Understanding the High Prevalence of Low-Prevalence Chronic Disease Combinations: Databases and Methods for Research. Centers for Medicare & Medicaid Services Grouper Systems


Hierarchical Condition Categories (HCC)


1 Pope GC, Kautter J, Ellis RP, et al. Risk Adjustment of Medicare Capitation Payments Using the CMS-HCC Model. Health Care Financ Rev. 2004;25(4):119–141.
2 Medpac. Issues for risk adjustment in Medicare Advantage. Chapter 4. Report to Congress. Medicare and Health Care Delivery System. June 2012; 94–111
3 Pope GC, Kautter J, Ingber MJ, et al. Evaluation of the CMS-HCC Risk Adjustment Model. Final Report. 2011. Contract No. HHSM-500-2005-000291I TO 0006
4 Sorace J, Wong HH, Worall C, et al. The Complexity of Diseases Combinations in the Medicare Population. Popul Health Manag. 2011; 14(4):161–6.

Sponsorship: Centers for Medicare & Medicaid Services
Description: The CMS HCC model was implemented in 2004 to adjust Medicare capitation payments to private health care plans for the health expenditure risk of their enrollees.1 CMS uses this model to risk adjust payments to health plans that participate in the Medicare Advantage program.2 This model uses enrollees’ demographics and medical conditions grouped into 70 categories to predict costliness.
Purpose/Use: To predict costliness of health plan enrollees.
Coding Family: International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)
Grouping Methodology: The HCC system begins by classifying over 14,000 ICD-9-CM diagnosis codes to 805 diagnostic cost groups (DCGs). Each diagnostic group represents a well-specified medical condition. Diagnostic groups are further aggregated into 189 condition categories. Condition categories represent a broad set of diseases that are related clinically and in terms of cost. Hierarchies are then imposed among condition categories, so that a patient is coded for only the most severe manifestation they have among related diseases. Out of 189 HCCs created, 70 are used in the CMS HCC model because they have been shown to strongly predict Part A and Part B medical expenditures.3 Approximately 3,000 ICD-9 codes are used in the final HCC model. This methodology results in three hierarchical levels of coding.
Level of Diagnosis Aggregation: ICD-9 codes are aggregated into 70 CMS-HCCs.
Number of Codes Included: 2,916 (for the 70 HCCs used in the CMS HCC model)
Number of Codes Excluded: 11,651 (for the 70 HCCs used in the CMS HCC model)
Methodological Considerations: The HCC model performs much better at predicting beneficiaries’ Medicare expenses relative to models based only on demographic characteristics. It has been shown to explain approximately 11 percent of the variation in beneficiaries’ costliness.2 However, the HCC model does not eliminate systematic prediction inaccuracies. The model is believed to leave approximately half or more of predictable variation unexplained. Further, for all enrollees with a given health condition, the HCC model adjusts payments by the same rate, which does not account for differences in severity. Additionally, it is calibrated using Medicare FFS data and must be re-calibrated if it were to be applied to other data sources.2
Related Data Sources: CMS Claims Data
Used in Disease Complexity Research: Yes4


Chronic Conditions Data Warehouse Algorithm


1 Buccaneer - Computer Systems & Services, Inc. Chronic Condition Data Warehouse: Medicare Administrative Data User Guide. Version 2.0. 2013. http://www.ccwdata.org/cs/groups/public/documents/document/ccw_userguide...

2 Centers for Medicare and Medicaid Services. Chronic Conditions among Medicare Beneficiaries, Chartbook, 2012 Edition. Baltimore, MD.2012.

3 Schneider KM, O’Donnell BE & Dean D. Prevalence and Multiple Chronic Conditions in the United States Medicare Population. Health and Quality of Life Outcomes. 2009; 7(82):1–11

Sponsorship: Centers for Medicare & Medicaid Services
Description: Under the Medicare Prescription Drug, Improvement and Modernization Act of 2003, CMS developed the Chronic Conditions Data Warehouse and corresponding CCW algorithm to support researchers in studying chronic illness in the Medicare population in the United States.1,2
Purpose/Use: To classify beneficiaries accordingly to one of 27 chronic condition categories for chronic conditions research.
Coding Family: International Classification of Diseases, Ninth Revision (ICD- 9)
Grouping Methodology:

The CCW algorithm assigns diagnosis codes to one of 27 pre- defined chronic conditions categories using a set of specific criteria:

1) ICD-9, CPT4 or HCPCS codes.

2) Claim types and counts.

3) Dates of service. Therefore, each chronic condition category is constructed based on diagnosis codes, but also on a reference period and the number of claims submitted for an individual.1

Level of Diagnosis Aggregation: ICD-9 codes are aggregated into 27 chronic condition categories.
Number of Codes Included: 581
Number of Codes Excluded: 13986
Methodological Considerations: Chronic condition categories in the CCW algorithm are designed to examine utilization patterns, which only serve as a proxy for identifying whether any given individual is receiving treatment for one of the actual conditions of interest. Chronic condition categories were also designed to be broad and more encompassing, rather than limiting. Therefore, researchers are expected to refine condition category specifications to fit particular research needs. As currently defined, the CCW condition categories are not necessarily designed to allow researchers to calculate straight population estimates without refinements.1
Related Data Sources: CMS Claims Data
Used in Disease Complexity Research: Yes3


Diagnosis Related Group (DRG)


1 Baker JJ. Medicare payment system for hospital inpatients: diagnosis-related groups. J Health Care Finance.2002;28(3):1–13.

2 Medpac. How Medicare pays for services: an overview. Report to Congress: Medicare Payment Policy. 2002. http://www.medpac.gov/publications/congressional_reports/mar02_ch1.pdf

3 Wynn BO, Beckett MK, Hilborne LH, Scott M, & Bahney B. Evaluation of Severity-Adjusted DRG Systems. 2007. WR-434-CMS. Prepared for the Centers for Medicare and Medicaid Services.

4 Hochlehnert, A et al. Psychiatric comorbidity in cardiovascular inpatients: Costs, net gain, and length of hospitalization. Journal of Psychosomatic Research. 2011; (70)2:135–139.

Sponsorship: Centers for Medicare & Medicaid Services & Yale School of Medicine, Division of Health Services Administration
Description: The Diagnosis Related Groups (DRGs) system provides a means by which to group patients according to diagnosis and healthcare resource use.1
Purpose/Use: Under the inpatient prospective payment system, diagnoses are categorized into DRGs. Each DRG is then assigned a payment weight, based on the average resources used to treat Medicare beneficiaries in that category. DRGs have also been used for risk adjustment, to study physician behavior and as a measure of healthcare quality.
Coding Family: International Classification of Diseases, Ninth Revision (ICD- 9)
Grouping Methodology: The DRG system is comprised of 538 categories. Patients are assigned to a DRG based upon principal diagnosis (ICD-9 codes), procedural codes, age, sex, discharge status, and the presence of comorbidities (up to 8 secondary diagnosis codes). These categories are designed to group patients together who are similar in terms of clinical conditions and who are expected to require similar amounts of hospital resources.2
Level of Diagnosis Aggregation: The DRG system groups patients into one of 538 categories.
Number of Codes Included: Not specified
Number of Codes Excluded: Not specified
Methodological Considerations: Within each diagnostic group are patients with similar pathology and treatment costs, which allows for a matching between services provided and hospital resources expended. However, evidence suggests that there is significant cost variation within individual DRGs.3
Related Data Sources: Claims data
Used in Disease Complexity Research: Yes4


View full report


"rpt_LowPrevMCCData.pdf" (pdf, 1.37Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®