Summary
A patient with diabetes and no HbA1c result during the measurement period counts in the numerator of CMS122 (poor control), not the denominator only. This trips up first-time implementers who assume "poor control > 9%" requires an actual result above 9%.
Detail
CMS122 measures the percentage of patients 18–75 with diabetes whose most recent HbA1c was greater than 9.0% during the measurement period. The measure treats a missing result the same as a poor-control result: if no HbA1c was performed, or the most recent result is absent, the patient is counted as poorly controlled.
The naive reading — "numerator = patients with a result > 9%" — undercounts, because it drops every patient who was never tested. The correct logic is closer to "numerator = patients whose most recent HbA1c is > 9% or who have no qualifying HbA1c result in the period."
Input that exposes the difference: a denominator-eligible diabetic patient with zero
Observation HbA1c results in the measurement period. Naive logic excludes them from the
numerator; the measure includes them.
Resolution
Resolved — this is the intended behavior of the published measure, not a defect. The numerator logic must count "missing result" as poor control. Implementations should include a no-result branch in the numerator definition and test it with a patient who has no HbA1c observation. This entry is seeded as a worked example of the corpus format; corrections and additional detail are welcome.
Repointed on 2026-08-01 from cms/diabetes-hba1c-poor-control version 13.0.0,
which was a hand-written package that disagreed with the upstream content on
steward, version and title. See cms122-2026-003.
The measure now also accepts a glucose management indicator result, so the
no-result branch this entry describes must consider both.