Health Services Research & Development

Veterans Crisis Line Badge
Go to the ORD website
Go to the QUERI website

NRI 03-268 – HSR&D Study

New | Current | Completed | DRA | DRE | Portfolios/Projects | Centers | Career Development Projects

NRI 03-268
Reducing Nosocomial Infection in ICUs by Active Screening Surveillance
Linda L McKinley BSN MPH
William S. Middleton Memorial Veterans Hospital, Madison, WI
Madison, WI
Funding Period: October 2008 - January 2013

BACKGROUND/RATIONALE:
Approximately 1.7 million healthcare-associated infections (HAIs) occur annually in the U.S. Multidrug-resistant organisms (MDROs) cause approximately 70% of HAIs with Methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE) accounting for 75% of MDRO HAIs. For every person infected with an MDRO, many more are colonized. Infected and colonized patients both serve as reservoirs for transmission. While clinical cultures (i.e., passive surveillance) identify patients with active infections requiring isolation (i.e., private room and use of glove/gown barriers by staff), active surveillance cultures (ASCs) are necessary to identify those with asymptomatic colonization who also require isolation.
The aim of the original proposal submitted in 2004 was to conduct a randomized trial to evaluate the effect of ASC and isolation for MRSA and VRE colonization in reducing HAIs in critically ill adults. However, the VA implemented MRSA ASCs as routine care in 2007 soon after the PI received notification of funding, resulting in the PI modifying the original study to a pilot project.
Because of lack of high-quality studies and difficulties in conducting randomized control trials for population-level interventions such as ASCs, this pilot study provides important feasibility evidence to consider when conducting these trials in the VA. This pilot study also provides preliminary results of the role of ASC in reducing VRE, an emerging pathogen, and the role of other potentially more important infection control interventions.

OBJECTIVE(S):
The primary aim of this study was to evaluate whether ASCs in the Intensive Care Unit (ICU) can reduce VRE HAIs. Secondary aims were to describe potentially confounding variables and describe the economic impact of the intervention. When the study was modified to a pilot study with a much smaller sample size, we revised the primary aims to focus more on feasibility and a framework to inform future interventions in larger studies.

METHODS:
The study was a multi-site quasi-experimental trial in three ICUs at two VA facilities, Madison and Milwaukee. The VRE pre-intervention phase of the study included baseline data collection at Madison (Jan-2010 to Feb-2011) and at Milwaukee (Jan-Dec 2010). The VRE intervention phase of the study was conducted at Madison (Mar-May 2011 and Nov-2012 to Apr-2013) and at Milwaukee (Jan-2011 to Nov-2012). In 2011, ICU patients providing written informed consent were enrolled in the intervention. In 2012, ICU patients providing verbal consent were enrolled in the intervention. There were no exclusion criteria. All ICU patients were evaluated for the study outcomes regardless of their intervention enrollment status.
In 2011, using written informed consent, a total of 155 subjects were enrolled into the intervention; the enrollment rate was 10%. Reasons for nonenrollment included refusal, inability to consent or transfer prior to swabbing. In 2012, using waiver of written consent, a total of 162 subjects were enrolled into the intervention; the enrollment rate was 63%. Swabs were not obtained because of refusal (24%) or missed (13%). There were a total of 5214 subjects evaluated for study outcomes.
The study variables collected included: 1. Patient-level data (demographics, comorbidities, severity index and outcomes); 2. Unit-level data (hand hygiene and isolation compliance, antibiotic and device utilization); and 3. Direct costs (i.e., nursing time and supplies). Data were collected from the electronic medical record, participating hospitals' reports, and national IPEC.
For VRE, MRSA, and CDiff, we analyzed infection rates per 1000 patient days and infection rates per 100 patient admissions. We used binary logistic regression models to explore potential associations between infection rates with Study Period (Pre-intervention, Intervention, or Post-intervention), Study Site (Madison or Milwaukee), and the following groups of independent variables: infection control compliance (hand hygiene, isolation, glove/gown use), antibiotic usage, and VA Inpatient Evaluation Center (IPEC) data (severity index and length of stay).

FINDINGS/RESULTS:
Unit-level analysis: For VRE infections per 1000 patient days, one predictor was associated with at least borderline statistical significance--Third generation cephalosporin usage (p=0.070). For VRE infections per 100 patient admissions, three predictors trending toward statistical significance were identified: First-generation cephalosporin usage (p=0.053), Glycopeptide usage (p=0.074), and Length of Stay (p=0.077).
Patient-level analysis: On preliminary multiple logistic regression analysis, there were three statistically significant risk factors for VRE colonization on admission: Regional Site (Madison or Milwaukee), History of C-Diff, and Severe Renal Disease. The odds in favor of VRE on Admission at Madison are 4.4 times as large as the odds in favor of VRE on Admission at Milwaukee; p=0.007. The odds in favor of VRE on admission are 5.6 times as great for patients with severe renal disease than for patients without severe renal disease; p=0.023. The odds in favor of VRE on admission are 11.2 times as great for patients with a history of C-Diff than for patients who do not have such a history; p=0.030.
Conclusions: Our study has several important findings. First, for infection control interventions that impact the entire unit (e.g., ASCs and MDRO transmission), individual informed consent is not recommended because lack of participation by even a small number of subjects does not allow the research question to be answered. Second, in the veteran population, perirectal swabs for VRE detection have moderate acceptability, as shown by our findings that even with waiver of consent, only 63% of subjects participated. Third, cluster randomized trials are important for emerging pathogens such as VRE, and our study provides a framework and an estimation of effect size to undertake such a study. Fourth, we have identified that antibiotic use is an important predictor of VRE acquisition. In this regard, our work complements the VA MDRO Program office's efforts to focus on antibiotic stewardship as a major intervention for infection prevention.

IMPACT:
ASCs for MDRO identification and isolation are an expensive and complex intervention. In this pilot study, we identified that antibiotic stewardship may be more important for prevention of VRE than ACS.
We provide an approved reference study with waiver of consent and a framework for implementing future cluster randomized trials of population-level infection control interventions.

PUBLICATIONS:

Conference Presentations

  1. McKinley L, Jucovy P, Moriarty H, Safdar N. Institutional Review Board Considerations for Cluster Trials in Infection Control. Paper presented at: Association for Professionals in Infection Control and Epidemiology Annual Conference; 2014 Jun 1; Fort Lauderdale, FL.
  2. McKinley L, Moriarty H, Short TH, Hagle M, Ranum A, Valentine S, Safdar N. Regional Differences in Vancomycin-resistant enterococcus and C. difficille. Paper presented at: Association for Professionals in Infection Control and Epidemiology Annual Conference; 2013 Jun 1; Fort Lauderdale, FL.
  3. McKinley L, Moriarty H, Short TH, Hagle M, Ranum A, Valentine S, Safdar N. Regional Differences in Vancomycin-resistant enterococcus and C difficile co-Colonization Rates in Critically Ill Veterans. Paper presented at: Infectious Diseases Week; 2012 Oct 18; San Diego, CA.


DRA: Health Systems, Infectious Diseases
DRE: Epidemiology, Prevention
Keywords: Best Practices, Nursing, Practice patterns, Quality assurance, improvement
MeSH Terms: none