Ionizing Radiation Exposure to Patients Admitted With Acute Myocardial Infarction in the United StatesClinical Perspective
Background—Invasive and noninvasive cardiovascular imaging is beneficial in the care of patients admitted with acute myocardial infarction. Little is known about patients' cumulative radiation exposure.
Methods and Results—All patients admitted with an acute myocardial infarction to any of 49 University HealthSystem Consortium member hospitals from 2006 to 2009 were reviewed for inpatient procedures involving ionizing radiation that included chest radiograph, computed tomogram scans, radionuclide imaging, diagnostic cardiac catheterization, and percutaneous coronary intervention. The average cumulative effective radiation dose per patient was estimated on the basis of published typical effective radiation doses for imaging procedures. Patients (n=64 071) admitted for acute myocardial infarction had a median age of 64.9 years. A total of 276 651 procedures involving ionizing radiation were performed during the study period, a median of 4.3 procedures per patient per admission. The majority of patients had invasive catheterization (77%), followed by computed tomogram scans (52%), mostly body examinations. The median cumulative effective radiation dose delivered was 15.02 mSv per patient per acute myocardial infarction admission. Postprocedural bleeding was a significant predictor of radiation exposure (odds ratio, 2.01; 95% confidence interval, 1.85 to 2.18), together with postprocedural mechanical complications resulting from device implantation (odds ratio, 2.86; 95% confidence interval, 2.61 to 3.13). Patients with higher underlying clinical complexity (defined by severity of illness scores) had higher radiation exposure and higher mortality (P<0.0001). There was also significant geographic variation in radiation exposure; patients in New England received the lowest cumulative exposure (odds ratio, 0.78; 95% confidence interval, 0.74 to 0.81).
Conclusions—Acute myocardial infarction inpatients are exposed to an approximate median radiation dose of 15 mSv. This exposure is a result of multiple cardiovascular and noncardiovascular procedures. Efforts should be made to understand the risks and benefits of radiation exposure per episode of care for acute myocardial infarction.
Each year, several billion diagnostic imaging studies and procedures are performed worldwide, of which at least one third are cardiovascular.1 Over the last decade, there has been significant growth in the use of advanced imaging studies, including those associated with ionizing radiation. Between 1995 and 2005, there was a 4-fold increase in the use rate of advanced imaging paid under the Medicare Physician Fee Schedule.2 Furthermore, between 1980 and 2006, it is estimated that there was a >7-fold increase in the collective dose attributed to radiation from medical tests and procedures.3
Editorial see p 2113
Clinical Perspective on p 2169
Recently, the American Heart Association Committee on Cardiac Imaging published a science advisory statement on ionizing radiation in cardiac imaging and procedures.4 The members of this committee commented that medical imaging is the largest controllable source of radiation exposure to the US population and that the key determinant is the ordering healthcare provider. The committee recommended (Class I, Level of Evidence C) that “physician education should emphasize that cardiac imaging studies that expose patients to ionizing radiation should be ordered only after thoughtful consideration of the potential benefit to the patient and in keeping with established appropriateness criteria.”
However, most research and educational efforts relative to radiation dose reduction have been directed toward quantifying the radiation dose per imaging study or procedure rather than decision making during episodes of care, which results in a cumulative exposure to patients and is an arguably substantial opportunity to reduce overall exposure. In this study, we sought to identify the number of tests involving ionizing radiation and total cumulative radiation exposure of patients admitted for a single episode of inpatient care for acute myocardial infarction (AMI). Additionally, we aimed to understand the variables that are associated with radiation exposure, which may be targets for radiation reduction strategies.
We performed a retrospective cohort study using billing and use data submitted to the Clinical Database/Resource Manager at the University HealthSystem Consortium (UHC) (https://www.uhc.edu). The UHC is an alliance of >100 academic medical centers in the United States and their affiliated hospitals, representing ≈90% of nonprofit US academic medical centers. Consortium members enrolled in the database submit inpatient billing and use data that are available to all enrolled members for performance improvement.
Data submitted by all UHC-enrolled hospitals participating in the Clinical Resource Manager database (49 academic medical centers) relating to inpatient admissions between January 1, 2006, and June 30, 2009, were analyzed. Imaging data were obtained from the 49 hospitals that had contributed to the Clinical Resource Manager database during the study period (Figure 1).
We included all adult patients (≥18 years of age) who presented to 1 of these 49 hospitals whose principal diagnosis at the time of discharge was AMI as defined by the International Classification of Diseases, ninth revision (ICD-9) codes 410.x0 and 410.x1 (unspecified and initial episode of care, respectively) but not 410.x2 (representing a subsequent episode of care). The Institutional Review Board of Duke University waived the requirement for informed consent. Before submission for publication, written approval was obtained from the UHC in keeping with the UHC policy governing the use of the data.
We reviewed patient characteristics, including demographics, chronic comorbidities, and postprocedural complications, during the encounter and discharge status. We also reviewed hospital characteristics, including size and geographic location.
For each patient in the study population, we examined the number of each of the following common diagnostic tests or procedures involving ionizing radiation: chest radiograph (CXR), computed tomogram (CT) of the head, CT of the chest, CT of the body, nuclear ventilation-perfusion scan, nuclear multiple gated acquisition scan, nuclear rest-stress scan, diagnostic cardiac catheterization, and percutaneous coronary intervention (PCI). To prevent dose overestimation from overcounting a single test that may have had >1 billing code, we ensured that only 1 specific billing code was associated with a given test. Furthermore, for the purposes of dose estimation, patients receiving diagnostic cardiac catheterization alone were considered separately from those receiving diagnostic cardiac catheterization with associated PCI. Those receiving PCI were assumed to have a single procedure but were allocated a higher dose.
Data on complications were obtained from the UHC Complication Profiler, which identified potentially avoidable complications not present at the time of admission. The UHC Complication Profiler gathers diagnosis-related group and ICD-9 code claims data submitted by each of the participating institutions.
Estimates of Radiation Dose
For each patient presenting with an AMI, we calculated an estimate of the cumulative effective radiation dose exposure during each admission. Because patient-specific effective doses were unavailable, we used estimates of typical effective radiation doses in millisieverts from the published literature5 (Figure 1), with the understanding that there is a range of reported dose values. Typical effective radiation doses for each imaging study were multiplied by the numbers of each imaging study performed to obtain an estimate of the total exposure during an AMI encounter.
Estimates of Underlying Clinical Complexity of the Study Population
An assessment of the underlying clinical risk and complexity of the study population (and their associated radiation doses) was made using the severity-of-illness (SOI) score created by the 3M All Patient Refined Diagnosis-Related Groups software (3M, St Paul, Minn). This tool provided a uniform process for qualifying the clinical differences in the inpatient populations. The 3M All Patient Refined Diagnosis-Related Groups severity and mortality subclasses are assigned according to a clinical logic that simultaneously evaluates the interactions of multiple comorbidities, age, procedures, and principal diagnosis. Radiation doses across SOI scores for all AMI patients, patients with and without revascularization, and patients with intra-aortic balloon pumps who also needed to be intubated and ventilated with and without cardiac arrest on admission were considered.
The SOI score is defined as the extent of physiological decompensation or loss of organ system function, and the subclasses are numbered sequentially as follows: 1=minor, 2=Moderate, 3=major, and 4=extreme SOI. In-hospital mortality was also determined for each subgroup.
We calculated the mean and median numbers of studies per AMI admission and the mean and median estimated cumulative effective radiation exposure per AMI admission (primary outcome) for the study population over the 3.5-year study period (Figure 1). We divided all AMI patient admissions by the amount of radiation received into the following 2 dose categories, which were used for comparison: less than the median dose and more than or equal to the median dose. We also separated all doses ≥50 mSv for the sake of interest (but not separate statistical comparison) because this value represents the annual occupational exposure limit for radiation workers.6 We then categorized the radiation exposure by the baseline variables as presented in Tables 1 through 3 to understand the influence of each variable on radiation exposure.
Descriptive statistics for patient characteristics were generated and compared between patients with median-high radiation exposure versus low (less than median) radiation exposure in each cohort by use of χ2 tests for categorical variables and Student t tests for normally distributed, continuous variables. Multivariate logistic regression models were constructed with radiation exposure as the dependent variable, and all listed variables and all identified 2-way interaction terms and confounders were used as independent variables. The χ2 and Mantel-Haenszel tests were used to assess all independent covariates for collinearity and as potential effect modifiers and confounders before inclusion in the logistic regression models. The models were constructed using backward stepwise variable selection, and a value of P<0.01 was used as the criterion for variable retention. The predictors in the final logistic regression models are listed in Table 4. Secondary outcomes (in-hospital complications and in-hospital mortality) were analyzed with Student t tests for normally distributed variables, Wilcoxon tests for skewed continuous variables, and logit models for dichotomous outcomes. A 1-way ANOVA test was used to compare the 4 SOI groups shown in Table 5. All statistical analyses were performed with SAS software (version 9.1.3, SAS Institute Inc, Cary, NC).
Baseline Patient Characteristics
Between January 1, 2006, and June 30, 2009, there were 64 071 admissions for AMI presenting to 1 of the identified 49 hospitals (Table 1). The median age was 64.9 years (interquartile range, 54.8 to 77 years), and 23 394 (37%) were female (Table 1). The majority (69%) of patients were white. Hypertension (55%), hyperlipidemia (48%), smoking (39%), and diabetes mellitus (35%) were the most common comorbidities. The median length of stay for treatment of AMI was 4 days. The average 30-day readmission rate to the same hospital was 13.82%. The average in-hospital mortality rate for all patients was 5.78%.
Cumulative Effective Radiation Dose per AMI Encounter
At least 276 651 studies involving ionizing radiation were performed on the sample population during the study period (Figure 1). On average, each patient was likely to receive 4.3 imaging studies involving radiation per AMI admission. The most common study was a CXR, performed in almost 83% of patients admitted with AMI, followed by PCI (53%), cardiac catheterization (24%), CT of the body (15%), CT of the head (12%), multiple gated acquisition (6%), nuclear stress test (5%), CT of the chest (3%), and nuclear ventilation-perfusion scan (1%). The estimated average cumulative effective radiation dose per AMI encounter was 14.63 mSv. Although there was heterogeneity in the hospital sample size in this analysis, the median radiation dose across each of the 49 hospitals showed minimal variability, 15.02 mSv (interquartile range, 7.1 to 15.1 mSv). Forty-eight percent of all AMI admissions received a cumulative effective radiation dose that was less than the median. Half of all patients were in the intermediate range of greater than or equal to the median exposure (15 mSv) but <49 mSv, and almost 2% received a cumulative radiation dose ≥50 mSv during their AMI admission. This distribution of cumulative effective radiation dose per AMI admission together with the gender distribution is shown in Figure 2.
Approximately half (49%) of the hospitals sampled had >500 beds, and most (62%) were in the Mid-Atlantic, Midcontinent, or Midwestern regions (Table 2). The remaining hospitals were in the New England, Southeastern, or Western regions. The most common payer (51%) was Medicare, followed by private insurance companies, which together reimbursed 35% of admissions in this cohort.
In-Hospital Complications and Radiation Exposure
The documented and coded in-hospital postprocedural complications and associated radiation exposure are presented inTable 3. Postprocedural bleeding, the most common complication, was reported in 5% of this population. Mechanical complications caused by devices or implants (predominantly cardiac devices and vascular devices such as pacemakers or dialysis catheters; ICD-9 codes 996.72 to 996.74) were seen in 4%. Both of these in-hospital complications were associated with a significant increase in radiation exposure (P>0.001).
Predictors of Radiation Exposure
The independent predictors of intermediate to high radiation exposure are presented inTable 4. It should be noted that postprocedural complications, including mechanical complications caused by device or implant (odds ratio, [OR], 2.86; 95% confidence interval [CI], 2.61 to 3.13), postprocedural hemorrhage or hematoma (OR, 2.01; 95% CI, 1.85 to 2.18), postprocedural gastrointestinal bleeding or ulcer (OR, 3.21; 95% CI, 2.64 to 3.90), and postprocedural pneumonia (OR, 1.89; 95% CI, 1.68 to 2.13), represented the strongest set of predictors for radiation exposure. Another marker of patient care complexity, the intensive care unit length of stay, was also a predictor of radiation (OR, 1.05; 95% CI, 1.04 to 1.05). Non–patient-related factors such as hospital region were also strongly associated with radiation exposure, with New England hospital location associated with less intermediate or high radiation exposure (OR, 0.78; 95% CI, 0.74 to 0.81). Finally, the average in-hospital mortality rate was higher (17.83%) for those patients receiving ≥50 mSv of radiation exposure compared with the average mortality rate for all patients (5.78%).
Underlying Clinical Complexity
The radiation doses for the different clinical risk and complexity groups, defined by the SOI score, including those with and without revascularization, are presented with their corresponding in-hospital mortality rates in Table 5. Although median radiation dose remained constant, the average radiation dose and in-hospital mortality increased with increasing SOI score. Patients undergoing revascularization had higher levels of radiation but a lower in-hospital mortality rate compared with those patients receiving medical treatment without revascularization. Furthermore, patients with intra-aortic balloon pumps who had been intubated and ventilated had significantly higher radiation exposures and almost twice the mortality rate of those not needing intubation/ventilation.
This study provides a first attempt at considering cumulative radiation exposure for a single episode of care for AMI. Our analysis of >64 000 AMI hospitalizations indicates that patients received a median radiation dose of 15.0 mSv per AMI admission.
For patients presenting with AMI, the majority of the estimated radiation dose per encounter was from invasive catheterization procedures, particularly PCI (Figure 3). The majority of the remainder of the estimated radiation exposure per AMI admission was accounted for by CT scans, in particular CT scans of the abdomen and pelvis. Although the specific indications for body CT scans are not known, it is probable that the observed 5% rate of postprocedural bleeding was clinically responsible for the ordering of most of the CT scans of the abdomen and pelvis. In patients without postprocedural bleeding, only 14% received body CT scans, whereas 32% of patients with bleeding complications received CT scans of the abdomen and pelvis. Therefore, the rate of CT scans of the abdomen and pelvis may be a potential marker for post–cardiac catheterization bleeding in AMI patients. Diagnostic angiography alone (without PCI) contributed 12% of the average cumulative radiation dose, which was less than the contribution from body CT scans.
In our model, the most significant predictors of higher radiation exposure were in-hospital complications, especially postoperative bleeding (OR, 2.01; 95% CI, 1.85 to 2.18) (possibly from cardiac catheterization) and mechanical complications (OR, 2.86; 95% CI, 2.61 to 3.13). Although it may appear reasonable that there is a higher use of imaging and procedures involving radiation in the setting of complications, individual physician and institutional practice patterns in response to these complications are variable, with almost one third of cases with complications falling into the lowest-exposure group. Moreover, the usual management of patients with postcatheterization bleeding is hemodynamic support and expectant care.7 Further work is needed to establish the specific indications and appropriate use criteria for procedures involving radiation in the setting of in-hospital complications. Nevertheless, these findings demonstrate that total radiation exposure during hospitalization may be a marker for in-hospital complications.
The use of imaging studies involving ionizing radiation increased with an increase in the underlying clinical complexity of the patient. Nevertheless, current guidelines indicate that the benefits of diagnostic catheterization and/or PCI in the setting of an AMI will always outweigh any radiation-associated risks. Therefore, the majority of AMI patients receive a minimum dose of radiation from angiographic procedures that are proven strategies to reduce morbidity and mortality.
However, the appropriateness of radiation from nonangiographic sources (ie, sources other than diagnostic catheterization or PCI), which accounted for 33% of the average cumulative radiation exposure during an AMI admission (majority from CT scans of the abdomen and pelvis), is unknown. It is in this situation that the risk-benefit relationship is less well understood.
We adjusted for underlying clinical risk by considering the SOI score and in-hospital mortality. National in-hospital AMI mortality rates have been reported as 9.4%.8,9 Our data suggest a slightly lower overall in-hospital mortality rate of 5.8%, which may reflect national quality improvement. Increasing SOI scores were associated with higher radiation exposure and worsening mortality. The only exception to this was in the setting of revascularization, in which increased radiation exposure was associated with improved mortality. This again confirms the incremental benefit of procedures involving radiation in this setting.
A minority of high-risk or critically ill patients represent the extreme outliers and tend to use more tests and procedures involving radiation, thereby driving up the average radiation doses. The sickest patients with major or extreme SOI scores drive the majority of mortality and radiation exposure. However, further evaluation of radiation use should be directed toward the minor and moderate SOI risk groups because, again, the risk-benefit relationship is less clear in this situation.
The median cumulative radiation exposure per AMI admission of 15 mSv is the equivalent of 750 CXRs. This median value remained constant regardless of clinical risk and represents ≈5 times the annual US background radiation of 3 mSv or the equivalent of 150 CXRs and ≈30% of the 50 mSv (2500 CXRs) annual limit for radiation workers.6 Although high doses of ionizing radiation (>100 mSv) have been linked to deleterious effects, including, but not limited to, cancer formation, the risk of malignancy attributable to the lower doses usually associated with medical imaging procedures remains controversial.10 However, on the basis of epidemiological studies, short-term exposure of doses of 10 to 50 mSv or protracted exposure of doses of 50 to 100 mSv has been associated with an increased risk of some cancers.11 Most authorities recommend a conservative strategy when considering tests and procedures involving ionizing radiation, in keeping with the “ALARA” (as low as reasonably achievable) principle while still being able to answer the clinical question.4
Thus far, the emphasis has focused on the radiation exposure for individual diagnostic tests and therapeutic procedures and their associated doses in isolation.12,–,14 One recent study calculated population-based rates of exposure and the annual cumulative effective radiation doses from imaging procedures.15 However, we believe a more important and actionable consideration may be the total cumulative radiation exposure that a patient receives during an episode of care for a given diagnosis. Rather than merely tracking total cumulative radiation exposure longitudinally, this new paradigm lends itself to an improved understanding of the specific predictors of radiation exposure by addressing radiation exposure per episodes of care. In so doing, ordering healthcare providers will be encouraged to consider the appropriateness of tests involving radiation in situations in which there is the greatest potential to effect change in practice patterns, ie, during an episode of care.
Interestingly, radiation exposure also varied by hospital region, with admission to a hospital in New England being associated with the lowest exposure (OR, 0.78; 95% CI, 0.74 to 0.81). There was a 10% variation in exposure based on geographical location. Of patients presenting to hospitals in the New England region, 47% received intermediate to high radiation doses, whereas 57% of patients presenting to a hospital in the Midwestern or Western regions were exposed to the same dose range. These findings highlight the role of local care patterns (as identified by geographic region) in determining radiation exposure for patients with AMI.
This study is the first to describe the cumulative radiation exposure using a large national sample. However, the results of our study should be interpreted within the framework of the following limitations. First, we used estimates of typical effective radiation doses because it was not possible to establish actual patient-specific doses. However, our estimates are conservative and likely underestimate actual exposure. Second, the study population was limited to the 49 academic hospitals. The ability to extrapolate these data to other nonacademic, community-based hospitals, where practice patterns may be different, is uncertain. However, it is likely that with the inclusion of more hospitals of various types, radiation exposure would have shown greater variation across hospitals. Third, only those complications with associated ICD-9–coded claims data were captured. Although unlikely, it is possible that other complications occurred but were not captured by this data set. Furthermore, coding practices vary between hospitals. Finally, we did not have any information on the clinical indications for which the studies were performed and therefore are unable to draw any conclusions on the appropriateness of testing or procedures.
Patients admitted to academic medical centers and their affiliated hospitals for the treatment for AMI are exposed to significant amounts of radiation. During a 6-day AMI hospitalization, patients are exposed to a median radiation dose of ≈15 mSv, a dose that is 5 times the annual background level and a third of the annual limit for radiation workers. The variables that independently predict radiation exposure are predominantly in-hospital complications, especially postprocedural complications, which may drive an increased use of CT scans.
In general, patients with higher underlying clinical complexity (SOI scores) tended to receive higher radiation exposure and had higher mortality. Additionally, non–patient-related factors such as hospital region are significantly associated with radiation exposure.
Future efforts should be aimed at reducing effective radiation exposure for a given episode of care with an appreciation of these variables. Specifically, research is needed to understand the appropriate use of in-hospital imaging and procedures involving radiation and to accurately measure radiation exposure for a given diagnosis.
We propose that rather than considering effective radiation exposure in isolation for a given imaging test, efforts should be made to document real-time total exposure per episode of care for a given diagnosis such as AMI. This paradigm has the greatest potential to inform decisions and change the care of AMI patients. Finally, effective radiation exposure per episode of care represents a potential safety metric for patients with common clinical conditions.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
- Received February 22, 2010.
- Accepted August 31, 2010.
- © 2010 American Heart Association, Inc.
Office of Inspector General. Growth in Advanced Imaging Paid Under the Medicare Physician Fee Schedule. Washington, DC: US Department of Health and Human Services; 2007. Report No: OEI-01-06-00260.
- Gerber TC,
- Carr JJ,
- Arai AE,
- Dixon RL,
- Ferrari VA,
- Gomes AS,
- Heller GV,
- McCollough CH,
- McNitt-Gray MF,
- Mettler FA,
- Mieres JH,
- Morin RL,
- Yester MV
- Baim DS
AHRQ Quality Indicators: Guide to Inpatient Quality Indicators: Quality of Care in Hospitals: Volume, Mortality, and Utilization. Rockville, Md: Agency for Healthcare Research and Quality; 2002.
National Registry of Myocardial Infarction. http://www.ncdr.com/WebNCDR/ACTION. Accessed April 28, 2010.
- Brenner DJ,
- Doll R,
- Goodhead DT,
- Hall EJ,
- Land CE,
- Little JB,
- Lubin JH,
- Preston DL,
- Preston RJ,
- Puskin JS,
- Ron E,
- Sachs RK,
- Samet JM,
- Setlow RB,
- Zaider M
- Gilbert ES
- Hall EJ,
- Brenner DJ
Over the last decade, there has been a significant growth in the use of imaging studies involving ionizing radiation. Most authorities recommend a conservative strategy when considering tests involving ionizing radiation, in keeping with the “ALARA” (as low as reasonably achievable) principle while still being able to answer the clinical question. In this large retrospective study of >64 000 acute myocardial infarction episodes, we found that during a median 4-day acute myocardial infarction hospitalization, patients were exposed to a median radiation dose of ≈15 mSv, a dose that is 5 times the annual background level and a third of the annual limit for radiation workers. Thus far, the emphasis has focused on the radiation exposure for individual diagnostic tests and their associated doses in isolation. We believe a more important and actionable consideration may be the total cumulative radiation exposure that a patient receives during an episode of care for a given diagnosis. Rather than merely tracking total cumulative radiation exposure longitudinally, this new paradigm lends itself to an improved understanding of the specific predictors of radiation exposure by addressing radiation exposure per episodes of care. In so doing, we hope that ordering healthcare providers might also be encouraged to carefully consider the appropriateness of tests involving radiation and become even more mindful of their potential risks.