Registry Says Fail; the Chart Says Otherwise

BP control (<140/90): The most recent BP in the measurement year is 138/84 - controlled. The registry is likely sampling the first BP of the year, or the worst, or failing to apply the measure's specified 'most recent' logic. Verify the measure specification before acting: most BP-control measures use the most recent reading in the measurement period, not an average or a first reading.

Tobacco screening: Screening was performed and documented at the 11/7 visit. The registry shows a fail because 'never' was recorded in narrative text rather than a structured field, or mapped to a non-reportable value. This is a data-extraction failure, not a care failure.

A1c poor control: A1c 7.4% two days before the 11/7 visit is controlled. The patient should not be in the poor-control numerator. Note that the June refusal note is irrelevant to this measure because a result exists for the year.

Conclusion: two and possibly three of the three reported results are wrong, and the practice is about to make outreach calls based on bad data. Before any patient contact, validate the registry against source charts - the registry is a hypothesis, not a finding.

Do not assume refusal creates an exclusion. Denominator exclusions are measure-specific and narrowly written. For diabetes A1c measures, patient refusal is generally not the pathway out - absence of a result often places the patient in the poor-control numerator, which is the opposite of what the practice wants. Some measures do permit a 'patient reason' exclusion for a specific service, but that is the exception, and it must be confirmed against the current specification for that measure and reporting year.

Correct documentation of a refusal:

• Structured, not free text in a telephone note.
• Date, who documented it, what was offered, the patient's stated reason, and whether an alternative was offered (e.g., point-of-care vs. lab draw, fingerstick).
• Documented at the time of the conversation, by the person who had it.
• Available to the reporting team for audit defense - a refusal record supports the clinical narrative and quality-improvement follow-up, even when it does not change the measure result.

Practical answer to the practice: the June refusal is a care-gap signal, not a reporting solution. The right response is a standing order or pre-visit lab protocol so the result arrives before the visit.

State the line clearly and without moralizing. Exclusions must reflect the patient's actual clinical status and must be documented in the record by the clinician. Adding a diagnosis to move a patient out of a denominator is measure manipulation - the quality-reporting analogue of upcoding, and it is equally indefensible in a payer audit.

Then redirect to legitimate levers, and be specific:

1. Data capture, not patient selection. Move tobacco status, BP, and lab values into structured fields so the registry reads what was actually done. This alone likely resolves two of the three failures in this chart.
2. Pre-visit planning. MA-driven standing orders for A1c and BP so the values exist before the provider walks in.
3. Gap lists by provider, reviewed monthly, with the caveat that registry output must be validated against source charts before it is used - as this case demonstrates.
4. Close the refusal loop with a documented alternative plan rather than a one-time note.

Framing: the practice's problem in this chart is not that patients don't qualify - it is that the work was done and the data didn't reach the registry. Fix the plumbing before touching the denominator.