Application of the New Centers for Disease Control and Prevention Surveillance Criteria for Ventilator-Associated Events to a Cohort of PICU Patients Identifies Different Patients Compared With the Previous Definition and Physician Diagnosis

Abstract

Objectives: We sought to compare the performance of the 2008 Centers for Disease Control and Prevention Pediatric criteria for ventilator-associated pneumonia, the 2013 Adult Ventilator-Associated Condition criteria, the new Draft Pediatric Ventilator-Associated Condition criteria, and physician-diagnosed ventilator-associated pneumonia in a cohort of PICU patients.

Design: Secondary analysis of a previously conducted prospective observational study.

Setting: PICU within a tertiary care children's hospital between April 1, 2010, and April 1, 2011.

Patients: Patients between 31 days and 18 years old, mechanically ventilated via endotracheal tube for more than 72 hours and no limitations of care.

Interventions: None.

Measurements and main results: Ventilator-associated pneumonia criteria applied in real time and ventilator-associated condition criteria applied retrospectively. Outcomes assessed between cases and noncases within criteria. Of the 133 eligible participants, 24 (18%) had ventilator-associated pneumonia by 2008 Pediatric criteria and 27 (20%) by physician diagnosis. Sixteen (12%) and 10 (8%) had ventilator-associated condition by 2013 Adult and Draft Pediatric criteria, respectively. We found significant overlap between cases identified with 2008 Pediatric criteria and physician diagnosis (p = 0.549), but comparisons between the other definitions revealed that the newer criteria identify different patients than previous Centers for Disease Control and Prevention ventilator-associated pneumonia criteria and physician diagnosis (p < 0.01). Although 20 participants were diagnosed with ventilator-associated pneumonia by 2008 Pediatric criteria and physician diagnosis, only three participants were identified by all four criteria. Three subjects uniquely identified by the Draft Pediatric criteria were noninfectious in etiology. Cases identified by all criteria except Draft Pediatric had higher ratios of actual ICU length of stay to Pediatric Risk of Mortality III-adjusted expected length of stay compared with noncases.

Conclusions: The Draft Pediatric criteria identify fewer and different patients than previous ventilator-associated pneumonia criteria or physician diagnosis, potentially missing patients with preventable harms, but also identified patients with potentially preventable noninfectious respiratory deteriorations. Further investigations are required to maximize the identification of patients with preventable harms from mechanical ventilation.

Figure 1.

Study Recruitment Diagram. A total of 483 patients admitted to the PICU were screened for study recruitment. One hundred forty subjects were consented, with 7 subjects determined to be ineligible for participation due to study protocol. A total of 133 patients were included in this study.

Figure 2.

Different Cohorts of Patients are Identified by Each Diagnostic Criteria. No significant difference was observed between the 2008 Pediatric criteria and Physician diagnosis (p=0.549; k=0.47, 95% CI: 0.21-0.74). All other comparisons were significantly different (p<0.01; except 2008 Pediatric and 2013 Adult p=0.185) with poor agreement in the patients identified.

Figure 3.. Differences in Clinical Outcomes by Diagnostic Criteria: VAP/VAC vs. No VAP/VAC.

(A) Significantly larger PICU length-of-stay ratios, defined as the ratio of actual ICU length-of-stay to PRISM-adjusted expected length of stay, were observed in VAP/VAC cases versus non-cases in all criteria tested except Draft criteria. (B) Significantly longer hospital lengths of stay among VAP/VAC cases versus non-cases were only observed by the 2008 Pediatric VAP and Physician Diagnosis criteria. (C) Significantly fewer ventilator free days were observed in VAP/VAC cases versus non-cases in all =criteria tested except Draft Criteria.