ICD-10-CM for substance use disorder is denser than it looks. The F10 to F19 block covers ten substance classes, each with its own remission, intoxication, withdrawal, and induced-disorder structure. The F19 (other psychoactive substances) bucket handles everything that does not fit cleanly into the other nine. A terminology engine that holds up under this load has to keep all of it queryable and current.
This is the four terminology engines that come up most in 2026 substance use treatment stacks. For FHIR resources for behavioral health teams the broader catalog has the rest of the picture.
For the higher-level brief on behavioral-health terminology services, the complete guide to FHIR terminology services for behavioral health in 2026 sets the context.
What the Brief Looks Like for SUD Coding
The expectations break into four pieces:
- Host the full F10 to F19 block with the modifiers (use, dependence, abuse, in remission) cleanly accessible.
- Surface the induced-disorder codes (anxiety induced, mood induced, psychotic induced) as related concepts during charting.
- Maintain a `$translate` map to DSM-5 substance-related disorders.
- Hold up under the lookup load that medication-assisted treatment generates, which often hits the server several times per encounter.
Engines that load ICD-10-CM as a flat list satisfy the first capability. The other three are where the work hides.
The 4 Terminology Engines Worth Knowing
- Termbox. A commercial terminology server with curated SUD subsets and a DSM-5 to ICD-10-CM `$translate` map covering the substance-related diagnoses.
- Ontoserver. A commercial server with strong general performance and an extensible model for SUD-specific value sets.
- HAPI FHIR Terminology Module. The open-source workhorse. Reliable lookup and expand once the content is loaded; curation of SUD-specific subsets is on the team.
- Smile Digital Health Terminology. A commercial offering that bundles content updates and an SLA, useful for programs that want one vendor for FHIR services.
What Tips the Choice in SUD Programs
Three concrete factors decide:
- Annual update cadence. ICD-10-CM revises every October; engines that ship updates as part of the contract save a manual content step.
- Cross-walk to DSM-5. Programs that chart in both systems benefit from a maintained `$translate` map for SUD diagnoses.
- 42 CFR Part 2 boundary. The terminology layer rarely enforces Part 2 directly, but engines that allow tagging Part 2-protected ValueSets reduce downstream policy work.
For the DSM-5 chart side, Top 5 FHIR terminology servers for DSM-5 coding in 2026 covers the diagnostic angle. For the hosting question (and the cost-shape trade-off that matters more for small programs), hosted vs self-hosted terminology servers for behavioral health practices walks through the deployment choices.
How to Pilot Cleanly
Pick the five most common SUD diagnoses (alcohol use disorder severe, opioid use disorder severe with remission, cannabis use disorder moderate, stimulant use disorder unspecified, polysubstance use). Run `$lookup`, `$expand` against the SUD ValueSet, and `$translate` to DSM-5 for each. Engines that handle all five on the first try are the ones worth a procurement call. Engines that need a content patch for any of them are not really SUD-ready yet.
Sources
- Confidentiality of SUD Patient Records final rule - Federal Register rule, SAMHSA + HHS OCR, 2024
- 42 CFR Part 2 current text - Federal regulation, US Government Publishing Office, 2025
- Using 42 CFR Part 2 revisions to integrate SUD treatment information into EHR - Peer-reviewed study, JAMIA via PMC, 2024