Ventilator-associated pneumonia surveillance using two methods

Abstract

Background: Ventilator-associated pneumonia surveillance is used as a quality indicator due to concerns that some cases may be preventable and may contribute to mortality. Various surveillance criteria exist for the purposes of national reporting, but a large-scale direct comparison has not been conducted.

Methods: A prospective cohort study applied two routinely used surveillance criteria for ventilator-associated pneumonia from the European Centre for Disease Control and the American Centers for Disease Control to all patients admitted to two large general intensive care units. Diagnostic rates and concordance amongst diagnostic events were compared.

Findings: A total of 713 at-risk patients were identified during the study period. The European surveillance algorithm returned a rate of 4.6 cases of ventilator-associated pneumonia per 1000 ventilation days (95% confidence interval 3.1-6.6) and the American surveillance system a rate of 5.4 (3.8-7.5). The concordance between diagnostic events was poor (Cohen's Kappa 0.127 (-0.003 to 0.256)).

Conclusions: The algorithms yield similar rates, but the lack of event concordance reveals the absence of inter-algorithm agreement for diagnosing ventilator-associated pneumonia, potentially undermining surveillance as an indicator of care quality.

Keywords: Critical care; Infection surveillance; Mechanical ventilation; ventilator-associated pneumonia.