You are on page 1of 6

Biases in palliative care access for elderly

patients dying in hospital: A prospective


study in acute care
Stefania Pandini 1, Sergio Defendi2, Carlo Scirè 1, Francesco Fiorini 3,
Gianfrancesco Fiorini 1,4
1
Internal Medicine Unit, ‘Ospedale Maggiore’ Hospital, Crema, Italy, 2Palliative Care Unit, ‘Ospedale Maggiore’
Hospital, Crema, Italy, 3King’s College London, School of Medicine, London, UK, 4Present address: Specialistic
Rehabilitation, Istituti Clinici Zucchi, Carate Brianza (MB), Italy

Background: Patients in the terminal phase of chronic illnesses are often admitted to acute care wards, with
the risk of receiving inappropriate intensive treatments as opposed to necessary palliative care (PC).
Objective: To assess patient features and possible service biases in the activation of PC pathways, or lack
thereof, for those dying in acute care.
Design and setting: This was a prospective observational study of all patients admitted to our acute medical
ward from the emergency department during 6 months.
Measurements: Need for palliation was evaluated for all patients by a physician and a nurse of the Internal
Medicine ward. Those proposed for PC were re-evaluated by a consultant and a nurse of the PC unit.
Clinical and epidemiological data were obtained for those selected for PC, and those deceased in the
acute ward without having received PC.
Results: A total of 781 patients were admitted to the acute care ward and screened during the study period.
Of the 56 patients assigned to the PC pathway, those that died in hospital (n = 30) had significantly poorer
Karnofsky, Braden, and ECOG scores compared to those alive at discharge (n = 26). Forty-eight more
patients died on the ward without having received PC. The prevalence of cardiopulmonary diseases was
significantly higher in this group, while more oncologic patients died in the PC cohort.
Conclusions: Palliative treatment needs to be augmented and better targeted in acute care settings. Special
attention should be given to patients with cardiopulmonary illnesses, a possible bias preventing the activation
of this type of care, and those at high risk of developing pressure ulcers. Further education and training in this
field are crucial for healthcare professionals working in acute wards.
Keywords: Acute care, Chronic disease, In-hospital dying, Palliative care, Non-neoplastic terminal patients, Performance scores

Introduction increasingly implemented particularly for the


The rate of non-acute mortality is expected to continue elderly,6,7 but access to PC remains limited.8
to grow in western countries,1 in spite of the overall Patients admitted to acute care wards from the
reduction in age-standardized mortality rates.2,3 This emergency department are often elderly, suffering
is linked to the constant ageing of the population from end-stage chronic diseases and frequent comor-
and the increasing impact of deaths due to cancer bidities. Their assessment is complex and PC packages
and other chronic diseases. are not always activated appropriately or promptly.
With the majority of deaths occurring in the acute The purpose of this study was to evaluate the clinical
hospital setting, countries such as US, Canada, and and epidemiological features of patients deceased in
Australia have made efforts to reduce the number of an acute ward who had entered PC pathways com-
in-hospital deaths.4 Extensive evidence suggests that pared to those discharged alive in PC and to those
a significant proportion of patients admitted to acute who died under active care, and to determine if there
care hospitals would be better treated with a palliative were biases in referral to PC.
approach.5 New pathways of care are being

Patients and methods


Correspondence to: Gianfrancesco Fiorini, Responsabile Area Internistica,
Riabilitazione specialistica 1° piano, Istituti Clinici Zucchi, Carate, MB, Italy.
All the patients (n = 781) admitted to our acute care
Email: gianfrancesco.fiorini@grupposandonato.it ward from the emergency department over 6 months

© 2016 Informa UK Limited, trading as Taylor & Francis Group


310 DOI 10.1080/09699260.2016.1230973 Progress in Palliative Care 2016 VOL. 24 NO. 6
Pandini et al. Palliative care for patients dying in acute wards

(September 2014–March 2015) were evaluated by a during the study period), a further 48 (57.1%) died
physician and a nurse to determine the need for PC. receiving only active care and no PC. When com-
The evaluation took place according to the Gold paring patients deceased in PC to those dying
Standards Framework (GSF) Prognostic Indicator without, we found significant differences in the
Guidance9 over two stages: concisely on admission and prevalence of main diagnostic clusters (Table 1).
more in depth after clinical stabilization (72 hours). Cardiopulmonary diseases (specifically 31/44
Where a PC package was requested, patients were then (70.5%) were heart failure) were more prevalent
evaluated by a PC consultant and head nurse of the among patients who did not receive PC (64.6%)
PC unit. This two-step process was implemented to than those who did (36.1%; P = 0.010). The oppo-
ensure the fulfillment of the five priorities proposed by site was true for oncologic pathologies, more likely
the Leadership Alliance for the Care of Dying among those assigned to PC (19.4%) than those
People.10 Ethical approval for this study was granted receiving only active care (2.1%; P = 0.007). These
by the ‘Ospedale Maggiore’ hospital ethics committee. differences accounted for an overall significant diag-
Patients referred to PC who then died in hospital nostic difference (P = 0.016).
were compared to those who died in hospital without The baseline demographic and clinical character-
having received PC and to those under PC and alive istics of the patients elected for PC are described in
at discharge. Table 2. Of these, 36 died in hospital and 20 were
A number of parameters were taken into consider- alive at discharge (four to hospice, five to intermedi-
ation, including demographic characteristics, clinical ate care, one to nursing home, one to residential
performance, and risk calculation scales relevant to care, nine home). The differences between these two
this patient group (ECOG,11 Karnofsky,12 Morse,13 groups are explained in Table 3. In summary,
and Braden scores).14 Diagnostic groups were deter- Karnofsky, ECOG, and Braden scores were signifi-
mined according to the main diagnosis present on dis- cantly lower in the deceased group (Karnofsky OR
charge forms, filled in accordance with International 0.90, 95% CI 0.83–0.97; ECOG OR 4.45, 95% CI
Classification of Diseases (ICD) criteria. Due to the 1.05–18.81; Braden OR 0.72, 95% CI 0.55–0.95).
limited sample size, diagnoses were further grouped After adjustment for age and gender, the association
into four clusters: cardiopulmonary, oncologic, infec- persisted for Karnofsky (OR 0.91, 95% CI 0.84–
tious, and miscellaneous diseases. 0.98) and Braden (OR 0.73, 95% CI 0.55–0.98)
The primary outcome was set as vital status at dis- scores, but not ECOG (OR 3.35, 95% CI 0.75–
charge. Differences in demographic or clinical vari- 14.86). Moreover, there was a significant correlation
ables between deceased patients and those alive at between Karnofsky and Braden scores (Spearman’s
discharge were estimated using a Chi-square test or r = 0.52, P < 0.0001).
Wilcoxon test for categorical and quantitative vari- On the contrary, no significant differences in demo-
ables, respectively. Correlations between performance graphic and clinical characteristics existed between
scales were estimated by calculating Spearman’s rho patients dying in PC and patients dying in active
(r) coefficient. To evaluate the specific associations care (data not shown).
of different performance scales with death, three The patients discharged alive in PC represented only
different logistic models were used: unadjusted, a small percentage (2.9%) of the 697 patients dis-
adjusted for age and gender, and fully adjusted step- charged alive.
wise. Results are presented as odds ratios and 95%
confidence intervals. Descriptive measures of demo- Table 1 Demographic and clinical variables for deceased
graphic and clinical variables are reported as fre- patients according to activation of PC or lack thereof
(standard care)
quency ( percentage), mean (standard deviation), or
median (interquartile range) as appropriate. Type of care for deceased patients
Analyses were performed using STATA software Palliative Standard P
package (2009, v. 11; StataCorp, TX, USA). Characteristic care care value

n (%) 36 (42.9) 48 (57.1)


Results Age, mean (SD) 82.6 (10.8) 84 (8.2) 0.478a
During the 6-month study period, 781 patients were Gender (female), n (%) 16 (44.4) 24 (50) 0.614b
Main diagnostic group, 0.016 b
admitted to our acute care ward from the emergency n (%)
department. Fifty-six (7.2%) were selected for acti- Cardiopulmonary 13 (36.1) 31 (64.6) 0.010 b
vation of a PC pathway; all the patients proposed for Neoplastic 7 (19.4) 1 (2.1) 0.007 b
Infection 6 (16.7) 6 (12.5) 0.589b
PC by the Internal Medicine team were judged suit- Other 10 (27.8) 10 (20.8) 0.460b
able by the PC team.
Note: Bold P values are statistically significant.
In addition to the deceased patients on the ward a
Wilcoxon test.
b
who were receiving PC (42.9% of total ward deaths Chi-square test.

Progress in Palliative Care 2016 VOL. 24 NO. 6 311


Pandini et al. Palliative care for patients dying in acute wards

Table 2 Baseline demographic and clinical characteristics deceased in hospital had not been assigned to a PC
of patients assigned to the PC pathway. Previous admissions:
at least 1 admission in the 3 months preceding the present
pathway.
admission. Time to assessment: days between patient Our data show that deceased patients who had been
admission and evaluation by the palliative care team assigned to PC had more frequently an oncologic
Characteristic Value disease compared to those who died receiving only
active care. This confirms a bias in referral that has
n 56 been previously described both for in-hospital
Age, mean (SD)t 83.7 (9.7)
Gender (female), n (%) 29 (48.2) patients15 and outpatients,16 despite our efforts to
Main diagnostic group, n (%) recognize early those requiring PC. Conversely,
Cardiopulmonary 20 (35.7)
Neoplasm 11 (19.6)
patients deceased under active care were more likely
Infection 7 (12.5) to suffer from a cardiopulmonary illness – in particu-
Other 18 (32.2) lar heart failure – than those who were selected for PC.
Previous admissions, n (%) 31 (55.4)
Caregiver This might represent another bias for PC activation, as
Family 54 (96.4) non-communicable diseases other than cancer may be
Other 1 (1.8) mistakenly and involuntarily regarded as less appro-
None 1 (1.8)
Time to assessment (days), median (IQR) 4 (2–8) priate for PC.
Karnofsky, mean (SD) 25.2 (9.3) Amongst patients who did receive PC, those who
ECOG, mean (SD) 3.82 (0.43)
Braden, mean (SD) 10.1 (2.4)
died scored significantly lower on a number of clinical
Morse, mean (SD) 88.3 (21.1) performance scales (Karnofsky, ECOG, and Braden)
Pain (NRS), mean (SD) 0.62 (1.52) than those alive at discharge. The poorer functional
Dysphagia, n (%) 20 (35.7)
Dyspnea, n (%) 32 (57.1) status described in these patients is not unexpected.
Confusion, n (%) 42 (75.0) However, when adjusting for age and gender ECOG
Anorexia, n (%) 39 (69.6) was not discriminatory. While this parameter is uni-
Hospitalization (days), median (IQR) 6 (4–11.5)
versally accepted and has been shown to improve
SD: standard deviation; IQR: interquartile range. risk adjustment models in oncologic patients,17 our
findings suggest that Karnofsky scores – which are
Table 3 Demographic and clinical variables for patients not influenced by age and gender – may better
receiving PC according to vital status at discharge define functional status in this broader group of
Vital status of PC patients at patients in the acute setting. Moreover, the signifi-
discharge cantly lower Braden scores (i.e. increased risk of devel-
P oping pressure ulcers) in the deceased group highlight
Characteristic Alive Deceased value the importance of pressure ulcers in the recognition
n (%) 20 (35.7) 36 (64.3) and management of patients nearing end of life. This
Age, mean (SD) 85.6 (6.9) 82.6 (10.8) 0.258a scale is often used in acutely ill elderly patients with
Gender (female), n (%) 13 (65.0) 16 (44.4) 0.140b a significant burden of comorbidities, and may be a
Main diagnostic group, 0.572b
n (%) useful complement to Karnofsky scoring.
Cardiopulmonary 7 (35.0) 13 (36.1) 0.934 Despite the best efforts of healthcare professionals,
Neoplastic 4 (20.0) 7 (19.4) 0.960
Infection 1 (5.0) 6 (16.7) 0.206
many patients hospitalized for acute illnesses suffer
Other 8 (40.0) 10 (27.8) 0.348 worsening of their conditions and death. This is
Previous admissions, n 14 (70.0) 17 (47.2) 0.100b often due to the progressive deterioration of pre-exist-
(%)
Time to assessment 4.5 (2.5–8.5) 4 (2–8) 0.870a ing comorbidities. As a consequence, these patients are
(days), median (IQR) at risk of receiving non-beneficial intensive treatment.
Karnofsky, mean (SD) 30 (10.1) 22.5 (7.8) 0.003 a The mismatch in patients’ needs and care provided is
ECOG, mean (SD) 3.65 (0.59) 3.91 (0.28) 0.033 a
Braden, mean (SD) 11.1 (2.9) 9.4 (1.85) 0.019 a often perceived by hospital personnel as frustrating.18
Morse, mean (SD) 87.5 (24.2) 88.7 (19.5) 0.724a Nevertheless, advances have been made in recent years
Pain, mean (SD) 0.45 (1.46) 0.71 (1.56) 0.370a to implement novel care pathways for such patients.
Dysphagia, n (%) 6 (30.0) 14 (38.9) 0.506b
Dyspnea, n (%) 11 (55.0) 21 (58.3) 0.809b These range from new care settings and facilities6 to
Confusion, n (%) 15 (75.0) 27 (75.0) 1.000b access to treatments like palliative care (PC), originally
Anorexia, n (%) 11 (55.0) 28 (77.8) 0.076b
designed for a different target population,15 to
Note: Bold P values are statistically significant.
a
‘comfort care’ for in-hospital patients who are very
Wilcoxon test.
b
Chi-square test.
close to death.19 Timely integration of appropriate
PC for patients in acute hospital settings leads to
better quality of life and prolonged survival; moreover,
Discussion it has been shown to permit a better utilization of
In spite of early evaluation of all patients admitted to resources in end-stage in-hospital patients20 but is
our acute care ward, the majority (57.1%) of those often prevented by many barriers. These include

312 Progress in Palliative Care 2016 VOL. 24 NO. 6


Pandini et al. Palliative care for patients dying in acute wards

difficulties in recognizing patients that could benefit performed statistical analysis. F.F.: Medical Student.
from PC,21 the type of clinical diagnosis,15,22 logistic Recently graduated (BMSB) at King’s College
problems and poorly defined indications for referral22 Medical School (London). He has an interest in neuro-
and ethnicity23; in spite of this, efforts are being sciences. He stored the data and wrote the manuscript
increasingly made for proper delivery of PC to the in English. G.F.: Head of Internal Medicine. He has
populations of patients who need this approach.24 worked in Internal Medicine for 30 years in various
The present work represents a contribution in this hospitals. He has been director at the Maggiore
direction. Hospital from December 2008 to 31 May 2015.
From the first of June 2015, he is consultant internist
Conclusions in a specialistic rehabilitation institution. Principal
Awareness and commitment of nurses and physicians investigator; coordinated the group and personally
are not sufficient to provide adequate PC to all took part in the majority of clinical assessments.
patients dying in acute hospital wards; there are
Funding There was no formal sponsor for this study.
biases in assigning terminal patients to palliative
care, an important one possibly being an oncologic Conflicts of interest The authors report no potential
versus a non-oncologic diagnosis. conflict of interest.
Further training in this field should be offered to Ethics approval This study was granted by the
acute hospital staff, together with the implementation ‘Ospedale Maggiore’ hospital ethics committee.
of appropriate performance scores.
A reliable panel of indicators is needed to recognize References
these patients early. In particular, we suggest the use of 1 Hasselaar J, Engels Y, Menten J, Jaspers B, Vissers K. The
the Karnofsky score in conjunction with the Braden burden of non-acute dying on society: dying of cancer and
score to evaluate the functional status of terminal chronic disease in the European Union. BMJ Support
Palliative Care 2012;2:334–8.
non-oncologic patients in these settings. Special atten- 2 Wheller L, Baker A, Griffiths C, Rooney C. Trends in avoidable
tion should be directed to the identification of pallia- mortality in England and Wales, 1993–2005. Health Stat Q
2007;34:6–25.
tion needs in elderly patients with heart failure and 3 Korda RJ, Butler JRG. Effect of healthcare on mortality: trends
other chronic cardiopulmonary diseases, as well as in avoidable mortality in Australia and comparisons with
Western Europe. J Public Health 2006;120:95–105.
pressure ulcers. 4 Gomes B, Higginson IJ. Where people die (1974–2030): past
trends, future projections and implications for care. Palliat
Med 2008;22:33–41.
Disclosure statement 5 Gardiner C, Gott M, Ingleton C, Seymour J, Cobb M, Noble B,
Contributors The activities of the authors and their et al. Extent of palliative care need in the acute hospital setting: a
contributions to the study are as follows: S.P.: Head survey of two acute hospitals in the UK. Palliat Med 2012;27:
76–83.
Nurse. She started to work with elderly and frail 6 Woodford HJ, George J. Intermediate care for older people in
patients, after graduation, in a nursing home in the UK. Clin Med 2010;10:119–23.
7 Fiorini G, Pandini S, De Matthaeis A, Seresini M, Dragoni R,
Crema (Fondazione Benefattori Cremaschi). Then Sfogliarini R, et al. Intermediate care as a means of improving
she moved to the Maggiore Hospital as Head Nurse mental status in post-acute elderly patients. Aging Clin Exp
Res 2013;25:337–41.
in the Internal Medicine Department. She has a 8 Wale J. Exploring key topics in palliative care: pain and palliative
master in vulnology and is responsible for consul- care for older people. Clin Med 2014;14:416–8.
9 Thomas K, et al. The GSF prognostic indicator guidance. 4th ed.
tations on advanced medications. She coordinated Gold Standards Framework Centre In End of Life Care CIC;
the nursing staff involved in the study and personally 2011. Available from: http://www.goldstandardsframework.org.
carried out various assessments of study participants. uk/cd-content/uploads/files/General%20Files/Prognostic%20
Indicator%20Guidance%20October%202011.pdf.
S.D.: Consultant in Palliative Care. At the beginning 10 Wise J. Five priorities of care for dying people replace Liverpool
of his career, he specialized in obstetrics and gynecol- care pathway. BMJ 2014;348:g4299.
11 Oken MM, Creech RH, Tormey DC, Horton J, Davis TE,
ogy, then he trained in Palliative Care and worked as a McFadden ET, et al. Toxicity and response criteria of the Eastern
palliativist in Bergamo before moving to the Maggiore Cooperative Oncology Group. Am J Clin Oncol 1982;5:649–55.
12 Evans C, McCarthy M. Prognostic uncertainty in terminal care:
Hospital, initially for a locum position and now with a can the Karnofsky index help? Lancet 1985;1(8439):11204–6.
permanent appointment as Palliative Care consultant. 13 Hill Bailey P, Lynn Rietze L, Moroso S, Szilva N. A description
He coordinated the evaluation and subsequent treat- of a process to calibrate the Morse fall scale in a long-term care
home. Appl Nurs Res 2011;24:263–8.
ment of the patients selected for a palliative care 14 Bergstrom N, Braden BJ, Laguzza A, Holman V. The Braden
package. C.S.: Internist. He has a specialization in Scale for predicting pressure sore risk. Nurs Res 1987;36:
205–10.
rheumatology and worked as a rheumatologist in 15 Dalgaard KM, Bergenholtz H, Nielsen ME, Timm H. Early
Pavia, in an outpatient clinic of the University integration of palliative care in hospitals: a systematic review
on methods, barriers, and outcome. Pall Supp Care 2014;12:
Hospital San Matteo. He has a degree in statistics. 495–513.
He works in the ward of medicine and is responsible 16 Hess S, Stiel S, Hofmann S, Klein C, Lindena G, Ostgathe C.
Trends in specialized palliative care for non-cancer patients in
for the rheumatology clinic. He cooperates with Germany – Data from the National Hospice and Palliative
regional authorities for epidemiological studies. He Care Evaluation (HOPE). Eur J Int Med 2014;25:187–92.

Progress in Palliative Care 2016 VOL. 24 NO. 6 313


Pandini et al. Palliative care for patients dying in acute wards

17 Young J, Badgery-Parker T, Dobbins T, Jorgensen M, Gibbs P, 21 Gardiner C, Cobb M, Gott M, Ingleton C. Barriers to the pro-
Faragher I, et al. Comparison of ECOG/WHO performance vision of palliative care for older people in acute hospitals. Age
status and ASA score as a measure of functional status. J Pain Ageing 2011;40:233–8.
Symptom Manage 2015;49:258–64. 22 Grudzen CR, Richardson LD, Hopper SS, Ortiz JM, Whang C,
18 Downar J, You JJ, Bagshaw SM, Golan E, Lamontagne F, Morrison RS. Does palliative care have a future in the emergency
Burns K, et al. Nonbeneficial treatment Canada: definitions, department? Discussions with attending emergency physicians. J
causes, and potential solutions from the perspective of healthcare Pain Symptom Manage 2012;43:1–9.
practitioners. Crit Care Med 2015;43:270–81. 23 Periyakoil VS, Neri E, Kraemer H. Patient-reported barriers to
19 Blinderman CD, Billings JA. Comfort care for patients dying in high-quality, end-of-life care: a multi-ethnic, multilingual,
the hospital. N Engl J Med 2015;373:2549–61. mixed-methods study. J Palliat Med 2015;18:1–7.
20 Foreman T, Kekewich M, Landry J, Curran D. Impact of pallia- 24 Sallnow L, Tishelman C, Lindqvist O, Richardson H, Cohen J.
tive care consultations on resource utilization in the final 48 to 72 Research in public health and end-of-life care – building on
hours of life at an acute care hospital in Ontario, Canada. J the past and developing the new. Progr Palliat Care 2016;24:
Palliat Care 2015;31:69–75. 25–30.

314 Progress in Palliative Care 2016 VOL. 24 NO. 6


Copyright of Progress in Palliative Care is the property of Taylor & Francis Ltd and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like