Neuroradiolog y/Head and Neck Imaging • Original Research Torok et al. Second Opinion Consultation in Neuroradiology
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Neuroradiology/Head and Neck Imaging Original Research
Neuroradiology Second Opinion Consultation Service: Assessment of Duplicative Imaging Collin M. Torok1 Cindy Lee Paul Nagy David M. Yousem Jonathan S. Lewin Torok CM, Lee C, Nagy P, Yousem DM, Lewin JS
OBJECTIVE. The purpose of this study was to characterize the performance of the Neuroradiology Second Opinion Consultation Service (NSOCS) at our institution to establish the rate, causes, and implications of requests for repeat imaging. MATERIALS AND METHODS. We queried 11,753 complete reports of all NSOCS studies for calendar year 2010 for the words “repeat” and “follow-up.” We categorized study limitations described in these reports into poor image quality, missing or inadequate MR sequences or CT reformats, lack of IV contrast administration where otherwise deemed appropriate, an “other” category for miscellaneous items, and a “clarification” category for indeterminate findings or recommendations for more advanced protocols. The corresponding available electronic medical records were reviewed. An estimated financial analysis of the NSOCS was additionally performed. RESULTS. Repeat imaging studies were recommended in 1.5% of cases. In 0.3% of all cases, a subsequent repeat examination was documented in the electronic medical records. Study limitations were most commonly due to poor image quality (77.5%), followed by missing or inadequate MR sequences or CT reformats (20.3%). The additional estimated cost of repeat imaging was calculated at $14,019.34, with an overall per-patient cost of $2.12 for the service. CONCLUSION. Reviewing outside studies generates a very low rate of requests for and performance of repeat studies, and is not a major additional health care expense.
I
Keywords: duplicative imaging, neuroradiology, outside studies, repeat imaging, second opinion DOI:10.2214/AJR.12.9429 Received June 17, 2012; accepted after revision February 28, 2013. 1
All authors: The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins Hospital, 600 N Wolfe St, Baltimore, MD 21287. Address correspondence to C. M. Torok ([email protected]).
AJR 2013; 201:1096–1100 0361–803X/13/2015–1096 © American Roentgen Ray Society
1096
n an age of progressively increasing health care costs in the United States [1] and increased pressure on radiologists to maximize cost-effectiveness, the elimination of duplicative imaging is an opportunity for potential cost savings. Patients frequently arrive from outside institutions or imaging facilities with radiology studies on CD, on hard copy, or occasionally through a virtual private network (VPN) transfer [2]. Loading these studies into the PACS provides immediate availability of imaging data throughout the enterprise. In turn, these imports enable the most appropriate and expedient diagnostic and therapeutic decisions to be made. Repeating these outside studies should be avoided. Such duplicative imaging encompasses a broad spectrum, ranging from a nonemergent brain MRI study repeated because of degradation of the original outside data through the transfer process or from intrinsically poor quality studies to a transferred trauma patient receiving a second “pan scan” in the emergen-
cy department when time is of the essence and rapid imaging data transfer is unavailable [3]. The Neuroradiology Second Opinion Consultation Service (NSOCS) at our institution frequently faces the former challenge. We have established a mechanism to streamline formal review of outside studies by subspecialty radiologists on request to provide second opinions for patients and clinicians. Outside studies can be scanned into our PACS at any time for clinicians at centralized locations during office hours and by emergency department technologists off-hours, allowing a mutually convenient and appropriate review rather than a cursory “curbside” opinion. Our department policy is that all outside studies are scanned into the PACS and a final report is generated and signed by a neuroradiologist for every study. Despite the availability of this service, however, a variety of reasons may still mandate repeat imaging. This potential for duplicative imaging of second opinion studies raises concern for excessive self-referral.
AJR:201, November 2013
Through analyzing the performance of the NSOCS, we sought to establish the rate, causes, and implications of requests for repeat imaging. We hypothesized that the rate of ordering repeat studies would be less than 10% and hoped that, through characterizing our performance and potentially elucidating problematic areas, we might identify opportunities to optimize the delivery of our service. Materials and Methods The institutional review board approved the retrospective review of patient data for this HIPAAcompliant study and waived the need for individual informed consent. We imported 11,753 complete reports of all NSOCS studies reviewed in calendar year 2010 into a Microsoft Excel spreadsheet from a radiology information system (RIS) search. The database was queried separately with the keyword search function for the words “repeat” and “follow-up.” In the latter case, coincidences with “repeat” were accounted for by additionally searching “follow-up AND repeat.” Spelling and punctuation variants of “follow-up” were accounted for by the search argument %follow%up%. This approach was further validated by random review of 100 complete radiology reports (via the RAND function in Excel). This secondary review yielded no additional repeat recommendations beyond those discovered using the search parameters noted. We reviewed the reports and categorized study limitations into the following five groups: poor image quality (hard copy scanned images, low MRI field strength, and motion artifact); missing or inadequate standard MRI sequences and CT reformats or reconstruction algorithms (missing or inadequate MRI sequences [T2-weighted, FLAIR, fat suppression, diffusion-weighted MRI, STIR], lack of sagittal CT reformats, no CT bone algorithm or window, poor bone subtraction on CT angiography, no source images on MR angiography, and no thin sections on CT angiography); lack of IV contrast administration where otherwise deemed appropriate, which included cases in which no contrast material was administered with indications such as clinically suspected brain mass or neck mass without an identifiable contraindication in available patient records; an “other” category for miscellaneous items, including an incomplete FOV, inability to link multiplanar images or use a measuring tool, inability to window, and poor reconstruction algorithm; and a “clarification” category that was defined to include studies for which repeat recommendations were made to clarify indeterminate findings with or without more advanced protocols or modalities. The corresponding available electronic medical records were reviewed to assess whether the repeat recommendation was performed at our in-
stitution. Additionally, a database of all internal neuroradiology studies for the same period was searched as an internal comparison. Faculty years of experience in practice were compared with individual repeat recommendation rates to determine if experience correlated with repeat request rate using a Spearman correlation coefficient test (n = 13 faculty members). A financial analysis was performed on the basis of the number of each individual type of examination performed during the study time period. This represents a hypothetical analysis of projected technical and professional fee charges rather than exact departmental financial records and actual reimbursement. The corresponding Current Procedural Terminology codes were input into a web-based fee calculator (2012 Medicare Part B Fee Schedule Calculator, Novitas Solutions), and global, technical, and professional fee output was obtained for Maryland Charge class 01 (Baltimore County), assuming a 100% Medicare reimbursement rate. Our institution is a major tertiary care medical center that in calendar year 2012 had 913 beds, 46,775 admissions, 84,982 emergency department visits, and 506,915 outpatient visits. Our neuroradiology division performed an average of 80,851 examinations per year for calendar years 2010 and 2011. Our department as a whole performed an average of 467,042 examinations per year for the same time period. Our practice setting is an academic neuroradiology division that handles referrals from across the country and the world, encompassing the full spectrum of inpatient, outpatient, and emergency department service. The studies arise predominantly from neurosurgery,
neurology, oncology, emergency department, and otorhinolaryngology referrers. On the basis of data categorization available for 2012, the vast majority of studies submitted to our neuroradiology division were on CD or DVD (97.7%) rather than film (2.3%). According to available partial data from 2010, 10,124 of 11,753 (86.1%) outside studies were submitted from the following sites: 87.0% (8806/10,124) outpatient, 10.3% (1038/10,124) inpatient, and 2.7% (280/10,124) emergency department. A separate analysis conducted at our institution in 2008 indicated the following representative pathology case distribution for outside studies: neoplastic (43.6%); infectious, inflammatory, or demyelinating (15.0%); and degenerative (8.3%)—with a wide variety of lesser contributions, including normal studies [4].
Results The keyword search for “repeat” yielded 287 reports, each of which was individually reviewed, and 159 studies were identified in which repeat imaging was recommended. The keyword search for “follow-up” yielded 1981 results, each of which was individually reviewed, yielding 23 additional studies that incorporated strong recommendations for repeating the examinations because of study limitations. The remaining “follow-up” keyword search reports either recommended routine follow-up, vaguely suggested followup if clinically indicated, or used “followup” in a manner unrelated to the impression (e.g., “When compared with the follow-up study, no change was noted”).
160 140 120
No. of Studies
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Second Opinion Consultation in Neuroradiology
100 80 60 40 20 0 Poor Image Missing or Quality Inadequate MRI/CT
Clarification
Lack of IV Contrast Administration
Other
Fig. 1—Graph shows data summary of overall categories. Note that multiple different types of limitations were reported for individual studies.
AJR:201, November 2013 1097
1098
Fig. 2—Graph of poor image quality category data summary shows most common factor was hard copy scanned images.
140 120
No. of Studies
100 80 60 40 20 0
Hard Copy Scanned
Motion Artifact
Low Magnetic Field Strength
14 12 10 No. of Studies
Repeat imaging studies were recommended by neuroradiologists in 1.5% of cases (182/11,753); 32 were CT and 150 were MRI (95% CI, 0.013–0.018) based on our search parameters. In 20.9% (38) of these cases (six CT and 32 MR; 0.154–0.277) and 0.3% of all cases (38/11,753; 0.002–0.004), a documented repeat examination was identified in the available clinical records. The search of internal neuroradiology studies for the same period as a control yielded 128 results. Each of these reports was individually reviewed, and 33 studies were identified in which repeat imaging was recommended, yielding a 0.1% (33/25,090; 0.001–0.002) radiologist repeat recommendation rate and a corresponding 0.04% (9/25,090; 0.0002–0.0008) actual repeat imaging rate based on follow-up clinic notes. Figures 1–4 summarize the data collected for each of the categories described in the Materials and Methods section. Poor image quality was the most common factor cited in recommendations for repeat examinations (77.5%; 141/182). The hard copy scanned nature of the images (i.e., a radiograph that was digitized) was directly cited in 51.1% (93/182) of cases (Fig. 5). A common speech recognition dictation macro stating either that the images are not of the quality usually generated at our institution or that the images were not in DICOM format and repeat examination should be considered was cited in 28.6% (52/182) of cases. This macro was accompanied by at least one definable limitation in the report body from the various categories described in 53.8% (28/52) of such cases. Of the remaining cases that were directly reviewed to determine a specific reason, 95.8% (23/24) involved hard copy scanned images, increasing the total number of repeat recommendation cases owing to the hard copy scanned nature of the images to 115 of 182. Motion artifact was cited in 20.9% (38/182) of cases. Low magnetic field strength was cited in 11.5% (21/182) of cases (1 T [n = 2], 0.7 T [n = 2], 0.6 T [n = 1], 0.3 T [n = 7], 0.2 T [n = 1], not otherwise specified [n = 8]). The category of missing or inadequate standard MRI sequences and CT reformats or reconstruction algorithms was the second most common category (20.3%; 37/182), followed by the “clarification” category (17.0%, 31/182). The category of lack of IV contrast administration where otherwise deemed appropriate was much less common as a source of repeat recommendations (6.6%; 12/182). In this category, two cases were for a clinically suspected brain mass,
8 6 4 2 0
s
s
ce
w ed us do ht d in ig eo e t e n q /W gh lla g -W ces m Se lS ei es ei ce T2 en al ith W s na W c r i n i R i o 2 n n M M equ Sp lg Sp io g l T ue e A al ia eq at al S us gin IR e x f xi u S f T a i A l A S on D Im eq pin R a B R R R M pin ad S T M M n M o C I S o o o N o N N N N n ue
eq
ed
ht
ce
n ue
Fig. 3—Graph of missing or inadequate MR/CT category data summary shows various cited study limitations. 4.5 4.0 3.5 No. of Studies
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Torok et al.
3.0 2.5 2.0 1.5 1.0 0.5 0
After Resection of GBM
Clinically Suspected Brain Mass
Neck Mass
Miscellaneous
Fig. 4—Graph of lack of IV contrast material administration category data summary shows various reported study indications. GBM = glioblastoma multiforme.
AJR:201, November 2013
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Second Opinion Consultation in Neuroradiology Fig. 5—Hard copy images show scanned study with poor image quality and limited ability to manipulate. White boxes have been applied to obscure patient identifiers.
four cases involved evaluation of residual tumor after resection of glioblastoma multiforme, two cases involved neck masses for further evaluation, one case was for suspected demyelinating disease, one case was for evaluation of a known vertebral lesion, one case was for evaluation of a known vestibular schwannoma, and one case was an MRI follow-up of a suspected meningioma on CT. No contraindications to IV contrast material were identified in the available clinical records. The “other” category accounted for the smallest number of miscellaneous examination limitations (2.7%). The mean time of follow-up was 39 days (range, 0–140 days; one case excluded with unavailable further records but clear indication of intention for repeat examination in the last available clinic note). The range of years in practice of faculty neuroradiologists was 1–26 years with a range of requesting repeats of 0.2–5.3%. However, there was no significant correlation between years in practice and repeat rate (Spearman correlation coefficient rs = −0.3; p = 0.2). If each examination interpreted as a second opinion study (11,753) were instead completely repeated and interpreted, the total global cost according to Medicare Part B charge class 01 (Baltimore County) would be approximately
$4,743,949.14. In an ideal situation, if no studies were repeated and purely interpreted as a second opinion, the professional fee total using Medicare rates for comparison would be approximately $837,743.36. This represents a saving of $3,906,205.78. As shown in this study, the repeat recommendation rate was very low (n = 182, 1.5%), and if every repeat were actually performed, this would amount to an additional cost of $65,927.71. The additional estimated cost given the actual repeat rate (n = 38, 0.3%) was $14,019.34. Given the total of 6604 patients for the year, the estimated per-patient cost of repeat imaging for the service was $2.12. Discussion Increased use of radiology services is multifactorial. A radiologist survey conducted by Lysdahl and Hofmann [5] summarized several of these reasons, including new technology, individual demand, clinician intolerance of uncertainty, expanded clinical indications, and availability. Other cited factors include self-referral abuse and unnecessary duplicative imaging [6]. The number of CT studies performed has quadrupled since 1992, leading to increased public and clinician concern regarding cumulative radiation dose and potential downstream cancer risks [7].
From the perspective of the radiologist, increased use is usually most strongly seen in the emergency department where emergency physicians must expedite care to manage the constant influx of patients. Radiography, sonography, CT, and increasingly MRI are all a mouse click away to minimize diagnostic uncertainty and medicolegal risk. Trauma transfers, in particular, have a high predilection for repeat imaging. Haley et al. [8] studied 410 patient transfers and found that 53% of patients received repeat imaging at an average cost of $2985 per patient, with the injury severity score as an independent predictor of study duplication. Gupta et al. [3] found that 58% of transferred trauma patients received duplicative CT studies, with head CT primarily repeated for their trauma indication and body CT primarily because of inadequate availability or quality from the referring institution. Excessive follow-up recommendations may depend on the confidence and experience level of the radiologist as well as the fear of medicolegal risk that is ubiquitous in modern medicine. A study by Sistrom et al. [9] showed an increase in recommendations for additional imaging from 6% to 12% over the period from 1995 to 2008 with increasing volumes and availability of various modalities. The odds of any single radiologist making such a recommendation decreased with experience by approximately 15% per decade. A negative correlation between years of experience and repeat recommendation rate was not shown to a level of significance among our neuroradiology staff. Duplicative imaging is an obvious target for cutting reimbursement. Our institutional data indicate that the NSOCS is not a significant source of duplicative imaging, with a repeat recommendation rate of 1.5% and an actual duplicative imaging rate of 0.3%. However, several studies have shown the added value of subspecialty radiologist second opinions in neuroradiology. Zan et al. [4] cited a 7.7% major discrepancy rate between internal and outside interpretations for neuroradiology studies, whereas Briggs et al. [10] found a 13% major discrepancy rate as well as a 21% minor discrepancy rate between subspecialty and general radiologists. These data suggest that a second opinion consultation service has a positive and economic impact on patient care without significant self-referred repeat studies. Repeat recommendations in our study were most commonly due to poor image quality from outside institutions, with the majority due to the submission of hard copy film rather than digital
AJR:201, November 2013 1099
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Torok et al. media, low magnetic field strength, and motion artifact. Inadequate or missing MRI sequences or CT reformatted images or reconstruction algorithms also frequently led to repeat recommendations. Many of these limitations are currently unavoidable given that images are distributed to patients on request using the limited media (film, CD, paper) described previously. Eventually, such obstacles may be avoided through the use of cloud-based imaging and adherence to universal standardized image acquisition and image transfer protocols. Streamlining this process for all parties involved is an important part of how we can add necessary value to our services for patients and referring physicians, with substantial potential savings. VPNs represent an important opportunity to provide the rapid transfer of imaging data between institutions without degrading image quality. However, the current implementation of VPNs is limited. Robinson and McNeeley [2] reported that only 8% of surveyed institutions receive 60% or more of their outside studies via network transfer because of the barriers of setting up these networks. On the contrary, this survey actually reported a doubling of non-DICOM format CDs as a major trend. Greater than 90% of studies at our institution are submitted on CD. A number of factors limit the accuracy of our results. One limitation is the keyword searches of the database. It is certainly possible to recommend a repeat study with phraseology that does not incorporate the word “repeat” or “follow-up.” We concluded that the likelihood of detecting a significant number of new applicable cases with broader expansion of the search limits was low and that “repeat” was the most likely terminology to convey inadequate quality of the original study. This was supported by the very low yield of the more expansive search with “follow-up.” Our random review of 100 sample radiology reports did not yield any
1100
additional cases beyond the keyword search, validating our methodology. Another limitation was a lack of available clinical information from outside institutions. Without complete medical records for cases that did not have a documented repeat examination, it was impossible to reliably determine whether a repeat study was performed at an outside institution that may not appear in our electronic medical record. The type of submitted media or other specific reasons for repeat recommendations were not always clearly described by the radiologist, leading to some ambiguity in data collection. Several cases simply stated that the study was limited and should be repeated; in these instances the patient’s images were reviewed to establish a more specific reason. Another challenge with establishing an accurate per-examination repeat recommendation rate was that the radiologist had the advantage in many cases of reviewing multiple bundled examinations over time, potentially allowing internal clarification of questionable findings without recommending a repeat study. An additional limitation of this study is that the results may not apply to practices of a different practice setting (tertiary care center vs rural), second opinion services rendered for subspecialties other than neuroradiology, or institutions that receive a different proportion of hard copy versus electronic media. As indicated by a prior analysis of our institutional pathology distribution for outside studies, our case mix is heavily weighted toward neoplasms in the setting of a major neurosurgical center [4]. Furthermore, our RIS has a common code for all outside studies, limiting analysis between different settings. Only partial data were available, indicating the general breakdown of how many outside studies were submitted in the inpatient, outpatient, and emergency department settings.
Reviewing outside studies generates a very low rate of requests for and performance of repeat studies but is beneficial for patient care [4]. This is not a major additional health care expense to the system and with improved technologies may allow substantial savings of health care dollars. References 1. Muennig PA, Glied SA. What changes in survival rates tell us about us health care. Health Aff (Millwood) 2010; 29:2105–2113 2. Robinson JD, McNeeley MF. Transfer patient imaging: a survey of members of the American Society of Emergency Radiology. Emerg Radiol 2012; 19:447–454 3. Gupta R, Greer SE, Martin ED. Inefficiencies in a rural trauma system: the burden of repeat imaging in interfacility transfers. J Trauma 2010; 69:253–235 4. Zan E, Yousem DM, Carone M, Lewin JS. Second-opinion consultations in neuroradiology. Radiology 2010; 255:135–141 5. Lysdahl KB, Hofmann BM. What causes increasing and unnecessary use of radiological investigations? A survey of radiologists’ perceptions. BMC Health Serv Res 2009; 9:155 6. DiLeo R, Spinelli R. Strategies for teaching nonradiologist physicians the appropriate use of imaging studies: use of radiology seminars. J Med Pract Manage 2006; 21:362–366 7. Lauer MS. Elements of danger: the case of medical imaging. New Engl J Med 2009; 361:841–843 8. Haley T, Ghaemmaghami V, Loftus T, Gerkin RD, Sterrett R, Ferrara JJ. Trauma: the impact of repeat imaging. Am J Surg 2009; 198:858–862 9. Sistrom CL, Dreyer KJ, Dang PP, et al. Recommendations for additional imaging in radiology reports: multifactorial analysis of 5.9 million examinations. Radiology 2009; 253:453–461 10. Briggs GM, Flynn PA, Worthington M, Rennie I, McKinstry CS. The role of specialist neuroradiology second opinion reporting: is there added value? Clin Radiol 2008; 63:791–795
AJR:201, November 2013
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Neuroradiology/Head and Neck Imaging Original Research
Neuroradiology Second Opinion Consultation Service: Assessment of Duplicative Imaging Collin M. Torok1 Cindy Lee Paul Nagy David M. Yousem Jonathan S. Lewin Torok CM, Lee C, Nagy P, Yousem DM, Lewin JS
OBJECTIVE. The purpose of this study was to characterize the performance of the Neuroradiology Second Opinion Consultation Service (NSOCS) at our institution to establish the rate, causes, and implications of requests for repeat imaging. MATERIALS AND METHODS. We queried 11,753 complete reports of all NSOCS studies for calendar year 2010 for the words “repeat” and “follow-up.” We categorized study limitations described in these reports into poor image quality, missing or inadequate MR sequences or CT reformats, lack of IV contrast administration where otherwise deemed appropriate, an “other” category for miscellaneous items, and a “clarification” category for indeterminate findings or recommendations for more advanced protocols. The corresponding available electronic medical records were reviewed. An estimated financial analysis of the NSOCS was additionally performed. RESULTS. Repeat imaging studies were recommended in 1.5% of cases. In 0.3% of all cases, a subsequent repeat examination was documented in the electronic medical records. Study limitations were most commonly due to poor image quality (77.5%), followed by missing or inadequate MR sequences or CT reformats (20.3%). The additional estimated cost of repeat imaging was calculated at $14,019.34, with an overall per-patient cost of $2.12 for the service. CONCLUSION. Reviewing outside studies generates a very low rate of requests for and performance of repeat studies, and is not a major additional health care expense.
I
Keywords: duplicative imaging, neuroradiology, outside studies, repeat imaging, second opinion DOI:10.2214/AJR.12.9429 Received June 17, 2012; accepted after revision February 28, 2013. 1
All authors: The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins Hospital, 600 N Wolfe St, Baltimore, MD 21287. Address correspondence to C. M. Torok ([email protected]).
AJR 2013; 201:1096–1100 0361–803X/13/2015–1096 © American Roentgen Ray Society
1096
n an age of progressively increasing health care costs in the United States [1] and increased pressure on radiologists to maximize cost-effectiveness, the elimination of duplicative imaging is an opportunity for potential cost savings. Patients frequently arrive from outside institutions or imaging facilities with radiology studies on CD, on hard copy, or occasionally through a virtual private network (VPN) transfer [2]. Loading these studies into the PACS provides immediate availability of imaging data throughout the enterprise. In turn, these imports enable the most appropriate and expedient diagnostic and therapeutic decisions to be made. Repeating these outside studies should be avoided. Such duplicative imaging encompasses a broad spectrum, ranging from a nonemergent brain MRI study repeated because of degradation of the original outside data through the transfer process or from intrinsically poor quality studies to a transferred trauma patient receiving a second “pan scan” in the emergen-
cy department when time is of the essence and rapid imaging data transfer is unavailable [3]. The Neuroradiology Second Opinion Consultation Service (NSOCS) at our institution frequently faces the former challenge. We have established a mechanism to streamline formal review of outside studies by subspecialty radiologists on request to provide second opinions for patients and clinicians. Outside studies can be scanned into our PACS at any time for clinicians at centralized locations during office hours and by emergency department technologists off-hours, allowing a mutually convenient and appropriate review rather than a cursory “curbside” opinion. Our department policy is that all outside studies are scanned into the PACS and a final report is generated and signed by a neuroradiologist for every study. Despite the availability of this service, however, a variety of reasons may still mandate repeat imaging. This potential for duplicative imaging of second opinion studies raises concern for excessive self-referral.
AJR:201, November 2013
Through analyzing the performance of the NSOCS, we sought to establish the rate, causes, and implications of requests for repeat imaging. We hypothesized that the rate of ordering repeat studies would be less than 10% and hoped that, through characterizing our performance and potentially elucidating problematic areas, we might identify opportunities to optimize the delivery of our service. Materials and Methods The institutional review board approved the retrospective review of patient data for this HIPAAcompliant study and waived the need for individual informed consent. We imported 11,753 complete reports of all NSOCS studies reviewed in calendar year 2010 into a Microsoft Excel spreadsheet from a radiology information system (RIS) search. The database was queried separately with the keyword search function for the words “repeat” and “follow-up.” In the latter case, coincidences with “repeat” were accounted for by additionally searching “follow-up AND repeat.” Spelling and punctuation variants of “follow-up” were accounted for by the search argument %follow%up%. This approach was further validated by random review of 100 complete radiology reports (via the RAND function in Excel). This secondary review yielded no additional repeat recommendations beyond those discovered using the search parameters noted. We reviewed the reports and categorized study limitations into the following five groups: poor image quality (hard copy scanned images, low MRI field strength, and motion artifact); missing or inadequate standard MRI sequences and CT reformats or reconstruction algorithms (missing or inadequate MRI sequences [T2-weighted, FLAIR, fat suppression, diffusion-weighted MRI, STIR], lack of sagittal CT reformats, no CT bone algorithm or window, poor bone subtraction on CT angiography, no source images on MR angiography, and no thin sections on CT angiography); lack of IV contrast administration where otherwise deemed appropriate, which included cases in which no contrast material was administered with indications such as clinically suspected brain mass or neck mass without an identifiable contraindication in available patient records; an “other” category for miscellaneous items, including an incomplete FOV, inability to link multiplanar images or use a measuring tool, inability to window, and poor reconstruction algorithm; and a “clarification” category that was defined to include studies for which repeat recommendations were made to clarify indeterminate findings with or without more advanced protocols or modalities. The corresponding available electronic medical records were reviewed to assess whether the repeat recommendation was performed at our in-
stitution. Additionally, a database of all internal neuroradiology studies for the same period was searched as an internal comparison. Faculty years of experience in practice were compared with individual repeat recommendation rates to determine if experience correlated with repeat request rate using a Spearman correlation coefficient test (n = 13 faculty members). A financial analysis was performed on the basis of the number of each individual type of examination performed during the study time period. This represents a hypothetical analysis of projected technical and professional fee charges rather than exact departmental financial records and actual reimbursement. The corresponding Current Procedural Terminology codes were input into a web-based fee calculator (2012 Medicare Part B Fee Schedule Calculator, Novitas Solutions), and global, technical, and professional fee output was obtained for Maryland Charge class 01 (Baltimore County), assuming a 100% Medicare reimbursement rate. Our institution is a major tertiary care medical center that in calendar year 2012 had 913 beds, 46,775 admissions, 84,982 emergency department visits, and 506,915 outpatient visits. Our neuroradiology division performed an average of 80,851 examinations per year for calendar years 2010 and 2011. Our department as a whole performed an average of 467,042 examinations per year for the same time period. Our practice setting is an academic neuroradiology division that handles referrals from across the country and the world, encompassing the full spectrum of inpatient, outpatient, and emergency department service. The studies arise predominantly from neurosurgery,
neurology, oncology, emergency department, and otorhinolaryngology referrers. On the basis of data categorization available for 2012, the vast majority of studies submitted to our neuroradiology division were on CD or DVD (97.7%) rather than film (2.3%). According to available partial data from 2010, 10,124 of 11,753 (86.1%) outside studies were submitted from the following sites: 87.0% (8806/10,124) outpatient, 10.3% (1038/10,124) inpatient, and 2.7% (280/10,124) emergency department. A separate analysis conducted at our institution in 2008 indicated the following representative pathology case distribution for outside studies: neoplastic (43.6%); infectious, inflammatory, or demyelinating (15.0%); and degenerative (8.3%)—with a wide variety of lesser contributions, including normal studies [4].
Results The keyword search for “repeat” yielded 287 reports, each of which was individually reviewed, and 159 studies were identified in which repeat imaging was recommended. The keyword search for “follow-up” yielded 1981 results, each of which was individually reviewed, yielding 23 additional studies that incorporated strong recommendations for repeating the examinations because of study limitations. The remaining “follow-up” keyword search reports either recommended routine follow-up, vaguely suggested followup if clinically indicated, or used “followup” in a manner unrelated to the impression (e.g., “When compared with the follow-up study, no change was noted”).
160 140 120
No. of Studies
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Second Opinion Consultation in Neuroradiology
100 80 60 40 20 0 Poor Image Missing or Quality Inadequate MRI/CT
Clarification
Lack of IV Contrast Administration
Other
Fig. 1—Graph shows data summary of overall categories. Note that multiple different types of limitations were reported for individual studies.
AJR:201, November 2013 1097
1098
Fig. 2—Graph of poor image quality category data summary shows most common factor was hard copy scanned images.
140 120
No. of Studies
100 80 60 40 20 0
Hard Copy Scanned
Motion Artifact
Low Magnetic Field Strength
14 12 10 No. of Studies
Repeat imaging studies were recommended by neuroradiologists in 1.5% of cases (182/11,753); 32 were CT and 150 were MRI (95% CI, 0.013–0.018) based on our search parameters. In 20.9% (38) of these cases (six CT and 32 MR; 0.154–0.277) and 0.3% of all cases (38/11,753; 0.002–0.004), a documented repeat examination was identified in the available clinical records. The search of internal neuroradiology studies for the same period as a control yielded 128 results. Each of these reports was individually reviewed, and 33 studies were identified in which repeat imaging was recommended, yielding a 0.1% (33/25,090; 0.001–0.002) radiologist repeat recommendation rate and a corresponding 0.04% (9/25,090; 0.0002–0.0008) actual repeat imaging rate based on follow-up clinic notes. Figures 1–4 summarize the data collected for each of the categories described in the Materials and Methods section. Poor image quality was the most common factor cited in recommendations for repeat examinations (77.5%; 141/182). The hard copy scanned nature of the images (i.e., a radiograph that was digitized) was directly cited in 51.1% (93/182) of cases (Fig. 5). A common speech recognition dictation macro stating either that the images are not of the quality usually generated at our institution or that the images were not in DICOM format and repeat examination should be considered was cited in 28.6% (52/182) of cases. This macro was accompanied by at least one definable limitation in the report body from the various categories described in 53.8% (28/52) of such cases. Of the remaining cases that were directly reviewed to determine a specific reason, 95.8% (23/24) involved hard copy scanned images, increasing the total number of repeat recommendation cases owing to the hard copy scanned nature of the images to 115 of 182. Motion artifact was cited in 20.9% (38/182) of cases. Low magnetic field strength was cited in 11.5% (21/182) of cases (1 T [n = 2], 0.7 T [n = 2], 0.6 T [n = 1], 0.3 T [n = 7], 0.2 T [n = 1], not otherwise specified [n = 8]). The category of missing or inadequate standard MRI sequences and CT reformats or reconstruction algorithms was the second most common category (20.3%; 37/182), followed by the “clarification” category (17.0%, 31/182). The category of lack of IV contrast administration where otherwise deemed appropriate was much less common as a source of repeat recommendations (6.6%; 12/182). In this category, two cases were for a clinically suspected brain mass,
8 6 4 2 0
s
s
ce
w ed us do ht d in ig eo e t e n q /W gh lla g -W ces m Se lS ei es ei ce T2 en al ith W s na W c r i n i R i o 2 n n M M equ Sp lg Sp io g l T ue e A al ia eq at al S us gin IR e x f xi u S f T a i A l A S on D Im eq pin R a B R R R M pin ad S T M M n M o C I S o o o N o N N N N n ue
eq
ed
ht
ce
n ue
Fig. 3—Graph of missing or inadequate MR/CT category data summary shows various cited study limitations. 4.5 4.0 3.5 No. of Studies
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Torok et al.
3.0 2.5 2.0 1.5 1.0 0.5 0
After Resection of GBM
Clinically Suspected Brain Mass
Neck Mass
Miscellaneous
Fig. 4—Graph of lack of IV contrast material administration category data summary shows various reported study indications. GBM = glioblastoma multiforme.
AJR:201, November 2013
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Second Opinion Consultation in Neuroradiology Fig. 5—Hard copy images show scanned study with poor image quality and limited ability to manipulate. White boxes have been applied to obscure patient identifiers.
four cases involved evaluation of residual tumor after resection of glioblastoma multiforme, two cases involved neck masses for further evaluation, one case was for suspected demyelinating disease, one case was for evaluation of a known vertebral lesion, one case was for evaluation of a known vestibular schwannoma, and one case was an MRI follow-up of a suspected meningioma on CT. No contraindications to IV contrast material were identified in the available clinical records. The “other” category accounted for the smallest number of miscellaneous examination limitations (2.7%). The mean time of follow-up was 39 days (range, 0–140 days; one case excluded with unavailable further records but clear indication of intention for repeat examination in the last available clinic note). The range of years in practice of faculty neuroradiologists was 1–26 years with a range of requesting repeats of 0.2–5.3%. However, there was no significant correlation between years in practice and repeat rate (Spearman correlation coefficient rs = −0.3; p = 0.2). If each examination interpreted as a second opinion study (11,753) were instead completely repeated and interpreted, the total global cost according to Medicare Part B charge class 01 (Baltimore County) would be approximately
$4,743,949.14. In an ideal situation, if no studies were repeated and purely interpreted as a second opinion, the professional fee total using Medicare rates for comparison would be approximately $837,743.36. This represents a saving of $3,906,205.78. As shown in this study, the repeat recommendation rate was very low (n = 182, 1.5%), and if every repeat were actually performed, this would amount to an additional cost of $65,927.71. The additional estimated cost given the actual repeat rate (n = 38, 0.3%) was $14,019.34. Given the total of 6604 patients for the year, the estimated per-patient cost of repeat imaging for the service was $2.12. Discussion Increased use of radiology services is multifactorial. A radiologist survey conducted by Lysdahl and Hofmann [5] summarized several of these reasons, including new technology, individual demand, clinician intolerance of uncertainty, expanded clinical indications, and availability. Other cited factors include self-referral abuse and unnecessary duplicative imaging [6]. The number of CT studies performed has quadrupled since 1992, leading to increased public and clinician concern regarding cumulative radiation dose and potential downstream cancer risks [7].
From the perspective of the radiologist, increased use is usually most strongly seen in the emergency department where emergency physicians must expedite care to manage the constant influx of patients. Radiography, sonography, CT, and increasingly MRI are all a mouse click away to minimize diagnostic uncertainty and medicolegal risk. Trauma transfers, in particular, have a high predilection for repeat imaging. Haley et al. [8] studied 410 patient transfers and found that 53% of patients received repeat imaging at an average cost of $2985 per patient, with the injury severity score as an independent predictor of study duplication. Gupta et al. [3] found that 58% of transferred trauma patients received duplicative CT studies, with head CT primarily repeated for their trauma indication and body CT primarily because of inadequate availability or quality from the referring institution. Excessive follow-up recommendations may depend on the confidence and experience level of the radiologist as well as the fear of medicolegal risk that is ubiquitous in modern medicine. A study by Sistrom et al. [9] showed an increase in recommendations for additional imaging from 6% to 12% over the period from 1995 to 2008 with increasing volumes and availability of various modalities. The odds of any single radiologist making such a recommendation decreased with experience by approximately 15% per decade. A negative correlation between years of experience and repeat recommendation rate was not shown to a level of significance among our neuroradiology staff. Duplicative imaging is an obvious target for cutting reimbursement. Our institutional data indicate that the NSOCS is not a significant source of duplicative imaging, with a repeat recommendation rate of 1.5% and an actual duplicative imaging rate of 0.3%. However, several studies have shown the added value of subspecialty radiologist second opinions in neuroradiology. Zan et al. [4] cited a 7.7% major discrepancy rate between internal and outside interpretations for neuroradiology studies, whereas Briggs et al. [10] found a 13% major discrepancy rate as well as a 21% minor discrepancy rate between subspecialty and general radiologists. These data suggest that a second opinion consultation service has a positive and economic impact on patient care without significant self-referred repeat studies. Repeat recommendations in our study were most commonly due to poor image quality from outside institutions, with the majority due to the submission of hard copy film rather than digital
AJR:201, November 2013 1099
Downloaded from www.ajronline.org by 73.46.105.161 on 11/04/15 from IP address 73.46.105.161. Copyright ARRS. For personal use only; all rights reserved
Torok et al. media, low magnetic field strength, and motion artifact. Inadequate or missing MRI sequences or CT reformatted images or reconstruction algorithms also frequently led to repeat recommendations. Many of these limitations are currently unavoidable given that images are distributed to patients on request using the limited media (film, CD, paper) described previously. Eventually, such obstacles may be avoided through the use of cloud-based imaging and adherence to universal standardized image acquisition and image transfer protocols. Streamlining this process for all parties involved is an important part of how we can add necessary value to our services for patients and referring physicians, with substantial potential savings. VPNs represent an important opportunity to provide the rapid transfer of imaging data between institutions without degrading image quality. However, the current implementation of VPNs is limited. Robinson and McNeeley [2] reported that only 8% of surveyed institutions receive 60% or more of their outside studies via network transfer because of the barriers of setting up these networks. On the contrary, this survey actually reported a doubling of non-DICOM format CDs as a major trend. Greater than 90% of studies at our institution are submitted on CD. A number of factors limit the accuracy of our results. One limitation is the keyword searches of the database. It is certainly possible to recommend a repeat study with phraseology that does not incorporate the word “repeat” or “follow-up.” We concluded that the likelihood of detecting a significant number of new applicable cases with broader expansion of the search limits was low and that “repeat” was the most likely terminology to convey inadequate quality of the original study. This was supported by the very low yield of the more expansive search with “follow-up.” Our random review of 100 sample radiology reports did not yield any
1100
additional cases beyond the keyword search, validating our methodology. Another limitation was a lack of available clinical information from outside institutions. Without complete medical records for cases that did not have a documented repeat examination, it was impossible to reliably determine whether a repeat study was performed at an outside institution that may not appear in our electronic medical record. The type of submitted media or other specific reasons for repeat recommendations were not always clearly described by the radiologist, leading to some ambiguity in data collection. Several cases simply stated that the study was limited and should be repeated; in these instances the patient’s images were reviewed to establish a more specific reason. Another challenge with establishing an accurate per-examination repeat recommendation rate was that the radiologist had the advantage in many cases of reviewing multiple bundled examinations over time, potentially allowing internal clarification of questionable findings without recommending a repeat study. An additional limitation of this study is that the results may not apply to practices of a different practice setting (tertiary care center vs rural), second opinion services rendered for subspecialties other than neuroradiology, or institutions that receive a different proportion of hard copy versus electronic media. As indicated by a prior analysis of our institutional pathology distribution for outside studies, our case mix is heavily weighted toward neoplasms in the setting of a major neurosurgical center [4]. Furthermore, our RIS has a common code for all outside studies, limiting analysis between different settings. Only partial data were available, indicating the general breakdown of how many outside studies were submitted in the inpatient, outpatient, and emergency department settings.
Reviewing outside studies generates a very low rate of requests for and performance of repeat studies but is beneficial for patient care [4]. This is not a major additional health care expense to the system and with improved technologies may allow substantial savings of health care dollars. References 1. Muennig PA, Glied SA. What changes in survival rates tell us about us health care. Health Aff (Millwood) 2010; 29:2105–2113 2. Robinson JD, McNeeley MF. Transfer patient imaging: a survey of members of the American Society of Emergency Radiology. Emerg Radiol 2012; 19:447–454 3. Gupta R, Greer SE, Martin ED. Inefficiencies in a rural trauma system: the burden of repeat imaging in interfacility transfers. J Trauma 2010; 69:253–235 4. Zan E, Yousem DM, Carone M, Lewin JS. Second-opinion consultations in neuroradiology. Radiology 2010; 255:135–141 5. Lysdahl KB, Hofmann BM. What causes increasing and unnecessary use of radiological investigations? A survey of radiologists’ perceptions. BMC Health Serv Res 2009; 9:155 6. DiLeo R, Spinelli R. Strategies for teaching nonradiologist physicians the appropriate use of imaging studies: use of radiology seminars. J Med Pract Manage 2006; 21:362–366 7. Lauer MS. Elements of danger: the case of medical imaging. New Engl J Med 2009; 361:841–843 8. Haley T, Ghaemmaghami V, Loftus T, Gerkin RD, Sterrett R, Ferrara JJ. Trauma: the impact of repeat imaging. Am J Surg 2009; 198:858–862 9. Sistrom CL, Dreyer KJ, Dang PP, et al. Recommendations for additional imaging in radiology reports: multifactorial analysis of 5.9 million examinations. Radiology 2009; 253:453–461 10. Briggs GM, Flynn PA, Worthington M, Rennie I, McKinstry CS. The role of specialist neuroradiology second opinion reporting: is there added value? Clin Radiol 2008; 63:791–795
AJR:201, November 2013
