Return to APA Collection Inventory Page
Text of document:
ADDENDUM C
GUIDELINES FOR THE FORMATION OF THE DRG'S
1. Except for principal diagnosis no importance w ill be attached to the ordering and/or sequencing of the diagnostic and surgical information on the patient record. Thus partitions must be based cn searches of the diagnostic and surgical information for specific diagnoses or procedures.
2. Whenever possible and appropriate, the initial partition of each major diagnostic category will bebased on the presence or absence of a surgical procedure performed in the operating room. The existence of an operating room surgical procedure differentiates the type of resouces a patient will receive. Thus, the surgery partition is considered as the primary partition from both a resource and a medical perspective. Exceptions would be psychiatric disorders and alcoholism where partitioning first in surgery would not be meaningful.
3. Whenever possible, and appropriate the initial partition of the surgical patients will. be based on the type of surgery performed.
4. Whenever possible,and appropriate the initial partition of the medical patients will be based on the principal diagnosis of the paticnts.
5. Any partition of the data based on the type of surgery or diagnostic information must be predicted on an underlying organizing principle. Variance reduction is a necessary but notsufficient condition for the specification of a split based on these variables. Examples of organizing principles include: major versus minor surgery, etiology, specification of organ involved and surgical approach.
6. Surgical partitions will be hierarchical,based in general on resource consumption. For example given that three surgical groups have been defined then patientswill be assigned to the most resource intensive surgical group if any surgery in that group is present on the record. If no surgery is found in the first group the record will be searched for surgeries in the next most resource intensive group and so on.
7. W henever possible all partitions based on surgical procedures and principal diagnosis should be completed before use of othsr variables such as age or secondary diagnosis. It is important to use principal diagnosis prior to thesecondary diagnosis since the principal diagnosis is generally more reliable than the secondary diagnosis. For example, if for patients with a particular type of surgery a partition Eased on principal diagnosis is to be used then that partition should occur before the use of any additional variable.
8. From a clinical perspective partitions should be as homogenous as possible. Medically unusual cases should be concentrated in a single "other” DRG. For example patients with a surgical procedure not necessarily expected given the principal diagnosis shculd he consolidated and not "contaminated" the remaining DRGs. Further, well defined and common particularly types of patients should form separate distinct DPG's.
[Page 2]
GUIDELINES FOR THE FORMATION OF THE DRGs
(continued)
9. If age is used in multiple places within the DRG definitions for a particular MDC, then the age grouping should be consistent unless there is a medical rationale to do otherwise For example, the age groupings within an MDC can not be at 65 years for one segment of the MDC and at 75 years for another segment without a precise medical explanaticn for why this would be expected. In this case, while the statistics recommend 65 and 75 years, if age were going to be used then a common age must be selected (e.g., 55 years). Pediatric patients will be treated separately.
10. Death can be used as a variable in the definition of the DRGs. However, its use should be avoided whenever possible. If death is used, it can only be used as the final partition in the DRG definitions.
11. System variables,such as payment source or discharge status to a nursing home,are not direct patient attributes and therefore, must not be used to define the DRGs. Such variables are characteristics of the health system in which care is rendered. Resource consumption may be affected by such variables but that effect is not related to the type of patient being treated.
12. Patient classes which occur very infrequently should not be formed unless they require highly specialized resources or are treated in particular types of hospitals. Thus, very rare patients receiving a kidney tranplant should form a separate group since they request highly specialized resources which are available only at few hospitals.
[Page 3]
CRITICISMS of OLD DRGs
1. Not Medically Meaningful
Insufficient physician input.
Cannot measure the impact of the pattern of care upon the patient.
Frequently contain what seems to be a random mix of diagnosis and procedures.
Use broad diagnostic categories, such as AMI.
Neither disease severity nor case severity are adequately captured, e.g. cancer diagnosis, and burn patients.
2. Insufficient Data Use to Form Patient Groups
Small nonrepresentative hospital patient' samples.
Rely on discharge abstract data that is incomplete and has classification and coding errors.
Not enough emphasis on age.
Patient's socioeconomic status should be used.
Type of admission and admission diagnosis should be used.
4. Not Economically Meaningful
Length of stay,not cost used as major grouping variable, e.g., Western Pennsylvania Study.
Used historical consumption of days, and hence is a norm not a standard.
4. Other
Fixed in time as a function o f current medical technology and practice.
Encourages surgery.
Inpatient use only.
Not reproduceable
[Page 4]
CHARACTERISTICS OF A PATIENT CLASSIFICATION SCHEME*
1. THE PATIENT CHARACTERISTICS USED IN THE DEFINITION OF THE PATIENT CLASS SHOULD BE LIMITED TO lNFORMATION ROUTINELY COLLECTED ON HOSPITAL ABSTRACT SYSTEMS.
2. THERE SHOULD BE A MANAGEABLE NUMBER OF PATIENT CLASSES W H IC H ENCOMPASS ALL PATIENTS . SEEN ON AN INPATIENT BASIS.
3. EACH PATIENT CLASS SHOULD'CONTAIN PATIENTS WITH A SIMILAR PATTERN OF RESOURCE INTENSITY: .
4. EACH PATIENT CLASS SHOULD CONTAIN PATIENTS . WHO ARE-SIMILAR FROM A CLINICAL PERSPECTIVE I.E., EACH CLASS SHOULD BE CLINICALLY COHERENT (MEDICALLY MEANINGFUL).
'YALE
[Page 5]
DATA BASIS USED
A. Commission on Professional and Hospital Activities (CPHA) Professional Activity Study Hospitals
Number o f Hospitals: 332 nationally distributed
Time Period: 3rd and 4th quarters, 1979
Number of Discharges: I.4 million
Sample: 320,000 discharges
Dependent Variable: Length of stay
Use of sample data'for construction of MDCs and DRGs.
Used full data set for statistical and reliability-procedures.
B. New Jersey State Department of Public Health
Number of Hospitals: 33
Time Period: 3rd and 4th quarters, 1979
Number of Discharges:. 334,000
Dependent Variable: Individual patient costs
-Used to determine whether there was a variation in patient costs in the defined DRGs.
[Page 6]
FREQUENTLY USED VARIABLES
PRINCIPAL DIAGNOSIS
SURGICAL RANKING
OPERATING ROOM PROCEDURE
PRESENCE, ABSENSE COMPLICATION/COMORBIDTYU :
AGE 17
AGE 70CC
MALIGNANCY
LENGTH OF STAY--AS THE DEPENDENT VARIABLE :
EXPLORED BUT NOT USED VARIABLES
PRINCIPAL PAYMENT METHOD
PREOPERATIVE LENGTH OF STAY
CHEMOTHERAPY
RADIOTHERAPY
PHYSICAL THRAPY
DIALYSIS
NUMBER R O F COMPLICATIONS
NUMBER OF COMORBIDITY
NUM BER OF OPERATIONS
NUMBER OF D IAGNOSiC PROCEDURES
AUTOPSY
LOCALITY
TYPE OF ADMISSION EMERGENCY
FROM OTHER ACUTE CARE FACILITY
FROM SNF
ER
READMIT
INTENSIVE CARE -
ICU
CCU
SPECIAL CARE UNIT
BIRTH W E I G H T
NUMBER OF SECOND D IAGNOSES
NUMBER OF MDCS INVOLVED
ADMISSION DATE
SEX
RACE
[Page 7]
BASIC SPLITTING APPROACH
[Page 8]
FIGURE IA
MAJOR DIAGNOSTIC CATEGORY 01:
DISEASES AND DISORDERS OF THE NERVOUS SYSTEM
Surgical Partitioning
[Page 9]
KEY TO SYMBOLS USED IN DIAGRMS
(Figures 1-23)
Symbol Definition
[Page 10]
FIGURE IB
MAJOR DIAGNOSTIC CATEGORY 01:
DISEASES AND DISORDERS OF THE NERVOUS SYSTEM
Medical Partitioning
[Page 11]
FIGURE IC
MAJOR DIAGNOSTIC CATEGORY 01:
DISEASES AND DISORDERS OF THE NERVOUS SYSTEM
Medical Partitioning (continued)
(End of text)