Use of healthcare claims to validate the Prevention of Arrhythmia Device Infection Trial cardiac implantable electronic device infection risk score

Aim: The Prevention of Arrhythmia Device Infection Trial (PADIT) infection risk score, developed based on a large prospectively collected data set, identified five independent predictors of cardiac implantable electronic device (CIED) infection. We performed an independent validation of the risk score in a data set extracted from U.S. healthcare claims. Methods and results: Retrospective identification of index CIED procedures among patients aged ≥18 years with at least one record of a CIED procedure between January 2011 and September 2014 in a U.S health claims database. PADIT risk factors and major CIED infections (with system removal, invasive procedure without system removal, or infection-attributable death) were identified through diagnosis and procedure codes. The data set was randomized by PADIT score into Data Set A (60%) and Data Set B (40%). A frailty model allowing multiple procedures per patient was fit using Data Set A, with PADIT score as the only predictor, excluding patients with prior CIED infection. A data set of 54 042 index procedures among 51 623 patients with 574 infections was extracted. Among patients with no history of prior CIED infection, a 1 unit increase in the PADIT score was associated with a relative 28% increase in infection risk. Prior CIED infection was associated with significant incremental predictive value (HR 5.66, P < 0.0001) after adjusting for PADIT score. A Harrell's C-statistic for the PADIT score and history of prior CIED infection was 0.76. Conclusion: The PADIT risk score predicts increased CIED infection risk, identifying higher risk patients that could potentially benefit from targeted interventions to reduce the risk of CIED infection. Prior CIED infection confers incremental predictive value to the PADIT score.