TY - JOUR PY - 1981// TI - Testing principles for severity index construction with application to burn injuries JO - Burns: journal of the International Society for Burn Injuries A1 - Krischer, JP A1 - Shuster, JJ A1 - Bingham, H. G. A1 - Melker, R SP - 1 EP - 5 VL - 8 IS - 1 N2 - A basic principle of severity indexes is that they define groups of patients with the expectation of similar outcomes given some standard of medical care. This principle is tested in an analysis of 933 burn patients where the injury descriptors themselves are used as an index. This form of analysis defines the capability of the data to detect differences which may be judged clinically significant and provides a means for testing severity indexes in terms of their contribution to increased group variation. The results showed that while patient injury descriptors succeeded in reducing variability in outcome (length of hospital stay), substantial variation remained. Also, age and total body surface area burn were the most powerful variables with respect to reducing outcome variability. Full thickness burn, respiratory involvement and burn type contributed less to defining homogeneous patient groupings. The results suggest that large relative differences in hospital stay are needed in comparative outcome studies to be statistically significant and this required difference is likely to increase with externally developed severity indexes.

LA - SN - 0305-4179 UR - http://dx.doi.org/ ID - ref1 ER -