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
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 1,000 ventilation days (95% confidence interval 31-6.6) 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
Ventilator-associated pneumoniaInfection surveillanceMechanical ventilationCritical care