Running UDS reporting as a year-round data pipeline

Every federally funded health center files a Uniform Data System report to HRSA each winter as a condition of its Section 330 grant. HRSA's own estimate of what that report costs is in the Federal Register: 238 hours of staff time per Universal Report for the 2025 reporting year, across 1,538 reporting organizations, or about 366,000 hours nationally. HRSA estimates 184 hours for 2026 after cutting and merging reporting lines. The federal Office of Management and Budget (OMB) approved the 2026 reporting requirements in September. Either number is more than a month of a skilled person's time, spent on data the health center collected over the whole year.

Who files the report, and what it costs them

In 2024, 1,359 health center organizations ran more than 16,300 sites and served 32.4 million patients, per HRSA data compiled by KFF. Medicaid was 45% of their revenue and the federal grant 11%. Net margins went negative in 2024, at -2.1%. Health centers often run on tight budgets with small IT teams, Healthcare Innovation reported in 2024, and HRSA has said that automating UDS reporting will eventually end the need for staff to enter UDS data by hand.

If staff time costs $60 an hour including benefits, an illustrative figure and not a survey result, 238 hours is about $14,000 of labor per health center per year. The report is due by mid-February, and HRSA reviewers then send change requests until corrected reports are due, generally by March 31.

The spreadsheet approach

When the report is built by hand, the work is a set of EHR exports, a spreadsheet one person owns, and denominators counted by hand. Many health centers also lean on outside groups for their data, and in 2024 one health center network told Healthcare Innovation it helps about 35 health centers on different EHRs pull out the data for their UDS reports. Clinical quality measures follow published electronic specifications, called eCQMs, and those specifications change. For 2026, every measure with an eCQM moves to the CMS 2026 specification. One example: the breast cancer screening denominator widens from ages 52 to 74 to ages 42 to 74, which changes every health center's number whether or not anyone re-reads the spec.

The 2026 changes reach many parts of the report: HRSA removes or merges lines in Tables 4, 5, 6A, 8A, 9D and 9E, and its estimate of total reporting time across health centers falls by about 72,700 hours. HRSA removes eleven lines from Table 6A, the table of selected diagnoses and services, and adds eleven new ones, including Type 1 diabetes, autism screening, and health-related social needs. Table 8A's cost reporting is redesigned. A spreadsheet that reproduced last year's tables will not reproduce this year's, and the person who built last year's version may have left.

UDS+ is deferred

HRSA's modernization plan, UDS+, moves most UDS data from aggregate tables to de-identified patient-level records, submitted electronically in the FHIR R4 standard. A minimum UDS+ submission was required for the 2024 reporting year, then postponed before the deadline. Trade guides describe it as deferred.

In practice, the data behind each table will eventually be submitted as patient records rather than totals. Values a person now reads out of notes and tallies, such as an A1c result, will have to exist as discrete coded data. In 2024, a network that supports health centers told Healthcare Innovation that some health centers had not been able to pay for EHR upgrades, and an EHR vendor said the health centers that tested UDS+ first were seeing their data in a way they never had before. Health centers whose data is already structured and checked will have less to fix whenever HRSA sets a new UDS+ deadline.

Building it as a pipeline

A UDS report follows pipeline steps: extract, clean, compute, format, and reconcile. Built once as a pipeline, the report runs on a schedule, and at the February deadline staff only have to check the output.

A reliable design has four parts.

1. A staging schema. Every EHR export lands in the same tables with the same column meanings, regardless of vendor. The connector is the only piece that changes between health centers.
2. Measure logic pinned to the specification year. Each clinical measure is written against the exact eCQM version the manual cites, with the citation in the code. When the version changes, the change is one file, reviewed once.
3. Hand-verified test patients. Synthetic patients whose correct answers are known, covering the boundary cases: the patient whose result depends on the last reading in the period, the missing lab that counts as poor control, the age cutoffs, the visit just outside the period. The pipeline must pass every one before it touches real data.
4. Reconciliation against last year's submission. Before anyone trusts the pipeline on this year's numbers, it must reproduce last year's submitted tables from last year's data. Every disagreement is investigated and documented. A wrong number in a federal submission is the risk this check exists to prevent.

The same pipeline can produce quarterly quality dashboards, so the measures serve management as well as compliance. A health center that sees its diabetes control rate in April can act before the annual number is final.

Cost and requirements

A reporting pipeline of this shape fits inside a fixed-price Data Warehouse and Reporting engagement, $25,000 to $60,000 depending on the EHR and the number of measures, with written acceptance criteria that include the reconciliation check and the test-patient suite. Handoff includes documentation and training to run it without outside help.

UDS work uses patient-level data, so it is PHI work. It proceeds only under a signed business associate agreement with errors-and-omissions and cyber insurance in force, minimum-necessary access, and a statement in the contract of exactly which tools touch which data. A health center that wants an assessment before committing to a build can start with the two-week AI Readiness Sprint ($9,500), which uses no patient data and starts by finding where your staff's hours go.

Questions for the CEO

The federal estimate puts the report at a month of someone's time. Ask who does that work, what happens if they leave, and whether the 2026 changes will live in a spreadsheet only one person understands. If the answers are unclear, start the pipeline by October.