Analysis to develop an automated denominator for the IHS Diabetes Audit.

In 1986 the Indian Health Service (IHS) Diabetes Program developed the Diabetes Audit as a method to assess the diabetes care provided in the various systems delivering healthcare to American Indians and Alaska Natives (AI/AN). This assessment is a self-audit of medical records. To promote uniformity of this self-audit, well-defined and widely tested criteria has been provided for identifying which patients should be included in the audit and what documentation must be present to ¿count¿ a preventive service as ¿provided¿ or a key surrogate variable as having ¿occurred.¿ However, concern has been raised about the ¿variable quality of the diabetes registry maintained at each facility and variable adherence to the medical record review definitions.¿ As part of an initiative to see if an automated assessment tool could be developed to use already existing data, the IHS Diabetes Program and the IHS Information Technology Support Center (ITSC) have undertaken a project to see if a logic could be designed that would allow them to select a valid, usable, understandable, and reproducible denominator of patients from the most widely used clinical information systems among the various Indian healthcare delivery systems, the Patient Care Component (PCC), the major clinical component of the IHS¿s integrated healthcare management system, the Resource and Patient Management System (RPMS). The purpose of this study was to determine if electronic indicators for these criteria could be developed, and to evaluate whether this electronic approach might work as well as skilled auditors trying to apply the criteria by manual chart review. Based on the results of this analysis, the authors decided that an appropriate set of patients upon whom a valid, usable, understandable, and reproducible audit could be conducted could be automatically selected from data exiting in the RPMS. While the authors concluded that an automated process is important in standardizing patient selection, they recognized that this is a first step towards a larger goal, a fully automated diabetes audit for which numerator data also come from automated analyses and no chart review is necessary.