Measurement issues in evaluating physicians' practice patterns using hospital billing and surgeon's self-report data.

The availability of computerized case mix data, which merge medical records information and patient billing data, provide opportunity for measuring physician-specific practice patterns. However, the validity of physician comparisons is a concern when total patient charges are measured at two points in time. In this physician-initiated multicenter study, comparisons using samples of uncomplicated, baseline cases, as determined from medical records chart reviews of five surgical procedures at three institutions, demonstrated a statistically significant reduction in average length of stay (ALOS) from 0.8 to 2.8 in seven of the 15 studies while total patient charges increased in 11 of the studies from one to 29 per cent. Billing data were useful in measuring ALOS and the frequency of specific laboratory tests ordered and then comparing these to physicians' self-report practice.