Background Randomized trials are the gold standard of evidence in healthcare. But their use in QI remains limited, held back by a perception that they do not suit the fast-paced, adaptable world of QI. It doesn’t have to be this way. At NYU Langone we draw on lessons from tech company A/B testing to build a learning healthcare system through RCTs. Now you can too.

Objectives Explain how NYU Langone’s RapidRCT lab designs, implements, and iterates rapid RCTs in our healthcare system – and how other systems can do this too.

Methods Rapidly-iterated RCTs, of different operational changes within NYU. Presentation of two examples; an iterative RCT of pediatric vaccination reminders and a more complex RCT to identify who we should target with reminder phone calls for screening appointments.

Results A cleanliness/safety focused text, and a one-off text did not improve pediatric vaccination rates. But, two reminder texts 48hrs apart were effective (0.4 vaccines per child in intervention, compared to 0.3 in the control arm (p=0.02)). Reminder phone calls were effective for all groups (6.9% of gaps in care closed for intervention vs 0.5% for controls (p<0.001) but were most effective for the quartile of patients predicted to be least likely to close a gap on their own.

Conclusions Healthcare systems can implement RCTs quickly and produce highly robust evidence without undue burden on frontline staff. It can allow systems to tailor interventions to their populations – initial rounds of pediatric vaccine trial showed null results (even though interventions were effective elsewhere), and the call reminder trial showed that our reminders were more effective for some of our patients than others.