PROPORTION OF DIAGNOSIS ERRORS PREVENTABLE BY DIAGNOSIS SUGGESTION SYSTEMS

Background: Diagnostic mistakes represent a significant proportion of medical errors. Diagnosis suggestion systems have been proposed as a way to reduce errors. Such systems operate on the principle that an incomplete list of possible diagnoses in the “practitioner memory workspace” is an important cause of diagnostic mistakes. Diagnosis suggestion systems address this by presenting diagnostic possibilities, sometimes generated by computational techniques, with or without an effort at probability ranking. However, it is not clear what proportion of diagnostic mistakes this strategy might actually prevent.

Methods: COPIC Insurance Co. provides professional liability insurance to approximately 8,000 physicians in Colorado and Nebraska. COPIC captures data on malpractice claims for covered providers – as well as reports of occurrences that might give rise to claims – in a structured database. Events are coded using COPIC’s “Taxonomy of Medical Errors.” We identified claims in which “Wrong or Delayed Diagnosis” was a factor. Two experienced clinicians reviewed brief, nurse-generated narratives for each claim. They sorted wrong diagnosis claims into 3 groups: a) Those potentially prevented if the practitioner had been prompted with a list that included the correct diagnosis; b) those where this would not have made a difference (e.g., lost lab report); and c) claims where a determination could not be made.

Results: Out of 33,189 reports (claims and occurrences) collected from 2002-2010, 4,141 events (12%) were coded as involving diagnostic error; of which 1,295 (31%, 4% of the total) were claims. After removing duplicate records in cases with multiple claims, 785 claims were reviewed. Reviewers agreed on 85 claims (10.8%) in which the error might have been prevented if the practitioner had been presented with a list of suggestions that included the correct diagnosis. They agreed that in 379 claims (48.3%), the error would not have been affected by diagnostic suggestions. They agreed that in 25 claims (3.2%) the information provided did not permit a decision.

Conclusion: In malpractice claims involving a wrong or delayed diagnosis, more complete diagnostic suggestions might be useful in averting the error in a small but significant number of cases. However, the majority of events in this series would not have been amenable to correction by diagnosis suggestion systems. Because of disagreement between reviewers, this study should be repeated with a strict definition of cases amenable to diagnostic suggestion and more reviewers, data sets, and care delivery settings.