Flexion Healthcare Quality Measure Calculation

Real-time digital quality measure calculation using the Next Generation Measures Calculator

Real-time Digital Quality Measure Calculation

Reusing the CMS-owned Next Generation Measures Calculator to return actionable results within days, not years

Flexion · Arbor Research Collaborative for Health

Measure calculator for ~70 converted dQMs

Quality Measurement at CMS Has a Timing Problem

≈ 2 years

Today

Care → data submission → feedback. By the time a provider learns a screening was missed, the measurement year has closed and the patient has moved on. The measures record what happened. They do not change what happens.

Days

With NGMC + EHR extraction

A running result inside the measurement period. The clinic sees which patients are missing a qualifying test while there is still time to order it. Performance improves because the care improved.

Concept of Operations — Three Zones

1

Provider Environment

EHR systems

Clinical data already captured in the course of care — encounters, diagnoses, labs, orders.

Data Extraction new build

Direct EHR integration · SMART on FHIR app · Bulk FHIR export · Data warehouse connector. Scoped to only the FHIR resources a measure requires.

2

Calculation Platform

Next Generation Measures Calculator exists today

Extracted from EQRS and generalized. A separation, not a rebuild — CMS already owns it and runs it at production scale.

  • Declarative ingestion
  • Data version control
  • Transformation
  • Calculation at scale

Provider-deployed open source or CMS-hosted shared service — one code base.

3

Actionable Results

Care Team

Rate plus the patient-level roster behind it — who is missing a test, in time to order it.

CMS / CCSQ

One reference implementation; every result reproducible and auditable.

Data-quality Feedback

Providers learn of their own coding gaps early.

1

Getting Consistent Data Out of Provider EHRs

The part that does not exist today

Direct Platform Integration

EHR-vendor supported extraction. Oracle/Cerner is supportive and recommends direct integration, the path already used with SSA and CMS CPI.

SMART on FHIR Application

Distributed through EHR vendor app galleries; a health system installs it without commissioning a custom integration project.

Bulk FHIR Export

Population-scale retrieval for a full measurement period.

Enterprise Data Warehouse Connector

Where an organization already consolidates clinical data.

Options are not mutually exclusive and can coexist per measure. Retrieval is scoped to only the FHIR resources a measure requires, and can be de-identified before calculation.

2

CMS Already Owns a Working Calculator — NGMC

Computes quality measures today for the End-Stage Renal Disease Quality Reporting System (EQRS). Architected as a decoupled measure calculation system from inception, so extraction is a separation rather than a rebuild.

Declarative Data Ingestion

Adding claims later is a configuration change, not a new system.

Data Version Control

Any disputed score reproduced against the exact inputs and logic version.

Data Transformation

Maps extracted resources to measure value sets; profiles coding gaps.

Measure Calculation at Scale

One published implementation per measure replaces many local interpretations.

Deployment — Both Paths from One Code Base

Provider-deployed, Open Source

Runs inside the organization's own environment. Patient-level data never leaves; results are readily available.

CMS-hosted Shared Service

For practices and facilities without a calculation stack of their own. Requires a security authorization — planned for, not assumed.

3

A Clinic Does Not Receive a Score Alone — It Receives a Roster

First Measure: CMS951FHIR — Kidney Health Evaluation for Patients with Diabetes

Did an adult aged 18–85 with diabetes receive both an eGFR and a uACR test this year? A gap a clinic can close in a single encounter.

Patient eGFR uACR Status
J. Public Mar 12 Order uACR
S. Rivera Order both
M. Chen Jan 30 Feb 4 Met

Illustrative roster. NGMC returns initial population, denominator, exclusions and numerator alongside it.

Why Kidney Health First

~37 million U.S. adults have chronic kidney disease; ~90% are unaware. Evaluation rates among Medicare patients with diabetes: 41.8%. CKD and ESRD spending exceeded $114B in 2016 — 23% of Medicare fee-for-service.

Which Measures This Fits

About half of converted dQMs compute in or near real time; another ~40% from longitudinal EHR data at episode end. Claims-risk-adjusted (CMS844, CMS529) and imaging-derived (CMS1074, 1206, 1056) measures come later.

Thin Vertical Slices — Scaled by Impact and Feasibility

1

Demonstrable Slice on Synthetic Data

CMS951FHIR end to end. Proves the decoupled NGMC, a targeted clinical schema and one complete calculation.

2

MVP at a Partner Provider

CMS951FHIR computed from live EHR data; rate and patient roster returned to that clinic. Demonstrable within weeks using a pilot clone of NGMC.

3

+5 Measures, Then Groups of ~10

Ordered by health impact of faster feedback against feasibility from data providers already capture. Re-prioritized after each group.

4

Later: Claims and Imaging

A claims source is a declarative ingestion change, bringing risk-adjusted measures into range. Imaging-derived values follow as translation matures.

Risks Surfaced Now, Not at Scale

Partial measure coverage · variable EHR coding quality · data-custody agreements · security authorization for a CMS-hosted path · adoption depends on returning something a clinic wants.

CCSQ Strategic Roadmap FY2025–2028

Goal 1

Prevention

The measures most amenable to rapid computation are predominantly screening measures.

Goal 2

Alignment

Aligning measures across settings and payers is easier when all parties compute from one reference implementation.

Goal 4

Interoperability

FHIR-based data flow and the transition to digital measures.

Goal 5

Less Redundant Reporting

One shared implementation of each dQM displaces the separate interpretation every vendor, registry and health system maintains today.

A result delivered two years late is a historical record; a result delivered this week is a chance to order the test.

Reusing a calculator CMS already owns is the fastest credible route to closing that interval.