Retrospective data for diabetic foot complications: only the tip of the iceberg?

Admission rates for diabetes-related foot complications to an Australian hospital were assessed by comparing the frequently used method of retrospectively identifying patients according to International Classification of Diseases (ICD) codes with that of prospectively identifying patients at the time of admission. The aim was to determine the true admission rate of diabetes-related foot complications and to assess the ability of ICD discharge codes to accurately represent the clinical severity of each identified admission. The retrospective study of ICD codes identified approximately one-third of the patients admitted during the prospective studies. Furthermore, ICD codes allocated in the prospective studies failed to accurately represent the clinical condition in 61% of cases and the corresponding Weighted Inlier Equivalent Separations weighting resulted in a $215 000/year deficit for admissions to a single hospital.