Potential MI cases were identified monthly. Counts of MI diagnoses generated by the VDW were calculated with each system update. These were overlaid with prior VDW updates, as well as those generated by the legacy sources. Cohort eligibility criteria such as age, which make use of other data sources, were factored into the comparisons. Plans were made to compare new data to a running mean by the use of a dashboard. Unexpected and nontrivial divergence in the number of new cases identified by

more »

... two approaches will prompt additional investigation. Results: The dashboard was implemented in September and October of 2012. Counts for MI diagnostic codes appearing in each month starting in January 2004 were compared across the September and October runs. The two series were nearly coincident between these successive versions of the VDW for events ranging from 2004 through February 2012. The divergence that began in March 2012 and continued through August 2012 represented the expected new case volume. Additional dashboard plots confirmed that the distribution of new cases by event month was not due to missed events or to variance in other data sources such as enrollment, demographics, pharmacy, and opt-outs. Conclusions: From the point of view of HVH, the dashboard monitors a transition between case-identification methods and assures comparability over time. From the point of view of GH, the dashboard also serves as a method of monitoring the quality of the VDW data updates and alerting the GH local VDW infrastructure team to warnings about unexpected deviations over time. This process can be expanded to monitor other major elements of the VDW and adopted as a local quality improvement effort.