Nosocomial bloodstream infections (BSI) due to vancomycin-resistant enterococci (VRE) represent an important infection control issue, as the burden of VRE infections has increased dramatically over the last few decades.Reference Chiang, Perencevich and Nair ¹ In a study of United States hospitals, the incidence of hospitalizations due to VRE doubled from 2000–2003 to 2003–2006.Reference Ramsey and Zilberberg ² Among Canadian hospitals, the prevalence of VRE infections tripled from 2007 to 2013.Reference Simner, Adam and Baxter ³ Other studies have reported similar increases in the incidence of VRE-BSI.Reference McKinnell, Patel, Shirley, Kunz, Moser and Baddley ^{4 –} Reference Ammerlaan, Harbarth and Buiting ⁶ Although these collective data suggest an increase in the burden of VRE infection, more current data from a national sample are needed. In this study, we aimed to quantify and evaluate trends in the incidence of nosocomial VRE-BSI in the Veterans Affairs (VA) healthcare system.

We conducted a national retrospective study of adult hospitalized patients admitted to any VA hospital between fiscal years (FY) 2004 and 2014 (ie, October 1, 2003, through September 30, 2014). Microbiologically confirmed cases were identified from inpatient clinical data from the VA Corporate Data Warehouse and were defined as (1) having ≥1 blood culture positive for an Enterococcus species demonstrating resistance to vancomycin as reported by institutional susceptibility testing results obtained during routine clinical care and (2) hospital admission ≥48 hours at the time of first positive blood culture. Additional data collected included patient demographics (eg, age, gender) and setting of onset (ie, intensive care unit [ICU] vs non-ICU). Age-specific hospitalization estimates were derived from Veterans Health Administration Support Service Center data. Cases from non–acute-care admissions were excluded from analysis. Age-specific incidences (per 10,000 hospitalizations and per 10,000 patient days) were computed per fiscal year and were analyzed over time by linear regression. Statistical analysis was performed using Prism version 7 software (GraphPad, La Jolla, California); P < .05 was considered statistically significant. The Kansas City VA institutional review board approved this study.

Over the study period, 4,572 cases of nosocomial VRE-BSI were observed over 7,269,927 hospitalizations (6.29 per 10,000 hospitalizations) and 46,732,419 patient days (0.98 per 10,000 patient days). In total, 114 hospitals across all 50 United States, the District of Columbia, and Puerto Rico contributed cases. Among these cases, 3,658 (80.0%) were due to Enterococcus faecium, 388 (8.5%) were due to Enterococcus faecalis, and 526 (11.5%) were due to other Enterococcus or unspecified species. Most cases occurred in a non-ICU setting (n=2,755; 60.3%). The median age was 65 years (interquartile range, 59–76 years) and 4,455 (97.4%) were male.

We detected notable differences in yearly age-specific incidence rates of nosocomial VRE-BSI over the study period (Table 1). Overall incidence increased during the first 5 years of study, peaking in FY2007–2008 (8.95 per 10,000 hospitalizations; 1.29 per 10,000 patient days; slope for linear trend for FY2004–2008, 0.06; r ²=0.79). The incidence of VRE-BSI steadily declined over the remainder of the study period and the lowest rates were observed in FY2014 (2.84 per 10,000 hospitalizations; 0.52 per 10,000 patient days; slope for linear trend for FY2008–2014, −0.12; r ²=0.9). Overall, the incidence density of nosocomial VRE-BSI significantly decreased over the study period (slope for linear trend for FY2004–2014, −0.07; 95% confidence interval, −0.10 to −0.03; r ²=0.69; P=.002). Similar year-by-year trends in incidence rates were observed consistently across all age groups, with the highest incidence observed in those aged 75–84 years.

TABLE 1 Age-Specific Incidence Rates of Nosocomial Vancomycin-Resistant Enterococcus (VRE) Bloodstream Infections Among United States Veterans Affairs Hospitals, Fiscal Years 2004–2014

In this national retrospective study of VA hospitals, we noted significant changes in the epidemiology of nosocomial VRE-BSI from FY2004 to FY2014. Consistent with previous data, we observed increasing incidence rates of VRE-BSI in the first few years of study.Reference Ramsey and Zilberberg ^{2 –} Reference McCracken, Wong and Mitchell ⁷ However, we report a marked decrease in the burden of nosocomial VRE-BSI over the last 5 years of study. The epidemiology of VRE infections was recently reviewed in a meta-analysis of multicenter studies.Reference Chiang, Perencevich and Nair ¹ In contrast to the present analysis, most studies reported increasing incidences of VRE infections, including BSI.Reference Chiang, Perencevich and Nair ¹ However, most of these investigations did not include data more recent than 2011.

The etiology for the decline we observed in nosocomial VRE-BSI incidence is uncertain. Similar reductions in methicillin-resistant Staphylococcus aureus (MRSA) and gram-negative bacillus BSI rates have recently been described in the VA population; they have been attributed to an expanded MRSA infection prevention initiative implemented across the healthcare system in early 2007.Reference Jain, Kralovic and Evans ^{8 ,} Reference Goto, O’Shea and Livorsi ⁹ This infection control program featured a number of horizontal infection control interventions that could have potentially reduced the incidence of other hospital-acquired infections.Reference Jain, Kralovic and Evans ⁸ Complete implementation of this initiative was mandated for all VA hospitals by FY2008, and the program details have been previously reviewed.Reference Jain, Kralovic and Evans ⁸ Consistent with this theory, a significant reduction in voluntary reporting of VRE infections was noted from 2007 to 2010 in 33 VA hospitals following the implementation of this infection control program.Reference Jain, Kralovic and Evans ⁸

Although the use of a national sample and microbiological confirmation of cases represent significant strengths of this study, the study had several limitations. This was a study of the VA population, which consists largely of Caucasian males of advanced age with a high comorbidity burden and poorer health than the general population.Reference Agha, Lofgren, VanRuiswyk and Layde ¹⁰ Therefore, the results of this study may not be generalizable to dissimilar populations. This study was not designed to evaluate potential etiologies for the changing epidemiology observed and further study is warranted.

In summary, we report a significant decrease in the national incidence of nosocomial VRE-BSI in recent years. This finding contrasts with older data suggesting that the incidence of VRE infection has been increasing. Whether the incidence of nosocomial VRE-BSI outside the VA system is similarly decreasing warrants further study.