Degree Name

Doctor of Public Health


Graduate School of Health and Medical Sciences


The Australian National Diagnosis Related Groups (AN-DRG) classification is intended to assign acute inpatient episodes to classes which are relatively homogeneous in terms of clinical attributes and the resources used in the provision of care. The purpose of this study was to determine the extent to which this objective was met in a sample of acute admitted patients with one or more diagnoses indicating the presence of diabetes meliitus.

The sample comprised all 2094 discharges with one or more diabetes diagnoses from acute care hospitals in the lllawarra Area Health Service in 1993-94. A subsample of 386 records was selected for the purpose of more detailed analysis by chart audit. Finally, another sample of 22 admitted patients was identified who were known to have diabetes because of their contacts with a community service, but whose diabetes had not been recorded in the discharge database.

There were three major findings. First, the discharges were distributed among many AN-DRGs in a way which was neither clinically coherent nor effective in terms of prediction of resource use. The logic of AN-DRG assignment, while effective for many types of care needs, appears to be less so where there is an underlying chronic condition. Compromises are unavoidable, but there is reason to conclude that chronic conditions have been given too little attention.

Second, there were many weaknesses in the data which are routinely assembled for the purpose of AN-DRG assignment. They included errors of medical documentation, abstraction and sequencing, and coding.

Third, the AN-DRG logic appears to ignore or under-estimate the effects of diabetes as a secondary condition. One critical finding which supports this view was that, where all diabetes diagnoses were deleted and the records reassigned to AN-DRGs, only 10 records in 1945 (0.5%) were assigned to different classes. Diabetes diagnoses have so little effect for one dominant reason: that the DRG logic only takes account of one more diagnosis after the principal, and a condition like diabetes is characterised by multiple problems.

It is concluded that, if the AN-DRG classification is to become more effective for cases with serious chronic conditions like diabetes, modifications will be needed in the simple and near-universal logic of assignment to a diagnosis or procedure cluster followed by (selective) splitting on one more condition and/or age. Some preliminary ideas are presented as to how greater precision and clinical meaning might be achieved.

02Whole.pdf (3580 kB)



Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.