Quaility Research · Report 01 · June 2026
The EHR quality gap.
Every federally funded health center reports its EHR vendor and its clinical quality rates to the government each year. Nobody had put the two together — so we did. Across 1,510 health centers serving 33.8 million patients, the gaps between EHR vendors are real. And the spread within every vendor is the bigger story.
The headline
Same measure. Same year. 14-point spread by EHR.
Median cervical cancer screening rates at health centers, grouped by the EHR vendor each center reports to HRSA.1 Centers on Epic's direct platform post a median of 60.6% — centers on NextGen, 46.3%. Every vendor's centers trail the top performers on their own platform by far more than that.
Cervical cancer screening, median rate by reported EHR vendor, UDS 2024.12 Centers with measure denominator ≥ 30; each center weighted equally. Hosted instances (OCHIN) reported separately by HRSA. Methodology ↓
What the data shows
Five findings — including the ones that complicate the story
We publish the findings that cut against a tidy narrative alongside the ones that fit it. That's the point of working from public data: you can check us.
The vendor gaps are real, and they repeat across measures
On cervical cancer screening, centers running Epic directly post a median of 60.6% vs 48.1% for eClinicalWorks and 46.3% for NextGen.1 On colorectal screening the ordering is similar: Epic (direct) 46.8%, eClinicalWorks 39.2%, NextGen 35.2%.1 These aren't sampling artifacts — the smallest group here is 159 centers.
Colorectal cancer screening · median rate by EHR vendor · UDS 2024, den ≥ 30
Greenway leads here on a smaller base (55 centers) — we show it because it's in the data, not because it's convenient.
Size explains part of the gap — and we checked
Epic centers are nearly twice the size of eClinicalWorks centers (median 21,865 patients vs 11,966),1 and bigger centers screen better on every platform. So we stratified by size. The cervical screening gap between Epic (direct) and eClinicalWorks centers persists within every size band — 9.5 points among centers under 10k patients, 13.1 points at 10–25k, 6.7 points above 25k.
Cervical cancer screening · median by vendor, within center-size bands · total patients per center
| EHR vendor | <10k patients | 10–25k | >25k |
|---|---|---|---|
| Epic (direct) | 51.6% | 62.1% | 63.2% |
| Athenahealth | 46.1% | 54.0% | 57.4% |
| OCHIN Epic (hosted) | 45.5% | 53.0% | 59.2% |
| eClinicalWorks | 42.1% | 49.0% | 56.5% |
| NextGen | 37.4% | 48.5% | 55.0% |
Read it both ways: within every vendor, bigger centers do better — and within every size band, the vendor ordering largely holds.1 Size is a real confounder; it is not the whole explanation. We have not yet stratified by payer mix — that's in the full methodology as a known limitation.
No vendor wins everything — the ordering flips on behavioral health
On depression screening and follow-up, the table inverts: NextGen centers lead at a 78.8% median and Epic (direct) centers come last at 71.6%.1 And on blood pressure control, every vendor's centers land within 1.9 points of each other (65.8–67.7%) — the EHR barely registers. Screening measures are workflow-and-recall problems where tooling shows up in the data; chronic disease control is a care-delivery problem where it doesn't.
Depression screening & follow-up · median rate by EHR vendor · UDS 2024, den ≥ 30
The spread within every vendor dwarfs the gap between them
This is the finding that matters most. The 10th-to-90th-percentile spread among centers on the same EHR runs 33 to 47 points — three times the largest between-vendor gap.1 The top decile of eClinicalWorks centers screens 67.8% of eligible patients for cervical cancer — beating the median Epic center on the same measure with the same software the bottom decile struggles with.
Which means the EHR is not destiny. The centers that beat their platform's median aren't running different software — they're running different operations on top of it: registries that stay current, recall that actually reaches patients, and data work that doesn't wait for the next reporting deadline.
Cervical cancer screening · 10th percentile — median — 90th percentile, by vendor · 0–100%
42% of health centers run more than one system — and they are not worse off
628 of 1,510 centers report operating multiple EHRs or data systems.1 The intuitive story — fragmentation drags down quality — is one we expected to find. The data says otherwise: multi-system centers post slightly higher medians (colorectal 41.8% vs 39.3%; cervical 52.0% vs 50.6%). They skew larger and better-resourced, and many have already invested in the integration layer that makes multiple systems workable.
We're publishing the result that contradicted our prior, because the honest lesson is sharper than the convenient one: what separates centers isn't how many systems they run — it's whether anything intelligent sits on top of them.
Why it matters
Behind every percentage point: patients.
Apply each center's own reported rate to its own reported eligible population, and the screening gaps at America's health centers add up to millions of people overdue for cancer screening.12
Computed as Σ(1 − rate) × denominator across all centers reporting each measure, UDS 2024. These are care gaps as defined by the federal measures — each one is a person a phone call could reach.
What this means
Your EHR isn't destiny. It isn't neutral either.
Most EHRs in this data were architected before the AI era — and it shows most in the measures where recall, registries, and outreach do the work. But the centers in the top decile of every vendor prove the ceiling isn't the software. They've built what the EHR doesn't ship: a layer that knows which patients are due, reaches them, and learns from what happened.
That layer is what Quaility is. We don't replace your EHR — eClinicalWorks, Epic, Athenahealth, NextGen, or several at once. We sit on top of it: a universal patient record across every source, AI and staff-led outreach that closes the gaps this report counts, and analytics that tie every closed gap back to the call that closed it.
Methodology & sources
Check our work
Every figure on this page is computed directly from the named public files — no modeling, no imputation. We publish the cryptographic hash of each source file so you can verify you're looking at the same data we are.
Methodology
- Universe: all 1,510 Health Center Program awardees and look-alikes in the HRSA UDS 2024 public files, serving 33,821,196 patients.
- Measures: clinical quality measures as reported to HRSA on UDS Tables 6B and 7 (aligned to CMS eCQM specifications). Rates are each center's own reported numerator/denominator.
- Inclusion: a center counts toward a measure only if its denominator is ≥ 30 — small-denominator rates are statistical noise.
- Statistic: medians across centers, each center weighted equally (not patient-weighted). Group sizes (n) are printed on every chart.
- EHR vendor: as reported by each center on the UDS Health IT form. Hosted instances reported separately by HRSA stay separate — "Epic (direct)" excludes OCHIN-hosted Epic; "NextGen" excludes OSIS-hosted.
- Known limitations: self-reported data; correlation not causation; size-stratified but not yet payer-mix-stratified; one reporting year (2024). We'll extend to multi-year trends as we archive future UDS releases.
Sources
-
Health Resources & Services Administration (HRSA), Uniform
Data System 2024 — Health Center Program awardee file.
H80-2024.xlsx, retrieved 2026-06-12. SHA-256
b33988f1b336… -
HRSA, Uniform Data System 2024 — look-alike file.
LAL-2024.xlsx, retrieved 2026-06-12. SHA-256
d67463ff6800… - HRSA Data Downloads hub (where both files are published): data.hrsa.gov/data/download. HRSA/CMS data is US-government public domain.
Spot an error? Tell us at quaility.com/contact — corrections are published, not buried. This report is part of the Quaility Index, our free public record of American healthcare quality.
See it on your own data
Watch our AI call a patient.
Then imagine it calling
thousands of yours.
A 30-minute demo: live AI outreach, your quality measures on a unified record, and an honest conversation about what we'd build for your workflows.