Screening Mammography Performance Metrics of 2D Digital Mammography Versus Digital Breast Tomosynthesis in Women With a Personal History of Breast Cancer.

BACKGROUND. Patients with a history of breast cancer are at higher risk of subsequent breast cancers and need close clinical and imaging follow-up. Limited data are available on screening of these patients with digital breast tomosynthesis (DBT) versus full-field digital mammography (FFDM). OBJECTIVE. The purpose of this study was to evaluate the screening mammography performance of DBT compared with FFDM among patients with a history of breast cancer undergoing imaging at a large academic oncology center. METHODS. This retrospective study included consecutively registered patients with a personal history of breast cancer treated with mastectomy or lumpectomy who underwent screening FFDM from October 2014 through September 2016 (5706 examinations of 4091 patients) or screening DBT from February 2017 through December 2018 (4440 examinations of 3647 patients). An institutional mammographic database was queried to obtain imaging type, breast density, history of mastectomy or lumpectomy, and BI-RADS category. An institutional breast cancer registry identified cancer diagnoses. Screening performance metrics were compared between FFDM and DBT groups. RESULTS. Recall rate was significantly lower with DBT than with FFDM (7.9% vs 10.1%; p < .001). DBT and FFDM did not differ in PPV1 (7.7% vs 6.1%; p = .36) or cancer detection rate (CDR) (6.1/1000 vs 6.0/1000; p = .97). Sensitivity was 96.4% for DBT and 71.4% for FFDM (p = .008). Specificity was 92.3% for DBT and 90.0% for FFDM (p < .001). With stratification by breast density, patients with nondense breast tissue had a lower recall rate with DBT than with FFDM (5.9% vs 8.8%; p < .001) and a nonsignificant increase in PPV1 (12.0% vs 6.4%; p = .05). The metrics were not otherwise different between DBT and FFDM among patients with nondense and those with dense breast tissue. Recall rates were lower with DBT than with FFDM among both patients who underwent mastectomy (7.8% vs 9.1%; p = .09) and those who underwent lumpectomy (7.9% vs 11.0%; p = .002). PPV1 and CDR were not different between DBT and FFDM among patients who underwent mastectomy and those who underwent lumpectomy. CONCLUSION. For patients with a personal history of breast cancer who have nondense breasts, the use of DBT as opposed to FFDM reduces recall rate and improves sensitivity and specificity. CDR and PPV1 remain unchanged. CLINICAL IMPACT. For women with a personal history of breast cancer and nondense breasts, DBT offers the potential to maintain the benefits of early cancer detection while reducing the potential harms of false-positive findings.

Clinical validation demonstrates concordance of qSOFA and POC lactate Bayesian model: Results from the ACDC Phase-2 program.

Sepsis is a common and lethal medical problem. The objective of this study was to validate a Bayesian Model that integrates qSOFA and prehospital Lactate, with a comparison analysis from a real clinical data of patients with sepsis. METHODS: We conducted a two tired validation study with one arm focusing on Bayesian modeling and a second retrospective observational arm addressing real data validation. For Bayesian modeling, sensitivity and specificity of prehospital lactate were attained from pooled meta-analysis data. Later, for clinical validation, we used data from 2016 to 2017 of ED patients diagnosed with sepsis. Pretest probabilities from qSOFA score where combined with prehospital lactate and inserted into a Bayesian model to calculate posttest probabilities. Absolute and relative diagnostic gains were calculated. Statistical significance was assessed via t-test, chi square and odds ratio. P value was set to be 0.05. RESULTS: For the Bayesian arm; meta-analysis data for prehospital lactate resulted in a positive likelihood ratio (LR+) of 1.69 and negative likelihood ratio (LR-) of 0.44. Integration of lactate and qSOFA demonstrated significant post-test improvements. On the Clinical Validation arm, 1470 patients were included with 176 patients meeting analysis criteria. When comparing qSOFA + Abnormal Lactate vs qSOFA and normal Lactate, the ICU vs Non-ICU cohorts were statistically different (p < 0.01) Odds Ratio: 2.35 (95% CI [1.22-4.6]). CONCLUSION: Bayesian mathematical model demonstrated that a qSOFA-based clinical decision can be complemented by the use of point of-care lactate. These results were confirmed by our clinical validation arm.

Variation in Follow-up Imaging Recommendations in Radiology Reports: Patient, Modality, and Radiologist Predictors.

Background Variation between radiologists when making recommendations for additional imaging and associated factors are, to the knowledge of the authors, unknown. Clear identification of factors that account for variation in follow-up recommendations might prevent unnecessary tests for incidental or ambiguous image findings. Purpose To determine incidence and identify factors associated with follow-up recommendations in radiology reports from multiple modalities, patient care settings, and imaging divisions. Materials and Methods This retrospective study analyzed 318 366 reports obtained from diagnostic imaging examinations performed at a large urban quaternary care hospital from January 1 to December 31, 2016, excluding breast and US reports. A subset of 1000 reports were randomly selected and manually annotated to train and validate a machine learning algorithm to predict whether a report included a follow-up imaging recommendation (training-and-validation set consisted of 850 reports and test set of 150 reports). The trained algorithm was used to classify 318 366 reports. Multivariable logistic regression was used to determine the likelihood of follow-up recommendation. Additional analysis by imaging subspecialty division was performed, and intradivision and interradiologist variability was quantified. Results The machine learning algorithm classified 38 745 of 318 366 (12.2%) reports as containing follow-up recommendations. Average patient age was 59 years ± 17 (standard deviation); 45.2% (143 767 of 318 366) of reports were from male patients. Among 65 radiologists, 57% (37 of 65) were men. At multivariable analysis, older patients had higher rates of follow-up recommendations (odds ratio [OR], 1.01 [95% confidence interval {CI}: 1.01, 1.01] for each additional year), male patients had lower rates of follow-up recommendations (OR, 0.9; 95% CI: 0.9, 1.0), and follow-up recommendations were most common among CT studies (OR, 4.2 [95% CI: 4.0, 4.4] compared with radiography). Radiologist sex (P = .54), presence of a trainee (P = .45), and years in practice (P = .49) were not significant predictors overall. A division-level analysis showed 2.8-fold to 6.7-fold interradiologist variation. Conclusion Substantial interradiologist variation exists in the probability of recommending a follow-up examination in a radiology report, after adjusting for patient, examination, and radiologist factors. © RSNA, 2019 See also the editorial by Russell in this issue.

A Bayesian decision support sequential model for severity of illness predictors and intensive care admissions in pneumonia.

BACKGROUND: Community-acquired pneumonia (CAP) is one of the leading causes of morbidity and mortality in the USA. Our objective was to assess the predictive value on critical illness and disposition of a sequential Bayesian Model that integrates Lactate and procalcitonin (PCT) for pneumonia. METHODS: Sensitivity and specificity of lactate and PCT attained from pooled meta-analysis data. Likelihood ratios calculated and inserted in Bayesian/ Fagan nomogram to calculate posttest probabilities. Bayesian Diagnostic Gains (BDG) were analyzed comparing pre and post-test probability. To assess the value of integrating both PCT and Lactate in Severity of Illness Prediction we built a model that combined CURB65 with PCT as the Pre-Test markers and later integrated the Lactate Likelihood Ratio Values to generate a combined CURB 65 + Procalcitonin + Lactate Sequential value. RESULTS: The BDG model integrated a CUBR65 Scores combined with Procalcitonin (LR+ and LR-) for Pre-Test Probability Intermediate and High with Lactate Positive Likelihood Ratios. This generated for the PCT LR+ Post-test Probability (POSITIVE TEST) Posterior probability: 93% (95% CI [91,96%]) and Post Test Probability (NEGATIVE TEST) of: 17% (95% CI [15-20%]) for the Intermediate subgroup and 97% for the high risk sub-group POSITIVE TEST: Post-Test probability:97% (95% CI [95,98%]) NEGATIVE TEST: Post-test probability: 33% (95% CI [31,36%]) . ANOVA analysis for CURB 65 (alone) vs CURB 65 and PCT (LR+) vs CURB 65 and PCT (LR+) and Lactate showed a statistically significant difference (P value = 0.013). CONCLUSIONS: The sequential combination of CURB 65 plus PCT with Lactate yielded statistically significant results, demonstrating a greater predictive value for severity of illness thus ICU level care.

Acute Care Diagnostic Collaboration: Bayesian modeling comparative diagnostic assessment of lactate, procalcitonin and CRP in risk stratified population by Mortality in ED (MEDS) score.

OBJECTIVE: To assess and compare the diagnostic value of lactate, procalcitonin (PCT) and C-reactive protein (CRP) in low, moderate, and high-risk stratified population applying Mortality in Emergency Department (MEDS) risk score using Bayesian statistical modeling. METHODS: MEDS criteria was used to risk stratify into low, moderate and high risk. Each population was attributed a percentage risk, and used as pre-test probability in the Bayesian nomogram. Sensitivity and specificity lactate, PCT and CRP were attained from pooled meta-analysis data. Absolute and relative diagnostic gains were calculated. RESULTS: Pooled diagnostic quality data obtained from a meta-analysis reflected sensitivity for PCT of 77% and specificity of 79%, for lactate sensitivity 49.1% and specificity 74.3% and CRP yielded a sensitivity of 75% and specificity 67%. likelihood ratios (LR) calculations for PCT were LR+ 3.67 and LR- 0.29; for lactate LR+ 1.88 and LR- 0.69; CRP LR+ 2.27 and LR- 0.37. When computed in Bayesian nomogram post-test probabilities for LR+ were as follows: for PCT low risk absolute gain of 11.7% and relative gain of 220%; moderate absolute gain 25.7% relative gain 148.5%; for high risk absolute gain 25.1% and relative gain 42.6%. Lactate LR+ results for low risk absolute gain of 4.7% and relative gain of 88.6%; moderate absolute gain 10.7% and relative gain 61.8%; high risk relative gain 14.1% and relative gain 23.9%. CRP results for low population and LR+ absolute gain 5.7% and relative gain 107.5%; moderate risk 14.7% absolute gain and 84.9% relative gain; high risk 77% post-test 18.1% absolute gain and 30.7% relative gain. CONCLUSION: Bayesian statistical model demonstrated the superior diagnostic quality of PCT. For ruling out severe disease, lactate yielded a higher benefit with increased relative gain with negative LR.

Bayesian comparative assessment of diagnostic accuracy of low-dose CT scan and ultrasonography in the diagnosis of urolithiasis after the application of the STONE score.

OBJECTIVE: The objective of our study was to assess the diagnostic quality of low-dose computed tomography (CT) when compared to ultrasound (US) in diagnosis of urolithiasis using STONE score as a predictor of pre-test probability and the Bayesian statistical model to calculate post-test probabilities (POST) for both diagnostic tests. METHODS: STONE score was used to form risk groups to obtain pre-test probabilities. Likelihood ratios (LR) were calculated from external data for low-dose CT and US. POST were obtained using pre-test probabilities and likelihood ratios with Bayesian nomogram. Absolute (ADG) and relative (RDG) gains in diagnostic value were calculated. RESULTS: Calculated +LR for US was 12 and -LR was 0.32; for CT, +LR was 19 and -LR 0.04. +LR and low STONE for US yielded POST 57% and RDG 470%; intermediate STONE POST 92% and RDG 84%; and high STONE POST 99% and RDG 10%. -LR and low STONE for US POST 3% and RDG -70%; intermediate POST 24% and RDG -52%; and high STONE POST 74% and RDG -17.7%. +LR and low STONE for CT POST 68% and RDG 580%; moderate STONE POST 95% and RDG 90%; and high STONE POST 99% and RDG 10%. -LR and low STONE for CT POST 0% and RDG -100%; intermediate POST 4% and RDG -92%; and high STONE POST 26% and RDG -71.1%. ANOVA calculations comparing CT vs US for +LR showed no statistical significance (P value = 0.9893; LR- P value = 0.5488). CONCLUSION: Bayesian statistical analysis demonstrated slight superiority of CT scan over US on STONE score low- and moderate-risk stratified subtypes, whereas no significant advantage was seen when evaluating high-probability patients.

Incremental diagnostic quality gain of CTA over V/Q scan in the assessment of pulmonary embolism by means of a Wells score Bayesian model: results from the ACDC collaboration.

OBJECTIVE: Our objective was to evaluate the diagnostic value of computed tomography angiography (CTA) and ventilation perfusion (V/Q) scan in the assessment of pulmonary embolism (PE) by means of a Bayesian statistical model. METHODS: Wells criteria defined pretest probability. Sensitivity and specificity of CTA and V/Q scan for PE were derived from pooled meta-analysis data. Likelihood ratios calculated for CTA and V/Q were inserted in the nomogram. Absolute (ADG) and relative diagnostic gains (RDG) were analyzed comparing post- and pretest probability. Comparative gain difference was calculated for CTA ADG over V/Q scan integrating ANOVA p value set at 0.05. RESULTS: The sensitivity for CT was 86.0% (95% CI: 80.2%, 92.1%) and specificity of 93.7% (95% CI: 91.1%, 96.3%). The V/Q scan yielded a sensitivity of 96% (95% CI: 95%, 97%) and a specificity of 97% (95% CI: 96%, 98%). Bayes nomogram results for CTA were low risk and yielded a posttest probability of 71.1%, an ADG of 56.1%, and an RDG of 374%, moderate-risk posttest probability was 85.1%, an ADG of 56.1%, and an RDG of 193.4%, and high-risk posttest probability was 95.2%, an ADG of 36.2%, and an RDG of 61.35%. The comparative gain difference for low-risk population was 46.1%; in moderate-risk 41.6%; and in high-risk a 22.1% superiority. ANOVA analysis for LR+ and LR- showed no significant difference (p = 0.8745, p = 0.9841 respectively). CONCLUSIONS: This Bayesian model demonstrated a superiority of CTA when compared to V/Q scan for the diagnosis of pulmonary embolism. Low-risk patients are recognized to have a superior overall comparative gain favoring CTA.

Acute Care Diagnostics Collaboration: Assessment of a Bayesian clinical decision model integrating the Prehospital Sepsis Score and point-of-care lactate.

UNLABELLED: Previous research demonstrated that shock index and respiratory rate are highly predictive of intensive care unit admissions. OBJECTIVE: The objective of the study is to evaluate the integration of the prehospital sepsis project score (PSP-S) and point-of-care lactate in assisting prediction of severity of illness using Bayesian statistical modeling. METHODS: The PSP-S incorporates fever (38°C [100.4°F]) allotted with 1 point, shock index greater than or equal to 0.7 given 2 points, and a respiratory rate greater than or equal to 22 breaths per minute given 1 point for a total maximum score of 4 points. The patient population was stratified based on the PSP-S: 1 point is low risk, 2 points is moderate risk, and 3 to 4 points is high risk. Percentage risk was obtained based on intensive care unit admissions and used as pretest probability. Prehospital lactate pooled data were obtained and used to calculate likelihood ratio (LR). Percentage risk used as pretest probability and LRs for prehospital lactate were charted into the Bayesian nomogram to obtain posttest probabilities. Absolute diagnostic gain (ADG) and relative diagnostic gains (RDG) were then calculated. RESULTS: Pooled data for prehospital point of care lactate demonstrated a positive LR of 1.6 and negative LR of 0.44. Posttest probability for low risk was 16% with an ADG of 6% and RDG of 160%. Moderate risk population yielded a posttest probability of 47%, ADG of 12.5%, and RDG of 136.2%. High-risk population resulted in a posttest probability of 72%, ADG of 12%, and RDG of 120%. CONCLUSION: We found that PSP-S can be clinically complemented with the use of point-of-care lactate.

Bayesian comparative model of CT scan and ultrasonography in the assessment of acute appendicitis: results from the Acute Care Diagnostic Collaboration project.

The objective of this study was to develop a comparative diagnostic model for computed tomography (CT) and ultrasound (US) in the assessment of acute appendicitis using Alvarado risk score as a predictor of pretest probability and Bayesian statistical model as a tool to calculate posttest probabilities for both diagnostic test. Stratification was made by applying the Alvarado score for the prediction of acute appendicitis. Likelihood ratios were calculated using sensitivity and specificity of both CT and US from a Meta-analysis. Posttest probabilities were obtained after inserting Alvarado score and likelihood ratios into Bayesian nomogram. Absolute and relative gains were calculated. ANOVA was used to assess statistical association. 4341 patients from 31 studies yielded a pooled sensitivity and specificity US of 83% (95% CI, 78%-87%) and 93% (95% CI, 90%-96%) and 94% (95% CI, 92%-95%) and 94% (95% CI, 94%-96%), respectively, for CT studies. Positive likelihood ratios (LR) for US were 12 and negative LR was 0.18; for CT +LR was 16 and -LR 0.06. Bayesian statistical modeling posttest probabilities for +LR and low Alvarado risk results yielded a posttest probability for US of 83.72% and 87.27% for CT, intermediate risk gave 95.88% and 96.88%, high risk 99.37% and 99.53 respectively. No statistical differences were found between Ultrasound and CT. This Bayesian analysis demonstrated slight superiority of CT scan over US low-risk patients, whereas no significant advantage was seen when evaluating intermediate and high risk patients. This study also favored elevated accuracy of the Alvarado score.

Patterns of Screening Recall Behavior Among Subspecialty Breast Radiologists.

RATIONALE AND OBJECTIVES: Determine whether there are patterns of lesion recall among breast imaging subspecialists interpreting screening mammography, and if so, whether recall patterns correlate to morphologies of screen-detected cancers. MATERIALS AND METHODS: This Institutional Review Board-approved, retrospective review included all screening examinations January 3, 2012-October 1, 2018 interpreted by fifteen breast imaging subspecialists at a large academic medical center and two outpatient imaging centers. Natural language processing identified radiologist recalls by lesion type (mass, calcifications, asymmetry, architectural distortion); proportions of callbacks by lesion types were calculated per radiologist. Hierarchical cluster analysis grouped radiologists based on recall patterns. Groups were compared to overall practice and each other by proportions of lesion types recalled, and overall and lesion-specific positive predictive value-1 (PPV1). RESULTS: Among 161,859 screening mammograms with 13,086 (8.1%) recalls, Hierarchical cluster analysis grouped 15 radiologists into five groups. There was substantial variation in proportions of lesions recalled: calcifications 13%-18% (Chi-square 45.69, p < 0.00001); mass 16%-44% (Chi-square 498.42, p < 0.00001); asymmetry 13%-47% (Chi-square 660.93, p < 0.00001) architectural distortion 6%-20% (Chi-square 283.81, p < 0.00001). Radiologist groups differed significantly in overall PPV1 (range 5.6%-8.8%; Chi-square 17.065, p = 0.0019). PPV1 by lesion type varied among groups: calcifications 9.2%-15.4% (Chi-square 2.56, p = 0.6339); mass 5.6%-8.5% (Chi-square 1.31, p = 0.8597); asymmetry 3.4%-5.9% (Chi-square 2.225, p = 0.6945); architectural distortion 5.6%-10.8% (Chi-square 5.810, p = 0.2138). Proportions of recalled lesions did not consistently correlate to proportions of screen-detected cancer. CONCLUSION: Breast imaging subspecialists have patterns for screening mammography recalls, suggesting differential weighting of imaging findings for perceived malignant potential. Radiologist recall patterns are not always predictive of screen-detected cancers nor lesion-specific PPV1s.

Predictors of malignancy in incidental adnexal lesions identified on CT in patients with prior non-ovarian cancer.

PURPOSE: To identify imaging features in incidental adnexal lesions which are associated with malignancy on portal venous phase contrast-enhanced CT in patients with known non-ovarian cancer. MATERIALS AND METHODS: This IRB-approved, HIPAA-compliant retrospective study was performed at a tertiary cancer center. Portal venous phase contrast-enhanced CT from January 2010 to December 2015 was reviewed to identify women with non-ovarian malignancy and incidental adnexal lesion, with mean 18 months (range 1-80 months) to definitive diagnosis or last imaging follow-up. Imaging features of adnexal lesions were recorded (size, laterality, shape, attenuation, and composition) and correlated with outcome (benign or malignant) using univariate and multivariate logistic regression analysis. A point-based system was used to predict likelihood of malignancy. RESULTS: Of 276 women (mean age 45 years), 216 (78.3%) had benign lesions, 58 (21.0%) ovarian metastasis, and 2 (0.7%) had primary ovarian malignancy. On logistic regression model, lesion size > 5 cm (p-value, OR, 95% CI 0.01, 9.11, 1.70-48.87), bilaterality (< 0.0001, 28.34, 7.46-107.67), irregular shape (0.01, 12.31, 1.61-94.05), higher-than-simple-fluid attenuation (< 0.0001, 28.27, 5.65-141.59), and heterogeneous composition (0.0017, 10.75, 2.45-47.23) were associated with malignant outcome (AUC 0.97). A point-based system incorporating these five features (possible 0-5 points) had AUC of 0.97. Rate of malignancy was 0% (0/147) if none of the features of malignancy were present, 12.7% (8/63) if one feature was present, 51.7% (15/29) if two features were present, and 100% (37/37) if three or more features present. CONCLUSION: Risk of malignancy of incidental adnexal lesions in women with prior non-ovarian cancer can be estimated based on lesion features seen on portal venous phase contrast-enhanced CT.

Electronic Worklist Improves Timeliness of Screening Mammogram Interpretation in an Urban Underserved Population.

OBJECTIVES: To evaluate the impact of an electronic workflow update on screening mammography turnaround time and time to diagnostic imaging for mammography performed on our urban mobile mammography van and at an urban community health center. METHOD: Prior to 10/15/2019, screening exams for the mammography van and urban community health center were made available for interpretation to a single designated radiologist via a manually generated paper list. On 10/15/2019, screening exams were routed electronically onto PACS for any breast radiologist across our Network to interpret. Screening mammogram turnaround time (defined as time form image acquisition to report finalization), time to diagnostic imaging, and time to tissue sampling were collected for pre- and post-implementation periods (6/1-9/30/2019 and 11/1/2019-2/29/2020, respectively) and compared via student t-test and statistical process control analyses. RESULTS: The number of screening exams in the pre- and post-implementation periods were 851 and 728 exams, respectively. Patients were predominately Black and/or African American (400/1579, 25%), non-English speaking (858/1579, 54%) and insured by Medicaid (751/1579, 48%). After implementation of the electronic workflow, turnaround time decreased from 101.0 to 36.4 hours (63.9%, P <0.001) and statistical process control analyses showed sustained decrease in mean turnaround time. However, mean time to diagnostic imaging and tissue sampling were unchanged after implementation (39 vs 45, days; P = 0.330 and 43 vs 59; P = 0.187, respectively). CONCLUSION: Electronic workflow management can reduce screening mammography turnaround time for underserved populations, but additional efforts are warranted to improve time to imaging follow-up for abnormal screening mammograms.

Representing narrative evidence as clinical evidence logic statements.

Objective: Clinical evidence logic statements (CELS) are shareable knowledge artifacts in a semistructured "If-Then" format that can be used for clinical decision support systems. This project aimed to assess factors facilitating CELS representation. Materials and Methods: We described CELS representation of clinical evidence. We assessed factors that facilitate representation, including authoring instruction, evidence structure, and educational level of CELS authors. Five researchers were tasked with representing CELS from published evidence. Represented CELS were compared with the formal representation. After an authoring instruction intervention, the same researchers were asked to represent the same CELS and accuracy was compared with that preintervention using McNemar's test. Moreover, CELS representation accuracy was compared between evidence that is structured versus semistructured, and between CELS authored by specialty-trained versus nonspecialty-trained researchers, using χ2 analysis. Results: 261 CELS were represented from 10 different pieces of published evidence by the researchers pre- and postintervention. CELS representation accuracy significantly increased post-intervention, from 20/261 (8%) to 63/261 (24%, P value < .00001). More CELS were assigned for representation with 379 total CELS subsequently included in the analysis (278 structured and 101 semistructured) postintervention. Representing CELS from structured evidence was associated with significantly higher CELS representation accuracy (P = .002), as well as CELS representation by specialty-trained authors (P = .0004). Discussion: CELS represented from structured evidence had a higher representation accuracy compared with semistructured evidence. Similarly, specialty-trained authors had higher accuracy when representing structured evidence. Conclusion: Authoring instructions significantly improved CELS representation with a 3-fold increase in accuracy. However, CELS representation remains a challenging task.

Integration of a Community Radiology Division into a Subspecialty-Focused Academic Radiology Department.

INTRODUCTION: Assimilate a general radiology division into a subspecialty-focused radiology department at an academic medical center. METHODS: This Institutional Review Board-approved quality improvement initiative was performed at an academic medical centers' subspecialty-focused academic radiology department, aiming to assimilate a general radiology division providing interpretive services for a distributed set of community ambulatory practices. An Oversight Committee charged by the department chair created a charter with unambiguous goal, timelines, clear decision-making, and conflict resolution processes. The Committee assessed the resources and clinical capabilities of the general radiologists, and the anticipated shift in exam volume from the community into subspecialty divisions. Primary outcome, percentage of targeted organ systems-specific interpretations by general radiologists based on assigned subspecialty division, and secondary outcome of report turnaround time (TAT) for all ambulatory exams, were compared before and after sub-specialization. RESULTS: Among 10 general radiologists, 4.5 were assigned to subspecialty divisions; 5.5 continued to cover an independent general radiology practice in a for-profit delivery network. In the 5 months' post-transition, a total 86.6% (11,668/13,477) of reports by the integrated general radiologists were within designated subspecialty divisions vs 23.9% (2,586/10,829) pre-transition (P < 0.01). There was no change in ambulatory radiology report TAT for non-urgent care center (UCC) or UCC exams pre- vs post-integration. DISCUSSION: A quality improvement initiative with unambiguous decision-making and conflict resolution processes incorporated a general radiology practice (radiologists and exams) into a subspecialty-focused academic radiology practice without negatively impacting TAT metrics. Future studies would be needed to assess impact on quality of interpretations.

Variation in Radiologists' Follow-Up Imaging Recommendations for Small Cystic Pancreatic Lesions.

OBJECTIVE: This study aimed to determine the incidence, identify imaging and patient factors, and measure individual radiologist variation associated with follow-up recommendations for small focal cystic pancreatic lesions (FCPLs), a common incidental imaging finding. METHODS: This institutional review board-approved retrospective study analyzed 146,709 reports from abdominal CTs and MRIs performed in a large academic hospital from July 1, 2016, to June 30, 2018. A trained natural language processing tool identified 4,345 reports with FCPLs, which were manually reviewed to identify those containing one or more <1.5-cm pancreatic cysts. For these patients, patient, lesion, and radiologist features and follow-up recommendations for FCPL were extracted. A nonlinear random-effects model estimated degree of variation in follow-up recommendations across radiologists at department and division levels. RESULTS: Of 2,872 reports with FCPLs < 1.5 cm, 708 (24.7%) had FCPL-related follow-up recommendations. Average patient age was 67 years (SD ± 11). In all, 1,721 (60.0%) reports were for female patients; 59.3% of patients had only one cyst. In multivariable analysis, older patients had slightly lower follow-up recommendation rates (odds ratio [OR]: 0.98 [0.98-1.00] per additional year), and lesions associated with main duct dilatation and septation were more likely to have a follow-up recommendation (ORs: 1.93 [1.11-3.36] and 2.88 [1.89-4.38], respectively). Radiologist years in practice (P = .51), trainee presence (P = .21), and radiologist gender (P = .52) were not associated with increased follow-up recommendations. There was significant interradiologist variation in the Abdominal Imaging Division (P = .04), but not in Emergency Radiology (P = .31) or Cancer Imaging Divisions (P = .29). DISCUSSION: Interradiologist variation significantly contributes to variability in follow-up imaging recommendations for FCPLs.

Radiologist Variation in the Rates of Follow-up Imaging Recommendations Made for Pulmonary Nodules.

OBJECTIVE: Determine whether differences exist in rates of follow-up recommendations made for pulmonary nodules after accounting for multiple patient and radiologist factors. METHODS: This Institutional Review Board-approved, retrospective study was performed at an urban academic quaternary care hospital. We analyzed 142,001 chest and abdominal CT reports from January 1, 2016, to December 31, 2018, from abdominal, thoracic, and emergency radiology subspecialty divisions. A previously validated natural language processing (NLP) tool identified 24,512 reports documenting pulmonary nodule(s), excluding reports NLP-positive for lung cancer. A second validated NLP tool identified reports with follow-up recommendations specifically for pulmonary nodules. Multivariable logistic regression was used to determine the likelihood of pulmonary nodule follow-up recommendation. Interradiologist variability was quantified within subspecialty divisions. RESULTS: NLP classified 4,939 of 24,512 (20.1%) reports as having a follow-up recommendation for pulmonary nodule. Male patients comprised 45.3% (11,097) of the patient cohort; average patient age was 61.4 years (±14.1 years). The majority of reports were from outpatient studies (62.7%, 15,376 of 24,512), were chest CTs (75.9%, 18,615 of 24,512), and were interpreted by thoracic radiologists (63.7%, 15,614 of 24,512). In multivariable analysis, studies for male patients (odds ratio [OR]: 0.9 [0.8-0.9]) and abdominal CTs (OR: 0.6 [0.6-0.7] compared with chest CT) were less likely to have a pulmonary nodule follow-up recommendation. Older patients had higher rates of follow-up recommendation (OR: 1.01 for each additional year). Division-level analysis showed up to 4.3-fold difference between radiologists in the probability of making a follow-up recommendation for a pulmonary nodule. DISCUSSION: Significant differences exist in the probability of making a follow-up recommendation for pulmonary nodules among radiologists within the same subspecialty division.

Integrity of clinical information in radiology reports documenting pulmonary nodules.

OBJECTIVE: Quantify the integrity, measured as completeness and concordance with a thoracic radiologist, of documenting pulmonary nodule characteristics in CT reports and assess impact on making follow-up recommendations. MATERIALS AND METHODS: This Institutional Review Board-approved, retrospective cohort study was performed at an academic medical center. Natural language processing was performed on radiology reports of CT scans of chest, abdomen, or spine completed in 2016 to assess presence of pulmonary nodules, excluding patients with lung cancer, of which 300 reports were randomly sampled to form the study cohort. Documentation of nodule characteristics were manually extracted from reports by 2 authors with 20% overlap. CT images corresponding to 60 randomly selected reports were further reviewed by a thoracic radiologist to record nodule characteristics. Documentation completeness for all characteristics were reported in percentage and compared using χ2 analysis. Concordance with a thoracic radiologist was reported as percentage agreement; impact on making follow-up recommendations was assessed using kappa. RESULTS: Documentation completeness for pulmonary nodule characteristics differed across variables (range = 2%-90%, P < .001). Concordance with a thoracic radiologist was 75% for documenting nodule laterality and 29% for size. Follow-up recommendations were in agreement in 67% and 49% of reports when there was lack of completeness and concordance in documenting nodule size, respectively. DISCUSSION: Essential pulmonary nodule characteristics were under-reported, potentially impacting recommendations for pulmonary nodule follow-up. CONCLUSION: Lack of documentation of pulmonary nodule characteristics in radiology reports is common, with potential for compromising patient care and clinical decision support tools.

Comparing Diagnostic Performance of Digital Breast Tomosynthesis and Full-Field Digital Mammography.

OBJECTIVE: Compare diagnostic performance of screening full-field digital mammography (FFDM), a hybrid FFDM and digital breast tomosynthesis (DBT) environment, and DBT only. MATERIALS AND METHODS: This institutional review board-approved, retrospective study consisted of all patients undergoing screening mammography at an urban academic medical center and outpatient imaging facility between January 1, 2011, and December 31, 2017. We used the electronic health record data warehouse to extract report data and patient demographics. A validated natural language processing algorithm extracted BI-RADS score from each report. An institutional cancer registry identified cancer diagnoses. Primary outcomes of recall rate, cancer detection rate (CDR), and positive predictive value 1 (PPV1) were calculated for three periods: FFDM-only environment, hybrid environment, and DBT-only environment. A χ2 test was used to compare recall rate, CDR, and PPV1. RESULTS: A total of 179,028 screening mammograms comprised the study cohort: 41,818 (23.3%) during the FFDM-only period, 83,125 (46.4%) during the hybrid period, and 54,084 (30.2%) during the DBT-only period. Recall rates were 10.4% (4,279 of 41,280) for the FFDM-only period, 10.6% (8,761 of 82,917) for the hybrid period, and 10.8% (5,850 of 54,020) for the DBT-only period (P = .96). CDR (cancers per 1,000 examinations) was 2.6 per 1,000, 4.9 per 1,000, and 6.0 per 1,000 for FFDM only, hybrid, and DBT only, respectively (P < .01). PPV1s (number of cancers per number of recalls) were 2.5% for the FFDM-only period, 4.6% for the hybrid period, and 5.6% for the DBT-only period (P < .01). CONCLUSION: Recall rates were not significantly different within the three periods in the breast imaging practice. However, PPV1 and CDR were significantly higher with DBT only.

Radiologists' Self-Assessment Versus Peer Assessment of Perceived Probability of Recommending Additional Imaging.

OBJECTIVE: Determine radiologist ability to accurately select the probability of recommendation of additional imaging (RAI) for themselves and colleagues when arrayed in a feedback report. METHODS: In this institutional review board-approved study, we analyzed 318,366 diagnostic imaging reports from examinations performed in the radiology department of a large quaternary teaching hospital during calendar year 2016. A validated machine learning algorithm identified reports containing RAI. A multivariable logistic regression model was then used to determine the probability of RAI. In 2018, an e-mailed survey asked radiologists to identify their own RAI probability and that of their colleagues from a report arrayed lowest to highest. Radiologists were grouped into quartiles based on their RAI probability. χ2 Analysis compared self-assessment and assessment of colleagues between quartiles. RESULTS: Forty-eight of 57 radiologists completed the survey (84.2%). Fourteen (29.2%) accurately self-identified their RAI probability (chose the correct quartile); 34 (70.8%) did not. There was no statistically significant difference between quartiles of radiologists and their ability to self-identify their RAI probability (ie, radiologists in the bottom or top quartile of RAI probabilities did not correctly predict their RAI probability). However, radiologists were better able to identify the RAI probability of their colleagues who were in the top and bottom quartiles. DISCUSSION: Radiologists were unable to estimate their own RAI probability but were better at predicting the RAI probability of colleagues. Given that radiologists, and physicians in general, may be poor evaluators of their own performance, objective assessment tools are likely needed to help reduce unwarranted variation.