A third of adults with lab-proven diabetes lacked a code, a prescription or both

Design

Retrospective national cohort, claims linked to laboratory results

Population

27,650 US adults with laboratory-confirmed diabetes and no prior diagnosis

Primary outcome

Type 2 diabetes diagnosis coding and drug initiation within 1 year

Effect

Both 66.6%; code only 19.7%; drug only 4.1%; neither 9.6%

Researchers linked US commercial and Medicare claims from 2017 to 2023 to laboratory results and followed 27,650 adults whose results met the threshold for diabetes and who had no prior diagnosis of diabetes or prediabetes. Over the following year, 66.6% received both a diagnosis code and a glucose-lowering prescription, 19.7% a code only, 4.1% a prescription only, and 9.6% neither.

Higher HbA1c made coding and treatment progressively more likely compared with 6.5–6.9%. Greater comorbidity made both less likely, so the patients with the most competing problems were the ones whose diabetes went unrecorded.

Claims cannot see lifestyle advice, so some of the untreated group near 6.5% may have been managed sensibly without a drug. But a missing diagnosis code means the condition is invisible to recall systems, eye and foot screening, and the next clinician. The fix is administrative and cheap: when you see a diabetic result, confirm it, record it and decide on treatment, all within the same episode of care.

• Act on every HbA1c of 6.5% or more: confirm it, record the diagnosis, and document the treatment decision
• Audit your own laboratory results for diabetic values with no diagnosis recorded
• Watch most closely in patients with many comorbidities, who were least likely to be coded
• If you choose lifestyle management alone, write that down so it is not mistaken for an omission
• Make sure a recorded diagnosis triggers eye, foot and kidney screening

Why it matters

The patients most likely to be missed were the sickest, which is the opposite of where attention should go.

Don't overread it

Claims data cannot see lifestyle advice or documented decisions, so not every uncoded or untreated case was neglect.

The statistics, in plain English

Adding the code-only, prescription-only and neither groups gives about 33% with an incomplete record. These are proportions from a large real-world dataset, not a trial; they describe what happened, not why.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.