ONC published USCDI v7 on July 23, 2026, adding 31 data elements to the baseline set of health data every certified electronic health record must be able to exchange nationwide. For most health IT teams, this is a certification and FHIR-mapping exercise. For medical coders and CDI staff, several of the new elements land directly on top of documentation gaps that already drive coding queries, denials, and risk-adjustment audit findings.
What ONC Actually Published
Standards Bulletin 2026-2 describes 30 new data elements and one significantly revised element, for 31 additions overall. Half the list is already in production: 15 of the new elements appear in specifications already required by the ONC Health IT Certification Program and are largely supported today through HL7 FHIR US Core. That leaves 16 elements representing genuinely new documentation and data-capture work.
Two new data classes appear in v7: Adverse Events and Healthcare Information Attributes. Notably, publication carries no certification deadline. USCDI v6 remains the actionable version, with an August 29, 2026 upgrade date under the SVAP Approved Standards for 2026, while v7 defines the eventual baseline coders and CDI teams should start preparing documentation workflows around now.
Adverse Events Becomes Its Own Data Class
The new Adverse Events class holds Adverse Event Condition, which captures a change in patient status that could be an unintended effect of a clinical intervention, and Adverse Event Outcome, documenting the resulting clinical status. ONC requires health IT to exchange both for at least medication and immunization adverse events.
That matters for coding because ICD-10-CM already asks coders to distinguish adverse effects from poisonings and to sequence T36-T50 codes against the causal drug or substance, a determination that today often sits buried in narrative progress notes. Once adverse event condition and outcome become structured, exchangeable fields, the causal link a coder currently has to query a physician to confirm arrives pre-structured instead.
Reason Not Performed Joins Healthcare Information Attributes
A Discrete Field for Discontinued Procedures
The second new class, Healthcare Information Attributes, gathers Diagnostic Report Date and Time and Reason Not Performed, while Indication and Performance Date and Time move in from other classes. Reason Not Performed gives coders a discrete, exchangeable field for documentation that currently lives inside operative or procedure notes and supports CPT modifiers such as 52 and 53, along with ICD-10-CM Z53 codes for procedures not carried out.
Diagnostic Imaging Reference Has Limits
The class also adds Diagnostic Imaging Reference, a computable pointer to imaging studies tied to an encounter. ONC is explicit that the pointer alone does not ensure image exchange from external PACS environments; realizing that benefit still depends on data-use agreements between the organizations hosting those systems.
Tobacco and Nicotine Product Use Retires the Smoking-Status Field
Smoking Status is being replaced outright by Tobacco and Nicotine Product Use, which expands coverage to e-cigarettes, vaping devices, nicotine pouches, heated tobacco products, and smokeless tobacco, matching the FDA’s definition. Systems that currently capture a binary or three-state smoking field will need to revisit both documentation templates and vocabulary bindings.
For coders, this pushes toward more granular product-type documentation to support the specific Z72.0 and F17 subcategory codes that Medicare Advantage risk-adjustment submissions increasingly expect, rather than a default tobacco-use code assigned from a vague chart note.
Expanded Coverage Data Has RCM Implications
Among the 15 already-supported elements, ONC includes a cluster of health insurance elements covering payer, plan, plan identifier, and coverage period. That data feeds directly into CMS’s Interoperability and Prior Authorization Final Rule, which requires payers to expose coverage information through standardized FHIR APIs. For revenue cycle teams, eligibility and coverage-period mismatches that generate denials today become identifiable earlier in the workflow, ideally before a claim is ever coded and submitted.
Practically, closing the gap between what USCDI v7 makes exchangeable and what a coding workflow can currently act on takes a few concrete steps:
- Map Adverse Event Condition and Outcome fields to the T36-T50 and related causal-sequencing logic coders already apply manually.
- Build Reason Not Performed into claim-scrubbing rules so discontinued-procedure modifiers and Z53 codes are flagged automatically, not caught on post-submission audit.
- Update tobacco/nicotine documentation templates and code-mapping tables before payers start requesting the more granular Z72.0 and F17 subcategories.
- Reconcile expanded payer, plan, and coverage-period data against claims pre-submission to catch eligibility mismatches early.
- Treat USCDI as a moving target: the comment period for USCDI v8 is already open through September 28, 2026.
Where Agentic AI Closes the Gap
Standards move faster than EHR certification calendars, and certification moves faster than most coding teams can manually rebuild logic. That mismatch is exactly where agentic AI earns its keep: continuously mapping new structured fields like adverse event outcomes or reason-not-performed data to the right codes, flagging encounters missing a newly required element before a claim goes out, and updating mapping logic as ONC’s expansion queue moves, rather than waiting for a certification deadline to force the work.
Medikode’s automated medical coding platform is built to absorb exactly this kind of standards churn, keeping code assignment and claim validation current as USCDI, CMS, and payer requirements evolve, so coding teams spend less time rebuilding templates and more time on the cases that actually need a human review.