Trends in Operative Management of Pediatric Splenic Injury in a Regional Trauma System
http://www.100md.com
《小儿科》
Naval Medical Center Portsmouth, Portsmouth, Virginia
Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
Departments of Pediatric General and Thoracic Surgery, Division of Trauma and Surgical Critical Care
Anesthesia and Critical Care Medicine
Pediatrics, Division of Emergency Medicine, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
ABSTRACT
Objective. Selective nonoperative management of pediatric blunt splenic injury became the standard of care in the late 1980s. The extent to which this practice has been adopted in both trauma centers and nontrauma hospitals has been investigated sporadically. Several studies have demonstrated significant variations in practice patterns; however, most published studies capture only a selective population over a relatively short time interval, often without simultaneous adjustment for confounding variables. The objective of this study was to characterize the variation in operative versus nonoperative management of blunt splenic injury in children in nontrauma hospitals and in trauma centers with varying resources for pediatric care within a regionalized trauma system in the past decade.
Methods. The study population included all children who were younger than 19 years and had a diagnosis of blunt injury to the spleen (International Classification of Diseases code 865.00–865.09) and were admitted to each of the 175 acute care hospitals in Pennsylvania between 1991 and 2000. The proportion of patients who were treated operatively was stratified by trauma-level certification and adjusted for age and splenic injury severity. Multivariable logistic regression models were used to generate probabilities of splenectomy by age, injury severity, and hospital type.
Results. From 1991 through 2000 in Pennsylvania, 3245 children sustained blunt splenic injury that required hospitalization; 752 (23.2%) were treated operatively. Generally, as age and splenic injury severity increased, the proportion of patients who were treated operatively increased. Compared with pediatric trauma centers, the relative risk (with associated 95% confidence interval) of splenectomy was 4.4 (3.0–6.3) for level 1 trauma centers with additional qualifications in pediatrics; 6.2 (4.4–8.7) for level 1 trauma centers, 6.3 (5.3–7.4) for level 2 trauma centers, and 5.0 (4.2–5.9) for nontrauma centers. Significant variation in practice pattern was seen among hospital types and over time even after adjustment for age and injury severity.
Conclusions. The operative management of splenic injury in children varied significantly by hospital trauma status and over time during the past decade in Pennsylvania. Given the relative benefits of nonoperative treatment for children with blunt splenic injury, these results highlight the need for more widespread and standardized adoption of this treatment, particularly in hospitals without a large volume of pediatric trauma patients.
Key Words: trauma systems pediatric trauma care splenic injury blunt abdominal injury
Abbreviations: ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification PHC4, Pennsylvania Health Care Cost Containment Council PTC, pediatric trauma center AQTC, level 1 trauma center with additional qualifications in pediatric trauma L1TC, level 1 regional trauma center L2TC, level 2 regional trauma center NTC, nontrauma center CI, confidence interval RR, relative risk
Children are an important subgroup of patients who are treated in regional trauma systems and have unique requirements for optimal care. Many trauma systems have identified specialized resources for pediatric trauma care, including the designation of pediatric trauma centers and trauma centers with additional qualifications in pediatrics. Previous studies have examined outcomes of children who are treated in pediatric and nonpediatric trauma centers with conflicting results.1–4 Several studies have demonstrated no difference in predicted mortality for children who are treated in these different types of trauma centers.3,5 However, Potoka et al1 demonstrated that survival of children with blunt abdominal injury, in particular children with blunt splenic injury, is better for children who are treated at pediatric trauma centers.
After blunt injury to the spleen, splenectomy was the preferred method of management until the late 1970s. Practice patterns began to include several splenic salvage procedures in response to King and Schumacher's6 description in 1952 of overwhelming sepsis in children after splenectomy. Although the concept of nonoperative treatment of selected pediatric patients with splenic injury was introduced in 1968,7 it was not until the development of better diagnostic modalities in the 2 subsequent decades that nonoperative management became a more common treatment strategy in children.8 Currently, the standard of care for a hemodynamically stable child with a splenic injury documented by computed tomography scan is nonoperative treatment with close monitoring by an experienced surgical team.9–11
Numerous studies have described benefits of conservative nonoperative management of splenic injury in children. Children who are treated nonoperatively require fewer blood transfusions (44% vs 13%), have shorter lengths of hospital stay (7.8 vs 4.2 days), and have lower mortality compared with their laparotomy cohort.1,9 Morbidity and mortality are further compounded by the life-long risk of overwhelming postsplenectomy sepsis that is estimated to occur >85 times the rate of the normal population.12
Despite the well-documented benefits of nonoperative management, actual practice patterns for management of splenic injury in children vary widely. Previous research has documented that nonoperative management of splenic injury is more common in pediatric hospitals than in general trauma center and nontrauma center hospitals.1,13 Mooney et al13 described this variation in New Hampshire from 1991 to 1994. Even after adjustment for case mix, the splenectomy rate at the children's hospital was significantly lower (5.5%) than at the larger general hospitals (45.6%) and smaller community hospitals (17.6%). A comparison of the pediatric and nonpediatric trauma centers in Pennsylvania between 1991 and 1995 revealed 10-fold variation in performance of splenectomy across hospital types.1 Patrick et al4 recently published a descriptive account of decreasing trends of operative management of splenic injury in children at a single adult trauma center that suggested parity with pediatric trauma centers had been reached. These studies described practice patterns in a selective population over a relatively short time period during which initial reports of nonoperative management were published. The extent to which these original patterns have varied since is not known. In addition, no previous study has systematically examined the treatment of children with splenic injury at non–trauma center hospitals. Even in regions with organized trauma systems, a significant number of children are likely treated at non–trauma center hospitals. The diffusion of practice patterns developed in specialized trauma centers into these hospitals is largely unknown.
The objective of this study was to characterize the variation in operative versus nonoperative management of splenic injury in children among hospitals with varying resources for pediatric trauma care within a regionalized trauma system over a 10-year period. We hypothesized that the proportion of children who were treated operatively would vary inversely with the level of pediatric trauma care resources of the hospitals. In addition, we hypothesized that, within a group of hospitals with the same trauma-level certification, operative management would decrease over time.
METHODS
Using statewide hospital discharge data, we identified children who were 18 years of age or younger and had either a principal or a secondary diagnosis code for blunt injury to the spleen (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 865.00–865.09) and were admitted to all acute care hospitals in Pennsylvania from January 1, 1991, to December 31, 2000. Data were obtained from the Pennsylvania Health Care Cost Containment Council (PHC4), an independent state agency that collects demographic, clinical, and outcomes data on all patients who are discharged from all acute care hospitals in the state. Each of the 175 hospitals in Pennsylvania is required by law to provide administrative and coded diagnostic data on every patient discharge. Data available included age, gender, up to 9 diagnosis codes, up to 6 procedure codes, discharge status, length of stay, and a hospital identifier. This data source identified all children who were hospitalized with splenic injury in Pennsylvania. Children who died before hospital admission or those with penetrating abdominal trauma with splenic injury (865.1) were not included in the study.
Splenic injury severity was assigned using the Organ Injury Scaling system14 on the basis of information contained in the fifth-digit subclassification of the ICD-9-CM codes. Severity of injury was also determined using the ICDMAP90, a computerized program that converts ICD-9-CM diagnosis codes into Abbreviated Injury Scale scores and Injury Severity Scores. The ICDMAP90 was validated recently for use in pediatric injury research.15
Hospitals were stratified into 5 groups on the basis of their pediatric trauma care resources: pediatric trauma center (PTC), level 1 trauma center with additional qualifications in pediatric trauma (AQTC), level 1 regional trauma center (L1TC), level 2 regional trauma center (L2TC), and hospitals without trauma certification (NTC). The level of trauma certification of each hospital in the state for each year from 1991 to 2000 was verified by the Pennsylvania Trauma Systems Foundation and matched with the hospital identifier code in the PHC4 database. In the past decade, between 24 and 31 hospitals were certified trauma centers, some of which changed status over time. Currently, there are 2 PTC, 4 AQTC, 8 L1TC, 12 L2TC, and 149 NTC.
Operative management was defined as any of the following procedure codes: partial splenectomy (41.43), total splenectomy (41.5), repair and plastic operations of the spleen (41.95), and other operations on spleen (41.99). Nonoperative management was defined as a diagnosis code for blunt splenic injury (865.00–865.09) and the absence of any procedure code for surgery on the spleen.
Statistical Analysis
Although the unit of interest was the hospital, the patient was the unit of analysis for all statistical procedures. This method allowed for adjustment for differences in patient characteristics across hospitals and over time. Several patient-level characteristics such as age and injury severity were categorized after considering their distribution and the association between these factors and the risk of splenectomy. When appropriate, adjacent categories were collapsed to ensure the adequacy of sample sizes in each exposure group and to guard against overparameterization of regression models.
Annual proportions of operative versus nonoperative management were calculated and plotted by hospital type for each year over the 10-year study period. Rates are smoothed by robust nonlinear methods to enhance interpretation and uncover trends.16
The relationship between hospital trauma-level certification and the rate of operative management among children with splenic injury was examined by an odds ratio using multiple logistic regression as implemented in STATA V 7.0 (Stata Corp, College Station, TX) to control for potential confounding variables. Robust variances, confidence intervals (CIs), and P values were computed to account for the clustering of patients within hospitals by referral or practice patterns. Predicted probabilities of operative management adjusted for covariates were then computed from the results of the logistic regression. Relative risks (RR) with associated 95% CIs were calculated from logistic regression using the method of conditional standardization as implemented by a user-written program in STATA.17 Potential confounders were tested according to whether they changed the coefficients of interest (hospital type and time period). Interaction terms were tested using Wald tests. None was significant. Statistical significance was defined as P < .05.
RESULTS
A total of 3245 children with a diagnosis of blunt splenic injury were admitted to 175 acute care hospitals in Pennsylvania from 1991 to 2000. Excluded from the analysis were 249 children who had splenic injury and were admitted to rehabilitation facilities. Overall, 23.2% of children with blunt splenic injury were treated operatively. Of the 752 patients who were treated operatively, 56 (7.4%) were characterized as having had a partial splenectomy, 208 (27.7%) as a repair/plastic operation of the spleen, 484 (64.4%) as a total splenectomy, and 4 (<0.5%) as other operations on the spleen.
On average, children who had splenic injury and were treated operatively were older than those who were treated nonoperatively (P < .001). More than 90% of children who had splenic injury and were between 0 and 4 years of age were treated nonoperatively. The RR of operative management was >3 times higher for 15- to 18-year-olds than the youngest age group.
The mean length of stay was 7.9 ± 0.29 SEM days for children who were treated operatively versus 6.6 ± 0.46 SEM days for those who were treated nonoperatively (P < .001). Mortality was >3 times higher for children who had splenic injury and were treated operatively (P < .001).
The average Injury Severity Score was higher for children who were treated operatively (P < .001). The RR of operative management increased by >5-fold as the splenic injury grade increased from I/II to V.
Overall, 15.6% of children were treated at PTCs, 14.8% at AQTCs, 11.4% at L1TCs, 22.9% at L2TCs, and the remaining 35.3% at NTCs. The 10-year average proportion of operative management increased across the groups of hospitals as resources for pediatric trauma care decreased. The RRs of operative management were significantly higher in all types of hospitals when compared with the PTCs.
Given that the RR of operative management increased as the patients' age and splenic injury severity increased, hospital-specific rates of operative management were adjusted for these 2 potential confounders (Fig 2). Individual years were collapsed into 3 groups corresponding to the 3 time periods of relatively low (1991–1993), high (1993–1997), and variable (1998–2000) operative management in the nonpediatric trauma centers.
After adjustment for age and splenic injury severity, PTCs continued to demonstrate a consistent pattern of operative management in a minority of patients over the entire period of study. All other types of hospitals demonstrated a similar pattern of increasing operative management between 1993 and 1997 and then steady decreases over the final 2 years of the study period. At the end of the period, the average proportion of operative management varied from 21.7% in L2TCs to 12.4% in AQTCs. Of note, non–trauma center hospitals generally had rates of operative management that were between those of PTCs and nonpediatric trauma centers.
DISCUSSION
Results of this study demonstrate considerable variation in the management of blunt splenic injury in children by hospital type and over time in a statewide trauma system. Children who were treated in PTCs had consistently low rates of operative management across the entire period. Children who were treated in all other types of hospitals had rates of operative management that were lowest before 1993 then increased by 15% to 20% through 1997 and have subsequently decreased 3% to 9% since 1998. Among trauma centers, the rates of operative management varied inversely with resources allocated for pediatric trauma care.
Pediatric surgeons began using nonoperative management protocols for splenic injury in children in the early 1980s. On the basis of earlier studies, a management algorithm was proposed by pediatric surgeons in Toronto, stating that children who had blunt trauma and were hemodynamically stable after no more than 40% blood volume replacement with crystalloids and had a computed tomography scan that did not demonstrate other organ injuries could be treated safely and appropriately in a PICU with close surgical evaluation and management.18 Initial studies from pediatric trauma centers that have implemented this strategy reported favorable results. Reports from trauma centers that used this approach in adults in the late 1980s had less encouraging results, with varying success rates dependent on the grade of splenic injury, ranging from 92% for grade I injuries to 0% for grade IV.19 Therefore, nonoperative management of blunt splenic injury in adults was slowly accepted.
The peak of nonoperative management in the early part of the decade corresponded to the initial descriptions of its success in pediatric trauma centers in the late 1980s. Publications on this topic decreased in the mid-1990s, corresponding to the period of declining nonoperative management of pediatric splenic injury in all hospital types except PTCs. More recently, there has been a resurgence in the medical literature corresponding with the interest in and expansion of nonoperative management of splenic injury to the adult population.20,21 Our results suggest that the practice of nonoperative management of blunt splenic injury may vary with the amount of exposure that it received in the medical literature.
Previous studies have suggested that experience rather than training or resource allocation influences surgical practice patterns.22 Bratton et al22 examined whether the risk of operative treatment of children with intussusception varied by pediatric volume over a 10-year period in Washington State. The authors concluded that children who had intussusception and were treated in a specialized pediatric facility had a significantly lower risk of operative treatment, decreased length of stay, and lower hospital charges than children who were treated in facilities with lower pediatric volumes. The authors suggested that the greater physician experience with children in the large referral centers accounted for the differences rather than specialized pediatric training.22 In the current study, the average annual number of children with blunt splenic injury treated was 25 children in each PTC, 12 children in each AQTC, 5 children in each L1TC, 6 children in each L2TC, and <1 child in each NTC. Our results demonstrated that facilities with higher pediatric volumes used nonoperative management strategies for blunt splenic injury in children significantly more often than the lower volume facilities. However, non–trauma center hospitals had a nonoperative rate that was intermediate to that of pediatric-oriented trauma centers (PTCs and AQTCs) and adult-oriented trauma centers (L1TCs and L2TCs). This intriguing result is difficult to explain using the factors available in the PHC4 database. Given the relatively low annual number of children with blunt splenic injury treated at each NTC, we hypothesize that physicians at nontrauma centers have adopted practice patterns that are a composite of pediatric- and adult-oriented management strategies not reflective of experience or specialized training.
The decision regarding whether to operate on a child with blunt splenic injury is often complex and influenced by a number of factors, including the severity of the injury, the patient's physiologic stability, the presence of other serious injuries, physician preference, and local resources for care. In our study, children were more likely to be operated on when they were older and when they sustained a more serious splenic injury. The patterns of hospital-specific variation in management of blunt splenic injury in children persisted even after adjustment for age and splenic injury severity. However, differences in management strategy across hospitals may be attributable to additional factors that are not available in the PHC4 database. Information needed to determine the indication for surgery or the appropriateness of that decision was not available in our source of data and cannot be inferred by our results.
Information in the state discharge database was obtained primarily for nonresearch purposes; therefore, misclassification of some data elements is possible. For example, the use of a nonoperative protocol was assumed in the absence of surgical operative coding. It is likely that this misclassification is random across groups of hospitals; therefore, our estimates of the association between hospital type and probability of operative management may be biased toward the null. Our source of data allows stratification by organ injury severity providing more detailed information than other studies that have used more global injury severity scores, which may artificially inflate severity secondary to other associated injuries.
Researchers have shown that the physician decision-making process in acute medical problems is largely influenced by physician experience and knowledge and only somewhat influenced by external clinical evidence.23 Perhaps this is why management strategies such as the nonoperative management of blunt splenic injury in children are more likely to be adopted and sustained in centers with high volume experience. Published literature may have short-term impact that is not sustained. Careful reviews of the factors that significantly change physician behavior suggest that traditional didactic continuing medical education and passive distribution of information or clinical guidelines do not influence behavior or improve patient care.23 Systematic strategies to implement clinical practice guidelines or creation of clinical pathways have been shown to affect behavior and improve patient care, as have less appealing but effective financial incentives.23 Results of this study suggest the need for regionalized trauma care systems to implement evidence-based guidelines and to identify resources that are required to optimize care of the injured child across all types of hospitals.
ACKNOWLEDGMENTS
This study was supported by an Ambulatory Pediatric Association/AHRQ-Special Projects Grant, Endowed Chair of Pediatric Critical Care, Children's Hospital of Philadelphia, and Templeton Foundation for Research in Pediatric Surgery.
FOOTNOTES
Accepted Jun 14, 2004.
No conflict of interest declared.
Preliminary results were presented in poster format at the Pediatric Academic Societies Meeting; May 4–7, 2004; Baltimore, MD
REFERENCES
Potoka DA, Schall LC, Gardner MJ, Stafford PW, Peitzman AB, Ford HR. Impact of pediatric trauma centers on mortality in a statewide system. J Trauma. 2000;49 :237 –245
Jacobs IA, Kelly K, Valenziano C, Pawar J, Jones C. Non-operative management of blunt splenic and hepatic trauma in the pediatric population: significant differences between adult and pediatric surgeons Am Surg. 2001;67 :149 –154
Sherman HF, Landry VL, Jones LM. Should Level I trauma centers be rated NC-17 J Trauma. 2001;50 :784 –791
Patrick D, Moore EE, Bensard DD, Karrer FM. Operative management of injured children at an adult level 1 trauma center. J Trauma. 2000;48 :894 –901
Osler T, Vane DW, Tepas JJ, et al. Survival following trauma in children; do pediatric trauma centers have better survival rates Does ACS certification matter J Trauma. 1998;45 :189
King H, Schumacher HB. Splenic studies I: susceptibility to infection after splenectomy performed in infancy. Ann Surg. 1952;136 :239 –242
Upadhyaya P, Simpson JS. Splenic trauma in children. Surgery. 1968;126 :781 –790
Knudson MM, Maull KI. Non-operative management of solid organ injuries. Past, present, and future. Surg Clin North Am. 1999;79 :1357 –1371
Partrick DA, Bensard DD, Moore EE, Karrer FM. Non-operative management of solid organ injuries in children results in decreased blood utilization. J Pediatr Surg. 1999;34 :1695 –1699
Morse MA, Garcia VF. Selective non-operative management of pediatric blunt splenic trauma: risk for missed associated injuries. J Pediatr Surg. 1994;29 :23 –27
Sjovall A, Hirsch K. Blunt abdominal trauma in children: risks of non-operative treatment. J Pediatr Surg. 1997;32 :1169 –1174
Francke EL, Neu HC. Postsplenectomy infection. Surg Clin North Am. 1981;61 :135 –155
Mooney DP, Birkmeyer NJ, Udell JV, Shorter NA. Variation in the management of pediatric splenic injuries in New Hampshire. J Pediatr Surg. 1998;33 :1076 –1078; discussion 1079–1080
Moore EE, Cogbill TH, Jurkovich GJ, Shackford SR, Malangoni MA, Champion HR. Organ injury scaling: spleen and liver (1994 revision). J Trauma. 1995;38 :323 –324
Durbin D, Schwartz DF, Localio AR, MacKenzie EJ. Validation of the ICD/AIS MAP for pediatric use. Inj Prev. 2001;7 :96 –99
Velleman PF, Hoaglin DC. Applications, Basics, and Computing of Exploratory Data Analysis. Boston, MA: Duxbury Press; 1981
Flanders WD, Rhodes PH. Large sample confidence intervals for regression standardized risks, risk ratios, and risk differences. J Chron Dis. 1987;40 :697 –704
Wesson DE, Filler RM, Ein SH, Shandling B, Simpson JS, Stephens CA. Ruptured spleen—when to operate J Pediatr Surg. 1981;16 :324 –326
Cogbill TH, Moore EE, Jurkovich GJ, et al. Non-operative management of blunt splenic trauma: a multicenter experience. J Trauma. 1989;29 :1312 –1317
Cocanour CS, Moore FA, Ware DN, Marvin RG, Duke JH. Age should not be a consideration for non-operative management of blunt splenic injury. J Trauma. 2000;48 :606 –610; discussion 610–602
Harbrecht BG, Peitzman AB, Rivera L, et al. Contribution of age and gender to outcome of blunt splenic injury in adults: multicenter study of the eastern association for the surgery of trauma. J Trauma. 2001;51 :887 –895
Bratton SL, Haberkern CM, Waldhausen JH, Sawin RS, Allison JW. Intussusception: hospital size and risk of surgery. Pediatrics. 2001;107 :299 –303
Bauchner H, Simpson L, Chessare J. Changing physician behaviour. Arch Dis Child. 2001;84 :459 –462(Daniela H. Davis, MD, MSC)
Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
Departments of Pediatric General and Thoracic Surgery, Division of Trauma and Surgical Critical Care
Anesthesia and Critical Care Medicine
Pediatrics, Division of Emergency Medicine, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
ABSTRACT
Objective. Selective nonoperative management of pediatric blunt splenic injury became the standard of care in the late 1980s. The extent to which this practice has been adopted in both trauma centers and nontrauma hospitals has been investigated sporadically. Several studies have demonstrated significant variations in practice patterns; however, most published studies capture only a selective population over a relatively short time interval, often without simultaneous adjustment for confounding variables. The objective of this study was to characterize the variation in operative versus nonoperative management of blunt splenic injury in children in nontrauma hospitals and in trauma centers with varying resources for pediatric care within a regionalized trauma system in the past decade.
Methods. The study population included all children who were younger than 19 years and had a diagnosis of blunt injury to the spleen (International Classification of Diseases code 865.00–865.09) and were admitted to each of the 175 acute care hospitals in Pennsylvania between 1991 and 2000. The proportion of patients who were treated operatively was stratified by trauma-level certification and adjusted for age and splenic injury severity. Multivariable logistic regression models were used to generate probabilities of splenectomy by age, injury severity, and hospital type.
Results. From 1991 through 2000 in Pennsylvania, 3245 children sustained blunt splenic injury that required hospitalization; 752 (23.2%) were treated operatively. Generally, as age and splenic injury severity increased, the proportion of patients who were treated operatively increased. Compared with pediatric trauma centers, the relative risk (with associated 95% confidence interval) of splenectomy was 4.4 (3.0–6.3) for level 1 trauma centers with additional qualifications in pediatrics; 6.2 (4.4–8.7) for level 1 trauma centers, 6.3 (5.3–7.4) for level 2 trauma centers, and 5.0 (4.2–5.9) for nontrauma centers. Significant variation in practice pattern was seen among hospital types and over time even after adjustment for age and injury severity.
Conclusions. The operative management of splenic injury in children varied significantly by hospital trauma status and over time during the past decade in Pennsylvania. Given the relative benefits of nonoperative treatment for children with blunt splenic injury, these results highlight the need for more widespread and standardized adoption of this treatment, particularly in hospitals without a large volume of pediatric trauma patients.
Key Words: trauma systems pediatric trauma care splenic injury blunt abdominal injury
Abbreviations: ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification PHC4, Pennsylvania Health Care Cost Containment Council PTC, pediatric trauma center AQTC, level 1 trauma center with additional qualifications in pediatric trauma L1TC, level 1 regional trauma center L2TC, level 2 regional trauma center NTC, nontrauma center CI, confidence interval RR, relative risk
Children are an important subgroup of patients who are treated in regional trauma systems and have unique requirements for optimal care. Many trauma systems have identified specialized resources for pediatric trauma care, including the designation of pediatric trauma centers and trauma centers with additional qualifications in pediatrics. Previous studies have examined outcomes of children who are treated in pediatric and nonpediatric trauma centers with conflicting results.1–4 Several studies have demonstrated no difference in predicted mortality for children who are treated in these different types of trauma centers.3,5 However, Potoka et al1 demonstrated that survival of children with blunt abdominal injury, in particular children with blunt splenic injury, is better for children who are treated at pediatric trauma centers.
After blunt injury to the spleen, splenectomy was the preferred method of management until the late 1970s. Practice patterns began to include several splenic salvage procedures in response to King and Schumacher's6 description in 1952 of overwhelming sepsis in children after splenectomy. Although the concept of nonoperative treatment of selected pediatric patients with splenic injury was introduced in 1968,7 it was not until the development of better diagnostic modalities in the 2 subsequent decades that nonoperative management became a more common treatment strategy in children.8 Currently, the standard of care for a hemodynamically stable child with a splenic injury documented by computed tomography scan is nonoperative treatment with close monitoring by an experienced surgical team.9–11
Numerous studies have described benefits of conservative nonoperative management of splenic injury in children. Children who are treated nonoperatively require fewer blood transfusions (44% vs 13%), have shorter lengths of hospital stay (7.8 vs 4.2 days), and have lower mortality compared with their laparotomy cohort.1,9 Morbidity and mortality are further compounded by the life-long risk of overwhelming postsplenectomy sepsis that is estimated to occur >85 times the rate of the normal population.12
Despite the well-documented benefits of nonoperative management, actual practice patterns for management of splenic injury in children vary widely. Previous research has documented that nonoperative management of splenic injury is more common in pediatric hospitals than in general trauma center and nontrauma center hospitals.1,13 Mooney et al13 described this variation in New Hampshire from 1991 to 1994. Even after adjustment for case mix, the splenectomy rate at the children's hospital was significantly lower (5.5%) than at the larger general hospitals (45.6%) and smaller community hospitals (17.6%). A comparison of the pediatric and nonpediatric trauma centers in Pennsylvania between 1991 and 1995 revealed 10-fold variation in performance of splenectomy across hospital types.1 Patrick et al4 recently published a descriptive account of decreasing trends of operative management of splenic injury in children at a single adult trauma center that suggested parity with pediatric trauma centers had been reached. These studies described practice patterns in a selective population over a relatively short time period during which initial reports of nonoperative management were published. The extent to which these original patterns have varied since is not known. In addition, no previous study has systematically examined the treatment of children with splenic injury at non–trauma center hospitals. Even in regions with organized trauma systems, a significant number of children are likely treated at non–trauma center hospitals. The diffusion of practice patterns developed in specialized trauma centers into these hospitals is largely unknown.
The objective of this study was to characterize the variation in operative versus nonoperative management of splenic injury in children among hospitals with varying resources for pediatric trauma care within a regionalized trauma system over a 10-year period. We hypothesized that the proportion of children who were treated operatively would vary inversely with the level of pediatric trauma care resources of the hospitals. In addition, we hypothesized that, within a group of hospitals with the same trauma-level certification, operative management would decrease over time.
METHODS
Using statewide hospital discharge data, we identified children who were 18 years of age or younger and had either a principal or a secondary diagnosis code for blunt injury to the spleen (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 865.00–865.09) and were admitted to all acute care hospitals in Pennsylvania from January 1, 1991, to December 31, 2000. Data were obtained from the Pennsylvania Health Care Cost Containment Council (PHC4), an independent state agency that collects demographic, clinical, and outcomes data on all patients who are discharged from all acute care hospitals in the state. Each of the 175 hospitals in Pennsylvania is required by law to provide administrative and coded diagnostic data on every patient discharge. Data available included age, gender, up to 9 diagnosis codes, up to 6 procedure codes, discharge status, length of stay, and a hospital identifier. This data source identified all children who were hospitalized with splenic injury in Pennsylvania. Children who died before hospital admission or those with penetrating abdominal trauma with splenic injury (865.1) were not included in the study.
Splenic injury severity was assigned using the Organ Injury Scaling system14 on the basis of information contained in the fifth-digit subclassification of the ICD-9-CM codes. Severity of injury was also determined using the ICDMAP90, a computerized program that converts ICD-9-CM diagnosis codes into Abbreviated Injury Scale scores and Injury Severity Scores. The ICDMAP90 was validated recently for use in pediatric injury research.15
Hospitals were stratified into 5 groups on the basis of their pediatric trauma care resources: pediatric trauma center (PTC), level 1 trauma center with additional qualifications in pediatric trauma (AQTC), level 1 regional trauma center (L1TC), level 2 regional trauma center (L2TC), and hospitals without trauma certification (NTC). The level of trauma certification of each hospital in the state for each year from 1991 to 2000 was verified by the Pennsylvania Trauma Systems Foundation and matched with the hospital identifier code in the PHC4 database. In the past decade, between 24 and 31 hospitals were certified trauma centers, some of which changed status over time. Currently, there are 2 PTC, 4 AQTC, 8 L1TC, 12 L2TC, and 149 NTC.
Operative management was defined as any of the following procedure codes: partial splenectomy (41.43), total splenectomy (41.5), repair and plastic operations of the spleen (41.95), and other operations on spleen (41.99). Nonoperative management was defined as a diagnosis code for blunt splenic injury (865.00–865.09) and the absence of any procedure code for surgery on the spleen.
Statistical Analysis
Although the unit of interest was the hospital, the patient was the unit of analysis for all statistical procedures. This method allowed for adjustment for differences in patient characteristics across hospitals and over time. Several patient-level characteristics such as age and injury severity were categorized after considering their distribution and the association between these factors and the risk of splenectomy. When appropriate, adjacent categories were collapsed to ensure the adequacy of sample sizes in each exposure group and to guard against overparameterization of regression models.
Annual proportions of operative versus nonoperative management were calculated and plotted by hospital type for each year over the 10-year study period. Rates are smoothed by robust nonlinear methods to enhance interpretation and uncover trends.16
The relationship between hospital trauma-level certification and the rate of operative management among children with splenic injury was examined by an odds ratio using multiple logistic regression as implemented in STATA V 7.0 (Stata Corp, College Station, TX) to control for potential confounding variables. Robust variances, confidence intervals (CIs), and P values were computed to account for the clustering of patients within hospitals by referral or practice patterns. Predicted probabilities of operative management adjusted for covariates were then computed from the results of the logistic regression. Relative risks (RR) with associated 95% CIs were calculated from logistic regression using the method of conditional standardization as implemented by a user-written program in STATA.17 Potential confounders were tested according to whether they changed the coefficients of interest (hospital type and time period). Interaction terms were tested using Wald tests. None was significant. Statistical significance was defined as P < .05.
RESULTS
A total of 3245 children with a diagnosis of blunt splenic injury were admitted to 175 acute care hospitals in Pennsylvania from 1991 to 2000. Excluded from the analysis were 249 children who had splenic injury and were admitted to rehabilitation facilities. Overall, 23.2% of children with blunt splenic injury were treated operatively. Of the 752 patients who were treated operatively, 56 (7.4%) were characterized as having had a partial splenectomy, 208 (27.7%) as a repair/plastic operation of the spleen, 484 (64.4%) as a total splenectomy, and 4 (<0.5%) as other operations on the spleen.
On average, children who had splenic injury and were treated operatively were older than those who were treated nonoperatively (P < .001). More than 90% of children who had splenic injury and were between 0 and 4 years of age were treated nonoperatively. The RR of operative management was >3 times higher for 15- to 18-year-olds than the youngest age group.
The mean length of stay was 7.9 ± 0.29 SEM days for children who were treated operatively versus 6.6 ± 0.46 SEM days for those who were treated nonoperatively (P < .001). Mortality was >3 times higher for children who had splenic injury and were treated operatively (P < .001).
The average Injury Severity Score was higher for children who were treated operatively (P < .001). The RR of operative management increased by >5-fold as the splenic injury grade increased from I/II to V.
Overall, 15.6% of children were treated at PTCs, 14.8% at AQTCs, 11.4% at L1TCs, 22.9% at L2TCs, and the remaining 35.3% at NTCs. The 10-year average proportion of operative management increased across the groups of hospitals as resources for pediatric trauma care decreased. The RRs of operative management were significantly higher in all types of hospitals when compared with the PTCs.
Given that the RR of operative management increased as the patients' age and splenic injury severity increased, hospital-specific rates of operative management were adjusted for these 2 potential confounders (Fig 2). Individual years were collapsed into 3 groups corresponding to the 3 time periods of relatively low (1991–1993), high (1993–1997), and variable (1998–2000) operative management in the nonpediatric trauma centers.
After adjustment for age and splenic injury severity, PTCs continued to demonstrate a consistent pattern of operative management in a minority of patients over the entire period of study. All other types of hospitals demonstrated a similar pattern of increasing operative management between 1993 and 1997 and then steady decreases over the final 2 years of the study period. At the end of the period, the average proportion of operative management varied from 21.7% in L2TCs to 12.4% in AQTCs. Of note, non–trauma center hospitals generally had rates of operative management that were between those of PTCs and nonpediatric trauma centers.
DISCUSSION
Results of this study demonstrate considerable variation in the management of blunt splenic injury in children by hospital type and over time in a statewide trauma system. Children who were treated in PTCs had consistently low rates of operative management across the entire period. Children who were treated in all other types of hospitals had rates of operative management that were lowest before 1993 then increased by 15% to 20% through 1997 and have subsequently decreased 3% to 9% since 1998. Among trauma centers, the rates of operative management varied inversely with resources allocated for pediatric trauma care.
Pediatric surgeons began using nonoperative management protocols for splenic injury in children in the early 1980s. On the basis of earlier studies, a management algorithm was proposed by pediatric surgeons in Toronto, stating that children who had blunt trauma and were hemodynamically stable after no more than 40% blood volume replacement with crystalloids and had a computed tomography scan that did not demonstrate other organ injuries could be treated safely and appropriately in a PICU with close surgical evaluation and management.18 Initial studies from pediatric trauma centers that have implemented this strategy reported favorable results. Reports from trauma centers that used this approach in adults in the late 1980s had less encouraging results, with varying success rates dependent on the grade of splenic injury, ranging from 92% for grade I injuries to 0% for grade IV.19 Therefore, nonoperative management of blunt splenic injury in adults was slowly accepted.
The peak of nonoperative management in the early part of the decade corresponded to the initial descriptions of its success in pediatric trauma centers in the late 1980s. Publications on this topic decreased in the mid-1990s, corresponding to the period of declining nonoperative management of pediatric splenic injury in all hospital types except PTCs. More recently, there has been a resurgence in the medical literature corresponding with the interest in and expansion of nonoperative management of splenic injury to the adult population.20,21 Our results suggest that the practice of nonoperative management of blunt splenic injury may vary with the amount of exposure that it received in the medical literature.
Previous studies have suggested that experience rather than training or resource allocation influences surgical practice patterns.22 Bratton et al22 examined whether the risk of operative treatment of children with intussusception varied by pediatric volume over a 10-year period in Washington State. The authors concluded that children who had intussusception and were treated in a specialized pediatric facility had a significantly lower risk of operative treatment, decreased length of stay, and lower hospital charges than children who were treated in facilities with lower pediatric volumes. The authors suggested that the greater physician experience with children in the large referral centers accounted for the differences rather than specialized pediatric training.22 In the current study, the average annual number of children with blunt splenic injury treated was 25 children in each PTC, 12 children in each AQTC, 5 children in each L1TC, 6 children in each L2TC, and <1 child in each NTC. Our results demonstrated that facilities with higher pediatric volumes used nonoperative management strategies for blunt splenic injury in children significantly more often than the lower volume facilities. However, non–trauma center hospitals had a nonoperative rate that was intermediate to that of pediatric-oriented trauma centers (PTCs and AQTCs) and adult-oriented trauma centers (L1TCs and L2TCs). This intriguing result is difficult to explain using the factors available in the PHC4 database. Given the relatively low annual number of children with blunt splenic injury treated at each NTC, we hypothesize that physicians at nontrauma centers have adopted practice patterns that are a composite of pediatric- and adult-oriented management strategies not reflective of experience or specialized training.
The decision regarding whether to operate on a child with blunt splenic injury is often complex and influenced by a number of factors, including the severity of the injury, the patient's physiologic stability, the presence of other serious injuries, physician preference, and local resources for care. In our study, children were more likely to be operated on when they were older and when they sustained a more serious splenic injury. The patterns of hospital-specific variation in management of blunt splenic injury in children persisted even after adjustment for age and splenic injury severity. However, differences in management strategy across hospitals may be attributable to additional factors that are not available in the PHC4 database. Information needed to determine the indication for surgery or the appropriateness of that decision was not available in our source of data and cannot be inferred by our results.
Information in the state discharge database was obtained primarily for nonresearch purposes; therefore, misclassification of some data elements is possible. For example, the use of a nonoperative protocol was assumed in the absence of surgical operative coding. It is likely that this misclassification is random across groups of hospitals; therefore, our estimates of the association between hospital type and probability of operative management may be biased toward the null. Our source of data allows stratification by organ injury severity providing more detailed information than other studies that have used more global injury severity scores, which may artificially inflate severity secondary to other associated injuries.
Researchers have shown that the physician decision-making process in acute medical problems is largely influenced by physician experience and knowledge and only somewhat influenced by external clinical evidence.23 Perhaps this is why management strategies such as the nonoperative management of blunt splenic injury in children are more likely to be adopted and sustained in centers with high volume experience. Published literature may have short-term impact that is not sustained. Careful reviews of the factors that significantly change physician behavior suggest that traditional didactic continuing medical education and passive distribution of information or clinical guidelines do not influence behavior or improve patient care.23 Systematic strategies to implement clinical practice guidelines or creation of clinical pathways have been shown to affect behavior and improve patient care, as have less appealing but effective financial incentives.23 Results of this study suggest the need for regionalized trauma care systems to implement evidence-based guidelines and to identify resources that are required to optimize care of the injured child across all types of hospitals.
ACKNOWLEDGMENTS
This study was supported by an Ambulatory Pediatric Association/AHRQ-Special Projects Grant, Endowed Chair of Pediatric Critical Care, Children's Hospital of Philadelphia, and Templeton Foundation for Research in Pediatric Surgery.
FOOTNOTES
Accepted Jun 14, 2004.
No conflict of interest declared.
Preliminary results were presented in poster format at the Pediatric Academic Societies Meeting; May 4–7, 2004; Baltimore, MD
REFERENCES
Potoka DA, Schall LC, Gardner MJ, Stafford PW, Peitzman AB, Ford HR. Impact of pediatric trauma centers on mortality in a statewide system. J Trauma. 2000;49 :237 –245
Jacobs IA, Kelly K, Valenziano C, Pawar J, Jones C. Non-operative management of blunt splenic and hepatic trauma in the pediatric population: significant differences between adult and pediatric surgeons Am Surg. 2001;67 :149 –154
Sherman HF, Landry VL, Jones LM. Should Level I trauma centers be rated NC-17 J Trauma. 2001;50 :784 –791
Patrick D, Moore EE, Bensard DD, Karrer FM. Operative management of injured children at an adult level 1 trauma center. J Trauma. 2000;48 :894 –901
Osler T, Vane DW, Tepas JJ, et al. Survival following trauma in children; do pediatric trauma centers have better survival rates Does ACS certification matter J Trauma. 1998;45 :189
King H, Schumacher HB. Splenic studies I: susceptibility to infection after splenectomy performed in infancy. Ann Surg. 1952;136 :239 –242
Upadhyaya P, Simpson JS. Splenic trauma in children. Surgery. 1968;126 :781 –790
Knudson MM, Maull KI. Non-operative management of solid organ injuries. Past, present, and future. Surg Clin North Am. 1999;79 :1357 –1371
Partrick DA, Bensard DD, Moore EE, Karrer FM. Non-operative management of solid organ injuries in children results in decreased blood utilization. J Pediatr Surg. 1999;34 :1695 –1699
Morse MA, Garcia VF. Selective non-operative management of pediatric blunt splenic trauma: risk for missed associated injuries. J Pediatr Surg. 1994;29 :23 –27
Sjovall A, Hirsch K. Blunt abdominal trauma in children: risks of non-operative treatment. J Pediatr Surg. 1997;32 :1169 –1174
Francke EL, Neu HC. Postsplenectomy infection. Surg Clin North Am. 1981;61 :135 –155
Mooney DP, Birkmeyer NJ, Udell JV, Shorter NA. Variation in the management of pediatric splenic injuries in New Hampshire. J Pediatr Surg. 1998;33 :1076 –1078; discussion 1079–1080
Moore EE, Cogbill TH, Jurkovich GJ, Shackford SR, Malangoni MA, Champion HR. Organ injury scaling: spleen and liver (1994 revision). J Trauma. 1995;38 :323 –324
Durbin D, Schwartz DF, Localio AR, MacKenzie EJ. Validation of the ICD/AIS MAP for pediatric use. Inj Prev. 2001;7 :96 –99
Velleman PF, Hoaglin DC. Applications, Basics, and Computing of Exploratory Data Analysis. Boston, MA: Duxbury Press; 1981
Flanders WD, Rhodes PH. Large sample confidence intervals for regression standardized risks, risk ratios, and risk differences. J Chron Dis. 1987;40 :697 –704
Wesson DE, Filler RM, Ein SH, Shandling B, Simpson JS, Stephens CA. Ruptured spleen—when to operate J Pediatr Surg. 1981;16 :324 –326
Cogbill TH, Moore EE, Jurkovich GJ, et al. Non-operative management of blunt splenic trauma: a multicenter experience. J Trauma. 1989;29 :1312 –1317
Cocanour CS, Moore FA, Ware DN, Marvin RG, Duke JH. Age should not be a consideration for non-operative management of blunt splenic injury. J Trauma. 2000;48 :606 –610; discussion 610–602
Harbrecht BG, Peitzman AB, Rivera L, et al. Contribution of age and gender to outcome of blunt splenic injury in adults: multicenter study of the eastern association for the surgery of trauma. J Trauma. 2001;51 :887 –895
Bratton SL, Haberkern CM, Waldhausen JH, Sawin RS, Allison JW. Intussusception: hospital size and risk of surgery. Pediatrics. 2001;107 :299 –303
Bauchner H, Simpson L, Chessare J. Changing physician behaviour. Arch Dis Child. 2001;84 :459 –462(Daniela H. Davis, MD, MSC)