Error bars indicate 95% confidence interval.
Korley FK, Pham JC, Kirsch TD. Use of Advanced Radiology During Visits to US Emergency Departments for Injury-Related Conditions, 1998-2007. JAMA. 2010;304(13):1465–1471. doi:10.1001/jama.2010.1408
Author Affiliations: Departments of Emergency Medicine (Drs Korley, Pham, and Kirsch), and Anesthesia and Critical Care Medicine (Dr Pham), School of Medicine, and Bloomberg School of Public Health (Dr Kirsch), Johns Hopkins University, Baltimore, Maryland.
Context Excessive use of medical imaging increases health care costs and exposure to ionizing radiation (a potential carcinogen) without yielding significant benefits to all patients.
Objective To determine whether there has been a change in the prevalence of emergency department visits for injury-related conditions for which computed tomography (CT) or magnetic resonance imaging (MRI) was obtained and whether there has been a change in the diagnosis of life-threatening conditions and patient disposition.
Design, Setting, and Participants Retrospective cross-sectional analysis of emergency department visits using data from the National Hospital Ambulatory Medical Care Survey (1998-2007). Sampled visits were weighted to produce estimates for the United States.
Main Outcomes Measures Proportion of visits for injury-related conditions during which a CT or MRI was obtained, a life-threatening condition was diagnosed (eg, cervical spine fracture, skull fracture, intracranial bleeding, liver and spleen laceration), and which resulted in hospital and intensive care unit admission.
Results The prevalence of CT or MRI use during emergency department visits for injury-related conditions increased from 6% (95% confidence interval [CI], 5%-7%) (257 of 5237 visits) in 1998 to 15% (95% CI, 14%-17%) (981 of 6567 visits) in 2007 (P < .001 for trend). There was a small increase in the prevalence of life-threatening conditions (1.7% [95% CI, 1.2%-2.2%; 59 of 5237 visits] in 1998 and 2.0% [95% CI, 1.6%-2.5%; 142 of 6567 visits] in 2007; P = .04 for trend). There was no change in prevalence of visits during which patients were either admitted to the hospital (5.9% [95% CI, 4.9%-6.9%] in 1998 and 5.5% [95% CI, 4.7%-6.5%] in 2007; P = .50 for trend) or to an intensive care unit (0.62% [95% CI, 0.40%-1.00%] in 1998 and 0.80% [95% CI, 0.53%-1.21%] in 2007; P = .14 for trend). Visits during which CT or MRI was obtained lasted 126 minutes (95% CI, 123-131 minutes) longer than those for which CT or MRI was not obtained.
Conclusion From 1998 to 2007, the prevalence of CT or MRI use during emergency department visits for injury-related conditions increased significantly, without an equal increase in the prevalence of life-threatening conditions.
Injury-related conditions are among the most common reasons for visits to emergency departments in the United States. The most frequent injury mechanisms are unintentional falls and motor vehicle collisions.1 Some patients with injury-related emergency department visits are diagnosed with potentially life-threatening injuries, but the majority of patients sustain more benign conditions such as minor head injury, sprains, and contusions.
The widespread availability of advanced radiology (computed tomography [CT] and magnetic resonance imaging [MRI]) and the associated diagnostic superiority in identifying significant injuries have made these tools important in the evaluation of patients presenting to emergency departments. Several studies have reported increases in CT use in emergency departments,2- 7 although only a small percentage of these scans yield clinically significant findings.8- 10 Increased use of advanced radiology procedures is associated with increased health care expenditures, increased length of stay in the emergency department, and increased exposure to ionizing radiation.
To date, no studies of which we are aware have evaluated national trends in the use of advanced radiology during emergency department visits for injury-related conditions. Analyzing this trend is important because approximately 70% of injury-related visits are by persons who are younger than 45 years,1 and this group is at the greatest risk for potential long-term oncological effects of ionizing radiation exposure. It is also not known whether the use of advanced radiology has led to an increase in the diagnoses of life-threatening traumatic injuries in this patient population or a change in their disposition.
In this study, we investigated the national trends in the use of advanced radiology during emergency department visits for injury-related conditions over a 10-year period (1998-2007). We hypothesize that there has been an increase in the prevalence of CT or MRI use in emergency department visits for injury-related conditions without an equal increase in the prevalence of the diagnosis of life-threatening conditions.
We performed a retrospective cross-sectional analysis of CT or MRI use in patients with injury-related conditions who presented to emergency departments in the United States from 1998 to 2007. The institutional review board of the Johns Hopkins University School of Medicine reviewed the study protocol and determined it to be exempt from further review.
We analyzed data from the National Hospital Ambulatory Medical Care Survey, which is an annual, national probability-based sample of visits to emergency departments and outpatient departments of noninstitutional general and short-stay hospitals located in the 50 states, including the District of Columbia (excluding federal, military, and Veterans Affairs hospitals). Conducted since 1992 by the US Centers for Disease Control and Prevention's (CDC’s) National Center for Health Statistics, this survey uses a previously described 4-stage probability sampling design11 to collect data on the use and provision of ambulatory care services. Participating hospital staff is trained to collect data on a systematically selected sample of patient visits during a randomly assigned 4-week period each year. Patient charts are reviewed and relevant data are abstracted using a standardized patient record form. A field representative from the US Bureau of Census reviews the records used for visit sampling to determine if any cases are missing and also reviews completed forms to check for missing data. The data are then processed and coded by an independent company. The CDC's National Center for Health Statistics research ethics review board waived informed consent for study participants.
To identify visits for injury-related conditions, we selected visits in which at least one of a patient's verbatim reason for visit was coded as “injury” by type and/or location using the Reason for Visit Classification for Ambulatory Care codes 5001-5830.12 The primary outcome was the proportion of emergency department visits for injury-related conditions during which a CT or MRI was obtained. During the years when CT and MRI use data were reported separately (1998-2000 and 2005-2007), we combined the 2 variables into 1 outcome. The secondary outcomes included the proportion of visits during which a life-threatening condition was diagnosed (condition defined as any listed diagnosis[es] of a skull fracture [800.x-802.xx, 803.x-804.xx], cervical spine fractures [805.xx-806.10], intracranial hemorrhage [852.xx-853.xx], liver lacerations [864.xx], and/or spleen lacerations [865.xx] using codes from the International Classification of Diseases, Ninth Revision13). We chose these diagnoses because they are the most common life-threatening injury-related conditions that are diagnosed using CT.14 We determined the proportion of emergency department visits for injury-related conditions that resulted in admission to a hospital or an intensive care unit, and the mean lengths of stay when CT or MRI was obtained and when CT or MRI was not obtained.
We assessed whether there was a change in the following demographic and visit characteristics within the study population during the 10-year study period: age, sex, race, and insurance status, immediacy with which a patient should be seen, pain level, and hospital teaching status. Age was classified into 5 categories based on social milestones and frequency of presentation with injury-related conditions. We included patients who were younger than 3 years in a separate category because they are generally unable to verbalize symptoms reliably. Insured was defined as having Medicare, Medicaid, State Children's Health Insurance Program, worker's compensation, or private insurance as the expected source of payment for the emergency department visit. Uninsured was defined as self-pay, no charge, charity, other, or unknown. Based on the approach used in a prior study, we defined an academic emergency department as one in which a resident or intern saw more than 50% of the patients.15 In the categorization of race, there were small sample sizes for Asians, Native Hawaiian or other Pacific Islanders, and American Indian or Alaskan natives so these were combined into an “other” category to provide reliable estimates.
Data analysis was performed using Stata statistical software version 8.2 (StataCorp, College Station, Texas). To account for the survey sampling design, we used the svy set of commands from Stata. Each sampled patient visit was weighted to produce reliable national estimates. We report demographic and visit characteristics using point estimates and their corresponding 95% confidence intervals (CIs).
We hypothesized that there has been a 20% or more increase in the prevalence of CT or MRI use during emergency department visits for injury-related conditions, without an equal increase in the prevalence of life-threatening conditions. To determine whether there was a change in the demographics and visit characteristics of our study population during the study period, and whether there was a change in the proportion of visits for injury-related conditions during which advanced imaging was obtained, we performed trend tests using logistic regression analysis, and report P values for each trend. We used the t test to determine if there was a significant difference between 2 continuous variables (length of visit) and the Pearson χ2 test to determine if there was a significant difference between proportions.
To determine whether any patient or visit characteristics were independently associated with advanced radiology use, we constructed a multivariable logistic regression model. We included covariables associated with the predictor (year) and outcome (CT or MRI use). We excluded covariables with a P value of less than .05 on univariate analysis. The model was tested for colinearity by measuring the variance inflation factor. We removed all variables with a variance inflation factor of greater than 10 (there were none). The final model included the variables of age, insurance status, immediacy with which a patient should be seen, and pain level at presentation. A 2-tailed P value of less than .05 was considered statistically significant.
Between 1998 and 2007, a total of 324 569 emergency department visits were sampled. Of these, 65 376 presented with injury-related conditions (20%; 95% CI, 20%-21%). This represents an estimated average of 22.4 million (95% CI, 20.9-23.9 million) visits made to emergency departments in the United States each year for injury-related conditions. Visits were sampled from a mean of 370 hospitals per year (95% CI, 355-386 hospitals/year). There were variations in the distributions of age, sex, insurance status, pain level, and immediacy with which a patient should be seen during the study period. However, race and teaching institution status did not vary significantly (Table 1).
Of the 5237 visits for injury-related conditions sampled in 1998, 257 patients received CT or MRI (6%; 95% CI, 5%-7%). In 2007, of the 6567 visits sampled, 981 patients received CT or MRI (15%; 95% CI, 14%-17%) (P < .001 for trend). A similar increase was noted after adjusting for confounders (age, insurance status, pain, and immediacy with which a patient should be seen). Compared with patients who presented to the emergency department for an injury-related condition in 1998, those who presented to the emergency department for injury-related conditions in 2007 were significantly more likely (adjusted odds ratio [AOR], 3.43; 95% CI, 2.71-4.35) to receive CT or MRI. During the years when MRI and CT use data were collected separately (1998-2000 and 2005-2007), MRI use constituted only 0.2% (95% CI, 0.1%-0.4%) of the combined data for CT or MRI use. Hence, the increase in CT or MRI use effectively represents an increase in CT use.
A life-threatening condition was diagnosed in 59 of 5237 sampled visits (1.7%; 95% CI, 1.2%-2.2%) in 1998 compared with 142 of 6567 visits (2.0%; 95% CI, 1.6%-2.5%) in 2007 (P = .04 for trend; AOR for 2007 vs 1998, 1.20 [95% CI, 0.81-1.79]). There was no significant change in the proportion of injury-related visits that resulted in admission to the hospital (5.9% [95% CI, 4.9%-6.9%] in 1998 and 5.5% [95% CI, 4.7%-6.5%] in 2007; P = .50 for trend; AOR for 2007 vs 1998, 0.94 [95% CI, 0.73-1.20]) or admission to the intensive care unit (0.62% [95% CI, 0.40%-1.00%] in 1998 and 0.80% [95% CI, 0.53%-1.21%] in 2007; P = .14 for trend; AOR for 2007 vs 1998, 1.28 [95% CI, 0.69-2.40]; Figure).
Among the entire sample, there was no relationship between race and sex for CT or MRI use for injury-related conditions (Table 2). Patients aged 60 years or older were more likely to receive CT or MRI than those between the ages of 18 and 45 years. Patients aged 18 years or younger were less likely to receive a CT or MRI than those between the ages of 18 and 45 years. However, for patients between the ages of 3 and 18 years, the odds of getting a CT or MRI in 2007 was 2.16 times that in 1998 (4.4% [95% CI, 3.2%-6.0%] in 1998 and 9.1% [95% CI, 7.0%-11.9%] in 2007; P <.001 for trend).
Patients with severe pain were more likely than those with mild pain to receive CT or MRI. There was a higher likelihood for patients presenting to academic emergency departments to receive CT or MRI than those presenting to nonacademic emergency departments (OR, 1.55 [95% CI, 1.28-1.88]; AOR, 1.52 [95% CI, 1.22-1.90]). Length of stay was recorded in the National Hospital Ambulatory Medical Care Survey since 2001. The mean difference in length of stay for visits for injury-related conditions during which CT or MRI was obtained was 126 minutes (95% CI, 123-131 minutes) longer than for similar visits in which CT or MRI was not obtained (Table 3).
In this national probability-based sample, we report a 3-fold increase in the prevalence of CT or MRI obtained during emergency department visits for injury-related conditions between 1998 and 2007, without an equal increase in the prevalence of the diagnosis of life-threatening conditions or a change in the disposition of patients. Increase in CT use accounted for the majority of the increased CT or MRI use. A prior study demonstrated increases in CT use for patients with abdominal pain without any increase in the rate of diagnosis of significant intra-abdominal conditions.7 Some factors that may have contributed to this significant increase in CT use are the superiority of CT scans over x-rays for diagnosing conditions such as cervical spine fractures,16 the routine use of whole-body scanning for patients treated in some trauma centers,17- 19 the increased availability of CT scanners,20,21 the proximity of CT scanners to the patient care areas of most emergency departments, the speed of new-generation CT scanners leading to a decrease in the need to sedate pediatric patients,22 and concern about malpractice lawsuits for a missed diagnosis. Other reported contributors to increased use of advanced imaging are patient demand from direct-to-consumer advertising and in non–emergency department settings, the ability of physicians to refer patients to their own practices for imaging.23
Computed tomographic scans are more sensitive for detecting serious injuries, but monetary and nonmonetary costs are associated with their use. According to the US Government Accountability Office, annual spending on CT imaging more than doubled from 2000 to 2007 from $975 million to $2171 million. During the same period, spending for standard imaging (plain radiographs and ultrasounds) increased by 65%.23
Computed tomography also has other health risks associated with its use. Physiological risks include the potential for contrast-induced nephropathy and allergic reactions to the contrast, some of which may be life threatening.24 There is also a long-term risk of developing cancers, with a linear dose-response relationship between exposure to ionizing radiation and the development of certain neoplasms.25 It has been suggested that an estimated 1.0% to 2.0% of all cancers in the United Kingdom and the United States may be associated with exposure to ionizing radiation from CT.22,26 The carcinogenic effects of exposure to CT are especially important in patients presenting with injury-related conditions because they tend to be younger and receive multiple CTs.27 The 2006 National Academies' Biological Effects of Ionizing Radiation 7th Report25 estimated that 1 in 1000 individuals exposed to 10 mSv of radiation will develop a solid cancer or leukemia during their lifetime. This number is not trivial, given that approximately 15% of patients who presented with injury-related conditions in 2007 received a CT or MRI (the majority was CT). The risk is even higher in patients with injury-related conditions who receive a combination of CT scans for the head (2 mSv), cervical spine (6 mSv), chest (15 mSv), abdomen (8 mSv), and pelvis (6 mSv).26
Until recently, there has been little discussion in the emergency medicine community about the carcinogenic potential of ionizing radiation from CT. In a 2004 survey of emergency department physicians, only 9% were aware that CT can increase a patient's lifetime risk of cancer.28 The indirect effect of CT use can be increased length of visit in the emergency department29 due to the long waiting period for imaging results. This can contribute to emergency department crowding and can increase the risk of medical error. There are evidence-based decision rules for CT use in adult and pediatric patients with suspected cervical spine and brain injuries.30- 32 However, the level of adherence to these decision rules during routine clinical practice is unknown.
In our study, the increase in use of CT or MRI was less in the pediatric population than any other age group, but prevalence of use doubled between 1998 and 2007. Studies have supported the use of CT in the pediatric population for evaluating pediatric cervical spine injuries, appendicitis, and other conditions.33,34 Other studies have shown that between 2000 and 2006, use of cervical spine CT increased by 366% and chest CT by 435% in the pediatric population.2 For pediatric patients with head trauma, the rate of cranial CT has increased from 12.8% in 1995 to 28.6% in 2003.35 Despite the increased indications for CT in the pediatric population, it is possible that the smaller rate of increase in our study may reflect concern among physicians of the problem of radiation exposure.
The proportion of visits for injury-related conditions to academic emergency departments (a resident or intern sees >50% of patients) during which CT or MRI was obtained was significantly greater than in nonacademic emergency departments. This may be due to a number of factors including more severely injured patients at academic emergency departments, less experienced clinicians (residents) ordering tests, and increased availability of CT or MRI.36
Although our study was not designed to determine the association between use of CT or MRI and emergency department length of visit, injury-related visits during which CT or MRI was obtained lasted 2 hours more than those in which CT or MRI was not obtained. This finding is similar to previous estimates.29 This increase may reflect the time necessary for CT or MRI to be performed and interpreted by a radiologist, and the fact that patients with injury-related conditions who receive CT or MRI may be more severely ill, requiring additional treatment interventions. Although the rate of using CT or MRI during visits for injury-related conditions increased during the study period, the length of stay for these visits did not change by much. This is likely because visits for injury-related conditions during which CT or MRI was obtained constitute a small fraction of the total number (15% in 2007) of visits related to injuries. Other factors that may have contributed to this finding are the prioritization of the evaluation of patients with injury-related conditions (especially those with high-triage acuity levels) compared with patients with other presenting conditions, the advent of faster CT scanners, and the increased proximity of CT scanners to emergency department treatment areas. Whether CT or MRI use has led to an increase in the emergency department length of visit for other presenting conditions has yet to be determined.
Our study has several limitations. First, the National Hospital Ambulatory Medical Care Survey data set used is based on a probabilistic sampling of select visits in select emergency departments. For example, in 2007, 35 490 visits were selected as a representation of national emergency department visits. It is possible that the national estimates generated may not accurately represent true values. However, this seems unlikely because of the consistency and rigor of the methods and sampling framework. To ensure the reliability of our estimates, all of the outcomes we measured were based on more than 30 records of patient visits (the cut-off for reliability set by CDC's National Center for Health Statistics). It is also possible that the increased use of CT and MRI may increase the detection rate of other illnesses not reported herein, but the significance of this would be small because the admission rate did not increase during the study period.
Second, the combination of certain racial groups such as Asians, Native Hawaiian or other Pacific Islanders, and American Indians or Alaskan natives may have prevented us from detecting significant differences in CT use for injury-related visits in those groups. However, combining these groups allowed us to generate more reliable estimates.
Third, because data from this study are based on a retrospective review of patient charts, it is possible that the proportion of visits during which injury-related life-threatening conditions were diagnosed could have been underestimated if physicians did not record all the injury-related life-threatening conditions diagnosed during a particular visit. Finally, additional information related to the nature of injuries would have been informative. For instance, there was no way to directly adjust for disease severity, and the analysis did not include patient outcomes after admission for injury-related conditions.
From 1998 to 2007, use of CT or MRI during injury-related emergency department visits increased approximately 3-fold, without an equal increase in the diagnosis of injury-related, life-threatening conditions or admission rates associated with these visits. Time and monetary costs are associated with obtaining advanced radiology, and potential health risks occur for the patients receiving MRI or CT. Further work is needed to understand the patient, hospital, and physician factors responsible for this increase and to optimize the risk-benefit balance of advanced radiology use. The role of evidence-adoption strategies such as computerized decision support and audit and feedback in promoting adherence to decision rules for imaging needs to be further understood.
Corresponding Author: Frederick Kofi Korley, MD, 5801 Smith Ave, Davis Bldg, Ste 3220, Baltimore, MD 21209 (firstname.lastname@example.org).
Author Contributions: Drs Korley and Pham had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Korley, Pham, Kirsch.
Acquisition of data: Pham.
Analysis and interpretation of data: Korley, Pham.
Drafting of the manuscript: Korley, Pham, Kirsch.
Critical revision of the manuscript for important intellectual content: Korley, Pham, Kirsch.
Statistical analysis: Pham.
Administrative, technical, or material support: Korley.
Study supervision: Kirsch.
Financial Disclosures: None reported.
Funding/Support: Dr Korley is supported by the Robert E. Meyerhoff Professorship at the Johns Hopkins University School of Medicine, and by grant 5KL2RR025006 from the National Center Research Resources, a component of the National Institutes of Health (NIH), and the NIH Roadmap for Medical Research.
Role of the Sponsor: No funding organizations played a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
Disclaimer: The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official view of the National Center Research Resources or the National Institutes of Health.