'Learn From Every Patient': implementation and early results of a learning health system

Abstract

Aim: The convergence of three major trends in medicine, namely conversion to electronic health records (EHRs), prioritization of translational research, and the need to control healthcare expenditures, has created unprecedented interest and opportunities to develop systems that improve care while reducing costs. However, operationalizing a 'learning health system' requires systematic changes that have not yet been widely demonstrated in clinical practice.

Method: We developed, implemented, and evaluated a model of EHR-supported care in a cohort of 131 children with cerebral palsy that integrated clinical care, quality improvement, and research, entitled 'Learn From Every Patient' (LFEP).

Results: Children treated in the LFEP Program for a 12-month period experienced a 43% reduction in total inpatient days (p=0.030 vs prior 12mo period), a 27% reduction in inpatient admissions, a 30% reduction in emergency department visits (p=0.001), and a 29% reduction in urgent care visits (p=0.046). LFEP Program implementation also resulted in reductions in healthcare costs of 210% (US$7014/child) versus a Time control group, and reductions of 176% ($6596/child) versus a Program Activities control group. Importantly, clinical implementation of the LFEP Program has also driven the continuous accumulation of robust research-quality data for both publication and implementation of evidence-based improvements in clinical care.

Interpretation: These results demonstrate that a learning health system can be developed and implemented in a cost-effective manner, and can integrate clinical care and research to systematically drive simultaneous clinical quality improvement and reduced healthcare costs.