Evaluation of diagnosis-related groups in the National Health Service.

Evaluation of the use of diagnosis-related groups (DRGs) has revealed a number of technical problems in coding of diagnoses and operative procedures, as well as unresolved issues in the clinical acceptability of existing groupings. An investigation of the statistical homogeneity of DRGs, in terms of duration of patient stay, is described. Consideration of data relating to some 990,000 episodes of in-patient care in three English Regions discloses wide variations in statistical homogeneity, both between DRGs and in relation to individual clinical specialties. The greatest homogeneity is found in ENT surgery and gynaecology; and the least in general medicine and orthopaedic surgery. The need for improved data collection and coding procedures is discussed, together with the advisability of sensitivity in the interpretation of DRGs, as well as the need for a co-ordinated approach to their refinement for application in any wider introduction in the NHS.