Stephen J.LurieMD, PhD, Senior EditorIndividualAuthorJody W.ZylkeMD, Contributing EditorIndividualAuthor
To the Editor: Drs Hayward and Hofer1 document the poor reliability of physicians' retrospective
judgments of medical errors. They suggest that "previous interpretations of
medical error statistics are probably misleading." However, given the methodological
problems of their study, we question this interpretation.
First, the sample is small. Although the study started with 4198 deaths,
only 111 of them were included in the analysis. The conclusions are based
on multiple reviews, but these were conducted for only 62 patients; 59 patients
had only 1 review. Second, the problems were not representative. Patients
with a small number of fluid, electrolyte, and drug toxicities were oversampled.
In addition, the Department of Veterans Affairs hospitals tend to care for
sicker patients, but there was no standardization of disease severity. Third,
the magnitude of the adverse event problem appears even worse if the quality
of care that led to death is taken into account. The authors reported that
10.2% of deaths rated "borderline in care"; 22.7%, "possibly preventable by
optimal care"; 6.0%, "definitely preventable"; and 6.0%, "substandard"—together
leading to an unacceptable 44.9% of care deemed suboptimal or worse.
Barach P, Mohr JJ. Preventable Deaths From Medical Errors. JAMA. 2001;286(22):2813-2814. doi:10.1001/jama.286.22.2808