CENTER for BIOETHICS Strains on an ICU s Capacity to Provide Optimal Care Scott D. Halpern, M.D., Ph.D. Assistant Professor of Medicine and Epidemiology Deputy Director, Center for Health Incentives and Behavioral Economics, Leonard Davis Institute of Health Economics
ICU strain may lead to worse Single ICU from 92 95 1,050 patients 8 beds + 2 flex; ICU nurses for up to 6 beds
or maybe not APACHE data: 108 ICUs 02-05 Admission ICU census relative to that ICU s mean Iwashyna TJ, Kramer AA, Kahn JM. Crit Care Med 2009; 37:
ICU capacity strain: time-varying influence on a given ICU s ability to provide high-quality care on a given day Halpern SD. Curr Opinion Crit Care 2011; 17:
ICU capacity strain: time-varying influence on a given ICU s ability to provide high-quality care on a given day ICUs as balloons: able to accommodate increasing strain only up to a point, after which outcomes deteriorate Halpern SD. Curr Opinion Crit Care 2011; 17:
Key empirical questions re: Capacity 1. What factors contribute to ICU capacity strain (what increases pressure inside the balloon)? 2. What happens during periods of excessive strain (what happens if the balloon pops)? 3. What ICU characteristics enable some units to accommodate more strain than others (why do balloons differ in their elasticity)? Halpern SD. Curr Opinion Crit Care 2011; 17:
Informatio n captured Ease of data capture
Deriving a metric of capacity strain Prospective cohort of all patients admitted to Penn MICU (24 beds) over a 6-month period (July Dec, 2010) Objective assessment of potential elements of capacity strain each day from electronic records & chart review (candidate exposures) Daily survey of ICU attendings and charge nurses perceptions of capacity strain that day (outcomes) Sought to derive most parsimonious model (unit of analysis = day)
How strained do ICU clinicians On a scale from 1 10, how would you rate the overall strain or pressure on your team s capacity to provide high-quality critical care today? 1 indicates the least strain you ve ever felt, and 10 the most. 89% surveys completed (177 days) 17 MDs and 18 RNs Median strain score among charge nurses was 6 (IQR 3-7), and among attendings was 5 (IQR 3-7) Moderate correlation between attendings and charge nurses: ICC = 0.45 (0.30-0.60) Vranas K, Prasad M, Halpern SD. Presented at
Deriving parsimonious model of ICU Prasad M, Ratcliffe SJ, Vranas K, Harhay MO, Cooney E, Halpern SD (in
What might happen during times of Halpern SD. Curr Opinion Crit Care 2011; 17:
What might happen during times of Halpern SD. Curr Opinion Crit Care 2011; 17:
Developed by SCCM in 1996 Restrictions: ICUs participating in Project IMPACT for at least 1 year ICU-months with 10 patients with full clinical data Data provided through agreement with Cerner Corp.
census: 1.02 (1.00 1.04) average p(death): NS interaction: p = 0.03 Gabler NB, Ratcliffe SJ, Strom BL, Angus DC, Rubenfeld GD, Halpern SD. (in
Gabler NB, Ratcliffe SJ, Strom BL, Angus DC, Rubenfeld GD, Halpern SD. (in preparation) Effect modification by ICU staffing Main effect of standardized census on in-hospital mortality: OR: 1.02 (1.00 1.04) Closed ICUs: OR = 1.07 (1.02 1.11) Open ICUs: OR = 1.01 (0.99 1.03) p-value for Census X Staffing Model interaction = 0.03
Discharge day strain 209,075 patients eligible for ICU readmission at 163 ICUs participating in Project IMPACT from 2001 to 2008 Patient SOI and other factors fully adjusted for; ICU-year as fixed effect No significant interactions observed. Wagner JE, Ratcliffe SJ, Gabler NB, Brown SES, Halpern SD. (in
Future directions 1. Further explore optimal measures of ICU capacity strain by quantifying improvements in precision in an interventional RCT 2. Identify mechanisms (e.g., clinicians time, treatment intensity, prophylaxis provision) linking strain with outcomes 3. Evaluate influences of capacity strain on: equity (e.g., from whom are ICU beds/services withheld) patient-centered care (e.g., incidence of family meetings)
Acknowledgements Penn research team Sydney Brown Elizabeth Cooney, MPH Nicole Gabler, PhD Michael Harhay, MPH, MBE Max Herlim Meeta Prasad, MD, MSCE Sarah Ratcliffe, PhD Senior collaborators Derek Angus, MD, MPH David Asch, MD, MBA Gordon Rubenfeld, MD, MSc Cerner Andrew Kramer, PhD Maureen Stark Kelly Vranas, MD Jason Wagner, MD K08 HS018406 (Agency for Healthcare Research and Quality) Society of Critical Care Medicine VISION Grant (2009)