Clinical Compliance in Personalised Model-based Medical Decision Support: Do computers and interfaces yield better compliance?

Personalised and model-based clinical care is on the rise and offers significant opportunity to improve care and productivity in response to rising demographic demands and rising costs. Compliance in critical care is important for any protocolised clinical care. However, it is often overlooked and very infrequently studied. Critically ill patients often experience stress-induced hyperglycemia, which has been shown to result in increased morbidity and mortality. Safe, effective glycemic control (GC) can reduce mortality and improve outcomes, but is only effective if strong compliance is observed within the clinical practice. This study examines insulin-nutrition dosing compliance for STAR, a tablet-based protocol designed to easily adapt to variable clinical practice, its neonatal intensive care unit version GRYPHON, and a standard paper based clinical protocol (CHU). All interventions and changes were recorded for all three cohorts, and a sub-cohort was used to examine the validity of the data used. Compliance was over 99% for STAR, over 98% for GRYPHON and 80% for CHU. The differences is attributed primarily to interface design and its focus on ease of use and natural use for clinical staff. However, while compliance is higher, the reasons for good compliance in any such system remains to be more precisely specified with appropriate research tests.