HRGs for benchmarking and contractual arrangements: a poisoned chalice?

A study was undertaken to determine if healthcare resource groups (HRGs) are reliable, reproducible in different centres and accurately reflect clinical activity. During a 3 month period, 385 finished consultant episodes (FCEs) were assessed. Of these, 17 (4.4%) were grouped to U01, i.e. no group allocated as the primary diagnosis was not recorded. For 271 of these episodes, a clinician had also allocated a choice of what he thought the HRG should be. These choices of HRG were then compared with the official grouping determined by the software provided by the National Casemix Office (NCMO). Disparities between 'automatic grouping' and proposed HRG assignment by the consultants occurred in 20.66% (56/271). Version 3 should see some of these disparities and actual grouping software errors reduced, in particular through the introduction of a new vascular surgery chapter. However, the absence of a primary diagnosis will still produce the allocation of group U01 in version 3. Caution should be observed in using the HRG software and in the interpretation of data obtained from it. This could be especially important in benchmarking or contractual settings as there may be adverse implications to the unit/hospitals involved.