Dr. Lonsdale’s Research

Predictive Modeling for Improved Critical Care Availability

Abstract: As anesthesiologists, we regularly experience the challenge of critical care shortages, which lead to delays in essential surgery for our patients. This bottleneck in hospital capacity is an immediate and critical problem that will worsen as demands for healthcare continue to increase. Timely discharge of medically fit patients is essential to improving care capacity, but some patients wait extra days in hospital for completion of the preparation necessary to support their postdischarge needs. Anticipation of medical readiness for discharge with sufficient lead-time to complete preparations for home remains an unmet clinical need. One group of critical care patients with complex postdischarge needs are premature infants, who are discharged directly home from the neonatal intensive care unit (NICU). Prematurity occurs in 10% of births and has an annual societal cost of over $25 billion. Delayed NICU discharges are thought to be common but are under-studied and poorly quantified. Predictive modeling could be used to trigger timely interventions that improve NICU discharge planning, but essential steps such as prospective assessment of real-world model performance and understanding how clinicians could best use predictive models to improve patient outcomes are rarely studied.

My long-term aim is to optimize care availability by integrating predictive modeling into a comprehensive approach to reducing barriers to timely discharge.

For this IMRA, I will use a novel electronic health record-based tool to accurately quantify delayed NICU discharge; extensively test and refine my existing NEO-READY model, which estimates the probability of discharge within 5 days for premature NICU patients; and systematically identify barriers to discharge and targets for interventions that could be triggered by the model in a future intervention strategy. I will then use a future K23 award to develop and clinically implement a model-centric strategy to improve the incidence of delayed discharge.