Checkpoint to review end-to-end integration results

Background and Strategy

While the first two integration checkpoints (see Checkpoint to review the interpretation of source data and Checkpoint to review the implementation of semantic mapping and data mastering) focused on the sausage-making, the final checkpoint concerns itself mainly with the resulting sausage.

For this checkpoint, an experienced reviewer should start by reviewing the final set of results produced by the primary integration team in the previous three tasks: Run downstream assets and review data diagnostic measure results, Validate aggregate measure results against a gold standard, and Perform case review on a random sample of patients. Fresh eyes on these results might generate some new leads.

If any suspicious results are found, they should be presented back to the primary implementation team by the reviewer, who should explain why the results seem suspicious and convey any intuition about where the problem -- if one indeed exists -- might be found. It is then the primary implementation team's responsibility to investigate and resolve the issue. (The rationale for this approach is two-fold: first, it serves to reduce the time needed by the reviewer, might be serving in this role for several implementations; and, second, it maximizes the learning by the primary implementation team, whose members are likely the expert reviewers of the future.)

Any reviewer findings, and the result of any subsequent investigations from the primary integration team, should be documented, ideally in the same narrative document used for the first two checkpoint reports.

Conclusion of this last checkpoint represents the final significant development milestone in the integration effort.