A Granular View of Emergency Department Length of Stay: Improving Predictive Power and Extracting Real-Time, Actionable Insights.

STUDY OBJECTIVE: Improved understanding of factors affecting prolonged emergency department (ED) length of stay is crucial to improving patient outcomes. Our investigation builds on prior work by considering ED length of stay in operationally distinct time periods and using benchmark and novel machine learning techniques applied only to data that would be available to ED operators in real time. METHODS: This study was a retrospective review of patient visits over 1 year at 2 urban EDs, including 1 academic and 1 academically affiliated ED, and 2 suburban, community EDs. ED length of stay was partitioned into 3 components: arrival-to-room, room-to-disposition, and admit disposition to departure. Prolonged length of stay for each component was considered beyond 1, 3, and 2 hours, respectively. Classification models (logistic regression, random forest, and XGBoost) were applied, and important features were evaluated. RESULTS: In total, 135,044 unique patient encounters were evaluated for the arrival-to-room, room-to-disposition, and admit disposition-to-departure models, which had accuracy ranges of 84% to 96%, 66% to 77%, and 62% to 72%, respectively. Waiting room and ED volumes were important features in all arrival-to-room models. Room-to-disposition results identified patient characteristics and ED volume as the most important features for prediction. Boarder volume was an important feature of the admit disposition-to-departure models for all sites. Academic site models noted nurse staffing ratios as important, whereas community site models noted hospital capacity and surgical volume as important for admit disposition-to-departure prediction. CONCLUSION: This study identified granular capacity, flow, and nurse staffing predictors of ED length of stay not previously reported in the literature. Our novel methodology allowed for more accurate and operationally meaningful findings compared to prior modeling methods.

Risk of persistent air leaks following percutaneous cryoablation and microwave ablation of peripheral lung tumors.

OBJECTIVES: To compare the incidence of persistent air leak (PAL) following cryoablation vs MWA of lung tumors when the ablation zone includes the pleura. METHODS: This bi-institutional retrospective cohort study evaluated consecutive peripheral lung tumors treated with cryoablation or MWA from 2006 to 2021. PAL was defined as an air leak for more than 24 h after chest tube placement or an enlarging postprocedural pneumothorax requiring chest tube placement. The pleural area included by the ablation zone was quantified on CT using semi-automated segmentation. PAL incidence was compared between ablation modalities and a parsimonious multivariable model was developed to assess the odds of PAL using generalized estimating equations and purposeful selection of predefined covariates. Time-to-local tumor progression (LTP) was compared between ablation modalities using Fine-Gray models, with death as a competing risk. RESULTS: In total, 260 tumors (mean diameter, 13.1 mm ± 7.4; mean distance to pleura, 3.6 mm ± 5.2) in 116 patients (mean age, 61.1 years ± 15.3; 60 women) and 173 sessions (112 cryoablations, 61 MWA) were included. PAL occurred after 25/173 (15%) sessions. The incidence was significantly lower following cryoablation compared to MWA (10 [9%] vs 15 [25%]; p = .006). The odds of PAL adjusted for the number of treated tumors per session were 67% lower following cryoablation (odds ratio = 0.33 [95% CI, 0.14-0.82]; p = .02) vs MWA. There was no significant difference in time-to-LTP between ablation modalities (p = .36). CONCLUSIONS: Cryoablation of peripheral lung tumors bears a lower risk of PAL compared to MWA when the ablation zone includes the pleura, without adversely affecting time-to-LTP. KEY POINTS: • The incidence of persistent air leaks after percutaneous ablation of peripheral lung tumors was lower following cryoablation compared to microwave ablation (9% vs 25%; p = .006). • The mean chest tube dwell time was 54% shorter following cryoablation compared to MWA (p = .04). • Local tumor progression did not differ between lung tumors treated with percutaneous cryoablation compared to microwave ablation (p = .36).

Comparison of expected imaging findings following percutaneous microwave and cryoablation of pulmonary tumors: ablation zones and thoracic lymph nodes.

OBJECTIVE: To compare temporal changes of ablation zones and lymph nodes following lung microwave ablation (MWA) and cryoablation. METHODS: This retrospective cohort study compared lung ablation zones and thoracic lymph nodes following MWA and cryoablation performed 2006-2020. In the ablation zone cohort, ablation zone volumes were measured on serial CT for 12 months. In the lymph node cohort, the sum of bidimensional products of lymph node diameters was measured before (baseline) and up to 6 months following ablation. Cumulative incidence curves estimated the time to 75% ablation zone reduction and linear mixed-effects regression models compared the temporal distribution of ablation zones and lymph node sizes between modalities. RESULTS: Ablation zones of 59 tumors treated in 45 sessions (16 MWA, 29 cryoablation) in 36 patients were evaluated. Differences in the time to 75% volume reduction between modalities were not detected. Following MWA, half of the ablation zones required an estimated time of 340 days to achieve a 75% volume reduction compared to 214 days following cryoablation (p = .30). Thoracic lymph node sizes after 33 sessions (13 MWA, 20 cryoablation) differed between modalities (baseline-32 days, p = .01; 32-123 days, p = .001). Following MWA, lymph nodes increased on average by 38 mm2 (95%CI, 5.0-70.7; p = .02) from baseline to 32 days, followed by an estimated decrease of 50 mm2 (32-123 days; p = .001). Following cryoablation, changes in lymph nodes were not detected (baseline-32 days, p = .33). CONCLUSION: The rate of ablation zone volume reduction did not differ between MWA and cryoablation. Thoracic lymph nodes enlarged transiently after MWA but not after cryoablation. KEY POINTS: • Contrary to current belief, the rate of lung ablation zone volume reduction did not differ between microwave and cryoablation. • Transient enlargement of thoracic lymph nodes after microwave ablation was not associated with regional tumor spread and decreased within six months following ablation. • No significant thoracic lymph node enlargement was observed following cryoablation.

Active Versus Passive Thaw After Percutaneous Cryoablation of Pulmonary Tumors: Effect on Incidence, Grade, and Onset of Hemoptysis.

BACKGROUND. Hemoptysis is common after percutaneous image-guided cryoablation of pulmonary tumors. OBJECTIVE. The purpose of our study was to evaluate the effect of a final active thaw on the incidence, grade, and onset of hemoptysis after percutaneous cryoablation of pulmonary tumors. METHODS. This retrospective cohort study included 60 consecutive CT-guided cryoablation sessions targeting 95 pulmonary tumors in 47 patients from March 2017 to September 2020. The final thaw of a triple-freeze protocol was active (electrical, helium-free) in 27 of 60 sessions (45%, active group) and passive in 33 of 60 sessions (55%, passive group). The incidence, onset, and management of hemoptysis were recorded using prospectively collected data. Hemoptysis, pneumothorax, and hemothorax within 30 days after ablation were graded according to Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. The volume of immediate posttreatment changes on CT was quantified using semiautomated segmentation. Outcomes were compared between groups using generalized estimating equation models. A parsimonious multivariable model for hemoptysis incidence was developed using purposeful selection of predefined covariates followed by bootstrap resampling. Local tumor control was compared between groups using the Kaplan-Meier method and log-rank testing. RESULTS. Hemoptysis occurred after 26 of 60 (43%) sessions and was self-limited (CTCAE grade 1) in 22 of 26 (85%) sessions. The incidence of hemoptysis was lower in the active group than in the passive group (19% vs 64%, respectively; p = .002). The odds of hemoptysis adjusted for immediate posttreatment changes were 92% lower in the active group (odds ratio [OR], 0.08 [95% CI, 0.02-0.37]; p = .004). The odds of hemoptysis greater than grade 1 were 79% lower in the active group (OR, 0.21 [95% CI, 0.07-0.64]; p = .006). In the active group, the onset of hemoptysis was significantly delayed (OR, 0.75 [95% CI, 0.61-0.91]; p = .005). Pneumothorax (p = .60), hemothorax (p = .84), and local tumor control (p = .77) did not differ between groups. CONCLUSION. Active thaw after the final freeze reduces the incidence and grade of hemoptysis and delays the onset of hemoptysis after percutaneous cryoablation of pulmonary tumors without adversely affecting other procedural complications and local tumor control. CLINICAL IMPACT. Active thaw after the final freeze improves the safety profile of triple-freeze cryoablation of pulmonary tumors by reducing the incidence and grade of hemoptysis and by delaying the onset of hemoptysis beyond the immediate recovery period.

Impact of the Medicare hospital readmissions reduction program on vulnerable populations.

BACKGROUND: The Hospital Readmissions Reduction Program (HRRP) was established by the 2010 Patient Protection and Affordable Care Act (ACA) in an effort to reduce excess hospital readmissions, lower health care costs, and improve patient safety and outcomes. Although studies have examined the policy's overall impacts and differences by hospital types, research is limited on its effects for different types of vulnerable populations. The aim of this study was to analyze the impact of the HRRP on readmissions for three targeted conditions (acute myocardial infarction, heart failure, and pneumonia) among four types of vulnerable populations, including low-income patients, patients served by hospitals that serve a high percentage of low-income or Medicaid patients, and high-risk patients at the highest quartile of the Elixhauser comorbidity index score. METHODS: Data on patient and hospital information came from the Nationwide Readmission Database (NRD), which contained all discharges from community hospitals in 27 states during 2010-2014. Using difference-in-difference (DD) models, linear probability regressions were conducted for the entire sample and sub-samples of patients and hospitals in order to isolate the effect of the HRRP on vulnerable populations. Multiple combinations of treatment and control groups and triple difference (DDD) methods were used for testing the robustness of the results. All models controlled for the patient and hospital characteristics. RESULTS: There have been statistically significant reductions in readmission rates overall as well as for vulnerable populations, especially for acute myocardial infarction patients in hospitals serving the largest percentage of low-income patients and high-risk patients. There is also evidence of spillover effects for non-targeted conditions among Medicare patients compared to privately insured patients. CONCLUSIONS: The HRRP appears to have created the right incentives for reducing readmissions not only overall but also for vulnerable populations, accruing societal benefits in addition to previously found reductions in costs. As the reduction in the rate of readmissions is not consistent across patient and hospital groups, there could be benefits to adjusting the policy according to the socioeconomic status of a hospital's patients and neighborhood.

Identifying patterns in administrative tasks through structural topic modeling: A study of task definitions, prevalence, and shifts in a mental health practice's operations during the COVID-19 pandemic.

OBJECTIVE: This case study illustrates the use of natural language processing for identifying administrative task categories, prevalence, and shifts necessitated by a major event (the COVID-19 [coronavirus disease 2019] pandemic) from user-generated data stored as free text in a task management system for a multisite mental health practice with 40 clinicians and 13 administrative staff members. MATERIALS AND METHODS: Structural topic modeling was applied on 7079 task sequences from 13 administrative users of a Health Insurance Portability and Accountability Act-compliant task management platform. Context was obtained through interviews with an expert panel. RESULTS: Ten task definitions spanning 3 major categories were identified, and their prevalence estimated. Significant shifts in task prevalence due to the pandemic were detected for tasks like billing inquiries to insurers, appointment cancellations, patient balances, and new patient follow-up. CONCLUSIONS: Structural topic modeling effectively detects task categories, prevalence, and shifts, providing opportunities for healthcare providers to reconsider staff roles and to optimize workflows and resource allocation.