The data is not the problem

CGM sensors generate repeated glucose readings. The challenge is making the available data useful during clinical review.

Most CGM workflows fail because they present raw data and expect the clinician to do the pattern recognition. That is backwards. The system should identify patterns and present them for clinical judgment.

The three-screen rule

A usable CGM workflow fits in three screens:

Screen 1: Summary dashboard

The international consensus on time in range describes these CGM metrics for adults with type 1 or type 2 diabetes. Targets need individual clinical interpretation, including different considerations during pregnancy:

  • Time in range (70-180 mg/dL)
  • Time below range (< 70, < 54)
  • Time above range (> 180, > 250)
  • Glucose management indicator (GMI)
  • Coefficient of variation

One screen. No scrolling. Color-coded against targets.

Screen 2: Pattern analysis

Show the patterns the system detected:

  • Nocturnal hypoglycemia episodes
  • Postprandial spikes by meal period
  • Dawn phenomenon
  • Sensor gaps
  • Day-to-day variability trends

Each pattern should link to a recommended clinical action.

Screen 3: Comparison and note

Show current window versus prior window:

  • What improved
  • What worsened
  • What therapy changes happened in the interval

Then generate a structured interpretation note that the clinician can review, edit, and sign.

Automation boundaries

Some CGM interpretation is automatable. Pattern detection, metric calculation, trend comparison. These are deterministic and should be automated.

Clinical judgment is not automatable. "Should we increase basal insulin?" is a decision that depends on the patient, their goals, their adherence, their comorbidities. The system should propose. The clinician should decide.

The boundary matters. Cross it and you create liability. Respect it and you create leverage.

Documentation as a workflow output

The interpretation note should be a byproduct of the workflow, not a separate task. If the clinician reviews the summary, confirms the patterns, and adjusts the plan, the note should write itself from those actions.

This is where CPT 95251 becomes practical. The documentation exists because the workflow produced it, not because someone typed it after the fact.

Scaling across a panel

A single endocrinologist might manage 200+ patients on CGM. Manual review of each patient's raw data is not sustainable.

The scalable approach:

  • Automated triage: flag patients who need attention based on metrics thresholds
  • Batch review: group patients by urgency and pattern type
  • Exception-based workflow: only interrupt the clinician for exceptions

The goal is not "review every patient." The goal is "identify and act on every patient who needs a change."

The standard to hold

If your CGM workflow requires the clinician to manually download, interpret, and document each patient's data, you have not built a workflow. You have built a task list.

Frequently Asked Questions

How does Thyra handle CGM data from Dexcom or Libre safely?

Thyra includes a CGM interpretation and documentation workflow for physician review. Integration availability depends on your EHR and the workflow you want to use. We confirm the data source, permissions, and any read or write connection before a pilot begins. Confirm the supported device-data source and review its original information before using a draft.

Which tools integrate CGM data into daily workflows?

Switching between a chart and device portals can fragment CGM review. Thyra includes a CGM interpretation and documentation workflow for physician review. Integration availability depends on your EHR and the workflow you want to use. We confirm the data source, permissions, and any read or write connection before a pilot begins. Measure review time in your own practice rather than assuming a time saving.

Why does diabetes follow-up create continuous inbox work?

Diabetes management is inherently message-driven. CGM patients generate 288 glucose readings per day, and even well-managed patients produce questions about trends, dose adjustments, hypoglycemia episodes, and device supply refills between scheduled visits. Each message requires chart context (current regimen, last A1c, renal function, recent dose changes) that the inbox does not surface by default. This forces the physician to reconstruct the clinical picture for every message, turning each one into a miniature visit without the scaffolding of a visit.