Differences
in Trauma Care Among Pediatric Trauma Centers ByDon
K. Nakayama,
Wayne
S. Copes,
and William
and Nonpediatric Sacco
Pittsburgh, Pennsylvania and Bel-Air, Maryland l TO characterize
pediatric
trauma
care,
state
trauma
registry data from all designated trauma centers in Pennsylvania were divided into three categories, that from: (1) pediatric centers, (2) urban nonpediatric
centers, (3) and rural
nonpediatric centers. From October 1.1986 through September 30, 1889 (3 years), 4,615 patients less than 15 years old were admitted to 28 trauma centers in Pennsylvania. Nonpediatric centers cared for the majority of children (2,734, 59.2%). but the average number of children treated per nonpediatric institution (105.1 per year) was far fewer than the average treated in the pediatric centers (940.5). Pediatric trauma centers in the state treated a younger population (6.4 r 4.2 years, mean 2 SD) compared with urban and rural nonpediatric centers (8.4 + 4.2 and 8.1 f 4.3 years, respectively; P c .05). Pediatric centers received proportionately more children by transfer (56.2%). victims of falls (34.6%). pedestrian injuries (16.8%). and head and neck injuries (41.8%. all P < .05). Nonpediatric centers received children directly from the scene of injury more frequently than transferred from other hospitals. The male:female sex ratio in urban nonpediatric centers was significantly higher (70.1%. P c .05) than in the other two groups. Rural nonpediatric centers cared for a higher proportion of motor vehicle passengers (28.5%) and patients classified as “other” in the state registry, a category to which bicycle injuries are assigned (28.2%. P < .05). Mortality was highest in rural nonpediatric centers (6.2%). The death rate in pediatric centers and urban nonpediatric centers were similar (4.1%) and significantly lower (P < .05). Mortality from pedestrian injuries was higher in rural centers (15.1%); other injury mechanisms exhibited no regional differences in mortality. Using TRISS, I for rural centers (1.123) failed to reach statistical significance, whereas z for pediatric and urban nonpediatric centers reached significance (3.896 and 3.335, respectively). Using the probability of survival (P(s)) generated by TRISS analysis, no significant differences in survival were noted among the three groups of trauma centers when P(s) was stratified, although survival was slightly higher in pediatric centers for P(s) 1.3 and q.6. Important regional differences in pediatric trauma care exist, specifically major differences in patient age, injury, and referral source. Rural trauma centers have a higher pediatric mortality than urban centers, a possible result of differences in access to prehospital care. Copyright o 1992 by W.B. Saunders Company INDEX WORDS:
Pediatric trauma care.
0 The
NE FOURTH of injury victims are chiIdren, so pediatric care is a priority of all trauma centers. care of children with significant injuries often require special equipment and involve pediatric surgical specialists, and standards of pediatric trauma care have been published by the American College of Surgeons that address the unique requirements of children.’ In Pennsylvania a system of 28 accredited JournalofPediatric Surgery, Vol27, No 4 (April), 1992: pp 427-431
trauma centers has been developed to provide optimal care to the injured throughout the state; only the Children’s Hospital of Pittsburgh and the Children’s Hospital of Philadelphia have satisfied standards for designation as pediatric trauma centers. Therefore, many injured children receive care at general trauma centers. To characterize the care of injured children in both pediatric and nonpediatric trauma centers, we used data from the Pennsylvania trauma registry in an attempt to answer the following questions. How do trauma centers with pediatric designation differ from the many centers without special designation? How do centers differ in terms of the referral, transfer, patient age, injury type, and severity? What happens in areas where pediatric trauma centers don’t exist? Do regional differences affect mortality? MATERIALS
AND
METHODS
In 1985, the legislature of the Commonwealth
of Pennsylvania authorized the Pennsylvania Trauma Systems Foundation (PTSF) to formulate standards for the accreditation of hospitals as trauma centers. PTSF standards parallel published guidelines by the American College of Surgeons’ and the American College of Emergency Physicians.’ The PTSF confirms compliance with standards of trauma care through on-site surveys by out-of-state experts. Separate standards for pediatric trauma center designation, called Pediatric Regional Resource Trauma Centers, were also formulated. These were based on standards outlined by Ramenofsky and Morse,3 which are included with the guidelines of the American College of Surgeons. Pediatric trauma centers must demonstrate specific pediatric specialty facilities, equipment, personnel, training, research, and educational activities. Trauma centers without pediatric designation must have a volume of 600 trauma admissions annually; pediatric trauma centers must care for at least 150 pediatric trauma admissions annually. In this study, PTSF-designated trauma centers in the state were divided into three categories. “Pediatric trauma centers” refer to the two hospitals designated by PTSF as Pediatric Regional Resource Trauma Centers, the Children’s Hospital of Pittsburgh
From the Benedum Pediatric Trauma Program, the Department of Pediatric Surgery, the Children’s Hospital of Pittsburgh, the Department of Surgery, the University of Pittsburgh School of Medicine, Pittsburgh, PA, and Tri-Analylics, Inc, BeLAir MD. Date accepted: November I, 1991. Supported in part by the Pennsylvania Trauma Systems Foundation, the Claude Worthington Benedum Foundation, and the Health Services and Research Foundation, Pittsburgh, PA. Address reprint requests to Don I2 Nahayama, MD, Department of Pediatric Surgery, the Children S Hospital of Pittsburgh, 3705 Fifth Ave at DeSoto St, Pittsburgh, PA 15213. Copyright o I992 by WB. Saunders Company 0022-3468192/2704-0002$03.0010 427
NAKAYAMA,
428
and the Children’s Hospital of Philadelphia. Nonpediatric
centers
were divided into two groups: “urban centers,” those located in the
Pittsburgh and Philadelphia areas (“Pgh/Phl nonped”), regions in the immediate cachement area of the two pediatric centers, and “rural centers,” those located outside of the Pittsburgh and Philadelphia metropolitan areas (“other Penna”). PTSF maintains a trauma registry on all injured patients admitted to accredited trauma centers in the state satisfying the following criteria: all deaths; lengths of stay greater than 3 days; admissions to an intensive care unit; and all transfers into and out of the hospital. Gilliot et al have described the registry (termed the Pennsylvania Trauma Outcome Study) at length elsewhere.4 In brief, data elements collected are generally those of the Major Trauma Outcome Study (MTOS)? Other data items dealing with patient transfer, prehospital care, emergency department, and in-hospital phases of care are also collected. Tri-Analytics, Inc (Be1 Air, MD) receives data collected at the trauma centers, codes injury descriptions, and provides quarterly summaries of demographic, injury, and treatment data, quality assurance summaries, and outcome analyses based on the TRISS methodology,6 described in brief below. Inconsistent data, such as transposed dates of admission and discharge, are returned to the trauma center for revision. Trauma center registrars undergo training sessions on state trauma registry data collection to assure consistency in data abstraction. In addition, PTSF and Tri-Analytics, Inc. conduct periodic reviews of data collection submission on site at all trauma centers. The review matches registry data with data in hospital charts, reabstracted by a member of the state registry staff. Briefly, TRISS predicts the probability of survival (P(s)) for any patient from the patients’ Injury Severity Score (ISS, based upon the 1985 Revision of the Abbreviated Injury Scale),’ and the Revised Trauma Score (RTS) measured on arrival in the emergency department. A collection of P(s) for a given subset of patients (for example, all patients treated at a single hospital, or for the purposes of this report, a group of hospitals) gives thez statistic that reflects the difference between the actual number of survivors in a population subset and the expected number of survivors based on TRISS norms, which are generated from the entire MTOS group. When the actual number of survivors exceeds the number predicted from the baseline, z is positive; when it is less, .z is negative.
Absolute
values
of z exceeding
1.96
are
significant
(P < .OS). Far large samples,
small and clinically insignificant differences between actual and expected numbers of survivors can reach statistical significance. The W statistic provides perspective to the interpretation of significant z values. W represents the difference between the actual number of survivors and the expected number (predicted by TRISS) per 100 patients treated. The x2 statistic was used to compare frequencies and analysis of variance was used to compare differences between mean values. P < .05 was considered significant.
RESULTS
From October 1,1986 through September 30,1989 (3 years), 4,615 patients 14 years old and younger were admitted to PTSF-designated trauma centers in Pennsylvania, and had data entered into the state trauma registry. Table 1 summarizes the numbers of admissions by trauma center category (pediatric, nonpediatric in the Pittsburgh and Philadelphia metropolitan areas, and nonpediatric elsewhere in the state) and the numbers with all data required for TRISS. Nonpediatric centers admitted the majority
COPES, AND SACCO
Table 1. Completanets oi Data, 4,615 Trauma PatiofIts Lou Then 15 Years Old, Admitted to Trauma Centers in Pennsyhianle From 1986 to 1989 All Canters
Parameter Total no. of patients
PghlPRl Nonpea
Pediatric
Other Penne
4,615 28
1,881 2
1,293
1,441
16
10
Patients per center
164.8
940.5
80.8
144.1
TRISS suitability
3,704 (80.6)
1,498 (79.6)
1,040 (6b.4)
1,166 (80.9)
No. of centers
RTS defined
3,708 (80.3)
1,500 (79.7)
1,040 (4J.4)
1,168 (81.1)
ISS defined
4,610 (99.9)
1,879 (99.9)
1,292 (&X9)
1,439 (99.9)
NOTE. Percentages are shown in parentheses. Abbreviations: Pediatric, designated pediatric traljma centers; Pgh/ Phl Nonped,
trauma
metropolitan
areas surrounding
centers
without
pediatric
designation
in the
the pediatric trau/na centers (Pgh,
Pittsburgh: Phl, Philadelphia); Other Penna, other trabma centers in the state serving areas outside of the Pittsburgh and Phil delphia metropolBi itan areas; TRISS suitability, patients with sufficienti data elements to allow determination of probability of survival by mejhod described by Boyd et aL6 including defined measures for RTS, revised trauma score, and ISS. injury severity score.
of children, However, the average nu+bers of children cared for per pediatric institutiod was severalfold that cared for per nonpediatric idstitution. Approximately 80% of patients in each roup had all data required by TRISS. AImost with J,ut exception, patients with incomplete data are missing one or more RTS variables. Table 2 summarizes patient age an41 sex, mechanisms of injury, and mode of admission! The sex ratio of boys to girls was 2 to 1; it was highest in urban nonpediatric centers. Pediatric centqrs cared for significantly younger patients and received more patients by transfer than nonpediatric 4enters. Blunt injuries predominated in all centers, but significant differences existed in injury mechanigms. Pediatric centers received more pedestrian injujiies and falls, whereas rural nonpediatric centers cqred for more motor vehicle passengers and patient6 classified as “other” in the state registry, a catedory to which bicycle injuries are assigned (P < .OS). Table 3 summarizes mortaiity data. iMortality was highest in rural nonpediatric trauma denters (6.2%) for all age groups. The death rate in pediatric centers and urban nonpediatric centers were $imilar (4.1%) and significantly lower. Mortality was ihighest in the youngest age group (0 to 4 years) amodg all pediatric patients. For all children, death rates wpe highest for gunshot wounds (22.2%), pedestrian l&Juries (8.6%), and motor vehicle accidents (8.5%). Mortality from pedestrian injuries was highest in !rural centers (15.1%); other injury mechanisms exhi$ited no differences in mortality. No differences in m@rtality existed between transferred patients and direct admissions.
PEDIATRIC AND NONPEDIATRIC TRAUMA CENTERS
429
Table 2. Demographic Data in 4,615 Pediatric Trauma Patients in
Table 3. Mortality in 4,615 Pediatric Trauma Patients in Pennsylvania
Pennsylvania From 1966 to 1969
From 1966 to 1969
All Centers
Parameter Total
4,615
Pgh/Phl Nonped
Pediatric 1,881
1,293
Other Penna 1,441
Sex
All Centers
Total
4,615
Deaths
Male
3,095 (67.1) 1,231 (65.5)
906 (70.1)’
958 (66.5)
Female
1,516 (32.9)
Mean age (yr) o-4
Parameter
7.5 + 4.3 1,335 (26.9)
219 (4.1)
Pgh/Phl Nonped
Pediatric 1,881
1,293
77 (4.1)
Other Penna 1,441
53 (4.11
89 (6.2)*
Time to death
648 (34.5)t
387 (29.9)
483 (33.5)
< 24 hours
105 (47.4)
28 (36.4)’
31 (58.5)
46 (51.7)
6.4 f 4.2*
8.4 f 4.2
8.1 + 4.3
2 24 hours
114 (52.6)
49 (63.6)
22 (41.5)
43 (49.3)
706 (37.5)’
280 (21.7)
349 (24.2)
Deaths by age (yr)
5-9
1,607 (34.8)
682 (36.3)*
445 (34.4)
480 (33.3)
o-4
72 (5.4)
37 (5.2)
10 (3.6)
25 (7.2)
10-14
1,673 (36.3)
493 (26.2)+
568 (43.9)
612 (42.5)
5-9
78 (4.9)
28 (4.1)
17 (3.8)
33 (6.9)
10-14
69 (4.1)
12 (2.4)
26 (4.6)
31 (5.1)
Transfer
1,841 (39.2) 1,058 (56.2)’
325 (25.1)
458 (31.8)
Direct
2,688 (58.0)
966 (74.7)
969 (67.2)
197 (4.7)
70 (4.2)
43 (3.7)
84 (6.4)*
22 (4.9)
7 (3.6)
10 (8.1)
5 (3.8)
37 (9.0)
Type of admission
Unknown
88 (1.7)
753 (40.0) 70 (3.7)
2 (0.2)
14 (1.0)
Type of injury Blunt Penetrating
Deaths by type of injury Blunt Penetrating
4,162 (90.5) 1,684 (89.5) 453 (9.5)
197 (10.5)
1.169 (90.4)
1,309 (90.8)
124 (9.6)
132 (9.2)
Mechanism
Deaths by mechanism 78 (8.5)
24 (8.8)
17 (7.2)
920 (19.5)
273 (14.5)
237 (18.3)
410 (28.5)*
Motorcycle
l(1.4)
0 (0.0)
0 (0.0)
1 (4.0)
Motorcycle
72 (1.1)
20 (1.1)
27 (2.1)
25 (1.7)
Pedestrian
66 (8.6)
19 (6.0)
15 (6.3)
32 (15.1)*
Pedestrian
785 (16.8)
316 (16.8)*
237 (18.3)
212 (14.7)
Gunshot
20 (22.2)
90 (2.7)
32 (1.7)
34 (2.6)
24 (1.7)
Stabbing
2 (1.3)
Fall Other
Motor vehicle
Gunshot
Motor vehicle
159 (3.0)
76 (4.0)
47 (3.6)
36 (2.5)
Fall
1,360 (29.6)
651 (34.6)*
381 (29.5)
328 (22.8)
Other
1,249 (27.3)
513 (27.3)
330 (25.5)
406 (28.2)*
Stabbing
NOTE. Percentages are given in parentheses. ‘P < .05. tNo sex was recorded in two patients.
7 (21.9)
8 (23.5)
5 (20.8)
0 (0.0)
2 (4.3)
0 (0.0)
14 (1.0)
6 (0.9)
3 (0.8)
5 (1.5)
38 (3.0)
21 (4.11
8 (2.4)
9 (2.2)
Type of admission Transfer Direct
78 (4.2)
43 (4.1)
8 (2.5)
27 (5.9)
141 (5.2)
34 (4.5)
45 (4.7)
62 (6.4)
0 (0.0)
0 (0.0)
2 (0.2)
0 (0.0)
Unknown
NOTE. Percentages given in parentheses, refer to percent mortality for each category, except time to death, where percentages refer to
Table 4 summarizes injuries and percent survival by body region. A patient is included in statistics for each recorded injury. Head and neck injuries were most frequent (37.8%), followed by extremity injuries (33.1%). Pediatric trauma centers cared for a significantly larger proportion of head injuries (41.8% of all injuries v 32.3% and 37.6% for nonpediatric urban and rural centers, respectively), but the survival for patients with head and neck injuries among trauma center groups were similar. The lowest survival occurred in children with thoracic injuries (80.3%); the highest survival for thoracic injuries occurred in the pediatric centers. Table 5 gives the results of TRISS analysis for 3,705 pediatric trauma patients with all data required for TRISS. Rural nonpediatric trauma centers had a z score that failed to reach significance. z Scores for pediatric and nonpediatric trauma centers in Pittsburgh and Philadelphia were significant and positive. W for these groups indicated that 1.351 and 1.239 more survivors than expected from TRISS norms occurred per 100 patients, respectively. Figure 1 stratifies the survival rates for each group of trauma centers according to P(s), the probability of survival by TRISS analysis, into four intervals: P(s) 20.0and ~0.3, 10.3and ~0.6, z0.6and 0.5;
unexpected
> 0.5. See
textfor
deaths are those occurring in patients with P(s)
survivors are those patients surviving with PM description of z and W.
*P < .05 compared with other trauma center categories.
DISCUSSION
Important regional differences in pediatric trauma in Pennsylvania fall into three categories: trauma care, patient mix, and patient outcome. The first two were reflected in this study by a number of factors, and all had a potential effect on patient outcome: numbers of patients, age, sex, referral patterns, injury region and mechanisms, hospital organization, and characteristics of the community and region. The two pediatric specialty hospitals, the only designated pediatric trauma centers in the state, are located in major urban areas. They treat many injured children, especially younger patients and a large number with head injuries. They receive a majority of patients by transfer, not only from areas surrounding Pittsburgh and Philadelphia, but also from remote regions and from other nonpediatric trauma centers. Injury mechanisms reflect both the younger patient population (resulting in a higher proportion of admissions from falls) and an urban location (a relatively high rate of gunshot wounds and stabbings). The nonpediatric urban trauma centers care for far fewer children-a 10th of the number of children per institution treated by the pediatric specialty hospitals. The majority of their patients arrive directly from the injury scene. The children are older and more often boys. Injury mechanisms reflect older age groups, who are frequently victims of gunshot and stab wounds); teenage violence tends to affect male victims and to be treated in general hospitals in urban areas. Rural trauma centers outside the state’s two major metropolitan areas provide care for many children, an average of 140 patients per year-a number close to minimum standards for pediatric trauma care designation. Some provide pediatric specialty services, including pediatric surgery, neonatology, and
COPES, AND SACCO
pediatric critical care. Located in smaller cities than Pittsburgh and Philadelphia, trauma centers in Harrisburg, Hershey, Erie, and Allentown-Bethlehem serve smaller urban populations, but also provide tertiary care to large rural areas of the state. Children who come to these hospitals are injured in motor vehicle crashes more often than in the large cities. Like the nonpediatric centers in Pittsburgh and Philadelphia, rural centers receive most of their pediatric admissions from the scene of injury, and treat an older age group than the pediatric centers. The mortality rate was highest in rural centers, a statistic confirmed by TRISS. The rural hospitals also had a significantly higher rate of death from pedestrian injuries (15.1%). Our findings support those of Waller et al: who found higher deatb rates from injuries in rural areas than in urban, areas. They conducted their study during a time (the 1960s) when trauma center concepts were primitive and their development nonexistent, particularly in rural areas. Still, trauma victims in rural areas suffer delays in prompt transport to centers that can provide definitive care because discovery, notificationand transportation take longer in remote regions than in more populated areas, all with potentially adverse effects on survival. Why is the mortality similar in pediatric and nonpediatric trauma centers in the two major urban areas in the state? The system of pediatric patient referral that has evolved spontaneously between pediatric and nonpediatric trauma centers’ in Pittsburgh 100
urban nonped so iS .$ 60
z E 8
40
& n 20
0.3~PSSO.6
0.6
in Trauma Care Among Pediatric Trauma Centers ByDon
K. Nakayama,
Wayne
S. Copes,
and William
and Nonpediatric Sacco
Pittsburgh, Pennsylvania and Bel-Air, Maryland l TO characterize
pediatric
trauma
care,
state
trauma
registry data from all designated trauma centers in Pennsylvania were divided into three categories, that from: (1) pediatric centers, (2) urban nonpediatric
centers, (3) and rural
nonpediatric centers. From October 1.1986 through September 30, 1889 (3 years), 4,615 patients less than 15 years old were admitted to 28 trauma centers in Pennsylvania. Nonpediatric centers cared for the majority of children (2,734, 59.2%). but the average number of children treated per nonpediatric institution (105.1 per year) was far fewer than the average treated in the pediatric centers (940.5). Pediatric trauma centers in the state treated a younger population (6.4 r 4.2 years, mean 2 SD) compared with urban and rural nonpediatric centers (8.4 + 4.2 and 8.1 f 4.3 years, respectively; P c .05). Pediatric centers received proportionately more children by transfer (56.2%). victims of falls (34.6%). pedestrian injuries (16.8%). and head and neck injuries (41.8%. all P < .05). Nonpediatric centers received children directly from the scene of injury more frequently than transferred from other hospitals. The male:female sex ratio in urban nonpediatric centers was significantly higher (70.1%. P c .05) than in the other two groups. Rural nonpediatric centers cared for a higher proportion of motor vehicle passengers (28.5%) and patients classified as “other” in the state registry, a category to which bicycle injuries are assigned (28.2%. P < .05). Mortality was highest in rural nonpediatric centers (6.2%). The death rate in pediatric centers and urban nonpediatric centers were similar (4.1%) and significantly lower (P < .05). Mortality from pedestrian injuries was higher in rural centers (15.1%); other injury mechanisms exhibited no regional differences in mortality. Using TRISS, I for rural centers (1.123) failed to reach statistical significance, whereas z for pediatric and urban nonpediatric centers reached significance (3.896 and 3.335, respectively). Using the probability of survival (P(s)) generated by TRISS analysis, no significant differences in survival were noted among the three groups of trauma centers when P(s) was stratified, although survival was slightly higher in pediatric centers for P(s) 1.3 and q.6. Important regional differences in pediatric trauma care exist, specifically major differences in patient age, injury, and referral source. Rural trauma centers have a higher pediatric mortality than urban centers, a possible result of differences in access to prehospital care. Copyright o 1992 by W.B. Saunders Company INDEX WORDS:
Pediatric trauma care.
0 The
NE FOURTH of injury victims are chiIdren, so pediatric care is a priority of all trauma centers. care of children with significant injuries often require special equipment and involve pediatric surgical specialists, and standards of pediatric trauma care have been published by the American College of Surgeons that address the unique requirements of children.’ In Pennsylvania a system of 28 accredited JournalofPediatric Surgery, Vol27, No 4 (April), 1992: pp 427-431
trauma centers has been developed to provide optimal care to the injured throughout the state; only the Children’s Hospital of Pittsburgh and the Children’s Hospital of Philadelphia have satisfied standards for designation as pediatric trauma centers. Therefore, many injured children receive care at general trauma centers. To characterize the care of injured children in both pediatric and nonpediatric trauma centers, we used data from the Pennsylvania trauma registry in an attempt to answer the following questions. How do trauma centers with pediatric designation differ from the many centers without special designation? How do centers differ in terms of the referral, transfer, patient age, injury type, and severity? What happens in areas where pediatric trauma centers don’t exist? Do regional differences affect mortality? MATERIALS
AND
METHODS
In 1985, the legislature of the Commonwealth
of Pennsylvania authorized the Pennsylvania Trauma Systems Foundation (PTSF) to formulate standards for the accreditation of hospitals as trauma centers. PTSF standards parallel published guidelines by the American College of Surgeons’ and the American College of Emergency Physicians.’ The PTSF confirms compliance with standards of trauma care through on-site surveys by out-of-state experts. Separate standards for pediatric trauma center designation, called Pediatric Regional Resource Trauma Centers, were also formulated. These were based on standards outlined by Ramenofsky and Morse,3 which are included with the guidelines of the American College of Surgeons. Pediatric trauma centers must demonstrate specific pediatric specialty facilities, equipment, personnel, training, research, and educational activities. Trauma centers without pediatric designation must have a volume of 600 trauma admissions annually; pediatric trauma centers must care for at least 150 pediatric trauma admissions annually. In this study, PTSF-designated trauma centers in the state were divided into three categories. “Pediatric trauma centers” refer to the two hospitals designated by PTSF as Pediatric Regional Resource Trauma Centers, the Children’s Hospital of Pittsburgh
From the Benedum Pediatric Trauma Program, the Department of Pediatric Surgery, the Children’s Hospital of Pittsburgh, the Department of Surgery, the University of Pittsburgh School of Medicine, Pittsburgh, PA, and Tri-Analylics, Inc, BeLAir MD. Date accepted: November I, 1991. Supported in part by the Pennsylvania Trauma Systems Foundation, the Claude Worthington Benedum Foundation, and the Health Services and Research Foundation, Pittsburgh, PA. Address reprint requests to Don I2 Nahayama, MD, Department of Pediatric Surgery, the Children S Hospital of Pittsburgh, 3705 Fifth Ave at DeSoto St, Pittsburgh, PA 15213. Copyright o I992 by WB. Saunders Company 0022-3468192/2704-0002$03.0010 427
NAKAYAMA,
428
and the Children’s Hospital of Philadelphia. Nonpediatric
centers
were divided into two groups: “urban centers,” those located in the
Pittsburgh and Philadelphia areas (“Pgh/Phl nonped”), regions in the immediate cachement area of the two pediatric centers, and “rural centers,” those located outside of the Pittsburgh and Philadelphia metropolitan areas (“other Penna”). PTSF maintains a trauma registry on all injured patients admitted to accredited trauma centers in the state satisfying the following criteria: all deaths; lengths of stay greater than 3 days; admissions to an intensive care unit; and all transfers into and out of the hospital. Gilliot et al have described the registry (termed the Pennsylvania Trauma Outcome Study) at length elsewhere.4 In brief, data elements collected are generally those of the Major Trauma Outcome Study (MTOS)? Other data items dealing with patient transfer, prehospital care, emergency department, and in-hospital phases of care are also collected. Tri-Analytics, Inc (Be1 Air, MD) receives data collected at the trauma centers, codes injury descriptions, and provides quarterly summaries of demographic, injury, and treatment data, quality assurance summaries, and outcome analyses based on the TRISS methodology,6 described in brief below. Inconsistent data, such as transposed dates of admission and discharge, are returned to the trauma center for revision. Trauma center registrars undergo training sessions on state trauma registry data collection to assure consistency in data abstraction. In addition, PTSF and Tri-Analytics, Inc. conduct periodic reviews of data collection submission on site at all trauma centers. The review matches registry data with data in hospital charts, reabstracted by a member of the state registry staff. Briefly, TRISS predicts the probability of survival (P(s)) for any patient from the patients’ Injury Severity Score (ISS, based upon the 1985 Revision of the Abbreviated Injury Scale),’ and the Revised Trauma Score (RTS) measured on arrival in the emergency department. A collection of P(s) for a given subset of patients (for example, all patients treated at a single hospital, or for the purposes of this report, a group of hospitals) gives thez statistic that reflects the difference between the actual number of survivors in a population subset and the expected number of survivors based on TRISS norms, which are generated from the entire MTOS group. When the actual number of survivors exceeds the number predicted from the baseline, z is positive; when it is less, .z is negative.
Absolute
values
of z exceeding
1.96
are
significant
(P < .OS). Far large samples,
small and clinically insignificant differences between actual and expected numbers of survivors can reach statistical significance. The W statistic provides perspective to the interpretation of significant z values. W represents the difference between the actual number of survivors and the expected number (predicted by TRISS) per 100 patients treated. The x2 statistic was used to compare frequencies and analysis of variance was used to compare differences between mean values. P < .05 was considered significant.
RESULTS
From October 1,1986 through September 30,1989 (3 years), 4,615 patients 14 years old and younger were admitted to PTSF-designated trauma centers in Pennsylvania, and had data entered into the state trauma registry. Table 1 summarizes the numbers of admissions by trauma center category (pediatric, nonpediatric in the Pittsburgh and Philadelphia metropolitan areas, and nonpediatric elsewhere in the state) and the numbers with all data required for TRISS. Nonpediatric centers admitted the majority
COPES, AND SACCO
Table 1. Completanets oi Data, 4,615 Trauma PatiofIts Lou Then 15 Years Old, Admitted to Trauma Centers in Pennsyhianle From 1986 to 1989 All Canters
Parameter Total no. of patients
PghlPRl Nonpea
Pediatric
Other Penne
4,615 28
1,881 2
1,293
1,441
16
10
Patients per center
164.8
940.5
80.8
144.1
TRISS suitability
3,704 (80.6)
1,498 (79.6)
1,040 (6b.4)
1,166 (80.9)
No. of centers
RTS defined
3,708 (80.3)
1,500 (79.7)
1,040 (4J.4)
1,168 (81.1)
ISS defined
4,610 (99.9)
1,879 (99.9)
1,292 (&X9)
1,439 (99.9)
NOTE. Percentages are shown in parentheses. Abbreviations: Pediatric, designated pediatric traljma centers; Pgh/ Phl Nonped,
trauma
metropolitan
areas surrounding
centers
without
pediatric
designation
in the
the pediatric trau/na centers (Pgh,
Pittsburgh: Phl, Philadelphia); Other Penna, other trabma centers in the state serving areas outside of the Pittsburgh and Phil delphia metropolBi itan areas; TRISS suitability, patients with sufficienti data elements to allow determination of probability of survival by mejhod described by Boyd et aL6 including defined measures for RTS, revised trauma score, and ISS. injury severity score.
of children, However, the average nu+bers of children cared for per pediatric institutiod was severalfold that cared for per nonpediatric idstitution. Approximately 80% of patients in each roup had all data required by TRISS. AImost with J,ut exception, patients with incomplete data are missing one or more RTS variables. Table 2 summarizes patient age an41 sex, mechanisms of injury, and mode of admission! The sex ratio of boys to girls was 2 to 1; it was highest in urban nonpediatric centers. Pediatric centqrs cared for significantly younger patients and received more patients by transfer than nonpediatric 4enters. Blunt injuries predominated in all centers, but significant differences existed in injury mechanigms. Pediatric centers received more pedestrian injujiies and falls, whereas rural nonpediatric centers cqred for more motor vehicle passengers and patient6 classified as “other” in the state registry, a catedory to which bicycle injuries are assigned (P < .OS). Table 3 summarizes mortaiity data. iMortality was highest in rural nonpediatric trauma denters (6.2%) for all age groups. The death rate in pediatric centers and urban nonpediatric centers were $imilar (4.1%) and significantly lower. Mortality was ihighest in the youngest age group (0 to 4 years) amodg all pediatric patients. For all children, death rates wpe highest for gunshot wounds (22.2%), pedestrian l&Juries (8.6%), and motor vehicle accidents (8.5%). Mortality from pedestrian injuries was highest in !rural centers (15.1%); other injury mechanisms exhi$ited no differences in mortality. No differences in m@rtality existed between transferred patients and direct admissions.
PEDIATRIC AND NONPEDIATRIC TRAUMA CENTERS
429
Table 2. Demographic Data in 4,615 Pediatric Trauma Patients in
Table 3. Mortality in 4,615 Pediatric Trauma Patients in Pennsylvania
Pennsylvania From 1966 to 1969
From 1966 to 1969
All Centers
Parameter Total
4,615
Pgh/Phl Nonped
Pediatric 1,881
1,293
Other Penna 1,441
Sex
All Centers
Total
4,615
Deaths
Male
3,095 (67.1) 1,231 (65.5)
906 (70.1)’
958 (66.5)
Female
1,516 (32.9)
Mean age (yr) o-4
Parameter
7.5 + 4.3 1,335 (26.9)
219 (4.1)
Pgh/Phl Nonped
Pediatric 1,881
1,293
77 (4.1)
Other Penna 1,441
53 (4.11
89 (6.2)*
Time to death
648 (34.5)t
387 (29.9)
483 (33.5)
< 24 hours
105 (47.4)
28 (36.4)’
31 (58.5)
46 (51.7)
6.4 f 4.2*
8.4 f 4.2
8.1 + 4.3
2 24 hours
114 (52.6)
49 (63.6)
22 (41.5)
43 (49.3)
706 (37.5)’
280 (21.7)
349 (24.2)
Deaths by age (yr)
5-9
1,607 (34.8)
682 (36.3)*
445 (34.4)
480 (33.3)
o-4
72 (5.4)
37 (5.2)
10 (3.6)
25 (7.2)
10-14
1,673 (36.3)
493 (26.2)+
568 (43.9)
612 (42.5)
5-9
78 (4.9)
28 (4.1)
17 (3.8)
33 (6.9)
10-14
69 (4.1)
12 (2.4)
26 (4.6)
31 (5.1)
Transfer
1,841 (39.2) 1,058 (56.2)’
325 (25.1)
458 (31.8)
Direct
2,688 (58.0)
966 (74.7)
969 (67.2)
197 (4.7)
70 (4.2)
43 (3.7)
84 (6.4)*
22 (4.9)
7 (3.6)
10 (8.1)
5 (3.8)
37 (9.0)
Type of admission
Unknown
88 (1.7)
753 (40.0) 70 (3.7)
2 (0.2)
14 (1.0)
Type of injury Blunt Penetrating
Deaths by type of injury Blunt Penetrating
4,162 (90.5) 1,684 (89.5) 453 (9.5)
197 (10.5)
1.169 (90.4)
1,309 (90.8)
124 (9.6)
132 (9.2)
Mechanism
Deaths by mechanism 78 (8.5)
24 (8.8)
17 (7.2)
920 (19.5)
273 (14.5)
237 (18.3)
410 (28.5)*
Motorcycle
l(1.4)
0 (0.0)
0 (0.0)
1 (4.0)
Motorcycle
72 (1.1)
20 (1.1)
27 (2.1)
25 (1.7)
Pedestrian
66 (8.6)
19 (6.0)
15 (6.3)
32 (15.1)*
Pedestrian
785 (16.8)
316 (16.8)*
237 (18.3)
212 (14.7)
Gunshot
20 (22.2)
90 (2.7)
32 (1.7)
34 (2.6)
24 (1.7)
Stabbing
2 (1.3)
Fall Other
Motor vehicle
Gunshot
Motor vehicle
159 (3.0)
76 (4.0)
47 (3.6)
36 (2.5)
Fall
1,360 (29.6)
651 (34.6)*
381 (29.5)
328 (22.8)
Other
1,249 (27.3)
513 (27.3)
330 (25.5)
406 (28.2)*
Stabbing
NOTE. Percentages are given in parentheses. ‘P < .05. tNo sex was recorded in two patients.
7 (21.9)
8 (23.5)
5 (20.8)
0 (0.0)
2 (4.3)
0 (0.0)
14 (1.0)
6 (0.9)
3 (0.8)
5 (1.5)
38 (3.0)
21 (4.11
8 (2.4)
9 (2.2)
Type of admission Transfer Direct
78 (4.2)
43 (4.1)
8 (2.5)
27 (5.9)
141 (5.2)
34 (4.5)
45 (4.7)
62 (6.4)
0 (0.0)
0 (0.0)
2 (0.2)
0 (0.0)
Unknown
NOTE. Percentages given in parentheses, refer to percent mortality for each category, except time to death, where percentages refer to
Table 4 summarizes injuries and percent survival by body region. A patient is included in statistics for each recorded injury. Head and neck injuries were most frequent (37.8%), followed by extremity injuries (33.1%). Pediatric trauma centers cared for a significantly larger proportion of head injuries (41.8% of all injuries v 32.3% and 37.6% for nonpediatric urban and rural centers, respectively), but the survival for patients with head and neck injuries among trauma center groups were similar. The lowest survival occurred in children with thoracic injuries (80.3%); the highest survival for thoracic injuries occurred in the pediatric centers. Table 5 gives the results of TRISS analysis for 3,705 pediatric trauma patients with all data required for TRISS. Rural nonpediatric trauma centers had a z score that failed to reach significance. z Scores for pediatric and nonpediatric trauma centers in Pittsburgh and Philadelphia were significant and positive. W for these groups indicated that 1.351 and 1.239 more survivors than expected from TRISS norms occurred per 100 patients, respectively. Figure 1 stratifies the survival rates for each group of trauma centers according to P(s), the probability of survival by TRISS analysis, into four intervals: P(s) 20.0and ~0.3, 10.3and ~0.6, z0.6and 0.5;
unexpected
> 0.5. See
textfor
deaths are those occurring in patients with P(s)
survivors are those patients surviving with PM description of z and W.
*P < .05 compared with other trauma center categories.
DISCUSSION
Important regional differences in pediatric trauma in Pennsylvania fall into three categories: trauma care, patient mix, and patient outcome. The first two were reflected in this study by a number of factors, and all had a potential effect on patient outcome: numbers of patients, age, sex, referral patterns, injury region and mechanisms, hospital organization, and characteristics of the community and region. The two pediatric specialty hospitals, the only designated pediatric trauma centers in the state, are located in major urban areas. They treat many injured children, especially younger patients and a large number with head injuries. They receive a majority of patients by transfer, not only from areas surrounding Pittsburgh and Philadelphia, but also from remote regions and from other nonpediatric trauma centers. Injury mechanisms reflect both the younger patient population (resulting in a higher proportion of admissions from falls) and an urban location (a relatively high rate of gunshot wounds and stabbings). The nonpediatric urban trauma centers care for far fewer children-a 10th of the number of children per institution treated by the pediatric specialty hospitals. The majority of their patients arrive directly from the injury scene. The children are older and more often boys. Injury mechanisms reflect older age groups, who are frequently victims of gunshot and stab wounds); teenage violence tends to affect male victims and to be treated in general hospitals in urban areas. Rural trauma centers outside the state’s two major metropolitan areas provide care for many children, an average of 140 patients per year-a number close to minimum standards for pediatric trauma care designation. Some provide pediatric specialty services, including pediatric surgery, neonatology, and
COPES, AND SACCO
pediatric critical care. Located in smaller cities than Pittsburgh and Philadelphia, trauma centers in Harrisburg, Hershey, Erie, and Allentown-Bethlehem serve smaller urban populations, but also provide tertiary care to large rural areas of the state. Children who come to these hospitals are injured in motor vehicle crashes more often than in the large cities. Like the nonpediatric centers in Pittsburgh and Philadelphia, rural centers receive most of their pediatric admissions from the scene of injury, and treat an older age group than the pediatric centers. The mortality rate was highest in rural centers, a statistic confirmed by TRISS. The rural hospitals also had a significantly higher rate of death from pedestrian injuries (15.1%). Our findings support those of Waller et al: who found higher deatb rates from injuries in rural areas than in urban, areas. They conducted their study during a time (the 1960s) when trauma center concepts were primitive and their development nonexistent, particularly in rural areas. Still, trauma victims in rural areas suffer delays in prompt transport to centers that can provide definitive care because discovery, notificationand transportation take longer in remote regions than in more populated areas, all with potentially adverse effects on survival. Why is the mortality similar in pediatric and nonpediatric trauma centers in the two major urban areas in the state? The system of pediatric patient referral that has evolved spontaneously between pediatric and nonpediatric trauma centers’ in Pittsburgh 100
urban nonped so iS .$ 60
z E 8
40
& n 20
0.3~PSSO.6
0.6
