Docsity
Docsity

Prepare for your exams
Prepare for your exams

Study with the several resources on Docsity


Earn points to download
Earn points to download

Earn points by helping other students or get them with a premium plan


Guidelines and tips
Guidelines and tips

HCC Reclassification: Payment HCCs, HCC Groups, and Clinical Classification Changes, Study notes of Statistics

The process of examining HCC reconfigurations, payment HCC designation, HCC Groups, and clinical classification changes using 2016 data. discussions on potential changes such as adding new HCCs, splitting existing HCCs, and modifying HCC hierarchies. The document also covers revisions to the clinical classification, including payment and non-payment HCCs, clinical hierarchies, HCC Groups, and severe illness interactions.

Typology: Study notes

2021/2022

Uploaded on 09/27/2022

uzmaan
uzmaan 🇺🇸

3.1

(9)

216 documents

1 / 51

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
DEPARTMENT OF HEALTH & HUMAN SERVICES
Centers for Medicare & Medicaid Services
Center for Consumer Information & Insurance Oversight
200 Independence Avenue SW
Washington, DC 20201
Potential Updates to HHS-HCCs for the HHS-operated Risk Adjustment Program
June 17, 2019
EXECUTIVE SUMMARY ...................................................................................................................... 1
Table ES.1 Overview of Potential Changes across Adult, Child and Infant Models ........................ 3
Table ES.2 Summary of V06a HHS-HCC Risk Adjustment Model Changes .................................. 4
1.0 Overview ........................................................................................................................................ 6
1.1 Purpose and Structure of this Paper ........................................................................................... 6
1.2 Purpose of ICD-10 Reclassification ........................................................................................... 6
1.3 Other Revisions to the Clinical Classification ........................................................................... 7
1.4 Outside the Scope of this Clinical Reclassification ................................................................... 8
2.0 Brief Overview of HHS-HCC Diagnostic Classification and Criteria .......................................... 8
2.1 Criteria ....................................................................................................................................... 8
2.2 Overall Framework .................................................................................................................... 9
2.3 Elements ..................................................................................................................................... 9
3.0 Review and Reclassification Process ........................................................................................... 13
4.0 Potential V06a HHS-HCC Classification Updates ...................................................................... 16
4.1 Summary .................................................................................................................................. 16
4.2 Discussion of Key Payment Model Changes ........................................................................... 17
4.2.1 New Payment HCCs and New or Revised HCC Groups ................................................. 17
4.2.2 Significantly Reconfigured HCCs or Significant Hierarchy Changes ............................. 36
4.2.3 Infant Categorical Model Changes ................................................................................... 45
5.0 Overall Impact of Potential V06a Changes ................................................................................. 47
6.0 Other Potential Risk Adjustment Model Changes ....................................................................... 49
6.1 Other Long-Term Potential Risk Adjustment Model Changes ................................................ 50
6.2 Next Steps ................................................................................................................................ 51
EXECUTIVE SUMMARY
The Department of Health and Human Services Hierarchical Condition Category (HHS-HCC)
diagnostic classification is the foundation of the HHS-operated risk adjustment program for the
individual and small group markets under section 1343 of the Patient Protection and Affordable Care
Act (PPACA). The HHS risk adjustment model uses patient diagnoses and demographic information,
in addition to enrollment duration and a limited number of drugs for adults, to predict plan liability for
medical and drug spending.
The current HHS-HCC clinical classification, in place since 2014, was based on International
Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes. In
October 2015, the U.S. implemented ICD-10-CM diagnosis codes. This paper considers potential
pf3
pf4
pf5
pf8
pf9
pfa
pfd
pfe
pff
pf12
pf13
pf14
pf15
pf16
pf17
pf18
pf19
pf1a
pf1b
pf1c
pf1d
pf1e
pf1f
pf20
pf21
pf22
pf23
pf24
pf25
pf26
pf27
pf28
pf29
pf2a
pf2b
pf2c
pf2d
pf2e
pf2f
pf30
pf31
pf32
pf33

Partial preview of the text

Download HCC Reclassification: Payment HCCs, HCC Groups, and Clinical Classification Changes and more Study notes Statistics in PDF only on Docsity!

DEPARTMENT OF HEALTH & HUMAN SERVICES Centers for Medicare & Medicaid Services Center for Consumer Information & Insurance Oversight 200 Independence Avenue SW Washington, DC 20201

Potential Updates to HHS-HCCs for the HHS-operated Risk Adjustment Program

June 17 , 2019

EXECUTIVE SUMMARY ...................................................................................................................... 1 Table ES.1 Overview of Potential Changes across Adult, Child and Infant Models ........................ 3 Table ES.2 Summary of V06a HHS-HCC Risk Adjustment Model Changes .................................. 4

1.0 Overview........................................................................................................................................ 6 1.1 Purpose and Structure of this Paper ........................................................................................... 6 1.2 Purpose of ICD-10 Reclassification........................................................................................... 6 1.3 Other Revisions to the Clinical Classification ........................................................................... 7 1.4 Outside the Scope of this Clinical Reclassification ................................................................... 8

2.0 Brief Overview of HHS-HCC Diagnostic Classification and Criteria .......................................... 8 2.1 Criteria ....................................................................................................................................... 8 2.2 Overall Framework .................................................................................................................... 9 2.3 Elements ..................................................................................................................................... 9

3.0 Review and Reclassification Process........................................................................................... 13

4.0 Potential V06a HHS-HCC Classification Updates ...................................................................... 16 4.1 Summary .................................................................................................................................. 16 4.2 Discussion of Key Payment Model Changes ........................................................................... 17 4.2.1 New Payment HCCs and New or Revised HCC Groups ................................................. 17 4.2.2 Significantly Reconfigured HCCs or Significant Hierarchy Changes ............................. 36 4.2.3 Infant Categorical Model Changes ................................................................................... 45

5.0 Overall Impact of Potential V06a Changes ................................................................................. 47

6.0 Other Potential Risk Adjustment Model Changes ....................................................................... 49 6.1 Other Long-Term Potential Risk Adjustment Model Changes ................................................ 50 6.2 Next Steps ................................................................................................................................ 51

EXECUTIVE SUMMARY

The Department of Health and Human Services Hierarchical Condition Category (HHS-HCC) diagnostic classification is the foundation of the HHS-operated risk adjustment program for the individual and small group markets under section 1343 of the Patient Protection and Affordable Care Act (PPACA). The HHS risk adjustment model uses patient diagnoses and demographic information, in addition to enrollment duration and a limited number of drugs for adults, to predict plan liability for medical and drug spending.

The current HHS-HCC clinical classification, in place since 2014, was based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes. In October 2015, the U.S. implemented ICD-10-CM diagnosis codes. This paper considers potential

changes to the current HHS-HCC classification to better incorporate ICD-10 diagnosis codes. It is part of HHS’s continued assessment of modifications to its HHS-operated risk adjustment program for the individual and small group markets based on analysis of more recent data. In this paper, we describe our methodology for reviewing and restructuring the HHS-HCC classification to incorporate ICD- diagnosis codes, using the 2016 and 2017 benefit years masked enrollee-level External Data Gathering Environment (EDGE) claims data, which reflect the first two full years of ICD-10 diagnosis coding. This work was conducted by the Centers for Medicare and Medicaid Services (CMS) with our contractor, RTI International.

To conduct this reclassification analysis, we took the following the steps:

  1. Reviewed the current HHS-HCC full classification and risk adjustment model classification (V05), including an examination of disease groups with extensive ICD-10 classification changes, HCCs whose counts had changed considerably following ICD-10 implementation, clinical areas of interest (e.g., substance use disorders), and model under-prediction or over-prediction as identified by predictive ratios.

  2. Examined HCC reconfigurations, payment HCC designation, HCC Groups, and hierarchies to develop the preliminary regression analyses using 2016 data (V06).^1

  3. Conducted a series of clinical review calls to inform potential changes, discuss diagnosis and treatment of conditions, and review the potential classification changes.

  4. Reviewed the payment model and full classification regressions to compare frequencies and predicted incremental costs of HCCs.

  5. Repeated the preliminary regression analyses on 2017 data, reviewed regression results, and developed the new potential (V06a) HHS-HCC reclassification.^2

This paper focuses on discussing the rationale for potentially updating the classification of HHS-HCCs, including the details and reasoning for key potential HHS-HCC changes as a result of the analysis described above. Specifically, Section 1 of this paper provides the overview and purpose for updating the HHS-HCC diagnostic classification. Section 2 introduces the HHS-HCC diagnostic classification and identifies the criteria used in its formation and review. Section 3 describes the review and reclassification process. Section 4 presents the potential updates to the HHS-HCC classification referred to as V06a. This section includes an overview of changes, comparing V06a to the current V HHS-HCC classification, and discusses potential key changes that we are considering proposing for the HHS-HCC risk adjustment models in future rulemaking. Section 5 features summary statistics that summarize the overall impact of the current V05 to the potential V06a classification changes. Section 6 describes other potential model considerations (non-linear and count model specifications; and enrollment duration factors), and concludes with next steps.

(^1) Payment HCCs are those included in the HHS-HCC risk adjustment model. The full classification includes both payment and non-payment HCCs. HCC Groups refers to HCCs that are grouped together in the HHS-HCC risk adjustment model. (^2) To further clarify, V05 is the current classification model, V06 was our initial assessment of potential revisions to the classification model developed using the 2016 benefit year data, and V06a was our reassessment of potential revisions to the classification model that included 2017 benefit year data.

Table ES.2 Summary of V06a HHS-HCC Risk Adjustment Model Changes Condition Payment HCC Change

Summary of Changes

Payment HCC Changes Substance Use Disorders

+3 - Add 2 new HCCs for alcohol use disorders and 1 new HCC for lower severity drug use disorders.

  • Reconfigure drug dependence HCC to include drug use disorders with non-psychotic complications and a subset of drug poisoning (overdose) codes.
  • Impose a new combined hierarchy on drug use and alcohol use HCCs. Pregnancy +3 - Add 3 (ongoing) pregnancy-without-delivery HCCs, leaving them ungrouped in the adult model (to reflect differences in costs by level of complications) and grouping them in the child model (to address small sample sizes and unstable estimates).
  • Revise two existing pregnancy HCC Groups, separating out the ectopic/molar pregnancy HCC and the uncomplicated pregnancy-with- delivery HCC to better distinguish incremental costs. Diabetes: Type 1 +1 - Add diabetes type 1 additive HCC to the adult payment model to distinguish additional costs for diabetes type 1.
  • Remap hyperglycemia and hypoglycemia codes from the “chronic complications” HCC to the “without complication” HCC. Asthma +1 - Split current payment asthma HCC into two severity-specific HCCs.
  • Continue to group asthma HCCs with chronic obstructive pulmonary disease HCC in adult model and leave the 3 HCCs ungrouped to distinguish costs in child model. Fractures -
  • Delete a payment HCC (pathological fractures) to address a clinical distinction that may be inconsistently diagnosed/coded.
  • Reconfigure an existing payment HCC (hip fractures) to better distinguish fracture codes by site.
  • Add a new payment HCC (vertebral fractures) to better predict vertebral fractures, which may be indicative of chronic disease and frailty. Third Degree Burns and Major Skin Conditions

+2 - Reconfigure and add 2 HCCs (extensive third degree burns; major skin burns or conditions) to the payment model.

  • Impose a hierarchy.
  • Impose an a priori constraint^5 between extensive third degree burns and severe head injury in child model due to small sample size. Coma and Severe Head Injury

+1 - Add a new severe head injury HCC (represents a condition with ongoing care costs; similar to the inclusion of other injury HCCs).

  • Place in a hierarchy above the coma/brain compression HCC.
  • Impose an a priori constraint between extensive third degree burns and severe head injury in the child model due to small sample size. Traumatic Amputations +1 - Add a new HCC and reconfigure codes between the new HCC and current payment amputation status HCC to better distinguish early treatment and complication costs from long-term costs.
  • Impose a hierarchy with amputation status HCC.
  • Leave HCCs ungrouped in the adult model; group them in the child model for coefficient stability purposes due to small sample size. Narcolepsy and Cataplexy

+1 - Add HCC to both child and adult models because currently underpredicted and has associated treatment costs.

(^5) In a priori constraints, the HCC estimates are constrained to be equal to each other. These are applied to stabilize high cost estimates that may vary greatly due to small sample size.

Condition Payment HCC Change

Summary of Changes

Exudative Macular Degeneration

+1 - Add HCC to adult model because currently underpredicted; costs are primarily related to drug treatments. Congenital Heart Anomalies

new to adult

  • Add 3 underpredicted HCCs to adult model (already in the child and infant models). Group them in the adult model only. Changes in HCC Groups, Hierarchies Metabolic and Endocrine Disorders

N/A - Group HCCs 26 and 27 together in both the child and adult models to distinguish their significantly higher incremental costs from other HCCs (HCCs 28-30) previously in the full group (HCCs 26 and 27 are currently underpredicted in the models due to grouping).

  • Ungroup HCCs 29 and 30 in the adult model as they have adequate sample sizes and clinical and cost distinctions.
  • Group HCCs 28-29 in the child model due to small sample sizes, clinical similarity, and similar predicted costs.
  • Leave HCC 30 ungrouped in the child model because it is clinically distinct from HCCs 28-29. Necrotizing Fasciitis N/A - Ungroup the necrotizing fasciitis HCC (HCC 54) in the adult model to better predict higher incremental costs compared to HCC 55 (currently grouped with). Blood Disorders N/A - Revise groups in both adult and child models to move HCC 69 from its previous grouping, with HCCs 70 and 71, to the group with HCCs 67 and 68 to better reflect clinical severity and associated costs.
  • Reconfigure HCCs 69 and 71 based on clinical input. Mental Health N/A - Move delusional disorders/psychosis HCC above major depressive disorders (severe)/bipolar disorders HCC in the payment hierarchy and renumber the HCCs (i.e., HCCs 88 and 89 switch positions).
  • Relabel HCCs to align with ICD-10 categorizations. Cerebral Palsy and Spina Bifida

N/A - Refine hierarchies to exclude paralysis HCCs for enrollees with cerebral palsy HCCs, as ICD-10 coding guidelines prohibit these conditions from coding together.

  • Refine hierarchies to exclude hydrocephalus HCC for enrollees with spina bifida HCC for similar coding restriction purposes. Pancreatitis N/A - Reconfigure acute pancreatitis HCC to differentiate higher cost conditions.
  • Revise the hierarchy for pancreas transplant HCC to remove exclusion of pancreatitis HCCs because pancreas transplants are done primarily for diabetes and insulin conditions rather than pancreatitis. Liver N/A - Reconfigure codes in liver HCCs to reflect clinical distinctions.
  • Move acute liver failure HCC above chronic liver failure HCC in the hierarchy and renumber HCCs.

When ICD-10-CM was implemented in the U.S. on October 1, 2015, ICD-10 codes were cross- walked to ICD-9 codes and to the existing ICD-9-based HHS-HCC clinical classification.^9

One purpose for reclassifying HHS-HCCs as described in this paper is to update them to better incorporate coding changes made in the transition to ICD-10 diagnosis codes into the HHS-HCC models. This paper therefore considers potential changes to the current HHS-HCC classification for that purpose. We also used this opportunity to review and use the newly available 2016 and 2017 benefit years enrollee-level External Data Gathering Environment (EDGE) claims data^10 , which reflect the first two full years of ICD-10 diagnosis coding on claims. This allows us to evaluate potential changes to the HHS-HCC model classification on the population for which the models are targeted.

ICD-10 includes new clinical and classification concepts. The ICD-10 code set differs from the ICD-9 classification in four key aspects:

  1. Structure: ICD-10 codes are longer (7 characters versus 5) and use more alpha characters that allow for greater clinical detail and specificity (e.g., laterality) of codes;
  2. Volume: There are more than 70,000 ICD-10 codes as compared to approximately 14, ICD-9 codes;
  3. Multiple concepts within codes: ICD-10 has many more combination codes with two or more clinical concepts than ICD-9 did; and
  4. Clinical currency: Terminology and disease classifications have been updated to be consistent with current clinical practice.

1.3 Other Revisions to the Clinical Classification

In addition to analyzing current ICD-10 code mappings to HHS-HCCs, this reclassification examined other components of the clinical classification:

  1. Payment and non-payment HCCs;
  2. Certain clinical hierarchies (e.g., substance use disorder; pregnancy);
  3. HCC Groups^11 and a priori constraints on HCC coefficients; and
  4. Severe illness and other HCC interactions affected by potential changes.

(^9) The abbreviations ICD-10 and ICD-10-CM are used interchangeably in this paper to refer to the Tenth Revision diagnosis codes. Similarly, the abbreviations ICD-9 and ICD-9-CM are used interchangeably to refer to the Ninth Revision diagnosis codes. As a starting point for initial 2015 HHS-HCC crosswalks, we used the General Equivalence Mappings (GEMs) to backward map ICD-10 to ICD-9. GEMs were a tool developed by the Centers for Medicare & Medicaid Services (CMS) and the Centers for Disease Control and Prevention (CDC) to assist with the conversion of ICD-9 codes to ICD-10. (^10) In the 2018 Payment Notice, we finalized a policy to collect and use masked enrollee-level EDGE data to recalibrate the HHS risk adjustment models and to help update the HHS risk adjustment methodology. See 81 FR at 94101. (^11) Note on nomenclature: Because the terms group and groupings are used in multiple contexts throughout this paper, we will use HCC Groups when referring to HCCs that are grouped together in the HHS-HCC risk adjustment models. See Section 2.3 for a more detailed description of these HCC Groups.

1.4 Outside the Scope of this Clinical Reclassification

The scope of this reclassification review was limited to the HHS-HCCs. This reclassification work did not review the Prescription Drug Categories (RXCs) and RXC interactions.

2.0 Brief Overview of HHS-HCC Diagnostic Classification and Criteria

A diagnostic classification system provides the framework for developing a risk adjustment model that uses patient diagnoses and demographic information to predict medical service and drug spending. This section describes the HHS-HCC diagnostic classification, how the HHS-HCCs were selected and grouped for the HHS-HCC risk adjustment models, and other key components of the models.

2.1 Criteria

Determining which diagnosis codes should be included, how they should be grouped, and how the diagnostic groupings should interact for risk adjustment purposes was a critical step in the development of the HHS-HCC risk adjustment models. The following 10 principles, which were discussed in the proposed 2014 Payment Notice, guided the creation of the original HHS-HCC diagnostic classification system^12 and were used to develop the HCC classification system for the Medicare risk adjustment model.^13 These principles, which also guided the current reclassification, include:

Principle 1 — Diagnostic categories should be clinically meaningful.

Principle 2 — Diagnostic categories should predict medical (including drug) expenditures.

Principle 3 — Diagnostic categories that will affect payments should have adequate sample sizes to permit accurate and stable estimates of expenditures.

Principle 4 — In creating an individual’s clinical profile, hierarchies should be used to characterize the person’s illness level within each disease process, while the effects of unrelated disease processes accumulate.

Principle 5 — The diagnostic classification should encourage specific coding.

Principle 6 — The diagnostic classification should not reward coding proliferation.

Principle 7 — Providers should not be penalized for recording additional diagnoses (monotonicity).

Principle 8 — The classification system should be internally consistent (transitive).

Principle 9 — The diagnostic classification should assign all diagnosis codes (exhaustive classification).

(^12) See the HHS Notice of Benefit and Payment Parameters for 2014, Proposed Rule, 77 FR 73118 at 73128 (December 7, 2012). (^13) Report to Congress: Risk Adjustment in Medicare Advantage (December 2018) also discusses these principles in Section 2.3 under Principle for Risk Adjustment Models from Pages 14-16 and is available at: https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Downloads/RTC-Dec2018.pdf.

Because a single individual may be coded for zero, one, or more than one HCC, the HHS-HCC model can individually price millions of distinct clinical profiles. Thus, the model’s structure provides, and predicts from a detailed comprehensive clinical profile for each enrollee.

Figure 2.1: HCC Aggregations of Diagnosis Codes

Payment and Non-Payment HCCs

There are 267 HHS-HCCs in the current V05 HHS-HCC full diagnostic classification, among which 128 HHS-HCCs are used for the HHS risk adjustment adult, child and infant models. In addition to the established criteria based on the principles of risk adjustment, we utilize further criteria for selecting a subset of HCCs to distinguish costliness of plan enrollees. This set of payment HHS-HCCs is designed to support the success of the single risk pool and should:^15

  1. represent clinically significant medical conditions with significant costs for the target population;

(^15) These criteria are the same criteria for selection described in the discussion paper for the March 31, 2016, HHS-Operated Risk Adjustment Methodology Meeting (March 24, 2016). A copy of this discussion paper is available at: https://www.cms.gov/cciio/resources/forms-reports-and-other-resources/downloads/ra-march-31-white-paper-032416.pdf.

Diagnosis codes

Hierarchies Imposed

Diagnostic Groups (DXGs)

HHS-Condition Categories (HHS- CCs)

HHS-Hierarchical Condition Categories (HHS-HCCs)

Payment HHS-HCCs

Selection for HHS-HCC Payment Model

  1. contain sufficient sample size for stable results;
  2. exclude (or limit the impact of) diagnoses particularly subject to discretionary coding;
  3. exclude diagnoses that represent poor quality of care;
  4. identify chronic or systematic conditions that represent insurance risk selection or risk segmentation, rather than random acute events that represent insurance risk; and
  5. apply only to the appropriate model age group (infant, child, adult).

These criteria are intended to ensure that the payment HCCs target appropriate actuarial risk in the risk adjustment models.

HCC Groups and A Priori Constraints

To balance the competing goals of improving predictive power and limiting discretionary coding, as well as to create a risk adjustment model in which less experienced insurance carriers are not disadvantaged, a subset of payment HHS-HCCs are grouped into larger aggregate clusters, or HCC Groups. In HCC Groups, the HCC estimates are constrained to be equal to each other. We group payment HHS-HCCs for the following reasons:

  1. To reduce model complexity by limiting the number of effective payment HHS-HCCs;
  2. To avoid HHS-HCCs with low sample size and possibly unstable high-cost estimates;
  3. To limit diagnostic upcoding by severity within an HCC hierarchy; and
  4. To reduce additivity within disease groups (but not across disease groups) in order to decrease the sensitivity of the model to coding proliferation.

After grouping, the number of payment HHS-HCCs in the V05 risk adjustment model is effectively reduced from 128 to 104. From a single HCC Group, a person receives only one risk marker, i.e., the person receives a single indicator for the group if one or more of the constituent HCCs are present. The HHS-HCC model also incorporates a small number of “a priori constraints” (e.g., for organ transplants in the child model). HCC Group constraints and a priori constraints are applied in the initial phase of risk adjustment regression modeling. Other constraints may be applied in later stages depending on regression results. For example, HCCs may be constrained equal to each other if there is a hierarchy violation (a lower severity HCC has a higher estimate than a higher severity HCC in the same hierarchy). HCCs may be constrained to 0 if the estimates are negative. A priori constraints, on the other hand, are applied to stabilize estimates that might vary greatly due to small sample size.^16 These a priori constraints differ from the HCC Groups in how the corresponding estimates are counted. With a priori constraints, a person can have more than one indicated condition (each with the same magnitude coefficient) as long as the HCCs are not in the same hierarchy. With

(^16) For example, we previously finalized a constraint for the six transplant status HCC coefficients (other than kidney) in the child model, as the sample sizes of transplants are smaller in the child than the adult model. Because the levels and changes in the child transplant relative coefficients appeared to be dominated by random instability at the time, we believed the accuracy of the model was improved by constraining these coefficients. See the HHS Notice of Benefit and Payment Parameters for 2016, Final Rule, 80 FR 10749 (April 28, 2015).

disease interaction in the category. Finally, a hierarchy was imposed such that if an enrollee was in the high cost disease interaction category, the enrollee was excluded from the medium cost category. In sum, a person can have at most one disease interaction coefficient or incremental predicted expenditure in addition to the underlying HCCs. This constraint was imposed because clinical reasoning and empirical evidence indicated that a single one of the diagnostic markers sufficed to distinguish the most severely ill patients among those with the underlying interacted diagnosis.

Other Variables

To account for the observed underprediction of expenditures of people who are enrolled for less than the full year, beginning with the 2018 benefit year HHS risk adjustment methodology, the adult model included an enrollment duration factor as an additional risk factor. The enrollment duration factors are a series of 11 partial-year enrollment indicators. Variables ED_1, ED_2, etc., through ED_11 indicate people in the concurrent sample who are enrolled for one month, two months, and so on. Twelve months of enrollment (ED_12) is the reference group. The partial enrollment factors are not currently included in the child and infant models.

Beginning with the 2018 benefit year, the adult model also included 12 standalone prescription drug classes (RXCs), which were added as additional risk factors.^20 Ten of these RXCs (imputation/severity) enter the model specification separately as well as interacted with their related HCCs, while two of these RXCs are only included in interaction with their related HCCs (severity only). In the 2019 and 2020 benefit years, the severity only RXCs were dropped and the 10 imputation/severity prescription drug categories are included.

3.0 Review and Reclassification Process

This section describes the 2018-2019 review and reclassification process that we followed to develop the potential model changes outlined in this paper.

Analytic Files. For the initial analysis, separate analytic files were created using two data sources: 1) 2016 enrollee-level EDGE data and 2) 2016 Truven MarketScan ®^ Commercial Claims and Encounter data.^21 The enrollee-level EDGE data includes enrollees who are part of the individual and small group single risk pool in states where HHS is operating the risk adjustment program. Issuers in those states^22 are responsible for uploading enrollment, pharmaceutical claims, medical claims and supplemental diagnosis information to their respective EDGE servers. The enrollee-level EDGE data used in this analysis are masked person-level claims and enrollment data.^23 MarketScan ®^ data is from a national, proprietary database contributed to by large employers and health plans in all 50 states and the District of Columbia. MarketScan ®^ data includes claims and enrollment information on employees,

(^20) The 2018 benefit year adult risk adjustment models included 12 RXCs; however, starting with the 2019 benefit year, the two severity-only RXCs are removed from the adult risk adjustment models. See 83 FR at 16941. (^21) We used the 2016 enrollee-level EDGE and MarketScan ® (^) datasets because they were the most recently available data when we began the reclassification analysis. We created separate 2017 EDGE and MarketScan®^ analytic files once those datasets were available. (^22) In the 2016 benefit year, HHS operated the PPACA risk adjustment program in all states and the District of Columbia, except for Massachusetts. Beginning with the 2017 benefit year, HHS began operating the PPACA risk adjustment program in all 50 states and the District of Columbia. (^23) 45 C.F.R. § 153.720.

spouses, and dependents covered by employer-sponsored private health plans. The MarketScan ®^ data generally represent enrollees in the large self-insured employer plans.

For the analyses presented here, we focused on total expenditures, which are easier to interpret than plan liability when comparing relative costs of conditions and diseases. Using the 2016 analytic files, we calculated total person-level expenditure data for the full sample and three subpopulations— adults, children and infants. Results were presented at the ICD-10-CM diagnosis code level, DXG level, CC level, and HCC level. The total expenditure data included sample size, actual total expenditures, predicted total expenditures based on the current V05 HHS-HCC model, and predictive ratios.^24

Internal Review. We conducted a comprehensive review of the current HHS-HCC classification, encompassing these areas:

  • disease groups with extensive ICD-10 code classification changes (e.g., substance use disorder combination codes);
  • disease groups with new ICD-10 code severity breakouts that may enable better cost distinction by HCC (e.g., asthma);
  • ICD-10 codes whose initial mappings when based on GEMs suggested ICD-10 backward mappings to ICD-9 codes resulted in large changes in HCC counts (e.g., type 2 diabetes with hyperglycemia);
  • ICD-10 codes that include episode of care (initial, subsequent, sequela);
  • clinical areas of interest (e.g., substance use disorder, mental health, pregnancy); and
  • model underprediction or overprediction as identified by predictive ratios at the HCC, CC, DXG, or ICD-10 code level.

Analyses of Alternative Reclassifications. We examined various potential revisions at the HCC level. These revisions included: changing HCC payment status; removing, revising, or adding HCC Groups; changing HCC hierarchies; applying model restrictions (adult, child, infant) to new HCCs or removing model restrictions (e.g., congenital heart condition HCCs in the adult model); and revising infant model severity category assignments. Interim analyses were done, such as cross- tabulations of HCCs, to better understand the prevalence and overlap of conditions. We also considered and applied “within HCC” changes—splitting apart HCCs, deleting HCCs and redistributing their codes; and adding diagnoses to or removing diagnoses from payment and non- payment HCCs. Occasionally, for example when studying traumatic amputations, HCCs were temporarily split apart to better distinguish frequencies and costs by episode of care (initial encounter versus subsequent encounter).

For some disease groups, such as substance use disorders and pregnancy, we explored multiple model variations. In evaluating the options, we considered the predictive power, model complexity, and coding incentives. For substance use disorders, we tested different configurations to add new drug use disorder HCCs and alcohol use disorder HCCs to the HHS-HCC risk adjustment model—a single

(^24) Predictive ratios are calculated as the predicted expenditures divided by the actual expenditures. They reflect the accuracy of the model’s prediction for the given diagnostic group.

4.0 Potential V06a HHS-HCC Classification Updates

4.1 Summary

The V06a HHS-HCC reclassification resulted in these potential changes to the risk adjustment model used for payment:

14 HCCs were added as payment HCCs:

  • HCC 22 Type 1 Diabetes Mellitus, add-on to Diabetes HCCs 19-21 (adult model only)
  • HCC 83 Alcohol Use with Psychotic Complications^26
  • HCC 84 Alcohol Use Disorder, Moderate/Severe, or Alcohol Use with Specified Non-Psychotic Complications
  • HCC 85 Drug Use Disorder, Mild, Uncomplicated, Except Cannabis
  • HCC 123 Narcolepsy and Cataplexy (adult and child models only)
  • HCC 174 Exudative Macular Degeneration (adult model only)
  • HCC 210 (Ongoing) Pregnancy without Delivery with Major Complications (adult and child models only)
  • HCC 211 (Ongoing) Pregnancy without Delivery with Complications (adult and child models only)
  • HCC 212 (Ongoing) Pregnancy without Delivery with No or Minor Complications (adult and child models only)
  • HCC 218 Extensive Third Degree Burns
  • HCC 219 Major Skin Burn or Condition
  • HCC 223 Severe Head Injury
  • HCC 228 Vertebral Fractures without Spinal Cord Injury
  • HCC 234 Traumatic Amputations and Amputation Complications

1 existing payment HCC was split apart into 2 HCCs:

  • HCC 161.1 Severe Asthma
  • HCC 161.2 Asthma, Except Severe

3 existing payment HCCs were newly added to the adult model (i.e., adult model restriction removed; these HCCs are already in the child and infant models):

  • HCC 137 Hypoplastic Left Heart Syndrome and Other Severe Congenital Heart Disorders
  • HCC 138 Major Congenital Heart/Circulatory Disorders
  • HCC 139 Atrial and Ventricular Septal Defects, Patent Ductus Arteriosus, and Other Congenital Heart/Circulatory Disorders

1 payment HCC was deleted:

(^26) We continue to include all payment substance use disorder HCCs in the infant model. Although most infants who are affected by the mother’s substance use via placenta or breast milk are coded with a newborn-specific ICD-10 code from the P04 set, which in V06a maps to HCC 82, some infants are coded with substance use codes from the ICD-10 F10-F19 code sets, which map to V06a HCCs 81–86.4.

  • HCC 227 Pathological Fractures, Except of Vertebrae, Hip, or Humerus

In addition to the payment model designation changes, we also made code level changes to both payment and non-payment HCCs, hierarchy changes, and revisions to HCC Groups in the HHS- HCC risk adjustment model. Table 4-1 provides HCC, HCC Group, and ICD-10 code level summary statistics for the V05 and potential updates to V06a HHS-HCC classifications.

Table 4-1. Summary Statistics for the V05 and V06a HHS-HCC Classifications

Category V05 HHS-HCC Classification

V06a HHS-HCC Classification Total number of HCCs in full classification 267 274 Total number of payment HCCs in risk adjustment model

Adult model 115 132 Child model 119 131 Infant model 117 125 Number of HCC Groups in risk adjustment model Adult model 17 16 Child model 17 20 Effective number of payment HCCs in risk adjustment model (each HCC Group is counted as 1 HCC) Adult model 92 112 Child model 95 107 Total number of unique Fiscal Year (FY) 2019 ICD- 10 codes in full classification

Total number of unique FY2019 ICD-10 codes in risk adjustment model

Adult model 7,929 10, Child model 8,116 10, Infant model 6,812 7,

4.2 Discussion of Key Payment Model Changes

In this section, we discuss the potential substantive changes to the HHS-HCC risk adjustment models, including new payment HCCs, revised HCC Groups, and changes to HCC composition or hierarchies. Sections 4.2.1 and 4.2.2 focus more on how the V06a potential changes affect the adult and child models. Section 4.2.3 discusses potential V06a changes in the context of the infant categorical model.

4.2.1 New Payment HCCs and New or Revised HCC Groups

Substance Use Disorders

The current V05 HHS-HCC risk adjustment model includes two substance use HCCs:

HCC Group 9:

  • HCC 81 Drug Psychosis

HCCs 82 and 84, like their V05 counterparts, include drug and alcohol dependence (addiction) diagnoses. They also include drug and alcohol abuse and use with non-psychotic complication diagnoses (a smaller selected set of diagnoses in the case of alcohol). Additionally, V06a HCC 82, unlike V05 HCC 82, includes the diagnoses of drug poisoning (overdose) for select drugs, namely, opioids/narcotics (e.g., heroin, fentanyl, oxycodone), cocaine, hallucinogens (e.g., LSD), and psychostimulants (e.g., amphetamines, methamphetamines, MDMA/ecstasy).

Drug use disorder moderate/severe is included in V06a HCC 82 and drug use disorder, mild, is included in HCC 85.^28 Cannabis use disorder, mild, uncomplicated, is excluded from V06a HCC 85 because of concern about opportunities for gaming/overcoding of this diagnosis. Cannabis use disorder moderate/severe, or with complications, is included in HCC 82 (or HCC 81 in the presence of psychotic complications). The drug use, unspecified, uncomplicated code set is excluded from the HHS-HCC risk adjustment model because of lack of specificity of the ICD-10 codes as to the diagnosis or reason for treatment.

Alcohol use disorder moderate/severe is included in V06a HCC 84. Alcohol use disorder, mild, is excluded from the V06a model for empirical reasons (low estimated incremental predicted expenditures) and because of concern about opportunities for gaming/overcoding this diagnosis. The ICD-10 code for simple drunkenness (F10.129, alcohol abuse with intoxication, unspecified) and the alcohol poisoning (toxic effect of ethyl alcohol) codes are similarly excluded from the payment model because such diagnoses might be due to accidental overuse rather than indicative of higher than average actuarial risk enrollees.

Nicotine dependence (tobacco products) and non-psychoactive substance abuse are also excluded from the payment model.

Potential changes included in the V06a adult model and their impact:

V05 Adult Model V06a Adult Model HCC Count ParameterEstimate Group HCC Count ParameterEstimate Group 81 Drug Psychosis

3,180 $13, 473 G09 81 Drug Use with Psychotic Complications

3,180 $23,2 01

82 Drug Dependence

148,753 $13,473 G09 82 Drug Use Disorder, Moderate/Severe, or Drug Use with Non-Psychotic Complications

174,478 $12,

83 Alcohol Use with Psychotic Complications

5,224 $6,

84 Alcohol Use Disorder, Moderate/Severe, or Alcohol Use with Specified Non-Psychotic Complications

88,825 $5,

85 Drug Use Disorder, Mild, Uncomplicated, Except Cannabis

22,910 $3,

(^28) Note that ICD-10 maps both moderate and severe substance use disorders to the same codes, so it is not possible to distinguish moderate and severe use disorders of the same substance in ICD-10.

Potential changes included in the V06a child model and their impact:

Pregnancy

The current V05 HHS-HCC model includes six pregnancy HCCs, which are consolidated into two groups:

HCC Group 17:

  • HCC 203 Ectopic and Molar Pregnancy, Except with Renal Failure, Shock, or Embolism
  • HCC 204 Miscarriage with Complications
  • HCC 205 Miscarriage with No or Minor Complications

HCC Group 18:

  • HCC 207 Completed Pregnancy With Major Complications
  • HCC 208 Completed Pregnancy With Complications
  • HCC 209 Completed Pregnancy with No or Minor Complications

Pregnancy diagnosis codes differ between ICD-9 and ICD-10 classification systems in three key aspects: ICD- 9 ICD- 10 Episode of care Fifth character identified delivered or not

Separate codes for pregnancy by trimester; childbirth; and puerperium (the period of about six weeks after childbirth) Multiple gestation Separate codes Seventh character designation in subset of pregnancy or complications of delivery codes Ectopic or molar pregnancy complications

Subset of codes combined with miscarriage codes

Separate codes for complications following ectopic/molar pregnancy versus miscarriage

V05 Child Model V06a Child Model HCC Count ParameterEstimate Group HCC Count ParameterEstimate Group 81 Drug Psychosis

375 $25,060 G09 81 Drug Use with Psychotic Complications

356 $14,539 G09A

82 Drug Dependence

3,504 $25,060 G09 82 Drug Use Disorder, Moderate/Severe, or Drug Use with Non-Psychotic Complications

8,213 $14,539 G09A

83 Alcohol Use with Psychotic Complications

11 $2,461 G09B

84 Alcohol Use Disorder, Moderate/Severe, or Alcohol Use with Specified Non-Psychotic Complications

819 $2,461 G09B

85 Drug Use Disorder, Mild, Uncomplicated, Except Cannabis

1,932 $2,461 G09B