Asymptomatic Abdominal Aortic Aneurysm: Standardizing Reporting Recommendations at a Large Multistate Radiology Practice

ObjectiveAlthough often asymptomatic at initial diagnosis, abdominal aortic aneurysms (AAAs) require careful surveillance to prevent rupture, with rupture-associated mortality rates as high as 90%. The purpose of this study was to explore if a performance improvement initiative implemented across a large radiology practice successfully increased inclusion of best practice recommendations (BPRs) within the radiology report across the practice.MethodsAfter BPRs for asymptomatic AAA surveillance were developed, a structured reporting macro for follow-up recommendation was integrated into dictation software. Following a training period, inclusion of recommendations within the radiology report was monitored across 345 facilities within the practice. Performance was reported on scorecards distributed monthly. To measure practice improvement, inclusion of appropriate recommendation in radiology reports postimplementation was compared with pre-implementation data.ResultsDuring the period before AAA BPRs implementation, from 64,090 consecutive cross-sectional radiology reports reviewed during a 6-month period, 855 incidental AAAs (1.3%) were identified, with 783 aneurysms measuring 2.6 to 5.4 cm requiring imaging surveillance; only 17 (2.1%) included follow-up recommendations within the radiology report. Postimplementation, 2,641 of 148,807 cross-sectional imaging studies were positive for 2.6- to 5.4-cm AAAs requiring further management; 1,533 (58.0%) of these radiology reports included follow-up imaging recommendations (97.0% of which followed our AAA BPRs).DiscussionQuality improvement initiatives to develop BPRs for AAA surveillance and include these recommendations within the radiology report can be successfully implemented across large practices and are imperative to ensure imaging surveillance and avert AAA rupture.     

Ultrasound and Dual-Energy X-Ray Absorptiometry Report Transcription Error Rates and Strategies for Reduction

PurposeRadiologists play an essential role in patient care by providing accurate and timely results. An error-free radiology report is an expectation of both patients and referring physicians. Software is currently available that can eliminate measurement and side types of errors while saving radiologists and sonographers time. The objectives of this study were to evaluate the potential reduction in report errors, estimate the potential time savings associated with implementation, and conduct a cost-benefit analysis of implementing two software programs.MethodsData on the number of measurement errors and side errors in ultrasound and dual-energy x-ray absorptiometry reports were collected, and the time required for data entry that the software would reduce was measured by report type. Generalized estimating equations regression was used to estimate error rates and data entry times and corresponding 95% confidence intervals by report type for radiologists and sonographers. Current wages and report volumes were then applied to the time savings to estimate the annual wage savings. Projected volume increases were applied to the annual estimates to generate a 5-year savings estimate.ResultsOverall, measurement errors occurred in 6% to 28% of ultrasound reports, depending on the report type. Side errors were rare. It was estimated that over 5 years, the software could save $693,777 in radiologist wages and $130,771 in sonographer wages, a total of $824,548 (range, $621,866-$1,039,714).ConclusionsThe use of data integration software would both significantly reduce errors in ultrasound and dual-energy x-ray absorptiometry reports and save a considerable amount of time and money.     

Natural Language Processing of Radiology Reports in Patients With Hepatocellular Carcinoma to Predict Radiology Resource Utilization

ObjectiveRadiology is a finite health care resource in high demand at most health centers. However, anticipating fluctuations in demand is a challenge because of the inherent uncertainty in disease prognosis. The aim of this study was to explore the potential of natural language processing (NLP) to predict downstream radiology resource utilization in patients undergoing surveillance for hepatocellular carcinoma (HCC).Materials and MethodsAll HCC surveillance CT examinations performed at our institution from January 1, 2010, to October 31, 2017 were selected from our departmental radiology information system. We used open source NLP and machine learning software to parse radiology report text into bag-of-words and term frequency–inverse document frequency (TF-IDF) representations. Three machine learning models—logistic regression, support vector machine (SVM), and random forest—were used to predict future utilization of radiology department resources. A test data set was used to calculate accuracy, sensitivity, and specificity in addition to the area under the curve (AUC).ResultsAs a group, the bag-of-word models were slightly inferior to the TF-IDF feature extraction approach. The TF-IDF + SVM model outperformed all other models with an accuracy of 92%, a sensitivity of 83%, and a specificity of 96%, with an AUC of 0.971.ConclusionsNLP-based models can accurately predict downstream radiology resource utilization from narrative HCC surveillance reports and has potential for translation to health care management where it may improve decision making, reduce costs, and broaden access to care.     

Monitoring the Use of Extra Images on Chest Radiography Examinations

BackgroundA large number of chest radiography studies in our department include extra images due to incompletely imaged anatomy (eg, extra frontal view to include a truncated costophrenic angle). Negative impacts include: increased radiologist review time due to disruption of search pattern and the need to review additional films in slightly different obliquities, additional radiation exposure, and increased technologist time expenditure.PurposeTo determine the chest radiograph repeat rate, collaborate with technologists on a process to decrease the frequency, and incorporate the process into our quality control program.MethodsData collection was performed by using coded dictation macros to indicate the type of extra view (frontal, lateral) and whether the extra image was necessary due to patient habitus. Twelve weeks after the macro was instituted, baseline data were collected by querying the macro codes with a search engine (MONTAGE Search and Analytics). Intervention consisted of in-person tutorials of basic radiographic positioning principles with x-ray technologists and posting of checklists in all diagnostic radiology exam rooms. Twelve weeks of postintervention data were collected.ResultsBaseline data included 5645 examinations, of which 335 (5.9%) included extra images. Postintervention, 5943 examinations were performed and 295 (5.0%) included extra images. A significant decrease in the frontal view repeat rate was noted, decreasing from 4.6%-3.3% (P = 0.001). The repeat rate of lateral images did not change significantly (3.1%-3.2%).Conclusions and ImplicationsData monitoring and interprofessional collaboration led to a significant decrease in unnecessary extra radiographs. Ongoing monitoring may lead to sustained improvement and further reductions.     

Impact of PACS deployment strategy on dictation turnaround time of chest radiographs

RATIONALE AND OBJECTIVES: The aim of the study is to measure the impact of a picture archive and communication system (PACS) on dictation turnaround time of chest radiographs in a multisite hospital and relate variations across sites to local factors and implementation strategy. MATERIALS AND METHODS: The multisite hospital is composed of three sites. Dictation turnaround time was calculated by using data obtained from the radiology information system for examinations performed during three 90-day periods (immediately before PACS implementation, immediately after PACS implementation, and 1 year after implementation). Productivity, expressed as number of examinations dictated per full-time-equivalent radiologist, also was calculated. For each 3-month period, average interval delay was calculated. Values for average interval delay obtained during the different pre- and post-PACS periods were compared by using analysis of variance. This was done for each hospital. RESULTS: In the immediate post-PACS period at site 1, dictation turnaround time decreased 5% (P < .05), whereas productivity decreased 16.5%. The implementation strategy was revised for the next two sites, and dictation turnaround time decreased 21% (P < .001) in both sites in the immediate post-PACS period. Productivity increased 2% and 3% in these sites. One year after implementation, decreases in turnaround ranged from 28% to 55% (P < .001) in the three sites. CONCLUSION: Our experience suggests that PACSs cannot be isolated from their contexts; therefore, implementation strategy matters in the realization of projected benefits. In addition, regardless of differences in film-based environments before PACS, all three sites benefited from conversion to filmless operation, with the greatest benefits seen in the site that was least efficient before implementation.     

The Policy of Co–First Authorship and Co–Senior Authorship in Radiology Journals

ObjectiveIncreasingly, medical journals are recognizing “equally credited authors” (ECA) in the primary and senior authorship of articles. The aim of this study was to assess the policies of co–first authorship, co–senior authorship, and designation of a corresponding author in the radiology literature.MethodsWe identified 29 radiology journals based on impact factor ranking. Journal offices were contacted by phone and e-mail to ascertain their practices on first and senior authorship ECA designations. We surveyed the March, June, and December 2018 issues of each journal (when available) to assess the utilization of the co-designations in articles.ResultsTwenty-five of 29 journals responded to our survey (response rate: 86.2%). Of 25 journals, 20 (80%) allowed co–first authorship. Among these, 4 of 25 journals (16%) allowed more than two co–first authors. Among the 25 responses, 14 journals (56%) allowed co–senior authorship. Among the 24 journals who responded to this specific question, 23 (96%) approved designation of a corresponding author, different from the first or senior author. The review of March, June, December 2018 editions found co–first authorship and co–senior authorship ECA rates of 8.6% (range 0.0%-22.7%) and 1.8% (range 0.0%-13.3%), respectively. A corresponding author other than first or senior author was noted in 13.3% (range 0.0%-34.7%).DiscussionThere has been widespread acceptance of the concept of ECA in the policies of the top cited imaging journals particularly for first authors (80%). However, the utilization of these designations is uncommon for first authorship (8.6%) and rare (1.8%) for senior authorship based on our 2018 sampling.     

Trends in on-call workload in an academic medical center radiology department 1998-20021

RATIONALE AND OBJECTIVES: The workload in radiology departments is increasing rapidly. This study was designed to determine whether and to what extent the workload is being generated outside of traditional working hours (defined as 0800-1700 Monday thru Friday, excluding holidays). MATERIALS AND METHODS: Exam statistics were derived from the radiology department's automated examination scheduling and reporting system for four successive fiscal years. The distribution of the number of studies completed throughout the 24-hour day and the 7-day week was charted. RESULTS: A large proportion of studies are being completed outside of traditional working hours. Moreover, as the overall workload of the department increased, the proportion of studies being completed during nontraditional working hours was increasing at an even faster pace, particularly in the cross-sectional imaging modalities. Computed tomography, magnetic resonance imaging, and ultrasound have increased by 59%, 51%, and 30%, respectively, over 4 years. The on-call proportions have increased from 34% to 40% and 13% to 18% for computed tomography and ultrasound, respectively, over 4 years and from 44% to 50% for magnetic resonance imaging over 3 years. CONCLUSION: These trends have implications for radiologist and radiology technologist staffing. The department has already modified the scheduling of technologist staffing to provide in-house extended-hours coverage in most modalities. As the number of studies conducted outside of traditional working hours continues to expand and the demand for contemporaneous readings increases, radiologist staffing may need to be adjusted as well. Traditional on-call coverage may be insufficient to competently handle the growing workload. This may have particular implications for radiology residency programs.     

Automatic voice recognition for radiologic reports. Experiment with a prototype in an operational environment

Large vocabulary speech recognition, its techniques and its software and hardware technology, are being developed, aimed at providing the office user with a tool that could significantly improve both quantity and quality of his work: the dictation machine, which allows memos and documents to be input using voice and a microphone instead of fingers and a keyboard. The IBM Rome Science Center, together with the IBM Research Division, has built a prototype recognizer that accepts sentences in natural language from a 20,000-word Italian vocabulary. The unit runs on a personal computer equipped with a special hardware capable of giving all the necessary computing power. The first laboratory experiments yielded very interesting results and pointed out such system characteristics to make its use possible in operational environments. To this purpose, the dictation of medical reports was considered as a suitable application. In cooperation with the 2nd Radiology Department of S. Maria della Misericordia Hospital (Udine, Italy), a system was experimented by radiology department doctors during their everyday work. The doctors were able to directly dictate their reports to the unit. The text appeared immediately on the screen, and eventual errors could be corrected either by voice or by using the keyboard. At the end of report dictation, the doctors could both print and archive the text. The report could also be forwarded to hospital information system, when the latter was available. Our results have been very encouraging: the system proved to be robust, simple to use, and accurate (over 95% average recognition rate). The experiment was precious for suggestions and comments, and its results are useful for system evolution towards improved system management and efficiency.     

Radiologist Peer Review by Group Consensus

PurposeThe objective of this study was to evaluate the feasibility of the consensus-oriented group review (COGR) method of radiologist peer review within a large subspecialty imaging department.MethodsThis study was institutional review board approved and HIPAA compliant. Radiologist interpretations of CT, MRI, and ultrasound examinations at a large academic radiology department were subject to peer review using the COGR method from October 2011 through September 2013. Discordance rates and sources of discordance were evaluated on the basis of modality and division, with group differences compared using a χ² test. Potential associations between peer review outcomes and the time after the initiation of peer review or the number of radiologists participating in peer review were tested by linear regression analysis and the t test, respectively.ResultsA total of 11,222 studies reported by 83 radiologists were peer reviewed using COGR during the two-year study period. The average radiologist participated in 112 peer review conferences and had 3.3% of his or her available CT, MRI and ultrasound studies peer reviewed. The rate of discordance was 2.7% (95% confidence interval [CI], 2.4%-3.0%), with significant differences in discordance rates on the basis of division and modality. Discordance rates were highest for MR (3.4%; 95% CI, 2.8%-4.1%), followed by ultrasound (2.7%; 95% CI, 2.0%-3.4%) and CT (2.4%; 95% CI, 2.0%-2.8%). Missed findings were the most common overall cause for discordance (43.8%; 95% CI, 38.2%-49.4%), followed by interpretive errors (23.5%; 95% CI, 18.8%-28.3%), dictation errors (19.0%; 95% CI, 14.6%-23.4%), and recommendation (10.8%; 95% CI, 7.3%-14.3%). Discordant cases, compared with concordant cases, were associated with a significantly greater number of radiologists participating in the peer review process (5.9 vs 4.7 participating radiologists, P < .001) and were significantly more likely to lead to an addendum (62.9% vs 2.7%, P < .0001).ConclusionsCOGR permits departments to collect highly contextualized peer review data to better elucidate sources of error in diagnostic imaging reports, while reviewing a sufficient case volume to comply with external standards for ongoing performance review.     

From Being a Radiologist to Watching a Radiologist: Impact of Filmless Operation on the Training of Radiology Residents

RATIONALE AND OBJECTIVES: The authors performed this study to investigate the impact of changing from a film-based image interpretation system to one using digital image workstations on the training of radiology residents in the interpretation of radiographs. MATERIALS AND METHODS: Data were collected during a period when a conventional system of image interpretation with hard-copy images and multiviewers was used and during a period when digital image workstations were used. During each period, it was noted whether the first interpretation of the radiographs was performed by a radiology resident, by an attending radiologist, or as a group effort including both an attending radiologist and a radiology resident(s). In addition, it was noted whether a radiology resident or an attending radiologist dictated the report. RESULTS: The proportion of images first interpreted by the radiology resident alone decreased from 38% (53 of 139) when using the conventional system to 17% (34 of 199) after the switch to interpreting images on the workstations (P = .001). During the film-based period, radiology residents dictated 45% of reports (141 of 312), but during the workstation period, radiology residents dictated only 4% of reports (24 of 667; P = .001). CONCLUSION: The authors observed a decrease in autonomous participation by radiology residents in image interpretation and dictation of reports and an increase in "group reading" after the switch from a film-based system to a workstation system.     

Using a Natural Language Processing and Machine Learning Algorithm Program to Analyze Inter-Radiologist Report Style Variation and Compare Variation Between Radiologists When Using Highly Structured Versus More Free Text Reporting

PurposeTo use a natural language processing and machine learning algorithm to evaluate inter-radiologist report variation and compare variation between radiologists using highly structured versus more free text reporting.Materials and Methods28,615 radiology reports were analyzed for 4 metrics: verbosity, observational terms only, unwarranted negative findings, and repeated language in different sections. Radiology reports for two imaging examinations were analyzed and compared – one which was more templated (ultrasound – appendicitis) and one which relied on more free text (chest radiograph – single view). For each metric, the mean and standard deviation for defined outlier results for all dictations (individual and group mean) was calculated. The mean number of outlier metrics per reader per study was calculated and compared between radiologists and between the two report types. Wilcoxon rank test and paired Wilcoxon signed rank test were applied. The radiologists were also ranked based on the number of outlier metrics identified per study.ResultsThere was great variability in radiologist dictation styles – outlier metrics per report varied greatly between radiologists with the maximum 10 times higher than the minimum score. Metric values were greater (P < 0.0001) on the standardized reports using free text than the more structured reports.ConclusionsThe algorithm successfully evaluated metrics showing variability in reporting profiles particularly when there is free text. This variability can be an obstacle to providing effective communication and reliability of care.     

Academic radiology and the emergency department: does it need changing?

RATIONAL AND OBJECTIVES: The increasing importance of imaging for both diagnosis and management in patient care has resulted in a demand for radiology services 7 days a week, 24 hours a day, especially in the emergency department (ED). We hypothesized the resident preliminary reports were better than generalist radiology interpretations, although inferior to subspecialty interpretations. MATERIALS AND METHODS: Total radiology volume through our Level I pediatric and adult academic trauma ED was obtained from the radiology information system. We conducted a literature search for error and discordant rates between radiologists of varying experience. For a 2-week prospective period, all preliminary reports generated by the residents and final interpretations were collected. Significant changes in the report were tabulated. RESULTS: The ED requested 72,886 imaging studies in 2004 (16% of the total radiology department volume). In a 2-week period, 12 of 1929 (0.6%) preliminary reports by residents were discordant to the final subspecialty dictation. In the 15 peer-reviewed publications documenting error rates in radiology, the error rate between American Board of Radiology (ABR)-certified radiologists is greater than that between residents and subspecialists in the literature and in our study. However, the perceived error rate by clinicians outside radiology is significantly higher. CONCLUSION: Sixteen percent of the volume of imaging studies comes through the ED. The residents handle off-hours cases with a radiology-detected error rate below the error rate between ABR-certified radiologists. To decrease the perceived clinician-identified error rate, we need to change how academic radiology handles ED cases.     

AHRA (American Healthcare Radiology Administrators) survey. Staff utilization: Part II.

The AHRA Statistical Resource Committee is pleased to present the second part of survey results on staff utilization. This report covers multiple areas in radiology, including diagnostic radiology, CT, MRI, ultrasound, nuclear medicine and radiation therapy.     

The relationship of patient waiting time to capacity and utilization in emergency room radiology.

The relationship between patient waiting time and capacity and utilization is quantified. By deciding upon the average acceptable and maximum allowable waiting times, the required capacity and resulting utilization rate are fixed. The data needed for analysis include waiting time, patient volume, time required for specific types of examinations, and technologist and equipment capacity. Cost reduction is achieved without adversely affecting waiting time if volume variability can be reduced. Steps to increase productivity should also be considered as a means of reducing cost per examination, given the cost structure in emergency room radiology and the personnel costs.     

Adapting the waterproof BMS-96 diode array for isodose determination of dynamic beams.

We report the application of the Schuster BMS-96 waterproof linear diode array for isodose determination of dynamic beams. The array recorded beam profiles correctly, while depth dose distributions of dynamic beams with large variations in dose rate were registered erroneously. The deviations could be eliminated by appropriate software modifications. Until the software is revised, true isodoses can be obtained by rescaling each individual profile to the depth dose curve as measured with a single ionization chamber. After the corrections presented in the paper, isodoses interpolated from these corrected data sets agreed with ionization chamber measurements within 1-2%.