Abstract:
|
Biostatisticians routinely perform important work in analyzing healthcare data and creating predictive models. Although this work often leads to publications in medical research journals, the findings seldom lead to improved, sustainable health outcomes. When findings are applied, average time lag from discovery to practice is 17 years. To overcome these problems we created a Learning Healthcare System Platform. Using this infrastructure, we successfully completed several large randomized pragmatic trials of: 1) daily chlorhexidine bathing of ICU patients (n=9,340); 2) a comparison of saline vs. balanced crystalloid fluids (n=29,149); 3) the effectiveness of a post-discharge phone call on hospital readmissions (n=3,054). We have also tested whether real-time predictive models incorporated into the electronic health record can be used to focus prevention and improve outcomes in randomized controlled trials. Hospital operations leaders are equal partners in the learning healthcare system projects and now value having unbiased, rigorous, and efficient evaluations of new and existing hospital initiatives and treatments.
|