On approximations in treatment costing.

As detailed costing may be a time-consuming and expensive exercise within an evaluation, economists will be conscious of the possibilities of taking short-cuts. To explore the viability of such approaches in the context of acute care (the surgical treatment for colorectal cancer), we compare the results of a detailed costing study with reduced list costing and econometric estimation. We conclude, first, that use of a reduced list is likely to generate substantial research economies only at the expense of inaccuracy. Second, crude costing, based upon average costs of the specialty, is acceptable when the frame of reference is the aggregate. Such crude costing, however, is vulnerable to bias when specific sub-samples of patients are to be considered. Finally, total costs are predictable from a restricted list of cost and event variables, and with a high degree of accuracy, although ex ante specification of the functional form is problematic.