David Griffin , BBiomedSc, BSc (Hons), MBBS (Hons), MPH 1 , Irani Thevarajan , MBBS, FRACP, PhD 1 , 2 , Simon Iles , MBChB, FRACP, FCICM 3 , Thomas Rechnitzer , MBBS, FCICM 3 , Timothy Spelman , MBBS, FRACGP, PhD 2 , 4 , Deborah Barge , BN, CCRN 3 , Caroline Marshall , MBBS, PhD, FRACP, GradDipClinEpi 1 , 2 , Nerina Harley , MBBS, MD, PGDipEcho, FRACP, FCICM, AFRACMA 3 , 5 , 6 , Michael Richards , MBBS, FRACP, MD 1 , 2 , 4
26 November 2018
Ventilator-associated pneumonia (VAP) is a common complication of admission to intensive care units (ICU), and may be associated with significant morbidity, mortality and healthcare cost. While VAP surveillance is a desirable element of ICU infection prevention programmes, the lack of an easily applicable definition, providing accurate and clinically meaningful data limits implementation. We aimed to conduct a pilot study of ventilator associate event (VAE) surveillance, per Centre for Disease Control National Healthcare Safety Network (CDC NHSN) criteria, to describe the incidence, and outcomes for patients with VAE in our setting.
We conducted a prospective cohort study in our 24-bed mixed tertiary ICU in Melbourne, Australia. Adult patients requiring mechanical ventilation for ≥2 days between March and October, 2015, were included. We collected detailed clinical and laboratory data, including antibiotic duration and indication, and ICU and hospital length of stay. We applied the CDC NHSN criteria.
We included 202 patients (median age 58.1 ± 17.8 years, 32.7% female, 73% medical), over 1,390 ventilator days. Ventilator associated condition (VAC) occurred in 33 (16.3%) patients (23.7 per 1,000 ventilator days), Infection-related VAC (IVAC) in 15 (7.4%) patients (10.7 per 1,000 ventilator days), and possible VAP (PVAP) in 8 (3.9%) patients (5.75 per 1,000 ventilator days). In contrast, clinician-diagnosed VAP (CD-VAP) occurred in 37 (18.3%) patients (26.6 per 1,000 ventilator days). Patients with VAC had a greater median number of ventilator days (12 vs. 4, P < 0.001), ICU length of stay (LOS) (17 vs. 6 days, P < 0.005), hospital LOS (30 vs. 19 days, P = 0.005), and antibiotic days (12 vs. 5, P < 0.001), than those without VAC. CD-VAP was associated with VAC (OR 4.7, 95% CI 2.1–10.6, P < 0.001), but agreement was poor (kappa 0.29). The overall sensitivity of VAC for CD-VAP was 38%, specificity was 89%, PPV 48%, NPV 85%, while for PVAP these were 17, 99, 88 and 82%, respectively.
VAC is associated with important, measurable surveillance outcomes, but the agreement, sensitivity and predictive value of these criteria for CD-VAP are poor. Hence the CDC criteria may miss clinically important healthcare-associated infections and may not capture the most appropriate target group for VAP prevention.