Assessing the accuracy of diagnostic codes in administrative databases: the impact of the sampling frame on sensitivity and specificity.

The appropriate application of administrative data in clinical research requires an assessment of the accuracy of diagnostic codes against a gold standard of expert review of selected clinical charts. Typically accuracy is related to the degree of correlation between the administrative coding and the expert review. However, the choice of the sampling frame for the chart review can introduce bias that may alter the assessment of sensitivity and specificity of the coding process. This poster will explore advantages and disadvantages of common methods by which sensitivity and specificity are determined, and suggest possible methods of addressing bias.