Data Mining for Signal Detection of Targeted Therapy Related Drug Toxicity in Breast Cancer Patients

Journal Title
Journal ISSN
Volume Title

Application of signal detection methods using claims data can improve post-marketing drug surveillance. The aim of this study is to compare two routinely used approaches, the proportional reporting ratio (PRR) and Gamma Poisson Shrinker (GPS) with a tree-based scan statistic (TBSS). Using data from the Texas Cancer Registry and Surveillance, Epidemiology and End Results linked to Medicare from 2010-2014 we identified 8,949 patients with breast cancer treated with chemotherapy and 2,542 patients treated with trastuzumab in addition to chemotherapy. Inpatient and outpatient visits up to 1 year from start of therapy were used to identify adverse events (AEs). For each method two signaling thresholds were evaluated. Across all methods we found a total of 34 signals associated with use of trastuzumab. Clinical review determined that most identified signals represented known AEs or confounding. GPS on the highest signaling threshold failed to detect a well-established AE when time of follow-up was less than 6 months. Overall there was considerable agreement between methods with GPS being the most conservative. PRR and TBSS may be more appropriate in exploratory drug safety studies using this dataset.

data mining, adverse events, pharmacovigilance