Alternative outcome measures for pediatric clinical sepsis trials

Martha A. Q. Curley, RN, PhD, FAAN; Jerry J. Zimmerman, PhD, MD, FCCN

Objective: To review nonmortality outcome measures in clini-
cal trials of therapies to treat sepsis in children.
Data Source: Literature review using the search word terms “sepsis”
ⴙ “surrogate markers,” “sepsis” ⴙ “biomarkers,” or “sepsis” ⴙ
“outcomes.”
Study Selection: Articles were generally categorized as those
dealing with review of patient-centered outcomes, characteristics
of good surrogate markers, resolution of organ dysfunction, mor-
bidity and functional status, quality-of-life measures, intensive
care unit costs, and biomarkers.
Data Extraction and Synthesis: Information potentially relevant
for development of surrogate markers for pediatric sepsis trials
was extracted and organized as noted above.
Conclusions: Multiple potential surrogate markers are being ac-
tively investigated for their potential validity and utility in pediatric
clinical trials of sepsis and increasingly for adult sepsis trials. This is
important because as mortality decreases, it becomes an impractical
primary end point. Surrogate end points that address patient-related
morbidity and intensive care unit costs may also be quantified.
Treatment of sepsis and corresponding end points for clinical trials
should be aimed at both the duration and the quality of survival.
(Pediatr Crit Care Med 2005; 6[Suppl.]:S150 –S156)
KEY WORDS: surrogate markers; ventilator-free days; organ-
failure–free days; intensive care unit morbidity; intensive care
unit cost; quality of life; Pediatric Logistic Organ Dysfunction
score; event-free survival; functional health; health status; Pedi-
atric Overall Performance Category; functional health assessment;
Therapeutic Intervention Scoring System; biomarkers

Methodologic Quality of
Randomized Controlled Clinical
Trials in Sepsis
Problems with Using Mortality as an
Outcome Measure. The methodology of
randomized clinical trials in sepsis using
mortality as the primary end point has
improved over time (1–3). However, the
methodology of studies using surrogate
outcomes remains inadequate. In fact,
many of these studies do not report ex-
plicit outcomes or end points. What con-
stitutes sufficient outcome reporting in
sepsis trials continues to be controver-
sial. Most agree that it would include
short-term (28-day) all-cause mortality
along with secondary end points includ-
ing intensive care unit (ICU) and hospital
lengths of stay, organ dysfunction, renal
replacement therapy, physiologic vari-
ables, and complications such as nosoco-
mial infection (3). Some argue that the
From Children’s Hospital Boston, Boston, MA
(MAQC); and Children’s Hospital and Regional Medical
Center, Seattle, WA (JJZ).
This work was supported by the Mannion Family
Fund—Center for the Critically Ill Child, Division of
Critical Care Medicine at Children’s Hospital Boston,
the PALISI Network, and the ISF.
Copyright © 2005 by the Society of Critical Care
Medicine and the World Federation of Pediatric Inten-
sive and Critical Care Societies
DOI: 10.1097/01.PCC.0000161582.63265.B6
only real outcomes of importance in in-
tensive care are severity of illness–
adjusted mortality, residual morbidity,
and cost (4). Unfortunately, the tradi-
tional fixed end point of short-term mor-
tality is not practical in pediatric sepsis
trials because of the relatively low mor-
tality rate in pediatric sepsis, and thus far,
morbidity and cost can be difficult and
controversial to measure (5).
Many investigators continue to argue
that the primary outcome measure for
sepsis trials generally should be mortal-
ity; however, multiple problems exist
with the 28-day all-cause mortality end
point, particularly for pediatric trials:
1) A large number of subjects are re-
quired to demonstrate benefit, partic-
ularly as the risk of mortality contin-
ues to decrease.
2) The choice of 28 days is arbitrary
and may be too short to reflect the
effect of multiple organ dysfunction
syndrome on mortality.
3) Attributing death to sepsis as op-
posed to underlying disease is difficult,
and most children with sepsis have un-
derlying diseases that are unique from
adults.
4) Withdrawal-of-care issues cloud
mortality as an end point. This ques-
tion is critical because most death in
pediatric ICUs involves withdrawal of
care.
5) Mortality is insensitive to other im-
portant clinical outcomes. For exam-
ple, mortality does not account for
subjects with significant neurologic
morbidity after a sepsis event.
An illustrative power/sample size cal-
culation considering mortality as an end
point for a trial in severe pediatric sepsis
is instructive: to achieve a mortality re-
duction in severe pediatric sepsis from
24% to 20% (relative reduction of 16.7%,
roughly equal to that seen by Annane et
al. (6)) would require ⬎3,200 total sub-
jects to demonstrate this degree of mor-
tality reduction with an alpha of .05 and
power of .80 (7).
Alternative end points will increas-
ingly become important as mortality
from severe sepsis continues to decrease,
and under appropriate circumstances,
major morbidities could also be consid-
ered as primary end points. One measure
of morbidity is organ dysfunction, and
validated organ dysfunction scores for
children have been developed (8, 9). Pa-
tient-centered outcomes are also ex-
tremely important, including quality and
duration of life, quality of dying, and the
effect of a patient’s health on loved ones
(5). Finally, the cost of medical care must
also be considered.

S150 Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)

 Characteristics of Good Surrogate
Markers. A laboratory measurement or
physical sign may be used as a substitute
for clinically meaningful end points that
directly assess how a patient feels, func-
tions, or survives. Therapeutic-induced
changes on a surrogate end point are
expected to reflect important changes in
a clinically meaningful end point (10).
Surrogate markers must be known or
highly suspected to be in the causal path-
way of the related patient-centered out-
come (11). By definition, surrogate out-
comes are very sensitive to treatment
effects (and easier to measure) and there-
fore are considered to be more responsive
than usual patient-centered outcomes
(5). For example, a biomarker may occur
at some threshold level much sooner
than a corresponding related clinical
symptom. Not only would this marker be
more sensitive, but it may permit earlier
intervention, presumably with a therapy
that would affect the clinical sign by vir-
tue of its effect on the biomarker.
What is really at issue concerning sur-
rogate markers is the definition of clini-
cal benefit and how to measure it (12).
That is, there may be an important dis-
connect between the biological effect
(does it work?) and an actual clinical ben-
efit (does it help?). For example, in the
recent trial of the nitric oxide synthase
inhibitor, NG-methyl-L-arginine hydro-
chloride (546C88), in adults with severe
sepsis, the intervention promoted resolu-
tion of shock but actually seemed to in-
crease mortality (13, 14).
During the development of sepsis,
staging systems offer potential alternative
outcomes variables for clinical trials.
Staging itself can be associated with out-
come, although staging, like severity of
illness, is usually not considered an out-
come. For example, the PIRO staging sys-
tem assesses a patient’s predisposition,
the type of insult, the patient’s clinical
response, and organ dysfunction, which
generate potential outcome variables in-
volving the microbe and associated tox-
ins, clinical and genetic makeup of the
host, and number and intensity of organ
dysfunctions, respectively (15). Assessing
measures, markers, and mediator surro-
gate end points in clinical sepsis may
identify processes that mediate the dis-
ease state of interest at a time when an
intervention can reasonably alter disease
progression. A modification of Koch’s
postulates for identifying the role of me-
diators in sepsis has been proposed (15):
Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)
1) The mediator is present in all pa-
tients having the disease.
2) Administration of the mediator to an
experimental animal or human repro-
duces the clinical features of the dis-
ease.
3) Neutralization of the mediator be-
fore experimental induction of the dis-
ease should prevent the development
of the disease.
4) Neutralization of the mediator after
the experimental induction of the dis-
ease must attenuate its subsequent se-
verity.
Importance of illness severity and nat-
ural history of sepsis is also key in terms
of choosing appropriate outcome mea-
sures. For example, in the pediatric trial
examining bactericidal-permeability in-
creasing protein, the fulminant progres-
sion of meningococcal sepsis and the tim-
ing of intervention in relation to the
clinical explosion of the inflammatory
cascade likely contributed to the inability
to demonstrate a beneficial effect in re-
ducing mortality (16, 17). Investigators
involved in this trial emphasized the need
to develop end points other than mortal-
ity; for example, death vs. survival with
severe/moderate morbidity vs. survival
with no or mild morbidity (17). In de-
signing new trials using surrogate end
points, it will be important to establish
evidence of infection, biological plausibil-
ity of the intervention target, and illness
severity (8).
Although a mortality signal in pediat-
ric sepsis trials is unlikely to occur, dev-
astating morbidity events should be con-
sidered as valid outcome measures (18).
These events include moderate or severe
amputation, deterioration of the Pediat-
ric Overall Performance Category score of
two or more scales, and ongoing neuro-
logic or other organ dysfunction at day
28.
Organ Dysfunction Resolution. As
sepsis therapy is designed to support fail-
ing organ systems, it is rational to con-
sider rate of resolution of organ system
dysfunction as an important, morbidity-
related, alternate outcome. Essentially,
three methods may be utilized for quan-
tification of organ dysfunction in a sepsis
trial, namely, time to recovery of organ
failure, new organ failures, or organ-
failure–free days (19). As a number of
pediatric studies have clearly demon-
strated that the risk of mortality is related
to the number of organ dysfunctions
(20 –24), this alternate outcome measure
seems rational. Utilizing the Sequential
Organ Failure Assessment score, it was
demonstrated that activated protein C
(drotrecogin alfa) was associated with
faster resolution of cardiovascular (p ⫽
.009) and respiratory (p ⫽ .009) dysfunc-
tions and slower onset of hematologic (p
⫽ .041) organ dysfunction (25).
Multiple investigators have used vari-
ous scoring systems to define and quan-
titate organ dysfunction (20 –24), includ-
ing those of an International Pediatric
Severe Sepsis Consensus Conference con-
vened in 2002 (26). The only system that
has been validated is the Pediatric Logis-
tic Organ Dysfunction score (24). This
score considers neurologic, cardiovascu-
lar, renal, pulmonary, hematologic, and
hepatic organ dysfunctions and is highly
correlated with mortality. To date, no in-
vestigation has ascertained that a thera-
peutic intervention can result in faster
normalization of Pediatric Logistic Organ
Dysfunction scores in septic children. At
the current time, Eli Lilly and Company
is conducting the pediatric investigation
of activated protein C (F1K-MC-EVBP,
phase III study) with the primary objec-
tive being time to complete resolution of
a composite of cardiovascular, pulmo-
nary, and renal organ dysfunctions.
In practical terms, organ-failure–free
days are enumerated as the number of
days after enrollment to day 28 that a
patient is alive and free from clinically
significant organ failure as defined by the
Pediatric Logistic Organ Dysfunction pa-
rameters (24). This outcome reflects dif-
ferences in mortality, organ failure days
among survivors, or both, and it assumes
that any treatment that decreases the du-
ration of organ failure among the survi-
vor also increases the number of survi-
vors and that a decrease in organ failure
represents a decreased morbidity burden.
Accordingly, the number of days from
enrollment to day 28 in which a patient
does not experience organ dysfunction is
recorded, and on day 28, survivors are
assigned a score corresponding to the
number of days they did not experience
organ dysfunction. Nonsurvivors are as-
signed zero organ-failure–free days or a
score corresponding to the number of
days they did not experience organ dys-
function. This number is subtracted from
the lesser of 28 or the number of days to
death. The following summarizes one
way to calculate organ-failure–free days,
but it is not the only way (e.g., there are
multiple methods of imputation for miss-
S151
ing values). The most abnormal daily
value for each variable is used for scoring.
If a particular test was not done, previous
data are carried forward. If appropriate
data were never obtained, the value is
assumed to be normal, but if a particular
data point is missing, it is assumed to be
the worst possible variation. Organ-
failure–free days need not be contiguous.
When a patient is discharged from the
pediatric ICU to home or a rehabilitation
facility before day 28, the patient is con-
sidered to be organ failure free. If a pa-
tient is transferred to another ICU before
day 28, it is assumed that on the days for
which there is no information available,
the subject was stable from the last mea-
surement. Because organ-failure–free
days are not normally distributed, data
are usually represented as the median
number of days to day 28 that patients are
free from all organ failures, and these
data are analyzed utilizing a Wilcoxon’s
test, which weights mortality and organ
failure equally. In terms of pediatric clin-
ical trials, organ-failure–free days is be-
ing used in the pediatric activated protein
C trial noted above and has also been
used as a clinical end point for the pedi-
atric prone-positioning study (27) and in
a recent trial examining the potential
benefit of exogenous pulmonary surfac-
tant in children with acute lung injury
(D. F. Wilson, personal communication)
(28).
Other end points that reflect morbid-
ity may also be expressed in terms of
“free” days. Such end points include ven-
tilator-free days, shock-free days, compli-
cation-free days, vasoactive-drug–free
days, dialysis-free days, and ICU-free days
(7, 12). Of these, ventilator-free days has
received the most prominence and has
been utilized by the ARDS Network as an
efficacy outcome measure in clinical tri-
als of treatments for acute respiratory
distress syndrome (29). Ventilator-free
days are defined as the number of days
between successful weaning from me-
chanical ventilation to day 28 after study
enrollment. It reflects differences in mor-
tality, ventilator-free days among survi-
vors, or both and assumes that any treat-
ment that decreases the duration of
ventilation among survivors also in-
creases the number of survivors and that
the decrease in duration of ventilation
benefits the patient and care costs. Ven-
tilator-free days permits smaller sample
size if it is assumed that the treatment
simultaneously reduces the duration of
ventilation and improves mortality. It is
S152
unlikely that a treatment that led to
higher mortality could lead to a statisti-
cally significant improvement in ventila-
tor-free days, and this is especially true if
the treatment were also required to pro-
duce a nominal improvement in mortal-
ity (29). Ventilator-free days are enumer-
ated as the number of days from
enrollment to day 28 during which a pa-
tient breathed without assistance, if this
period of unassisted breathing lasted ⱖ48
consecutive hours. On day 28, survivors
are assigned a score corresponding to the
number of days they did not receive me-
chanical ventilation. Nonsurvivors are as-
signed zero ventilator-free days or a score
corresponding to the number of days they
were not supported with mechanical ven-
tilation (as long as that was ⬎48 hrs).
This number is subtracted from the lesser
of 28 or the number of days to death.
Survivors supported with mechanical
ventilation on day 28 are assigned a ven-
tilator-free days score of 0. Ventilator
days include bilevel positive airway pres-
sure and continuous positive airway pres-
sure of ⬎5 cm H2O. For patients with
tracheotomies, ventilator-free days are
measured as the number of days after
which a patient was returned to his or her
pre-illness level of support. Ventilator
days do not need to be contiguous—if the
patient is transferred before day 28, it is
assumed that on days for which no infor-
mation is available, the subject was re-
ceiving mechanical ventilation. Again as
ventilator-free days are not distributed
normally, this variable is usually analyzed
utilizing a Wilcoxon’s test, which weights
mortality and duration of ventilation
equally, but the Student’s t-test is valid
for non-normal distributions when the
sample size is large enough (29). Use of
ventilator-free days in terms of trial de-
sign requires use of study protocols that
decrease variation in ventilator manage-
ment, extubation readiness testing, and
ventilator weaning.
To control for patients who have not
been discontinued from mechanical ven-
tilation, but are probably ready to do so,
an alternate physiologic end point,
namely, time to recovery of acute lung
injury, may be utilized. This variable re-
flects the number of days from random-
ization to achieving a pulmonary crite-
rion, specifically, spontaneous breathing,
weaning of mechanical ventilator support
during the previous 24 hrs, and an oxy-
gen index of ⱕ6 (27). Time to recovery of
acute lung injury is calculated only for
patients who recover, and if the criteria
Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)
are not met by day 28, the case is cen-
sored. If transfer occurs before criteria
are met, the patient is followed to com-
pute the time to recovery of acute lung
injury. It is assumed that once the patient
is extubated, the oxygenation index is ⬍6.
Obviously, time to recovery of acute lung
injury represents an example of a more
generic physiologic end point, which is
time to recovery of any event. A similar
end point involves event-free survival, for
example, defined as the time from enroll-
ment into the study until the occurrence
of an adverse event or until the last con-
tact with the patient, whichever comes
first. Adverse events might include dis-
ease progression, diagnosis of a second
disease process, or death. Such end
points are frequently utilized in hematol-
ogy oncology clinical trials (30).
Evaluating Long-Term Outcomes.
Long-term outcome may be assessed as:
functional health, an individuals ability
to perform tasks of everyday life; quality
of life, an individual’s subjective experi-
ence of the effect of health and treatment
on one’s satisfaction with life; and health
status, a combination of both functional
health and quality of life. Various tools
are available for assessing long-term out-
come and require assessment of their va-
lidity, reliability, responsiveness, practi-
cality, and whether they are available in
the public domain. In terms of outcome
assessment after critical illness in chil-
dren, Pediatric Overall Performance Cat-
egory and Pediatric Cerebral Perfor-
mance Category scores have both been
evaluated (31). Baseline, discharge, and
delta Pediatric Overall Performance Cat-
egory and Pediatric Cerebral Perfor-
mance Category outcome scores have
been associated with pediatric ICU length
of stay and predicted risk of mortality
(32). In addition, Pediatric Overall Perfor-
mance Category is significantly related to
the Stanford–Binet Intelligence Quo-
tient, the Bayley Mental Developmental
Index, and the Vineland Adaptive Behav-
ior scores (33). For this outcome mea-
surement, delta scores are calculated as
the difference between baseline and dis-
charge scores, which controls for the ef-
fect of existing conditions on final out-
come. Typically, but not always, delta
scores are positive, reflecting overall de-
terioration. This score has been validated
in pediatric ICU populations and demon-
strates face and content validity, criterion
validity, and construct validity. These
scores seem to be responsive to individual
change over time. Interrater reliability is
adequate as reflected by interclass coeffi-
cients that range from 0.88 to 0.96. This
scale is easy to use, except for infants, for
whom there may be significant overlap in
the midrange scores. The Pediatric Over-
all Performance Category has been rec-
ommended by the American Academy of
Pediatrics, American Heart Association,
and the European Resuscitation Council
for the reporting of outcomes associated
with pediatric cardiopulmonary resusci-
tation (34, 35) and has been utilized in
multiple studies (34 –37).
Functional Health Assessment. Again,
multiple tools are available for this type
of evaluation, but two are considered
here, namely, WeeFIM® and Pediatric
Evaluation of Disability Inventory.
WeeFIM is the functional independence
measure for children and is probably the
most widely used scale for functional as-
sessment. It measures severity of disabil-
ity in terms of need for assistance in
performing basic every-day life activities,
for example, how much assistance is re-
quired for a child to complete each
WeeFIM activity above and beyond that
which is considered normal for age and
what resources are consumed to main-
tain a particular quality of life for that
child. Normative data are available, and
using this assessment, most children
achieve functional independence by 5– 6
yrs of age. Validity, reliability, and re-
sponsiveness are well documented (38,
39). WeeFIM assesses six domains,
namely, self-care, sphincter control,
transfers, locomotion, communication,
and social function, and is applicable for
children from 6 months to 8 yrs of age
and ⬎8 yrs of age in children with devel-
opmental disabilities (40). Thirteen mo-
tor and five cognitive items are each rated
on a seven-level ordinal scale that ranges
from complete independence to complete
dependence. The assessment is adminis-
tered by parent-interviewer observation
and requires ⬍20 mins to complete. One
limitation is that the WeeFIM is not sen-
sitive for finer gradations of change in
patients of ⬍18 months of age. WeeFIM
has been validated for use by direct ob-
servation or telephone interview (39, 41,
42) and has been accepted by the Joint
Commission on Accreditation of Health-
care Organizations to provide perfor-
mance measures for the Oryx initiative.
Measurement of long-term outcome
can also be assessed using the Pediatric
Evaluation of Disability Inventory, a dis-
criminative measure of functional limita-
tion (43, 44). It assesses three domains,
Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)
namely, motor, self-care, and social func-
tion, and can be use to assess children
between 6 months and 7.5 yrs of age. The
Pediatric Evaluation of Disability Inven-
tory is a 197-item test administered by
parent interview or observation and re-
quires approximately 45 mins to com-
plete. Normative data are available, and
validity and reliability are documented.
Quality-of-Life Assessment. Chal-
lenges involved in quantifying quality of
life include lack of tools that span the
pediatric population, lack of premorbid
quality-of-life baseline data, and lack of
comparable populations (45). Recently, a
benefit in terms of improved quality of
life was demonstrated for antithrombin
III in adult sepsis survivors in the Kyber-
Sept trial (46).
Clearly, long-term outcomes look be-
yond 28-day all-cause mortality. Obvi-
ously, the goal of treating a patient with
severe sepsis is to return that individual
to his or her premorbid health status. In
this regard, there may be conflicts be-
tween treatments that are, on the one
hand, life saving and, on the other, those
that affect quality of life. For example, a
long protective ventilation strategy for
acute respiratory distress syndrome may
reduce iatrogenic pulmonary injury and
decrease mortality but might lead to a
worse cognitive outcome. As nonsurvi-
vors do not contribute data to studies of
quality of life, treatments that increase
mortality may preferentially lead to the
death of debilitated patients who would
have very poor quality of life if they had
survived. Similarly, treatment that saves
the lives of very debilitated patients may
rescue these patients so that they are
healthy enough to contribute very poor
quality-of-life data to the study (47). Par-
ticular challenges to quality-of-life mea-
surements in the pediatric population in-
clude the need for proxy responders,
typically mothers or parents who live
with the patient, leading to potential bi-
ases with parent-proxy measures. This
may result in idealized views or expecta-
tions of a parent’s child, convergence of
the parents’ subjective experiences with
the perceived experience of their child, or
transference of the parents’ own sense of
debilitated function onto the child. An
ideal quality-of-life instrument would in-
clude both child and parent-proxy report
with concrete objective items (48). The
age at which a child is able to give a valid
and reliable response varies according to
the type of information sought and the
complexity of the questionnaire.
The Child Health Questionnaire mea-
sures overall health status of children
5–18 yrs of age (49). This tool utilizes 50
questions and generates two summary
scores, namely, psychosocial and physical
health. These summary scores are de-
rived from 12 subscales. The psychomet-
rics for the Child Health Questionnaire is
suitable for use as a primary health-
related quality-of-life outcome variable in
clinical trials (50 –53). This tool has been
utilized after surgery for D-transposition
of the great vessels (54).
Lastly, the Pediatric Quality of Life
Inventory (PedsQL 4.0) is increasingly
being utilized as a long-term outcome
measure after pediatric critical illness.
This is a modular measure of health-
related quality of life, which is reliable in
children aged 2–18 yrs and includes ge-
neric core scales (23 items) based on both
child self-report and parent proxy. Four
summary scores are generated, including
physical, emotional, social, and school
subscores. The Pediatric Quality of Life
Inventory can be integrated with disease-
specific measure, and psychometrics of
the total score are suitable as a summary
score for use as a primary health-related
quality-of-life outcome variable in clini-
cal trials (55–58).
Utilizing similar tools, long-term sur-
vival and state of health after pediatric
intensive care have been examined in 254
children of ⬎1 yr of age (59). In this
cohort, ICU mortality was 7.5%, hospital
mortality was 8.3%, and 1-yr mortality
was 10.5%. In this investigation, the out-
come assessment tool utilized was the
Multiattribute Health Status Classifica-
tion (60), which scores sensation, mobil-
ity, motion, cognition, self-care, and
pain. The score was compared preadmis-
sion and 1 yr after discharge. Irrespective
of the magnitude of change, overall
health status improved in 25.7% and de-
teriorated in 27.4%. Most changes in
overall health status were small, with
only one domain more or less affected.
Interestingly, no relation between acute
severity of illness (Pediatric Risk of Mor-
tality score) and health status before ad-
mission and after 1 yr were appreciated.
Some degree of health impairment 1 yr
after pediatric ICU discharge was present
in 66.4% of patients—most had some
health impairment before admission.
Overall state of health 1 yr after discharge
was unchanged or unimproved in 72.6%
(59).
Intensive Care Unit Costs. It would
seem logical that intensity of illness
S153
would be related to intensity of therapeu-
tic interventions and, accordingly, ICU
costs. Utilizing care costs as an outcome
measure in multiple-institution trials is
complicated by regional variations in
clinical practice, charges, and payor mix.
However, these difficulties may be over-
come by capitalizing on the relationship
between the sum of Therapeutic Inter-
vention Scoring System (TISS) (61, 62)
scores and ICU costs. In fact, the corre-
lation between total variable costs per
patient admission episode and the total
TISS per patient admission episode is
high (63). The relationship between TISS
and real pediatric ICU costs were exam-
ined in data collected from a ten-bed pe-
diatric ICU over 17 months for children
aged 1 month to 16 yrs. This cohort in-
cluded 611 consecutive admissions and
3190 patient days. For this group of pa-
tients, it was feasible to calculate total
direct medical costs based on a limited
number of readily available clinical vari-
ables related to patient characteristics
and treatment, of which TISS was the
most important determinant (63). Physi-
cian time, nursing time, and pharmaceu-
ticals contributed to 63% of the major
direct medical cost components, and lab-
oratory, material/disposables, and over-
head/depreciation contributed 11%, 11%,
and 10%, respectively. Although total pa-
tient costs per day were fairly normally
distributed, total costs per admission
were skewed to the left, with a long tail of
higher costs associated with prolonged
ICU admission for complex patients.
Resource utilization in the ICU has
also been assessed with computerized
TISS-based data (64). In this investiga-
tion, 1,229 adults were evaluated from
⬎1,372 admissions in eight ICUs within
the University of Pittsburgh system. TISS
was collected manually, and a computer-
ized TISS was obtained by developing a
map between TISS items and correspond-
ing charge items (transaction codes in
the billing database). Mean and median
manual and computerized TISS scores
were nearly identical. Ninety-five percent
of manually computed TISS scores were
⬍10. Comparison between the computer-
derived TISS score and the manually de-
rived TISS score exactly matched on
55.5% of days and nearly matched on
83.3% of days, and correlation between
the two was substantial (R2 ⫽ .85), with
the highest correlation noted for scores
of ⬍10. Sensitivity, specificity, positive
predictive value, and negative predictive
value for mechanical ventilation and va-
S154
soactive drug infusion were all very high
in terms of performance of the comput-
erized TISS mapping algorithm.
Biomarkers. It has been noted that the
common pathologic event in sepsis is
represented by the loss of the usual con-
trol mechanisms that regulate and com-
partmentalize inflammation (15). Ac-
cordingly, there has been wide interest in
identifying reliable biomarkers, not only
to detect sepsis but also to serve as a
reliable surrogate in terms of sepsis res-
olution in response to a therapy. Inter-
leukin 6 is known to be associated with
sepsis mortality (65). In the trial of the
assessment of safety and efficacy of
monoclonal anti–tumor necrosis factor
antibody fragment (MAK 195F) in pa-
tients with severe sepsis, patients with
interleukin-6 levels of ⬍1000 pg/mL
showed similar response to varying doses
of the anti–tumor necrosis factor anti-
body dosing. However, patients with ini-
tial interleukin-6 levels of ⬎1000 pg/mL
demonstrated a beneficial dose response
effect in terms of mortality reduction
with the anti–tumor necrosis factor in-
tervention (66).
As a secondary outcome marker for
goal-directed therapy in adults with se-
vere sepsis, lactate clearance was utilized
as a surrogate marker (67). Early sepsis
resuscitation included maintaining cen-
tral venous pressure at 10 ⫾ 2, mean
arterial pressure at ⬎65, and urine out-
put at ⬎0.5 mL·kg⫺1·hr⫺1. For patients
demonstrating ⱖ10% lactate clearance
during their first few hours of resuscita-
tion (as compared with ⬍10%), a signif-
icant decrease in mortality was docu-
mented. Low lactate clearance is also a
predictor of mortality in ICU patients
with severe sepsis (68). These data con-
firm earlier studies indicating that time
to clearance of lactate (lactime) was
strongly associated with mortality (69).
Pediatric sepsis mortality has also been
examined as a function of blood lactate
obtained 12 hrs after admission in 31
children with sepsis (70). At 24 hrs, a
continuing lactate of ⬎3 mM had a pos-
itive predictive value for death of 71%.
Area under the receiver-operator curve
plot for blood lactate assayed 12 hrs after
admission was 0.81.
A biomarker of sepsis reflects a sin-
gle biological aspect of a complex and
heterogeneous process and not the en-
tire spectrum of the disease. The sys-
temic inflammatory response syndrome
initiated by infection is complex, net-
worked, redundant, and genetically de-
Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)
termined and reflects microbial type
and burden, concurrent illness, and ac-
quired immunodeficiency (71). All of
these contingencies need to be valued
when considering a specific biomarker
as a surrogate marker for sepsis clinical
trials. For example, there seems to be
an overwhelming production of nitric
oxide in septic shock, which promotes
peripheral vasodilation, induces cate-
cholamine resistance, elicits cardiomy-
opathy, mediates cytopathic hypoxia,
and inhibits vasopressin release (72).
Multiple neonatal, pediatric, and adult
investigations have demonstrated a
clear relationship between the serum
load of nitrate ⫹ nitrite (catabolic end
products of nitric oxide) in relation to
sepsis mortality and organ dysfunction
(73–77). However, as noted above, an
inhibitor substrate for nitric oxide syn-
thase (546C88) reduced elevated plasma
nitrate ⫹ nitrite concentrations ob-
served in septic shock and improved
vascular tone, easing successful main-
tenance of a target mean arterial blood
pressure of ⬎70 mm Hg with a reduc-
tion in vasoactive agents without ad-
verse effects, but was associated with an
increase in mortality (13, 14).
Conclusion
Sepsis continues to represent a major
problem in both adult and pediatric pop-
ulations. Probably related to various rea-
sons for increasing acquired immunode-
ficiency, the prevalence of sepsis actually
seems to be increasing. For a number of
practical reasons, mortality as an end
point for clinical trials of severe sepsis,
particularly for pediatric studies, is not
useful. Accordingly, there is an active
search for meaningful, alternative surro-
gate markers. Currently, organ-failure–
free days and long-term outcome are re-
ceiving the most scrutiny in pediatric
clinical sepsis trials. An ideal biomarker
that is sensitive and specific and reflects
disease progression and resolution has
not yet been validated.
REFERENCES
1. Lefering R, Neugebauer EA: Steroid contro-
versy in sepsis and septic shock: A meta-
analysis. Crit Care Med 1995; 23:1294 –1303
2. Cronin L, Cook DJ, Carlet J, et al: Cortico-
steroid treatment for sepsis: A critical ap-
praisal and meta-analysis of the literature.
Crit Care Med 1995; 23:1430 –1439
3. Graf J, Doig GS, Cook DJ, et al: Randomized,
controlled clinical trials in sepsis: Has meth-
 odological quality improved over time? Crit
Care Med 2002; 30:461– 472
4. de Keizer NF, Bonsel GJ, Al MJ, et al: The
relation between TISS and real paediatric
ICU costs: A case study with generalizable
methodology. Intensive Care Med 1998; 24:
1062–1069
5. Rubenfeld GD: Surrogate measures of pa-
tient-centered outcomes in critical care. In:
Surviving Intensive Care. Angus DC, Carlet J
(Eds). Berlin, Springer-Verlag, 2002
6. Annane D, Sebille V, Charpentier C, et al:
Effect of treatment with low doses of hydro-
cortisone and fludrocortisone on mortality in
patients with septic shock. JAMA 2002; 288:
862– 871
7. Vincent JL: Endpoints in sepsis trials: More
than just 28-day mortality? Crit Care Med
2004; 32(Suppl):S209 –S213
8. Cohen JC, Guyatt G, Bernard GR, et al: New
strategies for clinical trials in patients with
sepsis and septic shock. Crit Care Med 2001;
29:880 – 886
9. Bernard GR: Sepsis trials: Intersection of in-
vestigation, regulation, funding and practice.
Am J Respir Crit Care Med 1995; 152:4 –10
10. Temple RJ: A regulatory authority’s opinion
about surrogate endpoints. In: Clinical Mea-
surement in Drug Evaluation. Nimmo WS,
Tucker GT (Eds). New York, J Wiley, 1995, p
329
11. Prentice RL: Surrogate endpoints in clinical
trials: Definition and operational criteria.
Stat Med 1989; 8:431– 440
12. Bernard GR: Research in sepsis and acute
respiratory distress syndrome: Are we chang-
ing course? Crit Care Med 1999; 27:434 – 436
13. Bakker J, Grover R, McLuckie A, et al: Ad-
ministration of the nitric oxide synthase in-
hibitor NG- methyl L-arginine hydrochloride
(546C88) by intravenous infusion for up to
72 hours can promote resolution of shock in
patients with severe sepsis: Results of a ran-
domized, double-blind, placebo-controlled
multicenter study (study no. 144-002). Crit
Care Med 2004; 32:1–12
14. Watson D, Grover R, Anzueto A, et al: Car-
diovascular effects of the nitric oxide syn-
thase inhibitor NG-methyl-L-arginine hydro-
chloride (546C88) in patients with septic
shock: Results of a randomized, double blind,
placebo-controlled multicenter study (study
no. 144-002). Crit Care Med 2004; 32:13–20
15. Marshall JC, Vincent JL, Fink MP, et al: Mea-
sures, markers, and mediators: Toward a
staging system for clinical sepsis. Crit Care
Med 2003; 31:1560 –1567
16. Giroir BP, Quint PA, Barton P, et al: Prelim-
inary evaluation of recombinant amino-
terminal fragment of human bactericidal/
permeability-increasing protein in children
with severe meningococcal sepsis. Lancet
1997; 350:1439 –1443
17. Levin M, Quint PA, Goldstein B, et al: Re-
combinant bactericidal/permeability increas-
ing protein (rBPI21) as adjunctive treatment
for children with severe meningococcal sep-
Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)
sis: A randomized trial. Lancet 2000; 356:
961–967
18. Giroir BP, Scannon PJ, Levin M: Bactericidal
permeability-increasing protein: Lessons
learned from the phase III, randomized, clin-
ical trial of rBPI21 for adjunctive treatment of
children with severe meningococcemia. Crit
Care Med 2001; 29(Suppl):S130 –S135
19. Bota DP, Melot C, Lopes Ferreira F, et al: The
Multiple Organ Dysfunction Score (MODS)
versus the Sequential Organ Failure Assess-
ment (SOFA) score in outcome prediction.
Intensive Care Med 2002; 28:1619 –1624
20. Wilkinson JD, Pollack MM, Ruttiman UE, et
al: Outcome of pediatric patients with mul-
tiple organ system failure. Crit Care Med
1986; 14:271–274
21. Wilkinson JD, Pollack MM, Glass NL, et al:
Mortality associated with multiple organ sys-
tem failure and sepsis in pediatric intensive
care unit. J Pediatr 1987; 111:324 –328
22. Proulx F, Gauthier M, Nadeau D, et al: Tim-
ing and predictors of death in pediatric pa-
tients with multiple organ system failure.
Crit Care Med 1994; 22:1025–1031
23. Tantalean JA, Leon RJ, Santos AA, et al: Mul-
tiple organ dysfunction in children. Pediatr
Crit Care Med 2003; 4:181–185
24. Leteurtre S, Martinot A, Duhamel A, et al:
Validation of the paediatric logistic organ
dysfunction (PELOD) score: Prospective, ob-
servational, multicenter study. Lancet 2003;
362:192–197
25. Vincent JL, Angus DC, Artigas A, et al: Effects
of drotrecogin alfa (activated) on organ dys-
function in the PROWESS trial. Crit Care
Med 2003; 31:834 – 840
26. Goldstein B, Giroir B, Randolph A, et al:
International Pediatric Severe Sepsis Con-
sensus Conference: Definitions for sepsis and
organ dysfunction in pediatrics. Pediatr Crit
Care Med 2005; 6:2– 8
27. Curley MAQ, Fineman LD, Cvijanovich N, et
al: Prone positioning in pediatric acute lung
injury: Results of a randomized controlled
clinical trial. Presented at the 15th ESPNIC
Medical and Nursing Annual Congress, Lon-
don, September, 2004
28. Wilson DF, Zaritsky A, Bauman LA, et al:
Instillation of calf lung surfactant extract
(calfactant) is beneficial in pediatric acute
hypoxemic respiratory failure. Crit Care Med
1999; 27:188 –195
29. Schoenfeld DA, Bernard GR: Statistical eval-
uation of ventilator-free days as an efficacy
measure in clinical trials of treatments for
acute respiratory distress syndrome. Crit
Care Med 2002; 30:1772–1777
30. Grier HE, Krailo MD, Tarbell NJ, et al: Addi-
tion of ifosfamide and etoposide to standard
chemotherapy for Ewing’s sarcoma and
primitive neuroectodermal tumor of bone.
N Engl J Med 2003; 348:694 –701
31. Fiser DH: Assessing outcome of pediatric in-
tensive care. J Pediatr 1992; 121:68 –74
32. Fiser DH, Tilford JM, Robertson PK: Rela-
tionship of illness severity and length of stay
to functional outcomes in the pediatric in-
tensive care unit: A multi-institutional study.
Crit Care Med 2000; 28:1173–1179
33. Fiser DH, Long N, Robertson PK, et al: Re-
lationship of pediatric overall performance
category and pediatric cerebral performance
category scores at pediatric intensive care
unit discharge with outcome measures col-
lected at hospital discharge and 1- and
6-month follow-up assessments. Crit Care
Med 2000; 28:2616 –2620
34. Schindler MB, Bohn D, Cox PN, et al: Out-
come of out-of-hospital cardiac or respira-
tory arrest in children. N Engl J Med 1996;
335:1473–1479
35. Torres A Jr, Pickert CB, Firestone J, et al:
Long-term functional outcome of inpatient
pediatric cardiopulmonary resuscitation. Pe-
diatr Emerg Care 1997; 13:369 –373
36. Hackbaarth RM, Rzeszutko KM, Sturm G, et
al: Survival and functional outcome in pedi-
atric traumatic brain injury: A retrospective
review and analysis of predictive factors. Crit
Care Med 2002; 30:1630 –1635
37. Goldstein B, Fiser DH, Kelly MM, et al: De-
complexification in critical illness and in-
jury: Relationship between heart rate vari-
ability, severity of illness, and outcome. Crit
Care Med 1998; 26:352–357
38. Ziviani J, Ottenbacher KJ, Shephard K, et al:
Concurrent validity of the Functional Inde-
pendence Measure for Children (WeeFIM)
and the Pediatric Evaluation of Disabilities
Inventory in children with developmental
disabilities and acquired brain injury. Phys
Occup Ther Pediatr 2001; 21:91–101
39. Sperle PA, Ottenbacher KJ, Braun SL, et al:
Equivalence reliability of the Functional In-
dependence Measure for Children (WeeFIM):
Administration methods. Am J Occup Ther
1997; 51:35– 41
40. WeeFIM System Clinical Guide: Version 5.01.
Buffalo, NY, State University of New York at
Buffalo, 1998
41. Msall ME, DiGaudio K, Duffy LC: Use of func-
tional assessment in children with develop-
mental disabilities. Phys Med Rehabil Clin
North Am 1993; 4:517–527
42. Ottenbacher KJ, Taylor ET, Msall ME, et al:
The stability and equivalence reliability of
the Functional Independence Measure for
Children (WeeFIM). Dev Med Child Neurol
1996; 38:907–916
43. Berg M, Jahnsen R, Froslie KF, et al: Reli-
ability of the Pediatric Evaluation of Disabil-
ity Inventory (PEDI). Phys Occup Ther Pedi-
atr 2004; 24:61–77
44. Kothari DH, Haley SM, Gill-Body KM, et al:
Measuring functional change in children
with acquired brain injury (ABI): Compari-
son of generic and ABI-specific scales using
the Pediatric Evaluation of Disability Inven-
tory (PEDI). Phys Ther 2003; 83:776 –785
45. Herridge MS, Cheung AM, Tansey CM, et al:
One-year outcomes in survivors of the acute
respiratory distress syndrome. N Engl J Med
2003; 348:683– 693
46. Rublee D, Opal SM, Schramm W, et al: Qual-
ity of life effects of antithrombin III in sepsis
S155
 survivors: Results from the KyberSept trial
[ISRCTN22931023]. Crit Care 2002;
6:349 –356
47. Rubenfeld G: Looking beyond 28-day all-
cause mortality. Crit Care 2002; 6:293–294
48. Parsons SK, Brown AP: Evaluation of quality
of life of childhood cancer survivors: A meth-
odological conundrum. Med Pediatr Oncol
1998; Suppl 1:46 –53
49. Landgraf JM, Abetz L, Ware JE: Childhealth
Questionnaire (CHQ) Users Manual. First
Edition. Boston, Health Institute, New En-
gland Medical Center, 1996
50. Georgalas C, Tolley N, Kanagalingam J: Mea-
suring quality of life in children with adeno-
tonsillar disease with the Child Health Ques-
tionnaire: A first UK study. Laryngoscope
2004; 114:1849 –1855
51. Panepinto JA, O’Mahar KM, DeBaun MR, et
al: Validity of the child health questionnaire
for use in children with sickle cell disease.
J Pediatr Hematol Oncol 2004; 26:574 –578
52. Warschburger P, Landgraf JM, Petermann F,
et al: Health-related quality of life in children
assessed by their parents: Evaluation of the
psychometric properties of the CHQ-PF50 in
two German clinical samples. Qual Life Res
2003; 12:291–301
53. Klassen AF, Landgraf JM, Lee SK, et al:
Health related quality of life in 3 and 4 year
old children and their parents: Preliminary
findings about a new questionnaire. Health
Qual Life Outcomes 2003; 1:81
54. Dunbar-Masterson C, Wypij D, Bellinger DC,
et al: General health status of children with
D-transposition of the great arteries after the
arterial switch operation. Circulation 2001;
104(12 Suppl 1):I138 –I142
55. Varni JW, Seid M, Kurtin PS: PedsQL 4.0:
Reliability and validity of the Pediatric Qual-
ity of Life Inventory version 4.0 generic core
scales in healthy and patient populations.
Med Care 2001; 39:800 – 812
56. Varni JW, Seid M, Rode CA: The PedsQL in
pediatric rheumatology: Reliability, validity,
and responsiveness of the Pediatric Quality of
S156
Life Inventory Generic Core Scales and
Rheumatology Module. Arthritis Rheum
2002; 46:714 –725
57. Varni JW, Burwinkle TM, Katz ER, et al: The
PedsQL in pediatric cancer: Reliability and
validity of the Pediatric Quality of Life Inven-
tory Generic Core Scales, Multidimensional
Fatigue Scale, and Cancer Module. Cancer
2002; 94:2090 –2106
58. Varni JW, Seid M, Knight TS, et al: The
PedsQL 4.0 Generic Core Scales: Sensitivity,
responsiveness, and impact on clinical deci-
sion-making. J Behav Med 2002; 25:175–193
59. Gemke RJBJ, Bonsel GJ, van Vught AJ: Long
term survival and state of health after paedi-
atric intensive care. Arch Dis Child 1995;
73:196 –201
60. Feeny D, Furlong W, Barr RD, et al: A com-
prehensive multiattribute system for classi-
fying the health status of survivors of child-
hood cancer. J Clin Oncol 1992; 10:923–928
61. Cullen DJ, Civetta JM, Briggs BA, et al: Ther-
apeutic intervention scoring system: A
method for quantitative comparison of pa-
tient care. Crit Care Med 1974; 2:57– 60
62. Pollack MM, Yeh TS, Ruttiman UE, et al:
Evaluation of pediatric intensive care. Crit
Care Med 1984; 12:376 –383
63. Dickie H, Vedio A, Dundas R, et al: Relation-
ship between TISS and ICU cost. Intensive
Care Med 1998; 24:995–966
64. Clermont G, Angus DC, Linde-Zwirble WT, et
al: Measuring resource use in the ICU with
computerized Therapeutic Intervention
Scoring System– based data. Chest 1998;
113:434 – 442
65. Damas P, Ledoux D, Nys M, et al: Cytokine
serum levels during severe sepsis in humans:
IL-6 as a marker of severity. Ann Surg 1992;
215:356 –362
66. Reinhart K, Weigand-Löhnert C, Grim-
minger F, et al: Assessment of the safety and
efficacy of the monoclonal anti-tumor necro-
sis factor antibody-fragment, MAK 195F, in
patients with sepsis and septic shock: A mul-
ticenter, randomized, placebo-controlled,
Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)
dose-ranging study. Crit Care Med 1996; 24:
733–742
67. Nguyen HB, Rivers EP, Knoblich BP, et al:
Early lactate clearance is associated with im-
proved outcome in severe sepsis and septic
shock. Crit Care Med 2004; 32:1637–1642
68. Levraut J, Ichai C, Petit I, et al: Low exoge-
nous lactate clearance as an early predictor
of mortality in normolactatemic critically ill
septic patients. Crit Care Med 2003; 31:
705–710
69. Bakker J, Gris P, Coffernils M, et al: Serial
blood lactate levels can predict the develop-
ment of multiple organ failure following sep-
tic shock. Am J Surg 1996; 171:221–226
70. Duke TD, Butt W, South M: Predictors of
mortality and multiple organ failure in chil-
dren with sepsis. Intensive Care Med 1997;
23:684 – 692
71. Zimmerman JJ: A question of balance. Crit
Care Med 1999; 27:7– 8
72. Vincent JL, Zhang H, Szabo C, et al: Effects
of nitric oxide in septic shock. Am J Respir
Crit Care Med 2000; 161:1781–1785
73. Yuan S, Li HQ, Shen CK, et al: Plasma nitric
oxide levels in newborn infants with sepsis.
J Pediatr 1993; 123:435– 438
74. Doughty LA, Kaplan SS, Carcillo JA: Inflam-
matory cytokine and nitric oxide responses
in pediatric sepsis and organ failure. Crit
Care Med 1996; 24:1137–1143
75. Wong HR, Carcillo JA, Burckart G, et al:
Increased serum nitrite and nitrate concen-
trations in children with the sepsis syn-
drome. Crit Care Med 1995; 23:835– 842
76. Kraft-Jacobs B, Brilli R, Szabo C, et al: Cir-
culating methemoglobin and nitrite/nitrate
concentrations as indicators of nitric oxide
overproduction in critically ill children with
septic shock. Crit Care Med 1997; 25:
1588 –1593
77. Spack L, Havens PL, Griffith OW: Measure-
ments of total plasma nitrite and nitrate in
pediatric patients with the systemic inflam-
matory response syndrome. Crit Care Med
1997; 25:1071–1078