Outcomes and delivery of care in pediatric injury
John C. Densmorea,b,*, Hyun J. Limb, Keith T. Oldhama,b,c, Karen S. Guicea,b,c
aDepartment of Surgery, Division of Pediatric Surgery, Medical College of Wisconsin, Milwaukee, WI 53226, USA
bChildren’s Research Institute, Milwaukee, WI 53226, USA
cAPSA Outcomes Center, Milwaukee, WI 53226, USA
Purpose: To design effective pediatric trauma care delivery systems, it is important to correlate site of
care with corresponding outcomes. Using a multistate administrative database, we describe recent
patient allocation and outcomes in pediatric injury.
Methods: The 2000 Kids’ Inpatient Database, containing 2,516,833 inpatient discharge records from 27
each discharge using ICDMAP-90 (Tri-Analytics, Inc, Forest Hill, MD). After weighting to estimate
of Children’s Hospitals and Related Institutions–designated site of care. Measured outcomes included
mortality, length of stay, and total charges. Analysis was completed using Student’s t test and v2.
Results: Among 79,673 injury cases, mean age was 12.2 F 6.2 years and ISS was 7.4 F 7.6. Eighty-nine
percent of injured children received care outsideof children’shospitals. In the subgroup of patients aged0
to 10 years with ISS of greater than 15, the mean ISS for adult hospitals and children’s hospitals was not
significantly different (18.9 F 9.1 vs. 19.4 F 9.3, P = .08). However, in-hospital mortality, length of stay,
and charges were all significantly higher in adult hospitals (P b.0001).
Conclusions: Younger and more seriously injured children have improved outcomes in children’s
hospitals. Appropriate triage may improve outcomes in pediatric trauma.
D 2006 Elsevier Inc. All rights reserved.
Injury is the single greatest cause of morbidity and
mortality among children and adolescents in the United
Statesafter theageof1year.In 2002,the mostrecent yearfor
which data are available, there were 18,953 injury-related
deaths among individuals aged 1 to 20 years . Of these,
13,280 deaths resulted from unintentional injuries, making it
the leading cause of death in this age group. Motor-vehicle
crashes were the most common mechanism of injury in this
category, accounting for 8816 deaths. Unintentional injury
isthesixthleadingcauseofdeathforchildren less than1year
of age, with 946 deaths reported in 2002.
Incidence of nonfatal injuries is more difficult to assess.
Injured children receive evaluation and treatment in a
variety of settings including homes, schools, physician
offices, and emergency departments. In 2002, an estimated
10,393,392 children, aged less than 1 to 20 years, were seen
0022-3468/$ – see front matter D 2006 Elsevier Inc. All rights reserved.
Presented at the 36th Annual Meeting of the American Pediatric Surgical
Association, Phoenix, AZ, May 29-June 1, 2005.
This was supported in part by grant # 5 H03 MC 00011-04 from the
Maternal and Childhood Health Bureau of the Health Research and
Services Administration, US Department of Health and Human Services.
* Corresponding author. Division of Pediatric Surgery, Medical
College of Wisconsin and Children’s Hospital of Wisconsin, Milwaukee,
WI 53226, USA. Tel.:+ 1 414 266 6557; fax: +1 414 456 6473.
E-mail address: email@example.com (J.C. Densmore).
Length of stay;
Journal of Pediatric Surgery (2006) 41, 92–98
in emergency departments with nonfatal injuries. The
leading causes of nonfatal injury in this group were falls
and being struck by or against an object or person .
The central premise of trauma systems is that injured
patients are provided rapid access to life-saving care.
Children, because of their unique physiology, have particular
needs and present challenges to predominantly adult-
oriented trauma care delivery systems. In 1999, the
American College of Surgeons (ACS) Committee on Trauma
stated that care for injured children bmay be optimally
provided in the environment of a children’s hospital with a
demonstrated commitment to trauma careQ . Previous
ACS guidelines recommended the presence of a pediatric
surgeon, when possible, to treat pediatric trauma . This
recognition highlights a longstanding question of whether
specialized pediatric trauma care improves outcomes for
these youngest, sickest, and most vulnerable patients.
Multiple investigators have sought to further define
optimal pediatric trauma care by using the best evidence
available. Studies of single institution experiences and state
databases (eg, Pennsylvania, New Jersey, New York,
Oregon, Washington) have demonstrated the positive role
of a statewide trauma system in decreasing mortality [4,5].
Others have demonstrated higher mortality rates for
pediatric trauma patients in a rural setting , decreased
mortality with pediatric intensive care capabilities at
receiving centers , equivalence in mortality rates with
and without pediatric surgery care providers [8-12],
increased operative rates and mortality at adult centers
[13,14], and improved functional outcome with care in
pediatric centers . Such studies have provided helpful
indicators, but are limited in power and often reflect region-
specific trauma care patterns.
Osler et al  used the National Pediatric Trauma
Registry to study survival rates of children in pediatric vs
adult trauma centers admitted from 1985 to 1996. Using this
self-reporting database spanning a period of nearly a decade,
we found that injury severity–adjusted mortality rates
between center types were not different. Interestingly,
pediatric patients entering adult trauma centers in this series
had a mean Injury Severity Score (ISS) of 9.4 vs 6.9 in the
pediatric trauma centers.
Recently, Segui-Gomez et al  described delivery of
pediatric trauma care in 18 states using state hospital
discharge databases from 1997 to 1998. This well-powered
study revealed that only 10% of pediatric trauma patients
received care in pediatric trauma centers. Although most
of these patients received care in large urban areas
(N106persons), triage to pediatric centers was not correla-
ted with injury severity. Mortality among pediatric trauma
patients was higher in adult trauma centers with pediatric
trauma designation when compared to pediatric-only
trauma centers (2% vs 1%, P b .05). These data would
suggest that factors other than acuity of injury drove
pediatric trauma triage (eg, proximity to pediatric trauma
center, financial considerations).
Loss of revenue by adult trauma centers with implemen-
tation of regional pediatric trauma triage was examined
by MacKenzie et al  using the National Hospital
Data Survey. This study concluded that triaging pediatric
trauma patients to designated centers would have minimal
impact upon the existing trauma care providers. Neverthe-
less, their models estimate that 91% of pediatric trauma care
An additional concern of increasing triage of pediatric
trauma patients to pediatric centers is the possibility of
overtriage. Engum et al  calculated that the overtriage
rate was 71% among 1285 pediatric patients using the ACS
triage criteria. Other groups have identified a manpower
shortage in pediatric surgery and related subspecialties to
care for pediatric trauma patients [14,20]. D’Amelio et al
 calculated that the use of pediatric surgeons to evaluate
all pediatric trauma would result in 1200 cases per year
per pediatric surgeon. The cost of overtriage could be
significant and has been confirmed in the context of air
transport in pediatric trauma patients [21,22].
The purpose of this study was to characterize pediatric
trauma care by the National Association of Children’s
Hospitals and Related Institutions (NACHRI) hospital type
and identify associated outcomes, especially in the youn-
gest, most seriously injured children. This analysis uses the
2000 Kids’ Inpatient Database (KID), an administrative
database containing discharge information for ~2.5 million
US children in the year 2000. The database represents
27 states and greater than 70% of the total inpatient pediatric
population in the United States. This type of database is
unique as it provides a large and comprehensive sample
over 1 year’s time, hence minimizing the effects of ever-
changing systems of care.
1. Materials and methods
The 2000 KID, used for these analyses, is part of the
Healthcare Cost and Utilization Project sponsored by the
Agency for Healthcare Research and Quality (AHRQ).
The KID is a derivative database composed of sampled
pediatric discharges from 27 state inpatient databases (SIDs)
from the Healthcare Cost and Utilization Project SIDs.
The KID sampling frame includes all pediatric discharges
from community, nonrehabilitation hospitals stratified by
uncomplicated in-hospital births, complicated in-hospital
births, and nonbirth events. The resulting records contained
in the KID include 10% of uncomplicated in-hospital births
and 80% of all other pediatric discharges from each frame
hospital, selected using systematic random sampling.
Weights can be applied to KID records to provide a
national estimate. The weighting factor is based on
6 characteristics of the poststratified hospitals (geographic
region, ownership [public, private not-for-profit, proprie-
tary], urban or rural location, teaching status, bed size
[small, medium, or large], and type of hospital [children’s or
Outcomes and delivery of care in pediatric injury93
other]). An additional weighting factor for hospital charges
at the time of discharge is also available to produce national
estimates of hospital charges.
The 2000 KID includes 2,516,833 records for children
aged 20 or younger at the time of hospital admission. Data
elements, collected from the 27 participating states, are
standardized to allow for comparisons of pediatric hospital
use across the United States. Limitations of the dataset
include an incomplete hospital universe for sampling and
possible biases in the state-submitted data.
The KID contains pediatric hospitalization data from
2784 hospitals out of 2788 in the sampling frame, which
is selected from 3247 hospitals within the SID with
pediatric discharges (excluding rehabilitation hospitals).
The American Hospital Association buniverseQ of hospi-
tals for the same period is 4839. In addition, regional
representation of American Hospital Association hospitals
within the KID is only 30% for the Midwest (approxi-
mately 90% for the Northeast, 57% for the South, and
76% for the West).
Hospital discharge datasets reflect only those hospitals
that participate and can change annually. Requirements
for data element reporting vary among the states,
particularly with regard to E-code reporting. Depending
on the state’s ability to link records, duplicate records may
exist for one individual. For example, a patient may be
transferred from one hospital to another or transferred to a
rehabilitation unit within the hospital and readmitted. In
addition, hospital coding practices may vary, and the
principal diagnosis field may not reflect the actual reason
Records contained in the 2000 KID were first filtered for
the presence of an E-code, a specific ICD-9CM code
descriptive of specific injury type and location, in any one
of 15 diagnosis fields. The resulting 210,895 records
were coded for injury severity using ICDMAP-90.
ICDMAP-90 translates ICD-9CM diagnosis codes into
abbreviated injury scale and ISSs and has been validated
[23,24]. Records for which abbreviated injury scale and
ISSs could not be computed were deleted, leaving 123,347
records for analysis.
These records were further filtered to include only those
children who were urgently or emergently admitted to a
hospital. Records containing birth codes, lacking informa-
tion regarding in-hospital mortality or age, or those
containing a principal diagnosis code for which an ISS
could not be calculated, were also deleted. Remaining
records totaled 79,673 and were divided into 2 groups:
(1) all pediatric injury patients and (2) those aged 0 to
10 years with an ISS of greater than 15. The selected ISS
threshold of 15 is consistent with previously published
studies and allows grouping for comparison. Both groups
were carried separately through all analyses. Means were
derived for demographic information including age, sex,
ISS, and total charges. A frequency analysis was used to
report sites of care for injured children by NACHRI
hospital type. Fewer than 10 children received care in
NACHRI-designated children’s specialty hospitals. These
cases were removed, leaving the sites children’s hospital,
children’s units in general hospitals, and nonchildren’s
hospitals for further analysis.
Although mortality was expressed as a simple percent-
age, length of stay and total charges were expressed as
percentage of each group lying above a derived threshold.
Length of stay of 8 days or more corresponds to the upper
10th percentile of hospital stay periods for all injured
children. Charges of more than $15,000 delineate total
hospital charges above the 50th percentile for all injured
children. Outcomes were additionally analyzed using a
generalized linear model controlling for ISS as a covariate.
This regression analysis was performed to assess outcome
measurement differences across hospital types after adjust-
ing for patient injury severity. Using a frequency procedure
for principal diagnosis, we identified the top 3 injury
diagnoses among injured children as fractures, intracranial
injury, and internal injury. Prevalence and associated
mortality of each injury type by site of care were derived.
Finally, rates of procedure performed by diagnosis and site
of care were analyzed.
Significance for all outcomes data was confirmed using
v2analysis for categorical variables and analysis of variance
(ANOVA) to compare means. Variables showing P values
less than or equal to .05 were considered significantly
different. Analysis was completed using SAS statistical
software version 8 (SAS Institute, Cary, NC).
Among 79,673 pediatric injury cases (weighted equiva-
lent), mean age was 12.2 F 6.2 years and ISS was 7.4 F
7.6. Table 1 shows the age, sex distribution, mean ISS, and
total hospital charges for injured pediatric patients as well
as the 0- to 10-year ISS N15 subgroup. As expected,
this subgroup had higher total charges, ISS, and lower age.
2.2. Trauma triage
A large percentage of injured patients received care
outside of children’s hospitals. Seventy-three percent of the
0-10 y, ISS N15 All patients
Mean ISS (SD)
3.9 F 5.0 y
U 38%; h 62%
19.5 F 9.6
$24,435 F 59,015
12.2 F 6.2 y
U 31%; h 69%
7.4 F 7.6
$15,383 F 30,494
J.C. Densmore et al.94
youngest, most seriously injured patients and 89% of all
patients fell into this category. That the predominance of
care is provided outside the children’s hospitals reflects
triage biases and limited access in some states. The 16%
difference between groups is a measure of the actual
adjustment of the current triage system to the youngest
and most seriously injured patients. Table 2 shows
distribution of care by NACHRI type with bchildren’s
unitQ representing a floor, wing, or designated children’s
area within an adult general hospital.
Table 3 demonstrates ISS scores among site type within
each group. In both groups, there was no significant
difference in ISS between children’s hospitals and adult
hospitals. Children’s unit mean ISS was higher than
children’s hospitals in the 0- to 10-year ISS N15 group
(20.0 F 10.6 vs 18.9 F 9.3, P = .003). The physiological
significance of the observed statistical difference would
appear to be minimal; however, this difference prompted
further regression analysis to control for ISS as a covariate.
Considering all patients, we found that children’s units were
different than both other sites (P = .0001), although there
was no significant difference between children’s hospitals
and adult hospitals.
2.3. Outcomes by site of care
Mortality, frequency of length of stay of higher than the
90th percentile, and frequency of total hospital charges
higher than the 50th percentile were all significantly greater
in children’s units and adult hospitals when compared to
children’s hospitals. Using these outcome measures, we
found the children’s unit in an adult hospital to have the
least desirable outcomes. This observation held in the
youngest, most seriously injured subgroup as well. How-
ever, children’s units have patients with statistically higher
ISS than other sites.
To further refine this analysis, we completed a linear
regression of outcomes by site of care controlling for ISS
differences. In the category of all pediatric patients,
outcomes represented in Table 4B were found to be
significant (P V.05). In the 0- to 10-year ISS N15 subgroup,
injury severity–adjusted mortality in adult hospitals and
children’s units was significantly higher than in children’s
hospitals (P V .007). Severity-adjusted length of stay and
total charges were significantly higher in children’s units
than in adult hospitals or children’s hospitals (P V .01). In
these 2 outcomes after injury severity adjustment, adult
hospitals and children’s hospitals were not found to be
2.4. Injury patterns and mortality by site of care
Fractures, intracranial injury, and internal injury were the
most frequent diagnoses among all pediatric injury patients.
Table 5A and B shows where each injury type was treated.
Interestingly, intracranial injury, associated with poorer
outcomes in pediatric trauma, did not affect site of care.
This was also the case for internal injuries, the surgical
treatment for which has been historically different between
Mortality by injury type and site of care is shown in
Table 6A and B. For all injury types in both groups, the
Site of care (NACHRI designation)
0-10 y, ISS N15 (%) All patients (%)
Injury severity score by site of care
0-10 y, ISS N15 All patients
18.9 F 9.3*
20.0 F 10.6*
19.4 F 9.4
7.0 F 6.6**
8.4 F 8.7**,***
7.0 F 7.3***
* ANOVA, P = .0003.
** ANOVA, P = .0001.
*** ANOVA, P = .0001.
Outcomes by site of care
z8 d (%)
A. 0-10 y, ISS N15
Children’s hospital 4.9
P b .0001, v2P b .0001, v2P b .0001, v2
B. All patients
Children’s hospital 0.9
P b .0001, v2P b .0001, v2P b .0001, v2
Injury type by site of care
A. 0-10 y, ISS N15
B. All patients
Outcomes and delivery of care in pediatric injury 95
lowest mortality rates for the most common diagnoses
occurred in a children’s hospital.
As injury systems continue to evolve, benchmarks for
their efficacy are becoming ever more important. The data
presented here from a large administrative database indicate
that in the year 2000, 79,673 pediatric injury patients were
delivered to 3 basic hospital types in the United States.
Nearly 90% of these children received care outside of
children’s hospitals. With regard to the youngest, most
seriously injured patients, the present health care system
directs an additional 16%, but by no means all of these
children, to dedicated children’s facilities. Is this level of
increased triage to the children’s hospital enough?
To answer this question requires a second question: are
outcomes for these children improved in children’s
hospitals and by how much? In this study, measures of
mortality, length of stay, and total charges were uniformly
higher in adult vs children’s hospitals. Interpretation of the
children’s unit outcomes data is tempered by statistically
higher ISS scores. After controlling for the variance in ISS
at each site of care, mortality, length of stay, and charges
are significantly higher in adult hospitals and highest in
children’s units. In the youngest, most severely injured
subgroup, injury severity–adjusted mortality follows the
same pattern with significant differences. Interestingly,
severity-adjusted length of stay and total charges in this
subgroup demonstrate a convergence of outcomes between
adult and children’s hospitals and significant difference
when compared to children’s units. Further characterization
of the geographic location of children’s units, their
capabilities, and proximity to other hospital types may
be of interest to direct care for young, severely injured
patients. Hospitals characterized as children’s units may
have expertise as simple as a dedicated pediatric floor or
as sophisticated as a bhospital within a hospital.Q We
initially hypothesized that some pediatric expertise (eg,
children’s units) would provide improved outcomes relative
to adult-oriented care institutions when caring for injured
children. The reported data do not support this hypothesis
as measured outcomes were worse in children’s units
than in adult hospitals. Alternatively, children’s units may
serve as a site of transfer from adult hospitals or preferred
triage when barriers exist to acquiring care in a children’s
hospital. Such barriers may be economic, geographic,
social, or dictated by the medical condition of the patient.
Unfortunately, the 2000 KID does not allow identifica-
tion of cases of transfer or distance between sites of care.
However, the increased ISS in children’s units relative
to other hospital types suggests that medical condition
and geographic factors may necessitate children’s units to
act as referral centers in the current national pediatric
Previous investigators have argued that mortality for all
pediatric trauma is low and is indistinguishable between
adult centers and pediatric centers [8,14,15]. Although
this appears correct in single state or institutional studies,
the improved power of an administrative database and
subgroup analysis shows significant differences at the
national level in contemporary US pediatric trauma care.
The major limitation of this database is the limited depth of
clinical information. Nevertheless, from a systems stand-
point, the data indicate that the majority of the youngest
and most severely injured children receive care in hospitals
with poorer outcomes. Young children with intracranial
injury and ISS of greater than 15 are no exception to this
rule, despite well-documented improvement in clinical
morbidity and mortality in a children’s specific environ-
The geographic disparity and limited number of pediatric
surgeons, specialists, and children’s hospitals have been
cited as the reason for the current pediatric trauma care
schema in the United States [8,14,20,27,28]. In contrast,
Segui-Gomez et al  demonstrated that bvery severely
injured children were being treated in nontrauma facilities,
even in states with pediatric trauma designation systems in
place.Q There appears to be substantial room for improve-
ment in current pediatric trauma triage systems, and this
study suggests that significantly improved outcomes may
result. Alternatives including multitiered systems have been
proposed [19,29]. These data indicate that some multitier
triage occurs presently, but we conclude it is inadequate and
that broad examination of pediatric trauma systems in the
United States is warranted.
 Web-Based Injury Statistics Query and Reporting System (WIS-
QARS). Centers for Disease Control and Prevention National Center
for Injury Prevention and Control, 2003 [Accessed 3/02/2005]
 Resources for optimal care of the injured patient. Chicago (Ill)7
American College of Surgeons; 1999.
Mortality differences for injury types by site of care
A. 0-10 y, ISS N15
Children’s hospital 4.7
P = .003, v2
P = .03, v2
P = .03, v2
B. All patients
Children’s hospital 0.7
P b .0001, v2P b .0001, v2P b .0001, v2
J.C. Densmore et al.96
 Tepas III JJ, Wesson DE, Harris BH. Evaluation and management of
the injured child. American College of Surgeons Committee on
Trauma. Bull Am Coll Surg 1995;80:36-9.
 Hulka F, Mullins RJ, Mann NC, et al. Influence of a statewide trauma
system on pediatric hospitalization and outcome. J Trauma 1997;
 Vavilala MS, Cummings P, Sharar SR, et al. Association of hospital
trauma designation with admission patterns of injured children.
J Trauma 2004;57:119-24 [discussion 124].
 Ehrlich PF, Ortega J, Mucha PM. The need for a statewide pediatric
trauma program. W V Med J 2002;98:66-9.
 Farrell LS, Hannan EL, Cooper A. Severity of injury and mortality
associated with pediatric blunt injuries: hospitals with pediatric
intensive care units versus other hospitals. Pediatr Crit Care Med
 Fortune JB, Sanchez J, Graca L, et al. A pediatric trauma center
without a pediatric surgeon: a four-year outcome analysis. J Trauma
1992;33:130-7 [discussion 137-139].
 Knudson MM, Shagoury C, Lewis FR. Can adult trauma surgeons
care for injured children? J Trauma 1992;32:729-37 [discussion
 Bensard DD, McIntyre Jr RC, Moore EE, et al. A critical analysis of
acutely injured children managed in an adult level I trauma center.
J Pediatr Surg 1994;29:11-8.
 Eichelberger MR, Bowman LM, Sacco WJ, et al. Trauma score versus
revised trauma score in TRISS to predict outcome in children with
blunt trauma. Ann Emerg Med 1989;18:939-42.
 Kaufmann CR, Maier RV, Kaufmann EJ, et al. Validity of applying
adult TRISS analysis to injured children. J Trauma 1991;31:691-7
 Potoka DA, Schall LC, Gardner MJ, et al. Impact of pediatric trauma
centers on mortality in a statewide system. J Trauma 2000;49:237-45.
 D’Amelio LF, Hammond JS, Thomasseau J, et al. bAdultQ trauma
surgeons with pediatric commitment: a logical solution to the pediatric
trauma manpower problem. Am Surg 1995;61:968-74.
 Potoka DA, Schall LC, Ford HR. Improved functional outcome for
severely injured children treated at pediatric trauma centers. J Trauma
2001;51:824-32 [discussion 824-32].
 Osler TM, Vane DW, Tepas JJ, et al. Do pediatric trauma centers have
better survival rates than adult trauma centers? An examination of the
National Pediatric Trauma Registry. J Trauma 2001;50:96-101.
 Segui-Gomez M, Chang DC, Paidas CN, et al. Pediatric trauma care:
an overview of pediatric trauma systems and their practices in 18 US
States. J Pediatr Surg 2003;38:1162-9.
 MacKenzie EJ, Morris Jr JA, de Lissovoy GV, et al. Acute hospital
costs of pediatric trauma in the United States: how much and who
pays? J Pediatr Surg 1990;25:970-6.
 Engum SA, Mitchell MK, Scherer LR, et al. Prehospital triage in the
injured pediatric patient. J Pediatr Surg 2000;35:82-7.
 Vernon DD, Bolte RG, Scaife E, et al. Alternative model for a
pediatric trauma center: efficient use of physician manpower at a
freestanding children’s hospital. Pediatr Emerg Care 2005;21:18-22.
 Moront ML, Gotschall CS, Eichelberger MR. Helicopter transport of
injured children: system effectiveness and triage criteria. J Pediatr
Surg 1996;31:1183-6 [discussion 1187-8].
 Eckstein M, Jantos T, Kelly N, et al. Helicopter transport of pediatric
trauma patients in an urban emergency medical services system: a
critical analysis. J Trauma 2002;53:340-4.
 MacKenzie EJ, Shapiro S, Eastham JN. The Abbreviated Injury Scale
and Injury Severity Score. Levels of inter- and intrarater reliability.
Med Care 1985;23:823-35.
 MacKenzie EJ, Steinwachs DM, Shankar B. Classifying trauma
severity based on hospital discharge diagnoses. Validation of an ICD-
9CM to AIS-85 conversion table. Med Care 1989;27:412-22.
 Acierno SP, Jurkovich GJ, Nathens AB. Is pediatric trauma still a
surgical disease? Patterns of emergent operative intervention in the
injured child. J Trauma 2004;56:960-4 [discussion 965-6].
 Nakayama DK, Copes WS, Sacco WJ. The effect of patient age upon
survival in pediatric trauma. J Trauma 1991;31:1521-6.
 Rhodes M, Smith S, Boorse D. Pediatric trauma patients in an dadultT
trauma center. J Trauma 1993;35:384-92 [discussion 383-92].
 Watson RS. Location, location, location: regionalization and outcome
in pediatric critical care. Curr Opin Crit Care 2002;8:344-8.
 Barnett SJ, Saltzman DA. Efficacy of pediatric-specific trauma
centers. Pediatr Crit Care Med 2004;5:93-4.
Randall Burd, MD (New Brunswick, NJ): I enjoyed your
presentation. I agree with the basic conclusions of your
study but have a few questions about how you
Did you consider the trauma center designation of the
hospitals? Among the nonpediatric hospitals, there is a
mix of hospitals that have no trauma designation as well
as hospitals that have a trauma designation. If you added
trauma designation as a variable in your analysis, that
might have impacted your results.
My second question relates to the use of ISS alone for
risk stratification. ISS as an isolated risk stratifier has a
pretty good ability to discriminate mortality but it really
does not perform well across the entire probability range
of mortality, that is, it has poor calibration when used in
isolation. How stringently did you consider the adequacy
of your risk stratification method?
Again, I think the conclusions of your study are very
important but I think we need to be very careful when
presenting this kind of data since we really need to give
our adult colleagues a fair shot too.
John Densmore, MD (response): Those are excellent
questions. Thank you.
With regard to trauma center designation, that is
outside the scope of this database and the information
that is available. I think the strength of our study is that it
is a snapshot in a single year of the large number of
injured children and where they wound up and the
outcomes from that. However, we were unable to resolve
which children actually wound up at a pediatric trauma
center versus a nontrauma center.
With regard to your second question, we did in fact
risk stratify our outcomes using injury severity con-
trolled as a covariate and all of the P values that I
showed you used that linear regression modeling to
David Mooney, MD (Boston, MA): Just one word of
caution. In the KID 2000 database there is an up to
4 times variation in inpatient mortality rates between the
different states. Some of this might be because of
variation in local regulations regarding the ability of
prehospital providers to declare death in the field. For
example, in our state children who are dying are brought
Outcomes and delivery of care in pediatric injury97
to the nearest hospital to be declared, which is typically Download full-text
an adult hospital. This creates an inaccurate measure of
that facility s care. So similar to the previous comment,
you need to be very careful about the conclusions based
upon the data set that you are using.
John Densmore, MD (response): Thank you for your
comment. I would agree that one of the shortcomings
of this database is that we are not able to determine
interfacility transfers, so we are unable to capture how
often that occurs. However, our mortality rates that we
demonstrate are significant and they actually match with
previously discovered estimates.
Jonathan Groner, MD (Columbus, OH): At least according
to what is in the abstract, the patients at the adult general
hospitals had a higher ISS and also had a higher
mortality, which is not exactly surprising. Since most
pediatric trauma patients do, in fact, survive their
injuries, I think that mortality is not the best indicator
of how we are doing. Is there any opportunity to look at
other indicators such as functional independence mea-
surement as an indicator of how well kids do at adult
hospitals versus pediatric hospitals?
John Densmore, MD (response): I would agree with you
that mortality is a crude indicator of quality of care
delivered. This particular data set allowed us to
demonstrate significance, whereas previous data sets
struggled with that. The limitations of this administrative
database are that we can simply look at very simple
outcomes like mortality, length of stay, and charges.
Functional-type studies are certainly relevant; however,
are outside of the scope of this database.
J.C. Densmore et al.98