 1 Patient Characteristics No PCAD Pre PCAD PCAD (medicine) (oncology) (oncology) Characteristic n 14 n 10 n 15 P-Value Age, yrs 73 9 71 12 72 14 0.91 Two or more hospitalizations in past six months m, (%) 6 (43) 3 (30) 5 (33) 0.78 Admitted from nursing homes or other hospitals, n (%) 9 (64) 2 (20) 6 (40) 0.09 Primary cancer diagnosis, n (%) 6 (43) 10 (100) 13 (87) 0.003 Length of stay, days n (range) 10 (2, 75) 10 (3, 40) 23 (4, 94) 0.04 All restraints and intravenous antibiotics were withdrawn on PCAD patients during the last 3 days. In contrast, restraints were withdrawn on none of the Non-PCAD patients and intravenous antibiotics were withdrawn on only 14.3% of these patients. Similarly, intravenous hydration was withdrawn more often on PCAD than Non-PCAD patients during the last 3 days (66.7 vs. 22.2%). Medication review and withdrawal of medications not considered to be consistent with the goals of care was documented in 40% of patients and in none of Non-PCAD patients. Because these comparisons involve small groups with zero frequencies of occurrence, P-values are not reported in Table 2. Symptom Assessment and Management During the Last 3 Days of Life PCAD patients were assessed for symptoms more frequently than Non-PCAD patients: median of 6 symptoms per patient versus 4 symptoms for Non-PCAD patients (P 0.004). Although some of the symptoms assessed were not problematic, the number of symptoms which required management using PCAD guidelines was higher (P 0.02). Feedback Evaluation The completion of PCAD templates by the seven disciplines (physician, nurse, social work service, chaplain, pharmacist, psychologist, and dietitian) was monitored as a measure of staff acceptance of the pathway and the extent of implementation into routine practice. Completion rate was defined as at least one template completed by the discipline during the patient s length of stay (LOS) on PCAD. Figure 2 demonstrates the completion of templates for the first 20 PCAD patients. Template completion was poor for the first 8 patients with three or fewer templates completed (0 to 42.8%). For the next 12 patients, completion of templates improved, Table 2 Selected CAT Items Non PCAD Patients (n 24) No PCAD Pre-PCAD P-value No PCAD PCAD Patients P-value (medicine) (oncology) No PCAD vs. Pre PCAD (oncology) Non-PCAD CAT Item n 14 n 10 Pre PCAD n 24 n 15 vs. PCAD Decision-making at the end of life Patients identified as dying, n (%) 5 (35.7) 5 (50) 1.0 10 (41.6) 15 (100) 0.0001 Review of goals of care, n (%) 6 (42.9) 2 (20) 0.39 8 (33.3) 10 (66.7) 0.06 Plan of comfort care, n (%) 1 (7.1) 0 (0) 1.0 1 (4.2) 15 (100) 0.0001 Interventions in last 3 days of life CPR, n (%) 5 (35.7) 2 (20) 0.65 7 (29.2) 1 (6.7) 0.12 Intravenous fluids, n (%) 13 (92.9) 5 (50) 0.05 18 (75) 3 (20) 0.001 Intravenous antibiotics, n (%) 9 (64.3) 5 (50) 0.68 14 (58.3) 3 (20) 0.02 Blood draws, n (%) 13 (92.9) 9 (90) 1.00 22 (91.6) 0 (0) 0.0001 Restraints, n (%) 3 (21.4) 1 (10) 0.61 4 (16.6) 1 (6.7) 0.63 Interventions withdrawn in last 3 days Restraints, n (%) 0/3 (0) 0/1 (0) NA 0/4 (0) 1/1 (100) NA Intravenous antibiotics, n (%) 1/ 9 (11.1) 1/5 (20) NA 2/14 (14.3) 3/3 (100) NA Intravenous fluids, n (%) 3/13 (23.1) 1/5 (20) NA 4 /18 (22.2) 2/3 (66.7) NA Medication(s) reviewed and/or 0 (0) 0 (0) NA 0 (0) 6 (40) NA withdrawn, n (range) Total number of symptoms assessed 4 (2, 8) 4.1 (2, 7) 0.61 4 (2,8) 6 (3, 12) 0.004 Total number of symptoms managed 0 (0, 3) 1 (0, 4) 0.29 0.5 (0,4) 1 (0,5) 0.02

Vol. 29 No. 6 June 2005 Improving Care of Dying Patients 549 Number of templates completed 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 1112 1314151617181920 Patient number Fig. 2. Number of templates completed by the seven disciplines for the first 20 patients on PCAD. with 5 to 7 templates (71.4 to 100%) completed for the majority of patients. Patients 11 and 17 on Figure 2 were on PCAD for one day, suggesting that poor documentation in these patients was most likely due to their short LOS on PCAD. Attending physician feedback questionnaires were completed for 33 of the 42 patients (78.6%) who died off PCAD on the oncology unit. The attending physicians indicated that 32 patients (97%) who died off PCAD were terminally ill and for 23 patients (70%), the attending physician was not surprised the patient died during the hospitalization. For 16 patients (48%), the attending indicated that, in retrospect, the patient should have been placed on PCAD. The three major reasons given for these patients not being on PCAD: (1) the patient was expected to go home; (2) the hospitalization was too short to initiate PCAD; and (3) the patient or his family wanted the patient to have CPR. Because patients were eligible for PCAD even if they wanted everything to be done, including resuscitation, this result indicated to the Team that additional education of the attending physicians was needed. These results were discussed at monthly staff meetings and provided important feedback to the physicians. Discussion The PCAD pathway was developed by Bookbinder et al. 2,7 at Beth Israel Medical Center in response to the need to improve EOL care for inpatients in acute care settings. Although the pilot study conducted at Beth Israel did not confirm that PCAD drove the improvement seen in the care of the dying inpatient, positive trends were noted in the documentation of outcomes related to EOL care. 2 The results of this validation study, in which PCAD was implemented on an oncology unit in a VA Medical Center, adds further support to the feasibility and effectiveness of PCAD in improving EOL care in the acute care setting and provides preliminary evidence of the positive effects of PCAD on patient and system outcomes. The establishment of a QI interdisciplinary team to lead the modification of the PCAD pathway and the education of the clinical staff assisted in the successful implementation of this clinical pathway on the oncology unit. The majority of patients (64.7%) who died on the oncology unit from June 2002 to November 2003 were placed on PCAD. Another indicator of acceptance of the pathway by staff on the unit was completion of the discipline-specific templates. The majority of these templates were completed for most patients after the ninth patient was entered on PCAD. One of the most important implementation strategies in the original pilot at Beth Israel, in which PCAD was implemented on three study units, was the identification of a local unit champion, a nurse or physician leader to interact on a regular basis with the unit staff. Although in the VA study PCAD was implemented on only one unit, the commitment of nursing and physician leadership was a key factor in the success of implementation of the pathway. Of note is the importance of physician education and compliance, as they played a central role in the implementation of PCAD. Attending physicians were required to initiate the order for PCAD, to conduct the family meeting, to review goals of care and to complete a questionnaire if a patient died off the PCAD pathway. The attending feedback questionnaire may have enhanced early acceptance of PCAD by the physicians. The attending physicians indicated that 97% of patients who died off PCAD were terminally ill and for 70% of these patients, physicians were not surprised that the patient died during the hospitalization. In retrospect, the physicians felt that 48% of these patients should have been placed on PCAD. The completion of this questionnaire allowed physicians to review the patient s hospital course in the context of his terminal illness and may have

550 Luhrs et al. Vol. 29 No. 6 June 2005 enhanced the use of PCAD. In addition, staff meetings provided feedback to the physicians on their awareness and familiarity with PCAD, addressing some of the barriers known to affect physician acceptance of clinical pathways. 11,12 Acceptance of PCAD represented an important change in clinical behavior on the oncology unit but this did not contribute to discernable improvement in the care of the dying inpatient on the acute medicine wards. This finding is supported by the analysis of patients who died on acute medicine nine months after PCAD was initiated on the oncology unit. In contrast, PCAD seemed to have contributed to the way dying inpatients were managed throughout the hospital in the Beth Israel pilot. 2 In that study, it was suggested that system change implemented on a few units may be sufficient to provide diffusion and wider improvement in care and that the presence of a Palliative Care Team may have contributed to this spread. The experience in this VA study does not support this finding, as the improvements were only seen on the oncology unit where PCAD was implemented. The implementation of PCAD resulted in improvement in several aspects of EOL care: documentation of a review of goals of care and an explicit plan of comfort care, the number of interventions in the last three days of life, and symptom assessment and management. Analysis of chart review data from patients who died on PCAD (PCAD patients) and off PCAD (Non- PCAD patients) demonstrates a statistically significant difference in the number of symptoms assessed and managed according to PCAD guidelines. Future studies might benefit from using standardized symptom assessment scales that quantify symptoms and include global indices of symptom control. Barriers to the implementation of PCAD included the complexity of the discipline-specific computerized templates, the need for continued education of the staff, the reluctance to initiate the pathway on weekends or holidays and the initial difficulty in identifying PCADeligible patients. Initially, patients were considered eligible for PCAD by an affirmative answer from any member of this team to the question Is this patient likely to die during this admission?. The identification of patients was made easier by changing this question to Is this patient a candidate for comfort care?. The increased length of stay for patients on PCAD was unexpected and cannot be readily explained by the patient population studied. One possibility is that physicians were reluctant to discharge imminently dying patients to avoid changing the venue of care and the healthcare team. Future studies of this clinical pathway, which will include standardized surveys to evaluate family perception of care, may demonstrate that quality of EOL care was improved despite the longer length of stay. This study of the PCAD pathway has several limitations. The family satisfaction surveys could not be evaluated due to poor return, with only 22% of the families of PCAD patients responding. To improve the response of families, future studies of PCAD will include a standardized, validated telephone survey of families. Revisions of the pathway are underway and include simplification of the discipline-specific templates, standardization of patient and staff educational materials, preparation of more detailed algorithms for symptom management and a more structured approach to bereavement follow-up and grief counseling. Our assessment approach also has limitations. The BI EOL-CAT does not directly assess patient outcomes but focuses on process variables and retrospective outcome variables. Future studies testing the PCAD intervention should have more specific outcome measures of symptom management, the death experience, 13 and family perspectives of EOL care. 14,15 The recently published National Consensus Guidelines for Quality Palliative Care 16 presents guidelines for EOL care, with goals similar to those of PCAD. Among the important issues addressed in these guidelines are the recognition of the imminently dying patient, symptom control, the interdisciplinary approach to patient care, attention to the emotional, psychological and spiritual issues of the patient and family, and patient preferences. The PCAD intervention, which addresses many of these issues, offers a structured approach to disseminate these best practices and to promote institutional change to raise the standard of care of the dying. Acknowledgments The authors wish to thank the clinical and administrative staff of the Brooklyn campus of

Vol. 29 No. 6 June 2005 Improving Care of Dying Patients 551 the VA-New York Harbor Healthcare System for their commitment and support of this project. We especially thank all of the nurses and physicians on the oncology unit and the following members of the team: Jennifer Amato, RD, Joseph Vitanza, PharmD, Daniel Garry, Quality Management, and Fr. Thomas Mullin. References 1. Bookbinder M, Romer A. Raising the standard of care for imminently dying patients using quality improvement. Innovations in End of Life Care 2001;3(4), www.edc.org/lastacts. 2. Bookbinder M, Blank A, Arney E, et al. Improving end-of-life care: development and pilot-test of a clinical pathway. J Pain Symptom Manage 2005; 29(6):529 543. 3. SUPPORT Principal Investigators. A controlled trial to improve care for seriously ill hospitalized patients: The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT). JAMA 1995;274:1591 1598. 4. Weeks J, Cook F, O Day SJ, Peterson L, et al. Relationship between cancer patients predications of prognosis and their treatment preferences. JAMA 1998;2789:1709 1714. 5. Institute of Medicine (IOM). In: Field MJ, Cassel CK, eds. Approaching death: Improving care at the end of life. Committee on care at the end of life, Division of Health Care Services, Institute of Medicine. Washington DC: National Academy Press, 1997, pp 259 271. 6. Shortell S, Bennett C, Byck GR. Assessing the impact of continuous quality improvement on clinical practice: What it will take to accelerate progress? Milbank Quarterly 1998;76:593 624. 7. Bookbinder M. Improving the quality of care across settings. In: Ferrell B, Coyle N, eds. Textbook of palliative nursing. Oxford: Oxford University Press, 2001:503 530. 8. Gordon DB. Critical pathways: a road to institutionalizing pain management. J Pain Symptom Manage 1996;11:252 259. 9. Ellershaw J. Clinical pathways for care of the dying: an innovation to disseminate clinical excellence. Innovations in End-of-Life Care 2001;3(4), www.edc.org/lastacts. 10. Scott L, BrintzenhofeSzoc KB, Shaffer H. Developing effective tools for social work documentation within a changing healthcare system. Poster presentation. Association of Oncology Social Work 17th Annual Conference, Cleveland, Ohio, May 2001. 11. Cabana MD, Rand CS, Powe NR, et al. Why don t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999;282: 1458 1465. 12. Cabana MD, Rushton JL, Rush AJ. Implementing practice guidelines for depression: applying a new framework to an old problem. Gen Hosp Psychiatry 2002;24:35 42. 13. Ternestedt BM, Andershed B, Eriksson M, Johansson I. A good death: development of a nursing model of care. J Hospice Palliat Nurs 2002;4:153 160. 14. Teno JM, Clarridge BR, Casey V, et al. Family perspectives on end-of-life care at the last place of care. JAMA 2004;291(1):88 93. 15. Teno J. Center to improve care of the dying. Toolkit of instruments to measure end-of-life. Available from http://www.gwu.edu/. Accessed August 14, 1998. 16. Clinical guidelines for quality palliative care, 2004. Available from www.nationalconsensusproject. org. Accessed 2004.