Yao RJR, et al. - A total of 24 studies, comprising 21 from North America or Scandinavia were recognized by the researchers in order to evaluate the validity of atrial fibrillation (AF, most prevalent arrhythmia and a major matter of stroke and healthcare utilization) case definitions in administrative databases. In 10, 4 and 10 studies, hospital, ambulatory and mixed data sources were evaluated, respectively. Nine various AF case definitions were assessed, most on the basis of ICD-9 or 10 codes. In patients diagnosed with AF, 22 studies evaluated case definitions and therefore, could generate positive predictive values alone. Half the studies sampled unrestricted populations comprising a mix of those with and without AF to evaluate sensitivity. Only 13 studies involved ECG confirmation as a gold standard. Therefore, relatively few studies investigated sensitivity, and fewer still incorporated rhythm monitoring in the gold standard comparison. Administrative data may fail to recognize an important proportion of patients with AF. Moreover, this, in turn, may bias estimations of quality of care and prognosis.