This study tackles interobserver variability with respect to specialty training in manual segmentation of non-small cell lung cancer (NSCLC). Four readers included for segmentation are: a data scientist (BY), a medical student (LS), a radiology trainee (MH), and a specialty-trained radiologist (SK) for a total of 293 patients from two publicly available databases. Sørensen–Dice (SD) coefficients and low rank Pearson correlation coefficients (CC) of 429 radiomics were calculated to assess interobserver variability. Cox proportional hazard (CPH) models and Kaplan-Meier (KM) curves of overall survival (OS) prediction for each dataset were also generated. SD and CC for segmentations demonstrated high similarities, yielding, SD: 0.79 and CC: 0.92 (BY-SK), SD: 0.81 and CC: 0.83 (LS-SK), and SD: 0.84 and CC: 0.91 (MH-SK) in average for both databases, respectively. OS through the maximal CPH model for the two datasets yielded c-statistics of 0.7 (95% CI) and 0.69 (95% CI), while adding radiomic and clinical variables (sex, stage/morphological status, and histology) together. KM curves also showed significant discrimination between high- and low-risk patients (p-value < 0.005). This supports that readers’ level of training and clinical experience may not significantly influence the ability to extract accurate radiomic features for NSCLC on CT. This potentially allows flexibility in the training required to produce robust prognostic imaging biomarkers for potential clinical translation.

No abstract

Background Seasonal fluctuations in antibiotic prescribing for respiratory tract diagnoses (RTDs) have been identified, but characteristics and appropriateness of these variations have not been well described. The objectives of this study were to describe seasonal variations for RTDs and to determine whether seasonal variation in prescribing is associated with inappropriate use. Methods From July 1, 2016 through June 30, 2017, antibiotic prescribing was analyzed for 31 primary care practices comparing winter (October-March) and summer (April-September) months. ICD-10 codes for RTDs were described as tier 1, 2, or 3 based on whether antibiotics are almost always, sometimes, or almost never indicated, respectively. Twenty visits from each of 60 providers were randomly selected and manually reviewed to determine a gold standard of antibiotic appropriateness in order to characterize the appropriateness of these seasonal variations. Associations between season and diagnosis tier, season and appropriateness, and individual provider seasonal changes in antibiotic prescribing and provider characteristics were determined. Results There was a lower proportion of visits with tier 3 diagnoses in winter months (68% v. 74%, p&lt; 0.01), but a greater proportion of tier 2 diagnoses (29% v. 23%, p&lt; 0.01). There were greater proportions of visits in which an antibiotic was prescribed for both tier 2 (80% vs 74%, p&lt; 0.01) and tier 3 diagnoses (23% v. 16%, p&lt; 0.01) in winter months. Using medical record review, inappropriate antibiotics were prescribed more frequently for RTDs in winter compared to summer months (73% v. 64%, p&lt; 0.01). Greater individual provider difference in proportion of RTD visits in which an antibiotic was prescribed from summer to winter was associated with family medicine v. internal medicine specialty (8.2% v. 5.1%, p&lt; 0.01), nonteaching v. teaching practice (8.1% v. 2.9%, p&lt; 0.01), and nonurban v. urban setting (9.1% v. 3.9%, p&lt; 0.01). Conclusion Although there was a greater proportion of tier 2 compared to tier 3 RTDs in winter months, winter months were associated with more inappropriate prescribing than in summer months. More investigation is needed to understand the drivers for seasonal variations in RTDs and inappropriate prescribing. Disclosures Kathleen Degnan, MD, Gilead: Grant/Research Support Michael Z. David, MD PhD, Contrafect: Grant/Research Support|GSK: Advisor/Consultant|Johnson and Johnson: Advisor/Consultant.