Rationale and objectives

Radiology turnaround time is an important quality measure that can impact hospital workflow and patient outcomes. We aimed to develop a machine learning model to predict delayed turnaround time during non-business hours and identify factors that contribute to this delay.

Materials and methods

This retrospective study consisted of 15,117 CT cases from May 2018 to May 2019 during non-business hours at two hospital campuses after applying exclusion criteria. Of these 15,177 cases, 7,532 were inpatient cases and 7,585 were emergency cases. Order time, scan time, first communication by radiologist, free-text indications, and other clinical metadata were extracted. A combined XGBoost classifier and Random Forest natural language processing model was trained with 85% of the data and tested with 15% of the data. The model predicted two measures of delay: when the exam was ordered to first communication (total time) and when the scan was completed to first communication (interpretation time). The model was analyzed with the area under the curve (AUC) of receiver operating characteristic (ROC) and feature importance. Source code: https://bit.ly/2UrLiVJ RESULTS: The algorithm reached an AUC of 0.85, with a 95% confidence interval [0.83, 0.87], when predicting delays greater than 245 minutes for "total time" and 0.71, with a 95% confidence interval [0.68, 0.73], when predicting delays greater than 57 minutes for "interpretation time". At our institution, CT scan description (e.g. "CTA chest pulmonary embolism protocol"), time of day, and year in training were more predictive features compared to body part, inpatient status, and hospital campus for both interpretation and total time delay.

Conclusion

This algorithm can be applied clinically when a physician is ordering the scan to reasonably predict delayed turnaround time. Such a model can be leveraged to identify factors associated with delays and emphasize areas for improvement to patient outcomes.

Background Though incidental pulmonary nodules are common, rates of guideline-recommended surveillance and associations between surveillance and mortality are unclear. In this study, we describe adherence (categorized as complete, partial, late and none) to guideline-recommended surveillance among patients with incidental 5-8 mm pulmonary nodules and assess associations between adherence and mortality. Methods This was a retrospective cohort study of 551 patients (≥35 years) with incidental pulmonary nodules conducted from September 1, 2008 to December 31, 2016, in an integrated safety-net health network. Results Of the 551 patients, 156 (28%) had complete, 87 (16%) had partial, 93 (17%) had late and 215 (39%) had no documented surveillance. Patients were followed for a median of 5.2 years [interquartile range (IQR), 3.6-6.7 years] and 82 (15%) died during follow-up. Adjusted all-cause mortality rates ranged from 2.24 [95% confidence interval (CI), 1.24-3.25] deaths per 100 person-years for complete follow-up to 3.30 (95% CI, 2.36-4.23) for no follow-up. In multivariable models, there were no statistically significant associations between the levels of surveillance and mortality (p > 0.16 for each comparison with complete surveillance). Compared with complete surveillance, adjusted mortality rates were non-significantly increased by 0.45 deaths per 100 person-years (95% CI, -1.10 to 2.01) for partial, 0.55 (95% CI, -1.08 to 2.17) for late and 1.05 (95% CI, -0.35 to 2.45) for no surveillance. Conclusions Although guideline-recommended surveillance of small incidental pulmonary nodules was incomplete or absent in most patients, gaps in surveillance were not associated with statistically significant increases in mortality in a safety-net population.