Ward Capacity Strain: A Novel Predictor of Delays in Intensive Care Unit Survivor Throughput

• Published in: Annals of the American Thoracic Society Vol. 16; no. 3; pp. 387 - 390
• Main Authors: Kohn, Rachel, Harhay, Michael O, Weissman, Gary E, Anesi, George L, Bayes, Brian, Greysen, S Ryan, Ratcliffe, Sarah J, Halpern, Scott D, Prasad Kerlin, Meeta
• Format: Journal Article
• Language: English
• Published: United States American Thoracic Society 01.03.2019
• Subjects: Censuses; Clinical outcomes; Comorbidity; Emergency Service, Hospital - statistics & numerical data; Hospital Bed Capacity - statistics & numerical data; Hospitalists; Humans; Illnesses; Intensive care; Intensive Care Units - statistics & numerical data; Letters; Mortality; Patients; Retrospective Studies; Sepsis; Studies; Survivors; Telemetry; Time-to-Treatment - trends; United States; Workloads; United States
• ISSN: 2329-6933; 2325-6621; 2325-6621
• DOI: 10.1513/AnnalsATS.201809-621RL

In prior research of intensive care units (ICUs) (2, 3) and emergency departments (EDs) (4), higher patient volume, turnover, and severity of illness have shown associations with patient outcomes (e.g., mortality), patient throughput, and care delivery (5-11). Variables included ward-level measures of patient volume (census, numbers of admissions and discharges), staff workload (presence of any off-ward transports and transfusions, and numbers of medications administered and respiratory therapy orders), and overall acuity (number of patients on telemetry, presence of any patients with quick Sequential Organ Failure Assessment >2 [24], and ICU transfers). To identify strain variables associated with longer admission wait times and to quantify the variance explained by ward capacity strain, we estimated two least absolute shrinkage and selection operator (LASSO) (25) regressions, a method for building parsimonious regression models that selects variables in descending order of variance explained and incorporates a penalty for adding more variables. [...]LASSO regression is only one of multiple penalized regression and alternate methods to identify predictors of ward admission wait times.