The Actuarial Standards Board has approved an exposure draft of a proposed revision of ASOP No. 45, "The Use of Health Status-Based Risk Adjustment Methodologies," with a comment deadline of September 1, 2026. The revision arrives at a critical juncture: CMS deferred its V28 risk model recalibration for Medicare Advantage 2027, states are expanding Medicaid managed care into behavioral health carve-ins where risk adjustment calibration is especially uncertain, and ACA individual market issuers face ongoing questions about how coding intensity variation distorts risk adjustment transfers. For any actuary certifying an ACA rate filing, building a Medicare Advantage bid, or setting Medicaid capitation rates, the revised standard creates new documentation and analysis obligations that reshape how risk adjustment transfer amounts are validated and how population morbidity assumptions are supported.
This article walks through the specific actuarial workflow of applying a health-status risk adjustment model in an ACA individual market rate filing and identifies where the revised ASOP 45 imposes new requirements at each step.
What Changed: Scope Narrows to Model Use
The current ASOP 45, adopted in 2011, covers actuaries "when performing actuarial services with respect to the use of health status-based risk adjustment methodologies." The revised exposure draft sharpens this boundary considerably. It explicitly states that the standard applies to selecting or using risk adjustment models to quantify morbidity differences across organizations, populations, programs, or time periods. It does not apply to actuaries when designing, reviewing, or modifying health status-based risk adjustment models, including recalibration.
The transmittal memorandum explains the rationale: "The task force intentionally drafted this proposed ASOP to apply to the 'use of' health status based risk adjustment methodologies, as opposed to the creation of these methodologies." Many commercial risk adjustment platforms are built by non-actuarial teams at firms like Verisk, Milliman, and Cotiviti. The standard now reflects that reality by focusing its requirements on the actuary who applies the output, not the team that builds the model.
This distinction matters for consulting actuaries who both build and apply proprietary risk adjustment models. Under the revised standard, the same actuary working on model design in one engagement and model application in another must document which role they occupy for each assignment. The standard's analysis and documentation requirements attach only when the actuary is wearing the "model user" hat, but the boundary between use and design is not always obvious, particularly when an actuary recalibrates off-the-shelf model coefficients for a local population before applying them in a rate filing.
Step 1: Model Selection and Population Fit
An ACA individual market rate filing begins with model selection. The actuary must choose among the HHS-HCC concurrent model (the federal default for ACA risk adjustment transfers), a prospective model variant, or in some cases a carrier-proprietary model layered on top of the federal risk scores. Section 3.1 of the revised ASOP 45 requires the actuary to consider whether the model was designed for the specific estimation goal, whether the population and program to which the model is applied "is reasonably consistent with those used to develop the model," and what behavioral incentives the risk adjustment system may create.
The population-fit requirement carries real teeth in the current ACA environment. Consider an issuer whose enrolled population shifted significantly during the 2023-2024 Medicaid unwinding. Redetermination churn pushed roughly 4.6 million people from Medicaid into the individual market between April 2023 and December 2024, according to CMS tracking data. These enrollees often carried limited prior commercial claims history and different coding patterns than the population on which the HHS-HCC model was originally calibrated. The revised ASOP 45 requires the actuary to evaluate whether that calibration population is sufficiently similar to the issuer's current book and, where material differences exist, to document the assessment and any adjustments made.
From tracking ACA rate filings across multiple states, this population-fit question is especially acute for issuers with dominant Medicaid managed care enrollment in their service area. When a large share of new ACA enrollees come from Medicaid, their diagnostic coding patterns may reflect Medicaid fee schedules and provider networks rather than the commercial environment the HHS-HCC model assumes. The revised standard does not prescribe a specific test for "sufficiently similar," but it does require the actuary to document the analysis, which creates an implicit expectation that the work was performed.
Step 2: Data Period Alignment
Once the model is selected, the actuary applies it to claims data from a base period to project risk scores for the rating period. Section 3.1.5 of the revised ASOP addresses timing considerations that were underspecified in the 2011 version. The actuary must evaluate differences between the model's development data and the application data regarding collection periods, estimation periods, and claims run-out timing.
In practice, this means the actuary filing a 2027 ACA rate may use 2024 claims experience as the base period while applying an HHS-HCC model calibrated using 2020, 2021, and 2022 EDGE data (per the 2026 NBPP final rule recalibration). The gap between calibration vintage and application vintage creates a data-period alignment issue. Coding intensity drifted meaningfully between 2020 (a COVID-disrupted year with suppressed elective utilization and diagnostic coding) and 2024 (a year of normalized utilization and post-pandemic coding catch-up). The revised standard now requires the actuary to assess whether that drift affects the model's applicability and to document the assessment.
A concrete illustration: suppose the 2020-2022 calibration period understates HCC prevalence in respiratory categories because COVID-era visit suppression reduced the frequency of COPD, asthma, and other chronic respiratory diagnoses in the claims record. If the actuary applies that model to 2024 data where respiratory coding has rebounded to pre-pandemic levels, the model may systematically underpredict respiratory risk scores. The revised ASOP 45 expects the actuary to identify this kind of temporal mismatch and either adjust for it or document why no adjustment is necessary.
Step 3: Coding Intensity and Sensitivity Analysis
Section 3.2.3 of the revised standard addresses coding variation directly: "The actuary should consider the impact of differences in levels of diagnosis codes or services coded across organizations and time periods." This section intersects with one of the most contentious issues in health risk adjustment: how much does variation in provider coding intensity across issuers distort the risk adjustment transfer mechanism?
In the ACA individual market, the HHS risk adjustment program transfers funds from issuers with lower-than-average risk scores to issuers with higher-than-average risk scores within each state market. An issuer whose providers code more completely, capturing more HCCs per member, will generate higher risk scores and receive larger inbound transfers, even if its enrollees' actual health status is identical to those of a competitor with less aggressive coding. CMS applies a coding intensity adjustment at the plan level through its Risk Adjustment Data Validation (RADV) program, but the adjustment operates with a multi-year lag and a $10,000 de minimis threshold that leaves smaller issuers' coding differences largely uncorrected.
The revised ASOP 45 does not require the actuary to quantify the exact magnitude of coding distortion. However, it does require consideration of how coding variation affects model output. In practical terms, this means the actuary should perform sensitivity analysis testing how changes in coding intensity affect projected risk scores and the resulting transfer payment. A reasonable approach: test how a +/- 5% coding intensity shift (measured by HCC prevalence rates relative to the statewide average) would affect the issuer's projected risk score and the net transfer position. If a 5% swing in coding intensity moves the issuer from a net transfer payer to a net transfer receiver, the risk adjustment assumptions underlying the rate filing merit closer scrutiny and more thorough documentation.
This guidance also intersects with the Medicare Advantage chart review controversy. CMS finalized the exclusion of unlinked chart review diagnoses from CY 2027 MA risk scores, a change with a -1.53% payment impact. While ACA risk adjustment does not use the same chart review mechanism, the underlying coding intensity dynamics are analogous. ACA issuers with prospective coding programs capture HCCs differently from those relying on retrospective claims review, and the revised standard puts the burden on the actuary to assess how those differences affect their risk-adjusted pricing.
Step 4: Credibility Blending for Small Enrollee Pools
Section 3.4 of the revised ASOP 45 addresses individuals with limited claims data: new enrollees, newborns, and individuals transitioning from uninsured status. The standard requires the actuary to "consider the minimum criteria required for an individual to be included in the risk adjustment analysis" and offers several acceptable approaches, including assigning age-gender-only factors, using average risk scores from the scored population, or excluding limited-data members entirely.
For ACA issuers, the limited-data population can be substantial. In states with high rates of coverage churn, 15% to 25% of an issuer's January enrollment may consist of new members with no prior claims history on which to generate HCC-based risk scores. These members receive a default demographic risk factor under the HHS model, but the actuary still must make a morbidity assumption for the rate filing that accounts for whether the default factor adequately predicts this subpopulation's actual claims experience.
This is where ASOP 45's requirements intersect with ASOP No. 25 (Credibility Procedures). When an issuer's enrollee pool is too small for the risk adjustment model to produce stable individual risk scores, the actuary must blend model-predicted morbidity with manual rate assumptions using a credibility-weighting framework. ASOP 25 requires the actuary to select a credibility method (limited fluctuation or Buhlmann/least-squares), identify the complement of credibility, and document the rationale for both choices.
A worked example clarifies the interaction. Suppose an issuer has 3,000 individual market members in a rural service area. Of these, 2,200 have a full year of prior claims data generating HCC-based risk scores, and 800 are new enrollees with demographic-only factors. The actuary projects a plan-level risk score of 1.08 based on the 2,200 scored members. For the 800 new members, the actuary must decide whether the demographic default factor (say, 0.95 for the age-gender mix) is credible on its own or should be blended with the 1.08 from the scored population.
Under a limited fluctuation approach, the actuary tests whether 800 member-months is sufficient to meet a chosen credibility standard (commonly 90% probability of being within 5% of the true mean). For most health insurance applications, 800 lives falls well short of full credibility. The actuary then assigns partial credibility to the demographic factor and supplements it with the complement, often the scored population's risk score or a broader market average. ASOP 25 requires documenting the credibility standard, the partial credibility percentage, the complement, and the rationale for selecting each. ASOP 45 now explicitly reinforces these obligations in the risk adjustment context, creating a dual-standard documentation trail that the actuary must satisfy.
The Cross-Standard Documentation Web
The revised ASOP 45 does not operate in isolation. It creates explicit cross-references to three other ASOPs that collectively govern the risk adjustment workflow:
- ASOP No. 23 (Data Quality): When the actuary identifies data limitations, such as coding variation, incomplete run-out, or population mismatch, ASOP 23 requires assessment of data quality and documentation of any limitations that could materially affect the analysis. The revised ASOP 45's data-alignment requirements are a specific application of ASOP 23's general data quality framework.
- ASOP No. 25 (Credibility Procedures): As discussed above, the credibility intersection governs how the actuary blends risk-adjusted and non-risk-adjusted morbidity assumptions. Any rate filing that relies on risk adjustment must navigate both standards simultaneously.
- ASOP No. 56 (Modeling): Because a risk adjustment model is, fundamentally, a model, ASOP 56's requirements for model identification, assumption documentation, output validation, and limitation disclosure apply in addition to ASOP 45's risk-adjustment-specific requirements. The actuary must document the model's intended purpose, known limitations, and any modifications made for the specific application.
Navigating this web of interrelated standards is not new, but the revised ASOP 45 makes the connections more explicit. An actuary certifying an ACA rate filing who applies the HHS-HCC model must now demonstrate compliance with ASOP 45 (model use), ASOP 23 (data quality), ASOP 25 (credibility, if applicable), and ASOP 56 (modeling), with each standard's documentation requirements layering onto the others. The practical effect is a thicker actuarial memorandum with more detailed discussion of why the selected model, data, and assumptions are appropriate for the specific filing.
Broader Implications: MA Bids and Medicaid Capitation
While this article has focused on the ACA individual market workflow, the revised ASOP 45 applies equally to Medicare Advantage bid construction and Medicaid managed care capitation rate development. For MA actuaries, the standard's coding intensity guidance arrives just as CMS deferred the V28 recalibration and finalized unlinked chart review exclusions worth -1.53% of plan payment. MA actuaries now face a standard that requires them to document their assessment of how chart review practices, health risk assessments, and audio-only telehealth coding affect the risk scores underlying their bids.
For Medicaid managed care, the timing coincides with the ASB's parallel exposure draft of ASOP No. 49 (Medicaid Managed Care Capitation Rates), also open for comment through September 1, 2026. States expanding behavioral health carve-ins face a particular challenge: the CDPS+Rx risk adjustment models commonly used in Medicaid were not originally calibrated on populations with integrated behavioral health coverage, and the revised ASOP 45 requires the certifying actuary to assess whether the calibration population is sufficiently similar to the population being served under the expanded benefit package.
This is part of what actuary.info has called "the busiest standard-setting cycle in a generation." ASOP Nos. 12, 20, 30, 36, 39, 41, 45, and 49 are all in active revision simultaneously, and the ASOP 12 revision addressing algorithmic bias in risk classification adds another layer for actuaries whose risk adjustment models incorporate machine learning components. Practitioners should review the ASOP 45 exposure draft before the September 1 comment deadline and assess how the revised requirements affect their current documentation practices for risk-adjusted rate filings.