328

Journal of Pain and Symptom Management

Vol. 24 No. 3 September 2002

Original Article

A Retrospective Observation of Corticosteroid

Use at the End of Life in a Hospice
Craig Gannon, MB ChB, MRCGP, Dip Pall Med,
and Penny McNamara MB BS, MRCGP
Princess Alice Hospice (C.G.) Surrey, and Sue Ryder Care-St. Johns (P.McN.),
Bedford, United Kingdom

Abstract
This study aimed to clarify corticosteroid prescribing during final hospice care, realizing the
clinical and ethical dilemmas that may be associated with this therapy. A retrospective review
was performed of deaths occurring at our unit during a 6-month period. Corticosteroid use
was recorded from drug charts and cross-referenced by case note review. Fifty-one percent of 178
patients received corticosteroids, which were continued until death in 53%. Only 2% were
switched from oral to parenteral corticosteroids. The reason for using corticosteroids was
documented in 67% of patients. The main indications included treatment for raised
intracranial pressure and to give a boost. The foremost reason for withdrawing
corticosteroids was loss of the oral route. These data confirm the high prevalence of
corticosteroid use in the terminal phase, even until death. This contrasted with the near
absolute withdrawal of corticosteroids once the oral route was lost. The study suggests a need
for greater vigilance in corticosteroid prescribing, and identified issues to be addressed in the
prescribing of these drugs. J Pain Symptom Manage 2002:24:328334. U.S. Cancer
Pain Relief Committee, 2002.
Key Words
Research, corticosteroids, steroids, palliative care, terminal care, end of life

Introduction
Corticosteroids are extensively prescribed in
palliative care and can seemingly deliver dramatic symptomatic benefit in a variety of conditions (Table1).14 However best practice for
corticosteroid use in palliative care is not clear,
and treatment requires an empirical approach.
The use of corticosteroid therapy raises numerous concerns. There is limited evidence of
efficacy, and uncontrolled studies may be bi-

Address reprint requests to: Craig Gannon, MB ChB,

Princess Alice Hospice, West End Lane, Esher, Surrey, KT10 8NA, United Kingdom.
Accepted for publication: November 17, 2001.
U.S. Cancer Pain Relief Committee, 2002
Published by Elsevier, New York, New York

ased by placebo response that may be high (up

to 50%) and there are numerous confounders
(e.g., malignant bowel obstruction may spontaneously resolve). Long-term use has a high incidence of adverse effects,15 some of which
can be severe and occasionally devastating (Table 2).6 In many clinical settings, corticosteroid
monitoring is not sufficiently stringent,1,4,5 and
polypharmacy risks significant drug interactions.4,5,79
The longer-term use of corticosteroids for
non-specific indications are more concerning because there is no robust evidence base to
detail any symptom benefits experimentation
has been discouraged10and there is no rationale, as any shorter-term benefit would rapidly
cease. Cumulative adverse effects would con0885-3924/02/$see front matter
PII S0885-3924(02)00487-6

Vol. 24 No. 3 September 2002

Corticosteroid Use at the End of Life

329

Table 1
Corticosteroids in Palliative Care1,5,8
1. Tumor compression
a) Raised intracranial pressure
b) Spinal cord compression
c) Superior vena cava obstruction
d)Nerve damage (palsy/pain)
e) Lymphoedema
f) Lymphangitis carcinomatosis
g) Ureteric obstruction
h)Bronchial obstruction
i) Tracheal compression, stridor
j) Bowel obstruction
k) Bone pain, metastatic joint pain
l) Dysphagia
m) Dysphagia caused by a tumor mass
2. Anticancer therapy
Disease modification
3. To combat adverse effects of radiation
Post-cranial radiotherapy, Pneumonitis
4. Hypercalcemia
Limited role, particularly since the advent of bisphosphonates
5. Bronchospasm

tinue to increase4 with this treatment. In such

cases, corticosteroids may be better avoided in
lieu of alternative measures.10
Corticosteroid withdrawal poses a new set of
issues.1 Even low corticosteroid doses suppress
the hypothalamic-pituitary-adrenal axis4,5,10 and
require complex withdrawal and additional corticosteroids for stressors (e.g., pain, fractured femur, and dying) during treatment. Stress doses
may be needed possibly months after discontinuation, depending on the duration of corticosteroid use.1113 Corticosteroid withdrawal syndrome is yet to be detailed in palliative care
patients, but could be mistaken for progressive
disease. Possibly, subtle symptoms may go unnoticed (Table 3),10 particularly in dying patients. These symptoms may be distressing for
the patient. Finally, discontinuation of corticosteroids in the terminal phase could mistakenly

Anorexia
Pain
Nausea/vomiting
Weakness/fatigue
Mood/well-being
Dyspnea
Fever/itch
Rectal discharge

be implicated as the cause of a terminal deterioration.

For all these reasons patients receiving corticosteroids require supervision to ensure that
benefit is accrued and not outweighed by toxicity
(the shortest time and the lowest dose).1,4,14
Furthermore, it should be recognized that this
treatment poses sufficient clinical and ethical
dilemmas during patients final care to warrant
study.
The present study arose from informal observations suggesting that a high number of patients received corticosteroids during their terminal phase, and at this time corticosteroid doses
were reduced more rapidly within hospices
than advocated in other settings. Recently published work focusing on corticosteroids in palliative care either has not addressed this issue
specifically16 or has provided information for a

Table 2
Corticosteroid Side Effects1,4,5,13,15

Immunosuppression (e.g. oral thrush, risk of varicella, or masking the signs of septicemia)
Proximal myopathy
Sodium and water retention (and potassium loss)/peripheral edema/hypertension
Psychological changes (insomnia, agitation, behavioral changes, depression, paranoid psychoses)
Cushingoid features (moon facies, truncal obesity, buffalo hump, striae, acne)
Skin changes (striae, acne, bruising, thinning, poor healing, mild depigmentation, increased hair growth)
Hyperglycemia/loss of diabetic control
Dyspepsia (particularly with concurrent nonsteroidal anti-inflammatory drugs [NSAID])
Increased appetite (e.g., distressing if accompanying dysphagia)
Cataract (long-term useyears)
Osteoporosis/avascular necrosis of femoral head/joint pains/corticosteroid pseudorheumatism
Suppression of hypothalamic-pituitary-adrenal axis

330

Gannon and McNamara

Vol. 24 No. 3 September 2002

Table 3
The Potential Disadvantage of Stopping Corticiosteroids Abruptly 1,4,5,10,11,13
Loss of symptom control if
persisting corticosteroid benefit
Specific withdrawal
symptoms

Addisonian crisis
when corticosteroids
stopped or subsequent infection

small subgroup of the hospice population as

part of a broader audit.17 The aim of this study
was to clarify the present use and documentation of corticosteroid therapy, specifically in
patients dying in the hospice and receiving ongoing specialist input. Data included indications for initiating/maintaining corticosteroids,
the level of monitoring of corticosteroid response, the indications for withdrawing corticosteroids in terminal hospice admissions, and
the pattern of discontinuing corticosteroids. It
was hoped that the data collected could aid
standard setting, suitable for later clinical audits. The preliminary findings of this study
were presented in poster/abstract format.18

Methods
A retrospective case note review was performed on all deaths occurring at the Princess
Alice Hospice over the 6-month period immediately prior to the start of the research (April
1997 to September 1997). The hospice has a
26-bedded inpatient unit, accepting patients
from a 10-mile radius southwest of London who
require specialist palliative care. The retrospective design was required, as death while a hospice inpatient was an entry requirement. This
served to limit some biasas at the time of prescribing corticosteroids in the study period, no
one knew their practice was later to be surveyed. The retrospective design also alleviated
the need for prospective consent, as well as
many other possible barriers to enrollment.19,20

Rebound symptoms (e.g., headache, confusion, agitation,

neuropathic pain, nausea/vomiting, dyspnea)
Painmyalgia, arthralgia, and corticosteroid
pseudorheumatism
Abdominal pain
Nausea/vomiting
Weight loss
Malaise/depression
Fatigue
Fever
Hypotension/dizziness
Benign intracranial hypertension
Rhinitis, conjunctivitis, painful/itchy skin nodules
Cardiovascular shock
Confusion/coma
Can be fatal

Computer records identified the patients who

had died in the hospice during the study period. All drug charts for those patients whose
notes could be retrieved were checked for any
prescribed corticosteroid, excluding topical preparations. Patients who did not receive corticosteroids were not considered further for study.
Patients taking corticosteroids had their notes
reviewed to complete a simple table designed
to ensure inter-observer consistency, allow checking, and facilitate subsequent audit. Details were
obtained from the drug charts on the choice of
corticosteroid, the dosage, and routes used across
the period of their final admission, with the dates
of any changes. The medical and nursing case
notes were then cross-referenced to identify documentation relating to the indication(s) for prescribing corticosteroids and the rationale behind any changes in corticosteroid management.
Where the reasons were not documented, but
there were sufficient clinical details to identify
the likely indication, this deduced indication
was noted (e.g., corticosteroids within a month
of whole brain radiotherapy). In addition, the
use of a syringe driver during the terminal phase
(for any drug) was noted in a 1-month sample of
the study population for comparison.
The date of death and some baseline demographic details, namely age, sex, and diagnosis,
were also noted to allow comparison against the
normal hospice population, along with the
patients ID number to allow rechecking. A further 1-month sample of patients discharged during the study period was identified to provide a
comparison of corticosteroid prescribing.

Vol. 24 No. 3 September 2002

Corticosteroid Use at the End of Life

Results
The total number of admissions was 329.
The total number of deaths was 200 and the
proportion of admissions that ended in death
was 60.8%. The total number of notes retrieved was 178, and the percentage of case
notes retrieved for deceased patients was
89.0%. The mean age of the sample was 65
years and the men:women ratio was 4:3.
The total number of patients receiving corticosteroids during their terminal admission was
90 (50.6%). Of these, treatments included dexamethasone (n  77), prednisolone (n 13),
and parenteral corticosteroid (n  3). The
number dying with ongoing corticosteroid prescription was 48 and the number who had the
corticosteroids stopped on day of death was 7.
The percentage of patients who received corticosteroids during their final admission that received corticosteroids on that day of death was
61.1%. The percentage that received a regular
prescription for parenteral corticosteroids was
3.3% and 2.2% were switched from oral to
parenteral corticosteroids.
Documentation of the indication for corticosteroid use was clear in notes for 67%, deduced
for 9%, and not clear for 24%. A wide range of
indications was noted, with raised intracranial
pressure and general boost accounting for
nearly half of cases (Table 4). All three patients
who received parenteral corticosteroids while
at the hospice for their final stay had tumor
compression, raised intracranial pressure in
two and superior vena cava obstruction in the
other.

Table 4
Observed Indications for Corticosteroids During a
Terminal Hospice Admission (n  90)
Indication

Raised intracranial pressure

Boost/improve well-being
Paina
Disease modificationb
Post-radiotherapyc
Dyspnea
Anorexia
Superior vena cava obstruction
Spinal cord compression

24
22
13
13
10
7
4
3
3

aMainly

liver pain (in 89% of cases).

example, in hematological malignancies, carcinoma of the
prostate, or chronic obstructive pulmonary disease.
cMainly following crainial radiotherapy (86% of cases).
bFor

331

The overwhelming reason cited for withdrawing corticosteroids was loss of the oral route (in
79% of cases). The only other cited reason for
stopping corticosteroids was failed therapeutic
trial in 7% of cases. In the remaining 14% of
cases, there was no documented or evident reason for stopping corticosteroids.

Discussion
The study sample included 90 patients who
received corticosteroids during their terminal
stay in a hospice. This is the largest series of
this type published to date. Case note retrieval
appeared sufficient, with only 11% missing as
misfiled or inaccessible as in use elsewhere,
for example, for bereavement follow-up. The
patient group selected sufficiently mirrored
the hospices patient demographics. They were
slightly younger and had more males than expected, possibly because of an excess of glioma
and carcinoma of the prostate patients (receiving corticosteroids as hormone manipulation).
There were limitations to this studys retrospective case note review design. The data
available and any deductions may reflect documentation rather than practice. Equally important, the short time frame and the single unit
studied means the results may reflect a handful
of palliative medicine practitioners, now dating
back four years, rather than a current standard approach. Additionally, it was not possible to determine adequately the preceding duration of corticosteroid use, and, in particular,
the rate of tapering corticosteroids in those patients admitted for their final stay who were already receiving corticosteroids. A prospective
study of all hospice admissions may help in this
regard. Nonetheless, our results do appear fairly
consistent with other published work.1,3,16,17

Prevalence of Corticosteroid Prescribing

Fifty-one percent of patients received corticosteroids at some point in their final admission, with 22% of these receiving their corticosteroids merely for a boost. Approximately
60% of these patients received corticosteroids
on the day they died. This means that corticosteroids were prescribed on the day of death in
27% of all deaths over the 6-month study period (178 patients). Although there are many
arguments for prescribing corticosteroids in the
terminal phase (Table 5), these figures may

332

Gannon and McNamara

represent overenthusiastic initiation of corticosteroids. The 51% appears to reflect higher corticosteroid prescribing during the terminal
phase, rather than a general trend at our unit,
and was substantially higher than the 33% of
patients discharged from the hospice on corticosteroids seen in a 1-month sample. A causal
relationship between dying and the prescription of corticosteroidsnamely, corticosteroids
being prompted by the symptoms, possibly
unrecognized, of dyingwould be worthy of
further study. Alternatively, inadequate supervision to discontinue corticosteroids appropriately during earlier management could result
in the observed high use of corticosteroids at
the end of life (Table 6). Why the withdrawal
of corticosteroids may be unnecessarily delayed until the terminal phase is not clear. It
could be that the issue of discontinuing corticosteroids is more difficult to address before
patients reach the very final stages for the patient, their carers, and staff.

Abrupt Withdrawal of Corticosteroids Once the

Oral Route Is Lost
The high number of patients dying on oral
corticosteroids suggests that corticosteroids were
perceived to have a major role in symptom
control, even up until a patients death. However, this contrasted with the low use of
parenteral corticosteroids in the terminal phase.
Only 2% of patients were switched to parenteral
corticosteroids (0.4% of all deaths) and no patient had corticosteroids initiated parenterally.
Though unnecessary drugs are discontinued
in the terminal phase, our usual practice would
be to replace key oral agents with parenteral
formulations, where available, to maintain symptom control. In a 1-month sample of 34 patients
dying at our unit, 100% received syringe driver
infusions. The overwhelming reason for the
abrupt withdrawal of corticosteroids in the study
was loss of the oral route, implying a per-

Vol. 24 No. 3 September 2002

ceived change in the relative risks/benefits of

corticosteroids once the oral route is lost as
part of a terminal decline. This perception
could imply a reduced likelihood of detectable
corticosteroid benefit, which could not be
matched with alternative approaches, or a reduced likelihood of distress from (or detectable) corticosteroid withdrawal effects requiring intervention. It could also suggest a changed
clinical picturea clearer prognosis measured in
days allowing a different therapeutic approach
leading to the discontinuation of nonessential
drugs, for example, antihypertensives. Finally,
it may suggest a desire to avoid the additional burden of a second syringe driver related to known incompatibility problems.
We suspect that these practices reflect the
convention within other hospices. Historically, loss of the oral route has been seen as a
natural or convenient time to stop corticosteroids in the terminal phase. However true,
our clinical practice must still reflect best pharmacological evidence. Despite anecdotal reassurance,8 there is little evidence to justify the
abrupt withdrawal of corticosteroids seen in
the terminal phase. In view of the well-established disadvantages (Table 3) and the suggestion that an abrupt corticosteroid reduction
may exacerbate terminal restlessness,16 there
appears some justification in maintaining accepted protocols from other settings (to withdraw chronic corticosteroids slowly), as there is
no comparable evidence set in the terminal
phase. In particular, the loss of the oral route
alone may be insufficient justification for discontinuation of corticosteroids when parenteral
formulations exist. That said, existing guidelines for the general patient population7,10,12 may
not be as practical in the terminal care setting,
for example, reducing prednisolone by 1 mg
every month.12
There is a paradoxical argument for continuing patients long-term corticosteroids when

Table 5
Possible Reasons for Prescribing Corticosteroids in the Terminal Phase
To ensure adequate symptom control (e.g., malignant spinal cord compression) and within 3 months of whole brain
radiotherapy.21
Potentially life-prolonging treatment (e.g., in raised intracranial pressure, and superior vena cava obstruction).
Empirically, to ensure any reversible components responsible for the patients decline are fully addressed.
To match the physiological increase in cortisol for dying patients with a suppressed hypothalamic-pituitary-adrenal axis.
To avoid the fear that the discontinuation of corticosteroids in the terminal phase could mistakenly be implicated as the cause
of the patients death.

Vol. 24 No. 3 September 2002

Corticosteroid Use at the End of Life

333

Table 6
Reasons for Concern at the High Corticosteroid Prescribing Remaining in the Terminal Phase
51% appears in excess of previously published hospice figures of corticosteroid use of 1 in 3.1
Little robust evidence to support present use.4,5
Overall only 33% of patients would be expected to benefit from corticosteroids 3, and these would be time limited.1
Corticosteroid side effects are cumulative; exacerbated by hypoalbuminemia; and can have a disproportionate impact in the
terminal phase (e.g., myopathy and oral thrush).
Polypharmacy/drug interactions.
Corticosteroids in the terminal phase may be considered an unnecessary medicalization, sending misleading messages of
positive management (i.e., perceived as life-prolonging treatment).
Though it appears increasingly difficult to wean patients off corticosteroids in the terminal phase as death approaches, this
leads to heightened issues around abrupt withdrawal when the oral route is lost.

they are no longer receiving oral or parenteral

hydration with a prognosis measurable in days.
By this point, many of the advantages of stopping their corticosteroids would be lost. Arguably, the risks of abrupt withdrawal are heightened and the limited time would prevent a
gradual withdrawal (Table 7).
Thus, these findings may indicate underprescription of parenteral corticosteroids, which are
acknowledged as appropriate for specific indications.17 such as raised intracranial pressure,
superior vena cava obstruction, and malignant
spinal cord compression. These indications
were present in nearly one-third of the study
patient group. The appropriate withdrawal of
corticosteroids in the terminal phase following
a failed therapeutic trial (of less than 3 weeks
duration) only accounted for a tiny percentage
of abrupt withdrawals.

Limited Documentation
An important study finding was the incomplete documentation of corticosteroid use. Adequate documentation is essential to continuity
of care with all interventions, but particularly
with corticosteroids in view of the ramifications
discussed. Though deemed essential,23 the

reason(s) for prescribing corticosteroids was

only documented in two-thirds of cases, and
the reason for withdrawal in less. In no set of
notes was a treatment plan pertaining to corticosteroids found sufficient to guide subsequent management.
The study suggested opportunities for greater
vigilance in corticosteroid prescribing and documentation to limit any excess in corticosteroid use leading up to the terminal phase. We
suspect that in a number of patients, particularly those receiving longer-term corticosteroids for non-specific indications, these drugs
could have been discontinued safely before the
terminal phase, as any short-term benefits would
have waned. This would have prevented unnecessary toxicity, and avoided the difficult management decisions surrounding corticosteroids
around the time of death. To achieve this, we
would advocate better adherence to time-limited
trials (of 23 weeks), an avoidance of maintenance doses, and efforts to ensure that doses
are continually reduced to the minimum sustaining benefit (or stopped when benefit lost).
In addition, we believe that there is a subgroup of patients whose corticosteroids should

Table 7
Reasons to Maintain Corticosteroids in the Terminal Phase22
To avoid uncertainty (in staff and relatives) as to the possible contribution of a reduction in corticosteroids to the patients
ongoing decline.
To maintain established optimal symptom control measures, rather than embarking on empirical management when time is
short and risking rebound symptoms.
To avoid inducing withdrawal symptoms, that may require additional medications.
Any distress resulting from corticosteroid withdrawal in the terminal phase may not be obvious; for example, as non-specific
(e.g. increased stiffness) or in locked in patients unable to communicate.
Benefits from stopping corticosteroids are not likely to be realized in the available time frame. Specifically, there is little
evidence to support the fear that corticosteroids prolong dying in the terminal phase, and parenteral corticosteroids would
only be needed for a short time and arguably not particularly burdensome.
There is sufficient time for a controlled withdrawal as per recommendations.14
There is a rationale for increasing corticosteroids in the terminal phase (to match physiological response).

334

Gannon and McNamara

be continued to maintain optimal symptom

control. These patients require parenteral corticosteroids once the oral route is lost, though
this role remains to be proven. If ongoing corticosteroid benefits are monitored and documented as standard practice, the identification
of this group of patients should be easier. However, when corticosteroids offer uncertain benefit in the terminal phase, it is still unclear how
they should be withdrawn (if at all).
In order to address these issues specific corticosteroid guidelines have been prepared for
our unit, although practice remains empirical
while more robust evidence is awaited.

Acknowledgments
The authors would like to thank The Princess Alice Hospicein particular, the Clinical
Secretaries Department.

References
1. Hardy J. Corticosteroids in palliative care. Eur J
Pall Care 1998;5(2):4650.
2. Hanks GW, Trueman T, Twycross RG. Corticosteroids in terminal cancera prospective analysis of
current practicepostgraduate. Med J 1983;59:702
706.
3. Twycross R. The risks and benefits of corticosteroids in advanced cancer. Drug Safety 1994;11(3):
164178.
4. Tiernan E. The use of steroids. In: Kaye P, Ed.
Tutorials in Palliative Medicine. Northampton,
United Kingdom: EPL, 1997, 295312.
5. Anonymous. Corticosteroids: useful palliative
role in cancer but use with caution. Drugs Ther Perspectives 1995;5(7):911.

Vol. 24 No. 3 September 2002

formulary. Oxon, United Kingdom: Radcliffe Medical Press, 1998.

9. Committee on Safety of Medicines and Medicines Control Agency. Using long-term systemic corticosteroids safely. Curr Prob Pharmacovigilance
1997;23:5.
10. Kountz DS, Clark C. Safely withdrawing patients
from chronic glucocorticoid therapy. Am Fam Phys
1997;55(2):521530.
11. Wood AJJ. Corticosteroid therapy in severe illness. N Engl J Med 1997;37(18):12851292.
12. Page RC. How to wean a patient off corticosteroids. Prescribers J 1997; 37:1;1116.
13. Lovinger RD. Use and abuse of adrenal corticosteroids. Virginia Med 1983;110:319321.
14. Committee on Safety of Medicines and Medicines Control Agency Expert Working Group. Focus
on corticosteroids. Curr Prob Pharmacovigilance
1998;24:57.
15. Howser RL. What you need to know about corticosteroid therapy. Am J Nurs 1995;95(8):4448.
16. Hardy JR, Rees E, Ling J, et al. A prospective survey of the use of dexamethasone on a palliative care
unit. Palliat Med 2001;15:38.
17. Wand J, Campbell C. Dexamethasone administration in the last 14 days of life in patients dying
with primary brain tumors. Poster, 1st Congress of
the Research Network of the European Association
for Palliative Care, Berlin, December 2000.
18. Gannon C, McNamara P. A retrospective study
of steroids during terminal care. Poster at Palliative
Care Congress University of Warwick, March 2000
[published abstract Palliat Med 2000;14:4:350].
19. Calman K, Hanks GWC. Clinical and health services research. In: Doyle D, Hanks GWC, MacDonald N, eds. Oxford textbook of palliative medicine, 2nd edition. United Kingdom: Oxford University
Press, 1998:159165.
20. Grande GE, Todd CJ. Why are trials in palliative
care so difficult? Palliat Med 2000;14:6974.

6. Ralls JE. To dex or not to dex: a descriptive

account of corticosteroid use in palliative medicine.
Aust Fam Phys 1996;25(4):713715.

21. Oneschuk D, Bruera E. Palliative management

of brain metastases. Support Care Cancer 1998;6:
365372.

7. Anonymous. British National Formulary 37.

London, United Kingdom: British Medical Association and Royal Pharmaceutical Society of Great Britain, 1999.

22. Gannon C. The prescription of steroids in the

terminal phase [letter]. Palliat Med 2001;15:522.

8. Twycross R, Wilcock A, Thorp S. Palliative care

23. Twycross R. Corticosteroids in advanced cancerif they are not working, stop them. BMJ 1992;
305;969999.