CMS122 / #001 — Diabetes: glycemic status > 9%

An inverse measure: lower performance is better. A patient is “performance met” when the most recent HbA1c (or GMI) result is greater than 9.0%, missing, or not performed during the measurement period.

Denominator

• Age 18–75 (see the age anchors).
• A qualifying encounter in the measurement period. The eCQM’s encounter value sets include nutrition-services visits and exclude virtual encounters.
• Diabetes active during the measurement period:
  • Condition.code in the Diabetes value set;
  • an encounter-diagnosis should have a recordedDate in the period;
  • a problem-list-item should be active, recurrent or relapsed.
  • If the EHR’s conditions don’t cover the period at all, apply the check leniently rather than excluding everyone.

Finding the A1c

• Search Observation for the HbA1c value set (and GMI where the specification includes it) in the measurement period. Keep a short list of well-known HbA1c LOINC codes as a floor, in case an EHR’s value set lookup comes back incomplete.
• Accept only final results (final, amended, corrected). An ordered test with no value is not a result.
• Read valueQuantity; strip a trailing %. Values like >14 or "7.2%" may arrive as valueString — decide how to parse them and do it consistently.
• Select the most recent day, then the lowest value on that day.
• Prefer coded results over uncoded ones on the same day, and reject uncoded values above 30 — they are on the glucose scale, not a percentage (see eClinicalWorks).
• Some EHRs label A1c only in text (code.text containing “hemoglobin A1c”). A text fallback recovers these from poorly coded sources.

Exclusions

• Hospice care in the measurement period.
• Palliative care.
• Age 66+ in long-term care.
• Age 66+ with frailty and either advanced illness (diagnosis in the measurement period or the year before) or a dispensed dementia medication.

Because the measure is inverse, keep the A1c values for excluded patients for audit, but report them as excluded.

QRDA I specifics

When the data arrives as QRDA I rather than FHIR:

• Combine each entry’s primary code and translation codes before matching value sets.
• Encounter diagnoses are not QDM Diagnosis entries.
• Date the result by the lab test’s own effectiveTime; use the nested result’s time only as a fallback.
• Read the value from the nested result observation, and treat values outside a plausible range (roughly 2–18%) as errors.