Continuous Glucose Monitoring
Help us take CGM data from sensor to endpoint
CDISC is forming a focused CGM Sub-team inside its Digital Health Technologies team: a multi-stakeholder group to carry continuous glucose monitoring examples from SDTM through analysis to endpoints, working alongside FDA's 2026 CGM technical specification.
1 · Background
The CDISC Digital Health Technologies project
Digital health technologies (wearables, sensors, apps and other connected devices) are generating data in clinical research that doesn't look like a CRF page or a lab transfer. CDISC, in partnership with the Digital Medicine Society (DiMe), runs a volunteer team working out how that data should be represented using CDISC standards so it is interoperable, comparable and submission-ready.
The team's output is the CDISC DHT Portal: worked examples that take a real device data stream and show the biomedical concepts, device identification and SDTM representation end to end, with considerations explaining each modeling choice.
Focus area
Key DHT concepts
A shared vocabulary and concept map for device-generated data.
Focus area
Device attributes
How sensors, readers and apps are identified and related.
Focus area
Digital endpoints
Linking data to the endpoints in DiMe's library of digital endpoints.
Focus area
Best practices
Applying CDISC and DiMe resources to DHT data consistently.
2 · Where we are
The CGM work so far
Two CGM examples were developed for the portal, built from de-identified data from two different CGM devices. They are the only DHT examples that carry individual sensor readings rather than device summaries, which makes them the best test case the team has for granular, high-frequency device data.
Select a stage to see what each example already covers and where the Sub-team will extend it.
Stage 1 · Device & concepts
Drafted
Biomedical concepts and device identification
The examples define the assessment concept “Glucose Measurement” and device concepts for the glucose sensor and the signal-processing device (reader or phone app), with Device Identifiers (DI) records and a proposed Related Devices (RELDEV) structure to represent a sensor plus app as one system.
Open question for the Sub-team to address: at what level should the sponsor device identifier sit (device type, brand, or individual sensor), and how should the UDI be carried?
Stage 2 · SDTM data
Drafted · revisiting
SDTM representation
Individual glucose readings are represented as findings in the Laboratory Test Results (LB) domain, with timestamps, specimen (interstitial fluid) and the device that produced each value. The two examples show how data exported from different systems ends up in the same structure.
Revisiting for the Sub-team: test and method terminology, timestamp precision, dataset splitting, SI versus conventional units, and how to represent gaps and non-wear periods.
Stage 3 · Analysis (ADaM)
Sub-team scope
Analysis datasets
Not yet built. The FDA specification describes an ADCGM analysis dataset, handling of missing epochs, and paired comparisons of CGM values with lab or glucose-meter readings. The Sub-team will design CDISC-conformant ADaM structures that meet those expectations and document the derivations from SDTM.
Stage 4 · Endpoints
Sub-team scope
Endpoints and analysis concepts
Not yet built. Time in range is the most common CGM endpoint in DiMe's library of digital endpoints; others include time below range, mean glucose and glycemic variability. The Sub-team will generalize these into reusable analysis concepts and show the full trace from reading to endpoint.
3 · Why now
FDA published a CGM technical specification
In May 2026, partway through the portal's first public review, FDA released Submitting Continuous Glucose Monitoring Data in Clinical Trials: a technical specification with SDTM and ADaM expectations for CGM data, appendix datasets, and clear signals about what the agency wants to see.
CDISC's Medical Devices, SDS and ADaM teams reviewed the technical specification. Some recommendations appear inconsistent with current CDISC standards; others set expectations beyond current sponsor practice. CDISC presented these findings to the FDA and is awaiting a response. Rather than publish examples that conflict with the specification, the team withdrew the CGM examples from the current public review so they can be reworked properly.
- May 2026FDA publishes the CGM technical specification.
- June to July 2026CDISC Medical Devices, SDS and ADaM teams review it; a consolidated one-pager of comments is prepared.
- July 23, 2026CDISC presents its review at the CDISC–FDA Technical Meeting.
- August 27, 2026Duke-Margolis workshop on digitally derived endpoints reinforces FDA's priorities: full timestamps, missing-data handling and traceability from raw readings to summaries.
- September 2026CGM examples withdrawn from the second public review and returned to development. CGM Sub-team proposed.
The questions the Sub-team will work on
1. Granularity: every reading, not just summaries
The specification expects epoch-level data (each individual glucose reading) with full timestamps, so readings can be assigned unambiguously to analysis windows. Many sponsors today receive only device-generated summaries and rarely identify individual sensors.
2. Device identification and the UDI
The specification suggests populating the sponsor device identifier with the sensor's UDI. A CGM is really two devices (sensor plus reader or app), each with its own UDI, and the UDI is a long, opaque string. CDISC's view is that the UDI belongs in the Device Identifiers dataset and that the combined system needs a sponsor-level identifier. The right level of identification is still open.
3. Missing data and non-wear
FDA wants to know when a sensor was not providing data and why: warm-up periods, sensor failures, gaps between sensors, and periods when nothing was worn. From device data alone the reason is often unknowable, and CDISC has no established representation for non-wear. This is the hardest and most reusable problem on the list.
4. Terminology
The specification uses the test "Glucose" and the method "CGM". CDISC terminology treats the CGM result as Plasma Equivalent Glucose (the specimen is interstitial fluid, converted), and treats CGM as a device, not a method. Alignment is needed in both directions.
5. From SDTM to analysis
The specification includes an analysis dataset (ADCGM), paired comparisons of CGM values against lab or glucose-meter values, and handling of missing epochs. Time in range is the dominant endpoint. None of this exists yet in the CDISC examples, and it is where the Sub-team's analytical expertise matters most.
4 · The ask
The CGM Sub-team: a multi-stakeholder working group
A Sub-team is a small, focused group of subject-matter experts attached to the DHT team. It owns one topic, works between the DHT team's meetings every two weeks (and meets on its own when useful), and brings modeling questions back to the wider team and CDISC governance. This CGM Sub-team will be the inaugural Sub-team.
Its remit is to take the CGM examples from data representation through to analysis and endpoints, resolve the open questions set out in the FDA tech spec section, and build a consistent, standards-conformant way of submitting CGM data that works alongside the FDA specification. We already have CGM analysis expertise from the Jaeb Center for Health Research and sponsor experience from the existing team. We need the rest of the picture.
Who we're looking for
SDTM and Medical Devices implementers
You've mapped device or lab data into SDTM and know the SDTMIG for Medical Devices.
ADaM and analysis experts
You build analysis datasets or define derived variables, and want to shape ADCGM-style structures.
CGM data analysts and statisticians
You analyze CGM data in trials or research: time in range, glycemic variability, accuracy comparisons.
Sponsors running CGM studies
Your organization collects CGM data in trials and has views on what is practical to collect and submit.
Device manufacturers and data providers
You know what the sensor and app actually export, and what can be shared under a data transfer agreement.
Clinical, regulatory and endpoint specialists
Diabetes clinicians, regulatory scientists, patient-research organizations and endpoint developers.
What the commitment looks like
Every two weeks
DHT team meeting (1 hour, Mondays, 10:00 a.m. US Eastern). Sub-team members attend when CGM is on the agenda.
2–3 weeks
Typical turnaround on a Sub-team task: a review, a draft dataset, a worked analysis, a written recommendation.
6 months+
Suggested minimum involvement, starting with a baseline review of the FDA specification and CDISC's comments.
Your choice
Contribute as a doer (building parts of the example) or a reviewer (checking and advising). Both are needed.
CDISC volunteering
Three levels of volunteer commitment
Level 1 · Leadership
Industry Co-Lead
Up to 20% of your time
- Actively works with the CDISC lead to plan and progress the project to publication
- Plans and attends team and monthly all-hands meetings
- Provides expertise in the development of the standard
Level 2 · Doing
Active Contributor
3–4 hours a week
- Attends team and monthly all-hands meetings
- Provides expertise in the development of the standard
- Participates in the internal and public review of the standard
Level 3 · Checking
Reviewer
About 2 hours a month during development
- 4–8 hours to complete internal and public review of deliverables
- Attends monthly all-hands meetings
Meeting cadence varies by project: the DHT team meets every two weeks, with Sub-teams working in between.
5 · Next step
Get involved: join the CDISC DHT team
- Choose your commitment level. Industry Co-Lead, Active Contributor or Reviewer. See what each level involves
- Open the DHT project page. Go to the CDISC Digital Health Technologies page and open the Volunteer area. CDISC members and non-members are welcome; however, you will need a free cdiscID to access the volunteering page.
- Sign up and mention CGM. Complete the volunteer sign-up, check the DHT – CGM Sub-team box and describe why you are interested in the CGM Sub-team. We'll then send you information, including a questionnaire about your capacity to be involved and your areas of interest.
- Join the kickoff. The Sub-team's first sessions walk through the FDA specification and CDISC's comments, then agree on the work plan.
Please forward this page to colleagues and partners who work with CGM data.