Assessment of Adverse Events in the Demise of Pediatric Burn ...

The Journal of TRAUMA Injury, Infection, and Critical Care
Assessment of Adverse Events in the Demise of Pediatric
Burn Patients
Dennis C. Gore, MD, Hal K. Hawkins, MD, David L. Chinkes, PhD, Dai H. Chung, MD,
Arthur P. Sanford, MD, David N. Herndon, MD, and Steven E. Wolf, MD
Background: Given the contention
that survival is to be expected from even
the most severely burned child, then, intuitively,
at least some pediatric burn victims
die because of suboptimal care. The
purpose of this study is to assess the
impact of any adverse events that may
have contributed to the death of burned
children.
Methods: Four surgeons with specialty
training in pediatric burn care reviewed
the clinical course and autopsy
findings of 71 burned children who died
after admission to a burn center during a
10-year interval. Reviewers were asked to
determine the predominant factor or factors
contributing to each child’s demise
and to assess the significance of any deviations
from optimal care.
Results: For the 10 years under review,
overall mortality for all pediatric
burns was 2.4%. Of these deaths, 25%
had burns encompassing less than 50%
body surface area. The reviewers identified
lung damage as the most frequent
cause of death, which was deemed largely
unpreventable. Conversely, hypovolemia
related to inadequate prehospital fluid resuscitation
and failure to obtain and maintain
a patent airway were considered the
second and third most common factors in
a child’s death and deemed preventable
under ideal circumstances.
Conclusions: This review implies
that deficiencies in health care contribute
to the demise of many burned children.
The most notable areas for improvement
are in fluid resuscitation and airway control.
This suggests that quality assurance
and educational initiatives to improve
these aspects of care may have the greatest
impact on further improving survival of
burned children.
Key Words: Patient safety, Quality
assurance, Medical error.
J Trauma. 2007;63:814–818.
Historically the main determinants of mortality from burn
injury have been age, the extent of burn, and the presence
or absence of inhalation injury. 1–3 Yet with the
remarkable advances in burn care and the associated improvements
in survival, the “traditional” determinants of skin
and lung damage may no longer have a dominant influence
on survival of the young. 2 This contention is supported by
Sheridan et al., 4 who examined the clinical course of burned
children who subsequently died at the Shriners Burns Hospital
in Boston. By comparing patients cared for from 1974 to
1980 with those patients seen between 1991 and 1997, and
stratified by burn size, these authors concluded that the
chance of survival after a burn has greatly improved for
children to the extent that even the very young and severely
burned child should survive. Dr. Saffle, 5 a recent president of
the American Burn Association, echoed this sentiment, commenting
that for adolescents and young adults, almost no
burn is too extensive to preclude recovery. Demonstrating
that such an ambitious goal is possible, we recently reviewed
Submitted for publication February 1, 2006.
Accepted for publication April 13, 2007.
Copyright © 2007 by Lippincott Williams & Wilkins
From the Shriners Hospitals for Children (H.K.H., D.L.C., A.P.S.,
D.N.H.) and Department of Surgery (D.C.G., D.H.C., D.N.H.), The University
of Texas Medical Branch, Galveston, and the Department of Surgery
(S.E.W.), University of Texas Health Science Center, San Antonio, Texas.
Address for reprints: Dennis C. Gore, MD, Department of Surgery, The
University of Texas Medical Branch, 301 University Boulevard, Galveston,
TX 77555-1172; email: dcgore@utmb.edu.
DOI: 10.1097/TA.0b013e31811f3574
the outcome of pediatric patients with burns of greater than
80% of their body surface. 6 Amazingly, more than 60% of
these severely burned victims survived. Yet, regardless of
these therapeutic advancements and miraculous survivals,
victims of burn injury still die. At our institution alone there
were 18 deaths during a past year. Because these publications
imply that survival is to be expected, one must conclude that
at least some pediatric burn victims die because either the
correct intervention was not performed, was not performed in
a timely fashion, or was not performed correctly.
The purpose of this study was to appraise the major
determinants of death in burned children and assess the impact
of any adverse event that may have contributed to the
demise of these children. It is hoped that this review and
analysis may identify those factors, which are potentially
correctable and thereby guide future maneuvers to improve
survival.
PATIENTS AND METHODS
Records
Of a total of 3,005 admissions to the Shriners Hospitals
for Children in Galveston, TX from January 1992 through
December 2001, there were 72 deaths. Except for one child
whose family refused autopsy on religious edict, autopsies
were performed on all (71) children. All but four of these
autopsies were performed by a single pathologist (H.K.H.).
Three of these autopsies were performed at the adjacent
University Pathology Department of the University of Texas
Medical Branch with the findings recorded in the identical
format prescribed by the specialty pathologist. The single
814 October 2007

Adverse Events in Burn Care
remaining autopsy was performed by the county coroner. In
addition to the autopsy findings, the pathologist also reviewed
each patient’s hospital records and reported a summary
of the clinical course for each patient. These clinical
case summaries were available for all 72 deaths. For all
admissions to the Shriners hospital during this 10-year interval,
patient characteristics and demographics were available
on computerized hospital records.
Review Process
After approval from the Institutional Review Board, autopsies
and clinical course records were reviewed by four
surgeons, all of whom have experience in burn care; three
having completed postresidency fellowships in burn care at a
Shriners Hospitals for Children, the other with fellowship
training in pediatric surgery and with more than 10 years
experience as a physician consultant at the Shriners Hospitals
for Children. Using a standardized form, each reviewer was
asked to note the probable cause of death and any associated
factors contributing to death. Reviewers were also asked to
scale the contribution of any deficiencies in care to the cause
of death, from 1 (denoting no error) to 4 (denoting a significant
impact of negligence on the child’s death). Also scaled
from 1 to 4 by each reviewer were (1) the percent chance that
a misadventure caused death and (2) the possibility of surviving
to discharge had the error not occurred. For grading
items 1 and 2, reviewers agreed on the following scale: 1
0% to 15%, 2 16% to 50%, 3 51% to 84%, and 4 85%
to 100%. Each reviewer was screened from the decisions of
the other reviewers.
Statistics
Patient characteristics between survivors and deaths
were compared using a Student’s independent t test. Differences
in the incidence of inhalation injury between survivors
and deaths were assessed using 2 . Median values between
raters were used to report the incidence of an event. Median
and mean values corresponded closely for all variables, so mean
values are not reported. Inter-rater agreement was assessed using
simultaneous -statistics for variables that had an incidence of at
least 10%. values of greater than 0.7 were considered significant
and indicative of strong inter-rater agreement.
RESULTS
The overall mortality for all hospital admissions during
the 10 years of review was 2.4%. Comparison of patient
characteristics between survivors and deaths is shown in
Table 1 with deaths having a significantly greater extent of
burn and a higher incidence of inhalation injury. In Table 2,
patient’s characteristics are stratified by their severity of burn
demonstrating a significant association between a greater
surface area burn to increasing mortality. Of note, there were
19 deaths (mortality rate 0.7%) of patients admitted to the
hospital with burns encompassing 50% of their body surface
area.
Table 1 Patient Characteristics
Survivors
Deaths
N 2,933 72
Age (yr) 6.4 2.1 5.7 1.8
Body weight (kg) 27.6 3.9 26.1 5.1
TBSA burn (%) 26 5 66 3*
TBSA third burn (%) 14 4 60 3*
Inhalation injury (%) 16 60 †
Values are mean SEM.
* p 0.01 comparison by Student’s t test.
† p 0.05 comparison by 2 .
TBSA, total body surface area.
Table 2 Deaths Stratified by Severity of Burn
Burns 50%
(n 2,602)
Burns 50; 80%
(n 259)
Burns 80%
(n 72)
Number of
19 31 22
deaths
Mortality (%) 0.7 12* 30*
Age (yr) 5.4 1.4 7.1 2.8 6.8 3.0
TBSA burn (%) 40 3 71 2* 91 2*
Inhalation injury,
n (%)
8 (42) 19 (61) 16 (73) †
Values are mean SEM.
* p 0.01 comparison by Student’s t test.
† p 0.06 comparison by 2 .
Table 3 Primary Causes of Death and Factors
Contributing to Death
Adverse Event
Primary Cause of
Death ( )
Contributing
Factor ()
Diffuse alveolar lung damage (%) 25 (0.32) 51 (0.32)
Hypovolemia/inadequate
16 (0.24) 23 (0.32)
resuscitation (%)
Airway loss/aspiration (%) 15 (0.48) 22 (0.47)
Pneumonia (%) 13 (0.24) 23 (0.34)
Burn wound sepsis (%) 13 (0.31) 33 (0.53)
Anoxia (%) 7 20 (0.53)
Drug reaction (%) 1
Data are median values.
-statistics for inter-rater agreement 0.7 considered
significant.
The reviewers determined that diffuse alveolar damage
of the lungs was the most prevalent primary cause of death
(Table 3). Hypovolemia and inadequate emergency resuscitation
were considered the second most common cause of
death. Loss of airway or aspiration, pneumonia, and burn
wound sepsis were designated as the third, fourth, and fifth
most common primary causes of death, respectively. Death
related to anoxic brain damage occurring at the time of the
fire was noted as the cause of death in five children; the sixth
most frequent cause of death. One child died from an apparent
anaphylactic drug reaction. For the five most frequent
causes of death, there was no significant agreement between
reviewers with a maximum score for airway or aspiration at
Volume 63 • Number 4 815

The Journal of TRAUMA Injury, Infection, and Critical Care
Table 4 Significance of Medical Error to Death
Cause of Death
Average Score
Diffuse alveolar lung damage 1.7
Hypovolemia/inadequate resuscitation 3.0
Airway loss/aspiration 3.5
Pneumonia 2.3
Burn wound sepsis 2.5
Anoxia 1.5
Drug reaction 3.9
Cumulative score 2.4 ( 0.06).
Scale: 1 0%–15%; 2 16%–50%; 3 51%–84%; 4
85%–100%.
-statistics for inter-rater agreement 0.7 considered significant.
Table 5 Inter-Rater Variability for Percent Chance
that Error Caused Death
Reviewers
Scores
1 2 3 4
1 51 (72) 16 (22) 29 (41) 25 (35)
2 5 (7) 12 (17) 1 (1) 17 (24)
3 7 (10) 7 (10) 5 (7) 8 (11)
4 8 (11) 36 (51) 36 (51) 21 (29)
Mean SD 1.61 1.06 2.89 1.26 2.37 1.83 2.35 1.24
Values presented as no. patients (%) for each score given by a
reviewer.
Cumulative score 2.3 ( 0.06).
Table 6 Inter-Rater Variability for Percent Chance of
Surviving had Error not Occurred
Reviewers
Scores
1 2 3 4
1 50 (71) 21 (29) 27 (38) 31 (44)
2 6 (8) 23 (32) 6 (8) 21 (29)
3 6 (8) 13 (18) 13 (18) 4 (6)
4 9 (13) 14 (20) 25 (35) 15 (21)
Mean SD 1.61 1.06 2.89 1.10 2.17 1.73 2.04 1.16
Values presented as no. patients (%) for each score given by a
reviewer.
Cumulative score 2.0 ( 0.06).
0.48. As to whether adverse events contributed to death,
reviewers noted diffuse alveolar lung damage as the most
common contributing factor to death (Table 3). Burn wound
sepsis, pneumonia, hypovolemia or inadequate resuscitation,
and airway loss or aspiration completed the five most commonly
noted factors contributing to a child’s demise. There
was no significant inter-rater agreement in designating an
adverse contributing factor with a maximum score for burn
wound sepsis at 0.53.
Reviewer’s assessment as to the significance of any error
in each child’s death is shown in Table 4. The cumulative
score for all patients on a1to4scale was 2.4. Thus, in the
aggregate opinion of the reviewers, suboptimal care contributed
significantly in the demise of approximately 50% of the
burn victims. Stratifying the significance of any deficiency in
care by cause of death, drug reaction, airway loss or aspiration,
and hypovolemia or inadequate resuscitation was considered
by the reviewers to be frequently associated with
negligence. In contrast, error was not deemed of much importance
to either anoxic brain injury or diffuse alveolar lung
damage. The impact of a medical misadventure on the demise
of each child was also assessed by the reviewer’s determination
of a percentage chance that error contributed to a child’s
death (Table 5). Also scaled 1 to 4, physician reviewers
assessed a cumulative score for this impact of error at 2.3,
translating into the reviewer’s perception that medical misadventure
contributed to a child’s demise is about 50% of all
burn victims. There was, however, a wide range of disagreement
as to the impact of error with a score of only 0.06.
Another determination as to the significance of medical error
was assessed by the reviewers as the percentage chance that
the child would have survived had the error not occurred.
Again scaled 1 to 4, reviewers assigned this an overall 2.0
score, meaning approximately one-third of all deaths could
have been prevented had a deficiency in care not occurred
(Table 6). Again inter-rater variability was high with a
score of 0.06.
Although there was a wide range of disagreement concerning
the incidence and severity of medical error, this
disparity of opinion did not seem to arise from differences as
to the seriousness of an adverse event because the range of
values that raters gave for significance of error was usually
relatively narrow. In general, there were only systematic
differences between raters. For example, rater 3 assessed the
significance of error for pneumonia as 1 (i.e. no error)
whereas the other reviewers typically gave ratings of 2 or 3.
Rater 3 also assessed any deficiencies in care for burn wound
sepsis as 1 in about half the cases, whereas the other reviewers
almost universally assigned burn wound sepsis to a rating
of 2 or 3. Reviewer 4 often rated the significance of medical
error for anoxia as a3to4suggesting that in his opinion
something could have been performed in the initial resuscitation
to preserve brain function. In contrast, the other reviewers
rated the significance of error for anoxia as a1or2
indicative of their opinion as to the futility of any medical
intervention to forego the anoxic brain death. Opinions also
varied as to the significance of error with diffuse alveolar
lung damage. For example, rater 2 consistently assigned a
significance of error of 3 for children dying primarily from
lung injury and thereby conveying his opinion that something
might have been performed to forego pulmonary failure.
Other reviewers consistently gave diffuse alveolar damage an
error significance score of 1.
DISCUSSION
The overall mortality for this 10-year review was only
2.4%; far less than that reported from the national burn
registry. 7 As to be expected, there was a significantly greater
extent of burn injury and a higher incidence of inhalation
816 October 2007

Adverse Events in Burn Care
injury in those dying at our institution. However, the mean
value for all deaths was a body surface area burn of only 66%
and an incidence of inhalation injury of only 60%. Furthermore,
19 children died with burns of less than 50% body
surface area of which only 8 were noted to have an inhalation
injury. These values demonstrate that some deaths were the
result of neither devastating skin nor lung damage. This
shows that, regardless of the extent of burns, something other
than the actual injury must have contributed to the demise of
many of these children. One probable assumption would be
that a deficit in care, either as an omission or delay in
providing appropriate care or possibly inappropriate care,
may be a factor in some deaths.
In this review, experts in pediatric burn care identified
lung damage as the most frequent cause of death in these
injured children. In general, these reviewers thought that
death from pulmonary failure, a consequence predominantly
of either the initial smoke injury or as sequelae from sepsis,
was largely not preventable. Conversely, these physician reviewers
in aggregate considered hypovolemia or inadequate
resuscitation as the second, and failure of airway control or
aspiration as the third, most common cause of death. It was
their opinion that such deaths were likely preventable. Death
from infection, noted separately as pneumonia or burn wound
sepsis, was also considered a frequent primary cause of death.
Reviewers generally scaled these infectious deaths as potentially
preventable, but with a varied opinion as to the responsibility
of the medical community regarding infection control
versus a feeling of pessimism for any intervention in an
individual patient with severe sepsis.
The main goal of this study was to categorize failures in
the care of injured children and thereby provide a guide to
direct future quality care initiatives and ultimately improve
patient care and survival. One presumption is to focus on
those failures deemed most preventable; for example, an
educational initiative aimed at optimizing fluid resuscitation
and airway control of a severely burned child. With the
findings of this study demonstrating that hypovolemia and
loss of airway, a common facet in the death of some children,
such a quality assurance directive would intuitively be beneficial.
However, an acutely burned child is relatively infrequent
for even an experienced emergency room physician,
thus any postevent education is unlikely to have much impact
in the future care of burn children as a rarely repeated event. 8
Likewise, pre-emptive education such as that provided by the
Advanced Burn and Trauma Life Support courses designed to
reduce errors in resuscitation and airway management may
fail to have a prolonged impact with possibly many years
preceding an occurrence. As an alternative to a traditional
continuing medical education initiative, advances in information
technology, such as telemedicine, 9,10 may provide an
avenue for improving the emergency care and transfer of burn
victims by allowing closer, more timely interaction of a burn
expert to the initial care of an injured child. A key aspect of
any such interaction is the prompt contact from the initial
emergency room to the burn specialist and the ready availability
of the technology. Furthermore, a concentrated effort
at education and training for personnel dedicated to repeatedly
providing safe transport would also likely be beneficial.
Likewise, many physicians who work at a burn center are
residents in training who lack experience in fluid resuscitation
and airway intubation, especially for severely burned
children. Emerging facilities such as the Human Patient
Simulator 11 may greatly aid in resident physician training by
providing a manner for skills acquisition before such a physician
is thrust into a life-or-death situation. In contrast to
emergency resuscitation, death from pulmonary failure was
generally considered nonpreventable. Yet, considering the
prominent contribution of pulmonary damage to the demise
of these patients, any improvements in this regard may have
a substantial impact on outcome. However, considering
health care negligence was not deemed to have a substantive
influence regarding these respiratory deaths, therapeutic advances,
not educational, quality assurance initiatives, are
likely to be a more viable avenue for improving outcome
from this cause of death.
There was no significant agreement between reviewers
as to either the cause of death or the relevance of medical
error in the demise of these children, despite having both
autopsy records and a summary of their clinical course. Such
a discrepancy between well-trained experts in a clearly defined
speciality of care in a single specialized hospital setting
emphasizes the difficulty and subjectivity in deciding on
those factors contributing to the death of these injured children.
Much of this disparity resulted from our inability to
distinguish the relative contribution to a child’s demise from
the initial injury versus any exacerbation of that injury related
to an inadequate or delayed resuscitation or damage related to
sepsis. Furthermore, because children have such an extensive
physiologic reserve and the vast majority have excellent cardiac
function, the progression through multiple organ failure
to death was often prolonged. This common scenario of
prolonged, multiorgan failure often precluded a clear interpretation
of the exact inciting factor in the child’s death.
Autopsy findings did little to relieve this uncertainty because
the postmortem examination almost universally showed a
combination of extensive inflammatory damage within the
lung, bacterial and fungal infiltration within the lungs (i.e.
pneumonia), and infection or contamination within any residual
wound. Thus, autopsies were not particularly helpful in
identifying either a predisposing cause of death or any associated
medical error. Because autopsies are known to disagree
with the clinical diagnosis as much as 40% of the
time, 12,13 any study with the intent of determining the nature
and preventability of death would be expected to have substantial
error without inclusion of the autopsy findings. Yet,
this disparity between the pre- and postmortem cause of death
is largely evident in review of adult patients. Furthermore,
this review of injured children circumvents any variable of a
“natural” death rate because the vast majority of these burn
Volume 63 • Number 4 817

The Journal of TRAUMA Injury, Infection, and Critical Care
victims were in good health before their injury, which negates
much of the uncertainty by a reviewer as to the influence of
the disease process versus error in the demise of these children.
Therefore, a finding of this study suggests that, at least
in regard to injured children, accurate records of a child’s
clinical care, and not the autopsy, are of utmost importance in
assessing any medical misadventures.
REFERENCES
1. Muller MJ, Pegg SP, Rule MR. Determinants of death following
burn injury. Br J Surg. 2001;88:583–587.
2. Ryan CM, Schoenfeld DA, Thorpe WP, et al. Objective estimates of
the probability of death from burn injuries. N Engl J Med. 1998;
338:362–366.
3. Smith DL, Cairns BA, Ramadan F, et al. Effect of inhalation injury,
burn size, and age on mortality: a study of 1447 consecutive burn
patients. J Trauma. 1994;37:655–659.
4. Sheridan RL, Remensnyder JP, Schnitzer JJ, et al. Current
expectations for survival in pediatric burns. Arch Pediatr Adolesc
Med. 2000;154:245–249.
5. Saffle JR. Predicting outcomes of burns. N Engl J Med. 1998;
338:387–388.
6. Spies M, Herndon DN, Rosenblatt JI, et al. Prediction of
mortality from catastrophic burns in children. Lancet. 2003;
361:989–994.
7. Saffle JR, Davis B, Williams P. Recent outcomes in the treatment
of burn injury in the United States: a report from the American
Burn Association Patient Registry. J Burn Care Rehabil. 1995;
16:219–232.
8. Brigham PA, McLoughlin E. Burn incidence and medical care use in
the United States: estimates, trends, and data sources. J Burn Care
Rehabil. 1996;17:95–107.
9. Aucar J, Granchi T, Liscum K, et al. Is regionalization of trauma
care using telemedicine feasible and desirable? Am J Surg. 2000;
180:535–539.
10. Tachakra S, Jaye P, Bak J, et al. Supervising trauma life support by
telemedicine. J Telemed Telecare. 2000;6(Suppl 1):S7–S11.
11. Holcomb JB, Dumire RD, Crommett JW, et al. Evaluation of trauma
team performance using an advanced human patient simulator for
resuscitation training. J Trauma. 2002;52:1078–1086.
12. Goldman L, Sayson R, Robbins S, et al. The value of the autopsy in
three medical eras. N Engl J Med. 1983;308:1000–1005.
13. Mort TC, Yeston NS. The relationship of premortem diagnoses and
post mortem findings in a surgical intensive care unit. Crit Care
Med. 1999;27:299–303.
818 October 2007