DOI: 10.3171/2014.6.PEDS13682 ©AANS, 2014
Utility of surveillance imaging after minor blunt head trauma Clinical article Joshua J. Chern, M.D., Ph.D.,1,2 Samir Sarda, B.S.,1 Brian M. Howard, M.D., 2 Andrew Jea, M.D., 3 R. Shane Tubbs, P.A.-C., Ph.D., 4 Barunashish Brahma, M.D.,1,2 David M. Wrubel, M.D.,1,2 Andrew Reisner, M.D.,1,2 and William Boydston, M.D.1,2 Pediatric Neurosurgery Associates, Children’s Healthcare of Atlanta; 2Department of Neurosurgery, Emory University, Atlanta, Georgia; 3Department of Neurosurgery, Texas Children’s Hospital, Houston, Texas; and 4 Pediatric Neurosurgery, Children’s of Alabama, Birmingham, Alabama 1
Object. Nonoperative blunt head trauma is a common reason for admission in a pediatric hospital. Adverse events, such as growing skull fracture, are rare, and the incidence of such morbidity is not known. As a result, optimal follow-up care is not clear. Methods. Patients admitted after minor blunt head trauma between May 1, 2009, and April 30, 2013, were identified at a single institution. Demographic, socioeconomic, and clinical characteristics were retrieved from administrative and outpatient databases. Clinical events within the 180-day period following discharge were reviewed and analyzed. These events included emergency department (ED) visits, need for surgical procedures, clinic visits, and surveillance imaging utilization. Associations among these clinical events and potential contributing factors were analyzed using appropriate statistical methods. Results. There were 937 admissions for minor blunt head trauma in the 4-year period. Patients who required surgical interventions during the index admission were excluded. The average age of the admitted patients was 5.53 years, and the average length of stay was 1.7 days; 15.7% of patients were admitted for concussion symptoms with negative imaging findings, and 26.4% of patients suffered a skull fracture without intracranial injury. Patients presented with subdural, subarachnoid, or intraventricular hemorrhage in 11.6%, 9.19%, and 0.53% of cases, respectively. After discharge, 672 patients returned for at least 1 follow-up clinic visit (71.7%), and surveillance imaging was obtained at the time of the visit in 343 instances. The number of adverse events was small and consisted of 34 ED visits and 3 surgeries. Some of the ED visits could have been prevented with better discharge instructions, but none of the surgery was preventable. Furthermore, the pattern of postinjury surveillance imaging utilization correlated with physician identity but not with injury severity. Because the number of adverse events was small, surveillance imaging could not be shown to positively influence outcomes. Conclusions. Adverse events after nonoperative mild traumatic injury are rare. The routine use of postinjury surveillance imaging remains controversial, but these data suggest that such imaging does not effectively identify those who require operative intervention. (http://thejns.org/doi/abs/10.3171/2014.6.PEDS13682)
Key Words • hospital readmission • blunt head trauma • return to system • surveillance imaging • pediatrics
M
inor head trauma that results in concussive symptoms, skull fracture, or minor intracranial injury is common. Methods of diagnosis,4,11 hospital admission criteria,1,15 identification of child abuse victims,2,17 and return-to-play criteria13,14 are just some of the active areas of research. Follow-up care of pediatric patients with minor head trauma, accompanied by surveillance imaging when appropriate, is rendered by primary physicians, neurosurgeons, and sports medicine programs among others. Herein, we document the practice pattern of a large pediatric neurosurgery group in rendering care to this patient group. Adverse clinical events following minor blunt
Abbreviation used in this paper: ED = emergency department.
J Neurosurg: Pediatrics / July 11, 2014
head trauma were identified. In addition, the pattern of follow-up imaging utilization was documented and analyzed. The primary focus of this study is to provide a natural history of these patients after discharge and to estimate the incidence of adverse events. We hypothesized that given the rarity of these adverse events, postdischarge surveillance imaging does not effectively identify the patients at risk.
Study Population
Methods
There were 1155 admissions to the neurosurgery service after head trauma over a 4-year period (May 1, 2009 through April 30, 2013). Surgical procedures were per1
J. J. Chern et al. formed during 212 admissions and these were excluded. Five patients with CSF shunts and 1 patient who was undergoing treatment for a known brain tumor were excluded. Patients who were admitted to the general surgery service for multi-organ trauma were excluded. The remaining 937 patients were admitted to the neurosurgery service from the ED after minor blunt head trauma and composed the study population. The criteria for admission varied among treating physicians but were overall similar: patients were admitted if they had abnormal findings on clinical examination, abnormal findings on imaging, parental concerns, high-energy mechanism of injury, or were aged less than 1 year. Social situations may affect the decision: a patient is more likely to be admitted late in the night especially if he or she lives far away from the hospital. The reliability of the caretaker is another consideration. The lack of clearly defined admission criteria reflects the complexity of the admission decision process but is also the result of the retrospective nature of this study. Data Fields Used
Demographic and clinical characteristics were prospectively collected in the administrative and inpatient databases, including Epic ASAP (emergency department application), EpicCare Inpatient (universal hospital system), EpicCare Ambulatory (outpatient medical record application), and ADT (inpatient and outpatient admission-discharge-transfer application). Data from each database were then merged into a single operational database on an Oracle platform (Oracle 11g). Sequential hospital system visits, including ED, urgent care center, and outpatient clinic visits within 180 days of discharge were then queried and imported into a Microsoft Excel file for further analysis. Demographic data were collected at the time of admission (index admission) and included age, primary language, race/ethnicity, zip code, and primary payer status (public assistance, self-pay vs private insurance). Two data fields were derived from the zip code in conjunction with US population census data (2010): average income and whether the zip code belongs to the Atlanta metropolitan area. All of the radiology reports from studies performed during the index admission were reviewed manually for each patient. Actual images were reviewed when the reports failed to yield needed information. Postdischarge outpatient clinic, radiology department, operating room, and ED encounters were analyzed. Surveillance imaging, when acquired, was performed on the same day as the follow-up clinic visit. These were imaging studies that were ordered at the time of discharge. The decision to order imaging studies was made by discharging physicians or nurse practitioners. For the ED visits, ED and neurosurgery physician notes were reviewed in detail to determine whether the return visit was related to the index admission and whether the visit could have been prevented (preventability).
Statistical Methods
Statistical analysis was performed with IBM SPSS
2
Statistics 20. The association between dichotomous variables was analyzed with contingency tables. Binary logistic regression analysis was used to assess whether significant risk factors identified from univariate analysis independently contributed to the increased hazard of dependent variables. Tolerance statistics below 0.20 were considered to reflect colinearity. A 2-tailed p value less than 0.05 was considered statistically significant.
Results Patient Characteristics in the Study Cohort
The mean age of the included patients was 5.5 ± 5.4 (SD) years and the mean length of hospital stay was 1.7 ± 2.6 days. Commensurate with the fact that only patients with minor head trauma were included, 807 of 937 patients had a length of stay of less than 48 hours, and 873 patients had a length of stay of less than 72 hours. At presentation, 15.7% of patients were admitted for concussion symptoms with normal radiological findings, and 26.4% of patients had a skull fracture without intracranial injury. Patients presented with subdural, subarachnoid, and intraventricular hemorrhage in 11.6%, 9.19%, 0.53% of cases, respectively. None of the patients in the study cohort required surgical intervention during the index admission. Descriptive statistics and findings of imaging studies performed at the time of admission are provided in Table 1.
Follow-Up Visits and Surveillance Imaging
After discharge, 265 patients failed to attend followup clinic. In the remaining 672 patients who did return, the average interval between discharge and the follow-up clinic visit was 48 days. At the discretion of the treating neurosurgeons, imaging was obtained on the day of follow-up in 343 of the 672 clinic visits. These radiological studies included CT scans (n = 219), skull radiographs (n = 102), and MRI (n = 22). MRI studies were obtained when there were unexpected incidental findings that warranted further investigation.
Adverse Events After Discharge
ED Visits Within 30 Days of Discharge. Adverse clinical events after discharge were first categorized as related or unrelated to the initial head trauma. The related events were then categorized as care-appropriate or care-inappropriate. Care-inappropriate events were those in which a deficiency in care was identified. The rate of 30-day return to system after discharge is a commonly used quality-of-care measurement. There were 63 ED visits within 30 days following discharge. Twenty-nine visits were unrelated: the most common reasons for the visits in this category were a second, unrelated trauma (n = 11) and fever, respiratory, and gastrointestinal symptoms (n = 11). In the 34 related visits, nausea and vomiting were the presenting symptoms in 15 cases, and headache, poor sleeping, and irritability were the presenting symptoms in 8 cases. Three patients returned for facial weakness that was not present at time of discharge. As documented below, 3 of the 34 related ED visits led to surgical interventions. Careful review suggested that J Neurosurg: Pediatrics / July 11, 2014
Outcome and surveillance imaging after minor head trauma TABLE 1: Patient characteristics and other variables associated with minor blunt head trauma in 937 cases* Variable age (years) mean ± SD range length of stay (days) mean ± SD range household income per zip code ($) median range ICU admission English as primary language Caucasian Atlanta metropolitan zip codes Medicaid as primary payee imaging findings normal skull fracture only epidural hemorrhage only subdural hemorrhage only subarachnoid hemorrhage only intraventricular hemorrhage contusion/intraparenchymal hemorrhage epidural & subdural hemorrhage brain hemorrhage NOS
Value 5.53 ± 5.39 0–19.5 1.70 ± 2.56 1–62 50,432 13,084–114,674 124 (13.2) 868 (92.7) 661 (70.6) 438 (46.8) 419 (44.8) 147 (15.7) 247 (26.4) 68 (7.26) 109 (11.6) 86 (9.19) 5 (0.53) 85 (9.08) 160 (17.1) 29 (3.10)
* Values represent number of patients (%) unless otherwise specified. NOS = not otherwise specified.
more comprehensive discharge instructions and concussion education might have prevented 13 of the return ED visits. These involved patients who presented with expected scalp swelling (n = 3); mild nausea, vomiting, and headache (n = 9); and in 1 case, dried blood found in the ear canal.
Surgical Interventions. In the 6-month period following the index admission, 4 elective surgeries were completed for pathologies deemed to be unrelated to the initial injury. These included 2 cavernous malformation resections, 1 tumor resection, and 1 intracranial pressure monitor placement for incidentally found lückenschädel skull appearance. Two depressed skull fractures of the forehead were electively repaired in 2 patients and were deemed to be related, but care-appropriate, events. In both instances, the attending neurosurgeons and family observed the patients until the full resolution of forehead swelling, which revealed the cosmetic extent of the depression, before deciding to proceed with surgery. Three surgeries were related to the initial trauma, and deficiency in care was identified in these cases. A 5-month-old infant presented with a very small subdural hematoma and a nondisplaced skull fracture. The etiolJ Neurosurg: Pediatrics / July 11, 2014
ogy of the injury was determined to be child abuse. He was brought to the ED 63 days later with seizures and required a subdural-peritoneal shunt. The child missed 2 clinic appointments in the interval between discharge and surgery. In the second case, a 2-month-old child, also a victim of abuse, presented initially with a combination of a small epidural and subdural hemorrhage. He was sent to the ED by his primary care physician for evaluation of accelerated head growth at 27 days postdischarge. There was not a neurosurgery clinic appointment scheduled following the discharge. A ventriculoperitoneal shunt was placed. In the third case, a 15-year-old presented with a ruptured arachnoid cyst after minor sports-related trauma. He was observed for 4 days and was asymptomatic at time of discharge. He presented 10 days later with hemorrhage into the ruptured cyst, which required emergent evacuation. A follow-up appointment had been scheduled for 15 days after discharge. Statistical Analysis
Statistical methods, including univariate regression analysis and contingency tables, were used to detect significant relationships between demographic and clinical factors with the occurrences of adverse events (ED visits and surgeries). Age of the patient, length of hospital stay, household income, intensive care unit admission, primary language, race/ethnicity, proximity to the hospital, and Medicaid payee status all failed to correlate with the occurrence of adverse events. Types of injury found on initial imaging did not correlate with the occurrence of adverse events (Table 2, postdischarge ED visits). The pattern of surveillance imaging utilization and its impact on outcome was examined. The use of surveillance imaging after discharge significantly correlated with the identities of the attending physicians (range 20%–69%, Table 3, p < 0.001). However, unexpected ED visits after discharge were equally distributed among the groups of patients who 1) did not return to clinic for follow-up, 2) returned to clinic but did not undergo surveillance imaging, and 3) returned to clinic and underwent surveillance imaging (Table 4). Even though the number of patients who required surgical intervention is small (n = 3), because it occurred only in the group of patients that did not return for follow-up, it did reach statistical significance (p = 0.022).
Discussion Incidence of Adverse Events After Minor Head Trauma
Minor blunt head trauma is common in the pediatric population. A heightened awareness of concussion signs and symptoms together with increased use of CT imaging has resulted in more ED visits4 and, correspondingly, neurosurgeons’ involvement. The neurosurgical literature contains abundant small case series and reports documenting complications that may develop subacutely after these seemingly innocuous injuries, including growing skull fractures, 5,6,9,10 cranial nerve injuries3,8,12 and posttraumatic hydrocephalus.7,16 While one intuitively assumes that these are rare events, the actual incidence is 3
J. J. Chern et al. TABLE 2: Surveillance imaging utilization and adverse events categorized by initial imaging findings* Initial Imaging Findings
No. of Pts
Postdischarge ED Visits (surgeries)
No. of Pts w/ Surveillance Imaging
normal CT skull fracture only epidural hemorrhage subdural hemorrhage only subarachnoid hemorrhage intraventricular hemorrhage contusion/intraparenchymal hemorrhage epidural & subdural hemorrhage brain hemorrhage NOS total
147 247 68 109 86 5 85 160 30 937
3 7 (1) 5 4 (1) 4 0 4 7 (1) 0 34
33 68 44 49 32 4 35 68 10 343
* Pt = patient.
not known. In the present study, the rate of adverse events, when defined as related ED visits, was determined to be 3.8%. Severe adverse events that required surgical intervention were rarer, at 0.3% (3/938). These rates, while low, are likely over-estimations of the true rate of adverse events following minor traumatic head injury because the study cohort consisted only of patients admitted from the ED and not all patients evaluated. A substantial number of patients were evaluated at the ED and deemed to be stable for discharge instead of being admitted. Additionally, some patients were directed to outpatient clinic by the primary physicians for followup and not to ED for evaluation. Overall, the rates of adverse events, in particular those that require delayed neurosurgical intervention, are exceedingly low in pediatric patients that suffer minor traumatic head injury. Unexpected Findings
There were 2 main unexpected findings. First, even though our physicians had generally recommended outpatient follow-up in 2–3 weeks after discharge, we were surprised to find that the actual time interval was 48 days. This was a valuable finding for our practice. Upon further investigation, we found that this occurs because the outpatient clinic is usually fully booked 4 weeks ahead of time. Since this study we have begun a new clinic staffed by mid-level providers to serve this patient population. Secondly, there were 33 surveillance imaging studies TABLE 3: Physician identity and use of surveillance of imaging
4
Physician Identity
No. of Pts Attending Follow-Up Clinic
% of Pts w/ Surveillance Imaging
1 2 3 4 5 total
156 139 178 148 51 672
52.0% 42.8% 46.4% 69.4% 20.0%
obtained in 147 patients with negative initial CT scans. Review of the discharge documentation showed that at least half of the studies were reflexively ordered by discharging nurses and resident physicians without a review of the initial imaging results. The results of this study were used as education material to address this issue. Follow-Up Imaging After Minor Head Trauma
The low number of adverse events, especially those that required surgical intervention (3 cases), raises questions about the optimal way to follow these patients after discharge. The results of this study support judicious use of surveillance imaging. Overall, no patient required neurosurgical intervention based on routine surveillance imaging. In the 2 surgically treated patients who ultimately required CSF diversion (both victims of child abuse), increasing head circumference and a tense anterior fontanel would have likely heralded the need for further imaging and intervention. The third patient who underwent delayed surgical intervention was a 15-year-old boy who required craniotomy for evacuation of delayed, symptomatic subdural hematoma following rupture of an arachnoid cyst. The patient was observed in the hospital and discharged with follow-up 2 weeks later. On postdischarge Day 10 the patient became symptomatic and came to the ED. The clinical change was acute, and routine imaging in the asymptomatic postdischarge period would likely not have revealed new pathology. Acquisition of postdischarge imaging following minor head trauma remains a common practice. However, TABLE 4: Types of follow-up after discharge and adverse events Follow-Up Visit & Surveillance Imaging
No. of Pts
No. of ED Visits
No. of Surgeries
did not return for follow-up returns w/ imaging returns w/o imaging total
265 343 329 937
11 11 12 34
3 0 0 3
J Neurosurg: Pediatrics / July 11, 2014
Outcome and surveillance imaging after minor head trauma these data demonstrate that, in a large series of consecutive pediatric patients, routine imaging in the absence of clinical signs or symptoms did not lead to any further neurosurgical intervention. In this study, we reported the incidence of severe adverse events that required surgical intervention to be 0.3%. If one were to design a prospective 2-group interventional trial to alter this outcome (that is, obtain imaging only for patients under 6 months of age or a particular injury type), power analysis suggests that approximately 40,000 patients would have to be recruited to demonstrate a 50% reduction. This is not likely to be feasible or justifiable for a disease with an overwhelmingly benign course. Given the limited utility of routine imaging and the risk of additional radiation exposure, we advocate for close, outpatient follow-up and judicious use of surveillance imaging in the population of pediatric patients who suffer minor blunt head trauma. One may further suggest that all surveillance imaging after discharge should be abandoned. While the proposal may be provocative and may represent a radical change to some physicians’ current practice, our study, while lacking adequate power to support this recommendation, does not contradict such a conclusion. At-Risk Patient Groups
Close attention should be cast on very young patients and those whose initial injury was the result of child abuse. Indeed, 2 of the 3 patients who required delayed neurosurgical intervention in this series were less than 1 year of age, and both were victims of child abuse. These cohorts pose similar challenges for the neurosurgeon in that neurological changes are often more subtle. In addition, those who are the victims of child abuse are anecdotally less likely to return for follow-up at clinic due to the complexity of social issues. Consequently, special emphasis should be placed on bringing these patients back for follow-up evaluation.
Conclusions Adverse events after nonoperative mild traumatic head injury are rare in the pediatric population. This study suggests that routine surveillance imaging following discharge does not influence delayed operative intervention and close clinical observation alone may be sufficient in the great majority of cases following minor traumatic head injury. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Chern, Jea. Acquisition of data: Chern. Analysis and interpretation of data: Sarda, Tubbs. Drafting the article: Howard. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Chern.
J Neurosurg: Pediatrics / July 11, 2014
References 1. Af Geijerstam JL, Britton M, Marké LA: Mild head injury: observation or computed tomography? Economic aspects by literature review and decision analysis. Emerg Med J 21:54– 58, 2004 2. Clarke NMP, Shelton FRM, Taylor CC, Khan T, Needhirajan S: The incidence of fractures in children under the age of 24 months—in relation to non-accidental injury. Injury 43:762– 765, 2012 3. Coello AF, Canals AG, Gonzalez JM, Martín JJA: Cranial nerve injury after minor head trauma. Clinical article. J Neurosurg 113:547–555, 2010 4. Colvin JD, Thurm C, Pate BM, Newland JG, Hall M, Meehan WP III: Diagnosis and acute management of patients with concussion at children’s hospitals. Arch Dis Child 98:934– 938, 2013 5. Diyora B, Nayak N, Kamble H, Kukreja S, Gupte G, Sharma A: Surgical treatment and results in growing skull fracture. Neurol India 59:424–428, 2011 6. Donahue DJ, Sanford RA, Muhlbauer MS, Chadduck WM: Cranial burst fracture in infants: acute recognition and management. Childs Nerv Syst 11:692–697, 1995 7. Honeybul S, Ho KM: Incidence and risk factors for post-traumatic hydrocephalus following decompressive craniectomy for intractable intracranial hypertension and evacuation of mass lesions. J Neurotrauma 29:1872–1878, 2012 8. Jin H, Wang S, Hou L, Pan C, Li B, Wang H, et al: Clinical treatment of traumatic brain injury complicated by cranial nerve injury. Injury 41:918–923, 2010 9. Leung GK, Chan KH, Hung KN: Growing skull fracture in an adult nine years after blunt head trauma. J Clin Neurosci 18:855–857, 2011 10. Liu XS, You C, Lu M, Liu JG: Growing skull fracture stages and treatment strategy. Clinical article. J Neurosurg Pediatr 9:670–675, 2012 11. Morrison J, Mâsse B, Ouellet P, Décarie JC, Gravel J: Fourfilm X-ray series is more sensitive than 2-film for diagnosis of skull fractures in children. Pediatr Emerg Care 29:1189– 1193, 2013 12. Sandford AA, Davidson TM, Herrera N, Gilbert P, Magit AE, Haug K, et al: Olfactory dysfunction: a sequela of pediatric blunt head trauma. Int J Pediatr Otorhinolaryngol 70:1015– 1025, 2006 13. Sarsfield MJ, Morley EJ, Callahan JM, Grant WD, Wojcik SM: Evaluation of emergency medicine discharge instructions in pediatric head injury. Pediatr Emerg Care 29:884–887, 2013 14. Schnadower D, Vazquez H, Lee J, Dayan P, Roskind CG: Controversies in the evaluation and management of minor blunt head trauma in children. Curr Opin Pediatr 19:258–264, 2007 15. Stein SC, Burnett MG, Glick HA: Indications for CT scanning in mild traumatic brain injury: a cost-effectiveness study. J Trauma 61:558–566, 2006 16. Strahle J, Garton HJ, Maher CO, Muraszko KM, Keep RF, Xi G: Mechanisms of hydrocephalus after neonatal and adult intraventricular hemorrhage. Transl Stroke Res 3 (Suppl 1): 25–38, 2012 17. Vogelbaum MA, Kaufman BA, Park TS, Winthrop AL: Management of admissuncomplicated skull fractures in children: is hospital ion necessary? Pediatr Neurosurg 29:96–101, 1998 Manuscript submitted December 19, 2013. Accepted June 11, 2014. Please include this information when citing this paper: published online July 11, 2014; DOI: 10.3171/2014.6.PEDS13682. Address correspondence to: Joshua J. Chern, M.D., Ph.D., 5455 Meridian Mark Rd. NE, Ste. 540, Atlanta, GA 30342. email: joshua. [email protected].
5
Utility of surveillance imaging after minor blunt head trauma Clinical article Joshua J. Chern, M.D., Ph.D.,1,2 Samir Sarda, B.S.,1 Brian M. Howard, M.D., 2 Andrew Jea, M.D., 3 R. Shane Tubbs, P.A.-C., Ph.D., 4 Barunashish Brahma, M.D.,1,2 David M. Wrubel, M.D.,1,2 Andrew Reisner, M.D.,1,2 and William Boydston, M.D.1,2 Pediatric Neurosurgery Associates, Children’s Healthcare of Atlanta; 2Department of Neurosurgery, Emory University, Atlanta, Georgia; 3Department of Neurosurgery, Texas Children’s Hospital, Houston, Texas; and 4 Pediatric Neurosurgery, Children’s of Alabama, Birmingham, Alabama 1
Object. Nonoperative blunt head trauma is a common reason for admission in a pediatric hospital. Adverse events, such as growing skull fracture, are rare, and the incidence of such morbidity is not known. As a result, optimal follow-up care is not clear. Methods. Patients admitted after minor blunt head trauma between May 1, 2009, and April 30, 2013, were identified at a single institution. Demographic, socioeconomic, and clinical characteristics were retrieved from administrative and outpatient databases. Clinical events within the 180-day period following discharge were reviewed and analyzed. These events included emergency department (ED) visits, need for surgical procedures, clinic visits, and surveillance imaging utilization. Associations among these clinical events and potential contributing factors were analyzed using appropriate statistical methods. Results. There were 937 admissions for minor blunt head trauma in the 4-year period. Patients who required surgical interventions during the index admission were excluded. The average age of the admitted patients was 5.53 years, and the average length of stay was 1.7 days; 15.7% of patients were admitted for concussion symptoms with negative imaging findings, and 26.4% of patients suffered a skull fracture without intracranial injury. Patients presented with subdural, subarachnoid, or intraventricular hemorrhage in 11.6%, 9.19%, and 0.53% of cases, respectively. After discharge, 672 patients returned for at least 1 follow-up clinic visit (71.7%), and surveillance imaging was obtained at the time of the visit in 343 instances. The number of adverse events was small and consisted of 34 ED visits and 3 surgeries. Some of the ED visits could have been prevented with better discharge instructions, but none of the surgery was preventable. Furthermore, the pattern of postinjury surveillance imaging utilization correlated with physician identity but not with injury severity. Because the number of adverse events was small, surveillance imaging could not be shown to positively influence outcomes. Conclusions. Adverse events after nonoperative mild traumatic injury are rare. The routine use of postinjury surveillance imaging remains controversial, but these data suggest that such imaging does not effectively identify those who require operative intervention. (http://thejns.org/doi/abs/10.3171/2014.6.PEDS13682)
Key Words • hospital readmission • blunt head trauma • return to system • surveillance imaging • pediatrics
M
inor head trauma that results in concussive symptoms, skull fracture, or minor intracranial injury is common. Methods of diagnosis,4,11 hospital admission criteria,1,15 identification of child abuse victims,2,17 and return-to-play criteria13,14 are just some of the active areas of research. Follow-up care of pediatric patients with minor head trauma, accompanied by surveillance imaging when appropriate, is rendered by primary physicians, neurosurgeons, and sports medicine programs among others. Herein, we document the practice pattern of a large pediatric neurosurgery group in rendering care to this patient group. Adverse clinical events following minor blunt
Abbreviation used in this paper: ED = emergency department.
J Neurosurg: Pediatrics / July 11, 2014
head trauma were identified. In addition, the pattern of follow-up imaging utilization was documented and analyzed. The primary focus of this study is to provide a natural history of these patients after discharge and to estimate the incidence of adverse events. We hypothesized that given the rarity of these adverse events, postdischarge surveillance imaging does not effectively identify the patients at risk.
Study Population
Methods
There were 1155 admissions to the neurosurgery service after head trauma over a 4-year period (May 1, 2009 through April 30, 2013). Surgical procedures were per1
J. J. Chern et al. formed during 212 admissions and these were excluded. Five patients with CSF shunts and 1 patient who was undergoing treatment for a known brain tumor were excluded. Patients who were admitted to the general surgery service for multi-organ trauma were excluded. The remaining 937 patients were admitted to the neurosurgery service from the ED after minor blunt head trauma and composed the study population. The criteria for admission varied among treating physicians but were overall similar: patients were admitted if they had abnormal findings on clinical examination, abnormal findings on imaging, parental concerns, high-energy mechanism of injury, or were aged less than 1 year. Social situations may affect the decision: a patient is more likely to be admitted late in the night especially if he or she lives far away from the hospital. The reliability of the caretaker is another consideration. The lack of clearly defined admission criteria reflects the complexity of the admission decision process but is also the result of the retrospective nature of this study. Data Fields Used
Demographic and clinical characteristics were prospectively collected in the administrative and inpatient databases, including Epic ASAP (emergency department application), EpicCare Inpatient (universal hospital system), EpicCare Ambulatory (outpatient medical record application), and ADT (inpatient and outpatient admission-discharge-transfer application). Data from each database were then merged into a single operational database on an Oracle platform (Oracle 11g). Sequential hospital system visits, including ED, urgent care center, and outpatient clinic visits within 180 days of discharge were then queried and imported into a Microsoft Excel file for further analysis. Demographic data were collected at the time of admission (index admission) and included age, primary language, race/ethnicity, zip code, and primary payer status (public assistance, self-pay vs private insurance). Two data fields were derived from the zip code in conjunction with US population census data (2010): average income and whether the zip code belongs to the Atlanta metropolitan area. All of the radiology reports from studies performed during the index admission were reviewed manually for each patient. Actual images were reviewed when the reports failed to yield needed information. Postdischarge outpatient clinic, radiology department, operating room, and ED encounters were analyzed. Surveillance imaging, when acquired, was performed on the same day as the follow-up clinic visit. These were imaging studies that were ordered at the time of discharge. The decision to order imaging studies was made by discharging physicians or nurse practitioners. For the ED visits, ED and neurosurgery physician notes were reviewed in detail to determine whether the return visit was related to the index admission and whether the visit could have been prevented (preventability).
Statistical Methods
Statistical analysis was performed with IBM SPSS
2
Statistics 20. The association between dichotomous variables was analyzed with contingency tables. Binary logistic regression analysis was used to assess whether significant risk factors identified from univariate analysis independently contributed to the increased hazard of dependent variables. Tolerance statistics below 0.20 were considered to reflect colinearity. A 2-tailed p value less than 0.05 was considered statistically significant.
Results Patient Characteristics in the Study Cohort
The mean age of the included patients was 5.5 ± 5.4 (SD) years and the mean length of hospital stay was 1.7 ± 2.6 days. Commensurate with the fact that only patients with minor head trauma were included, 807 of 937 patients had a length of stay of less than 48 hours, and 873 patients had a length of stay of less than 72 hours. At presentation, 15.7% of patients were admitted for concussion symptoms with normal radiological findings, and 26.4% of patients had a skull fracture without intracranial injury. Patients presented with subdural, subarachnoid, and intraventricular hemorrhage in 11.6%, 9.19%, 0.53% of cases, respectively. None of the patients in the study cohort required surgical intervention during the index admission. Descriptive statistics and findings of imaging studies performed at the time of admission are provided in Table 1.
Follow-Up Visits and Surveillance Imaging
After discharge, 265 patients failed to attend followup clinic. In the remaining 672 patients who did return, the average interval between discharge and the follow-up clinic visit was 48 days. At the discretion of the treating neurosurgeons, imaging was obtained on the day of follow-up in 343 of the 672 clinic visits. These radiological studies included CT scans (n = 219), skull radiographs (n = 102), and MRI (n = 22). MRI studies were obtained when there were unexpected incidental findings that warranted further investigation.
Adverse Events After Discharge
ED Visits Within 30 Days of Discharge. Adverse clinical events after discharge were first categorized as related or unrelated to the initial head trauma. The related events were then categorized as care-appropriate or care-inappropriate. Care-inappropriate events were those in which a deficiency in care was identified. The rate of 30-day return to system after discharge is a commonly used quality-of-care measurement. There were 63 ED visits within 30 days following discharge. Twenty-nine visits were unrelated: the most common reasons for the visits in this category were a second, unrelated trauma (n = 11) and fever, respiratory, and gastrointestinal symptoms (n = 11). In the 34 related visits, nausea and vomiting were the presenting symptoms in 15 cases, and headache, poor sleeping, and irritability were the presenting symptoms in 8 cases. Three patients returned for facial weakness that was not present at time of discharge. As documented below, 3 of the 34 related ED visits led to surgical interventions. Careful review suggested that J Neurosurg: Pediatrics / July 11, 2014
Outcome and surveillance imaging after minor head trauma TABLE 1: Patient characteristics and other variables associated with minor blunt head trauma in 937 cases* Variable age (years) mean ± SD range length of stay (days) mean ± SD range household income per zip code ($) median range ICU admission English as primary language Caucasian Atlanta metropolitan zip codes Medicaid as primary payee imaging findings normal skull fracture only epidural hemorrhage only subdural hemorrhage only subarachnoid hemorrhage only intraventricular hemorrhage contusion/intraparenchymal hemorrhage epidural & subdural hemorrhage brain hemorrhage NOS
Value 5.53 ± 5.39 0–19.5 1.70 ± 2.56 1–62 50,432 13,084–114,674 124 (13.2) 868 (92.7) 661 (70.6) 438 (46.8) 419 (44.8) 147 (15.7) 247 (26.4) 68 (7.26) 109 (11.6) 86 (9.19) 5 (0.53) 85 (9.08) 160 (17.1) 29 (3.10)
* Values represent number of patients (%) unless otherwise specified. NOS = not otherwise specified.
more comprehensive discharge instructions and concussion education might have prevented 13 of the return ED visits. These involved patients who presented with expected scalp swelling (n = 3); mild nausea, vomiting, and headache (n = 9); and in 1 case, dried blood found in the ear canal.
Surgical Interventions. In the 6-month period following the index admission, 4 elective surgeries were completed for pathologies deemed to be unrelated to the initial injury. These included 2 cavernous malformation resections, 1 tumor resection, and 1 intracranial pressure monitor placement for incidentally found lückenschädel skull appearance. Two depressed skull fractures of the forehead were electively repaired in 2 patients and were deemed to be related, but care-appropriate, events. In both instances, the attending neurosurgeons and family observed the patients until the full resolution of forehead swelling, which revealed the cosmetic extent of the depression, before deciding to proceed with surgery. Three surgeries were related to the initial trauma, and deficiency in care was identified in these cases. A 5-month-old infant presented with a very small subdural hematoma and a nondisplaced skull fracture. The etiolJ Neurosurg: Pediatrics / July 11, 2014
ogy of the injury was determined to be child abuse. He was brought to the ED 63 days later with seizures and required a subdural-peritoneal shunt. The child missed 2 clinic appointments in the interval between discharge and surgery. In the second case, a 2-month-old child, also a victim of abuse, presented initially with a combination of a small epidural and subdural hemorrhage. He was sent to the ED by his primary care physician for evaluation of accelerated head growth at 27 days postdischarge. There was not a neurosurgery clinic appointment scheduled following the discharge. A ventriculoperitoneal shunt was placed. In the third case, a 15-year-old presented with a ruptured arachnoid cyst after minor sports-related trauma. He was observed for 4 days and was asymptomatic at time of discharge. He presented 10 days later with hemorrhage into the ruptured cyst, which required emergent evacuation. A follow-up appointment had been scheduled for 15 days after discharge. Statistical Analysis
Statistical methods, including univariate regression analysis and contingency tables, were used to detect significant relationships between demographic and clinical factors with the occurrences of adverse events (ED visits and surgeries). Age of the patient, length of hospital stay, household income, intensive care unit admission, primary language, race/ethnicity, proximity to the hospital, and Medicaid payee status all failed to correlate with the occurrence of adverse events. Types of injury found on initial imaging did not correlate with the occurrence of adverse events (Table 2, postdischarge ED visits). The pattern of surveillance imaging utilization and its impact on outcome was examined. The use of surveillance imaging after discharge significantly correlated with the identities of the attending physicians (range 20%–69%, Table 3, p < 0.001). However, unexpected ED visits after discharge were equally distributed among the groups of patients who 1) did not return to clinic for follow-up, 2) returned to clinic but did not undergo surveillance imaging, and 3) returned to clinic and underwent surveillance imaging (Table 4). Even though the number of patients who required surgical intervention is small (n = 3), because it occurred only in the group of patients that did not return for follow-up, it did reach statistical significance (p = 0.022).
Discussion Incidence of Adverse Events After Minor Head Trauma
Minor blunt head trauma is common in the pediatric population. A heightened awareness of concussion signs and symptoms together with increased use of CT imaging has resulted in more ED visits4 and, correspondingly, neurosurgeons’ involvement. The neurosurgical literature contains abundant small case series and reports documenting complications that may develop subacutely after these seemingly innocuous injuries, including growing skull fractures, 5,6,9,10 cranial nerve injuries3,8,12 and posttraumatic hydrocephalus.7,16 While one intuitively assumes that these are rare events, the actual incidence is 3
J. J. Chern et al. TABLE 2: Surveillance imaging utilization and adverse events categorized by initial imaging findings* Initial Imaging Findings
No. of Pts
Postdischarge ED Visits (surgeries)
No. of Pts w/ Surveillance Imaging
normal CT skull fracture only epidural hemorrhage subdural hemorrhage only subarachnoid hemorrhage intraventricular hemorrhage contusion/intraparenchymal hemorrhage epidural & subdural hemorrhage brain hemorrhage NOS total
147 247 68 109 86 5 85 160 30 937
3 7 (1) 5 4 (1) 4 0 4 7 (1) 0 34
33 68 44 49 32 4 35 68 10 343
* Pt = patient.
not known. In the present study, the rate of adverse events, when defined as related ED visits, was determined to be 3.8%. Severe adverse events that required surgical intervention were rarer, at 0.3% (3/938). These rates, while low, are likely over-estimations of the true rate of adverse events following minor traumatic head injury because the study cohort consisted only of patients admitted from the ED and not all patients evaluated. A substantial number of patients were evaluated at the ED and deemed to be stable for discharge instead of being admitted. Additionally, some patients were directed to outpatient clinic by the primary physicians for followup and not to ED for evaluation. Overall, the rates of adverse events, in particular those that require delayed neurosurgical intervention, are exceedingly low in pediatric patients that suffer minor traumatic head injury. Unexpected Findings
There were 2 main unexpected findings. First, even though our physicians had generally recommended outpatient follow-up in 2–3 weeks after discharge, we were surprised to find that the actual time interval was 48 days. This was a valuable finding for our practice. Upon further investigation, we found that this occurs because the outpatient clinic is usually fully booked 4 weeks ahead of time. Since this study we have begun a new clinic staffed by mid-level providers to serve this patient population. Secondly, there were 33 surveillance imaging studies TABLE 3: Physician identity and use of surveillance of imaging
4
Physician Identity
No. of Pts Attending Follow-Up Clinic
% of Pts w/ Surveillance Imaging
1 2 3 4 5 total
156 139 178 148 51 672
52.0% 42.8% 46.4% 69.4% 20.0%
obtained in 147 patients with negative initial CT scans. Review of the discharge documentation showed that at least half of the studies were reflexively ordered by discharging nurses and resident physicians without a review of the initial imaging results. The results of this study were used as education material to address this issue. Follow-Up Imaging After Minor Head Trauma
The low number of adverse events, especially those that required surgical intervention (3 cases), raises questions about the optimal way to follow these patients after discharge. The results of this study support judicious use of surveillance imaging. Overall, no patient required neurosurgical intervention based on routine surveillance imaging. In the 2 surgically treated patients who ultimately required CSF diversion (both victims of child abuse), increasing head circumference and a tense anterior fontanel would have likely heralded the need for further imaging and intervention. The third patient who underwent delayed surgical intervention was a 15-year-old boy who required craniotomy for evacuation of delayed, symptomatic subdural hematoma following rupture of an arachnoid cyst. The patient was observed in the hospital and discharged with follow-up 2 weeks later. On postdischarge Day 10 the patient became symptomatic and came to the ED. The clinical change was acute, and routine imaging in the asymptomatic postdischarge period would likely not have revealed new pathology. Acquisition of postdischarge imaging following minor head trauma remains a common practice. However, TABLE 4: Types of follow-up after discharge and adverse events Follow-Up Visit & Surveillance Imaging
No. of Pts
No. of ED Visits
No. of Surgeries
did not return for follow-up returns w/ imaging returns w/o imaging total
265 343 329 937
11 11 12 34
3 0 0 3
J Neurosurg: Pediatrics / July 11, 2014
Outcome and surveillance imaging after minor head trauma these data demonstrate that, in a large series of consecutive pediatric patients, routine imaging in the absence of clinical signs or symptoms did not lead to any further neurosurgical intervention. In this study, we reported the incidence of severe adverse events that required surgical intervention to be 0.3%. If one were to design a prospective 2-group interventional trial to alter this outcome (that is, obtain imaging only for patients under 6 months of age or a particular injury type), power analysis suggests that approximately 40,000 patients would have to be recruited to demonstrate a 50% reduction. This is not likely to be feasible or justifiable for a disease with an overwhelmingly benign course. Given the limited utility of routine imaging and the risk of additional radiation exposure, we advocate for close, outpatient follow-up and judicious use of surveillance imaging in the population of pediatric patients who suffer minor blunt head trauma. One may further suggest that all surveillance imaging after discharge should be abandoned. While the proposal may be provocative and may represent a radical change to some physicians’ current practice, our study, while lacking adequate power to support this recommendation, does not contradict such a conclusion. At-Risk Patient Groups
Close attention should be cast on very young patients and those whose initial injury was the result of child abuse. Indeed, 2 of the 3 patients who required delayed neurosurgical intervention in this series were less than 1 year of age, and both were victims of child abuse. These cohorts pose similar challenges for the neurosurgeon in that neurological changes are often more subtle. In addition, those who are the victims of child abuse are anecdotally less likely to return for follow-up at clinic due to the complexity of social issues. Consequently, special emphasis should be placed on bringing these patients back for follow-up evaluation.
Conclusions Adverse events after nonoperative mild traumatic head injury are rare in the pediatric population. This study suggests that routine surveillance imaging following discharge does not influence delayed operative intervention and close clinical observation alone may be sufficient in the great majority of cases following minor traumatic head injury. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Chern, Jea. Acquisition of data: Chern. Analysis and interpretation of data: Sarda, Tubbs. Drafting the article: Howard. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Chern.
J Neurosurg: Pediatrics / July 11, 2014
References 1. Af Geijerstam JL, Britton M, Marké LA: Mild head injury: observation or computed tomography? Economic aspects by literature review and decision analysis. Emerg Med J 21:54– 58, 2004 2. Clarke NMP, Shelton FRM, Taylor CC, Khan T, Needhirajan S: The incidence of fractures in children under the age of 24 months—in relation to non-accidental injury. Injury 43:762– 765, 2012 3. Coello AF, Canals AG, Gonzalez JM, Martín JJA: Cranial nerve injury after minor head trauma. Clinical article. J Neurosurg 113:547–555, 2010 4. Colvin JD, Thurm C, Pate BM, Newland JG, Hall M, Meehan WP III: Diagnosis and acute management of patients with concussion at children’s hospitals. Arch Dis Child 98:934– 938, 2013 5. Diyora B, Nayak N, Kamble H, Kukreja S, Gupte G, Sharma A: Surgical treatment and results in growing skull fracture. Neurol India 59:424–428, 2011 6. Donahue DJ, Sanford RA, Muhlbauer MS, Chadduck WM: Cranial burst fracture in infants: acute recognition and management. Childs Nerv Syst 11:692–697, 1995 7. Honeybul S, Ho KM: Incidence and risk factors for post-traumatic hydrocephalus following decompressive craniectomy for intractable intracranial hypertension and evacuation of mass lesions. J Neurotrauma 29:1872–1878, 2012 8. Jin H, Wang S, Hou L, Pan C, Li B, Wang H, et al: Clinical treatment of traumatic brain injury complicated by cranial nerve injury. Injury 41:918–923, 2010 9. Leung GK, Chan KH, Hung KN: Growing skull fracture in an adult nine years after blunt head trauma. J Clin Neurosci 18:855–857, 2011 10. Liu XS, You C, Lu M, Liu JG: Growing skull fracture stages and treatment strategy. Clinical article. J Neurosurg Pediatr 9:670–675, 2012 11. Morrison J, Mâsse B, Ouellet P, Décarie JC, Gravel J: Fourfilm X-ray series is more sensitive than 2-film for diagnosis of skull fractures in children. Pediatr Emerg Care 29:1189– 1193, 2013 12. Sandford AA, Davidson TM, Herrera N, Gilbert P, Magit AE, Haug K, et al: Olfactory dysfunction: a sequela of pediatric blunt head trauma. Int J Pediatr Otorhinolaryngol 70:1015– 1025, 2006 13. Sarsfield MJ, Morley EJ, Callahan JM, Grant WD, Wojcik SM: Evaluation of emergency medicine discharge instructions in pediatric head injury. Pediatr Emerg Care 29:884–887, 2013 14. Schnadower D, Vazquez H, Lee J, Dayan P, Roskind CG: Controversies in the evaluation and management of minor blunt head trauma in children. Curr Opin Pediatr 19:258–264, 2007 15. Stein SC, Burnett MG, Glick HA: Indications for CT scanning in mild traumatic brain injury: a cost-effectiveness study. J Trauma 61:558–566, 2006 16. Strahle J, Garton HJ, Maher CO, Muraszko KM, Keep RF, Xi G: Mechanisms of hydrocephalus after neonatal and adult intraventricular hemorrhage. Transl Stroke Res 3 (Suppl 1): 25–38, 2012 17. Vogelbaum MA, Kaufman BA, Park TS, Winthrop AL: Management of admissuncomplicated skull fractures in children: is hospital ion necessary? Pediatr Neurosurg 29:96–101, 1998 Manuscript submitted December 19, 2013. Accepted June 11, 2014. Please include this information when citing this paper: published online July 11, 2014; DOI: 10.3171/2014.6.PEDS13682. Address correspondence to: Joshua J. Chern, M.D., Ph.D., 5455 Meridian Mark Rd. NE, Ste. 540, Atlanta, GA 30342. email: joshua. [email protected].
5
