Background/Purpose:

For projects such as the ACR Rheumatology Informatics System for Effectiveness (RISE) to succeed for quality improvement, practice benchmarking, & research, the data included must be high fidelity and represent the majority of patients in participating practices.  To prepare a large academic practice for RISE participation, we synergistically harnessed the power of institutional informatics coupled with systems engineering.  We focused on two Rheumatoid Arthritis (RA) assessments: Routine Assessment of Patient Index Data (RAPID3) & Clinical Disease Activity Index (CDAI).  Baseline data was extremely low (13% & 27% respectively).  The aim was to improve compliance for these instruments in the Rheumatology clinics to > 75% by December 2016.

Methods:

High quality data was extracted from our RA registry established in April 2015.  Inclusion criteria: RA patients seen by a rheumatology provider twice in the last 24 months & at least once in last 12 months.  By Feb 2017, the registry had 543 patients.  The project team utilized Six Sigma DMAIC (Define, Measure, Analyze, Improve & Control) data-driven improvement cycle for problem-solving & process improvement.  Tools & techniques included brainstorming, project charter, process maps, fishbone diagrams, Pareto chart, & feedback communications.  Interventions were aimed towards standardization of clinic work flow, and motivating better EMR documentation practices.

Results:

We examined growth of compliance (proportion of completed assessments) over time.  We found very strong effect sizes (r > .90, p < .001) for both completion of the Rapid3 & CDAI (average growth rates over 3% a month).  Their compliance was also highly related; but Rapid3 compliance was not the sole explanatory variable for CDAI compliance as demonstrated by mediational analyses.

We have created modified control charts (Tables 1 & 2).  Given our goal to demonstrate differences over time, the standard error of difference (SED; (σ/N)*√ (2)*1.96) within month was calculated and used to create the bounds around the observed compliance rates.  Any month’s proportion rate outside of other months’ SED bounds is statistically different at α = .05.

Conclusion:

Combining the power of informatics & systems engineering led to impressive improvements in provider behaviors.  This methodology helped to identify opportunities to overcome barriers & to motivate providers.  High quality processes facilitated high quality data, and vice versa.  Performance expectations were clear, improvements were sustained, and we have believable long-term RA outcomes data that can be used to improve outcomes at the practice level.