The effect of age on the development and outcome of adult sepsis*

Greg S. Martin, MD, MSc; David M. Mannino, MD; Marc Moss, MD

Objective: Sepsis is an increasingly common and lethal med-
ical condition that occurs in people of all ages. The influence of
age on sepsis risk and outcome is incompletely understood. We
sought to determine the independent effect of age on the inci-
dence, severity, and outcome of adult sepsis.
Design: Longitudinal observational study using national hos-
pital discharge data.
Setting: Approximately 500 geographically separated nonfed-
eral acute care hospitals in the United States.
Patients: Patients were 10,422,301 adult sepsis patients hos-
pitalized over 24 yrs, from 1979 to 2002.
Interventions: None.
Measurements and Main Results: Incident sepsis cases were
age adjusted and characterized by demographics, sources and
types of infection, comorbid medical conditions, and hospital
discharge status. Elderly patients (>65 yrs of age) accounted for
12% of the U.S. population and 64.9% of sepsis cases, yielding a
relative risk of 13.1 compared with younger patients (95% confi-
dence interval, 12.6 13.6). Elderly patients were more likely to
have Gram-negative infections, particularly in association with
pneumonia (relative risk, 1.66; 95% confidence interval, 1.63
1.69) and to have comorbid medical conditions (relative risk, 1.99;
95% confidence interval, 1.922.06). Case-fatality rates increased
linearly by age; age was an independent predictor of mortality in
an adjusted multivariable regression (odds ratio, 2.26; 95% con-
fidence interval, 2.172.36). Elderly sepsis patients died earlier
during hospitalization, and elderly survivors were more likely to
be discharged to a nonacute health care facility.
Conclusions: The incidence of sepsis is disproportionately
increased in elderly adults, and age is an independent predictor of
mortality. Compared with younger sepsis patients, elderly non-
survivors of sepsis die earlier during hospitalization and elderly
survivors more frequently require skilled nursing or rehabilitative
care after hospitalization. These findings have implications for
patient care and health care resource prioritization and provide
insights for expanded scientific investigations and potential pa-
tient interventions. (Crit Care Med 2006; 34:1521)
KEY WORDS: critical illness; factual database; intensive care;
epidemiology; epidemiologic methods; intensive care; outcome
assessment; sepsis; sepsis syndrome; septicemia; United States

T he U.S. population continues
to grow, increasing 13.2%
from 1990 to 2000. Twelve
percent or 35 million individ-
uals in the United States are over age 65,
and the proportion of people over age 90
increased by 42% in the past decade (1).
Health care resource consumption for pa-
tients over age 65 is increasing, recently
estimated at $387 billion per year, and
will continue to increase due to greater
life expectancy and approach of the baby
boomers to this elderly distinction.
General utilization of expensive intensive
care unit (ICU) resources continues to
*See also p. 234.
From the Division of Pulmonary, Allergy and Crit-
ical Care, Department of Medicine, Emory University
(GSM, MM); and the Division of Pulmonary and Critical
Care, University of Kentucky (DMM).
Supported, in part, by grants HL K23-67739 (GSM)
and AA R01-11660 (MM) from the National Institutes of
Health.
The authors have no financial relationship to dis-
close relative to this work.
Copyright 2005 by the Society of Critical Care
Medicine and Lippincott Williams & Wilkins
DOI: 10.1097/01.CCM.0000194535.82812.BA
increase and accounts for as much as
30% of all hospital costs (2, 3). Intensive
care utilization increases with age, and
half of all ICU days are currently used by
patients older than 65 yrs of age (4). El-
derly patients will use more ICU re-
sources in the future, as the overall pop-
ulation is projected to increase 50% by
the year 2050 while the population over
age 65 increases by 115% (5).
Sepsis, a life-threatening inflamma-
tory disorder representing the immune
response to an infection, is a common
disorder affecting nearly 700,000 people
annually in the United States (6 8). The
more seriously ill sepsis patients, defined
by the presence of acute organ system
dysfunction, are termed severe sepsis,
which may occur in one quarter of all ICU
admissions and account for up to half of
ICU bed-days (9 11). Unfortunately, sep-
sis remains a lethal disease with case-
fatality rates of 20 40% and contributes
to nearly 20% of all in-hospital deaths
(8). It is the leading cause of death in
noncoronary ICUs and the tenth leading
cause of death overall in the United States
(12, 13). As the majority of severe sepsis
patients require ICU care, sepsis patients
contribute substantially to health care
costs, estimated at $17 billion annually in
the United States (14). Of concern, the
frequency of sepsis is increasing by 5%
per year, in excess of the growth and
aging of our population (8).
Sepsis appears to be a disease of the
elderly. A composite profile of severe sep-
sis patients from 1995 hospital discharge
records in seven U.S. states reported a
mean age of 63.8 yrs with an incidence of
severe sepsis that increased with age (14).
Similarly, in Europe, the median age of
ICU patients who met criteria for severe
sepsis was 65 yrs of age (10, 15). In addi-
tion, the average age of sepsis patients in
the United States is increasing over time,
exceeding 65 yrs in the year 2000 (8).
Case-fatality rates have been associated
with age in one study, increasing from
10% in children to 40% in patients 85
yrs of age (14).
Further characterization of the epide-
miology of sepsis in older patients has
been targeted as a critical focus for both
patient care and research (16). An inter-
national collaboration has been estab-

Crit Care Med 2006 Vol. 34, No. 1 15

Table 1. Comorbid conditions among sepsis patients, stratified by age
65 Yrs of Age
(n 3,654,421)
Cancer
Chronic renal failure
Congestive heart failure
COPD
Coronary artery disease
Diabetes
Hepatic cirrhosis
HIV
Hypertension
COPD, chronic obstructive pulmonary disease; HIV, human immunodeficiency virus. Values are
percentages. Differences are statistically significant between age groups for each condition at the p
.01 level.
lished to improve care and reduce mor-
tality with sepsis, requiring better
understanding of afflicted patients (17).
Equally important, health care providers
need to understand the effect of age on
the risk and outcome with sepsis to guide
treatment decisions and prognostic dis-
cussions. In this article, we used a vali-
dated method to define the influence of
age on the risk of sepsis and relevant
clinical outcomes, with particular atten-
tion to the elderly population as defined
by age 65 yrs. We used nationally col-
lected hospitalization data from 1979
through 2002 to examine temporal
changes in the incidence and outcome of
sepsis, differences in the distribution of
responsible organisms, and chronic med-
ical conditions across age groups.
METHODS
Data Sources. The National Center for
Health Statistics has conducted the National
Hospital Discharge Survey (NHDS) continu-
ously since 1965 (18). Since 1979, the NHDS
has conformed to the guidelines of the Uni-
form Hospital Discharge Data Set (UHDDS)
for consistency of record reporting. The NHDS
is composed of a sample of all nonfederal acute
care hospitals in the United States, applied
across approximately 500 hospitals with equal
geographic representation. The NHDS con-
tains inpatient discharge records from each
participating hospital, representing approxi-
mately 350,000 discharges annually in the
United States, or 1% of all hospitalizations.
The following information is abstracted from
each patient discharge record: patient demo-
graphics (age, gender, ethnic background,
geographic locale, and marital status), seven
diagnostic codes and four procedural codes
(Clinical Modification of the International
Classification of Diseases, Ninth Revision;
ICD-9-CM), dates of hospital admission and
discharge, sources of payment, and patient
discharge disposition. Information regarding
16
65 Yrs of Age
(n 6,767,880)
19.2 16.0
13.1 12.1
8.1 23.7
7.6 13.1
6.3 18.1
17.7 19.1
5.8 2.0
5.5 0.02
14.0 18.0
the total U.S. population and racial/ethnic,
gender, and age subgroups was taken from the
U.S. Census Bureau 1980 to 2002 data files (5).
Comparative census information for 1979 was
linearly extrapolated from the subsequent 23
yrs. This project was considered exempt from
the requirement for informed consent accord-
ing to federal regulations of human subjects
protection 45 CFR 46.101(b).
Definitions. Cases were identified from dis-
charge records in the NHDS of patients 18
yrs of age that included any discharge diagno-
sis of sepsis. Sepsis was defined by the follow-
ing ICD-9-CM codes: 038 (septicemia), 020.0
(septicemic), 790.7 (bacteremia), 117.9 (fun-
gemia), 112.5 (disseminated Candida infec-
tion), and 112.81 (disseminated fungal endo-
carditis), as previously published (8). Organ
failure was defined by a combination of ICD-
9-CM and CPT codes, consistent with those
previously used (8, 14). Chronic comorbid
medical conditions were assessed using previ-
ously established methods that are consistent
with accepted comorbidity indexes (e.g.,
Charlson-Deyo score) (8, 19 21). Postsepsis
nonacute care hospitalizations were defined to
include any nonacute care medical facility,
whether it be a short or long-term rehabilita-
tion facility, a skilled nursing facility, or a
chronic care facility.
Data Presentation and Statistical Analysis.
Frequency counts of sepsis cases were divided
into age deciles, from 18 29 yrs to 90 99 yrs,
or dichotomized at age 65 yrs or 65 yrs.
Incidence rates were calculated from the year-
specific census data and normalized to the
2000 U.S. census population distribution. Es-
timates are presented according to accepted
guidelines for accuracy of NHDS data, requir-
ing absolute unweighted samples sizes 60
with relative standard error (RSE) measures
30% to be included for data analysis. RSE
was calculated according to NHDS designs us-
ing SUDAAN software (22), as outlined in the
RSE tables as part of the NHDS documenta-
tion (18). Standard error was calculated by
multiplying the RSE by the estimate, and 95%
confidence intervals (CI) were constructed
from these standard error measures. Contin-
uous data were compared by analysis of vari-
ance, and categorical data were compared us-
ing the chi-square test. Predictors of clinical
outcomes with sepsis were assessed by multi-
variate logistic regression modeling including
potentially relevant predictors (age, gender,
race, source of infection [categorized as respi-
ratory; genitourinary; gastrointestinal; skin,
soft tissue, and bone; and other], chronic co-
morbid medical conditions [categorized as in
Table 1], and severity of illness by cumulative
organ dysfunction) using a backward elimina-
tion technique described by Kleinbaum (23).
Regression models reported adjusted odds ra-
tios (OR) with Wald confidence intervals.
When race was missing from the sample for a
given year (120% in any given year), these
persons were excluded from the race-specific
rate calculations but were included in all other
rate calculations. Analyses were performed us-
ing SAS System version 9.1 for Windows (SAS
Institute, Cary, NC) (24). All p values are two-
sided with a threshold of .05 used to assign
significance.
RESULTS
Demographics. The number of hospi-
talizations during the period remained
static, totaling 717 million, of which
1.6% involved sepsis. Of these 10,422,301
sepsis cases, the mean age increased over
time from 64.1 yrs to 68.2 yrs (median
age increased from 67 to 71 yrs, both p
.001). Fifty-three percent of patients were
female, and the racial distribution was
72.6% Caucasian, 14.6% African Ameri-
can, and 12.8% other races. Sepsis was
accompanied by acute organ dysfunction
(i.e., severe sepsis) in 3,216,996 patients
(annual average of 30.9%) during the
24-yr study period.
Sepsis Incidence. The cumulative
24-yr age-specific incidence of sepsis in-
creased exponentially across all age de-
ciles, from 29.6 cases per 100,000 indi-
viduals in the 18 29 age decile to 2,422.3
cases per 100,000 in the 90 99 age decile
(Fig. 1). When stratified at age 65, the
relative risk for sepsis was 13.1 times
higher for those age 65 or above (95% CI,
12.6 13.6). During the 24-yr study pe-
riod, patients 65 yrs of age accounted
for 37.3% of hospitalizations and 64.9%
of sepsis cases, with sepsis occurring in
2.5% of these hospitalizations. These pa-
tients contributed disproportionately to
the longitudinal increases in the inci-
dence of sepsis, with incidence rates in-
creasing 20.4% faster than rates in the
younger cohort (mean increase 11.5% vs.
9.5% per year, p .001, Fig. 2). There
were no significant differences in the oc-
Crit Care Med 2006 Vol. 34, No. 1

Figure 1. Incidence rate (filled circles, left abscissa) and case-fatality rates (open boxes, right abscissa)
for sepsis, adjusted and stratified by age deciles over the 24-yr study period. Points are mean values;
I-bars represent the SEM.
Figure 2. Age-adjusted incidence rates of sepsis among hospitalized patients, stratified by age 65
(dashed line) or 65 (solid line). Points are mean values; I-bars represent the SEM.
Figure 3. Distribution of sources of infection among sepsis patients, stratified by age. GI, gastroin-
testinal infections; GU, genitourinary infections.
currence or types of acute organ dysfunc-
tion based upon age.
Organisms and Source of Infection.
Forty-six percent of infections were with-
Crit Care Med 2006 Vol. 34, No. 1
differences in the types or sources of in-
fections according to age.
out an identified organism. In those cases
of sepsis with a classified organism,
51.0% of infections were from Gram-
positive bacteria, compared with 43.4% of
Gram-negative bacteria, 1.9% anaerobes,
and 3.7% fungi. Elderly patients were
1.31 times more likely to have Gram-
negative infections (95% CI, 1.271.35),
whereas patients under age 65 were 1.19
times more likely have Gram-positive in-
fections (95% CI, 1.151.23) and 1.78
times more likely to have fungal infec-
tions (95% CI, 1.76 1.80). Respiratory
infections accounted for the most sepsis
cases during the study period at 34.2%,
whereas genitourinary infections ac-
counted for 25.5%; gastrointestinal infec-
tions for 12.4%; skin, soft tissue, and
bone for 5.5%; and other sources (cardio-
vascular, central nervous system, blood-
stream, unclassified) for 22.7%. For those
patients 65 yrs of age, respiratory infec-
tions (relative risk [RR], 1.29; 95% CI,
1.251.33) and genitourinary infections
(RR, 1.38; 95% CI, 1.321.44) were more
common as the cause of sepsis, whereas
gastrointestinal infections were less com-
mon (RR, 0.72; 95% CI, 0.68 0.76) com-
pared with the younger patient cohort
(Fig. 3). Pneumonia was the single most
common cause of sepsis, and in this sub-
group Gram-negative infections ac-
counted for 34.1% of cases for those age
65, compared with 20.5% in those pa-
tients 65 yrs of age (RR, 1.66; 95% CI,
1.631.69). There were disproportionate
longitudinal increases in Gram-positive
infections during the study period, but no
Comorbid Medical Conditions.
Chronic comorbid medical conditions
were present in 71.7% of sepsis patients
during the 24-yr study period. Sepsis pa-
tients over age 65 were twice as likely to
have at least one comorbid medical con-
dition (RR, 1.99; 95% CI, 1.922.06) and
more likely to have a Charlson-Deyo
comorbidity index 0 (67.5% vs. 55.2%,
p .001) compared with younger sepsis
patients. Although most comorbid condi-
tions were overrepresented in the elderly
sepsis population, this was not true for
conditions that alter immunity (HIV,
cancer, chronic renal failure, and diabe-
tes, Table 1). Sepsis patients under age 65
carried a coincident diagnosis of HIV in
5.5% of cases, compared with only 0.02%
of cases among older patients. Elderly
sepsis patients were three times more
likely to have a coincident diagnosis of
coronary artery disease (RR, 2.93; 95%
CI, 2.873.00) or congestive heart failure
(RR, 3.00; 95% CI, 2.94 3.06).
Clinical Outcomes and Hospital Dis-
position. Overall, the case fatality rate for
17

Figure 4. Fatality rates of sepsis among hospitalized patients longitudinally from 1979 to 2002 (top)
and by Kaplan-Meier style hospital survival plot (bottom), stratified by age 65 (dashed line) or 65
(solid line). Points are mean values; I-bars represent the SEM.
the sepsis cohort was 24.4% and it de-
clined more rapidly over time in the el-
derly cohort (analysis of variance p
.001, Fig. 4, top). Case fatality increased
linearly across age deciles (Fig. 1) and
averaged 27.7% for those over age 65 vs.
17.7% for those 65 yrs of age (p
.001). Thus, the risk of dying was 1.56
times greater for those over age 65 (95%
CI, 1.521.61). In a multivariate logistic
regression model adjusting for race, gen-
der, severity of illness, source of infec-
tion, and chronic comorbid medical con-
ditions, age 65 yrs was independently
associated with a 2.3 times higher risk of
18
death among patients with sepsis (ad-
justed OR, 2.26; 95% CI 2.172.36, Table
2).
The mean length of stay for sepsis
patients decreased over time to 15.9 days
and was shorter for those over the age of
65 (15.5 days vs. 16.7 days, p .001).
Among nonsurvivors of sepsis, elderly pa-
tients were 26% more likely to die during
the first week of hospitalization (51.3%
vs. 40.6%, p .001). Age-stratified differ-
ences in survival were apparent within 24
hrs but approached equal rates of decline
after 14 days (Fig. 4, bottom). Among
survivors of sepsis, elderly patients were
less likely to return home (54% vs. 76%,
p .001), and the use of nonacute health
care facilities after hospital discharge was
greater in the elderly population (37% vs.
15%, p .001).
DISCUSSION
In this study, it is apparent that sepsis
is a disease defined by age, with a mean
age that exceeds the common definition
for elderly in this country. Despite the
fact that people over age 65 account for
only one eighth of the U.S. population,
they account for two thirds of all sepsis
cases. Longitudinal increases in the inci-
dence of sepsis are weighted toward the
elderly population, where incidence rates
are rising the fastest. Furthermore, age is
an independent predictor of death with
sepsis, and elderly patients die earlier
during sepsis-related hospitalizations
whereas elderly survivors more fre-
quently require skilled nursing or reha-
bilitative care after hospitalization.
The reason that age is strongly asso-
ciated with both risk and outcome with
sepsis is likely multifactorial. Incidence
rates for sepsis in humans are known to
increase with age (14), and susceptibility
to sepsis has been documented as a func-
tion of age in animal models (25). This
may reflect age-related differences in im-
mune function, ranging from failed anti-
gen processing by leukocytes (26) to al-
tered inflammatory cytokine expression
(2730). The consistent finding of age as
a predictor of outcomes in related criti-
cally ill populations (3137) may be uni-
fied by an impaired immune response in
the elderly (25, 27, 29, 30). Aside from
alterations in immunity, the type of in-
fection and its propensity for causing sep-
sis may be influenced by age, as we found
pneumonia to be more common in the
elderly population and pneumonia more
frequently leads to a septic physiologic
response (6). Although comorbidities are
highly relevant for predicting outcomes
in critically ill patients with sepsis (38
41), it is unlikely that they fully account
for the differences in susceptibility to sep-
sis given the similar distributions of co-
morbidities that alter immunity. How-
ever, age-related conditions such as
dementia or immobility may contribute
and may be amenable to health care in-
terventions. The finding of age an inde-
pendent predictor of death in sepsis, al-
though in part a tautology of human
lifespan, may relate to key differences in
access to health care, delivery of health
Crit Care Med 2006 Vol. 34, No. 1
 tives. This will create greater inpatient dysfunction) but could relate to limita-

T
he incidence of
sepsis is dispropor-
tionately in-
creased in elderly adults,
and age is an independent
predictor of mortality.
care, and patient care preferences, which
require further investigation to under-
stand.
Our findings have broad implications
for both the care of sepsis patients and
the provision of health care. Interven-
tional sepsis trials invariably exclude pa-
tients at high risk of death, often defined
in terms of age and associated comorbid
medical conditions. Ironically, these are
the individuals most likely to develop sep-
sis and thus should be the targets of such
controlled investigations. The limited
available evidence regarding sepsis ther-
apies in the elderly suggests that these
therapies are similarly effective at im-
proving survival (42, 43). Increasing life
expectancy and the shift of baby boomers
toward the elderly age group (currently
40 58 yrs of age), combined with greater
rates of sepsis in the elderly population,
may increase the need for intensive care
resources, depending on patient prefer-
ences and utilization of advance direc-
Table 2. Multivariate logistic regression model of the risk of death among patients with sepsis
Variable Coefficient ()
hospital expenditures for sepsis, costing
$20,000 60,000 per case (14, 44), and
disproportionate allocation of resources
for sepsis care based on shifting popula-
tion demographics, with the elderly pop-
ulation increasing by 44% during the
study period compared with the younger
population increasing by only 26%. The
inpatient burden of health care resource
consumption does not include sepsis pa-
tients treated outside the hospital, the
greater utilization of posthospitalization
medical care for the elderly cohort, or
necessary home care for all patients after
sepsis hospitalization. Taken together,
these factors portend an increasing chal-
lenge for the health care system in the
United States but do not support health
care rationing based on chronological
age.
These data advance our understanding
of the factors that influence the risk and
outcome with sepsis and may help to
optimize care for these patients. The in-
clusion of age in a decision-analysis
model may assist in the management of
sepsis patients by considering the fre-
quency of organisms or sources of infec-
tion as a function of age. Age may also
lead to additional diagnostic consider-
ations, such as HIV, which is relatively
prevalent in younger sepsis patients and
an increasingly common reason for ICU
admission (45). It remains unknown why
elderly sepsis patients died earlier during
hospitalization despite apparently similar
severity of illness (by quantitative organ
p Value OR 95% CI
tions on care imposed by either the pa-
tient or the physician, among other fac-
tors (46 48). Examining provider
delivery of care and patient preferences
for care across age groups is equally im-
portant to optimizing patient-centered
outcomes (47 49).
The utilization of administrative data-
sets for critical care health services re-
search is recognized as an essential tool
yet has limitations that are partially offset
by the large sample size of patient-level
data (11, 50). The utilization of ICD-9-CM
codes to identify specific medical condi-
tions has limitations and precludes differ-
entiation of community-acquired sepsis
from nosocomial sepsis, for example. Our
validation steps demonstrate the 038
code to carry a positive predictive value of
88.9% for identifying true cases of sepsis
(51), whereas other validation studies re-
port the 038 code to have a sensitivity of
87.7% (52). Incorporating a conservative
estimate of sepsis prevalence at 2%, the
specificity and negative predictive value
may be calculated at 98.8% and 98.6%,
respectively (8). In addition, ICD-9 cod-
ing schema may change over time, thus
influencing incidence rates, although
temporal changes would be unlikely to
vary by age. Discharge disposition may
vary depending on the patients premor-
bid locale, with increasing use of chronic
long-term care facilities potentially in-
creasing the rate of apparent long-term
care after sepsis. Finally, we cannot de-
termine mortality specifically attribut-
able to sepsis, differentiate preconceived
physician prejudices relating to age that
may affect age-related outcomes, or nec-
essarily extrapolate these findings to pa-
tients outside the United States.

Age, 65 yrs 0.816 .001 2.26 2.172.36 CONCLUSIONS

Gender, Male 0.090 .001 1.09 1.051.14
Source of infectiona We have shown that age is a critical
Pulmonary 0.460 .001 1.58 1.521.65 factor in determining both the risk for
Genitourinary 0.626 .001 0.53 0.510.56
sepsis and subsequent vital outcomes.
Skin/soft tissue 0.484 .001 0.62 0.560.68
Central nervous system 0.480 .001 1.62 1.341.94 These data provide important informa-
Comorbid conditionsa tion on infectious causes and types of
Cirrhosis 0.512 .001 1.67 1.521.83 sepsis in older patients as well as a clin-
HIV 0.601 .001 1.83 1.592.09 ical correlate to previous hypotheses re-
Cancer 0.657 .001 1.93 1.842.02
Diabetes 0.142 .001 0.87 0.830.91 garding immune system senescence. Fu-
Severity of illness ture investigations should identify the
0 dysfunctional organs Referent 1.00 reasons why sepsis is more common in
1 dysfunctional organ 0.956 .001 2.60 2.542.67 the elderly, why it is growing at a more
2 dysfunctional organs 1.912 .001 6.77 6.447.11
rapid rate, and why elderly patients die
3 dysfunctional organs 2.868 .001 17.61 16.3518.96
earlier than younger sepsis patients. Fi-
OR, odds ratio; CI, confidence interval; HIV, human immunodeficiency virus. nally, utilization of robust datasets or
a
For variables categorized underneath an asterisk, the referent group is composed of all sepsis large prospective cohorts may elucidate
cases not possessing the condition of interest (e.g., pulmonary vs. nonpulmonary sources of infection). the complex interactions between source

Crit Care Med 2006 Vol. 34, No. 1 19

or type of infection, chronic comorbid
diseases, and age, thus allowing for the
creation of an accurate risk prediction
model for the development of sepsis and
important clinical outcomes.
REFERENCES
1. Meyer J. Age 2000: Census 2000 Brief. Avail-
able at: http://www.census.gov/population/
www/cen2000/briefs.html. Accessed June 1,
2005
2. Joint position statement: Essential provi-
sions for critical care in health system re-
form. Society of Critical Care Medicine.
American Association of Critical Care
Nurses. Crit Care Med 1994; 22:20172019
3. Halpern NA, Pastores SM, Greenstein RJ:
Critical care medicine in the United States
19852000: An analysis of bed numbers, use,
and costs. Crit Care Med 2004; 32:
1254 1259
4. Angus DC, Kelley MA, Schmitz RJ, et al:
Current and projected workforce require-
ments for care of the critically ill and pa-
tients with pulmonary disease: Can we meet
the requirements of an aging population?
JAMA 2000; 284:27622770
5. U.S. Interim Projections by Age, Sex, Race
and Hispanic Origin. Available at: http://
www.census.gov/ipc/www/usinterimproj. Ac-
cessed June 1, 2005
6. Wheeler AP, Bernard GR: Treating patients
with severe sepsis. N Engl J Med 1999; 340:
207214
7. Cross AS, Opal SM: A new paradigm for the
treatment of sepsis: Is it time to consider
combination therapy? Ann Intern Med 2003;
138:502505
8. Martin GS, Mannino DM, Eaton S, et al: The
epidemiology of sepsis in the United States
from 1979 through 2000. N Engl J Med 2003;
348:1546 1554
9. Brun-Buisson C, Doyon F, Carlet J, et al:
Incidence, risk factors, and outcome of se-
vere sepsis and septic shock in adults. A mul-
ticenter prospective study in intensive care
units. JAMA 1995; 274:968 974
10. Padkin A, Goldfrad C, Brady AR, et al: Epide-
miology of severe sepsis occurring in the first
24 hrs in intensive care units in England,
Wales, and Northern Ireland. Crit Care Med
2003; 31:23322338
11. Moss M, Martin GS: A global perspective on
the epidemiology of sepsis. Intensive Care
Med 2004; 30:527529
12. Parrillo JE, Parker MM, Natanson C, et al:
Septic shock in humans. Advances in the
understanding of pathogenesis, cardiovascu-
lar dysfunction, and therapy. Ann Intern Med
1990; 113:227242
13. Arias E, Anderson RN, Kung HC, et al:
Deaths: Final data for 2001. Natl Vital Stat
Rep 2003; 52:1115
14. Angus DC, Linde-Zwirble WT, Lidicker J, et
al: Epidemiology of severe sepsis in the
United States: Analysis of incidence, out-
20
come, and associated costs of care. Crit Care
Med 2001; 29:13031310
15. Brun-Buisson C, Meshaka P, Pinton P, et al:
EPISEPSIS: A reappraisal of the epidemiol-
ogy and outcome of severe sepsis in French
intensive care units. Intensive Care Med
2004; 30:580 588
16. High KP: Why should the infectious diseases
community focus on aging and care of the
older adult? Clin Infect Dis 2003; 37:196 200
17. Dellinger RP, Carlet JM, Masur H, et al: Sur-
viving sepsis campaign guidelines for man-
agement of severe sepsis and septic shock.
Crit Care Med 2004; 32:858 873
18. National Hospital Discharge Survey descrip-
tion. Available at: http://www.cdc.gov/nchs/
about/major/hdasd/nhdsdes.htm. Accessed
June 1, 2005
19. Charlson ME, Pompei P, Ales KL, et al: A new
method of classifying prognostic comorbidity
in longitudinal studies: Development and
validation. J Chronic Dis 1987; 40:373383
20. Deyo RA, Cherkin DC, Ciol MA: Adapting a
clinical comorbidity index for use with ICD-
9-CM administrative databases. J Clin Epide-
miol 1992; 45:613 619
21. Charlson M, Szatrowski TP, Peterson J, Gold
J: Validation of a combined comorbidity in-
dex. J Clin Epidemiol 1994; 47:12451251
22. Bieler GS, Williams RL: Analyzing survey
data using SUDAAN Release 7.5. Research
Triangle Park, NC, Research Triangle Insti-
tute, 1997
23. Kleinbaum DG: Logistic Regression: A Self-
Learning Text. New York, Springer, 1996
24. SAS/STAT User Guide. Users Guide for SAS/
STAT. Second Edition. Cary, NC, SAS Insti-
tute, 1990
25. Turnbull IR, Wlzorek JJ, Osborne D, et al:
Effects of age on mortality and antibiotic
efficacy in cecal ligation and puncture.
Shock 2003; 19:310 313
26. Pawelec G, Solana R, Remarque E, et al:
Impact of aging on innate immunity. J Leu-
koc Biol 1998; 64:703712
27. Tateda K, Matsumoto T, Miyazaki S, et al:
Lipopolysaccharide-induced lethality and cy-
tokine production in aged mice. Infect Im-
mun 1996; 64:769 774
28. Bruunsgaard H, Skinhoj P, Qvist J, et al:
Elderly humans show prolonged in vivo
inflammatory activity during pneumococ-
cal infections. J Infect Dis 1999; 180:
551554
29. Gabriel P, Cakman I, Rink L: Overproduction
of monokines by leukocytes after stimulation
with lipopolysaccharide in the elderly. Exp
Gerontol 2002; 37:235247
30. Yamamoto K, Shimokawa T, Yi H, et al: Ag-
ing accelerates endotoxin-induced thrombo-
sis: Increased responses of plasminogen ac-
tivator inhibitor-1 and lipopolysaccharide
signaling with aging. Am J Pathol 2002; 161:
18051814
31. Azoulay E, Adrie C, De Lassence A, et al:
Determinants of postintensive care unit mor-
tality: A prospective multicenter study. Crit
Care Med 2003; 31:428 432
32. Higgins TL, McGee WT, Steingrub JS, et al:
Early indicators of prolonged intensive care
unit stay: Impact of illness severity, physician
staffing, and pre-intensive care unit length of
stay. Crit Care Med 2003; 31:4551
33. Ely EW, Wheeler AP, Thompson BT, et al:
Recovery rate and prognosis in older persons
who develop acute lung injury and the acute
respiratory distress syndrome. Ann Intern
Med 2002; 136:2536
34. Stephan F, Cheffi A, Bonnet F: Nosocomial
infections and outcome of critically ill elderly
patients after surgery. Anesthesiology 2001;
94:407 414
35. Ely EW, Evans GW, Haponik EF: Mechanical
ventilation in a cohort of elderly patients
admitted to an intensive care unit. Ann In-
tern Med 1999; 131:96 104
36. Seneff MG, Wagner DP, Wagner RP, et al:
Hospital and 1-year survival of patients ad-
mitted to intensive care units with acute
exacerbation of chronic obstructive pulmo-
nary disease. JAMA 1995; 274:18521857
37. Chelluri L, Pinsky MR, Donahoe MP, et al:
Long-term outcome of critically ill elderly
patients requiring intensive care. JAMA 1993;
269:3119 3123
38. Tran DD, Groeneveld AB, van der MJ, et al:
Age, chronic disease, sepsis, organ system
failure, and mortality in a medical inten-
sive care unit. Crit Care Med 1990; 18:
474 479
39. Knaus WA, Wagner DP, Draper EA, et al: The
APACHE III prognostic system. Risk predic-
tion of hospital mortality for critically ill
hospitalized adults. Chest 1991; 100:
1619 1636
40. Pittet D, Thievent B, Wenzel RP, et al: Im-
portance of pre-existing co-morbidities for
prognosis of septicemia in critically ill pa-
tients. Intensive Care Med 1993; 19:265272
41. Walter LC, Brand RJ, Counsell SR, et al:
Development and validation of a prognostic
index for 1-year mortality in older adults
after hospitalization. JAMA 2001; 285:
29872994
42. Ely EW, Angus DC, Williams MD, et al: Dro-
trecogin alfa (activated) treatment of older
patients with severe sepsis. Clin Infect Dis
2003; 37:187195
43. Bernard GR, Vincent JL, Laterre PF, et al:
Efficacy and safety of recombinant human
activated protein C for severe sepsis. N Engl
J Med 2001; 344:699 709
44. Zhan C, Miller MR: Excess length of stay,
charges, and mortality attributable to medi-
cal injuries during hospitalization. JAMA
2003; 290:1868 1874
45. Rosenberg AL, Seneff MG, Atiyeh L, et al:
The importance of bacterial sepsis in inten-
sive care unit patients with acquired im-
munodeficiency syndrome: Implications
for future care in the age of increasing
antiretroviral resistance. Crit Care Med
2001; 29:548 556
46. Hakim RB, Teno JM, Harrell FE Jr, et al:
Factors associated with do-not-resuscitate
orders: Patients preferences, prognoses,
Crit Care Med 2006 Vol. 34, No. 1
 and physicians judgments. SUPPORT In-
vestigators. Study to Understand Prog-
noses and Preferences for Outcomes and
Risks of Treatment. Ann Intern Med 1996;
125:284 293
47. Hamel MB, Davis RB, Teno JM, et al: Older
age, aggressiveness of care, and survival for
seriously ill, hospitalized adults. SUPPORT
Investigators. Study to Understand Prog-
noses and Preferences for Outcomes and
Risks of Treatments. Ann Intern Med 1999;
131:721728
Crit Care Med 2006 Vol. 34, No. 1
48. Hamel MB, Teno JM, Goldman L, et al:
Patient age and decisions to withhold life-
sustaining treatments from seriously ill,
hospitalized adults. SUPPORT Investiga-
tors. Study to Understand Prognoses and
Preferences for Outcomes and Risks of
Treatment. Ann Intern Med 1999; 130:
116 125
49. Knaus WA, Harrell FE Jr, Lynn J, et al: The
SUPPORT prognostic model. Objective esti-
mates of survival for seriously ill hospitalized
adults. Study to understand prognoses and
preferences for outcomes and risks of treat-
ments. Ann Intern Med 1995; 122:191203
50. Pronovost P, Angus DC: Using large-scale
databases to measure outcomes in critical
care. Crit Care Clin 1999; 15:615-631
51. Eaton S, Burnham E, Martin GS, et al: The
ICD-9 code for septicemia maintains a high
positive predictive value for clinical sepsis.
Am J Respir Crit Care Med 2002; 165:A471
52. Ollendorf DA, Fendrick AM, Massey K, et al:
Is sepsis accurately coded on hospital bills?
Value Health 2002; 5:79 81
21