Length of intensive care unit stay and the apparent efficacy of antimicrobial-based versus non-antimicrobial-based ventilator pneumonia prevention interventions within the Cochrane review database.

Abstract

Background: The risk of acquiring ventilator-associated pneumonia (VAP) increases with intensive care unit (ICU) length of stay (LOS). The objectives here are to estimate, using data derived from randomized concurrent control trials (RCCTs) of non-antimicrobial versus antimicrobial interventions, the relation of LOS with firstly, apparent VAP prevention effect, and secondly, with VAP incidence in control and intervention groups.

Methods: Control and intervention group data derived from 13 Cochrane reviews of 78 RCCTs of antimicrobial-based interventions versus 111 RCCTs of various non-antimicrobial-based VAP prevention interventions.

Results: In meta-regression models of VAP prevention effect versus group mean LOS, the effect size of non-antimicrobial-based interventions regress towards the null (+0.028; +0.002 to +0.054) whereas antimicrobial-based interventions regress away from the null (-0.043; -0.08 to -0.004). The day 9-10 VAP incidence increase is 1.28 (0.97-1.6) percentage points among the control groups of antimicrobial interventions per day. By contrast, these increases among antimicrobial- (0.45; 0.19-0.71) and non-antimicrobial- (0.58; 0.29-0.87) intervention groups and in control groups of non-antimicrobial- (0.76; 0.46-1.05) interventions are all similar.

Conclusions: Antimicrobial-based versus non-antimicrobial-based interventions show overall greater apparent VAP prevention which is most apparent with longer group mean LOS. The basis for this surprising relationship with LOS resides, paradoxically, within the control rather than the intervention groups. This discrepancy implicates indirect (spill-over) effects, inapparent within individual antimicrobial-based RCCTs, which could spuriously conflate the appearance of VAP prevention.