A discrete event simulation (DES) model of patient flow incorporating infection control policy for vancomycin-resistant Enterococcus (VRE) and methicillin-resistant Staphylococcus aureus (MRSA)
Background: Infection control policies are critical determinants of patient placement during hospital admission, when patients must be matched to beds based on acuity, service and gender, as well as VRE and MRSA colonization status. We designed and validated a novel DES model of patient flow.
Methods: The model uses a bed-allocation algorithm matching patients to beds in a simulated hospital (Figure). Patients are matched to appropriate beds based on acuity, service, gender and observed colonization status, which may be discordant with their true colonization status, discovered only through diagnostic testing. A data repository of 104,725 historical admissions over a 2-year period was used to populate the model with patients arriving hourly with characteristics drawn from a joint distribution of acuity (12% observation; 68% general; 12% step-down and 8% ICU), service (53% medicine; 47% surgery) , gender (49% female) and colonization status (91% non-colonized, 4% VRE, 3% MRSA, 2% MRSA/VRE). áStochastic inputs included: probability distributions of hourly time-varying acuity changes and discharges, published estimates of true colonization, culture and nucleic acid test characteristics, and VRE and MRSA transmission. The model was validated to mean length of stay (LOS, d) and mean occupancy to ensure accurate capture of patient flow, and run over a 5-year period.
Results: The model reliably assigned patients to appropriate beds, including when patients experienced changes in acuity or observed colonization status. Mean LOS (▒ standard deviation) in the data repository was 4.7 ▒ 5.5d; the model-estimated LOS after five years and >248,000 admissions was 4.9 ▒ 5.0d. We achieved a valid occupancy of 83.8 ▒ 4.7% compared to hospital-reported 82.9 ▒ 1.7%. Other reportable model outcomes include false-cohorting when cohorted patients differ on true colonization status; queues for available resources; and VRE and MRSA transmission under varying surveillance strategies.
Conclusion: DES for modeling patient flow has important applications in infection control. Expansion of the model to incorporate transmission under strategies employing diagnostics to more accurately identify patients with underlying VRE and MRSA colonization will increase its utility to clinicians and policy makers.
E. S. Shenoy,
T. Hou, None
E. Ryan, None
J. Cotter, None
W. Ware, None
D. Hooper, None
R. Walensky, None