Use of an electronic data warehouse to enhance cardiac surgical site infection surveillance at a large Canadian centre

Abstract

Surgical site infection surveillance (enumeration, and reporting of cases) reduces infection incidence. Data-driven "trigger" mechanisms focus surveillance on high-probability cases, yet often lack specificity. We aimed to develop trigger mechanisms with greater specificity for surveillance of cardiac surgical site infection. We developed these mechanisms in a two part study: systematic review to identify potential trigger factors; and nested case-control study to derive trigger mechanisms from a novel information structure called a data warehouse. Among 158 studies, we identified 570 trigger factors, which we grouped into themes, using the top 33 in the case-control study Using 203 cases and 516 controls, we derived two models for surveillance trigger mechanisms. These models provided true positive rates of 0.941 and 0.931 respectively (non-inferior to the current trigger mechanism), with false positive rates of 0.1085 each (superior to the current trigger mechanism). These trigger mechanisms may standardize and automate surgical site infection surveillance triggering.