© U.S. Cancer Pain Relief Committee, 1998 0885-3924/98/$–see front matterPublished by Elsevier Science, New York, New York PII S0885-3924(98)00093-1

374 Journal of Pain and Symptom Management Vol. 16 No. 6 December 1998

Original Article

A Retrospective Review of the Frequency of Infections and Patterns of Antibiotic Utilization on a Palliative Care Unit

Jose Pereira, MD, Sharon Watanabe, MD, and Gary Wolch,* MD

Regional Palliative Care Program, Grey Nuns Community Hospital, Edmonton, Canada

Abstract

Various disease-related and therapy-induced factors make cancer patients susceptible to infections. The epidemiology and management of infections in neutropenic cancer patients and patients with primary hematological malignancies has been widely reported, but very few studies have characterized infections and their management in palliative care patients. We conducted a retrospective review of 100 consecutive admissions to an acute palliative care unit with the objective of assessing overall and site-specific frequencies of infections, the pathogens involved and their antibiotics sensitivities, and the pattern of antibiotic utilization. The mean

age was 64

6

11.5 years, the mean length of stay was 29.8

6

28.2 days, and 70% of patients died on the unit. Only one case of neutropenia was identified. Fifty-five of the 100 patients were diagnosed with a total of 74 separate infections. The most frequent sites of infections were the urinary tract (39.2%), the respiratory tract (36.5%), skin and subcutaneous tissues (12.2%), and blood (5.4%). Fifty-four culture-positive infections were identified. Overall, the most common organisms were

Escherichia coli

(22.9%),

Staphylococcus aureus

(20%), and

Enterococcus

(11.4%). Fifty-three of the74 (71.6%) infections were treated with antibiotics. The decision-making process regarding treatment versus nontreatment of an infection can be complex in terminally ill patients and needs to be individualized. Symptom control is the primary objective in the majority of cases. The appropriate management of infections, with specific attention to measures that would improve patients’ quality of life, should be a research priority in patients with advanced cancer.

J Pain

Symptom Manage 1998;16:374–381.

© U.S. Cancer Pain Relief Committee, 1998.

Introduction

Cancer patients are particularly susceptibleto infections due to a variety of disease-related

and therapy-induced factors.

1

Infections arethe cause of significant morbidity and mortal-

ity in patients with cancer.

2,3

The epidemiol-ogy of infections in cancer patients who areneutropenic or have a hematological malig-nancy has been extensively studied, and nu-merous guidelines for the management andantibiotic utilization in these patient are avail-able.

4–7

However, very little information existson the frequency, distribution, and appropri-ate management of infections in non-neutro-penic patients with advanced, nonhematologi-cal cancers.

8–10

In an attempt to characterize the epidemiol-ogy and management of infections in this lat-ter group of patients, we conducted a retro-spective review of 100 consecutive admissions

*Presently Palliative Care Consultant and FamilyMedicine Practice, Calgary, Alberta, Canada

Address reprint requests to:

Dr. Jose Pereira, RegionalPalliative Care Program, Grey Nuns Community Hos-pital, Room 5211, 1100 Youville Drive West, Edmon-ton, Alberta, Canada T6L 5X8.

Accepted for publication: March 31, 1998.

Vol. 16 No. 6 December 1998 Frequency of Infections in Palliative Care 375

to an Acute Palliative Care Unit (PCU). Theobjective was to assess overall and site-specificfrequencies of infections, the bacterial patho-gens involved and their antibiotic sensitivities,and the pattern of antibiotic utilization.

Methods

The charts of 100 consecutive cancer pa-tients admitted to a 14-bed, tertiary PCU in Ed-monton, Canada were reviewed. The unit ad-mits terminally ill patients who are sufferingfrom distressing physical and psychosocialproblems related to their illness. It is a teach-ing unit with affiliations to a medical univer-sity. Information collected included patient de-mographics, the documentation of a clinicaldiagnosis of an infection, and physicians’ or-ders and clinical chart notes regarding the in-fection. The retrospective nature of this studyand the influence of factors peculiar to thecare of terminally ill patients made it difficultto utilize stringent criteria such as those of theCenters for Disease Control (CDC) for the di-agnosis of an infection.

11

For the purpose ofthis study an “infection” was defined as thepresence of symptoms and physical signs thatwere interpreted and assessed by the attendingphysicians to have been caused by a microbialagent and led the attending physicians to doc-ument it as such in the patient’s chart. In mostcases the clinical diagnosis was confirmed bylaboratory and radiological investigations.

Nosocomial infections are commonly de-fined as infections that develop 72 hours ormore after admission.

12

Some authors havesuggested a shorter time frame of 48 hours tobe more appropriate for immunocompro-mised patients.

13

However, infections incubat-ing at the time of the patient’s admission to thehospital serve as a ready source of infection forother patients or personnel and they, too,cause significant morbidity and impact on thequality of life of patients in the PCU. There-fore this study included all infections, includ-ing those diagnosed in the first 72 hour afteradmission.

Positive bacterial cultures were identifiedand the microbial pathogens and their antibi-otic sensitivities were noted. These cultureswere derived from urine or sputum samples, orswabs of cutaneous/gastrostomy site discharges.The pattern of antibiotic utilization, includingthe antibiotic selected, the route, and the dura-

tion of administration, was recorded. Whereavailable, the reasons for not initiating or fordiscontinuing antibiotic therapy were noted.The presence of some risk factors, such as uri-nary catheters, subcutaneous and intravenouslines, and concurrent corticosteroid use, wasnoted.

A comparison of independent proportionswas performed using the normal approximation.

Results

Patient characteristics and demographics areshown in Table 1. Only one case of neutrope-nia (white cell count

,

4

3

10

9

/L ) was identi-fied in evaluable full blood counts of 84 pa-tients.

Fifty-five of the 100 patients (55%) were di-agnosed with a total of 74 separate infections.Forty-three patients were diagnosed with oneinfection, 8 patients were diagnosed with twoseparate infections, and 3 patients were diag-nosed with three infections. One patient expe-rienced four separate infections during the ad-mission. Infections were diagnosed after amedian of 9.5 days (range 0–112 days) fromthe day of admission. Twenty-one of all the in-fections (28.4%) were diagnosed within 48hours of admission. Of these, the infection hadbeen identified prior to the admission in 9cases (42.9%).

The most frequent sites of infection areshown in Table 2. Two of the four reportedcases of septicemia were confirmed by positiveblood cultures; the other two were suspectedon clinical grounds. Apart from the patients di-agnosed with septicemia, one other patient was

Table 1

Patient Characteristics

Total number of patients 100Age (mean

6

SD) 64

6

11.5Length of stay on unit

(mean

6

SD) 29.8

6

28.2 daysDied:Discharged 70:30Male:female 43:57Primary cancer

Gastrointestinal 30 (30%)Genitourinary 21 (21%)Lung 17 (17%)Breast 17 (17%)Head and neck 4 (4%)Hematological 2 (2%)Other 5 (5%)Unknown 4 (4%)

376 Pereira et al. Vol. 16 No. 6 December 1998

diagnosed with two concurrent infections atseparate sites. The infections were found in theurinary and the respiratory tracts.

Thirty-eight patients had a total of 54 cul-ture-positive infections with 70 different organ-isms involved. Overall, the most common or-ganisms were

Escherichia coli

,

Staphylococcusaureus

, and

Enterococcus

(Table 3). The mostfrequent urinary tract pathogens were

E. coli

,

Enterococcus

, and

Klebsiella pneumoniae.

The ma-jority of these were sensitive to norfloxacin, ni-trofurantoin, trimethoprim–sulfamethoxazole,ampicillin, and cefazolin (Table 4). The mostcommon respiratory tract pathogens obtainedby sputum culture were

S. aureus

,

Hemophilusinfluenzae

, and

Pseudomonas aeruginosa.

Thesewere most commonly sensitive to trimetho-prim–sulfamethoxazole and ciprofloxacin.

Sta-phylococcus aureus

accounted for the majority ofsubcutaneous infections and was sensitive totrimethoprim–sulfamethoxazole, cephalothin,oxacillin, clindamycin, and erythromycin.

Fifty-three of the 74 (71.6%) infections thatwere diagnosed were treated with antibiotics.The reasons for not treating the remaining 21infections were documented in 10 cases. In 5cases, very poor general condition and immi-nent death were cited. Three symptomatic in-fections were not treated because the patientswere unable to take medications by mouth andopted not to be treated with parenteral antibi-otics. In 2 cases, the charts indicated that thepatients and/or their families opted not to betreated with antibiotics despite the medicalstaff’s opinion that antibiotic therapy in thoseparticular patients would have probably beenbeneficial. In one case, chart documentationindicated that the family of a very ill, deliriouspatient insisted on antibiotic therapy despitereservations by the medical staff about the ef-

fectiveness of antimicrobial therapy at thetime. The patient died 2 days after initiatingparenteral antibiotic therapy.

Table 5 describes the pattern of antibiotic se-lection. Of the total of 72 antibiotic orders,both as single or combined therapy, trimetho-prim–sulfamethoxazole was the most com-monly prescribed antibiotic, making up 32 ofthe 72 orders (44.4%). This was followed byciprofloxacin in 17 cases (23.6%). The route ofadministration was oral in 52 cases (72.2%), in-travenous in 19 cases (26.4%), and intramuscu-lar in one case (1.4%). The latter case was a pa-tient with pneumonia who had severe coughingand dyspnea, was unable to swallow, and hadno access to the intravenous route. Single anti-biotic therapy was prescribed in the overwhelm-ing number of cases (65 of 72). Treated pa-tients were on antibiotics for a mean of 6.4

6

3.3 days (median: 7 days, range: 1–21 days).Eighteen of the 55 infected patients (32.7%)

had been catheterized prior to or at the time ofinfection, as compared to13 of the 45 nonin-fected patients (28.9%). This difference wasnot significant. The number of patients withsubcutaneous sites, namely 48 (87.3%) in theinfected group and 41 (91.1%) in the nonin-fected group was not significantly different be-tween the two groups. Twenty-six of the 55 in-fected patients (47.3%) were on a course ofcorticosteroids at the time of infection, as com-pared to 22 of 45 noninfected patients (48.9%);this difference too was not significant. Dexa-methasone, apart from one patient who receivedprednisone, was the corticosteroid prescribedin all cases. The maximum daily doses of dex-amethasone during the admission were thesame between the two groups, a median of 30mg/day (range 1–40 mg) (

P

5

0.22).

Table 2

Sites of Infection

Sites of infection

n

(%)

Urinary tract 29 (39.2)Respiratory tract 27 (36.5)Skin/subcutaneous tissues 9 (12.2)Blood/septicemia 4 (5.4)Colostomy/gastrostomy stomas 2 (2.7)Parotiditis 1 (1.4)Oral floor 1 (1.4)Unknown 1 (1.4)Total 74 (100)

Table 3

Overall Frequencies of Organisms

Organism

n

(%)

Escherichia coli

16 (22.9)

Staphylococcus aureus

14 (20.0)

Enterococcus

species 8 (11.4)

Klebsiella pneumoniae

5 (7.2)

Pseudomonas aeruginosa

4 (5.7)Coagulase-negative staphylococci 3 (4.3)

Hemophilus influenzae

3 (4.3)

Proteus mirabilis

3 (4.3)

Citrobacter freundi

3 (4.3)Other 12 (18)Total 70 (100)

Vol. 16 No. 6 December 1998 Frequency of Infections in Palliative Care 377

Discussion

This study indicates a high frequency of in-fections among non-neutropenic terminally illcancer patients with solid tumors being caredfor in a PCU (55% of patients). This high fre-quency supports the findings of two previousstudies in similar population groups. Lagmanet al. reported a frequency of 52% of patients

8

while a previous study by our group reported

an incidence rate of 13.0

6

6.7 infections per1000 patient days.

9

This incidence rate was sig-nificantly higher than rates reported by medi-cal and surgical acute care units and long-termcare units. The factors that are responsible forthis increased susceptibility are varied and in-clude impaired immunity, malnutrition, asthe-nia, decreased level of consciousness, immobil-ity, the failure of host barriers, and the use of

Table 5

Patterns of Antibiotic Prescriptions

Site of infection AntibioticTimes

prescribedRoute

(

n

)Empiric (E) or culture (C) (

n

)

Urinary tract First line TMP-SMX 15 PO Eciprofloxacin 1 PO Eamoxicillin 1 PO Eceftriaxone 1 IM Cclindamycin

1

ceftaxidime 1 IV CSecond line ciprofloxacin 5 PO C(4), E(1)

ceftazidime 1 IV CTMP-SMX 1 PO C

Respiratory tract First line TMP-SMX 12 PO Eciprofloxacin 5 PO Eciprofloxacin

1

metronidazole 1 IV Eclindamycin

1

ceftriaxone 1 IV Eamoxicillin 1 PO Ecefuroxime 1 PO E

Second line ciprofloxacin 3 PO C(1), E(2)Skin/subcutaneous

tissue/stomasFirst line TMP-SMX 3 PO E(1), C(2)

TMP-SMX

1

metronidazole 1 PO Eciprofloxacin 1 PO Ccloxacillin 1 PO Ccefazolin

1

metronidazole 1 IV ESecond line cloxacillin 1 PO C

Blood/bacteremia First line ceftriazone 1 IM Epiperacillin

1

gentamycin

1

cloxacillin

1 IV E

ciprofloxacin 1 PO EParotid First line clindamycin 1 IV EMouth floor First line metronidazole

1

clindamycin

Abbreviations: TMP/SMX

5

trimethoprim-sulfamethoxazole; PO

5

oral; IM

5

intramuscular; IV

5

intravenous.

Table 4

Urinary Tract Organisms and Antibiotic Sensitivities

% Sensitivity

Organisms

n

(%) AMP CFZ NTF TMP/SMX NOR CIP

E. coli

14 (38.9) 86 93 00 93 100 —

a

Enterococcus

sp. 7 (19.4) 86 — 100 — 86 —

K. pneumoniae

4 (11.1) 33 100 100 75 100 —Coagulase-negative staphylococci 3 (8.3) — — — — — —

P. mirabilis

3 (8.3) 100 100 0 100 100 100

P. aeruginosa

2 (5.6) — — — — — 100

E. agglomerans

1 (2.8) 0 100 100 100 100 —

Serratia marcescens

1 (2.8) 0 — 0 100 100 —

S. aureus

1 (2.8) — — 100 100 100 —Total 36 (100)

Abbreviations: AMP

5

ampicillin; CFZ

5

cefazolin; NTF

5

nitrofurantoin; TMP/SMX

5

trimethoprim-sulfamethox-azole; NOR

5

norfloxacin; CIP

5

ciprofloxacin.

a

—

5

not available

378 Pereira et al. Vol. 16 No. 6 December 1998

various foreign bodies in the care of these pa-tients.

14–17

Almost 60% of infections that were presentat the time of admission had not been recog-nized prior to the admission. This may reflect areduced intensity of surveillance for infectionsin these patients. It may also reflect the uniquechallenges to diagnosing infections in patientswith advanced cancer. Common indicators ofinfection such as a leucocytosis or fever are notalways present.

18–20

Drugs such as corticoster-oids, nonsteroidal anti-inflammatory drugs,and acetaminophen are frequently used inthese patients, and may potentially blunt thefebrile response.

21

Corticosteroids may pro-duce a leucocytosis, complicating the labora-tory diagnosis of an infection.

21

Cognitive im-pairment, which occurs relatively frequently inthese patients, may prevent patients from re-porting the symptoms of infections.

22

A fever,when present, may not necessarily represent aninfective process but could rather be neo-plasm-induced.

23

The two most commonly infected sites werethe urinary tract and the respiratory tract—39.2% and 36.5%, respectively. The high fre-quency of urinary tract infections is similar todata from acute and long-term care settingsthat consistently indicate the urinary tract tobe the most often affected, responsible for 35to 45% of all hospital-acquired infections.

17,24–

26

Pneumonias account for approximately 20%of all nosocomial infections

24,25,27

and surgicalsite infections account for 24%.

24,28

Only inten-sive care units and units caring for patientswith advanced hematological malignancieshave reported the lower respiratory tract to bethe most commonly infected site, accountingfor about 37 to 54% of infections.

10,13

Lagmanet al.

8

found that 31% and 22% of their pallia-tive care patients had septicemia and pneumo-nia, respectively, whereas Kuehn et al. reportedpneumonia in 16% of their patients, bactere-mia in 2%, and skin/soft tissue infections in upto 22%.

9

The relatively low frequency of bacteremia/septicemia in this study needs to be inter-preted with caution. Septicemia can be notori-ously difficult to diagnose clinically, particu-larly in the context of multiple concurrentorgan involvement by malignancy. Therefore,bacteremias may often go unnoticed.

Surgical site infections are not frequent inpatients admitted to PCUs since in most cases

the patients are not in a peri- or immediatepostoperative period. However, skin and sub-cutaneous site infections still account for12.5% to 22% of infections in a PCU.

8,9

Careshould be taken when inserting and maintain-ing a subcutaneous site. The use of subcutane-ous sites in this study reflects the frequent needfor alternative routes to administer drugs, suchas opioids and anti-emetics, in terminally illcancer patients. Hypodermoclysis, if appropri-ate, may also be administered via this route.

29,30

Overall, the types and patterns of bacterialpathogens appear to be similar to those re-ported in most studies of hospital-associated in-fections.

10,24

Although this study focused onbacterial infections, viruses and fungi can alsobe the causative agents. In reports from oncol-ogy centers, bacterial organisms were isolatedin over 75% of nosocomial infections and fun-gal pathogens were identified in approximately3–10% of the infections.

10,31

Viruses were iden-tified in only 2% and unusual pathogens, suchas

Pneumocystis carinii

, accounted for the re-mainder. It must be noted that these studies in-cluded patients with hematological malignan-cies and patients at various cancer stages.

The trends of pathogen distribution by sitein this study also appear to be very similar tothose reported in other hospital-based stud-ies.

10,24

The most common causative organismsfor urinary tract infections are the enterobacte-riaceae and enterococci.

16

Studies of catheter-associated hospital infections have reportedthat, with long periods of catheterization,

E.coli

are replaced by more resistant enterococciand other gram-negative bacilli, such as

Provi-dencia

and

Pseudomonas

species.

17

The organ-isms most commonly reported in this study tobe associated with pneumonia appear to besimilar to those reported in hospital-basedstudies (aerobic gram-negative bacilli and

S.aureus

).

32

It must be noted that the pathogenprofile was obtained by sputum cultures. Thisrelatively noninvasive method is not ideal andthe pathogen pattern may reflect colonizationrather than actual infection.

33

Generally, theGroup A streptococci and

S. aureus

appear tobe the two organisms most involved in skin andsubcutaneous infections.

34

In necrotic areas suchas those caused by invasive tumors, the patho-gens tend to be polymicrobial and are oftenanaerobic as a result of poor local blood supply.

The small number of culture-positive bacter-emias in this study does not allow for detailed

Vol. 16 No. 6 December 1998 Frequency of Infections in Palliative Care 379

analysis. The literature indicates that the pat-terns of pathogen involvement between hema-tological and solid tumor malignancies are dif-ferent.

10,35–39

Although

E. coli and S. aureus arecommon bacteremic organisms to both groups,patients with hematological infections tend tohave more P. aeruginosa, K. pneumoniae, coagu-lase-negative staphylococci and S. pneumoniaeinfections than do patients with solid tumors.36–38

More recently, yeast infections have beenfound to cause septicemia and the syndrome isoften indistinguishable from gram-negativesepticemia.38

Unfortunately, the retrospective nature ofthis study did not allow for adequate analysis ofsymptom response to antibiotic therapy. This isan issue that needs to be addressed in prospec-tive studies using stricter diagnostic criteria.The effectiveness of antibiotics in treating se-vere symptoms in carefully selected patientshas been documented by various authors.Bruera reported seven patients with large ul-cerated head and neck tumors who developeda sudden increase in pain intensity due to sec-ondary infections of the ulcers.40 In all pa-tients, antibiotic treatment produced a markedand rapid improvement in pain control. Greenet al. described the effective control of symp-toms related to infections in two cancer pa-tients who were treated with antibiotics.41 Onepatient had severe respiratory symptoms as aresult of pneumonia and a second had sepsis-induced delirium. MacKey and MacDonald de-scribed a patient in which a large psoas abscessresulted in severe abdominal pain.42 Antibiotictherapy and surgical drainage resulted in goodsymptom control.

The decision whether to treat or not to treatan infection with antibiotics in a patient with aterminal illness may be difficult at times. Thereexists some reluctance to prescribe antibioticsin the palliative care setting because it may beconsidered to be life-sustaining treatment.43

The decision-making process is made morecomplex by the knowledge that infection is acommon cause of death in cancer patients andit is often perceived to be an “old-man’sfriend.”41 On the other hand, treating an infec-tion may be the most appropriate means ofsymptom control when life prolongation is notthe goal. The underlying diagnosis, the stageof the illness, the level of multisystem deterio-ration, uncontrolled pain and other symptoms,and the patient and family’s own wishes all

have to be considered in the decision-makingprocess.41

When selecting an antibiotic regimen, fac-tors such as potential adverse effects, the needfor laboratory monitoring, and comfort andconvenience to the patient become rele-vant.44,45 Compromises that try to manage thesymptoms while addressing specific patientneeds may need to be made and regimensneed to be individualized. Parenteral antibioticregimens may be appropriate in some cases, es-pecially where good quality of life is still ex-pected for a significant period of time.

It is not always possible to predict whetherantibiotic therapy will result in a favorable effecton symptom control. Under these circumstances,it may be reasonable to initiate a therapeutictrial of antimicrobial therapy. If deteriorationoccurs despite antibiotic therapy, a decision todiscontinue the antibiotics would be reason-able. Future research should attempt to fur-ther clarify patient subgroups that would bene-fit optimally from treatment of their infections.

Local surveillance of pathogenic patternsand susceptibility to antibiotics is useful in de-termining infectious trends. Reports of antibi-otic-resistant bacteria in hospitals have ap-peared with increasing frequency.46 Bacterialresistance often goes unnoticed until it is epi-demic.

It has been reported that 30% of all re-ported nosocomial infections are prevent-able.24 One study found that 36% of all cathe-ter days were unnecessary and that promptcatheter removal would have theoretically pre-vented 40% or more of all infections.47 Avoid-ance of urinary catheters is not always possiblein terminally ill patients. Several authors havereported a high incidence of urinary retentionand incontinence in these patients.48 Profoundasthenia may prevent patients from going tothe bathroom or using urinals.48

An increased risk for infection also is recog-nized as a possible drawback of corticosteroidtherapy.49 However, the benefits to be gainedby using corticosteroids in selected advancedcancer patients are numerous.50–53 In this studythere did not appear to be a significant differ-ence in the number of infected patients versusnoninfected patients receiving corticosteroids.On the other hand, clinical prudence suggeststhat the presence of an infection should beviewed as a relative contraindication to initiat-ing or maintaining corticosteroid treatment.

380 Pereira et al. Vol. 16 No. 6 December 1998

ConclusionThis study indicates a high frequency of

symptomatic infections in advanced cancer pa-tients. Respiratory and urinary tract infectionsare common and a wide spectrum of microbi-als are responsible for these infections. Unlikethe acute surgical and medical settings wheresymptomatic infections are, with few excep-tions, always treated, the decision whether totreat or not to treat in the palliative care settingcan be complex and needs to be individual-ized. The goal of antibiotic therapy in palliativecare is symptom control, whereas the goal inthe acute medical and surgical setting is the de-crease of mortality and morbidity. Diagnosticinterventions need to be carefully selected soas to avoid unacceptable discomfort to the pa-tient. When the decision to initiate antibiotictreatment is made, it is helpful to be aware ofpathogen patterns and antibiotic susceptibilityto assist with antibiotic selection. The appropri-ate management of infections in patients withadvanced cancer should be a research priority.

References1. Chanock S. Evolving risk factors for infectious

complications of cancer therapy. Hematol/OncolClin North Am 1993;7(4):771–793.

2. Inagaki J, Rodriguez V, Bodey GP. Causes ofdeath in cancer patients. Cancer 1974;33(2):568–573.

3. Klastersky J, Daneau D, Verhest A. Causes ofdeath in patients with cancer. Europ J Cancer 1972;8:149–154.

4. Rossini F. Prognosis of infections in elderly pa-tients with haematological diseases. Support CareCancer 1996;4:46–50.

5. Rossi C, Klastersky J. Initial empirical antibiotictherapy for neutropenic fever: analysis of the causesof death. Support Care Cancer 1996;4:207–212.

6. Giamarellou H. Empiric therapy for infectionsin the febrile, neutropenic, compromised host. In:Cullen JH, Jacobs CC, eds. The medical clinics ofNorth America. Philadelphia: W.B. Saunders, 1995;79(3):559–580.

7. Klastersky J. Therapy of infections in cancer pa-tients. In: Klastersky J, Schimpff SC, Senn HJ, eds.Handbook of supportive care in cancer. New York:Marcel Dekker, 1995;1:1–44.

8. Lagman R, Panta R, Walsh D. Infectious Compli-cations in Advanced Cancer. 8th International Sym-posium, Abstracts. Support Care Cancer 1996;4:226.

9. Kuehn N, MacEachern T, Miller M, SpachynskiK. Nosocomial infections (NI) on a palliative care

unit (PCU): a prospective comparison study. Posterfor 9th International Congress on Care of the Ter-minally Ill, Montreal Canada, 1992; Oct. 31–Nov. 2.

10. Rotstein C, Cummings KM, Nicolaou AL, et al.Nosocomial infection rates at an oncology center.Infect Control Hosp Epidemiol 1988;9(1)13–19.

11. Ayliffe GA, Casewell MW, Cookson BD, et al. Na-tional prevalence survey of hospital acquired infec-tions: definitions. J Hospital Infect 1993;24:69–76.

12. Garner JS, et al. CDC definitions for nosocomialinfections. Am J Infect Control 1988;16:128–140.

13. Hughes WT, Flynn PM, Williams BG. Nosoco-mial infection in patients with neoplastic diseases.In: Mayhall CG, ed. Hospital epidemiology and in-fection control. Baltimore: Williams and Wilkins,1996;43:618–631.

14. Curnette JT, Boxer LA. Clinically significantphagocytic cell defects. In: Remington J, Swartz M,eds. Current clinical topics in infectious diseases.New York: McGraw Hill, 1985;6:103–156.

15. Keusch GT. Nutrition and infection. In: Rem-ington JS, Swartz NM, eds. Current clinical topics ininfectious disease. New York: McGraw Hill, 1984;5:106–123.

16. Glauser MP, Zinner SH. Mechanisms of acquisi-tion and development of bacterial infections in can-cer patients. In: Klastersky J, ed. Infections in cancerpatients. New York: Raven Press, 1982:13–24.

17. Warren JW. Nosocomial urinary tract infections.In: Mandell GL, Bennett JE, Dolin R, eds. Principlesand practice of infectious diseases, 4th ed. NewYork: Churchill Livingstone, 1995;283:2607–2616.

18. Shapiro M, Greenfield S. The complete bloodcount and leucocyte differential count. Ann InternMed 1987;106:65–74.

19. Wasserman M, Levinstein M, Keller E, et al. Util-ity of fever, white blood cells, and differential countin predicting bacterial infections in the elderly. J AmGeriatr Soc 1989;37:537–543.

20. Hodkinson HM. Common symptoms of diseasein the elderly, 2nd ed. Oxford: Blackwell, 1989:6–20.

21. Haynes RC. Adrenocorticotropic hormone;adrenocortical steroids and their synthetic analogs;inhibitors of the synthesis and actions of adrenocor-tical hormones. In: Goodman L, Gilman A, eds. Thepharmacological basis of therapeutics. New York:Pergamon Press, 1990:1431–1462.

22. Pereira J, Hanson J, Bruera E. The frequencyand clinical course of cognitive impairment in pa-tients with terminal cancer. Cancer 1997;79:835–842.

23. Johnson M. Neoplastic fever. Palliative Med1996;10:217–224.

24. Martone WJ, Jarvis WR, Culver DH, Haley RW.Incidence and nature of endemic and epidemicnosocomial infections. In: Bennett JV, BrachmanPS, Sanford JP, eds. Hospital infections, 3rd ed. Bos-ton: Little, Brown, 1992;27:577–596.

Vol. 16 No. 6 December 1998 Frequency of Infections in Palliative Care 381

25. Magaziner J, Tenney JH, DeForge B, et al. Preva-lence and characteristics of nursing home–acquiredinfections in the aged. JAGS 1991;39:1071–1078.

26. Smith MA, Duke WM. A retrospective review ofnosocomial infections in an acute rehabilitative andchronic population at a large skilled nursing facility.JAGS 1994;42:45–49.

27. LaForce FM. Lower respiratory tract infections.In: Bennett JV, Brachman PS, Sanford JP, eds. Hos-pital infections, 3rd ed. Boston: Little, Brown, 1992;29:611–639.

28. Wong ES. Surgical site infections. In: MayhallCG, ed. Hospital epidemiology and infection con-trol. Baltimore: Williams & Wilkins, 1996;11:168–175.

29. Fainsinger RL, MacEachern T, Miller MJ, et al.The use of hypodermoclysis for rehydration in ter-minally ill cancer patients. J Pain Symptom Manage1994;9(5):298–302.

30. Bruera E, Legris MA, Kuehn N, Miller MJ. Hypo-dermoclysis for the administration of fluids and nar-cotic analgesics in patients with advanced cancer. JPain Symptom Manage 1990;5(4):218–220.

31. Robinson GV, Tegtmeier BR, Zaia JA. Brief re-port: nosocominal infection rates in a cancer treat-ment center. Infect Control 1984;5:289–294.

32. George DL. Nosocomial pneumonia. In: May-hall CG, ed. Hospital epidemiology and infectioncontrol. Baltimore: Williams & Wilkins, 1996;12:175–195.

33. Woods GL, Washington JA. The clinician andthe microbiology laboratory. In: Mandell GL, Ben-nett JE, Dolin R, eds. Principles and practice of in-fectious diseases, 4th ed. New York: Churchill Living-stone, 1995;11:169–199.

34. Cote L. Clinical characteristics of skin and softtissue infections. Canadian J CME 1995;37–45.

35. Morrison VA, Peterson BA, Bloomfield CD.Nosocomial septicemia in the cancer patient: the in-fluence of central venous access devices, neutrope-nia, and type of malignancy. Med Pediatr Oncol1990;18:209–216.

36. Singer C, Kaplan MH, Armstrong D. Bacteremiaand fungemia complicating neoplastic disease: astudy of 364 cases. Am J Med 1977;62:731–742.

37. Schimpff SC. Gram-negative bacteremia. Sup-port Care Cancer 1993;1:5–18.

38. Pittet D, Wenzel RP. Nosocomial bloodstreaminfections. Arch Intern Med 1995;155:1177–1184.

39. Mayo JW, Wenzel R. Rates of hospital-acquiredbloodstream infections in patients with specific ma-lignancy. Cancer 1982;50:187–190.

40. Bruera E. Intractable pain in patients with ad-

vanced head and neck tumors: a possible role of lo-cal infection. Cancer Treat Rep 1986;70(5):691–692.

41. Green K, Webster H, Watanabe S, Fainsinger R.Case report: management of nosocomial respiratorytract infections in terminally ill cancer patients. JPalliative Care 1994;10(4):31–34.

42. MacKey J, Birchall I, MacDonald N. Occult in-fection as a cause of hip pain in a patient with meta-static breast cancer. J Pain Symptom Manage 1995;10(7):569–572.

43. Council on Ethical and Judicial Affairs, Ameri-can Medical Association. Decisions near the end oflife. JAMA 1992;267(16):2229–2233.

44. Craig WA, Andes DR. Parenteral versus oral an-tibiotic therapy. In: Cunha BA, ed. The medical clin-ics of North America. Philadelphia: W.B. Saunders,1995;79(3):497–508.

45. Ma MY, Rho JP. Considerations in antimicrobialprescribing. In: Cunha BA, ed. The medical clinicsof North America. Philadelphia: W.B. Saunders,1995;79(3):537–549.

46. French GL, Phillips I. Antimicrobial resistancein hospital flora and nosocomial infections. In: May-hall CG, ed. Hospital epidemiology and infectioncontrol. Baltimore: Williams & Wilkins, 1996;74:980–999.

47. Hartstein AI, Garber SB, Ward TT, Jones SR,Morthland VH. Nosocomial urinary tract infection:a prospective evaluation of the 108 catheterized pa-tients. Infect Control 1981;2:380–386.

48. Fainsinger RL, MacEachern T, Hanson J, BrueraE. The use of urinary catheters in terminally ill can-cer patients. J Pain Symptom Manage 1992;7(6):333–338.

49. Stuck AE, Minder CE, Frey FJ. Risk of infectiouscomplications in patients taking glucocorticoster-oids. Rev Infect Dis II 1989;(6):954–963.

50. Watanabe S, Bruera E. Corticosteroids as adju-vant analgesics. J Pain Symptom Manage 1994;9(7):442–445.

51. Chye R, Lickiss N. Palliative care rounds: the useof corticosteroids in the management of bilateralmalignant ureteric obstruction. J Pain SymptomManage 1994;9(8):537–540.

52. Fainsinger RL, Spachynski K, Hanson J, BrueraE. Symptom control in terminally ill patients withmalignant bowel obstruction. J Pain Symptom Man-age 1994;9(1):12–18.

53. Bruera E, Seifert L, Watanabe S, Babul N, DarkeA, Harsanyi Z, Suarez-Almazor M. Chronic nausea inadvanced cancer patients: a retrospective assessmentof a metoclopramide-based antiemetic regimen. JPain Symptom Manage 1996;11(3):147–153.