Reliability of administrative data for evaluating the quality of systemic treatment for cancer.

208Background: There is ongoing interest in leveraging administrative data to examine quality but methodological concerns persist. We evaluated the reliability of a previously established panel of administrative data derived quality measures for systemic cancer treatment. Methods: The study cohort consisted of women diagnosed with early stage (stage I-III) breast cancer (ESBC) in Ontario, Canada, in 2010. Performance on 11 quality indicators evaluated using deterministically linked healthcare administrative databases has been reported previously. The sensitivity and specificity of these 11 indicators were examined using the chart as the gold standard. Results: The administrative cohort consisted of 6,795 women with ESBC from which a validation cohort of 705 patients was randomly selected from among patients who underwent cancer surgery at one of five hospitals chosen to balance feasibility and institutional characteristics.Sensitivity and specificity varied by indicator (Table). Reliability of some indica...