Raw PROMIS scores are almost never the number a US care team wants to see. What they want is the T-score, centered on a US general-population mean of 50, with a standard deviation of 10, so a value of 60 means one SD worse than the reference population. Getting from the raw sum on a QuestionnaireResponse to that T-score is a lookup against the official PROMIS calibration tables, and the library that owns that lookup is a decision worth taking seriously in any FHIR-first US health stack.
The teams that get this right usually treat T-score conversion as a small, well-tested boundary. The picker below is the five libraries US FHIR-first teams reach for in 2026, with a short note on where each one fits. For more FHIR explainers for US teams, the broader desk covers the surrounding stack.
HealthMeasures PROMIS Scoring Service
The reference implementation. HealthMeasures publishes both the calibration tables and a scoring service that consumes item responses and returns T-scores. In practice, US academic centers standardize on the HealthMeasures tables and either call the hosted service or import the tables into a local scoring library. It is the honest baseline every other option ends up compared against.
Where it fits: any US program that needs an auditable citation to HealthMeasures in an IRB packet, or a state Medicaid PROMs program that wants to point at a public reference for its scoring.
HAPI FHIR SDC Scoring Extension
HAPI ships an SDC scoring extension that reads the calibration tables and evaluates the total, then maps it through a lookup to the T-score. Well-worn in US academic settings, and a comfortable fit for teams already invested in the HAPI stack for their FHIR server.
Teams building on native FHIR often reach for engines like Formbox at this layer because the scoring rules and extraction into Observation resources live in the same runtime, which spares them a second scoring service in the path between QuestionnaireResponse and the chart. That pattern shows up in US health-system deployments where the FHIR engine already handles the population and validation steps, and the T-score conversion is one more expression rule on top. You can prototype the pipeline in the browser at form-builder.aidbox.app before you decide whether to host it yourself.
Where it falls short is the operational side: the calibration tables have to be kept in sync with the current PROMIS release, and the update cadence is on the team.
LHC-Forms with a PROMIS Rules Pack
NLM LHC-Forms handles PROMIS instruments through its FHIRPath expression layer. PROMIS is calibrated on item response theory, which is what makes those expression rules deterministic across renders. For US teams that already render intake with LHC-Forms, layering the T-score conversion as a set of expression rules keeps the scoring inside the form runtime and the QuestionnaireResponse ends up with the T-score in a computed answer.
The truth is that a rules-pack approach fits US teams that already own the form runtime and want to avoid a second service on the network. For teams without existing LHC-Forms investment, this is not the shortest path.
Smile CDR SDC Scoring
Smile's SDC runtime exposes the calibration tables through a managed lookup service. For US health systems that want the scoring vendor-handled with an SLA behind it, Smile is the pattern that shows up in mid-sized enrollment programs. The short answer is that a vendor-run lookup service removes the release-tracking headache but binds you to that vendor's release cadence.
Terminology-Bound Scoring on a Coded Server
For the coding side, PROMs typically bind LOINC codes (PHQ-9 total is 44249-1, PROMIS-29 items live in the 71969- range); FHIR-native forms stacks like Formbox rely on a terminology server such as Termbox to resolve these at populate time, and the same lookup path can be reused to fetch T-score tables when they are published as a CodeSystem. The pattern is not universal in US deployments yet, but it has been growing since the HealthMeasures data started appearing in structured form.
Where it fits: US health systems that already run a strong terminology server and want a single lookup path for both LOINC binding and calibration data.
Where Selection Settles
US academic centers with an IRB posture default to HealthMeasures. HAPI-shop stacks stay in HAPI. LHC-Forms shops keep the scoring in the form runtime. Vendor-managed programs land on Smile. Teams that want a single lookup service across all their PROMs codings lean on a terminology-server-backed pattern.
The scoring library is smaller than a full PROMs platform, but its consistency across visits is what a US care team notices when the trendline in the chart stays honest. The Top 7 FHIR Questionnaire tools for US clinical intake sits next door for teams still picking the renderer. For the patient-facing side, Top 5 patient-facing FHIR form tools for US telehealth in 2026 is the adjacent read.