Background

• Diagnosis Related Groups (DRGs™) is an American patient classification system that describes the types of patients treated by a hospital (i.e. its case mix).
  
  
• The DRGs™ work by grouping the 10,000+ ICD-9-CM codes into a more manageable number of meaningful patient categories (close to 500 now). Patients within each category are similar clinically and in terms of resource use. (The DRG™ grouper uses administrative data to group patients.)

DRG™ Assignment

DRG™ grouper software (licensed from Health Systems Consultants) uses "principal diagnosis, secondary diagnoses, surgical procedures, age, sex and discharge status of the patients treated" (DRG™ Definitions Manual, 1994) to assign inpatient records to a specific DRG. The grouper requires that diagnoses and procedures be classified using ICD-9-CM. DRG™ information (e.g., DRG™, RDRG®, ADRGs) is attached to each record in the MCHP hospital databases. Diagnoses 2 through 16 are scanned for complications/comorbidities. In the Hospital Discharge Abstracts Database, all 12 procedure code variables are scanned by the grouper to select the most resource-intensive one.

1. Most DRG™/RDRG® data are grouped by Primary DX. For DRG™/RDRG® the hospital diagnoses are temporarily modified so that DX01 is the PRIMARY (as required by DRG™ and as recorded at MB Health prior to April, 1987), and MOST RESPONSIBLE (if different) becomes one of the secondary diagnoses. This is essential since newborns must have their birth DX as DX01 in order for RDRGs® to work properly. All possible principal diagnoses are divided into 25 mutually exclusive categories referred to as Major Diagnostic Categories (MDCs) which are loosely based on organ systems.

2. Cases are then separated into surgical and medical cases. (the presence of a surgical procedure has a significant impact on resource use.) A case is considered surgical if there is a procedure performed that would normally require the use of an operating room. (E.g. A bronchoscopy would not be considered surgical because no operating room is required).

3. Surgical patients are further divided based on the actual surgical procedure performed, while medical patients are further divided based on the actual principal diagnosis. Usually, in each MDC, surgical patients are divided according to the extent of the surgical procedure performed. (Different classes of procedures are grouped within each MDC reflecting different resource requirements.) Because patients can have more than one surgical procedure performed at one time, or during one hospital stay, the DRG™ system has a hierarchy of surgical procedures within each MDC and the most resource intensive procedure (highest on the hierarchy) is used for DRG assignment. (Thus the ordering of the surgical procedures on the hospital discharge abstract has no influence on the assignment of DRG™.)


• In general, specific groups of principal diagnoses are defined for medical patients within each MDC and usually include a class for neoplasms, symptoms and specific conditions relating to the organ system involved.
  
  
• There are a few exceptions to the assignment algorithm described above. Because certain patient groups are extremely resource intensive, they are put into separate DRGs before MDC assignment (i.e., liver transplant, bone marrow transplant, tracheostomy).

DRG™ Variables

• DRG™

• ADRG (adjacent DRGs™) - rolled-up DRGs™. They collapse any adjacent DRGs that are different only on the basis of age or presence/absence of complications/comorbidities. These are useful if you do not need to distinguish 'with' versus 'without' complications/comorbidities and age breaks.

• RDRG® (Refined Diagnosis-Related Group). They account for the impact on resource use of severity of illness by dividing DRGs™ (ADRGs actually) into three (for medical categories) or four (for surgical categories) levels of severity based on the presence of comorbidities and complications and their impact on resource use.

Versions

Although there have been countless versions of DRGs™ since they were first developed in the 1960s, the official version number associated with each DRG™ grouper refers to the version since the implementation of the prospective payment system (PPS) in the United States in 1983. The version of DRG™ used then was given the version number 1.0.

A new version of DRG™ is developed each year (effective Oct. 1) because of the yearly ICD-9-CM changes, as well as changes in clinical practice and resource use. Currently in the U.S., version 12 is used. It is important to use the correct version. For example, for newborn analyses, DRG™ 9/5 must be corrected using the DR5 file.

Because the RDRGs® were developed after the implementation of PPS, the version numbers for the RDRGs® do not correspond to the version numbers of the DRGs™. Thus, the RDRGs® that correspond to the version 9 DRGs™ are version 5. Likewise, version 7 RDRGs® correspond to version 11 DRGs™.

Because Manitoba hospitals respond to ICD-9-CM changes 6 months after they occur (i.e. if a change occurs Oct 1, 1993 in ICD it is not captured in the Manitoba Health data until April 1, 1994) a DRG™ grouper for 93/94 fits perfectly with Manitoba Health data for the fiscal year 94/95.

Two versions of DRG™/RDRG® are available for the 1993/94 hospital data:


• Version 5/9 (RDRG®/DRG™) - effective Oct. 1, 1991; fits 92/93 data
  
  
• Version 7/11 - effective Oct. 1, 1993; fits 94/95 data.


See: DRG_versions.txt (internal access only)

Notes


• We have done some code editing so that changes in ICDs that occurred before or after the particular grouper in question are taken into consideration (making the grouper in question and the data a better fit).
  
  
• RDRG® 2.3 information is only available for data sets where it currently exists; the capability to run it no longer exists.
  
  
• For the hospital years 1989-96, the DRG™ software is a perfect fit with MCHP data. For other hospital years, the database has been edited to allow a given DRG version to fit other years of data. A list of DRG™/RDRG® versions for each year of MCHP data is provided in DRG_versions.txt file.

Cautions

Conducting Comparative Studies

An important difference between Canadian and American data should be noted for comparative analyses, although results will usually be similar. "Principal diagnosis" refers to the diagnosis established after study to be most responsible for causing the patient's ADMISSION to hospital. MCHP uses "most responsible diagnosis" instead, "the one diagnosis which describes the most significant condition of a patient which causes his STAY in hospital" (Manitoba Health, 1997).

Deciding which Grouper to Use

Applying a DRG™ grouper to years of data other than that intended. The decision about which grouper to use when using a particular year's data (e.g., 93/94) depends on the research design:


• Cross-sectional: If only one year of data is to be used, it is usually best to use the more recent DRG version because it will more accurately capture current practices. For example, the 7/11 grouper (RDRG® V.7 and DRG™ V.11) has two new DRGs™ for laparoscopic cholecystectomies. Using an older grouper (e.g., 5/9), no distinction would be made between lap cholecystectomies and other cholecystectomies. It is thus preferable to use version 7/11 for 1993/94 in this case since it differentiates types of cholecystectomy.
  
  
• Longitudinal: This type of analysis may require using more than one grouper (depending on the focus of the study).

MORE INFORMATION:

• Brownell M (1993). [memo]
  
• Clinical Coding: St. Anthony's, February 1997.
  
• Health Systems Consultants (New Haven, CT) - contact: Karen C. Schneider
  
• Health Systems Consultants: DRG Refinement Grouper, 1994. (MCHP has also obtained earlier versions)