Garg, Sushil K

PurposeTo develop a two‐stage three‐dimensional (3D) convolutional neural networks (CNNs) for fully automated volumetric segmentation of pancreas on computed tomography (CT) and to further evaluate its performance in the context of intra‐reader and inter‐reader reliability at full dose and reduced radiation dose CTs on a public dataset.MethodsA dataset of 1994 abdomen CT scans (portal venous phase, slice thickness ≤ 3.75‐mm, multiple CT vendors) was curated by two radiologists (R1 and R2) to exclude cases with pancreatic pathology, suboptimal image quality, and image artifacts (n = 77). Remaining 1917 CTs were equally allocated between R1 and R2 for volumetric pancreas segmentation [ground truth (GT)]. This internal dataset was randomly divided into training (n = 1380), validation (n = 248), and test (n = 289) sets for the development of a two‐stage 3D CNN model based on a modified U‐net architecture for automated volumetric pancreas segmentation. Model’s performance for pancreas segmentation and the differences in model‐predicted pancreatic volumes vs GT volumes were compared on the test set. Subsequently, an external dataset from The Cancer Imaging Archive (TCIA) that had CT scans acquired at standard radiation dose and same scans reconstructed at a simulated 25% radiation dose was curated (n = 41). Volumetric pancreas segmentation was done on this TCIA dataset by R1 and R2 independently on the full dose and then at the reduced radiation dose CT images. Intra‐reader and inter‐reader reliability, model’s segmentation performance, and reliability between model‐predicted pancreatic volumes at full vs reduced dose were measured. Finally, model’s performance was tested on the benchmarking National Institute of Health (NIH)‐Pancreas CT (PCT) dataset.ResultsThree‐dimensional CNN had mean (SD) Dice similarity coefficient (DSC): 0.91 (0.03) and average Hausdorff distance of 0.15 (0.09) mm on the test set. Model’s performance was equivalent between males and females (P = 0.08) and across different CT slice thicknesses (P > 0.05) based on noninferiority statistical testing. There was no difference in model‐predicted and GT pancreatic volumes [mean predicted volume 99 cc (31cc); GT volume 101 cc (33 cc), P = 0.33]. Mean pancreatic volume difference was −2.7 cc (percent difference: −2.4% of GT volume) with excellent correlation between model‐predicted and GT volumes [concordance correlation coefficient (CCC)=0.97]. In the external TCIA dataset, the model had higher reliability than R1 and R2 on full vs reduced dose CT scans [model mean (SD) DSC: 0.96 (0.02), CCC = 0.995 vs R1 DSC: 0.83 (0.07), CCC = 0.89, and R2 DSC:0.87 (0.04), CCC = 0.97]. The DSC and volume concordance correlations for R1 vs R2 (inter‐reader reliability) were 0.85 (0.07), CCC = 0.90 at full dose and 0.83 (0.07), CCC = 0.96 at reduced dose datasets. There was good reliability between model and R1 at both full and reduced dose CT [full dose: DSC: 0.81 (0.07), CCC = 0.83 and reduced dose DSC:0.81 (0.08), CCC = 0.87]. Likewise, there was good reliability between model and R2 at both full and reduced dose CT [full dose: DSC: 0.84 (0.05), CCC = 0.89 and reduced dose DSC:0.83(0.06), CCC = 0.89]. There was no difference in model‐predicted and GT pancreatic volume in TCIA dataset (mean predicted volume 96 cc (33); GT pancreatic volume 89 cc (30), p = 0.31). Model had mean (SD) DSC: 0.89 (0.04) (minimum–maximum DSC: 0.79 −0.96) on the NIH‐PCT dataset.ConclusionA 3D CNN developed on the largest dataset of CTs is accurate for fully automated volumetric pancreas segmentation and is generalizable across a wide range of CT slice thicknesses, radiation dose, and patient gender. This 3D CNN offers a scalable tool to leverage biomarkers from pancreas morphometrics and radiomics for pancreatic diseases including for early pancreatic cancer detection.

Introduction: Clostridioides difficile infection (CDI) is associated with substantial emergency department (ED) and inpatient burden. To date, few studies have evaluated the ED burden of CDI. Using the Nationwide Emergency Department Sample, we evaluated trends in ED use, ED and inpatient charges, admission and mortality rates, length of stay, and independent risk factors for hospital admission and mortality after an ED visit. Methods: Using Nationwide Emergency Department Sample for 2006 through 2014, we identified all patients with the primary diagnosis of CDI (using diagnostic codes). We determined the trends in ED visits and used survey logistic regression analysis to identify factors associated with hospital admission. Results: Overall, 909,236 ED visits for CDI resulted in 817,935 admissions (90%) to the hospital. The number of visits increased from 76,709 in 2006 to 106,869 in 2014, and the admission rate decreased from 92.4% to 84.4%. ED charges adjusted for inflation went up from US$1433.0 to 2900, a significant rise even accounting for inflation. The overall length of hospital stay decreased from 7 to 5.8 days. Independent predictors of admission after ED visits included smoking, use of alcohol, and presence of multiple comorbidities. Independent risk factors for mortality in admitted patients include increasing age and presence of comorbidities. Conclusions: Although ED use for CDI increased, rates of hospital admission decreased over 9 years. Identification of predictors of admission and in-hospital mortality will help guide policies and interventions to reduce the burden on health care resources.

Background: Alcohol abuse and liver disease are associated with high rates of 30-day hospital readmission, but factors linking alcoholic hepatitis (AH) to readmission are not well understood. We aimed to determine the incidence rate of 30-day readmission for patients with AH and to evaluate potential predictors of readmission. Methods: We used the Nationwide Readmissions Database to determine the 30-day readmission rate for recurrent AH between 2010 and 2014 and examined trends in readmissions during the study period. We also identified the 20 most frequent reasons for readmission. Multivariate survey logistic regression analysis was used to identify factors associated with 30-day readmission. Results: Of the 61,750 index admissions for AH, 23.9% were readmitted within 30-days. The rate of readmission did not change significantly during the study period. AH, alcoholic cirrhosis, and hepatic encephalopathy were the most frequent reasons for readmission. In multivariate analysis female sex, leaving against medical advice, higher Charlson comorbidity index, ascites, and history of bariatric surgery were associated with earlier readmissions, whereas older age, payer type (private or self-pay/other), and discharge to skilled nursing-facility reduced this risk. Conclusions: The 30-day readmission rate in patients with AH was high and stable during the study period. Factors associated with readmission may be helpful for development of consensus-based expert guidelines, treatment algorithms, and policy changes to help decrease readmission in AH.

Introduction: Chronic pancreatitis (CP) is a common reason for emergency department (ED) visits, but little research has examined ED use by patients with CP. Materials and Methods: The Nationwide Emergency Department Sample (2006 to 2012) was interrogated to evaluate trends in adult ED visits for a primary diagnosis of CP (International Classification of Disease, 9th revision, Clinical Modification code: 577.1), the rates of subsequent hospital admission, and total charges. A survey logistic regression model was used to determine factors associated with hospitalization from the ED. Results: We identified 253,753 ED visits with a primary diagnosis of CP. No significant trends in annual incidence were noted. However, the ED-to-hospitalization rates decreased by 3% per year (P<0.001) and mean ED charges after adjusting for inflation increased by 11.8% per year (P<0.001). Higher Charlson comorbidity index, current smoker status, alcohol use, and biliary-related CP were associated with hospitalization. In hospitalized patients, length of stay decreased by 2.2% per year (P=0.003) and inpatient charges increased by 2.9% per year (P=0.004). Conclusions: Patient characteristics associated with higher risk of hospitalization from the ED deserve further attention.

Introduction: Acute pancreatitis (AP) is the most common gastroenterology-related reason for hospital admission, and a major source of morbidity and mortality in the United States. This study examines the National Emergency Database Sample, a large national database, to analyze trends in emergency department (ED) utilization and costs, risk factors for hospital admission, and associated hospital costs and length of stay (LOS) in patients presenting with AP. Methods: The National Emergency Database Sample (2006 to 2012) was evaluated for trends in ED visits, ED charges, hospitalization rates, hospital charges, and hospital LOS in patients with primary diagnosis of AP (further subcategorized by age and etiology). A survey logistic-regression model was used to determine factors predictive of hospitalization. Results: A total of 2,193,830 ED visits were analyzed. There was a nonsignificant 5.5% ( P =0.07) increase in incidence of ED visits for AP per 10,000 US adults from 2006 to 2012, largely driven by significant increases in ED visits for AP in the 18 to <45 age group (+9.2%; P =0.025), AP associated with alcohol (+15.9%; P =0.001), and AP associated with chronic pancreatitis (+59.5%; P =0.002). Visits for patients aged ≥65 decreased over the time period. Rates of admission and LOS decreased during the time period, while ED and inpatient costs increased (62.1%; P <0.001 and 7.9%; P =0.0011, respectively). Multiple factors were associated with increased risk of hospital admission from the ED, with the strongest predictors being morbid alcohol use [odds ratio (OR), 4.53; P <0.0001], advanced age (age>84 OR, 3.52; P <0.0001), and smoking (OR, 1.75; P <0.0001). Conclusions: Despite a relative stabilization in the overall incidence of ED visits for AP, continued increases in ED visits and associated costs appear to be driven by younger patients with alcohol-associated and acute on chronic pancreatitis. While rates of hospitalization and LOS are decreasing, associated inflation-adjusted costs are rising. In addition, identified risk factors for hospitalization, such as obesity, alcohol use, and increased age, should be explored in further study for potential use in predictive models and clinical improvement projects.

Introduction: Although gastric cancer (GC) rates have been declining in the United States, it continues to be a major cause of morbidity. This study examined trends in hospital admissions, in-hospital mortality, length of stay (LOS), and inpatient costs related to GC. In addition, various factors associated with in-hospital mortality, LOS, and inpatient costs were examined. Methods: National inpatient sample—the largest publicly available all-payer inpatient care database—was interrogated to obtain information about various demographic and hospital-related factors (including those mentioned above) in patients who were primarily admitted for GC between the years 1998 to 2013. These trends were analyzed. Multivariate analysis was also performed to identify risk factors associated with LOS, costs, and mortality. Results: A total of 679,330 hospital discharges with the principal diagnosis of GC were obtained. Hospital stays increased by approximately 340 stays per year (±110; P=0.00079). However, inpatient mortality rate and LOS declined by 0.36% per year (±0.024%; P<0.0001), and 0.11 days per year (±0.01; P<0.0001), respectively. The inpatient charges have increased at the rate of $3241 per year (±133.3; P<0.0001). Differences in mortality rate, LOS, and inpatient costs were affected by multiple factors. Conclusions: Despite the overall decline in GC incidence, the incidence of hospitalizations per 100,000 US population related to GC did not change significantly. Although LOS and mortality declined, inpatient charges increased over the study period.