Nationwide Emergency Department Imaging Practices for Pediatric Urolithiasis: Room for Improvement Emilie K. Johnson,* Dionne A. Graham, Jeanne S. Chow and Caleb P. Nelson From the Department of Urology (EKJ, CPN), Center for Patient Safety and Quality Research (DAG), and Department of Radiology (JSC), Boston Children’s Hospital, Boston, Massachusetts
Abbreviations and Acronyms CT ¼ computerized tomography ED ¼ emergency department HCUP ¼ Healthcare Cost and Utilization Project NCCT ¼ noncontrast computerized tomography NEDS ¼ Nationwide Emergency Department Sample NHAMCS ¼ National Hospital Ambulatory Medical Care Survey SU ¼ suspected urolithiasis U.S. ¼ United States Accepted for publication January 6, 2014. Supported by NIDDK Grant K23-DK088943 (CPN) and AHRQ/ARRA Recovery Act 2009 T32 HS19485 National Research Service Award in Expanding Training in Comparative Effectiveness for Child Health Researchers (EKJ). * Correspondence: Department of Urology, Boston Children’s Hospital, 300 Longwood Ave., HU 3rd Floor, Boston, Massachusetts 02115 (telephone: 734-678-1594; FAX: 617-730-0174; e-mail: [email protected]).
Purpose: Children are particularly vulnerable to the harmful effects of medical radiation, and children with urolithiasis comprise a group that may undergo repeated radiation intensive imaging tests. We sought to characterize imaging practices for children presenting to the emergency department with suspected urolithiasis and to determine factors associated with the choice of imaging study. Materials and Methods: Using the 2006 to 2010 Nationwide Emergency Department Sample, we conducted a retrospective cohort study of patients younger than 18 years presenting with suspected urolithiasis. We determined imaging practices for visits to emergency departments where billing codes for computerized tomography and ultrasound were reliably reported. Logistic regression was used to delineate patient and hospital level factors associated with the use of computerized tomography vs ultrasound. Results: We identified 18,096 pediatric visits for suspected urolithiasis in the 1,191 Nationwide Emergency Department Sample emergency departments with reliable imaging codes. A total of 11,215 patients underwent computerized tomography alone, ultrasound alone or both. Of the patients 9,773 (87%) underwent computerized tomography alone. Computerized tomography use peaked in 2007 and declined thereafter. On multivariate analysis several factors were associated with the use of computerized tomography alone, including smaller proportion of pediatric patients treated at the emergency department, older age, location in the Midwest or South, evaluation at a nonteaching hospital and visit on a weekend. Conclusions: Computerized tomography use is highly prevalent for children presenting with suspected urolithiasis. The lowest computerized tomography use is in emergency departments that care for more children. Ultrasound is used infrequently regardless of site. Awareness regarding risks of computerized tomography and consideration of alternatives including ultrasound are warranted in caring for these patients. Key Words: emergency treatment; pediatrics; tomography, x-ray computed; ultrasonography; urolithiasis
DURING the last 2 decades a consensus has developed that noncontrast computerized tomography is the gold standard for evaluation of suspected urolithiasis.1,2 However, multiple
200
j
www.jurology.com
recent studies have established a link between radiation exposure from computerized tomography in children and subsequent cancer risk.3,4 Given such concerns, the radiology
0022-5347/14/1921-0200/0 THE JOURNAL OF UROLOGY® © 2014 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC.
http://dx.doi.org/10.1016/j.juro.2014.01.028 Vol. 192, 200-206, July 2014 Printed in U.S.A.
EMERGENCY DEPARTMENT IMAGING FOR PEDIATRIC UROLITHIASIS
community developed the ALARA (as low as reasonably achievable) principle to guide imaging choices.5,6 Previous investigations have evaluated the usefulness of alternative imaging algorithms for SU in pediatric patients, comparing NCCT with ultrasound. Although NCCT is somewhat more sensitive for stone detection, in approximately 75% of instances children can be evaluated and treated for urolithiasis without NCCT.7,8 While there has been intensive attention to ALARA and CT alternatives at pediatric centers, approximately 90% of emergency care for children in the U.S. is provided at adult focused hospitals.9 These hospital systems may not be as equipped to execute ALARA based strategies in children compared to pediatric centers. We sought to describe the national use of CT vs ultrasound imaging for pediatric patients seen at EDs for suspected urolithiasis. Our aims were to 1) determine whether pediatric focused EDs use less ionizing radiation when evaluating for suspected urolithiasis in children compared to adult focused EDs, and 2) identify patient and facility characteristics associated with increased likelihood of undergoing CT.
MATERIALS AND METHODS Study Design and Data Source We conducted a retrospective cohort study of children 0 to 17 years old evaluated at EDs throughout the U.S. for urolithiasis and/or flank pain. We used the 2006 to 2010 NEDS, a large, nationally representative database of ED visits. NEDS was developed and is maintained by the HCUP of the Agency for Healthcare Research and Quality, and consists of combined records from HCUP State ED and State Inpatient Databases. NEDS data sets are constructed annually using a sampling scheme that approximates a 20% stratified sample of hospital based EDs nationally, allowing for the calculation of national estimates of ED visit parameters. The study was determined to be exempt by the institutional review board of our hospital.
Selection of Participants We identified all ED visits where the patient was younger than 18 years and had an ICD-9 billing code for urolithiasis (592.0, 592.1 and/or 274.11) and/or renal colic (788.0). We created an algorithm to identify hospitals that reliably report codes for CT plus ultrasound within the ED. (Codes for imaging procedures performed at certain facilities may not appear in the NEDS database due to billing factors such as a private radiology group servicing a given hospital, thus generating a separate bill.) A hospital was designated as reliably coding the relevant radiological tests for a given study year if it billed for at least 1 abdominal CT and 1 abdominal ultrasound for a diagnosis of abdominal pain or appendicitis during that study year. Only patients seen at EDs at a “reliable coding” hospital were included in the final analysis cohort.
201
Variables Examined We extracted multiple patient level variables from the data set, including age, insurance status, gender, income quartile (stratified by patient zip code), disposition status, weekday vs weekend visit and comorbidities. Comorbidities were classified according to organ system based categories. There were an insufficient number of infants in the cohort to be used in the final analyses. If patients older than 19 years made up more than 90% of patients seen at a given ED for any diagnosis, then that hospital was classified as nonpediatric focused using a previously published definition.10 We also stratified hospitals by quartiles of rate of pediatric patients seen at each hospital for any diagnosis, with 1st quartile hospitals having the lowest percentage of pediatric patients and 4th quartile hospitals the highest percentage. This stratification resulted in the following groupings for the 4 quartiles. The first quartile consisted of 0% to 19.2% pediatric patients, 2nd quartile 19.3% to 23.1%, 3rd quartile 23.1% to 26.6% and 4th quartile 26.6% to 100%. Other hospital level characteristics examined included U.S. Census region, urban/rural location and teaching hospital status.
Outcome Measures The primary outcome was the imaging that occurred during each ED visit. Our goal was to compare the use of CT vs ultrasound specifically. Therefore, patients who underwent no imaging or only plain x-ray were excluded from the data set used to analyze imaging strategies. Patients who underwent x-ray in addition to CT and/or ultrasound were included and classified according to whether they underwent CT and/or ultrasound. Patients undergoing the imaging modalities of interest were then grouped as follows. Group 1 underwent CT alone, and group 2 underwent ultrasound alone or ultrasound plus CT. The rationale for including patients who underwent both imaging studies in group 2 was that clinical situations exist where an ultrasound done for SU first could be equivocal and necessitate CT for clarification or confirmation of a diagnosis. Since we were unable to determine the chronological order of the studies from this database, it seemed reasonable to assume that in cases where both imaging studies were performed the ultrasound was likely the initial study. However, it is plausible that CT was done first in some patients, with ultrasound done subsequently to establish a baseline for followup. Nevertheless, it seemed reasonable to classify these patients with others who underwent ultrasound only evaluation and to assume that ultrasound was the initial study.
Statistical Analysis Descriptive statistics were used to characterize the population of children with SU who were seen at “reliable coding” NEDS emergency departments, as well as the subset of patients contained within our analytical cohort who underwent CT alone or ultrasound with or without CT. Groups 1 (CT alone) and 2 (ultrasound with or without CT) were compared using chi-square testing to examine the bivariate relationships between choice of imaging modality and potentially relevant patient and hospital level factors. Weighted logistic regression was
202
EMERGENCY DEPARTMENT IMAGING FOR PEDIATRIC UROLITHIASIS
Table 1. Characteristics of children seen at U.S. emergency departments from 2006 to 2010 for suspected urolithiasis
Table 1 (continued ) Cohort
Cohort
No. hospitals represented/ total No. pts Yearly visits/hospital (median, IQR) No. yr (%): 2006 2007 2008 2009 2010 No. age (%): Less than 1 yr 1e4 Yrs 5e12 Yrs 13e17 Yrs Unknown No. gender (%): Male Female Unknown No. insurance status (%): Private Public Other Unknown No. comorbidities (%): Any comorbidity Cardiac Pulmonary Gastrointestinal Endocrine Oncologic Hematological Neurological Psychiatric Autoimmune No. region (%): Northeast Midwest South West No. median household income by zip code quartile (%): 1 ($1e$39,999) 2 ($40,000e$40,999) 3 ($50,000e$65,999) 4 ($66,000þ) Unknown No. pediatric focused hospital (%) No. children seen at ED (%): Quartile 1 (fewest) Quartile 2 Quartile 3 Quartile 4 (most) No. rural/urban status (%): Urban Suburban or rural No. teaching hospital (%) No. disposition status (%): Discharged home Hospitalized No. day seen (%): Weekday Weekend Unknown No. imaging strategy (%): Any imaging reported CT Ultrasound
Reliably Reported
Analytical
1,191/18,096
1,105/11,215
8 (3, 17)
6 (2, 12)
2,769 3,036 3,921 4,008 4,362
(15.3) (16.8) (21.7) (22.2) (24.1)
1,664 1,724 2,446 2,570 2,811
(14.8) (15.4) (21.8) (22.9) (25.1)
223 (1.3) 308 (1.7) 3,179 (17.6) 14,356 (79.3) 20 (0.11)
61 (0.54) 1,778 (15.9) 9,376 (83.6) 0 (0)
5,176 (37.7) 8,556 (62.3) 7 (0.01)
4,253 (37.9) 6,962 (62.1) 0 (0)
10,302 5,457 2,272 65
*
(56.9) (30.2) (12.6) (0.42)
6,720 3,017 1,441 37
(59.9) (26.9) (12.9) (0.33)
3,778 (20.9) 264 (1.5) 1,006 (5.6) 1,391 (7.7) 238 (1.3) 52 (0.29) 310 (1.7) 540 (3.0) 1,200 (6.6) 127 (0.70)
1,828 80 459 625 97 18 115 170 558 42
(16.3) (0.71) (4.1) (5.6) (0.86) (0.16) (1.0) (1.5) (5.0) (0.37)
3,994 (22.1) 1,885 (10.4) 10,462 (57.8) 1,755 (9.7)
2,485 (22.2) 1,008 (9.0) 6,706 (59.8) 1,016 (9.1)
4,630 4,722 4,306 4,109 329 620
(25.6) (26.1) (23.8) (22.7) (1.8) (3.4)
2,753 2,904 2,639 2,747 172 173
(24.6) (25.9) (23.5) (24.5) (1.5) (1.5)
3,680 4,176 4,156 6,084
(20.3) (23.1) (23.0) (33.6)
2,496 2,683 2,614 3,442
(22.3) (23.9) (23.3) (30.5)
14,997 (82.9) 3,099 (17.1) 6,753 (37.3)
9,270 (82.7) 1,945 (17.3) 3,659 (32.6)
15,709 (86.8) 2,387 (13.2)
11,023 (98.3) 192 (1.71)
13,179 (72.8) 4,916 (27.2) 1 (0.01)
8,156 (72.7) 3,058 (27.3) 1 (0.01)
11,926 (65.9) 7,905 (56.3) 1,457 (8.1)
11,215 (100) 10,185 (90.8) 1,442 (12.9)
Abdominal x-ray Excretory urography No. coded diagnosis (%): Urolithiasis with or without flank pain Flank pain only
Reliably Reported
Analytical
2,040 (11.3) 63 (0.35)
1,388 (12.4) 13 (0.12)
14,687 (81.2)
9,448 (84.2)
3,409 (18.8)
1,767 (15.8)
* Infants were excluded from analysis data set due to insufficient sample size.
then performed to determine adjusted bivariate and multivariate associations between patient and hospital level factors and the outcome of CT administration alone (vs ultrasound with or without CT). Variables were entered into our regression model based on statistical significance on bivariate analyses, taking into consideration collinearity among several hospital level factors. Data analysis was conducted using SASÒ, version 9.3. Weighted analyses were based on the NEDS stratified sampling scheme using the weight, cluster and strata variables provided by HCUP. A p value of less than 0.05 was considered significant.
RESULTS Characteristics of Study Subjects We identified 30,045 pediatric ED visits for SU, representing a weighted total of 134,834 visits nationally from 2006 to 2010 (95% CI 127,494e142,175). Of these visits 18,096 were at hospitals that reliably coded for CT and ultrasound, and 11,215 visits involved performance of CT and/or ultrasound. The characteristics of all patients seen at EDs that reliably reported radiological codes and the subset in our analysis data set are outlined in table 1. For the reliably reported cohort the majority of patients were adolescent (79.3%), female (62.3%), privately insured (56.9%) and evaluated in the South (57.8%). The most notable difference between the reliably reported and analytic cohorts was the percentage of patients admitted to the hospital (13.2% and 1.7%, respectively). Otherwise, the cohorts were similar regarding all other parameters examined. Principal Findings Figure 1 illustrates the distribution of ED imaging practices for pediatric patients with SU. Among the overall cohort 48% underwent CT alone. Other combinations of imaging were used in 17% of patients and 35% underwent no imaging during the ED visit. Overall less than 9% of patients underwent ultrasound. Among the analytic cohort patients in group 1 (CT alone, 9,773 patients, 87.1%) were compared to those in group 2 (ultrasound with or without CT, 1,442 patients, 12.9%) to determine associations
EMERGENCY DEPARTMENT IMAGING FOR PEDIATRIC UROLITHIASIS
203
Figure 1. Distribution of imaging studies for pediatric patients with suspected urolithiasis
with the use of CT alone. CT use appeared to peak in 2007, then decline gradually from 2007 to 2010 (fig. 2, chi-square trend test p
Abbreviations and Acronyms CT ¼ computerized tomography ED ¼ emergency department HCUP ¼ Healthcare Cost and Utilization Project NCCT ¼ noncontrast computerized tomography NEDS ¼ Nationwide Emergency Department Sample NHAMCS ¼ National Hospital Ambulatory Medical Care Survey SU ¼ suspected urolithiasis U.S. ¼ United States Accepted for publication January 6, 2014. Supported by NIDDK Grant K23-DK088943 (CPN) and AHRQ/ARRA Recovery Act 2009 T32 HS19485 National Research Service Award in Expanding Training in Comparative Effectiveness for Child Health Researchers (EKJ). * Correspondence: Department of Urology, Boston Children’s Hospital, 300 Longwood Ave., HU 3rd Floor, Boston, Massachusetts 02115 (telephone: 734-678-1594; FAX: 617-730-0174; e-mail: [email protected]).
Purpose: Children are particularly vulnerable to the harmful effects of medical radiation, and children with urolithiasis comprise a group that may undergo repeated radiation intensive imaging tests. We sought to characterize imaging practices for children presenting to the emergency department with suspected urolithiasis and to determine factors associated with the choice of imaging study. Materials and Methods: Using the 2006 to 2010 Nationwide Emergency Department Sample, we conducted a retrospective cohort study of patients younger than 18 years presenting with suspected urolithiasis. We determined imaging practices for visits to emergency departments where billing codes for computerized tomography and ultrasound were reliably reported. Logistic regression was used to delineate patient and hospital level factors associated with the use of computerized tomography vs ultrasound. Results: We identified 18,096 pediatric visits for suspected urolithiasis in the 1,191 Nationwide Emergency Department Sample emergency departments with reliable imaging codes. A total of 11,215 patients underwent computerized tomography alone, ultrasound alone or both. Of the patients 9,773 (87%) underwent computerized tomography alone. Computerized tomography use peaked in 2007 and declined thereafter. On multivariate analysis several factors were associated with the use of computerized tomography alone, including smaller proportion of pediatric patients treated at the emergency department, older age, location in the Midwest or South, evaluation at a nonteaching hospital and visit on a weekend. Conclusions: Computerized tomography use is highly prevalent for children presenting with suspected urolithiasis. The lowest computerized tomography use is in emergency departments that care for more children. Ultrasound is used infrequently regardless of site. Awareness regarding risks of computerized tomography and consideration of alternatives including ultrasound are warranted in caring for these patients. Key Words: emergency treatment; pediatrics; tomography, x-ray computed; ultrasonography; urolithiasis
DURING the last 2 decades a consensus has developed that noncontrast computerized tomography is the gold standard for evaluation of suspected urolithiasis.1,2 However, multiple
200
j
www.jurology.com
recent studies have established a link between radiation exposure from computerized tomography in children and subsequent cancer risk.3,4 Given such concerns, the radiology
0022-5347/14/1921-0200/0 THE JOURNAL OF UROLOGY® © 2014 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC.
http://dx.doi.org/10.1016/j.juro.2014.01.028 Vol. 192, 200-206, July 2014 Printed in U.S.A.
EMERGENCY DEPARTMENT IMAGING FOR PEDIATRIC UROLITHIASIS
community developed the ALARA (as low as reasonably achievable) principle to guide imaging choices.5,6 Previous investigations have evaluated the usefulness of alternative imaging algorithms for SU in pediatric patients, comparing NCCT with ultrasound. Although NCCT is somewhat more sensitive for stone detection, in approximately 75% of instances children can be evaluated and treated for urolithiasis without NCCT.7,8 While there has been intensive attention to ALARA and CT alternatives at pediatric centers, approximately 90% of emergency care for children in the U.S. is provided at adult focused hospitals.9 These hospital systems may not be as equipped to execute ALARA based strategies in children compared to pediatric centers. We sought to describe the national use of CT vs ultrasound imaging for pediatric patients seen at EDs for suspected urolithiasis. Our aims were to 1) determine whether pediatric focused EDs use less ionizing radiation when evaluating for suspected urolithiasis in children compared to adult focused EDs, and 2) identify patient and facility characteristics associated with increased likelihood of undergoing CT.
MATERIALS AND METHODS Study Design and Data Source We conducted a retrospective cohort study of children 0 to 17 years old evaluated at EDs throughout the U.S. for urolithiasis and/or flank pain. We used the 2006 to 2010 NEDS, a large, nationally representative database of ED visits. NEDS was developed and is maintained by the HCUP of the Agency for Healthcare Research and Quality, and consists of combined records from HCUP State ED and State Inpatient Databases. NEDS data sets are constructed annually using a sampling scheme that approximates a 20% stratified sample of hospital based EDs nationally, allowing for the calculation of national estimates of ED visit parameters. The study was determined to be exempt by the institutional review board of our hospital.
Selection of Participants We identified all ED visits where the patient was younger than 18 years and had an ICD-9 billing code for urolithiasis (592.0, 592.1 and/or 274.11) and/or renal colic (788.0). We created an algorithm to identify hospitals that reliably report codes for CT plus ultrasound within the ED. (Codes for imaging procedures performed at certain facilities may not appear in the NEDS database due to billing factors such as a private radiology group servicing a given hospital, thus generating a separate bill.) A hospital was designated as reliably coding the relevant radiological tests for a given study year if it billed for at least 1 abdominal CT and 1 abdominal ultrasound for a diagnosis of abdominal pain or appendicitis during that study year. Only patients seen at EDs at a “reliable coding” hospital were included in the final analysis cohort.
201
Variables Examined We extracted multiple patient level variables from the data set, including age, insurance status, gender, income quartile (stratified by patient zip code), disposition status, weekday vs weekend visit and comorbidities. Comorbidities were classified according to organ system based categories. There were an insufficient number of infants in the cohort to be used in the final analyses. If patients older than 19 years made up more than 90% of patients seen at a given ED for any diagnosis, then that hospital was classified as nonpediatric focused using a previously published definition.10 We also stratified hospitals by quartiles of rate of pediatric patients seen at each hospital for any diagnosis, with 1st quartile hospitals having the lowest percentage of pediatric patients and 4th quartile hospitals the highest percentage. This stratification resulted in the following groupings for the 4 quartiles. The first quartile consisted of 0% to 19.2% pediatric patients, 2nd quartile 19.3% to 23.1%, 3rd quartile 23.1% to 26.6% and 4th quartile 26.6% to 100%. Other hospital level characteristics examined included U.S. Census region, urban/rural location and teaching hospital status.
Outcome Measures The primary outcome was the imaging that occurred during each ED visit. Our goal was to compare the use of CT vs ultrasound specifically. Therefore, patients who underwent no imaging or only plain x-ray were excluded from the data set used to analyze imaging strategies. Patients who underwent x-ray in addition to CT and/or ultrasound were included and classified according to whether they underwent CT and/or ultrasound. Patients undergoing the imaging modalities of interest were then grouped as follows. Group 1 underwent CT alone, and group 2 underwent ultrasound alone or ultrasound plus CT. The rationale for including patients who underwent both imaging studies in group 2 was that clinical situations exist where an ultrasound done for SU first could be equivocal and necessitate CT for clarification or confirmation of a diagnosis. Since we were unable to determine the chronological order of the studies from this database, it seemed reasonable to assume that in cases where both imaging studies were performed the ultrasound was likely the initial study. However, it is plausible that CT was done first in some patients, with ultrasound done subsequently to establish a baseline for followup. Nevertheless, it seemed reasonable to classify these patients with others who underwent ultrasound only evaluation and to assume that ultrasound was the initial study.
Statistical Analysis Descriptive statistics were used to characterize the population of children with SU who were seen at “reliable coding” NEDS emergency departments, as well as the subset of patients contained within our analytical cohort who underwent CT alone or ultrasound with or without CT. Groups 1 (CT alone) and 2 (ultrasound with or without CT) were compared using chi-square testing to examine the bivariate relationships between choice of imaging modality and potentially relevant patient and hospital level factors. Weighted logistic regression was
202
EMERGENCY DEPARTMENT IMAGING FOR PEDIATRIC UROLITHIASIS
Table 1. Characteristics of children seen at U.S. emergency departments from 2006 to 2010 for suspected urolithiasis
Table 1 (continued ) Cohort
Cohort
No. hospitals represented/ total No. pts Yearly visits/hospital (median, IQR) No. yr (%): 2006 2007 2008 2009 2010 No. age (%): Less than 1 yr 1e4 Yrs 5e12 Yrs 13e17 Yrs Unknown No. gender (%): Male Female Unknown No. insurance status (%): Private Public Other Unknown No. comorbidities (%): Any comorbidity Cardiac Pulmonary Gastrointestinal Endocrine Oncologic Hematological Neurological Psychiatric Autoimmune No. region (%): Northeast Midwest South West No. median household income by zip code quartile (%): 1 ($1e$39,999) 2 ($40,000e$40,999) 3 ($50,000e$65,999) 4 ($66,000þ) Unknown No. pediatric focused hospital (%) No. children seen at ED (%): Quartile 1 (fewest) Quartile 2 Quartile 3 Quartile 4 (most) No. rural/urban status (%): Urban Suburban or rural No. teaching hospital (%) No. disposition status (%): Discharged home Hospitalized No. day seen (%): Weekday Weekend Unknown No. imaging strategy (%): Any imaging reported CT Ultrasound
Reliably Reported
Analytical
1,191/18,096
1,105/11,215
8 (3, 17)
6 (2, 12)
2,769 3,036 3,921 4,008 4,362
(15.3) (16.8) (21.7) (22.2) (24.1)
1,664 1,724 2,446 2,570 2,811
(14.8) (15.4) (21.8) (22.9) (25.1)
223 (1.3) 308 (1.7) 3,179 (17.6) 14,356 (79.3) 20 (0.11)
61 (0.54) 1,778 (15.9) 9,376 (83.6) 0 (0)
5,176 (37.7) 8,556 (62.3) 7 (0.01)
4,253 (37.9) 6,962 (62.1) 0 (0)
10,302 5,457 2,272 65
*
(56.9) (30.2) (12.6) (0.42)
6,720 3,017 1,441 37
(59.9) (26.9) (12.9) (0.33)
3,778 (20.9) 264 (1.5) 1,006 (5.6) 1,391 (7.7) 238 (1.3) 52 (0.29) 310 (1.7) 540 (3.0) 1,200 (6.6) 127 (0.70)
1,828 80 459 625 97 18 115 170 558 42
(16.3) (0.71) (4.1) (5.6) (0.86) (0.16) (1.0) (1.5) (5.0) (0.37)
3,994 (22.1) 1,885 (10.4) 10,462 (57.8) 1,755 (9.7)
2,485 (22.2) 1,008 (9.0) 6,706 (59.8) 1,016 (9.1)
4,630 4,722 4,306 4,109 329 620
(25.6) (26.1) (23.8) (22.7) (1.8) (3.4)
2,753 2,904 2,639 2,747 172 173
(24.6) (25.9) (23.5) (24.5) (1.5) (1.5)
3,680 4,176 4,156 6,084
(20.3) (23.1) (23.0) (33.6)
2,496 2,683 2,614 3,442
(22.3) (23.9) (23.3) (30.5)
14,997 (82.9) 3,099 (17.1) 6,753 (37.3)
9,270 (82.7) 1,945 (17.3) 3,659 (32.6)
15,709 (86.8) 2,387 (13.2)
11,023 (98.3) 192 (1.71)
13,179 (72.8) 4,916 (27.2) 1 (0.01)
8,156 (72.7) 3,058 (27.3) 1 (0.01)
11,926 (65.9) 7,905 (56.3) 1,457 (8.1)
11,215 (100) 10,185 (90.8) 1,442 (12.9)
Abdominal x-ray Excretory urography No. coded diagnosis (%): Urolithiasis with or without flank pain Flank pain only
Reliably Reported
Analytical
2,040 (11.3) 63 (0.35)
1,388 (12.4) 13 (0.12)
14,687 (81.2)
9,448 (84.2)
3,409 (18.8)
1,767 (15.8)
* Infants were excluded from analysis data set due to insufficient sample size.
then performed to determine adjusted bivariate and multivariate associations between patient and hospital level factors and the outcome of CT administration alone (vs ultrasound with or without CT). Variables were entered into our regression model based on statistical significance on bivariate analyses, taking into consideration collinearity among several hospital level factors. Data analysis was conducted using SASÒ, version 9.3. Weighted analyses were based on the NEDS stratified sampling scheme using the weight, cluster and strata variables provided by HCUP. A p value of less than 0.05 was considered significant.
RESULTS Characteristics of Study Subjects We identified 30,045 pediatric ED visits for SU, representing a weighted total of 134,834 visits nationally from 2006 to 2010 (95% CI 127,494e142,175). Of these visits 18,096 were at hospitals that reliably coded for CT and ultrasound, and 11,215 visits involved performance of CT and/or ultrasound. The characteristics of all patients seen at EDs that reliably reported radiological codes and the subset in our analysis data set are outlined in table 1. For the reliably reported cohort the majority of patients were adolescent (79.3%), female (62.3%), privately insured (56.9%) and evaluated in the South (57.8%). The most notable difference between the reliably reported and analytic cohorts was the percentage of patients admitted to the hospital (13.2% and 1.7%, respectively). Otherwise, the cohorts were similar regarding all other parameters examined. Principal Findings Figure 1 illustrates the distribution of ED imaging practices for pediatric patients with SU. Among the overall cohort 48% underwent CT alone. Other combinations of imaging were used in 17% of patients and 35% underwent no imaging during the ED visit. Overall less than 9% of patients underwent ultrasound. Among the analytic cohort patients in group 1 (CT alone, 9,773 patients, 87.1%) were compared to those in group 2 (ultrasound with or without CT, 1,442 patients, 12.9%) to determine associations
EMERGENCY DEPARTMENT IMAGING FOR PEDIATRIC UROLITHIASIS
203
Figure 1. Distribution of imaging studies for pediatric patients with suspected urolithiasis
with the use of CT alone. CT use appeared to peak in 2007, then decline gradually from 2007 to 2010 (fig. 2, chi-square trend test p
