Original Research

PNEUMONIA

Resource Utilization of Adults Admitted

to a Large Urban Hospital With
Community-Acquired Pneumonia
Caused by Streptococcus pneumoniae*
Heather K. Sun, PharmD; David P. Nicolau, PharmD, FCCP; and
Joseph L. Kuti, PharmD

Objective: To determine if penicillin-nonsusceptible Streptococcus pneumoniae, among other

variables, was significantly associated with greater hospital costs among patients with community-
acquired pneumonia (CAP).
Design: Retrospective, cohort study.
Setting: Eight hundred ten-bed, urban, private, teaching hospital.
Patients: Adult patients admitted between 1999 and 2003 with CAP caused by S pneumoniae.
Intervention: Clinical criteria and costs (inflated to 2004 dollars) were collected from the medical
charts and detailed hospital bills for each individual patient. Costs were compared according to
classification by penicillin susceptibility. Multivariate linear regression was utilized to determine
variables independently associated with increased hospital costs and length of stay.
Results: Of 168 patients included, 44 patients (26%) had CAP caused by penicillin-nonsusceptible S
pneumoniae. Median total hospital costs were $8,654 (25th to 75th percentile, $5,457 to $16,027), with
no difference between susceptible and nonsusceptible groups. Bed costs accounted for 55.6% of total
costs, followed by laboratory (9.9%) and pharmacy (9.8%) costs. Regression analyses determined that
ICU admission (p < 0.001), unexplained delays in discharge (p 0.001), and neoplasm (p < 0.04)
were independently predictive of both total hospital costs (adjusted r2 0.46) and increasing length
of stay (adjusted r2 0.30). Hospital mortality, bacteremia, and congestive heart failure were also
associated with at least one of the dependent variables.
Conclusion: In the current era in which more potent antibiotics are empirically utilized to treat
CAP, it does not appear that a simple classification of penicillin nonsusceptibility complicates the
economic impact of S pneumoniae infection. Focused efforts to reduce length of stay, including
minimizing prolonged and unnecessary observation of patients, should have the most profound
effect on reducing total costs. (CHEST 2006; 130:807 814)

Key words: costs; pneumococcal; pneumonia; resistance; Streptococcus pneumoniae

Abbreviations: CAP community-acquired pneumonia; CHF congestive heart failure; MIC minimum inhibitory
concentration; PSI pneumonia severity illness

P neumonia is the sixth-leading cause of death in

the United States and the number-one cause of
death from an infectious disease.1,2 The subset of
*From the Center for Anti-Infective Research and Development,
Hartford Hospital, Hartford, CT.
This study was funded through a competitive Hartford Hospital
Research Endowment grant.
This paper was presented, in part, at the American College of
Clinical Pharmacy 2005 Annual Meeting, San Francisco, CA,
October 2005.
There are no proprietary data presented in this article, and the
authors have no conflicts of interest to disclose.
patients who acquire pneumonia while not residing
in a hospital or long-term care facility for at least 2
weeks before the onset of symptoms are defined as
having community-acquired pneumonia (CAP).1 Of
the 4 to 5 million cases of CAP treated annually, a
Reproduction of this article is prohibited without written permission
from the American College of Chest Physicians (www.chestjournal.
org/misc/reprints.shtml).
Correspondence to: Joseph L. Kuti, PharmD, Center for Anti-
Infective Research and Development, Hartford Hospital, 80
Seymour St, Hartford, CT 06102; e-mail: jkuti@harthosp.org
DOI: 10.1378/chest.130.3.807

www.chestjournal.org CHEST / 130 / 3 / SEPTEMBER, 2006 807

minority (approximately 0.5 to 1 million) are treated
in a hospital.3 Yet, this minority has accounted for
the majority of health-care costs associated with the
infection; $8 billion was spent on hospitalization
costs in 1994, compared with $0.4 billion spent
among outpatients.4
Although most studies1,2 have been unable to
identify the cause of CAP in approximately 50% of
patients, the most common bacteria isolated in most
reports have been Streptococcus pneumoniae. In a
large metaanalysis5 of 7,000 CAP cases, S pneu-
moniae accounted for two thirds of those for whom
an etiologic diagnosis was made, as well as two thirds
of the patients with bacteremic pneumonia. De-
creasing susceptibility to penicillin is the most com-
mon epidemiologic marker for tracking resistance
among S pneumoniae. A recent surveillance study6
conducted during the 2002 to 2003 flu season deter-
mined the overall rate of penicillin resistance (ie,
nonsusceptibility) was 34.2% at 45 centers through-
out the United States; moreover, resistance had not
changed significantly since 1998. Similar resistance
rates were also apparent for second-generation ceph-
alosporins, macrolides, and trimethoprim-sulfame-
thoxazole. Many of these therapies remain first-line
treatment options for patients with CAP; however, to
date studies711 have shown mixed results regarding
the effect of penicillin resistance on patient out-
comes.
Additionally, antimicrobial resistance has become
an increasing health economic concern. Infections
caused by numerous species of resistant Gram-
positive and Gram-negative bacteria have been asso-
ciated with increased health-care costs.1216 While
newer therapies for drug-resistant bacteria are rou-
tinely more expensive than the older standard of
care antibiotics, the majority of attributable costs
associated with resistance result from a delayed
clinical response due to inappropriate therapy, an
increased severity of illness among those acquiring a
resistant organism, or the perception that patients
require extended treatment or observation.17 All can
result in increased hospital length of stay, thereby
significantly escalating costs. However, few cost-of-
illness studies18,19 have attempted to evaluate the
economic impact of penicillin susceptibility in pa-
tients with CAP caused by S pneumoniae. The
primary objective of this study was to determine
resource utilization and costs associated with the
in-hospital treatment of CAP caused by S pneu-
moniae, as stratified by susceptibility to penicillin. A
secondary objective was to determine other patient
covariates that were predictive of a more costly
hospital stay.
808
Materials and Methods
Study Design
This was a retrospective cohort study of adult patients admitted
to Hartford Hospital between 1999 and 2003 with documented
CAP caused by S pneumoniae. Hartford Hospital is an 810-adult-
bed, nonprofit, private, teaching hospital located in urban Hart-
ford, CT. The institutional review board approved this study.
Informed consent was not required since the study was retro-
spective in design and all protected health information was
destroyed on completing data collection. Each patient received a
subject number corresponding to a consecutively reviewed med-
ical record and hospital bill.
Patients
All patients admitted to the hospital from January 1999 to
January 2003 with respiratory or blood culture specimens testing
positive for S pneumoniae were identified through the microbi-
ology laboratory computer database. Examination of the medical
records of these subjects was conducted to identify the cohort of
patients who had the diagnosis of CAP caused by S pneumoniae.
Patients were included if they were 18 years old; had at least
one sputum culture (with presence of 10 squamous epithelial
cells and 25 polymorphonuclear cells per 100 magnification
field; or be obtained from semiquantitative culture) or two blood
cultures positive for S pneumoniae; and had signs and symptoms
consistent with the diagnosis of CAP including the presence of a
new infiltrate on chest radiograph and at least two of the
following within 1 day of the first positive culture: documentation
of fever ( 38.0C) or hypothermia ( 35.0C); WBC count
10,000/L or 15% bands or leukopenia (WBC 4,500/L);
documentation of auscultory findings on pulmonary examination
and/or evidence of pulmonary consolidation; documentation of
new cough with or without sputum production; documentation of
new-onset dyspnea or tachypnea; or hypoxemia with a Po2 60
mm Hg on room air. Patients were excluded if their total
hospitalization was 2 days, if they had impaired immune
function (ie, AIDS, HIV, leukocyte count 1,000/L), known or
suspected tuberculosis, known or suspected Pneumocystis ji-
roveci, or concomitant pneumonia or other infection at baseline
caused by viruses, fungi, or other bacteria except intracellular
pathogens (Mycoplasma pneumoniae, Chlamydophila pneu-
moniae, Legionella pneumoniae), Haemophilus influenzae, or
Moraxella catarrhalis.
Data Collection
Medical charts were reviewed for all identified patients, and
data were collected using a standardized collection tool. Infor-
mation documented included patient demographics; comorbid
conditions (diabetes, COPD, neoplasm, congestive heart failure
[CHF], cirrhosis, or other chronic liver disease); microbiology
data including penicillin susceptibility and culture source; admis-
sion year; hospital length of stay; presence of ICU stay; pneumo-
nia severity illness (PSI) score20; antibiotic therapy (including
route); daily maximum temperature and WBC count; and crude
mortality at end of hospitalization. Penicillin susceptibility was
defined according to current Clinical and Laboratory Standards
Institute guidelines21 as susceptible (penicillin minimum inhibi-
tory concentration [MIC] 0.06 g/mL), intermediate resistant
(MIC, 0.12 to 1.0 g/mL), or resistant (MIC 2.0 g/mL).
Comorbid conditions were positive if they were present at
baseline except for neoplasm, which was any cancer except basal
or squamous cell cancer of the skin that was active at time of
presentation or diagnosed within 1 year of presentation.20
Original Research
 All resource utilization and economic data were derived from
the patients detailed medical bill. Each bill was separated by the
following pertinent charge departments: hospital bed charges
(further divided by non-ICU bed or ICU bed), pharmacy,
antibiotic specific charges, laboratory, radiology, respiratory care,
rehabilitation, and other. At Hartford Hospital, the accounting
system considers fixed indirect and direct costs (ie, nursing and
physician services, housekeeping, electricity, administration)
within the hospital bed charge, and the quantity of services
utilized will be reflected in the bed charge (eg, ICU bed charge
greater than non-ICU bed charge because of greater intensity of
services). All charges were converted to costs using department
specific cost-to-charge ratios and inflated to 2004 using the
medical component of the consumer price index for all consum-
ers in US cities (www.bls.gov/data/home.htm). Total costs
equaled the sum of all department costs. Costs associated with
specific services known not to be associated with the patients
admission for CAP (eg, open-heart surgery) were excluded.
Statistics
Economic analyses were conducted from the hospital perspec-
tive. The primary analysis separated the cohort into two groups
based on penicillin susceptibility (susceptible vs nonsusceptible).
For all comparisons, the nonsusceptible group included both
intermediate-resistant and resistant S pneumoniae. Continuous
data were compared using a Student t test for normally distrib-
uted data or Mann-Whitney U test for nonnormally distributed
data (eg, costs and length of stay). 2 or Fisher exact test were
used to compare proportions between the two groups. For cost
comparison between susceptible, intermediate, and resistant
groups, data were analyzed by the Kruskal-Wallis rank-sum test.
Multivariate linear regression was utilized to control for con-
founding variables and determine covariates that predicted total
costs and length of stay for the entire cohort. For all regression
analyses, total costs and length of stay were log-transformed so
that parametric tests could be utilized. In both regression models,
all variables (patient demographics, comorbidities, S pneumoniae
susceptibility, ICU stay, admission year, presence of bacteremia,
PSI score, mortality, and receipt of oral antibiotic therapy) were
inserted into the model at once to control for covariance. An
additional variable, termed unexplained delayed discharge, was
added into the models to improve predictability. Knowing that at
our hospital the ultimate decision regarding patient discharge is
made by the attending physician, this variable was designed to
capture prolonged observation of the patient. A patient was
defined as having a delayed discharge if they remained in the
hospital for 48 h after they clinically met criteria for discharge.
Criteria for discharge included stable normalization of tempera-
ture and WBC, and the receipt of oral antibiotics or the ability to
receive oral antibiotic therapy, which ever came later. All patients
who died in the hospital were automatically excluded from being
defined as a delayed discharge.
A p value 0.05 was considered significant during all statistical
analysis. All statistical analysis was conducted using statistical
software (SPSS/PC; SPSS; Chicago, IL).
Results
Of the 543 patients identified with S pneumoniae-
positive blood and/or respiratory culture findings,
168 patients met inclusion/exclusion criteria. Pa-
tients were excluded from the analysis for the fol-
lowing reasons: 203 patients had HIV infection, 72
www.chestjournal.org
patients did not meet clinical criteria or did not have
a positive chest radiographic finding, 65 patients
were discharged from the emergency department or
admitted to the hospital for 2 days, and 35 patients
were infected with another bacteria at baseline other
than those allowed in the inclusion/exclusion criteria.
Overall, patients were elderly (mean age, 63 years)
and were equally divided among male and female
gender (Table 1). Twenty-seven percent of patients
were directly admitted to the ICU, and 42% had
bacteremic pneumonia. The high severity of illness
of this population was also apparent in the PSI score
with 33% of patients in class IV and 23% in class V.
Penicillin susceptibility was 74%; 11% were interme-
diate resistant, and 16% displayed high-level resis-
tance. Crude mortality was 13% but was 21% among
subjects within PSI classes IV and V. Ninety-five
percent of patients who died were in PSI classes IV
and V; one patient in PSI class III also died. There
were no significant differences in patient demo-
graphics or clinical characteristics when the cohort
was divided by penicillin susceptibility except for
bacteremic pneumonia, which was more common
among patients infected with penicillin susceptible
isolates (48% vs 25%, p 0.012). Lastly, penicillin
susceptibility did not affect the median hospital
length of stay for patients, which was 6 days in both
groups (p 0.725), nor did it influence admission to
an ICU (p 0.497). However, patients admitted to
the ICU did have a significantly longer median
length of stay (11.0 days [25th to 75th percentile, 6.8
to 19.8 days] vs 5.0 days [25th to 75th percentile, 4.0
to 8.0 days]; p 0.001) compared with those treated
outside the ICU.
Table 2 displays the list of antibiotic regimens
utilized to empirically treat CAP in these patients for
at least the first 24 h. The majority of patients
received a -lactam as monotherapy (32.7%), or in
combination with a macrolide (32.7%), or mono-
therapy with a fluoroquinolone (9.5%). The remain-
ing 25% of patients received alternative regimens,
but in all but 4.8% of cases, a -lactam, macrolide, or
fluoroquinolone was utilized as at least one of the
drugs in the regimen. Only one patient received IV
penicillin G, and the majority of -lactam use was
cefuroxime, ceftriaxone, or cefepime. Eighty-three
patients (49.4%) received an oral antibiotic during
their hospital stay for the treatment of CAP, with all
but two being transitioned from an IV regimen. A
greater percentage of patients in the penicillin-
susceptible group were transitioned from IV to oral
compared with the nonsusceptible group, but the
difference did not reach statistical significance
(53.2% vs 38.6%, p 0.137). For those patients who
received oral therapy during hospitalization, the
CHEST / 130 / 3 / SEPTEMBER, 2006 809
 Table 1Patient Demographic and Clinical Characteristics for All Patients With CAP Caused by S pneumoniae
and Then Comparable Results for Groups Defined by Penicillin Susceptibility*

Patients With Penicillin- Patients with Penicillin-

All Patients Susceptible S pneumoniae Nonsusceptible S pneumoniae
Characteristics (n 168) (n 124) (n 44) p Value

Age, yr 63 19 63 18 63 19 0.983
Male gender 88 (52) 64 (52) 24 (55) 0.874
Admission to ICU 45 (27) 31 (25) 14 (32) 0.497
Bacteremic pneumonia 71 (42) 60 (48) 11 (25) 0.012
PSI score
I 13 (8) 11 (9) 2 (5) 0.677
II 23 (14) 17 (14) 6 (14)
III 38 (23) 26 (21) 12 (27)
IV 55 (33) 43 (35) 12 (27)
V 39 (23) 27 (22) 12 (27)
Median (25th to 75th percentile) 6 (410) 6 (410.5) 6 (49.5) 0.725
hospital length of stay, d
Penicillin susceptibility
Susceptible 124 (74) 124 (100) ND
Intermediate 18 (11) 18 (41)
Resistant 26 (16) 26 (59)
Comorbid conditions
Diabetes 39 (23) 27 (22) 12 (27) 0.593
COPD 46 (27) 37 (30) 9 (20) 0.316
CHF 26 (15) 15 (12) 11 (25) 0.073
Neoplasm 21 (13) 17 (14) 4 (9) 0.596
Liver disease 14 (8) 11 (9) 3 (7) 0.916
Admission year
1999 28 (17) 17 (14) 11 (25) 0.441
2000 47 (28) 35 (28) 12 (27)
2001 37 (22) 27 (22) 10 (23)
2002 51 (30) 41 (33) 10 (23)
2003 5 (3) 4 (3) 1 (2)
Crude mortality 21 (13) 17 (14) 4 (9) 0.596
*Data are presented as mean SD or No. (%) unless otherwise noted.
Comparison between penicillin-susceptible and penicillin-nonsusceptible groups. Penicillin susceptible penicillin MIC 0.06 g/mL;
penicillin nonsusceptible penicillin MIC 0.12 g/mL; ND not done.

median day to transition was day 5 (25th to 75th Total and departmental costs are depicted in
percentile, day 4 to day 7). Penicillin susceptibility Table 3. There was no significant difference on any
had no effect on the day to oral transition cost level between patients infected with penicillin
(p 0.718). susceptible vs nonsusceptible S pneumoniae. The
greatest percentage of total cost resources utilized
were allocated to hospital bed costs (including both
Table 2Empiric Antibiotic Regimens Utilized for
> 24 h non-ICU and ICU beds) at 55.6%, followed by
laboratory (9.9%) and pharmacy (9.8%) costs (Fig 1).
Antibiotic Regimens No. (%) Antibiotics accounted for 48% of total pharmacy
-Lactam monotherapy* 55 (32.7) costs. Median costs were also not significantly differ-
-Lactam plus macrolide 55 (32.7) ent when separated into penicillin-susceptible, pen-
Fluoroquinolone monotherapy 16 (9.5)
icillin-intermediate, or penicillin-resistant S pneu-
-Lactam plus vancomycin 8 (4.8)
-Lactam plus macrolide plus fluoroquinolone 7 (4.2) moniae groups: $8,503 (25th to 75th percentile,
-Lactam plus clindamycin 6 (3.6) $5,425 to $15,862) vs $6,347 (25th to 75th percen-
Macrolide monotherapy 5 (3.0) tile, $4,889 to $13,099) vs $10,809 (25th to 75th
-Lactam plus macrolide plus vancomycin 3 (1.8)
percentile, $6,118 to $22,209), respectively
-Lactam plus fluoroquinolone 3 (1.8)
Macrolide plus fluoroquinolone 2 (1.1) (p 0.477).
Other 8 (4.8) Independent variables included in the final multi-
*One patient received IV penicillin G for treatment of pneumonia.
variate models predicting total hospital costs and
All other -lactams included ampicillin-sulbactam, piperacillin/ lengths of stay are shown in Table 4. Only baseline/
tazobactam, cefuroxime, cefotaxime, ceftriaxone, or cefepime. history of neoplasm, ICU admission, and delayed

810 Original Research

 Table 3Median Costs (2004 US$) by Cost Center for All Patients With CAP Caused by S pneumoniae and Then
Comparable Results for Groups Defined by Penicillin Susceptibility*

Patients With Penicillin- Patients With Penicillin-

All Patients Susceptible S pneumoniae Nonsusceptible S pneumoniae
Variables (n 168) (n 124) (n 44) p Value

Total costs 8,654 (5,45716,027) 8,503 (5,42515,862) 9,441 (5,53116,822) 0.617

Costs by cost center
Non-ICU bed 3,841 (2,3796,124) 3,827 (2,3206,124) 3,874 (2,7645,835) 0.740
ICU bed 0 (01,636) 0 (0629) 0 (03,282) 0.460
Pharmacy 733 (4621,379) 744 (4741,301) 718 (4591,695) 0.853
Antibiotics 368 (259642) 365 (256606) 372 (273696) 0.877
Laboratory 767 (4981,588) 821 (4881,660) 626 (5071,482) 0.537
Radiology 264 (115999) 256 (137964) 307 (861,605) 0.986
Respiratory care 315 (55915) 293 (72826) 336 (381,108) 0.748
Rehabilitation 0 (0144) 0 (0153) 0 (0167) 0.817
Other 1,033 (6012,682) 1,029 (6012,272) 1,229 (6623,096) 0.350
*All costs are presented as median (25th to 75th percentile).
Comparison between penicillin-susceptible and penicillin-nonsusceptible groups.

discharge were significantly associated with both
higher total costs (adjusted r2 0.456) and increas-
ing length of stay (adjusted r2 0.300). Additionally,
hospital mortality (p 0.039) was also associated
with higher total costs, and baseline/history of CHF
(p 0.048) and presence of bacteremia (p 0.026)
also significantly predicted increasing length of stay.
In both models, admission to the ICU had the
greatest positive -coefficient, suggesting it was
highly predictive of the dependent variables. Peni-
cillin susceptibility was not significantly associated
with total costs (p 0.349) or length of stay
(p 0.409).
Eighty-eight patients (52%) in our cohort met the
definition of unexplained delayed discharge. Exclud-
ing patients who died in the hospital, the median
length of stay was significantly longer in patients with
unexplained delayed discharge: 7 days (25th to 75th
percentile, 4 to 11 days) vs 5 days (25th to 75th
percentile, 3.25 to 7 days) [p 0.005]. There was no
Figure 1. Percentage of total costs (2004 dollars) accounted for
by each cost center. White area represents ICU bed costs, and
gray area represents antibiotic costs.
apparent difference in age, comorbidities, or PSI
score among patients meeting the definition of un-
explained delayed discharge and those who did not.
Additionally, penicillin susceptibility had no effect
on patients with unexplained delayed discharge
(p 0.847). The addition of delayed discharge im-
proved the coefficient of determination (r2) signifi-
cantly in both models.
Discussion
Health-care costs attributed to managing CAP in
the United States are now estimated to be $10
billion annually, with the majority of these costs
credited to treatment in the hospital.3,22 Therefore,
studies aimed at identifying key factors associated
with increased health-care costs are essential to
improving cost-effectiveness and quality of care. In
the current study, we assessed the factors predictive
of increasing hospital costs and length of stay for
patients admitted with pneumococcal pneumonia to
a large, urban, private, teaching hospital over a
5-year period. We observed that hospital bed costs
accounted for 50% of the total resources utilized
and that admission to the ICU, neoplasm, and an
unexplained delayed discharge were independently
associated with total hospital costs and extended
length of stay. Additionally, end of hospital mortality
was a significant predictor of costs, and the presence
of CHF and bacteremia were positive predictors of
length of stay.
However, the primary objective of this study was
to determine if resistance to penicillin among S
pneumoniae had an adverse effect on hospital costs
or length of stay. Total median hospital cost for
patients infected with penicillin susceptible isolates

www.chestjournal.org CHEST / 130 / 3 / SEPTEMBER, 2006 811

 Table 4 Final Multiple Linear Regression Models Predicting Total Hospital Costs and Length of Stay

Total Hospital Cost (Adjusted r2 0.46) Length of Stay (Adjusted r2 0.30)

Independent Variables Coefficient SE p Value Coefficient SE p Value

Age 0.001 0.002 0.502 0.001 0.001 0.804

Male gender 0.050 0.047 0.284 0.076 0.04 0.079
Diabetes 0.048 0.054 0.382 0.035 0.050 0.486
CHF 0.124 0.066 0.061 0.121 0.061 0.048*
COPD 0.066 0.054 0.226 0.061 0.050 0.223
Neoplasm 0.151 0.072 0.039* 0.166 0.067 0.014*
Admission year 0.003 0.021 0.885 0.001 0.020 0.953
Bacteremia 0.082 0.048 0.091 0.100 0.044 0.026*
Delayed discharge 0.172 0.052 0.001* 0.165 0.048 0.001*
ICU Admission 0.512 0.055 0.001* 0.318 0.051 0.001*
Mortality 0.189 0.090 0.039* 0.072 0.083 0.389
Penicillin susceptibility 0.051 0.054 0.349 0.041 0.050 0.409
PSI score 0.050 0.059 0.398 0.058 0.055 0.292
Oral transition 0.028 0.053 0.595 0.076 0.049 0.126
*Indicates significance.

was $8,503, compared with $9,441 for those infected
with nonsusceptible isolates (p 0.617). As a result,
penicillin susceptibility was not a significant variable
in any of the multivariate regression models predict-
ing hospital costs or length of stay. The conclusions
remained similar even when the definition of peni-
cillin nonsusceptibility was changed to fully resistant
(MIC 2 g/mL) vs susceptible/intermediate (data
not shown), or when the total costs were compared
for susceptible, intermediate, and resistant indepen-
dently ($8,503 vs $6,347 vs $10,809, respectively;
p 0.477).
These results conflict with other studies1216,18,19
that have identified increased costs attributed to
infection with numerous resistant organisms, includ-
ing S pneumoniae. In particular, our results conflict
with a retrospective, cohort study of similar size but
different time period by Klepser and colleagues19;
these investigators evaluated health-care resource
utilization for the treatment of penicillin-susceptible
and penicillin-nonsusceptible isolates of S pneu-
moniae and found that total hospitalization costs
were significantly greater for patients infected with
penicillin nonsusceptible isolates ($10,309.25 vs
$7,801.54, p 0.0006). The primary reason for the
higher cost was a greater length of stay among the
nonsusceptible group (14 days vs 10 days, p 0.05).
Length of stay in general was much lower for
patients in our study (median, 5 days) and not
different between susceptible and nonsusceptible
groups. It is most likely that the advances in under-
standing the time frame and clinical characteristics
of the stable CAP patient in the current era (2000 to
present) have dramatically reduced the duration of
hospital stay compared with when the study by
Klepser et al19 was conducted (1995 to 1998), there-
fore making it more difficult to find a difference if
one were to exist.
Another potential reason for the discordance be-
tween our results and others is due to the fact that
we defined groups according to penicillin suscepti-
bility, yet only one patient actually received penicil-
lin in our study. Instead, most were treated with
second-generation or third-generation cephalospo-
rins in combination with a macrolide or with fluoro-
quinolone monotherapy. Later-generation cephalo-
sporins and fluoroquinolones are known to be more
effective against S pneumoniae harboring intermedi-
ate-level resistance to penicillin, as well as isolates
that are fully resistant with an MIC of 2 g/mL.1,2,23
-Lactam efficacy is only questioned when the pen-
icillin MIC is 4 g/mL.23 In the study by Klepser
et al,19 patients also received a variety of different
antibiotic cocktails; however, it was not possible to
determine which -lactams were used, or if any were
penicillin. We had susceptibility data reported solely
for penicillin, and only in isolated cases did the
microbiology laboratory have data for other antibiot-
ics. This made it difficult to associate antibiotic
resistance specific to the antibiotic received with that
of hospital cost or length of stay. It also made it
impossible for us to evaluate the appropriateness of
empiric antibiotic therapy based on in vitro suscep-
tibilities and its effect on the dependent variables.
Therefore, from these data we cannot make a final
conclusion regarding an association between antibi-
otic resistance among S pneumoniae and increased
costs, except that penicillin resistance matters little
in light of few patients still receiving this agent
empirically. These data do, however, suggest that
differentiation among S pneumoniae using solely
penicillin susceptibility may not be sufficient.
The present results do confirm findings from other
studies19,24 27 demonstrating that hospital bed costs
are the major contributor to total costs for CAP. In
the study by Klepser et al,19 room costs accounted

812 Original Research

for 34% of total hospital costs and nursing costs
accounted for 37%. Although we were not able to
separate nursing costs from that of room costs in our
study, our hospital bed costs accounted for both
direct and indirect fixed costs. When added together,
the percentage of resources dedicated to room and
nursing was similar with our reported hospital bed
cost percentage of 55.6%. Likewise, in a separate
study to assess resource utilization in the treatment
of CAP, Orrick and colleagues27 found median costs
to be $2,430 for their population, with hospital room
costs constituting 83.7% of total costs, followed by
antibiotic (4.6%), radiology (2.6%), and respiratory
care (0.9%) costs. It is not surprising then that
admission to an ICU in our study was significantly
associated with an increased cost and length of stay,
as daily bed cost and total length of stay (11 days vs
5 days, p 0.001) were both greater for patients in
the ICU compared with a normal hospital bed.
Importantly, these data taken together confirm that
antibiotic costs account for only a small percentage
( 5%) of total costs in the treatment of hospitalized
patients with CAP; moreover, any efforts to reduce
total or ICU length of stay, such as IV-to-oral
transition or clinical pathways, will have the most
profound effect on reducing the economic burden of
CAP.28,29
The only variable that was significantly different
between the susceptible and nonsusceptible groups
in this study was the presence of bacteremia. In our
analysis, 48% of patients infected with penicillin-
susceptible S pneumoniae were also bacteremic,
compared with 25% of patients with nonsusceptible
isolates (p 0.012). These results are consistent
with numerous other reports79,18,19 in which inva-
sive disease was more common among susceptible
pneumococci. This is probably related to organism
serotype, with certain serotypes commonly more
susceptible than others, but also more virulent.30 We
were unable to confirm this because we did not have
access to the S pneumoniae isolates in our study to
perform serotyping.
One interesting and perhaps still controversial
observation in our study was the significance of the
variable we defined as unexpected delayed discharge
in predicting total costs and increased length of stay.
Although difficult, it was our intent to characterize a
measure of prolonged and unnecessary observation.
We used clinical criteria (normalization of tempera-
ture and WBC) as well as transition to oral antibiotic
therapy to determine whether the physicians in our
hospital might by using unnecessarily longer obser-
vation periods and not discharging patients when
they were stable. Excluding those patients who died,
median length of stay was 2 days longer in patients
who met our definition of delayed discharge. If
www.chestjournal.org
patients were discharged within 1 day of meeting our
clinical definition and receiving oral antibiotic ther-
apy, the median length of stay for these 88 patients
would have been 3 days (range, 2 to 6.75 days), with
an estimated savings of approximately $507,000.
These potential savings are likely underestimated
given there was often a delay between IV-to-oral
transition and the first day of stable clinical criteria;
furthermore, we conservatively used the later day to
define discharge eligibility. Our study was conducted
at a single hospital, and while our patient population
is likely similar to other large urban hospitals in the
United States, one study31 has documented the
distinct differences in clinical criteria used by physi-
cians from institution to institution to determine
when patients are stable for discharge, which may
have influenced the impact of this variable. Addition-
ally, clinical outcomes have not been adversely af-
fected by decreases in length of stay (ie, discharge
once patients are clinically stable without observa-
tion), suggesting that patients with CAP admitted to
hospitals with historically long lengths of stay might
be treated just as effectively with shorter hospital
stays.32 It is important to note the limitations in
attempting to define this variable retrospectively.
Although there was no affect of age, comorbid
illness, or PSI score, prolonged lengths of stay in
these patients may be a reflection of numerous other
factors in addition to defervescence, WBC normal-
ization, and time to oral transition. Specifically, social
issues and physician preferences, which cannot be
accounted for in any severity of illness score, are
important factors and difficult to account for given
this study design. Nevertheless, given the significant
impact that this defined variable had on our model,
it will be sensible for us to further evaluate criteria
for discharge within our institution and determine if
changes in practice or discharge management can be
addressed to reduce length of stay for CAP.
In conclusion, at our large, urban hospital, we
observed that hospital bed costs accounted for
50% of the total cost of care for patients admitted
with CAP due to S pneumoniae. Admission to an
ICU, neoplasm, and an unexplained delay in dis-
charge were independently associated with both total
hospital costs and increased length of stay; therefore,
focused efforts to reduce total and ICU length of
stay, including minimizing prolonged and unneces-
sary observation of patients, should have the most
profound effect on reducing total costs. Finally, in
the current era in which more potent antibiotics are
empirically utilized to treat CAP, it does not appear
that a simple classification of penicillin nonsuscepti-
bility complicates the clinical or economic impact of
S pneumoniae infection.
CHEST / 130 / 3 / SEPTEMBER, 2006 813
ACKNOWLEDGMENT: We thank Sheryl Horowitz, PhD, for
guidance and suggestions regarding statistical analysis.
References
1 Bartlett JG, Dowell SF, Mandell LA, et al. Practice guidelines
for the management of community-acquired pneumonia in
adults. Clin Infect Dis 2000; 31:347382
2 Niederman MS, Mandell LA, Anzueto A, et al. Guidelines for
the management of adults with community-acquired pneu-
monia: diagnosis, assessment of severity, antimicrobial ther-
apy, and prevention. Am J Respir Crit Care Med 2001;
163:1730 1754
3 Colice GL, Morley MA, Asche C, et al. Treatment costs of
community-acquired pneumonia in an employed population.
Chest 2004; 125:2140 2145
4 Niederman MS, McCombs JS, Unger AN, et al. The cost of
treating community-acquired pneumonia. Clin Ther 1998;
20:820 837
5 Fine MJ, Smith MA, Carson CA, et al. Prognosis and
outcomes of patients with community-acquired pneumonia.
JAMA 1996; 275:134 141
6 Doern GV, Richter SS, Miller A, et al. Antimicrobial resis-
tance among Streptococcus pneumoniae in the United States:
have we begun to turn the corner on resistance to certain
antimicrobial classes? Clin Infect Dis 2005; 41:139 148
7 Palleres R, Linares J, Vadillo M, et al. Resistance to penicillin
and cephalosporin and mortality from severe pneumococcal
pneumonia in Barcelona, Spain. N Engl J Med 1995; 333:
474 480
8 Metlay JP, Hofmann J, Cetron MS, et al. Impact of penicillin
susceptibility on medical outcomes for adult patients with
bacteremic pneumococcal pneumonia. Clin Infect Dis 2000;
30:520 528
9 Aspa J, Rajas O, Rodriguez de Castro F, et al. Drug-resistant
pneumococcal pneumonia: clinical relevance and related
factors. Clin Infect Dis 2004; 38:787798
10 Flaco V, Almirante B, Jordano Q, et al. Influence of penicillin
resistance on outcome in adult patients with invasive pneu-
mococcal pneumonia: is penicillin useful against intermedi-
ately resistant strains? J Antimicrob Chemother 2004; 54:
481 488
11 Bonnard P, Lescure FX, Douadi Y, et al. Community-
acquired bacteraemic pneumococcal pneumonia in adults:
effect of diminished penicillin susceptibility on clinical out-
come. J Infect 2005; 51:69 76
12 Webb M, Riley LW, Roberts RB. Cost of hospitalization for
and risk factors associated with vancomycin-resistant Entero-
coccus faecium infection and colonization. Clin Infect Dis
2001; 33:445 452
13 Cosgrove SE, Kaye KS, Eliopoulous GM, et al. Health and
economic outcomes of the emergence of third-generation
cephalosporin resistance in Enterobacter species. Arch Intern
Med 2002; 162:185190
14 Engemann JJ, Carmelis Y, Cosgrove SE, et al. Adverse
clinical and economic outcomes attributable to methicillin
resistance among patients with Staphylococcus aureus surgi-
cal site infection. Clin Infect Dis 2003; 36:592598
15 Lodise TP, McKinnon PS. Clinical and economic impact of
methicillin resistance in patients with Staphylococcus aureus
bacteremia. Diagn Microbiol Infect Dis 2005; 52:113122
16 Cosgrove SE, Qi Y, Kaye KS, et al. The impact of methicillin
814
resistance in Staphylococcus aureus bacteremia on patient
outcomes: mortality, length of stay, and hospital charges.
Infect Control Hosp Epidemiol 2005; 26:166 174
17 Cosgrove SE, Carmeli Y. The impact of antimicrobial resis-
tance on health and economic outcomes. Clin Infect Dis
2003; 36:14331437
18 Einarsson S, Kristjansson M, Kristinsson KG, et al. Pneumo-
nia caused by penicillin-non-susceptible and penicillin-sus-
ceptible pneumococci in adults: a case-control study. Scand
J Infect Dis 1998; 30:253256
19 Klepser ME, Klepser DG, Ernst EJ, et al. Health care
resource utilization associated with treatment of penicillin-
susceptible and -nonsusceptible isolates of Streptococcus
pneumoniae. Pharmacother 2003; 23:349 359
20 Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to
identify low-risk patients with community-acquired pneumo-
nia. N Engl J Med 1997; 336:243250
21 Performance standards for antimicrobial susceptibility test-
ing: fifteenth informational supplement; CLSI document
M100 S15. Wayne, PA: Clinical and Laboratory Standards
Institute, 2005
22 Agency for Healthcare Research and Quality. Healthcare
costs: AHRQ publication No. 02-P033, September 2002.
Available at: www.ahrq.gov/news/costsfact.pdf. Accessed Oc-
tober 5, 2005
23 Heffelfinger JD, Dowell SF, Jorgenson JH, et al. Manage-
ment of community-acquired pneumonia in the era of pneu-
mococcal resistance: a report from the Drug-resistant Strep-
tococcus pneumoniae Therapeutic Working Group. Arch
Intern Med 2000; 160:1399 1408
24 Nicolau DP. The challenge of prescribing treatment for
respiratory tract infections. Am J Manag Care 2000; 6:S419
S426
25 Dresser LD, Niederman MS, Paladino JA. Cost-effectiveness
of gatifloxacin vs ceftriaxone with a macrolide for the treat-
ment of community-acquired pneumonia. Chest 2001; 119:
1439 1448
26 Bartoleme M, Almirall J, Morera J, et al. A population-based
study of the costs of care for community-acquired pneumo-
nia. Eur Respir J 2004; 23:610 616
27 Orrick JJ, Segal R, Johns TE, et al. Resource use and cost of
care for patients hospitalized with community acquired pneu-
monia: impact of adherence to Infectious Diseases Society of
America guidelines. Pharmacoeconomics 2004; 22:751757
28 Kuti JL, Capitano B, Nicolau DP. Cost-effective approaches
to the treatment of community acquired pneumonia in the
era of resistance. Pharmacoeconomics 2002; 20:513528
29 Fine MJ, Pratt HM, Obrosky DS, et al. Relation between
length of hospital stay and costs of care for patients with
community-acquired pneumonia. Am J Med 2000; 109:378
385
30 Sandgren A, Sjostrom K, Olsson-Liljequist B, et al. Effect of
clonal and serotype-specific properties on the invasive capac-
ity of Streptococcus pneumoniae. J Infect Dis 2004; 189:785
796
31 Fine MJ, Medsgar AR, Stone RA, et al. The hospital dis-
charge decision for patients with community-acquired pneu-
monia. Arch Intern Med 1997; 157:4756
32 McCormick D, Fine MJ, Coley CM, et al. Variation in length
of hospital stay in patients with community-acquired pneu-
monia: are shorter stays associated with worse medical out-
comes. Am J Med 1999; 107:512
Original Research