Sequence Symmetry Analysis as a Signal Detection Tool for Potential Heart Failure Adverse Events in an Administrative Claims Database

The potential for routine sequence symmetry analysis (SSA) signal detection in health claims databases to detect new safety signals of medicines is unknown.

Our objective was to assess the potential utility of SSA as a signal detection tool in health claims data for detecting medicines with potential heart failure (HF) adverse event signals.

We applied the SSA method to all subsidized single-ingredient medicines in Australia. The source of data was the Australian Government Department of Veterans’ Affairs (DVA) administrative claims database using data collected between 2002 and 2011. We used first ever HF hospitalization and frusemide initiation as indicators for HF. A signal was considered to be present if the lower limit of the 95 % confidence interval for the adjusted sequence ratio was greater than one. To identify potential new signals of HF, we excluded medicines where HF or edema was listed in the product information (PI) of that medicine or for any other medicine in the same class. We also excluded medicines that were used in HF treatment and medicines indicated for diseases that may contribute to the development of HF.

We tested 691 medicines. HF signals were detected for 12 % (80/691) using the hospitalization event and 22 % (153/691) using frusemide initiation. Among medicines that did not have HF listed in the PI, SSA found 11 % (44/397) associated with HF hospitalization and 15 % (60/397) associated with frusemide initiation. Of the medicines tested in which no other medicine in the same class had HF or edema in the PI, and where the medicine was not indicated for a disease that is a risk factor for HF, potential new signals were generated for 2–3 % of these medicines tested (12 of 397 medicines using HF hospitalization and 9 of 397 medicines using frusemide initiation).

SSA generated potential new signals of HF for some anti-glaucoma and anti-dyspepsia medicines. For some of the potential signals, the event is biologically plausible and some have pre-marketing and post-marketing case reports to support the finding. Confirmation of these signals using cohort studies is required.