A systematic review of reviews on ventriculostomy related infection definitions: A fundamental problem.

Abstract

Despite being a common, well-recognized and important complication to External Ventricular Drainage (EVD), a consensus definition for Ventriculostomy Related Infections (VRI) has not yet been established. We conducted a review to qualitatively assess definition heterogeneity and objectivity among Randomized Controlled Trials (RCTs); and investigated systematic reviews, meta-analyses, and reviews of the literature for definition citation accuracy and common methodological approaches and points of discussion related to VRI definitions. RCTs were grouped into arbitrarily chosen infection rate brackets to examine the hypothesized correlation between broader definitions and higher infection rates in RCTs. A literature search was conducted via Ovid in the Embase, MedLine and Cochrane databases from all years until the 8th of January 2025. Using Covidence, two authors (MH, AA) independently evaluated records, including studies that had ≥ 1 VRI definitions and numerical VRI rates. We identified 12 definitions in 13 RCTs, documenting pronounced disagreement. Cumulative rates for "definitive" VRI (8.4%) were lower than "suspected" VRI (13.5%). Qualitatively assessed, studies with narrow definitions presented lower VRI rates. All 17/17 meta-analyses and systematic reviews, and 15/19 literature reviews cited ≥ 1 definition inaccurately. Trial results may change based on definition choice. Definition heterogeneity was not sufficiently accounted for in meta-analyses. All literature-based studies were confounded by definition heterogeneity. Previously reported findings based on meta-analytical methodologies may be invalid, and inaccurately presented definitions could give a false impression of trial comparability. A consensus set of definitions are necessary to allow comparison between studies, and should be constructed to account for the intended use since sensitivity and specificity may have different weight depending on the context.