Vancomycin-resistant enterococci (VRE) infections result in increased hospital costs and lengths of stay.Reference Butler, Olsen, Merz and Guth 1 Previously, we described a hospital-wide “reflex” testing program for detecting VRE intestinal colonization as an intervention to limit nosocomial transmission.Reference Bodily, McMullen and Russo 2 Inpatients that had diarrheal stools submitted to the clinical microbiology laboratory for Clostridium difficile (C. difficile) toxin testing also had screening cultures for VRE. In July 2010, this program was discontinued, and in the following 18 months, the monthly incidence of healthcare-associated VRE increased by 71%.
Based on these findings, the VRE reflex testing program was reinstated in January 2012. We examined the effect of reinstatement of the VRE reflex testing program on healthcare-associated VRE incidence.
METHODS
Barnes-Jewish Hospital is a 1,250-bed academic tertiary care hospital in Saint Louis, Missouri. We examined the healthcare-associated VRE rate between January 2009 and December 2015. Reflex testing for VRE was discontinued in July 2010 and was reinstated in January 2012. Clinicians were notified of these changes to VRE reflex testing. Throughout the study period, clinicians could order stool or perirectal cultures for VRE testing at their discretion. The hospital policy during the study was to place VRE colonized or infected patients on contact precautions in a private room. During the reflex screening period 1 (January 2009 to July 2010), stool specimens were plated on Enterococcosel Agar (Becton Dickinson, Sparks, MD). During screening period 2 (January 2012 to December 2015), chromID VRE (bioMerieux, Durham, NC) was used. No differences in the screening cultures performed that were positive for VRE were detected for either screening period: 2,457 of 9,637 (25.5%) for period 1 versus 6,289 of 24,920 (25.2%) for period 2 (P=.62).
All hospitalized patients with a positive urine or blood culture for VRE were identified. A healthcare-associated VRE case was defined as the first positive specimen per patient, where VRE was detected in blood or urine ≥3 calendar days after admission. A VRE case was considered present on admission if VRE was detected in blood or urine <3 calendar days after admission. 3 In the original study, we defined healthcare-associated VRE as an initial positive culture within 48 hours after admission. The VRE incidence was expressed as cases per 10,000 patient days.
The effect of reflex testing on the incidence of healthcare-associated VRE was evaluated using regression models with autoregressive integrated moving average (ARIMA) errors. In addition to reflex testing, VRE prevalence on admission, central line utilization, urinary catheter utilization, and overall temporal trend were considered in the model to evaluate the effects of these factors on healthcare-associated VRE rate during the study period. The Washington University Human Research Protection Office approved this study.
RESULTS
In the initial reflex testing period, there were 99 cases of healthcare-associated VRE; 36 cases (36.4%) were identified via blood cultures and 63 cases (63.6%) were identified via urine cultures, for an overall rate of 2.3 VRE cases per 10,000 patient days. During the period when reflex testing was discontinued, there were 166 cases of healthcare-associated VRE; 63 cases (38.0%) were identified via blood cultures and 103 cases (62.0%) were identified via urine cultures, for an overall rate of 3.7cases per 10,000 patient days. During the study period after reflex testing was reinstated, there were 218 cases of VRE; 57 cases (26.1%) were identified via blood cultures and 161 cases (73.9%) were identified via urine cultures, for an overall rate of 1.8 cases per 10,000 patient days (Figure 1). Accounting for the overall temporal trend, healthcare-associated VRE decreased at a rate of 36.4% (−36.4; 95% CI, −50.9 to −17.7) per month over the entire study period. The average healthcare-associated VRE rate was 1.9 cases per 10,000 patient days during the reflex testing periods versus 3.7 cases per 10,000 patient days during the nontesting period. When adjusted for VRE prevalence on admission and overall temporal trend, reflex testing was associated with a 32.0% reduction (−32.0; 95% CI, −48.2 to −10.8) in healthcare-associated VRE incidence compared to the non–reflex testing period.
FIGURE 1 Monthly incidence of healthcare-associated vancomycin resistant enterococci (VRE) infections, January 2009 - December 2015
DISCUSSION
In this follow-up study, we noted that hospital-wide reinstitution of a VRE reflex screening program led to a decrease in hospital-associated VRE incidence to a baseline rate similar to pre-discontinuation period, which further suggested a causal relationship between the program and a reduction in hospital-associated VRE transmission.Reference Hill 4 Most hospitals, often due to the lack of routine active surveillance, do not identify a large proportion of colonized patients who are potential sources for ongoing hospital transmission.Reference Muto, Jernigan and Ostrowsky 5 , Reference Siegel, Rhinehart, Jackson and Chiarello 6 Some studies have demonstrated the benefit of active surveillance cultures to control VRE transmission in hospitals; however, these were generally conducted during an outbreak in which multiple interventions were introduced simultaneously.Reference Calfee, Gianetta, Durbin, Germanson and Farr 7 , Reference Lucet, Armand-Lefevre and Laurichesse 8 In our previous study, we found that discontinuation of reflex VRE testing of each stool sample submitted for testing for C. difficile at our hospital resulted in a 71% increase in the endemic healthcare-associated VRE rate (Figure 1). This long-term, follow-up analysis indicated that when reflex VRE testing was reimplemented, healthcare-associated VRE rates returned to the prediscontinuation baseline.
Our study has several limitations. Our data were collected from a single care center and did not include patient-level characteristics. Another concern is the change in the screening method of stool specimens. However, the percentage of positive screening tests did not change significantly. The strengths of our study include repeated treatment designReference Harris, McGregor and Perencevich 9 and the use of regression models with ARIMA errors to appropriately account for correlated observations over time and to adjust for the overall temporal trend and important confounders such as VRE prevalence on admission. Furthermore, no other specific infection prevention measures were implemented over the study period, and laboratory methods for identifying VRE from urine and blood cultures did not change. Also, there were no outbreaks of C. difficile during this period.
In conclusion, we found that the use of VRE reflex testing of stool submitted for C. difficile testing was effective in reducing the incidence of hospital-associated VRE infections when combined with a contact precautions program. This strategy should be considered a valid method of reducing VRE transmission in hospital settings.
ACKNOWLEDGMENTS
Financial support: This study was funded in part by a grant from the CDC Prevention Epicenter Program (grant no. 1U54CK000162) to D.K.W.
Potential conflicts of interest: All authors report no conflicts of interest relevant to this article.
