438 Journal of Pain and Symptom Management Vol. 25 No.

5 May 2003

Original Article

Antimicrobial Use in Patients with Advanced

Cancer Receiving Hospice Care
Patrick H. White, BS, Heather L. Kuhlenschmidt, BS, Benjamin G. Vancura, BA,
and Rudolph M. Navari, MD, PhD
Walther Cancer Research Center, University of Notre Dame, Notre Dame, Indiana, USA

Abstract
Patients with advanced cancer receiving hospice and palliative care are highly susceptible to
infections. The decision whether to treat an active or suspected infection in end-of-life care may
be difficult. In order to develop guidelines for the use of antimicrobials (antibiotics and
antifungals) in palliative care, we discussed antimicrobial options with 255 patients with
advanced cancer at the time they entered a community-based hospice and palliative care
program. We subsequently documented the use and effectiveness of the antimicrobials employed
during the palliative care period. Most patients (79.2%) chose either no antimicrobials or
symptomatic use only. Choices were influenced by age, the desire for symptom control, life-
prolongation issues, and the condition of the patient. After admission, 117 patients had a
total of 129 infections, with the most common sites being urinary tract, respiratory tract,
mouth/pharynx, and skin/subcutaneous tissues. The use of antimicrobials controlled
symptoms in the majority of the urinary tract infections, but were less effective in controlling
symptoms in the other sites of infection. Survival was not affected by the patients’ choice of
whether to use antimicrobials, the prevalence of infections, or the actual use of antimicrobials.
Symptom control may be the major indication for antimicrobial use for patients receiving
hospice and palliative care. J Pain Symptom Manage 2003;25:438–443. © 2003
U.S. Cancer Pain Relief Committee. Published by Elsevier. All rights reserved.

Key Words
Antimicrobials, advanced cancer, hospice, symptom control

Introduction 102 patients (92% terminal malignant illness)

who died after admission to a tertiary care in-
There have been few reports on the use of
patient palliative care unit.
antimicrobials in patients with advanced can-
Thirty-seven patients were diagnosed with 42
cer who are receiving hospice or palliative
infections, with urinary tract, respiratory tract,
care.1,2 Vitetta et al.1 performed a retrospective
chart review on the prevalence of infections in blood, skin and subcutaneous tissues, and eyes
as the most common sites of infection. E. Coli
was the most common organism. Thirty-five of
37 patients were treated with antibiotics, with a
Address reprint requests to: Rudolph M. Navari, MD,
PhD, Walther Cancer Research Center, 250 Nieuw- symptom response in half of the patients
land Science Hall, University of Notre Dame, Notre treated; 2 of 37 patients were not treated with
Dame, IN 46556, USA. antibiotics due to survival limited to the day of
Accepted for publication: July 22, 2002. admission.

© 2003 U.S. Cancer Pain Relief Committee 0885-3924/03/$–see front matter

Published by Elsevier. All rights reserved. doi:10.1016/S0885-3924(03)00040-X
Vol. 25 No. 5 May 2003 Antimicrobial Use in Hospice Patients with Advanced Cancer 439

Pereira et al.2 reported a retrospective chart program included 309 consecutive patients
review of the prevalence of infections in 100 with advanced cancer, whose projected prog-
consecutive admissions to a tertiary care pallia- nosis was less than six months. Patients were
tive care unit. There were 74 infections in 55 asked about their preference about future anti-
patients, with urinary tract, respiratory tract, microbial use. A discussion was performed with
skin and subcutaneous tissue, blood, and the admitting hospice nurse using a uniform
mouth as the most common infection sites. E script, which included the potential advantages
coli, S. aureus, and Enterococcus were the most and disadvantages of the use of antimicrobials
common organisms. Twenty-one of the 74 in- in a palliative care setting. The antimicrobial
fections were not treated. The reasons for not use discussion complemented the standard dis-
using antimicrobials were documented in 10 cussion of the goals of the hospice and pal-
patients: very poor general condition in 5, not liative care program, including the use of re-
able to take oral antimicrobials and refusal of suscitative measures. Fifty-four patients were
parenteral antimicrobials in 3, and family re- excluded because they were taking antimicro-
fusal in 2. The retrospective nature of the study bials at the time of admission, did not have de-
did not allow for an adequate analysis of the cision-making capacity, or declined to partici-
symptom response to antibiotic therapy. pate in the discussion. Participating patients
The decision whether to treat an infection gave informed consent as approved by the Uni-
with antimicrobials in a patient with advanced versity Committee on the Protection of Human
cancer receiving palliative care may be diffi- Subjects, University of Notre Dame.
cult. Life-prolonging antimicrobials may or The hospice and palliative care program was
may not be appropriate in a palliative care set- the predominant palliative care provider in a
ting. It is difficult in many situations to predict five-county area, which included a city of ap-
whether the use of antimicrobials will provide proximately 300,000. All patients had a care-
symptom relief, affect survival, or prolong the giver and received their care at their place of
dying process. Although antimicrobials may be residence. Hospice nurses visited the patients
the most appropriate means of symptom con- on a routine schedule and were available at any
trol in the presence of certain infections, they time for additional visits.
may also be associated with symptom-produc-
ing interventions, such as laboratory testing, Procedures and Outcomes
venous access, and direct antimicrobial toxici- Patients were asked to elect one of three op-
ties. The patient’s overall condition and prog- tions: (1) Option A: full antimicrobial use for
nosis, the wishes of the patient and family, and suspected or established infections as would be
the potential for symptom control all are im- done in acute medical or surgical care, (2) Op-
portant aspects of the decision to prescribe an- tion B: antimicrobial use for symptomatic treat-
timicrobials. The patient’s status in the pallia- ment only, or (3) Option C: no antimicrobial
tive care setting and the goals of care may be use. For the purpose of the study, antimicro-
the most important determinants of the use of bials were defined as antibiotics or antifungal
antimicrobials. agents. Antimicrobial use for symptomatic treat-
In order to develop guidelines for the use of ment only was defined as the institution of an
antimicrobials in palliative care, we discussed antimicrobial to treat patients’ symptoms at-
antimicrobial options with patients having ad- tributed to an active or suspected infection. If a
vanced cancer at the time they elected end- patient was undecided about the use of antimi-
of-life palliative care and subsequently docu- crobials at the time of the initial discussion, op-
mented the use and effects of antimicrobials in tion A was employed in his/her care. Patients
this patient population. were invited to offer a rationale for their choice
and were informed that they could alter their
choice at any time. Patients in each group were
Methods subsequently followed for the duration of their
Setting and Participants participation in the hospice and palliative care
Patients were surveyed during a six-month program.
period in 2001. Admissions to a community- When an infection was suspected clinically,
based outpatient hospice and palliative care symptoms (fever, dyspnea, dysuria, pain) were
440 White et al. Vol. 25 No. 5 May 2003

recorded, clinically indicated cultures were ob- Table 1

tained, and antimicrobials were considered by Patient Characteristics
the attending physician according to the op- Number of Patients 255
tion chosen by the patient. For the purpose of Age Range 47–89 years
the study, an “infection” was defined as the Mean 63.1 years
Male: 50%; Female: 50%
presence of symptoms and physical signs that Spousal Status
were interpreted by the attending physician to Married 69.4%
have been caused by a microbial agent and was Divorced 12.1%
Single 9.0%
documented as such in the patient’s chart. The Widowed 9.5%
antimicrobial used, the route and duration of Caregiver
use, and the presence of risk factors for infec- Spouse 58.1%
Child 23.9%
tion, such as urinary catheters, central venous Parent 6.3%
catheters, and/or the use of corticosteroids, Other 11.7%
were recorded. Patients were subsequently moni- Diagnosis
Colon 21.9%
tored to determine the prevalence of infec- Breast 23.3%
tions, the actual use of antimicrobials, the ef- Lung 25.9%
fects of antimicrobials on infection related Prostate 5.9%
Ovarian 4.5%
symptoms, and the overall and infection re- Lymphoma 4.1%
lated survival. Colonization was considered if Leukemia 5.7%
there was an absence of inflammatory cells in Bladder 2.5%
Brain 1.6%
the cultured specimen. Treatment
Chemotherapy 85.1%
Statistics Radiation 71.9%
Surgery 72.1%
The Mann-Whitney and the Wilcoxon non- Hormonal 21.0%
parametric tests were used to compare groups Experimental 5.9%
with and without an outcome of interest for or- Payers
Medicare 49.6%
dinal and continuous data that were not nor- Private 35.7%
mally distributed. Medicaid 17.5%
HMO 33.5%
Uninsured 2.9%
Results
Patient Characteristics
Table 1 lists the characteristics of the study Option B or C were significantly older (P 
patients at the time of admission to the hospice 0.01), and had significantly lower Karnofsky
program. The majority of the patients had colon, scores (P  0.01), compared to patients who
breast, lung, or prostate cancer; had received ex- chose Option A. The three groups did not dif-
tensive treatment in the form of chemotherapy, fer according to gender, spousal status, care-
radiation, surgery, and/or hormonal therapy; givers, diagnosis, previous treatment, payers, or
and had various forms of health care insur- medication for pain, depression, or anxiety.
ance. Table 2 illustrates that over 80% of the The rationale offered by patients for their
patients had a Karnofsky performance of  choices were as follows: For Option A, 43 of 53
60%, and many were receiving medication for patients (81%) requested measures for prolon-
pain, depression, and anxiety. gation of life as well as symptom control. For
Option B, 107 of 123 patients (87%) requested
Patient Choices and Antimicrobial Use symptom control only. For Option C, 50 of 79
Table 3 shows the choices of the patients patients (63%) requested no interventions that
with regard to antimicrobial use and the actual might prolong life primarily due to their very
use during their participation in the hospice poor condition, 10 patients (13%) were not
program. Most of the patients (79.2%) chose able to take medications by mouth and refused
either no antimicrobials or symptomatic use parenteral antimicrobials, and 10 patients
only. The actual antimicrobial use was signifi- (13%) requested no further interventions. The
cantly higher (P  0.01) in option A compared remainder of the patients in each group of-
to Option B or Option C. Patients who chose fered no rationale.
Vol. 25 No. 5 May 2003 Antimicrobial Use in Hospice Patients with Advanced Cancer 441

Table 2 nosed with an infection. Twenty-two patients

Patients Status had one or more positive cultures.
Karnofsky Performance Scale Percent

100% Normal; no complaints, no evidence

Infection Sites and Treatment Response
of disease 0 One-hundred seventeen of the 255 patients
90%: Able to carry on normal activity; minor were diagnosed with a total of 129 infections.
symptoms 0
80%: Normal activity with effort; some
One-hundred eight patients were diagnosed
symptoms 5 with one infection, six patients were diag-
70%: Cares for self; unable to do active work 13 nosed with two separate infections, and three
60%: Requires occasional assistance; can
care for most needs 30
patients were diagnosed with three infec-
50%: Requires considerable assistance and tions.
frequent medical care 39 Table 4 shows the sites of infection, the anti-
40%: Disabled; requires special care and
assistance 10
microbials used, and the symptom response.
30%: Severely disabled, hospitalization is The majority of the patients treated for urinary
indicated 3 tract infections had improvement in their ini-
20%: Hospitalization necessary; active
supportive treatment 0
tial symptoms, but fewer than half the patients’
Medications symptoms responded to antimicrobial used for
Pain 77.6 infections of the respiratory tract, mouth/
Antidepressants 37.6
Anti-anxiety 41.8
pharynx, skin/subcutaneous tissues, or blood.
Eighty-one patients had a total of 99 culture-
positive infections, with 127 difference organ-
isms involved (Table 5). The most frequent
urinary tract organisms were E. coli, Enterococ-
Thirty of the 53 patients who chose Option
cus, and Klebsiella pneumoniae, with the majority
A received antimicrobials for diagnosed infec-
being sensitive to trimethoprim/sulfamethox-
tions. Twenty of the thirty patients had one or
azole (TMP/SMX), ciprofloxacin, and amox-
more positive cultures.
acillin. The most common respiratory tract
Forty-five of the 123 patients who chose Op-
pathogens were Staphylococcus aureus, Klebsiella
tion B received antimicrobials for a diagnosed
pneumoniae, and Pseudomonas aeruginosa, and
infection. Eleven additional patients who
these were most commonly sensitive to levof-
chose option B were judged to have an infec-
tion but did not receive antimicrobials due to
the absence of symptoms. Thirty-nine patients Table 4
in this group grew one or more organisms Infection Sites, Antimicrobials Prescribed, and
from the cultures taken. Symptom Response
Two of the 79 patients who chose Option C Symptom
received antimicrobials. Both of these patients Response
Infection Site (na) Antimicrobialb (nc) (%)
had severely symptomatic thrush and antimi-
crobial treatment was strongly recommended Urinary Tract (54) TMP/SMX (13) 92
despite their initial request for no antimicrobi- Ciprofloxacin (9) 89
Amoxacillin (5) 60
als. Thirty-one of the 79 patients were diag- Norfloxacin (3) 67
Respiratory Tract (45) Levofloxacin (13) 46
TMP/SMX (6) 50
Azithromycin (3) 33
Table 3 Gatifloxacin (2) 0
Patient Choices and Antimicrobial Use Clarithromycin (2) 0
Patient Choices and Antimicrobial Use n (%) Mouth/Pharynx (13) Fluconazole (6) 50
Nystatin (3) 33
Full Use (A) 53 (20.8) Skin/Subcutaneous (12) Cephalexin (6) 50
Symptomatic Use (B) 123 (48.2) TMP/SMX (3) 33
No Antimicrobials (C) 79 (31.0) Blood/Bacteremia (5) Cetriazone (2) 0
Levofloxacin (1) 0
Actual Antimicrobial Use n (%) a Number of infections.
bAntimicrobials were given by mouth except for parenteral adminis-
Option A 30 (56.6)
Option B 45 (36.6) tration in the bacteremias.
c Number of patients treated with this agent.
Option C 2 (2.5) TMP/SMX  trimethoprim-sulfamethoxazole.
442 White et al. Vol. 25 No. 5 May 2003

Table 5 Table 6
Overall Frequency of Organisms Patient Survival
Organism n (%) Infection
Antimicrobial Survival Related
Escherichia coli 29 (22.9) Option (median days) Deaths (n)
Staphylococcus aureus 19 (14.9)
Enterococcus species 14 (11.0) Option A 26.3 3
Klebsiella pneumoniae 10 (7.9) Option B 30.1 6
Candida albicans 9 (7.0) Option C 29.7 4
Proteus mirabilis 8 (6.3)
Pseudomonas aeruginosa 8 (6.3)
Coagulase-negative staphylococci 7 (5.5)
Hemophilus influenza 5 (3.9)
Other 18 (14.3) fections. Patients’ choices were based primarily
on symptom control, the condition of the pa-
tient, and whether to employ life-prolonging
interventions. The choice of the restricted use
loxacin and TMP/SMX. Oral thrush was clini- of antimicrobials was more common among
cally suspected in nine of the thirteen mouth/ patients who were older and had lower perfor-
pharynx infections and was treated with anti- mance levels. The choices were made after a
fungal agents. Staphylococcus aureus was respon- full disclosure and discussion of the potential
sible for the skin and subcutaneous infections advantages and disadvantages of the use of an-
and was sensitive to cephalexin and TMP/ timicrobials and were independent of the type
SMX. Vancomycin-resistant Enterococcus species of malignancy, gender, caregiver or spousal sta-
was isolated from one patient. There were two tus, previous treatment, health care payer, or
isolates of methicillin-resistant Staphyloccus au- medications for pain, depression, or anxiety.
reus in this patient population. Two retrospective chart reviews3,4 reported
Seventeen of the 117 infected patients that incurably ill dying hospitalized patients
(14.5%) had bladder catheters prior to, or at commonly received systemic antibiotics in
the time of infection, compared to 16 of the their last days or weeks of life, often empiri-
138 noninfected patients (11.6%) (PNS). cally. In these reviews, antimicrobial use was
Forty-three of the 117 infected patients common in patients with “do not resuscitate”
(36.7%) were receiving systemic corticoster- or “comfort measures only” orders. The cur-
oids at the time of infection, compared to 44 of rent study suggests that if these patients were
the 138 noninfected patients (31.9%) (PNS). given a choice of antimicrobial use, many may
All of the nine patients treated for oral thrush have restricted antimicrobial use.
were receiving systemic corticosteroids. The types of infections, the antimicrobials
employed, and the organisms found in this pa-
Patient Survival tient population were similar to that seen in
There was no significant difference in sur- two previous studies,1,2 despite the differences
vival between the patients with a diagnosed in- in location and setting (outpatient residential
fection (n117) and the group without an in- versus inpatient institutions). Forty-six percent
fection (n138). Patients’ overall survival and of the patients experienced at least one infec-
infection-related deaths were not significantly tion in this study, which compares to 36.3%
affected by the use of antimicrobials or the pa- and 55% in the two previous studies.1,2 Urinary
tients’ choices of antimicrobial use (Table 6). tract infections, infections of the respiratory
The thirteen infection-related deaths consisted tract, and infections of the skin/subcutaneous
of patients with infections of the respiratory tract tissue made up the majority of infections, with
(9), blood (3), and subcutaneous tissue (1). E. coli, Staphylococcus aureus, and Enterococcus the
main isolated organisms in the present and the
past studies. The similarity of the infections
Discussion may suggest that patients with advanced cancer
A large majority of patients with advanced may have common patterns of infection during
cancer who entered a community outpatient their final phase of care.
hospice program chose either not to use anit- The use of urinary catheters or the use of
microbials or limit their use to symptomatic in- corticosteroids did not increase the incidence
Vol. 25 No. 5 May 2003 Antimicrobial Use in Hospice Patients with Advanced Cancer 443

of overall infections in this group of patients. by their age, current condition, and concerns
There were more oral thrush infections in this about life-prolonging interventions. The use of
study compared to the rate reported in the lit- antimicrobials controlled symptoms from uri-
erature, and these infections certainly could be nary tract infections, but were less effective in
attributed to the use of corticosteroids. controlling symptoms in infections of the respi-
Although the use of antimicrobials im- ratory tract, mouth/pharynx, skin/subcutane-
proved symptoms in a large majority of pa- ous tissue, or blood. Although a formal quality of
tients with urinary tract infections, symptom life measure was not used in this study, patients
control was less successful with antimicrobial were clinically improved when their symptoms
use in infections of the respiratory tract, mouth/ responded to the use of antimicrobials. Overall
pharynx, skin/subcutaneous tissue, or blood. survival and infection related deaths were not al-
Since the majority of the organisms cultured tered by the use of antimicrobials. This informa-
were sensitive to the antimicrobials used, the tion should aid patients and clinicians in deci-
lack of symptom response in some patients sions concerning the use of antimicrobials for
may have been due to co-morbid conditions, infections in patients with advanced cancer re-
such as an immunocompromised state, malnu- ceiving hospice care. Additional studies will be
trition, the failure of host barriers, decreased needed to generate clinical guidelines for the
level of consciousness or immobility, or the use of antimicrobials in this patient population.
presence of a neoplasm in the symptomatic or-
gan. This pattern of symptom response was
also seen in the study of Vitetta et al.,1 who re- Acknowledgments
ported a similar symptom response of 40-50%
This work was supported by the Walther
in terminally ill hospice patients
Cancer Institute and The Reich Family Endow-
Symptom response to antimicrobials may
ment for Excellence on Care of the Whole Pa-
vary widely among patients with advanced life-
tient.
threatening diseases. Fabiszewski et al.5 re-
ported that treatment of fever with antibiotics
did not alter the outcome of fever in institu-
tionalized advanced Alzheimer patients, while References
some authors have reported improved pain 1. Vitetta L, Kenner D, Sali A. Bacterial infections in
control with the use of antimicrobials in a terminally ill hospice patients. J Pain Symptom Man-
age 2000; 20:326–334.
small number of patients with acute systemic
or local infections.6–8 2. Pereira J, Watanabe S, Wolch G. A retrospective
review of the frequency of infections and patterns of
Patient survival in this study was not affected antibiotic utilization on a palliative care unit. J Pain
by the presence of infection or the use of anti- Symptom Manage 1998; 16:374–381.
microbials. In the study by Vitetta et al.,1 pa- 3. Ahronheim JC, Morrison RS, Baskin SA, et al.
tients with documented infections had a longer Treatment of the dying in the acute care hospital.
median survival. The proposed explanation for Arch Intern Med. 1996; 156: 2094–2100.
the increased survival was that the probability of 4. Goodlin SJ, Winzeberg GS, Teno JM, et al. Death in
infection increases with the duration of survival the hospital. Arch Intern Med1998; 158: 1570–1572.
rather than an increased survival due to the use 5. Fabiszewski KJ, Volicer B, Volicer L. Effect of anti-
of antimicrobials. The effect of the use of anti- biotic treatment on outcome of fevers in institutional-
microbials on survival would be very important ized Alzheimer patients. JAMA 1990; 263: 3168–3172.
information to patients entering hospice care. 6. Bruera E. Intractable pain in patients with ad-
This information might strongly influence their vanced head and neck tumors: a possible role of lo-
choice of antimicrobial use. cal infection. Cancer Treat Rep 1986;70:691–692.
7. Green K, Webster H, Watanabe S, Fainsinger R.
Case report: management of nosocomial respiratory
Conclusions tract infections in terminally ill cancer patients. J
Palliative Care 1994;10:31–34.
The major conclusions of this study are that
8. MacKey J, Birchall I, MacDonald N. Occult infec-
patients with advanced cancer entering hospice tion as a cause of hip pain in a patient with meta-
care chose limited use of antimicrobials, mainly static breast cancer. J Pain Symptom Manage 1995:
for symptomatic care. The choices were affected 10:569–572.