The Measure Ahead

Quality measurement is moving upstream. This summer's developments from NCQA and CMS do not eliminate medical record retrieval, but they steadily narrow the circumstances in which late manual recovery can compensate for weak clinical data. The implication for quality leaders is operational rather than clinical: every rate you report is the output of a supply chain — source systems, interfaces, terminology mappings, validation rules, and accountable owners — and that supply chain is the part of the operation that rarely has a single owner.

Three decisions matter now. Which of your high-value rates still depend on late manual recovery? Which structured clinical feeds would have to replace that recovery? And who owns data quality before the information reaches the measure engine?

1. One Member, Eleven Supply Chains

Quality measurement is usually taught as a list of measures. It is more accurately a set of parallel data supply chains, each with its own format, its own owner, and its own characteristic failure. A single rate may depend on several of them at once. Here is the map most quality teams have never actually drawn:

Most root-cause analyses stop at the measure and never reach the row.

2. The Transition Is Already Underway

It is tempting to file digital measurement under "coming soon." NCQA's MY 2026 trending memo says otherwise. Three measures are in their first ECDS reporting year right now:

• Lead Screening in Children (LSC-E). Both the administrative and hybrid collection methods were removed. NCQA flags trending with caution for Medicaid — the product line that leaned hardest on chart collection for this measure.
• Statin Therapy for Patients With Cardiovascular Disease (SPC-E). The administrative method was removed, sex-specific age bands eliminated, the ASCVD lookback expanded, and the I-SNP/LTI exclusion removed. NCQA's determination is a break in trending across all product lines.
• Statin Therapy for Patients With Diabetes (SPD-E). Parallel changes, plus members with an ASCVD diagnosis now excluded from the denominator. Also a break in trending.

A break in trending means your MY 2026 rates for those measures cannot be compared to prior years — the baseline resets. For lead screening, the removal of hybrid collection means the rate is now precisely what your data feeds say it is, with no manual recovery available. Auditors will continue to benchmark every measure regardless of trending determination, and a movement that outruns NCQA's predictions will be treated as organization-specific.

These three rates are no longer abstraction outcomes. They are data integration outcomes — and MY 2026 is already in progress.

3. QRS Puts Hybrid-Dependent Measures on a Transition Clock

CMS published the Final 2026 Call Letter for the Quality Rating System and QHP Enrollee Experience Survey in June, finalizing measure-set changes that begin with the 2027 ratings year:

• Asthma Medication Ratio (AMR) leaves the QRS measure set in 2027, aligning with the measure steward's retirement. CMS continues to collect and score it for the 2026 ratings year.
• Medical Assistance With Smoking and Tobacco Use Cessation (MSC) also leaves in 2027, after collection and scoring in 2026.
• Childhood Immunization Status (CIS-E) and Immunizations for Adolescents (IMA-E) remain in the measure set, but change status. CMS proposed removing both and reversed course after comment. They are required through the 2026 ratings year; beginning in 2027, submission becomes optional through IDSS and the measures are no longer scored. CMS will continue to publish performance information for transparency.
• Two replacements enter collection in 2027 — a tobacco use screening and cessation intervention measure that expands the eligible population to include adolescents and leverages electronic clinical data, and a follow-up measure for acute and urgent care visits for asthma. Scoring begins in the 2028 ratings year at the earliest, with benchmark and performance information published through the QRS Proof Sheets during the first collection year.

Separately, CMS published anticipated timelines for ending hybrid reporting on the four hybrid measures remaining in the QRS set. These are transition plans, not finalized reporting-method changes — each will come through a future Call Letter.

Source: Final 2026 Call Letter, Exhibit 4. CMS will continue to allow hybrid reporting for the 2027 ratings year.

Why it matters

The strategic implication is not that chart retrieval disappears in 2027. It is that Exchange plans now have a defined, published window in which to identify and strengthen the structured clinical sources that must eventually replace hybrid recovery. PPC, CBP, WCC and GSD are precisely the measures that still rely materially on medical record review. Removing hybrid does not lower the bar; it moves the work upstream, onto whether blood pressure readings, weights, BMI percentiles and glycemic results arrive as structured, coded, plausible data. A BMI percentile documented in a pediatric note is an abstraction win today and a mapping problem in 2030.

Note the pattern across both HEDIS® and QRS: new measures are collected for a year before they are scored. That is a deliberate grace period, and it is the cheapest look you will ever get at your own data quality.

4. Data Quality Moves Upstream — Across FHIR® and Every Other Clinical Feed

What changed

NCQA is not eliminating primary source verification in MY 2027. It is preparing to move away from PSV as a universal requirement for organizations meeting criteria to be defined in a forthcoming roadmap, while retaining targeted, risk-based verification and the HEDIS Compliance Audit™. That roadmap is being written now: NCQA's Health Plan Data Quality Convening runs three virtual sessions from August through November 2026, with participation limited to one primary representative per organization, by application.

Alongside it, NCQA's HEDIS® Data Quality Specifications move standardized checks earlier in the data lifecycle — over 100 defined rules evaluating clinical data across dimensions including completeness, conformance and plausibility, spanning labs, medications, procedures, encounters, conditions, immunizations and demographics, with guidance for validating data both in real time and at rest.

Why it matters

Continuous monitoring rewards a different capability than seasonal recovery does: intake QA, rejection management, and governance documentation that holds up year-round rather than once, in April. Plans with weak intake QA will find that continuous monitoring makes their problems continuously visible.

This is where FHIR® transformation stops being an interoperability slide and becomes measure work — but it is one part of a broader operating model, not the whole of it. Claims, pharmacy, laboratory, registry, HL7® and survey feeds all remain material. A CCD or CCDA is a container, not a data source; the clinically meaningful content inside is frequently narrative text and locally-coded results. Turning that into a conformant, correctly-bound resource is the difference between a lab result that closes a gap and a lab result that sits in a rejection log. In our client work, unmapped local laboratory codes are one of the most common causes we encounter for a real clinical result never reaching a numerator.

Your move

• Map the actual shapes. For each supplemental source, document what physically arrives — HL7® v2 message, CCDA, flat file, proprietary extract, FHIR® bundle — not what the contract says arrives.
• Test conformance before volume. A source delivering ten million non-conformant records is worse than one delivering two million clean ones, because the first creates the appearance of coverage.
• Move the checks upstream. Completeness, conformance and plausibility belong at intake, before data reaches reporting or analytics.

5. Digital Health Accreditation Makes Data Exchange a Contracting Issue

NCQA launched Digital Health Engagement Accreditation on June 24 — a three-year accreditation for digital tools and applications, built from Core Standards plus two optional modules, Health Assessment and Digitally Enabled Interventions. It carries NCQA-defined standardized measures alongside organization-defined measures submitted on a standardized template. The program drew more than 1,400 public comments, and a learning collaborative tested goal setting and goal attainment as measures of meaningful engagement.

The most useful finding in NCQA's announcement is not about accreditation at all. From the learning collaborative: data structure is highly variable across digital health vendors — in the types of data collected, how those data are exchanged and stored, and how organizations use AI to manage them. NCQA's stated conclusions include that organizations must invest in structured data capture and improved data accessibility for quality reporting, and that the industry needs to align on minimum data expectations.

Many plans now run several digital point solutions across diabetes, behavioral health, maternity, musculoskeletal and remote monitoring. Each generates clinical and engagement data, and much of that data does not yet arrive in a form the measure engine can consume — which means the plan can pay twice: once for the program, and again for the care gap the program closed but could not document. Many organizations have not yet quantified that gap.

When a vendor category has a common quality framework, "what does your data export look like, and in what standard" becomes a procurement question rather than an implementation surprise.

Your move

• Add a data-exchange exhibit to every digital health contract at renewal covering format, cadence, terminology standard, patient matching approach, provenance, rejection handling, and who owns remediation.
• Inventory which vendor programs are closing gaps that never reach your supplemental data feed.

6. Two Weeks in August: Comment on What Gets Retired

NCQA has opened an ad-hoc public comment period, August 3 through August 17, on the proposed retirement of select HEDIS® measures for Measurement Year 2027.

The stated criteria are worth reading as a forecast. A measure becomes a retirement candidate if it shows limited impact on quality improvement, low actionability, small populations, consistently high performance with little variation, overlap with other measures, significant reporting burden relative to value, or reduced alignment with evolving clinical guidelines. Read from a data perspective: "consistently high performance with little variation" describes measures a plan has already solved, which means the list often includes measures an organization has invested in for years. And "significant reporting burden relative to value" is the criterion that eventually reaches every measure still leaning on manual collection.

Two weeks is short, and ad-hoc windows do not get the calendar attention the annual cycle does. Assign one owner — not a committee — and pull the list of measures where your plan carries a state contract obligation, a P4P arrangement, or a delegated vendor SLA. Those are where a retirement creates a contractual mismatch, and that is the comment NCQA can act on.

7. From Non-Compliance Lists to Actionable Data Sets

Most quality data requests produce the same artifact: a list of members who are non-compliant with a measure. It is accurate, it is defensible, and it changes very little. Everyone already knew the rate was low. The list restates the rate at member level.

An actionable data set is designed backward from a decision. Take a blood pressure measure. The non-compliance list says four thousand members do not have a controlled reading. The actionable version segments the same population by why: members with no reading anywhere in the data; members whose reading exists in a source the plan does not currently ingest; members with a reading captured but not coded in a way the measure recognizes; members with a documented elevated reading and no follow-up. Those are four different problems with four different owners — data engineering, source contracting, provider documentation and coding, and care management — and only the last one is what most plans set out to fix.

Producing the data and moving the data are different disciplines. The first answers what happened. The second requires knowing what would have to change for the number to move.

Designing that second kind of data set takes two kinds of judgment in the same room, and most organizations already have both — just not in the same conversation. Clinical and quality staff know what would actually change a decision at the point of care, and can tell whether a segment is clinically coherent or a statistical artifact. Data and analytics staff know what the sources can genuinely support, where the joins are reliable, and what a field means at the moment it was written. Neither view is sufficient alone. A technically valid pipeline can still misstate clinical timing or intent, and a clinically elegant specification can quietly require data that does not exist.

The organizations that navigate the next several measurement years well will not necessarily add headcount to either group. They will invest in enough shared vocabulary that a conversation about a rate movement can happen once, in one room, rather than in six sequential meetings — and they will treat data set design as a deliverable in its own right, with the same rigor they apply to an intervention.

8. Monitoring, Not Yet Acting

• CQMC core set maintenance. The Core Quality Measures Collaborative has scheduled maintenance reviews through the fall — Obstetrics and Gynecology and Pediatrics in August, Behavioral Health in September, Orthopedics in October, with Cardiology, HIV/Hepatitis C, Medical Oncology and ACO/PCMH/Primary Care in the fall. These are scheduled reviews, not finalized changes, but core set alignment is the mechanism by which a change in one program propagates into others.
• QHP Enrollee Survey protocol. CMS is continuing to seek comment on extending the telephone dialing timeframe, adding a third reminder email, adding screener questions to reduce respondent burden, and combining the separate race and ethnicity questions to align with OMB Statistical Policy Directive No. 15. These remain under Paperwork Reduction Act review rather than finalized. CMS is also moving CAHPS® analysis from the SAS macro to the R version, and does not anticipate an impact on issuer or survey vendor submission processes.
• Exchange rating display. Exchanges will display 2026 quality rating information during the plan year 2027 individual market open enrollment period. CMS will provide State-based Exchanges with the State Ratings Data Files through HIOS-MQM in September 2026.
• Medicare Advantage. The CY 2027 final rule made the largest changes to the Star Ratings measure set in years, including the continuation of the historical reward factor and the arrival of an ECDS-reported depression screening and follow-up measure measured on 2027 data. That deserves its own treatment; we will take it up in a coming issue.

What to Do Between Now and MY 2027