ORIGINAL STUDIES

Evaluation of a Continuous Quality

Improvement Initiative for End-of-Life
Care for Older Noncancer Patients
Jean Woo, MD, Joanna O. Y. Cheng, MSc, Jenny Lee, MRCP, Raymond Lo, MD, Elsie Hui, FRCP, C. M. Lum, FRCP,
K. H. Or, FRCP, Fanny Yeung, MSc, Florens Wong, BA, and Benise Mak, PhD

Objectives: The burden of suffering among patients measures for patients; quality-of-life and cost-of-
with end-stage chronic diseases may be greater than care index for family members; quality-of-life and
those of cancer patients, as a result of longer duration carer burden for staff; and use of various health
of illness trajectory and high prevalence of symptoms, care services.
yet they may be less likely to receive palliative care
services. To improve the quality of care of these Results: There were 80 and 89 participants in the pre-
patients, we carried out a continuous quality and post-intervention phase. The initiative resulted in
improvement initiative among medical and nursing shorter duration of stay, fewer investigations, fewer
staff of a convalescent facility. transfers back to the affiliated acute care hospital,
and more follow-up by the outreach team, with no
Design: Evaluation of a quality improvement initiative. significant difference in mortality after adjusting for
age and comorbidity. Symptoms of pain and cough
Setting: Nonacute institution in Hong Kong SAR, were reduced, while there was a trend toward more
China. constipation but less dizziness. Family members’ satis-
faction improved.
Participants: The participants were patients with
advanced chronic diseases not opting for active Conclusions: It is possible to improve quality-of-life
treatment. care for elderly patients with end-stage chronic
diseases by staff education, and culture and system
Intervention: The intervention was a continuous qual- change, not only without additional resources, but
ity improvement process carried out over a 3-month likely savings were achieved in terms of reduced use
period, consisting of service reengineering, provision of health care resources. (J Am Med Dir Assoc 2011;
of guidelines and educational material, and interac- 12: 105–113)
tive sessions to achieve culture change among staff.
Evaluation before and after the intervention included Keywords: End-of-life care; chronic disease; elderly;
patient symptoms checklist and quality-of-life continuous quality improvement

An increase in chronic disease burden inevitably accom-

panies population ageing, and increasing attention has been
Department of Medicine and Therapeutics and Department of Medicine and placed on end-of-life care for end-stage chronic diseases
Geriatrics, Shatin Hospital, The Chinese University of Hong Kong, Hong Kong such as chronic obstructive airways disease,1 heart failure,2 re-
SAR, China (J.W., J.O.Y.C., J.L., R.L., E.H., C.M.L., K.H.O., F.Y., F.W.); Cadenza: A
nal failure,3 neurodegenerative diseases,4 and dementia.5 The
Jockey Club Initiative for Seniors, Faculty of Social Sciences, The University of
Hong Kong, Hong Kong SAR, China (B.M.). burden of suffering among these patients may be greater than
This study was supported by the Hong Kong Jockey Club Charities Founda-
those of cancer patients, as a result of longer duration of
tion Cadenza Project. illness trajectory and high prevalence of symptoms.6,7
Address correspondence to Jean Woo, MD, Department of Medicine & Ther-
Although palliative care services have been developed
apeutics, Prince of Wales Hospital, Shatin, N.T., Hong Kong. E-mail: initially for cancer patients, with an emphasis on symptom
jeanwoowong@cuhk.edu.hk control and psychosocial and spiritual needs,8,9 in recent
Copyright Ó2011 American Medical Directors Association years guidelines of good practice for end-stage chronic
DOI:10.1016/j.jamda.2010.02.011 diseases have also been published.10,11

ORIGINAL STUDIES Woo et al 105

 However, uncertainties exist regarding when to initiate the same criteria was recruited. Patients recruited in the
such care pathways, and in general end-of-life care is started preintervention phase were not recruited again in subsequent
only in close proximity to death. It has been pointed out that readmissions.
end-of-life care should move away from a prediction of death Patients were screened for recruitment by senior geriatri-
to the creation of an environment of active management of cians. Patients (and relatives) not opting for aggressive/inva-
symptoms and end-of-life planning by the care team,12 and sive treatment with a diagnosis of advanced diseases (refer to
that palliative care should be both a philosophy of care as Appendix 1 for a list of prognostic indicators) were included
well as an organized, highly structured system for delivering in the study. Their informal family caregivers, if available,
care.13 A recent review of literature on end-of-life care for were invited by research assistants to participate on a volun-
older people highlighted the need for further research in cul- tary basis. Informal caregivers were defined by the status of
tures other than the United States, the needs of people older the official ‘‘next-of-kin’’ or proxy as recorded in the nursing
than 80 years, the preferences regarding end-of-life care, and notes. Research assistants approached patients and caregivers
the effect of generalizable palliative care models.14 to introduce the purpose of the study and obtained written in-
Although a quarter of Americans die in nursing homes, phi- formed consent from them. Consent-by-proxy was obtained
losophy and policies in long-term care facilities were targeted from caregivers of noncommunicable patients. Ethical
toward rehabilitation and improving function, and research in approval for the study was obtained from the Joint Clinical
end-of-life care in these settings were limited.15 In Hong Research Ethics Committee of the Chinese University of
Kong, dying at home or in a nursing home is not an available Hong Kong and the New Territory East Cluster of the Hospi-
option. Most deaths of the chronically ill occur in convales- tal Authority of Hong Kong.
cent facilities, where doctors and nurses play important roles
in the last stage of life. However, the focus of such facilities EVALUATION QUESTIONNAIRES PRE AND POST
was on rehabilitation and restoration of function, rather INTERVENTION
than end-of-life care. To improve the quality of care of pa-
Patients
tients with end-stage chronic diseases, we carried out a contin-
uous quality improvement initiative among medical and Two separate sets of questionnaires were used for communi-
nursing staff of a rehabilitation and convalescent facility. cable and noncommunicable (comatose/aphasic/cognitively
The process consisted of a series of seminars, workshops with impaired) patients. The set of questionnaires for communica-
role play, focus groups, audit reports, development of guide- ble patients included a palliative care Symptom Check List
lines, and new care pathways over a 3-month period. Evalua- (SCL) (0 5 no symptoms to 4 5 unbearable symptoms) mod-
tion was carried out before and after the initiative to document ified from the Support Team Assessment Schedule (STAS),16
feasibility, impact on patients’ symptom control, and health the Pain Assessment Chart (PAC),16 (range from 0 5 no pain
services use, as well as family caregiver and staff feedback. to 4 5 extreme pain) noting pain control, the Chinese Geri-
atric Depression Scale-417 (minimum 0, maximum 4), the ab-
SUBJECTS AND METHODS breviated Chinese Death Anxiety Inventory (CDAI)18
The initiative was carried out in a large rehabilitation and adapted for this study (minimum 5 0, maximum 5 4 for
convalescence unit in Hong Kong, consisting of 227 beds, 1 each item), the McGill Quality of Life Questionnaire—
geriatrics day hospital, and 1 palliative day hospital. All pa- Hong Kong version19 (minimum 5 0, maximum 5 10 for
tients were transferred from an affiliated university acute each item), and a 1-item, 10-point Likert satisfaction scale
care hospital after an average of 4 to 5 days stay for rehabili- on care quality. For noncommunicable patients, the question-
tation, convalescence, or step-down medical care. Doctors in naire consisted of only the SCL, with additional items on be-
charge were senior geriatricians and internal medicine resi- havioral expressions of depression, fear, anxiety, agitation,
dents without prior palliative care training. Nursing staff resistance to care, and pain as reported by nurses (proxy).
had had no palliative care training as well. Research assistants sought advice from nurses regarding
A quasi-experimental nonequivalent groups design was patients’ communicative and cognitive status and initiated
used. The project was run in 3 phases: preintervention, casual chats with patients to determine their ability to
wash out, and intervention. In the preintervention phase, complete the questionnaire. Questionnaires were delivered
patients who fulfilled criteria for end-stage chronic diseases through direct interviews with communicable patients and
and their family caregivers were recruited. They completed proxy ratings by nurses were obtained for noncommunicable
patient symptoms and caregiver assessments and received ones. Sociodemographic variables (age, sex, marital status,
none other than existing treatment. This was followed by educational status) and health-related variables were also
a wash-out phase of 3 months, during which assessments collected (underlying end-stage diseases, duration of stay in
stopped and staff training (intervention) took place. Three the facility before discharge, discharge destination, and
months was chosen arbitrarily because of limited time with referral to support services after discharge).
funding support. This duration was also deemed adequate
for the ‘‘wash-out period’’ between usual practice and the Family Caregivers
implementation of the quality improvement initiative, as Questionnaires for informal caregivers included the Short-
the average duration of stay rarely exceeds 60 days. After Form-12 (SF-12),20 a quality-of-life measure with a physical
the intervention, a nonoverlapping batch of patients fulfilling and a mental component, the Chinese version of the ‘‘Cost

106 Woo et al JAMDA – February 2011

Table 1. Comparisons of Patient Characteristics and Investigations Received during Current Stay (All Patients)
Items Preintervention (n 5 80) Postintervention (n 5 89) P Value
Female sex 55 (68.8) 54 (60.7) .273
Age, y, mean (SD) 83.5 (9.3) 85.1 (8.5) .164
Diagnosis
COPD 10 (12.5) 8 (9.0) .460
CHF* 14 (17.5) 5 (5.6) .015
CRF 13 (16.3) 19 (21.3) .398
Liver disease 2 (2.5) 7 (7.9) .174
Dementia 46 (57.5) 56 (62.9) .472
Stroke* 17 (21.3) 5 (5.6) .003
Other chronic neurological conditions 1 (1.3) 0 (0.0) .473
Marrow failure 2 (2.5) 3 (3.4) 1.000
Duration of current stay in convalescent facility (days) * 40.5 (56.0) 29.3 (22.0) .012
No. of blood specimens collected after recruitment* 8.6 (8.8) 4.8 (7.1) .000†
No. of x-rays taken after recruitment* 1.2 (1.6) 0.5 (0.9) .000†
Mean follow-up until death or Apr 2, 2009 (days)* 183.8 (20.2) 71.3 (7.86) \.001
Discharge destination
Home 18 (22.5) 18 (20.2) .718
Old age home 30 (37.5) 28 (30.5) .409
Transfer to acute care hospital* 10 (12.5) 0 (0.0) .000
Deceased* 22 (27.5) 43 (48.3) .005
Numbers are presented as n (%) unless otherwise indicated. A patient from the post-phase group was unclassified, having the diagnosis of
past hip fracture and suffering from pressure sore.
COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; CRF, chronic renal failure.
* P \.05.
† Comparison by t test.

of Care’’ Index (CCCI)21 (minimum 5 20, maximum 5 80) tions and trajectories of various end-stage diseases, common
and a 1-item, 10-point Likert satisfaction scale on the quality end-stage symptoms, and their management. Case vignettes
of care provided for the patient in the facility. Caregivers were presented to illustrate patients’ and relatives’ view of
were assessed within 1 week of patient recruitment. care quality and symptom control. An end-of-life care manual
Research assistants phoned the caregivers of referred pa- with key publications was placed in each nursing station as
tients to arrange a face-to-face questionnaire interview, training reference. Educational leaflets on end-of-life care
away from the patient. Where that could not be arranged, (EOLC), no-resuscitation orders, tube feeding, community
the questionnaires were delivered over the telephone with resources, bereavement and funeral arrangements, and care-
verbal consent. Sociodemographic data (age, sex, marital giver self-care were printed for families and patients. Condo-
status, educational level, occupation, relationship with the lence cards were printed and sent out 1 month after patients’
patient), health-related data (history of psychiatric illness, deaths. Forms to guide symptom monitoring, interventions,
present health complaints) and caregiving-related data (aver- and investigations were included in patients’ files
age hours per week spent on caregiving, participation in (Appendix 1). A care pathway was developed with on-
caregiver support service) were included. going review and consultation with staff. A new care category
(Med O) for EOLC was created to allow easy follow-up of pa-
Health Care Professionals tients. Staff were reminded by forms and case conferences to
The questionnaire consisted of the Chinese Maslach Burn- use new discharge options included telephone consultations
out Inventory (CMBI)22 (best mean score 51, worst 5 4) with with the parent ward staff, medical consultations in day hos-
3 subscales: emotional exhaustion, lack of personal achieve- pitals, direct admission to the same ward within the facility,
ment, and depersonalization, the Chinese Death Anxiety In- and referral to community geriatric nurses for follow-up of
ventory (CDAI),18 with a 12-item subscale on death and symptoms either at home or in long-term care nursing homes.
dying anxiety and a 11-item subscale on after-death anxiety Discharged Med O patients were entered into the ‘‘High-Risk
(minimum 5 1, maximum 5 5 per item) and the SF-12.20 Elderly Registry’’ in the computerized patient record system of
the Hong Kong Hospital Authority to alert the medical team
THE QUALITY IMPROVEMENT INTERVENTION of the affiliated acute care hospital of the EOLC status. These
The improvement model adopted the Institute for Health- high-risk patients were given priority bed-booking for early
care Improvement model with Plan-Do-Study-Act cycles23 transfer back to the convalescence facility for EOLC in their
and consisted of 3 cycles. First, a consensus meeting was subsequent readmissions, preferably to the same ward, for
held in which senior doctors agreed to the importance of im- continuity of care.
proving EOL care in the facility. This was followed by staff The next phase consisted of implementation of the plan
training and care pathway development. Each senior geriatri- on different wards, led by the physician in charge, in iden-
cian gave a seminar on specific chronic diseases for nursing tifying patients suitable for end-of-life care and checking
and junior medical staff. Seminar contents included defini- that the care plan was adhered to. Feedback sessions

ORIGINAL STUDIES Woo et al 107

Table 2. Comparison of Patient Symptoms in Pre- and Postintervention Phases and the Association between Patient Symptoms and the Interventions,
With Adjustment for Confounders
Symptoms Comparison of Patient Symptoms Association between Patient Symptoms n (%)†
in Pre- and Postintervention Phases and QI Interventions With Adjustments (No. Assessed)
n (%) P Value Odds 95% Confidence P Value
Ratio Interval
Pain Pre- 39 (48.8) .001* 0.256 0.13, 0.52 \.001* 169 (100%)
Post- 22 (24.7)
Anorexia Pre- 15 (34.1) .410 1.56 0.62, 3.93 .348 96 (56.8)
Post- 22 (42.3)
Nausea Pre- 2 (2.5 .168 3.00 0.54, 16.59 .209 162 (95.9)
Post- 6 (7.2)
Vomiting Pre- 6 (7.5) .120 0.34 0.06, 1.97 .214 166 (98.2)
Post- 2 (2.3)
Dysphagia Pre- 52 (67.5) .749 1.00 0.46, 2.19 .997 160 (94.7)
Post- 58 (69.9)
Oral discomfort Pre- 16 (20.3) .670 0.89 0.38, 2.10 .784 164 (97.0)
Post- 15 (17.6)
Constipation Pre- 27 (34.2) .032* 1.83 0.94, 3.55 .075 168 (99.4)
Post- 45 (50.6)
Diarrhea Pre- 6 (7.6) .397 0.53 0.14, 2.06 .358 168 (99.4)
Post- 4 (4.5)
Cough* Pre- 17 (21.5) .070 0.37 0.15, 0.93 .035* 168 (99.4)
Post- 10 (11.2)
Dyspnea Pre- 26 (32.5) .098 1.51 0.77, 2.95 .230 169 (100.0)
Post- 40 (44.9)
Edema Pre- 30 (37.5) .589 1.25 0.63, 2.46 .524 169 (100.0)
Post- 37 (41.6)
Lower limb Pre- 78 (97.5) .739 0.49 0.06, 3.93 .504 169 (100.0)
weakness
Post- 86 (96.6)
Fatigue Pre- 66 (82.5) .654 1.01 0.41, 2.48 .987 167 (98.8)
Post- 74 (85.1)
Dizziness Pre- 15 (18.8) .027* 0.33 0.11, 1.02 .055 154 (91.1)
Post- 5 (6.8)
Insomnia Pre- 13 (16.2) .479 0.72 0.28, 1.89 .507 161 (95.3)
Post- 10 (12.3)
Depressed mood Pre- 11 (15.5) .064 0.37 0.11, 1.30 .121 140 (82.8)
Post- 4 (5.8)
Fear Pre- 7 (9.9) .197 0.41 0.09, 1.77 .230 141 (83.4)
Post- 3 (4.3)
Anxiety Pre- 5 (7.0) .747 0.92 0.21, 4.01 .909 141 (83.4)
Post- 4 (5.7)
Agitation Pre- 18 (25.7) .187 0.70 0.29, 1.72 .441 142 (84.0)
Post- 12 (16.7)
Resistance to care Pre- 6 (8.5) .095 2.56 0.838, 7.81 .099 150 (88.8)
Post- 14 (17.7)
Adjusted for age, gender, heart failure, stroke.
QI, quality improvement.
* P \.05.
† N is smaller than 100% in some symptoms where assessment by proxy was not possible.

involving all levels of staff were held in small groups in each with the audience. Actors in the role play were nominated
ward, to encourage discussions of any logistic or conceptual by their colleagues as having good communication skills or
difficulties. As a result, modifications of the pathway were having prior experience in palliative care, such as nursing
carried out as needed. Problems with understanding of the staff working in the palliative care unit in the same facility.
end-of-life care concept as well as lack of confidence among
staff in discussing end-of-life care issues with patients and STATISTICAL ANALYSIS
relatives were identified. Based on the results of these discus- The sample size was calculated based on the improvement
sions, the next phase of the Continuous Quality Improve- in pain in a study of cancer pain control by palliative care
ment (CQI) cycle was implemented. This consisted of 2 teams,24 where the effect size was 0.49. Assuming noncancer
communication skills workshops, which included role plays pain might be less amendable to treatment such as opioids,25
of communication scenarios among doctors, nurses, patients, the effect size was reduced to 0.4 for the present study. A sam-
and family caregivers, followed by interactive discussions ple size of 61 per group was estimated to have an 80% chance

108 Woo et al JAMDA – February 2011

Table 3. Comparing Mean Scores of the SF-12, Chinese Cost of Care Index, and Satisfaction Survey of Caregivers and Staff Evaluation
Preintervention Postintervention P Value
Caregiver (n 5 30) (n 5 74)
SF-12
Physical Composite Scale 43.9 (9.2) 46.1 (10.4) .343
Mental Composite Scale 43.6 (10.2) 44.3 (10.1) .755
Chinese Cost of Care Index (min 20, max 80) 45.9 (6.5) 47.1 (7.3) .465
Satisfaction (max score 10)
Caregiver* 7.83 (1.50) 8.56 (1.26) .017
Patient (by proxy)* 7.54 (1.64) 8.24 (1.35) .029
Staff (n 5 118) (n 5 121)
SF-12
Physical Composite Scale 44.73 (8) 44.87 (7.49) .886
Mental Composite Scale 47.26 (9.54) 48.70 (8.53) .229
Chinese Maslach Burnout Inventory (best 5 1, worse 5 4)
Emotional exhaustion 2.52 (0.52) 2.44 (0.52) .201
Lack of personal achievement 1.83 (0.43) 1.79 (0.37) .401
Depersonalization 1.81 (0.59) 1.82 (0.52) .960
Chinese Death Anxiety Inventory (min 5 1, max 5 5)
Death and dying anxiety 3.22 (0.64) 3.26 (6.34) .686
After death anxiety 2.54 (0.61) 2.56 (0.71) .798
Data presented as mean (SD). Comparisons by t test.
SF-12, Short Form 12; min, minimum; max, maximum.
* P \.05.

to detect a difference in pain in the groups (G power 3). A for any of the other dimensions measured on the part of care-
30% attrition rate (because of unexpected death, withdrawal giver or staff (Table 3).
of consent, or becoming too sick to be interviewed) was esti- Among survivors of the index admission, there was in-
mated in this frail population. The sample size for each group creased use of the rapid-response medical clinic in the day
was thus 80. Student t test was used to compare continuous hospital and outreach services, but significantly fewer trans-
variables and the chi-square test was used to compare categor- fers back to the acute care hospital (Table 4). The reductions
ical variables. Cox regression was used to examine the impact in the number of investigations and bed days in both acute
of the intervention on patient survival, adjusting for age, sex, care hospitals and convalescence facilities for all postinter-
and diagnoses that were different in frequency at baseline. vention patients remained significant after adjusting for age,
sex, heart failure, and stroke (Table 5). Among the survivors,
RESULTS apart from significant reductions in acute care and convales-
cence bed days, there was also a marginal reduction in acute
Eighty and 89 subjects were recruited in the pre- and post- care hospital admission (Table 5).
intervention phases respectively. Table 1 shows the patients’
characteristics and number of investigations the patients
received during the 2 phases. There were more patients with DISCUSSION
congestive heart failure and stroke in the preintervention We have demonstrated the feasibility of improving the
phase. The quality improvement initiative resulted in short- quality of end-of-life care for older patients (mean age of 80
ened duration of stay, fewer investigations (blood tests and years) with end-stage chronic diseases in the context of a gen-
x-rays) and fewer transfers back to the acute care hospital. eral convalescence service, without additional resource, by
More patients died during the index admission in the postin- promoting a different philosophy of care as well as reengineer-
tervention phase. However, there was no significant differ- ing existing health care services. The design of this quality
ence in mortality between the 2 groups using Cox regression improvement initiative attempted to address the gap between
analysis (Hazard Ratio 1.345, 95% Confidence Interval the substantial body of literature, guidelines, and government
0.931, 1.944, P 5 .114) after adjusting for age, sex, presence policies articulating the need for improved end-of-life care
of heart failure, and stroke. for noncancer patients1–5,15,26–29 and descriptions and
After the intervention, more improvement in patient evaluation of models of actual service provision.30 The inter-
symptoms was achieved for pain and dizziness, although con- vention specifically targeted the rather prevalent view that
stipation worsened (Table 2). There was a trend toward less end-of-life care equates with terminal care (no resuscitation
cough and depressed mood (Table 2). Reductions remained order), a tendency toward preserving life at all cost versus in-
significant for pain and cough, although there was a tendency creasing emphasis on quality of care in the downward trajec-
toward more constipation and less dizziness after adjustments tory of many chronic diseases toward death, and a reluctance
for age, gender, and differences in diagnoses (Table 2). to discuss the dying process among many health care profes-
Caregiver and patient (by proxy) satisfaction improved sionals, patients, and relatives. This model of care could be
after the initiative. No significant differences were observed supplemented by consultation to palliative medicine

ORIGINAL STUDIES Woo et al 109

Table 4. Comparisons of Patient Characteristics and Follow-up Services Use among Survivors of the Current Stay
Items Preintervention Postintervention P Value
(n 5 58) (n 5 46)
Female sex 38 (65.5) 30 (65.2) .975
Age, y, mean (SD) 82.4 (9.3) 85.5 (8.6) .088
Diagnosis
COPD 8 (13.8) 4 (8.7) .419
CHF* 10 (17.2) 2 (4.3) .041
CRF 8 (13.8) 8 (17.4) .613
Liver disease 1 (1.7) 4 (8.7) .099
Dementia 32 (55.2) 29 (63.0) .418
Stroke* 15 (25.9) 3 (6.5) .010
Other chronic neurological conditions 1 (1.7) 0 (0.0) .371
Marrow failure 2 (3.4) 0 (0.0) .203
Follow-up services after discharge
Rapid-response medical clinic in day hospital* 3 (5.2) 11 (23.9) .005
Outreach team (home care or nursing home)* 11 (19.0) 24 (52.2) \.001
Mean follow-up after recruitment until death or 241.9 (23.6) 120.5 (10.7) \.001
Apr 2, 2009 (days, SE)*
Transfer back to acute care hospital* 10 (17.2) 0 (0.0) .003
Numbers are presented as n (%) unless otherwise indicated.
COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; CRF, chronic renal failure.
* P \.05.

specialists when necessary, rather than be dependent on the by sheer cultural change, without additional resources. The
palliative team to provide the service, and has the advantage shift toward continuing care in the community would also
of involvement of all staff with responsibility for caring for be compatible with patients’ wishes. A significant proportion
patients with end-stage chronic diseases. of these patients reside in long-term residential care, and
This quality improvement initiative resulted in a rather un- a recent survey of 1600 such residents in Hong Kong found
expectedly large impact on use of hospital services in terms of that over 80% preferred to receive comfort care as end-of-
fewer investigations and shorter duration of stay, with a shift life care and one third accepted receiving this care in their
toward use of community services. The reduction in acute residential care facility rather than be admitted to hospitals
care hospitalization rate from 12.5% to 0% as well as the (Chu et al, 2009, unpublished data).
reduction in hospital bed days would likely translate into The initiative was also able to achieve improvement in
cost savings, although we did not carry out a detailed eco- symptom control. It has been noted that the frequency of
nomic evaluation. Apart from that, it helped to avoid the un- pain among patients with end-stage chronic diseases is com-
desirable effects of acute care transfers on these ill patients31 parable to those with cancer32 (approximately 50%); yet pain

Table 5. Comparisons of Resource Use Pre- and Post-End-of-Life Intervention With Adjustments for Confounders*
Types of Resource Use Difference 95% Confidence P Value
in Use Interval
Postintervention of Change
Entire group (including patients deceased in current admission and survivors)
Current admission†
Current admission bed days (days)‡ 15.74 25.28, 6.19 .001
No. of blood tests in index admission after recruitment‡ 3.44 0.62, 0.86 .009
No. of x-ray tests in index admission after recruitment‡ 0.72 1.15, 0.30 .001
Acute care hospital bed days since recruitment until death or 5.05 7.86, 2.24 .001
censor date (days)‡
Convalescence bed days since recruitment until 11.22 19.87, 2.57 .011
death or censor date (days)‡
Among survivors of current admission
Acute care hospital bed days since recruitment until death 8.55 12.73, 4.36 \.001
or censor date (days)‡
Convalescence bed days since recruitment until 12.53 24.06, 0.99 .034
death or censor date (days)‡
Emergency room episodes after discharge .55 1.81, 0.71 .386
Acute care hospital admissions after discharge .78 1.56, 0.00 .050
Convalescence facility admissions after discharge 0.22 .31, 0.75 .422
* Confounders include age (per each additional year), female sex, heart failure, stroke, and for follow-up duration from recruitment until
death or April 2, 2009.
† Confounders as above, but not including follow-up duration.
‡ P \.05.

110 Woo et al JAMDA – February 2011

as a symptom receives more attention in cancer patients com- 2. NICE. National Institute of Clinical Excellence. United Kingdom:
pared with patients with nonmalignant end-stage diseases. Royal College of Physicians; 2003.
3. Neely KJ, Roxe DM. Palliative care/hospice and the withdrawal of
This likely reflects the comparatively well-developed pallia-
dialysis. J Palliat Med 2000;3:57–67.
tive care services for cancer patients compared with the 4. Voltz R, Borasio GD. Palliative therapy in the terminal stage of neurolog-
care of nonmalignant chronic diseases. Approximately 49% ical disease. J Neurol 1997;244:S2–10.
of the patients in this study at baseline had pain, and there 5. Robinson L, Hughes J, Daley S, et al. End-of-life and dementia. Rev Clin
was significant improvement in pain prevalence after the Gerontol 2005;15:135–148.
6. Harris D, Richard B, Khanna P. Palliative care for the elderly: A need
initiative. The magnitude of improvement was at least as
unmet. Geriatr Med 2006;36:29–33.
large as that achieved by palliative care teams.24,33 7. Fitzsimons D, Mullan D, Wilson JS, et al. The challenge of patients’ un-
Other than improvement in family caregiver satisfaction met palliative care needs in the final stages of chronic illness. Palliat Med
with services, we did not demonstrate any changes in quality 2007;21:313–322.
of life and other psychosocial measures among patients, care- 8. Hillier R. Palliative medicine. BMJ 1988;297:874–875.
9. Cancer pain relief and palliative care: report of a WHO Expert Commit-
givers, or staff. It is likely that many factors unrelated to
tee Geneva, World Health Organization, 1990 (WHO Technical Report
health care service delivered in the short term affect these Series, No. 804)
parameters. It is also possible that the psychological stress 10. Qaseem A, Snow V, Shekelle P, et al. Evidence-based interventions to
of caring for patients with end-stage chronic diseases is of improve the palliative care of pain, dyspnea, and depression at the end
a different nature or of a lower magnitude compared with of life: A clinical practice guideline from the American College of Phy-
patients with cancer, because of the longer disease trajectory sicians. Ann Intern Med 2008;148:141–146.
11. Richards M. The end of life care strategy. Available at: www.dh.gov.uk/
in those with chronic diseases, allowing for more gradual en/healthcare/intergratedcare/endoflifecare/index.htm. Accessed March
adaptation. Furthermore, the advance age of the patients 1, 2009.
may also be a factor in acceptance of end-of-life care. 12. Gibbins J, McCoubrie R, Alexander N, et al. Diagnosing dying in the
There are limitations in this study. We were unable to pro- acute hospital setting—are we too late? Clin Med 2009;9:116–119.
vide traditional CQI analysis with run charts/control charts as 13. Buxbaum RC. Subacute care: The road ahead. Ann Long-Term Care
2009;17:18–22.
relevant expertise was not available in our institution. From 14. Agren Bolmsjo I. End-of-life care for old people: A review of the litera-
the service point of view, this initiative provided an opportu- ture. Am J Hosp Palliat Care 2008;25:328–338.
nity for educating all staff in the philosophy and principles of 15. Oliver DP, Porock D, Zweig S. End-of-life care in US nursing homes: A
end-of-life care for the predominantly elderly patients with review of the evidence. J Am Med Dir Assoc 2005;6:S21–S30.
end-stage chronic conditions commonly encountered. In 16. Higginson IJ. Audit methods: A community schedule. In: Higginson I,
editor. Clinical Audit in Palliative Care. Oxford: Radcliffe Medical
spite of this limitation we have demonstrated that it is possi-
Press; 1993. p. 34–48.
ble to improve quality of end-of-life care for elderly patients 17. Chiu HF, Lee HC, Wing YK, et al. Reliability, validity and structure of
with end-stage chronic diseases by staff education, and the Chinese Geriatric Depression Scale in a Hong Kong context: A pre-
culture and system change, not only without additional re- liminary report. Singapore Med J 1994;35:477–480.
sources, but likely savings were achieved in terms of reduced 18. Wu AMS, Tang CSK, Yan ECW. Assessment of death anxiety among
Chinese: A study of reliability and validity (Chapter 7). Adv Psychol
use of hospital resources. In view of the increasingly large
Res 2003;28:119–128.
numbers of such patients with population ageing, such 19. Lo RS, Woo J, Zhoc KC, et al. Cross-cultural validation of the McGill
a model could be easily adopted, disseminated, and tested fur- Quality of Life questionnaire in Hong Kong Chinese. Palliat Med
ther in other hospitals or convalescence settings. Hopefully 2001;15:387–397.
this initiative may have contributed to improvement in the 20. Lam CL, Tse EY, Gandek B. Is the standard SF-12 health survey valid and
quality of dying.34 In the future, the incorporation of specially equivalent for a Chinese population? Qual Life Res 2005;14:539–547.
21. Tseh OY, Loke AY, Chan MF, Kwok T. Psychometric properties of the
designed forms in addition to computer categorization of Chinese version ‘‘Cost of Care Index’’ for measuring the caregiving bur-
types of patients into patients’ electronic records would dens among Hong Kong informal care givers. Asian Journal of Nursing
contribute to the sustainability of the quality improvement Studies 2005;8:4–13.
initiative. Application for funding for such a development 22. Tang CSK. Assessment of burnout for Chinese human service professionals:
is currently being considered A study of factorial validity and invariance. J Clin Psyc 1998;54:55–58.
23. Institute for Healthcare Improvement. Available at: http://www.ihi.org/
IHI/Topics/Improvement/ImprovementMethods. Accessed January 26,
CONCLUSION 2009.
24. Higginson IJ, Hearn J. A multicenter evaluation of cancer pain control
It is possible to improve quality-of-life care for elderly pa- by palliative care teams. J Pain Symptom Manage 1997;14:29–35.
tients with end-stage chronic diseases by staff education, 25. Kalso E, Edwards JE, Moore RA, McQuay HJ. Opioids in chronic non-cancer
and culture and system change, not only without additional pain: Systematic review of efficacy and safety. Pain 2004;112:372–380.
resources, but likely savings were achieved in terms of re- 26. Abrahm JL, Hansen-Flaschen J. Hospice care for patients with advanced
duced use of health care resources. lung disease. Chest 2002;121:220–229.
27. Seamark DA, Seamark CJ, Halpin DM. Palliative care in chronic ob-
structive pulmonary disease: a review for clinicians. J R Soc Med 2007;
100:225–233.
REFERENCES 28. Hanrahan P, Luchins DJ. Access to hospice programs in end-stage
1. NICE Clinical Guideline. Chronic obstructive pulmonary disease. dementia: A national survey of hospice programs. J Am Geriatr Soc
National clinical guideline on management of chronic obstructive pul- 1995;43:56–59.
monary disease in adults in primary and secondary care: Managing stable 29. NICE Clinical Guideline 42, developed by the National Collaborating
COPD. Thorax 2004;59:i39–i130. Centre for Mental Health. Dementia: supporting people with dementia

ORIGINAL STUDIES Woo et al 111

 and their careers in health and social care. National Institute for Health 32. Lo RS, Ding A, Chung TK, Woo J. Prospective study of symptom control
and Clinical Excellence, UK, November 2006. in 133 cases of palliative care inpatients in Shatin Hospital. Palliat Med
30. Department of Health. Government Response to House of Commons 1999;13:335–340.
33. Higginson IJ, Finlay IG, Goodwin DM, et al. Is there evidence that
Health Committee Report on Palliative Care 2003–2004. The Station-
palliative care teams alter end-of-life experiences of patients and their
ary Office Limited, UK. caregivers? J Pain Symptom Manage 2003;25:150–168.
31. Steinberg KE. Reducing unnecessary hospitalizations: Apple pie! J Am 34. MacLeod RD. Quality of dying—how are we doing? Qual Saf Health
Med Dir Assoc 2009;10:595–596. Care 2008;17:237–238.

APPENDIX 1

The EOLC Guidance Sheets Attached to Patient Files

Commencing Criteria (By Doctors)

1. Patient and relative not opting for aggressive/ invasive treatment AND Patient label
2. Existing Do-Not-Resuscitate order AND
3. Any of the following prognostic indicators of advanced disease (circle as
appropriate)

COPD Liver Disease

1A: Disabling SOB with reduced functional capacity 1. PT>5 sec above control or INR>1.5
1B: Progression as evidence by increasing AED 2. Serum albumin <2.5 g/dl
attendance / hospital admission 3. Plus any of the following:
2: PaO2 < 55 mmHg / SpO2 < 88% on supplemental a. Refractory ascites
oxygen or hypercapnia (PaCO2 > 50 mmHg) b. Spontaneous bacterial peritonitis
3: Cor pulmonale c. Hepatorenal syndrome
4: Unintentional weight loss> 10% over 6 months d. Encephalopathy with asterixis,
5: Resting tachycardia > 100 bpm. somnolence, coma
6. Already On optimal drug therapy e. Recurrent variceal bleeding
7. FEV 1 of 30 or less
8. LTOT+/- NIV

Chronic Heart Failure Dementia

1. No identifiable reversible precipitants 1. Cannot walk even with assistance, no speech,
2. Receiving optimum tolerated drugs double incontinence)
3. Deteriorating renal function 2. After the first episode of aspiration pneumonia
4. Failure to respond within two or three days to appropriate (despite feeding tube)
changes in diuretic, vasodilator drugs or fluid intake 3. Plus any of the following:
5. Ejection fraction ≤ 30% a. 10% weight loss in the previous 6 months
6. NYHA grade III-IV b. Serum albumin <25g/L
7. Serum sodium < 138mmol/L c. Recurrent fevers
8. Heart rate > 100bpm d. Pyelonephritis or UTI
9. Creatinine > 2.0mg/dL e. Multiple stage III or stage IV decubitus
ulcers

Chronic Renal Failure Stroke

1. With complications e.g.: pulmonary edema, Persistent vegetative or minimal conscious
hyperkalemia, acidosis, sepsis, hypertensive crisis state/dense paralysis/incontinence
2. Patient not seeking dialysis
3. Patient not a candidate for renal transplant
4. Plus any of the following: Neurological conditions
a. Creatinine clearance < 10 cc/min (without comorbid 1. Onset of symptoms that are related to, or a
conditions) complication of the neurological problem, e.g.
b. Creatinine clearance < 15 cc/min (with comorbid pneumonia, sepsis, coma, seizures.
diabetes or CHF) 2. Progressive decline
c. Creatinine clearance < 20 cc/min (with comorbid
diabetes and CHF)
d. Serum creatinine > 8.0 mg/dl (>6.0 for diabetics) Frailty

1. Complete functional dependency

2. Multiple organ failure (end-stage) despite
optimal medical intervention
3. Progressive wasting

112 Woo et al JAMDA – February 2011

 Conditions Review: Does the patient have the following?
Cognition & Psychological GI
- Agitation/ Restlessness - Anorexia
- Confusion/ Delirium - Dysphagia
- Anxiety/ Fear - Nausea/ Vomiting
- Depressed - Constipation
- Unconscious

General Physical Sleep problem

- Pain - Insomnia
- Dyspnea/ SOB - Intermittent sleep
- Increased airway secretions
- Edema
- Fatigue
- Lower Limb Weakness
- Pressure ulcer/ wounds

Interventions Review: Are the following really necessary to help the patient?
Monitoring: Investigations:
- SpO2 - Blood-taking
- H’stix - Endoscopy
- Excess BP or cardiac monitor - XR

Treatment: Drugs:
- BIPAP - Inotropes
- Diet Restriction - Antibiotics

Tubes and lines:

- IVF
- Feeding tubes
- Foley Catheters

ORIGINAL STUDIES Woo et al 113