Concept: Case Mix Groups (CMG™) - Overview

Introduction

This concept describes the use of a patient classification system called Case Mix Groups (CMG™) and their inclusion in hospital discharge abstract data held in the Data Repository at MCHP. Associated with CMGs are a day surgery classification system called Day Procedure Groups (DPG™) and a method to measure the relative value of resources that each CMG and DPG are expected to consume, called Resource Intensity Weights (RIW™).

Case Mix Groups (CMG™)

CMGs represent a Canadian patient classification system used to group and describe types of inpatients discharged from acute-care hospitals. Modeled after the American Diagnosis Related Groups (DRG's), CMGs were developed by the Canadian Institute for Health Information (CIHI) using four criteria:


1. patient groups had to make good clinical sense,
   
2. they had to be based on routinely collected data,
   
3. there had to be a manageable number of groups, and
   
4. they had to be statistically homogeneous with respect to length of hospital stay (Pink GH & Bolley HB, 1994).

How CMGs are Assigned

After a person is discharged from hospital, his/her chart is reviewed and an abstract is created based on the diagnoses and treatment received. Since the 1995/1996 fiscal year, Manitoba Health has submitted the abstract file on an annual basis to the Canadian Institute of Health Information (CIHI). CIHI uses a computer algorithm to assign each inpatient case in the hospital discharge abstracts file to a Case Mix Group (CMG). The CMG system is designed to assign similar inpatient cases to a single group. CMG is based on the most responsible diagnosis and describes types of inpatients discharged from acute care hospitals (Finlayson et al., 2009). Each patient case is assigned to one of 25 mutually exclusive major clinical categories (MCCs), which are based on body systems (e.g. circulatory, respiratory) and then further classified as a medical or surgical case. Finally, the CMG is assigned to create homogenous groups.

Cases within the same CMG are assigned to typical or atypical categories; typical cases represent the completion of a full course of treatment at a single hospital and are classified according to age group and complexity level using a complexity overlay from CIHI (CMG Plx™) (Finlayson et al., 2009). Atypical cases denote one of four categories: deaths, sign-outs, transfers and long-stay outliers.

For the latest information on CMGs™ see the CIHI website at: https://www.cihi.ca/en/data-and-standards/standards/case-mix/cmg

Day Procedure Groups (DPG™)

"Day Procedure Groups (DPG) is a national classification system for ambulatory hospital patients that focus on the area of day surgery. Patients are assigned to categories according to the principal or most significant procedure recorded on the patient abstract. Patients assigned to the same DPG group represent a homogeneous cluster with similar clinical episodes and requiring similar resources. The DPG 2008 grouping methodology is based on the CCI (Canadian Classification of Health Interventions) and is the result of an extensive review and revision process using Canadian case-cost data. Each DPG group is assigned a DPG Resource Intensity Weight (RIW™) value, which is used to standardize the expression of hospital day surgery volumes, recognizing that not all day surgery patients require the same health care resources."

"There are 115 DPG groups, each one defined by a set of intervention codes from CCI. In the 2008 version of DPG, several new DPG groups have been created using the data element anesthetic technique." (CIHI Web Site - Day Procedure Groups (DPG™ at https://www.cihi.ca/en/data-and-standards/standards/case-mix/day-procedure-groups-dpg - accessed October 22, 2015.

Resource Intensity Weight (RIW™)

To measure the expected use of hospital resources, CIHI developed weights known as Resource Intensity Weights (RIW™). Each CMG is assigned a weight that represents the relative value of resources that cases within that CMG are expected to consume when compared to other CMGs, and this value is called the RIW.

RIWs were originally based on U.S. charge data (1985 New York data for 1991 grouper; 1991/92 Maryland data for 1993 grouper) and Canadian length of stay data. Typical cases within a given CMG™ are assigned a single RIW™. For atypical cases, individual RIWs™ are estimated for each category.

A RIW is assigned to each patient on discharge based on the CMG to which they are assigned and other factors, such as age and co-morbidities (CIHI, 1996, 2005). The RIW allows fair comparisons across hospitals, regions, and provinces. CIHI computes the value of RIWs using micro-cost data from other Canadian provinces where these data are collected (see CIHI documentation for further information, CIHI, 1996, 2005).

The RIW reflects the relative amount of resources (i.e., the relative cost of care) for each individual. Patients who receive more complex and costly care would receive a higher RIW than those receiving less complex care (CIHI, 1996, 2005). For example: A patient with CMG assigned a value of 2 is expected to consume twice as many resources as a patient with a CMG assigned a value of 1. A patient undergoing hip replacement would consume more resources than a similar person treated for bronchitis.

With each CMG, CIHI calculates a weight based on one of three different age groups (0 to 17 years, 19 to 69 years, and 70 years and over) and one of the following complexity levels:

1 - no complexity
2 - complexity related to chronic condition(s)
3 - complexity related to serious/important condition(s)
4 - complexity related to potentially life-threatening condition(s), OR
9 - complexity not assigned

NOTE: Most cases with the same CMG/age group/complexity level will have the same RIW. However, atypical cases will be assigned a unique weight that reflects the expected level of resources consumed by that particular case. Atypical cases include cases that result from a transfer to or from an acute care facility, long-stay outliers, deaths, and those who leave the hospital against medical advice.

This approach recognizes that, even with a single CMG, there may be varying costs. For example, it may be more costly to provide care to an older person than a younger person, and multiple co-morbidities may also increase associated costs.

History of CMGs in MCHP Data

Over time, the CMG system has undergone changes, some major and some minor, to reflect changes in the International Classification of Diseases (ICD) coding system, as well as changes in patient medical / surgical conditions and the accompanying resource consumption measures. Four time frames related to significant changes in the use of CMGs at MCHP are identified below.

Prior to 1995/1996

Prior to 1995/96, MCHP purchased two CMG™ groupers from CIHI which were run on several years of data, as outlined below.

Grouper Hosp. File Location: See cmg_location (internal access only) .

NOTES:
• CMG™ files grouped at MCHP contain only a few key variables from the hospital claims prior to 95/96; the remainder are CMG™ variables.
  
• MCHP is unable to run these CMG™ groupers (mainframe software) because the U of M mainframe is not available.

MCHP Formats: see Filenames (formats) document.

Note: For a list of CMG™ titles (1995), divided into surgical, medical, and obstetrical see cmg95sm.txt . Lists for other available years are can be found in the system format library.

CAUTIONS:

Applying a CMG™ grouper to years of data other than that intended: This issue is described in the Diagnosis Related Groups (DRGs™) - Overview concept. An additional CMG-specific concern, however, is that the CMG™ number for a given group can actually change between grouper versions.

1995/1996

Beginning with 1995/96 hospital discharge abstracts, Manitoba Health arranged to have the data processed directly by CIHI. The outcome of this decision means that the CMG™, DPG™ and RIW™ values, as well as many other variables, such as Major Clinical Category (MCC) and RIWEX (typical / atypical RIW values) are added directly to the year-end file.

1997/1998 to 2006/2007

In April 1997, a refinement to the CMG system was introduced to reflect case complexity, commonly referred to as Complexity Overlay or "PLx™". Complexity is NOT a measure of clinical severity; it measures the impact of complications and comorbidity (i.e., multiple diagnoses) on length of stay (LOS) and the use of hospital resources.

Based on the presence of additional diagnoses, in addition to the most-responsible diagnosis, each case within a given CMG™ is assigned a PLx™ group ranging from Level 1 to Level 4 (level 4 representing the highest level of complexity and by definition, the most-resource intensive - i.e., prolonged LOS and more costly treatment is expected). Complexity levels are assigned to most inpatient surgical and medical CMGs.

The following MCCs are excluded since (a) case-complexity has already been defined for corresponding CMGs, or, (b) complexity does not improve clinical and statistical homogeneity.

MCCs Excluded from Complexity Overlay


• MCC 14 - Pregnancy and Childbirth
  
• MCC 15 - Newborns and Neonates
  
• MCC 19 - Mental Diseases and Disorders
  
• MCC 24 - HIV
  
• MCC 25 - Significant Trauma (*)
  
• MCC 99 - Ungroupable Data - CMG 910, CMG 912, CMG 997, CMG 998, and CMG 999.

  NOTE (*): In MCC 25, CMG 651-659, and CMG 674-679 are excluded from the complexity overlay, but complexity is applied to all remaining CMGs.


Once complexity has been assigned, regression analysis is used to generate an expected length of stay (ELOS). Variables included within regression models include CMG, complexity, age (0-17, 18-69, and 70+) and age * complexity.

Details describing the algorithm for assigning complexity, regression models and RIW™ calculation can be found in the following CIHI reference manuals (available at MCHP):


• Discharge Abstract Database (DAD) Length of Stay Indicators for Use with Complexity, 1997
  
• Discharge Abstract Database (DAD) Resource Indicators for use with Complexity, 1997

2007/2008

In 2007, CIHI introduced the CMG + (Plus) methodology that is "designed to aggregate acute care inpatients with similar clinical and resource-utilization characteristics. The CMG+ methodology was introduced in 2007 and was designed to take advantage of the increased clinical specificity of ICD-10-CA and CCI. This methodology, developed using multiple years of acute care inpatient activity and cost records, introduces and enhances several grouping factors to improve the ability to clinically group inpatients and to define length of stay and resource use indicators." (CIHI Web Site - CMG+ at: https://www.cihi.ca/en/data-and-standards/standards/case-mix/cmg - accessed October 22, 2015.

MORE INFORMATION:

• Programmers' meeting - January 28, 1997 - a great deal of useful information was disseminated at this meeting (e.g. comparisons between CMGs and DRGs, examples of programs for merging CMGs).