Background: : Diagnostic errors represent an important source of potentially preventable morbidity and mortality in hospitalized patients. Intensive care unit (ICU) patients may be at high risk for misdiagnosis because of high acuity and complexity of care. The spectrum of diagnostic errors in adult ICU populations is not well defined.
Methods: Objective─Estimate the frequency, severity, and principal causes of diagnostic error in adult ICU patients. Design─Systematic review of observational studies. Electronic (MEDLINE, EMBASE) and manual (references of eligible articles) search for articles (1960-2009). Terms (MeSH and Emtree) used included ICU, critical care, intensive care; AND diagnostic error, misdiagnosis, diagnostic delay, diagnostic error; AND necropsy, autopsy. Inclusions─Studies of ≥10 adult ICU patients with diagnostic errors confirmed by autopsy were included. Two independent reviewers selected studies, with differences adjudicated by a third. Patient characteristics, error rates, and error classes (using Goldman criteria for autopsy misdiagnoses) were abstracted. Differences were resolved by consensus.
Results: We examined 202 citations and reviewed 76 articles. 35 studies were included. Principal reasons for exclusion were lack of original data and lack of ICU-specific data. The 35 studies reported on diagnostic errors in 8,712 autopsies performed among 13,030 ICU deaths. For studies that provided both the number of deaths and the number of autopsies performed (30 studies), the average autopsy rate was 29% (range=3.7%-96%). A total of 1625 misdiagnoses were reported (19% of autopsies) with 1148 providing sufficient information to assign a Goldman class. Of these, Goldman Class I errors (major, likely lethal) were reported in 258 (22.5%), with a range of diagnoses being missed. Infection was the most common Class I error (8 studies) followed by pulmonary embolism (6 studies) in studies reporting these data. Class II errors (major, likely non-lethal) were identified in 501 (43.6%). Infection was again the most common (8 studies) followed by malignancy (4 studies). Minor Class III and IV errors were found in 233 (20.3%) and 156 (13.6%) cases.
Conclusion: Our results suggest that misdiagnosis in the adult ICU population may be a frequent occurrence. A major likely lethal diagnostic error is found in one-fifth of autopsied cases. It is uncertain whether autopsy data over- or underestimate the frequency or severity of diagnostic errors but the most frequent major diagnostic errors appear related to misdiagnosed infections. Further research is needed to better quantify diagnostic errors and to define potential strategies to reduce their frequency or mitigate misdiagnosis-related harm in the adult ICU.
See more of: The 32nd Annual Meeting of the Society for Medical Decision Making