At the Intersection of Health, Health Care and Policy Cite this article as: Michael J. Franczak, Madeline Klein, Flavius Raslau, Jo Bergholte, Leighton P. Mark and John L. Ulmer In Emergency Departments, Radiologists' Access To EHRs May Influence Interpretations And Medical Management Health Affairs, 33, no.5 (2014):800-806 doi: 10.1377/hlthaff.2013.0171
The online version of this article, along with updated information and services, is available at: http://content.healthaffairs.org/content/33/5/800.full.html
For Reprints, Links & Permissions: http://healthaffairs.org/1340_reprints.php E-mail Alerts : http://content.healthaffairs.org/subscriptions/etoc.dtl To Subscribe: http://content.healthaffairs.org/subscriptions/online.shtml
Health Affairs is published monthly by Project HOPE at 7500 Old Georgetown Road, Suite 600, Bethesda, MD 20814-6133. Copyright © 2014 by Project HOPE - The People-to-People Health Foundation. As provided by United States copyright law (Title 17, U.S. Code), no part of Health Affairs may be reproduced, displayed, or transmitted in any form or by any means, electronic or mechanical, including photocopying or by information storage or retrieval systems, without prior written permission from the Publisher. All rights reserved.
Not for commercial use or unauthorized distribution Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Emergency Care By Michael J. Franczak, Madeline Klein, Flavius Raslau, Jo Bergholte, Leighton P. Mark, and John L. Ulmer 10.1377/hlthaff.2013.0171 HEALTH AFFAIRS 33, NO. 5 (2014): 800–806 ©2014 Project HOPE— The People-to-People Health Foundation, Inc.
doi:
Michael J. Franczak is an undergraduate student in the College of Science at the University of Notre Dame, in South Bend, Indiana. At the time of the study he was a research assistant in the Department of Radiology at the Medical College of Wisconsin, in Milwaukee. Madeline Klein is a volunteer with Jesuit Volunteer Corps, in Boston, Massachusetts. At the time of the study she was a research assistant in the Department of Radiology at the Medical College of Wisconsin. Flavius Raslau is an assistant professor of radiology at the College of Medicine, University of Kentucky, in Lexington. At the time of the study he was a neuroradiology fellow in the Department of Radiology at the Medical College of Wisconsin. Jo Bergholte is a program manager at the Medical College of Wisconsin. Leighton P. Mark is a professor of radiology at the Medical College of Wisconsin. John L. Ulmer (julmer@ mcw.edu) is professor of radiology at the Medical College of Wisconsin.
800
In Emergency Departments, Radiologists’ Access To EHRs May Influence Interpretations And Medical Management ABSTRACT The adoption of electronic health records (EHRs) that meet federal meaningful-use standards is a major US national policy priority. Policy makers recognize the potential of electronic communication in delivering high-quality health care, particularly in an environment of expanding remote access to medical care and the ever-increasing need to transmit health care records across institutions. To demonstrate this principle, we sought to estimate the significance of EHR access in emergent neuroradiologic interpretations. Three neuroradiologists conducted a prospective expert-rater analysis of 2,000 consecutive head computed tomography (CT) exams ordered by emergency department (ED) physicians. For each head CT exam, the neuroradiologists compared medical information generated by ED physicians to information generated by the interpreting radiologists who had access to additional EHR-derived patient data. In 6.1 percent of the head CT exams, the neuroradiologists reached consensus—meaning two out of three agreed— that the additional clinical data derived from the EHR was “very likely” to influence radiological interpretations and that the lack of that data would have adversely affected medical management in those patients. Health care providers must recognize the value of implementing EHRs and foster their widespread adoption.
E
lectronic health records (EHRs) have been shown to improve the efficiency of health care delivery and have the potential to improve health care quality.1–3 The underlying principle is that sharing of, and immediate access to, patient health information among providers and institutions promotes more effective communication of critical data that may influence and guide clinical decision making. Thus, electronic communication has the potential to improve health care quality and delivery both within and across institutions and is particularly important in an environment of expanding remote access to medical care. Although this seems intuitive, data supporting this principle across
Health Affairs
May 2 014
33:5
health care fields are not substantial. Policy makers recognize the potential importance of the EHR. The adoption of EHRs that meet federal meaningful-use standards has been a major US national policy priority for several years. In 2009 the Health Information Technology for Economic and Clinical Health (HITECH) Act authorized nearly $30 billion in funding to support the adoption of EHRs that achieve federal meaningful-use standards.4 Hospitals and physicians’ offices are being rewarded for meeting (or penalized in later years for failing to meet) specific information system capability and reporting clinical quality measures.5,6 Since the incentives have gone into effect, there has been a substantial growth in EHR adoption.7 However,
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
there is a significant discrepancy in the rate of EHR adoption among providers, especially among those who are ineligible to receive such federal incentives.8,9 Catherine DesRoches and colleagues showed that as of 2012, 16.7 percent of US acute care hospitals used a comprehensive electronic record system, and 44.0 percent had a basic system.7 This increase appeared to be concentrated around large, Midwestern, and major teaching hospitals. In contrast, hospitals not eligible for federal incentives have dismally low rates of adoption of an EHR system. Only 12 percent of short-term acute care hospitals have basic EHR systems, compared with only 6 percent of long-term acute care hospitals.9 Access to EHRs and the electronic exchange of health information seems especially relevant to radiology, a technology-driven specialty that has been at the forefront of development of information technology including picture archiving and communication systems (PACS), web-based image distribution, and the expansion of teleradiology. The very technology-driven nature of radiology, which provides for remote access and delivery, also makes the field vulnerable to informational risks. The negative impact of incomplete clinical information on radiological interpretations has been well documented in several studies.10,11 To our knowledge, however, there are no studies investigating the interpretive value of EHRs in delivering radiologic services. Yet the field is ideally suited to test the hypothesis that EHRs can improve medical communication in a meaningful way to improve health care quality. We chose to study the use of EHRs in emergency department (ED) radiologic services because of the need for rapid turnaround time, the immediate initiation of therapy, and the impact of communication on early treatment. We compared medical information pertinent to emergent head computed tomography (CT) examinations as entered by referring ED physicians to additional EHR-derived information retrieved by interpreting radiologists. We sought to determine how often the EHR that was accessed by the interpreting radiologist was perceived to provide relevant additional information and to estimate the importance of those additional data in interpreting imaging studies.
Study Data And Methods Head Computed Tomography Exams And Work Flow The study was approved by the Medical College of Wisconsin Institutional Review Board. The Froedtert and the Medical College of Wisconsin, in Milwaukee, contains a statedesignated Level I regional trauma center that
provides twenty-four-hour ED head CT availability and interpretations year round. Radiologic interpretations for ED patients are provided via on-site teleradiology, using PACS and the EHR for exam reporting. PACS is a medical technology that provides storage, retrieval, communication, database, access to images from multiple modalities, and user interface components that in many ways resemble the EHR system. Combined with available and emerging web technology, PACS has the ability to deliver timely and efficient access to images, interpretations, and related data within and across institutions. Head CT requests from ED physicians require ED personnel to enter pertinent medical information into the electronic order entry system. As a matter of institutional protocol, interpretations are reported to ED physicians within one hour of the head CT examination. Interpretations within this one-hour time frame are delivered by neuroradiology faculty, neuroradiology fellows, or radiology residents. Interpreting radiologists have full access to the EHR. Thus, within one hour of the ED head CT request, the interpreting radiologist reviews the request, reviews additional EHR and imaging data, and reports the results via the EHR. Written interpretations include additional pertinent clinical information retrieved from the EHR in the indication section of the report. It is important to note that the EHR and imaging data may be reviewed simultaneously and that reviewing image and EHR data may trigger review of additional pertinent EHR data. In other words, the use of the EHR to support image interpretation is interactive. Patient Data Data on emergency department head CT scans were collected soon after the institutionwide implementation of EHRs. Retrospective data were obtained from 2,000 consecutive ED head CTexams acquired from October 1, 2010, to December 31, 2010. The institutional EHR stores all entered data, which could be accessed retrospectively. Thus, pertinent clinical information entered by ED personnel at the time of the request could be compared to any additional EHR-retrieved clinical information entered by the radiologist at the time of interpretation. Demographic patient information including age, sex, and race or ethnicity is linked to the image data and are, thus, always available. Of the 2,000 patient exams reviewed, 49.2 percent were males, and 50.8 percent were females. A mean patient age of fifty-four was observed, with a range from eighteen to ninety-nine years. The race or ethnicity distribution of patients scanned was 58.4 percent Caucasian, 34.4 percent African American, 3.2 percent Hispanic, and 4.1 percent other. M ay 2 0 1 4
33 : 5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Health A ffairs
801
Emergency Care Expert Analysis Three neuroradiologists with clinical imaging experience of two, sixteen, and thirty-one years, respectively, and identified for the purposes of this article as raters number 1, 2, or 3, respectively, independently compared the ED-entered clinical information for each head CT examination along with any additional EHR-retrieved information entered by the interpreting radiologist. The neuroradiologists rated discrepancies between the ED-entered clinical information and the information entered by the interpreting radiologist. The ratings were based on criteria described below. First, the medical content of ED-entered and radiologist-entered (EHR-retrieved) clinical data was rated using binary scoring of “yes” or “no” to indicate whether additional clinical data were entered by the radiologist into the radiology report. The rating was given without regard for a potential clinical impact of the additional clinical data. Correction of ED-entered spelling or wording errors were not considered additional data. An interrater reliability analysis was performed using the interclass correlation coefficient (ICC) in order to quantify the consistency among the three neuroradiologists in determining the presence of additional medical content. Second, when additional clinical data were present in the record, the neuroradiologists considered the likelihood that the additional EHRretrieved clinical information was clinically significant and “very likely,” “possibly,” or “unlikely” to influence radiological interpretations in a way that would affect medical management. In other words, was the additional information
Exhibit 1 Discrepancies In Medical Content Between Emergency Department (ED)–Entered and Electronic Health Record (EHR)–Retrieved Clinical Data On Head Computed Tomography (CT) Exams, October 1–December 31, 2010
attained from the EHR thought necessary? An effect on medical management might include an effect on imaging diagnoses or recommendations for additional imaging or laboratory tests, or might otherwise influence the workup or treatment algorithms. Third, we calculated the extent to which the neuroradiologists agreed that the additional clinical data could have “very likely” or “possibly” influenced radiological interpretations in a way that would affect medical management. Consensus agreement was set at two out of three raters, and total agreement occurred when all three raters gave the same response. Limitations Our study had several limitations. First, the investigation focused entirely on emergent head CTs performed in a Level 1 trauma center. Therefore, any conclusions drawn from this study might not be fully generalizable to other clinic settings. Any widespread extrapolation of results would be strengthened by multicenter studies. Second, our institution has full access to EHRs in all ED settings. Therefore, it is ethically untenable to design a randomized study where the EHR is not accessed in a significant proportion of patients. Consequently, we were unable to determine the actual impact of additional EHR-retrieved clinical data on radiologic interpretations or patient management; rather, we chose to calculate theoretical consequences. Third, we chose to study the impact on radiologic services in the ED setting, which requires rapid interpretation turnaround and is highly vulnerable to informational risks. Thus, the high estimated impact of EHR access might not extrapolate to other areas of radiology. Finally, larger patient series in multi-institutional studies would strengthen the statistical confidence of the results obtained in this preliminary investigation.
1,200
Study Results
Number of exams
1,000 800 600 400
54.9% 44.1%
48.9%
200 0 Rater 1
Rater 2
Rater 3
SOURCE Authors’ analysis of 2,000 head CT exams ordered by the ED at the Froedtert and the Medical College of Wisconsin, in Milwaukee, and reviewed by three neuroradiologists. NOTE Clinical experience for each of the neuroradiologist raters is as follows: rater 1, two years; rater 2, sixteen years; rater 3, thirty-one years.
802
Health Affairs
May 2 014
33:5
The three neuroradiologist raters identified that additional clinical data were present in 49.3 percent of the head CT exams. This represents an average of the raters’ findings, which ranged from 44.1 percent to 54.9 percent (Exhibit 1). For the remainder of cases, the raters determined that no additional clinical data were present in the record. The interrater reliability was ICC equals 0.824 (p < 0:001) and a 95 percent confidence interval of 0.810, 0.837, indicating strong agreement among the raters on the presence of additional medical content. On average, the additional clinical data available in the EHR was designated by individual expert raters as being “very likely” to affect in-
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
entered clinical data were exceedingly rare and may have been related to communication errors among ED personnel.
Discussion There is a broad consensus regarding the potential benefits of the EHR, especially the optimization of information sharing among providers. The HITECH Act authorized the Centers for Medicare and Medicaid Services to promote meaningful use of certified EHR technology, with penalties for lack of EHR use beginning in 2015. Health information exchange is another priority of HITECH, designed to improve the quality and efficiency of health care across institutions as well. Policy makers have been actively promoting the development of broad-based electronic health information exchange systems. In 2010 the Office of the National Coordinator for Health Information Technology (ONC) established the State Health Information Exchange Cooperative Agreement Program, which awarded more than $540 million to states in efforts to build capacity for exchanging health information among health care systems both within and across state lines. Barriers that have hindered widespread health information sharing include high cost, limited across-institution interoperability, and low motivation to share information in a fee-for-service environment.12–14 Julia Adler-Milstein and colleagues found that 70 percent of hospitals and 90 percent of ambulatory practices are not engaged in clinical data
Exhibit 2 Likelihood That Additional Information From The Electronic Health Record (EHR) Would Affect Interpretation And Medical Management Following A Head Computed Tomography (CT) Exam In The Emergency Department (ED), October 1–December 31, 2010 2,000
Number of exams
terpretations in a manner that would influence the medical management of head CT patients in 9.0 percent of cases, ranging from 4.9 percent to 13.4 percent among the three raters (Exhibit 2). In other words, raters predicted that in 179 of the 2,000 patients studied, interpretations would “very likely” have been adversely affected by the lack of clinical information that would have otherwise been readily available via interaction with the EHR. On average, additional clinical information was rated as “possibly” having a clinically significant impact on interpretations in 22.3 percent (ranging from 14.0 percent to 38.4 percent) of cases (Exhibit 2). There was consensus agreement that additional clinical information would “very likely” influence the interpretation and medical management in 6.1 percent of patients (Exhibit 3). Interestingly, total agreement among all three neuroradiologists that additional clinical data were “very likely” to have had an impact on interpretation and medical management occurred in 3.4 percent of cases, compared to a consensus, two out of three neuroradiologists, in 2.7 percent of cases. Consensus agreement that additional clinical data could have “possibly” had an impact on interpretation and medical management was 16 percent (Exhibit 3). Either ED-entered and EHR-retrieved clinical data for the remainder of the 2,000 exams exhibited medically identical information, or the additional information was thought to be of no significance. We found an inverse relationship between the estimated impact of additional clinical data and the experience level of neuroradiologists (Exhibit 2). Rater 3, who had thirty-one years of clinical experience, estimated the highest clinical impact of additional EHR-retrieved data. This was followed by rater 2 with sixteen years of experience, followed by rater 1 with only two years of clinical experience beyond fellowship. In other words, the more experience the rater had, the more “very likely” responses were given. An inspection of the types of additional clinically relevant information that the neuroradiologists identified in the records (Exhibit 4) revealed some commonalities including cancer history (such as glioma, lymphoma, breast cancer, and metastasis), treatment history (such as prior brain surgery or radiation therapy), elevated risk of hemorrhage, symptoms of infection, immunocompetency, immigrant status, pregnancy status, metabolic derangements, and laboratory values. Stroke symptoms (for example, limb weakness, aphasia, or visual deficits) and seizure symptoms that can be localized to a specific region in the brain were thought important by the interpreting radiologists to search those areas on the CT scans. Outright incorrect ED-
Very likely 81.1%
1,500
Possibly
Unlikely
76.7%
1,000
48.2% 38.4%
500 14.0%
0
4.9%
14.7% 8.6%
Rater 1
Rater 2
13.4%
Rater 3
SOURCE Authors’ analysis of 2,000 head CT exams ordered by the ED at the Froedtert and the Medical College of Wisconsin, in Milwaukee, with the likelihood (very likely, possibly, or unlikely) that additional information retrieved from the EHR would affect interpretation and medical management. NOTE Clinical experience for each of the neuroradiologist raters is as follows: rater 1, two years; rater 2, sixteen years; rater 3, thirty-one years.
M ay 2 0 1 4
33 : 5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Health A ffairs
803
Emergency Care Exhibit 3 Likelihood That Electronic Health Record (EHR) Information Would Affect The Interpretation Of Head Computed Tomography (CT) Exams, Based On Agreement Between Raters Agreement between raters Total agreement (3 out of 3) Consensus agreement (2 out of 3) Sum
Very likely to affect 3.4 % 2.7 6.1
Number of patients 68 55 123
Possibly would affect 5.9% 10.1 16.0
Number of patients 117 202 319
SOURCE Authors’ analysis of 2,000 head CT exams ordered by the emergency department at the Froedtert and the Medical College of Wisconsin, in Milwaukee, reviewed by three neuroradiologist raters between October 1 and December 31, 2010.
exchange. They also found that 74 percent of health information exchange efforts report struggling to develop a sustainable business model and that they are heavily dependent on grant funding, primarily from the federal government.15 Radiology is a field particularly suited to remote health care access and delivery, but at the same time it is vulnerable to informational risks, including lack of or inaccurate clinical data, that may influence the interpretation. Radiology is naturally shaped by advancements or failures in information management (such as PACS, network-based image distribution, digital imaging, and teleradiology). Teleradiology, in particular, bears the burden of additive communication risks, which is best tempered by physicianto-physician contact or, more practically, direct access to the EHR. For remote image viewing scenarios, whether within the same institution or between institutions, the integration of EHR
data into existing information technology has the potential to improve the quality and safety of radiologic practice. Our data are the first to show the potential positive benefit of accessing the EHR in optimizing radiological interpretations. Our results support the notion that EHR access is critical for accurate neuroradiology interpretations of ED head CTs. Specifically, we found that use of additional clinical information retrieved from the EHR was thought by expert consensus to very likely influence interpretations and affect patient management in an estimated 6.1 percent of ED head CT examinations. The differences in ED-entered clinical indications and what is retrieved from the EHR are likely related to the unpredictable influence of the clinical data on any given radiologic interpretation. It is virtually impossible for an ED physician to predict all of the medical information that might influence a radiologic inter-
Exhibit 4 Additional Clinical Information Retrieved From Electronic Health Records (EHRs) Reviewed By Three Neuroradiologist Raters, October 1–December 31, 2010 Patient number
ED-entered indication
6 113
Seizure HA
280
Pain
Possible seizure. H/o astrocytoma, grade II, post resection in May 2008 and chemoradiation therapy. 74-year-old female with h/o HIV and smooth muscle tumor of the brain with right-sided HA and occasional visual disturbance for last 10 days. 67-year-old female with HA and expressive aphasia
404
Pain
Fall, right-sided weakness, required intubation. H/o lung mets to brain, managed with whole brain irradiation.
412
AMS
Agitation, history of developmental delay, NF-1, resection of atypical meningioma on 12/30/2009. H/o bilateral MCA stroke and epilepsy. H/o optic glioma irradiated as a child.
522 1045
HA Dizzy
74-year-old female with h/o CLL in remission, presented with fever, HA and photophobia. Loss of balance, chills, hot flashes. Prior h/o cerebellar GBM, s/p resection.
1537
Dizziness
Several hour h/o dizziness, left leg weakness. H/o lymphoma in remission (presented with T12 mass).
1917 1958
AMS Leg numbness
Status epilepticus, recently admitted for seizures at CMH hospital. H/o breast cancer. INR-5.4 yesterday. 65-year-old male with h/o lung cancer with known brain mets, s/p cyber knife. Patient now experiencing right lower extremity numbness.
EHR-retrieved indication
SOURCE Examples of additional clinical data retrieved from the electronic health records rated as “very likely” to improve head CT interpretations and impact on patient management. NOTES ED is emergency department. HA is headache. H/o is history of. AMS is altered mental status. NF-1 is neurofibromatosis type 1. MCA is middle cerebral artery. CLL is chronic lymphocytic leukemia. GBM is glioblastoma multiforme.
804
H e a lt h A f fai r s
May 2 014
33:5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Our results suggest that access to the EHR is necessary to radiologic decision making in the ED.
pretation. Likewise, image interpretation via the PACS is supported and influenced by EHR clinical data, and the process is interactive. Analysis of image data often triggers manipulation of image data in the PACS as well as targeted mining of the EHR clinical data. The opposite is also true. Data discovered in the EHR may trigger additional targeted analysis of image data and additional targeted review of EHR clinical data. This interactive and integrated approach optimizes interpretation accuracy and is virtually impossible for ED physicians and radiologists alike to predict ahead of time. Additional information obtained from the patient or the EHR during the interval between ordering the head CT and its interpretation may contribute to informational discrepancies in some cases, particularly in scenarios where delay in obtaining the study may harm the patient. Additional information gleaned from the EHR that was rated as very likely to affect medical care included cancer history, treatment history, elevated risk of hemorrhage, symptoms of infection, immunocompetency, immigrant status, pregnancy status, metabolic derangements, and laboratory values. Symptoms of potential stroke and seizures that can be localized to a specific region in the brain were sought by interpreting radiologists to optimize head CT search patterns. Additional EHR-derived data can have a dramatic impact on the positive and negative predictive values of noncontrast head CT data, on diagnostic impressions, and on further workup recommendations, all of which may guide treatment algorithms. Take, for example, patient 1917, who presented with a provided history of altered mental status. This presentation is most often due to systemic causes and typically shows no active findings on a head CT. A diagnostic impression in this scenario would likely read “no active disease.” However, the additional information of a history of breast cancer and seizures would prompt the radiologist to inform the ED physician that a noncontrast CT is not sensi-
tive to detecting brain metastases, and a strong recommendation for obtaining an MRI would be made. Failure to do so would risk initiation of medical therapies to control symptoms, while allowing progression of untreated brain metastases until the patient presents with recurrent symptoms. Another example is patient 1537, who presented with a provided history of dizziness. This has many causes, most of which are benign, are treated medically, and typically show no findings evident on a head CT. The additional history of lymphoma and a spinal mass would immediately prompt a discussion of the insensitivity of a CT for detecting leptomeningeal and cranial nerve metastases, followed by recommendations for an MRI as well as a lumbar puncture for cerebrospinal fluid analysis. An interesting observation lies in the variability among raters’ scores. While there was high agreement in designating the presence of additional clinical data, there was considerable variability in the perceived interpretive significance of such additional data. This indicates that individual radiologists are influenced by clinical data differently. Some may tailor their impressions more closely to the stated clinical indication or may be more influenced by the provided information. It is interesting that the pattern of our small rater sample reveals that the most experienced neuroradiologist had the greatest sensitivity to the clinical history, and vice versa. As neuroradiologists gain experience, the integration of clinical and image data may enable a more pertinent and detailed result. However, this assumption would have to be confirmed with larger data sets. More importantly, the data suggest that a one-size-fits-all approach to providing important clinical information will be unsuccessful. Full access to known medical data through the EHR seems essential to empowering each neuroradiologist with the information he or she requires to render an accurate interpretation. The clinical indication entered directly by ED physicians remains critical as well, because this is the only mechanism to relay insights from physicians who have direct contact with patients and their current problems.
Conclusion In summary, our results suggest that access to the EHR is necessary to radiologic decision making in the ED setting. Health care providers must recognize the value of implementing EHRs and the potential harm that may come from their absence, while policy makers need to continue to work toward nationwide exchange of health information. ▪ M ay 2 0 1 4
33 : 5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Health A ffairs
805
Emergency Care
The authors thank Allen D. Elster of Wake Forest University Medical School, in Winston-Salem, North Carolina, for his thoughtful comments and input. The
authors attest that there are no conflicts of interest, financial or otherwise. The corresponding author acknowledges that he had full access to
all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
NOTES 1 Buntin MB, Burke MF, Hoaglin MC, Blumenthal D. The benefits of health information technology: a review of the recent literature shows predominantly positive results. Health Aff (Millwood). 2011;30(3):464–71. 2 Chaudhry B, Wang J, Wu S, Maglione M, Mojica W, Roth E, et al. Systematic review: impact of health information technology on quality, efficiency, and costs of medical care. Ann Intern Med. 2006;144(10): 742–52. 3 Black AD, Car J, Pagliari C, Anandan C, Cresswell K, Bokun T, et al. The impact of eHealth on the quality and safety of health care: a systemic overview. PLoS Med. 2011;8(1): e1000387. 4 Blumenthal D. Implementation of the federal health information technology initiative. N Engl J Med. 2011;365(25):2426–31. 5 Centers for Medicare and Medicaid Service. Medicare and Medicaid programs: electronic health record incentive program—stage 2. Final rule. Fed Regist. 2012;77(171): 53967–4162. 6 DesRoches CM, Worzala C, Bates S. Some hospitals are falling behind in
806
H e a lt h A f fai r s
M ay 2 0 1 4
33 : 5
7
8
9
10
11
meeting “meaningful use” criteria and could be vulnerable to penalties in 2015. Health Aff (Millwood). 2013;32(8):1355–60. DesRoches CM, Charles D, Furukawa MF, Joshi MS, Kralovec P, Mostashari F, et al. Adoption of electronic health records grows rapidly, but fewer than half of US hospitals had at least a basic system in 2012. Health Aff (Millwood). 2013;32(8):1478–85. Patel V, Jamoom EW, Hsiao CJ, Furukawa MF, Buntin M. Variation in electronic health record adoption and readiness for meaningful use: 2008–2011. J Gen Intern Med. 2013;28(7):957–64. Wolf L, Harvell J, Jha AK. Hospitals ineligible for federal meaningful-use incentives have dismally low rates of adoption of electronic health records. Health Aff (Millwood). 2012;31(3):505–13. Schuster DM, Gale ME. The malady of incomplete, inadequate, and inaccurate radiology requisition histories: a computerized treatment. AJR Am J Roentgenol. 1996;167(4): 855–9. Loy CT, Irwig L. Accuracy of diag-
12
13
14
15
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
nostic tests read with and without clinical information: a systemic review. JAMA. 2004;292(13):1602–9. Department of Health and Human Services. State Health Information Exchange Cooperative Agreement Program [Internet]. Washington (DC): HHS; [updated 2011 Feb 17; cited 2014 Mar 31]. Available from: http://www.healthit.gov/policyresearchers-implementers/statehealth-information-exchange Williams C, Mostashari F, Mertz K, Hogin E, Atwal P. From the Office of the National Coordinator: the strategy for advancing the exchange of health information. Health Aff (Millwood). 2012;31(3):527–36. Adler-Milstein J, Jha AK. Sharing clinical data electronically: a critical challenge for fixing the health care system. JAMA. 2012;307(16): 1695–6. Adler-Milstein J, Bates DW, Jha AK. Operational health information exchanges show substantial growth, but long-term funding remains a concern. Health Aff (Millwood). 2013;32(8):1486–92.
The online version of this article, along with updated information and services, is available at: http://content.healthaffairs.org/content/33/5/800.full.html
For Reprints, Links & Permissions: http://healthaffairs.org/1340_reprints.php E-mail Alerts : http://content.healthaffairs.org/subscriptions/etoc.dtl To Subscribe: http://content.healthaffairs.org/subscriptions/online.shtml
Health Affairs is published monthly by Project HOPE at 7500 Old Georgetown Road, Suite 600, Bethesda, MD 20814-6133. Copyright © 2014 by Project HOPE - The People-to-People Health Foundation. As provided by United States copyright law (Title 17, U.S. Code), no part of Health Affairs may be reproduced, displayed, or transmitted in any form or by any means, electronic or mechanical, including photocopying or by information storage or retrieval systems, without prior written permission from the Publisher. All rights reserved.
Not for commercial use or unauthorized distribution Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Emergency Care By Michael J. Franczak, Madeline Klein, Flavius Raslau, Jo Bergholte, Leighton P. Mark, and John L. Ulmer 10.1377/hlthaff.2013.0171 HEALTH AFFAIRS 33, NO. 5 (2014): 800–806 ©2014 Project HOPE— The People-to-People Health Foundation, Inc.
doi:
Michael J. Franczak is an undergraduate student in the College of Science at the University of Notre Dame, in South Bend, Indiana. At the time of the study he was a research assistant in the Department of Radiology at the Medical College of Wisconsin, in Milwaukee. Madeline Klein is a volunteer with Jesuit Volunteer Corps, in Boston, Massachusetts. At the time of the study she was a research assistant in the Department of Radiology at the Medical College of Wisconsin. Flavius Raslau is an assistant professor of radiology at the College of Medicine, University of Kentucky, in Lexington. At the time of the study he was a neuroradiology fellow in the Department of Radiology at the Medical College of Wisconsin. Jo Bergholte is a program manager at the Medical College of Wisconsin. Leighton P. Mark is a professor of radiology at the Medical College of Wisconsin. John L. Ulmer (julmer@ mcw.edu) is professor of radiology at the Medical College of Wisconsin.
800
In Emergency Departments, Radiologists’ Access To EHRs May Influence Interpretations And Medical Management ABSTRACT The adoption of electronic health records (EHRs) that meet federal meaningful-use standards is a major US national policy priority. Policy makers recognize the potential of electronic communication in delivering high-quality health care, particularly in an environment of expanding remote access to medical care and the ever-increasing need to transmit health care records across institutions. To demonstrate this principle, we sought to estimate the significance of EHR access in emergent neuroradiologic interpretations. Three neuroradiologists conducted a prospective expert-rater analysis of 2,000 consecutive head computed tomography (CT) exams ordered by emergency department (ED) physicians. For each head CT exam, the neuroradiologists compared medical information generated by ED physicians to information generated by the interpreting radiologists who had access to additional EHR-derived patient data. In 6.1 percent of the head CT exams, the neuroradiologists reached consensus—meaning two out of three agreed— that the additional clinical data derived from the EHR was “very likely” to influence radiological interpretations and that the lack of that data would have adversely affected medical management in those patients. Health care providers must recognize the value of implementing EHRs and foster their widespread adoption.
E
lectronic health records (EHRs) have been shown to improve the efficiency of health care delivery and have the potential to improve health care quality.1–3 The underlying principle is that sharing of, and immediate access to, patient health information among providers and institutions promotes more effective communication of critical data that may influence and guide clinical decision making. Thus, electronic communication has the potential to improve health care quality and delivery both within and across institutions and is particularly important in an environment of expanding remote access to medical care. Although this seems intuitive, data supporting this principle across
Health Affairs
May 2 014
33:5
health care fields are not substantial. Policy makers recognize the potential importance of the EHR. The adoption of EHRs that meet federal meaningful-use standards has been a major US national policy priority for several years. In 2009 the Health Information Technology for Economic and Clinical Health (HITECH) Act authorized nearly $30 billion in funding to support the adoption of EHRs that achieve federal meaningful-use standards.4 Hospitals and physicians’ offices are being rewarded for meeting (or penalized in later years for failing to meet) specific information system capability and reporting clinical quality measures.5,6 Since the incentives have gone into effect, there has been a substantial growth in EHR adoption.7 However,
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
there is a significant discrepancy in the rate of EHR adoption among providers, especially among those who are ineligible to receive such federal incentives.8,9 Catherine DesRoches and colleagues showed that as of 2012, 16.7 percent of US acute care hospitals used a comprehensive electronic record system, and 44.0 percent had a basic system.7 This increase appeared to be concentrated around large, Midwestern, and major teaching hospitals. In contrast, hospitals not eligible for federal incentives have dismally low rates of adoption of an EHR system. Only 12 percent of short-term acute care hospitals have basic EHR systems, compared with only 6 percent of long-term acute care hospitals.9 Access to EHRs and the electronic exchange of health information seems especially relevant to radiology, a technology-driven specialty that has been at the forefront of development of information technology including picture archiving and communication systems (PACS), web-based image distribution, and the expansion of teleradiology. The very technology-driven nature of radiology, which provides for remote access and delivery, also makes the field vulnerable to informational risks. The negative impact of incomplete clinical information on radiological interpretations has been well documented in several studies.10,11 To our knowledge, however, there are no studies investigating the interpretive value of EHRs in delivering radiologic services. Yet the field is ideally suited to test the hypothesis that EHRs can improve medical communication in a meaningful way to improve health care quality. We chose to study the use of EHRs in emergency department (ED) radiologic services because of the need for rapid turnaround time, the immediate initiation of therapy, and the impact of communication on early treatment. We compared medical information pertinent to emergent head computed tomography (CT) examinations as entered by referring ED physicians to additional EHR-derived information retrieved by interpreting radiologists. We sought to determine how often the EHR that was accessed by the interpreting radiologist was perceived to provide relevant additional information and to estimate the importance of those additional data in interpreting imaging studies.
Study Data And Methods Head Computed Tomography Exams And Work Flow The study was approved by the Medical College of Wisconsin Institutional Review Board. The Froedtert and the Medical College of Wisconsin, in Milwaukee, contains a statedesignated Level I regional trauma center that
provides twenty-four-hour ED head CT availability and interpretations year round. Radiologic interpretations for ED patients are provided via on-site teleradiology, using PACS and the EHR for exam reporting. PACS is a medical technology that provides storage, retrieval, communication, database, access to images from multiple modalities, and user interface components that in many ways resemble the EHR system. Combined with available and emerging web technology, PACS has the ability to deliver timely and efficient access to images, interpretations, and related data within and across institutions. Head CT requests from ED physicians require ED personnel to enter pertinent medical information into the electronic order entry system. As a matter of institutional protocol, interpretations are reported to ED physicians within one hour of the head CT examination. Interpretations within this one-hour time frame are delivered by neuroradiology faculty, neuroradiology fellows, or radiology residents. Interpreting radiologists have full access to the EHR. Thus, within one hour of the ED head CT request, the interpreting radiologist reviews the request, reviews additional EHR and imaging data, and reports the results via the EHR. Written interpretations include additional pertinent clinical information retrieved from the EHR in the indication section of the report. It is important to note that the EHR and imaging data may be reviewed simultaneously and that reviewing image and EHR data may trigger review of additional pertinent EHR data. In other words, the use of the EHR to support image interpretation is interactive. Patient Data Data on emergency department head CT scans were collected soon after the institutionwide implementation of EHRs. Retrospective data were obtained from 2,000 consecutive ED head CTexams acquired from October 1, 2010, to December 31, 2010. The institutional EHR stores all entered data, which could be accessed retrospectively. Thus, pertinent clinical information entered by ED personnel at the time of the request could be compared to any additional EHR-retrieved clinical information entered by the radiologist at the time of interpretation. Demographic patient information including age, sex, and race or ethnicity is linked to the image data and are, thus, always available. Of the 2,000 patient exams reviewed, 49.2 percent were males, and 50.8 percent were females. A mean patient age of fifty-four was observed, with a range from eighteen to ninety-nine years. The race or ethnicity distribution of patients scanned was 58.4 percent Caucasian, 34.4 percent African American, 3.2 percent Hispanic, and 4.1 percent other. M ay 2 0 1 4
33 : 5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Health A ffairs
801
Emergency Care Expert Analysis Three neuroradiologists with clinical imaging experience of two, sixteen, and thirty-one years, respectively, and identified for the purposes of this article as raters number 1, 2, or 3, respectively, independently compared the ED-entered clinical information for each head CT examination along with any additional EHR-retrieved information entered by the interpreting radiologist. The neuroradiologists rated discrepancies between the ED-entered clinical information and the information entered by the interpreting radiologist. The ratings were based on criteria described below. First, the medical content of ED-entered and radiologist-entered (EHR-retrieved) clinical data was rated using binary scoring of “yes” or “no” to indicate whether additional clinical data were entered by the radiologist into the radiology report. The rating was given without regard for a potential clinical impact of the additional clinical data. Correction of ED-entered spelling or wording errors were not considered additional data. An interrater reliability analysis was performed using the interclass correlation coefficient (ICC) in order to quantify the consistency among the three neuroradiologists in determining the presence of additional medical content. Second, when additional clinical data were present in the record, the neuroradiologists considered the likelihood that the additional EHRretrieved clinical information was clinically significant and “very likely,” “possibly,” or “unlikely” to influence radiological interpretations in a way that would affect medical management. In other words, was the additional information
Exhibit 1 Discrepancies In Medical Content Between Emergency Department (ED)–Entered and Electronic Health Record (EHR)–Retrieved Clinical Data On Head Computed Tomography (CT) Exams, October 1–December 31, 2010
attained from the EHR thought necessary? An effect on medical management might include an effect on imaging diagnoses or recommendations for additional imaging or laboratory tests, or might otherwise influence the workup or treatment algorithms. Third, we calculated the extent to which the neuroradiologists agreed that the additional clinical data could have “very likely” or “possibly” influenced radiological interpretations in a way that would affect medical management. Consensus agreement was set at two out of three raters, and total agreement occurred when all three raters gave the same response. Limitations Our study had several limitations. First, the investigation focused entirely on emergent head CTs performed in a Level 1 trauma center. Therefore, any conclusions drawn from this study might not be fully generalizable to other clinic settings. Any widespread extrapolation of results would be strengthened by multicenter studies. Second, our institution has full access to EHRs in all ED settings. Therefore, it is ethically untenable to design a randomized study where the EHR is not accessed in a significant proportion of patients. Consequently, we were unable to determine the actual impact of additional EHR-retrieved clinical data on radiologic interpretations or patient management; rather, we chose to calculate theoretical consequences. Third, we chose to study the impact on radiologic services in the ED setting, which requires rapid interpretation turnaround and is highly vulnerable to informational risks. Thus, the high estimated impact of EHR access might not extrapolate to other areas of radiology. Finally, larger patient series in multi-institutional studies would strengthen the statistical confidence of the results obtained in this preliminary investigation.
1,200
Study Results
Number of exams
1,000 800 600 400
54.9% 44.1%
48.9%
200 0 Rater 1
Rater 2
Rater 3
SOURCE Authors’ analysis of 2,000 head CT exams ordered by the ED at the Froedtert and the Medical College of Wisconsin, in Milwaukee, and reviewed by three neuroradiologists. NOTE Clinical experience for each of the neuroradiologist raters is as follows: rater 1, two years; rater 2, sixteen years; rater 3, thirty-one years.
802
Health Affairs
May 2 014
33:5
The three neuroradiologist raters identified that additional clinical data were present in 49.3 percent of the head CT exams. This represents an average of the raters’ findings, which ranged from 44.1 percent to 54.9 percent (Exhibit 1). For the remainder of cases, the raters determined that no additional clinical data were present in the record. The interrater reliability was ICC equals 0.824 (p < 0:001) and a 95 percent confidence interval of 0.810, 0.837, indicating strong agreement among the raters on the presence of additional medical content. On average, the additional clinical data available in the EHR was designated by individual expert raters as being “very likely” to affect in-
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
entered clinical data were exceedingly rare and may have been related to communication errors among ED personnel.
Discussion There is a broad consensus regarding the potential benefits of the EHR, especially the optimization of information sharing among providers. The HITECH Act authorized the Centers for Medicare and Medicaid Services to promote meaningful use of certified EHR technology, with penalties for lack of EHR use beginning in 2015. Health information exchange is another priority of HITECH, designed to improve the quality and efficiency of health care across institutions as well. Policy makers have been actively promoting the development of broad-based electronic health information exchange systems. In 2010 the Office of the National Coordinator for Health Information Technology (ONC) established the State Health Information Exchange Cooperative Agreement Program, which awarded more than $540 million to states in efforts to build capacity for exchanging health information among health care systems both within and across state lines. Barriers that have hindered widespread health information sharing include high cost, limited across-institution interoperability, and low motivation to share information in a fee-for-service environment.12–14 Julia Adler-Milstein and colleagues found that 70 percent of hospitals and 90 percent of ambulatory practices are not engaged in clinical data
Exhibit 2 Likelihood That Additional Information From The Electronic Health Record (EHR) Would Affect Interpretation And Medical Management Following A Head Computed Tomography (CT) Exam In The Emergency Department (ED), October 1–December 31, 2010 2,000
Number of exams
terpretations in a manner that would influence the medical management of head CT patients in 9.0 percent of cases, ranging from 4.9 percent to 13.4 percent among the three raters (Exhibit 2). In other words, raters predicted that in 179 of the 2,000 patients studied, interpretations would “very likely” have been adversely affected by the lack of clinical information that would have otherwise been readily available via interaction with the EHR. On average, additional clinical information was rated as “possibly” having a clinically significant impact on interpretations in 22.3 percent (ranging from 14.0 percent to 38.4 percent) of cases (Exhibit 2). There was consensus agreement that additional clinical information would “very likely” influence the interpretation and medical management in 6.1 percent of patients (Exhibit 3). Interestingly, total agreement among all three neuroradiologists that additional clinical data were “very likely” to have had an impact on interpretation and medical management occurred in 3.4 percent of cases, compared to a consensus, two out of three neuroradiologists, in 2.7 percent of cases. Consensus agreement that additional clinical data could have “possibly” had an impact on interpretation and medical management was 16 percent (Exhibit 3). Either ED-entered and EHR-retrieved clinical data for the remainder of the 2,000 exams exhibited medically identical information, or the additional information was thought to be of no significance. We found an inverse relationship between the estimated impact of additional clinical data and the experience level of neuroradiologists (Exhibit 2). Rater 3, who had thirty-one years of clinical experience, estimated the highest clinical impact of additional EHR-retrieved data. This was followed by rater 2 with sixteen years of experience, followed by rater 1 with only two years of clinical experience beyond fellowship. In other words, the more experience the rater had, the more “very likely” responses were given. An inspection of the types of additional clinically relevant information that the neuroradiologists identified in the records (Exhibit 4) revealed some commonalities including cancer history (such as glioma, lymphoma, breast cancer, and metastasis), treatment history (such as prior brain surgery or radiation therapy), elevated risk of hemorrhage, symptoms of infection, immunocompetency, immigrant status, pregnancy status, metabolic derangements, and laboratory values. Stroke symptoms (for example, limb weakness, aphasia, or visual deficits) and seizure symptoms that can be localized to a specific region in the brain were thought important by the interpreting radiologists to search those areas on the CT scans. Outright incorrect ED-
Very likely 81.1%
1,500
Possibly
Unlikely
76.7%
1,000
48.2% 38.4%
500 14.0%
0
4.9%
14.7% 8.6%
Rater 1
Rater 2
13.4%
Rater 3
SOURCE Authors’ analysis of 2,000 head CT exams ordered by the ED at the Froedtert and the Medical College of Wisconsin, in Milwaukee, with the likelihood (very likely, possibly, or unlikely) that additional information retrieved from the EHR would affect interpretation and medical management. NOTE Clinical experience for each of the neuroradiologist raters is as follows: rater 1, two years; rater 2, sixteen years; rater 3, thirty-one years.
M ay 2 0 1 4
33 : 5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Health A ffairs
803
Emergency Care Exhibit 3 Likelihood That Electronic Health Record (EHR) Information Would Affect The Interpretation Of Head Computed Tomography (CT) Exams, Based On Agreement Between Raters Agreement between raters Total agreement (3 out of 3) Consensus agreement (2 out of 3) Sum
Very likely to affect 3.4 % 2.7 6.1
Number of patients 68 55 123
Possibly would affect 5.9% 10.1 16.0
Number of patients 117 202 319
SOURCE Authors’ analysis of 2,000 head CT exams ordered by the emergency department at the Froedtert and the Medical College of Wisconsin, in Milwaukee, reviewed by three neuroradiologist raters between October 1 and December 31, 2010.
exchange. They also found that 74 percent of health information exchange efforts report struggling to develop a sustainable business model and that they are heavily dependent on grant funding, primarily from the federal government.15 Radiology is a field particularly suited to remote health care access and delivery, but at the same time it is vulnerable to informational risks, including lack of or inaccurate clinical data, that may influence the interpretation. Radiology is naturally shaped by advancements or failures in information management (such as PACS, network-based image distribution, digital imaging, and teleradiology). Teleradiology, in particular, bears the burden of additive communication risks, which is best tempered by physicianto-physician contact or, more practically, direct access to the EHR. For remote image viewing scenarios, whether within the same institution or between institutions, the integration of EHR
data into existing information technology has the potential to improve the quality and safety of radiologic practice. Our data are the first to show the potential positive benefit of accessing the EHR in optimizing radiological interpretations. Our results support the notion that EHR access is critical for accurate neuroradiology interpretations of ED head CTs. Specifically, we found that use of additional clinical information retrieved from the EHR was thought by expert consensus to very likely influence interpretations and affect patient management in an estimated 6.1 percent of ED head CT examinations. The differences in ED-entered clinical indications and what is retrieved from the EHR are likely related to the unpredictable influence of the clinical data on any given radiologic interpretation. It is virtually impossible for an ED physician to predict all of the medical information that might influence a radiologic inter-
Exhibit 4 Additional Clinical Information Retrieved From Electronic Health Records (EHRs) Reviewed By Three Neuroradiologist Raters, October 1–December 31, 2010 Patient number
ED-entered indication
6 113
Seizure HA
280
Pain
Possible seizure. H/o astrocytoma, grade II, post resection in May 2008 and chemoradiation therapy. 74-year-old female with h/o HIV and smooth muscle tumor of the brain with right-sided HA and occasional visual disturbance for last 10 days. 67-year-old female with HA and expressive aphasia
404
Pain
Fall, right-sided weakness, required intubation. H/o lung mets to brain, managed with whole brain irradiation.
412
AMS
Agitation, history of developmental delay, NF-1, resection of atypical meningioma on 12/30/2009. H/o bilateral MCA stroke and epilepsy. H/o optic glioma irradiated as a child.
522 1045
HA Dizzy
74-year-old female with h/o CLL in remission, presented with fever, HA and photophobia. Loss of balance, chills, hot flashes. Prior h/o cerebellar GBM, s/p resection.
1537
Dizziness
Several hour h/o dizziness, left leg weakness. H/o lymphoma in remission (presented with T12 mass).
1917 1958
AMS Leg numbness
Status epilepticus, recently admitted for seizures at CMH hospital. H/o breast cancer. INR-5.4 yesterday. 65-year-old male with h/o lung cancer with known brain mets, s/p cyber knife. Patient now experiencing right lower extremity numbness.
EHR-retrieved indication
SOURCE Examples of additional clinical data retrieved from the electronic health records rated as “very likely” to improve head CT interpretations and impact on patient management. NOTES ED is emergency department. HA is headache. H/o is history of. AMS is altered mental status. NF-1 is neurofibromatosis type 1. MCA is middle cerebral artery. CLL is chronic lymphocytic leukemia. GBM is glioblastoma multiforme.
804
H e a lt h A f fai r s
May 2 014
33:5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Our results suggest that access to the EHR is necessary to radiologic decision making in the ED.
pretation. Likewise, image interpretation via the PACS is supported and influenced by EHR clinical data, and the process is interactive. Analysis of image data often triggers manipulation of image data in the PACS as well as targeted mining of the EHR clinical data. The opposite is also true. Data discovered in the EHR may trigger additional targeted analysis of image data and additional targeted review of EHR clinical data. This interactive and integrated approach optimizes interpretation accuracy and is virtually impossible for ED physicians and radiologists alike to predict ahead of time. Additional information obtained from the patient or the EHR during the interval between ordering the head CT and its interpretation may contribute to informational discrepancies in some cases, particularly in scenarios where delay in obtaining the study may harm the patient. Additional information gleaned from the EHR that was rated as very likely to affect medical care included cancer history, treatment history, elevated risk of hemorrhage, symptoms of infection, immunocompetency, immigrant status, pregnancy status, metabolic derangements, and laboratory values. Symptoms of potential stroke and seizures that can be localized to a specific region in the brain were sought by interpreting radiologists to optimize head CT search patterns. Additional EHR-derived data can have a dramatic impact on the positive and negative predictive values of noncontrast head CT data, on diagnostic impressions, and on further workup recommendations, all of which may guide treatment algorithms. Take, for example, patient 1917, who presented with a provided history of altered mental status. This presentation is most often due to systemic causes and typically shows no active findings on a head CT. A diagnostic impression in this scenario would likely read “no active disease.” However, the additional information of a history of breast cancer and seizures would prompt the radiologist to inform the ED physician that a noncontrast CT is not sensi-
tive to detecting brain metastases, and a strong recommendation for obtaining an MRI would be made. Failure to do so would risk initiation of medical therapies to control symptoms, while allowing progression of untreated brain metastases until the patient presents with recurrent symptoms. Another example is patient 1537, who presented with a provided history of dizziness. This has many causes, most of which are benign, are treated medically, and typically show no findings evident on a head CT. The additional history of lymphoma and a spinal mass would immediately prompt a discussion of the insensitivity of a CT for detecting leptomeningeal and cranial nerve metastases, followed by recommendations for an MRI as well as a lumbar puncture for cerebrospinal fluid analysis. An interesting observation lies in the variability among raters’ scores. While there was high agreement in designating the presence of additional clinical data, there was considerable variability in the perceived interpretive significance of such additional data. This indicates that individual radiologists are influenced by clinical data differently. Some may tailor their impressions more closely to the stated clinical indication or may be more influenced by the provided information. It is interesting that the pattern of our small rater sample reveals that the most experienced neuroradiologist had the greatest sensitivity to the clinical history, and vice versa. As neuroradiologists gain experience, the integration of clinical and image data may enable a more pertinent and detailed result. However, this assumption would have to be confirmed with larger data sets. More importantly, the data suggest that a one-size-fits-all approach to providing important clinical information will be unsuccessful. Full access to known medical data through the EHR seems essential to empowering each neuroradiologist with the information he or she requires to render an accurate interpretation. The clinical indication entered directly by ED physicians remains critical as well, because this is the only mechanism to relay insights from physicians who have direct contact with patients and their current problems.
Conclusion In summary, our results suggest that access to the EHR is necessary to radiologic decision making in the ED setting. Health care providers must recognize the value of implementing EHRs and the potential harm that may come from their absence, while policy makers need to continue to work toward nationwide exchange of health information. ▪ M ay 2 0 1 4
33 : 5
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
Health A ffairs
805
Emergency Care
The authors thank Allen D. Elster of Wake Forest University Medical School, in Winston-Salem, North Carolina, for his thoughtful comments and input. The
authors attest that there are no conflicts of interest, financial or otherwise. The corresponding author acknowledges that he had full access to
all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
NOTES 1 Buntin MB, Burke MF, Hoaglin MC, Blumenthal D. The benefits of health information technology: a review of the recent literature shows predominantly positive results. Health Aff (Millwood). 2011;30(3):464–71. 2 Chaudhry B, Wang J, Wu S, Maglione M, Mojica W, Roth E, et al. Systematic review: impact of health information technology on quality, efficiency, and costs of medical care. Ann Intern Med. 2006;144(10): 742–52. 3 Black AD, Car J, Pagliari C, Anandan C, Cresswell K, Bokun T, et al. The impact of eHealth on the quality and safety of health care: a systemic overview. PLoS Med. 2011;8(1): e1000387. 4 Blumenthal D. Implementation of the federal health information technology initiative. N Engl J Med. 2011;365(25):2426–31. 5 Centers for Medicare and Medicaid Service. Medicare and Medicaid programs: electronic health record incentive program—stage 2. Final rule. Fed Regist. 2012;77(171): 53967–4162. 6 DesRoches CM, Worzala C, Bates S. Some hospitals are falling behind in
806
H e a lt h A f fai r s
M ay 2 0 1 4
33 : 5
7
8
9
10
11
meeting “meaningful use” criteria and could be vulnerable to penalties in 2015. Health Aff (Millwood). 2013;32(8):1355–60. DesRoches CM, Charles D, Furukawa MF, Joshi MS, Kralovec P, Mostashari F, et al. Adoption of electronic health records grows rapidly, but fewer than half of US hospitals had at least a basic system in 2012. Health Aff (Millwood). 2013;32(8):1478–85. Patel V, Jamoom EW, Hsiao CJ, Furukawa MF, Buntin M. Variation in electronic health record adoption and readiness for meaningful use: 2008–2011. J Gen Intern Med. 2013;28(7):957–64. Wolf L, Harvell J, Jha AK. Hospitals ineligible for federal meaningful-use incentives have dismally low rates of adoption of electronic health records. Health Aff (Millwood). 2012;31(3):505–13. Schuster DM, Gale ME. The malady of incomplete, inadequate, and inaccurate radiology requisition histories: a computerized treatment. AJR Am J Roentgenol. 1996;167(4): 855–9. Loy CT, Irwig L. Accuracy of diag-
12
13
14
15
Downloaded from content.healthaffairs.org by Health Affairs on June 19, 2015 at NANYANG TECHNOLOGICAL UNIVERSITY NTU
nostic tests read with and without clinical information: a systemic review. JAMA. 2004;292(13):1602–9. Department of Health and Human Services. State Health Information Exchange Cooperative Agreement Program [Internet]. Washington (DC): HHS; [updated 2011 Feb 17; cited 2014 Mar 31]. Available from: http://www.healthit.gov/policyresearchers-implementers/statehealth-information-exchange Williams C, Mostashari F, Mertz K, Hogin E, Atwal P. From the Office of the National Coordinator: the strategy for advancing the exchange of health information. Health Aff (Millwood). 2012;31(3):527–36. Adler-Milstein J, Jha AK. Sharing clinical data electronically: a critical challenge for fixing the health care system. JAMA. 2012;307(16): 1695–6. Adler-Milstein J, Bates DW, Jha AK. Operational health information exchanges show substantial growth, but long-term funding remains a concern. Health Aff (Millwood). 2013;32(8):1486–92.
