Background: In April 2019, Nebraska Public Health Laboratory identified an NDM-producing Enterobacter cloacae from a urine sample from a rehabilitation inpatient who had recently received care in a specialized unit (unit A) of an acute-care hospital (ACH-A). After additional infections occurred at ACH-A, we conducted a public health investigation to contain spread. Methods: A case was defined as isolation of NDM-producing carbapenem-resistant Enterobacteriaceae (CRE) from a patient with history of admission to ACH-A in 2019. We conducted clinical culture surveillance, and we offered colonization screening for carbapenemase-producing organisms to all patients admitted to unit A since February 2019. We assessed healthcare facility infection control practices in ACH-A and epidemiologically linked facilities by visits from the ICAP (Infection Control Assessment and Promotion) Program. The recent medical histories of case patients were reviewed. Isolates were evaluated by whole-genome sequencing (WGS). Results: Through June 2019, 7 cases were identified from 6 case patients: 4 from clinical cultures and 3 from 258 colonization screens including 1 prior unit A patient detected as an outpatient (Fig. 1). Organisms isolated were Klebsiella pneumoniae (n = 5), E. cloacae (n = 1), and Citrobacter freundii (n = 1); 1 patient had both NDM-producing K. pneumoniae and C. freundii. Also, 5 case patients had overlapping stays in unit A during February–May 2019 (Fig. 2); common exposures in unit A included rooms in close proximity, inhabiting the same room at different times and shared caregivers. One case-patient was not admitted to unit A but shared caregivers, equipment, and devices (including a colonoscope) with other case patients while admitted to other ACH-A units. No case patients reported travel outside the United States. Screening at epidemiologically linked facilities and clinical culture surveillance showed no evidence of transmission beyond ACH-A. Infection control assessments at ACH-A revealed deficiencies in hand hygiene, contact precautions adherence, and incomplete cleaning of shared equipment within and used to transport to/from a treatment room in unit A. Following implementation of recommended infection control interventions, no further cases were identified. Finally, 5 K. pneumoniae of ST-273 were related by WGS including carriage of NDM-5 and IncX3 plasmid supporting transmission of this strain. Further analysis is required to relate IncX3 plasmid carriage and potential transmission to other organisms and sequence types identified in this study. Conclusions: We identified a multiorganism outbreak of NDM-5–producing CRE in an ACH specialty care unit. Transmission was controlled through improved infection control practices and extensive colonization screening to identify asymptomatic case-patients. Multiple species with NDM-5 were identified, highlighting the potential role of genotype-based surveillance.

Funding: None

Disclosures: Muhammad Salman Ashraf reports that he is the principal investigator for a study funded by an investigator-initiated research grant.

Background: Nearly half of hospitalized patients with bacteriuria or treated for pneumonia receive unnecessary antibiotics (noninfectious or nonbacterial syndrome such as asymptomatic bacteriuria), excess duration (antibiotics prescribed for longer than necessary), or avoidable fluoroquinolones (safer alternative available) at hospital discharge.1–3 However, whether antibiotic overuse at discharge varies between hospitals or is associated with patient outcomes remains unknown. Methods: From July 2017 to December 2018, trained abstractors at 46 Michigan hospitals collected detailed data on a sample of adult, non–intensive care, hospitalized patients with bacteriuria (positive urine culture with or without symptoms) or treated for community-acquired pneumonia (CAP; includes those with the disease formerly known as healthcare-associated pneumonia [HCAP]). Antibiotic prescriptions at discharge were assessed for antibiotic overuse using a previously described, guideline-based hierarchical algorithm.3 Here, we report the proportion of patients discharged with antibiotic overuse by the hospital. We also assessed hospital-level correlation (using Pearson’s correlation coefficient) between antibiotic overuse at discharge for patients with bacteriuria and patients treated for CAP. Finally, we assessed the association of antibiotic overuse at discharge with patient outcomes (mortality, readmission, emergency department visit, and antibiotic-associated adverse events) at 30 days using logit generalized estimating equations adjusted for patient characteristics and probability of treatment. Results: Of 17,081 patients (7,207 with bacteriuria; 9,874 treated for pneumonia), nearly half (42.2%) had antibiotic overuse at discharge (36.3% bacteriuria and 51.1% pneumonia). The percentage of patients discharged with antibiotic overuse varied 5-fold among hospitals from 14.7% (95% CI, 8.0%–25.3%) to 74.3% (95% CI, 64.2%–83.8%). Hospital rates of antibiotic overuse at discharge were strongly correlated between bacteriuria and CAP (Pearson’s correlation coefficient, 0.76; P ≤ .001) (Fig. 1). In adjusted analyses, antibiotic overuse at discharge was not associated with death, readmission, emergency department visit, or Clostridioides difficile infection. However, each day of overuse was associated with a 5% increase in the odds of patient-reported antibiotic-associated adverse events after discharge (Fig. 2). Conclusions: Antibiotic overuse at discharge was common, varied widely between hospitals, and was associated with patient harm. Furthermore, antibiotic overuse at discharge was strongly correlated between 2 disparate diseases, suggesting that prescribing culture or discharge processes—rather than disease-specific factors—contribute to overprescribing at discharge. Thus, discharge stewardship may be needed to target multiple diseases.

Funding: This study was supported by the Society for Healthcare Epidemiology of America and by Blue Cross Blue Shield of Michigan and Blue Care Network.

Disclosures: Valerie M. Vaughn reports contracted research for Blue Cross and Blue Shield of Michigan, the Department of VA, the NIH, the SHEA, and the APIC. She also reports receipt of funds from the Gordon and Betty Moore Foundation Speaker’s Bureau, the CDC, the Pew Research Trust, Sepsis Alliance, and the Hospital and Health System Association of Pennsylvania.

Background: Approximately two-thirds of children aged <5 years receive out-of-home child care. Childcare attendees have an increased risk of infections compared to children not in childcare settings, possibly due to their close contact in a shared environment. As multidrug-resistant organisms (MDROs) increasingly move from healthcare-associated to community settings, childcare can provide a venue for further transmission of these pathogens. Our objective was to evaluate the bioburden of pathogens present on fomites in childcare centers and how surface contamination changes over time. Methods: The study was conducted in the single-room play area of an Ypsilanti, Michigan, childcare center caring for children aged 3–5 years. Polyester swabs were used to collect surface samples from 16 locations in the room, including (1) laminate, wood and plastic tabletops and furniture; (2) a stainless steel sink and adjacent plastic trash bin; and (3) wood, metal and plastic toys. A water sample was also collected at a 17th site. Samples were collected twice weekly for 5 of 6 weeks, followed by 1 additional collection (September–October 2019). Tryptic soy agar was used for standard plate counts and selective media were used to identify methicillin-resistant Staphylococcus aureus (MRSA), Vvancomycin-resistant Enterococcus (VRE), and extended-spectrum β-lactamase (ESBL)–producing Enterobacteriaceae. Single-plex RT-PCR was used to detect norovirus and adenovirus. Results: Among 175 samples collected on 11 days, MRSA and ESBL-producing Enterobacteriaceae were detected from 10.3% (18 of 175) and 8.0% (14 of 175), respectively, of environmental specimens. No specimens were positive for VRE or norovirus. Adenovirus was detected in 20 of 175 specimens (11.4%). Median bioburden by site ranged from 85 CFU/mL to 2,510 CFU/mL. The highest median bioburden was observed at the sink (2,510 CFU/mL), followed by the plastic building block table (1,620 CFU/mL), the small wood blocks (1,565 CFU/mL) and water from a water play area and an adjacent tabletop (1,260 and 1,100 CFU/mL respectively). The highest single day bioburden was 273,000 CFU/mL at the sink. Conclusion: The presence of MDROs on childcare center fomites raised concern for exposure to these pathogens among vulnerable populations. More study is needed to determine the degree to which these contaminated fomites drive transmission between children. We found the highest bioburdens on sites where children played or washed with water, identifying potential targets for more frequent cleaning.

Funding: None

Disclosures: Emily T. Martin reports a consulting from Pfizer.

Background: Historically, metronidazole was first-line therapy for Clostridioides difficile infection (CDI). In February 2018, the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA) updated clinical practice guidelines for CDI. The new guidelines recommend oral vancomycin or fidaxomicin for treatment of initial episode of CDI in adults. We examined the changes in treatment of CDI during 2018 across all types of healthcare settings in metropolitan Atlanta. Methods: Cases were identified through the Georgia Emerging Infections program (funded by the Centers for Disease Control and Prevention), which conducts active population-based surveillance in an 8-county area including Atlanta, Georgia (population, 4,126,399). An incident case was a resident of the catchment area with a positive C. difficile toxin test and no additional positive test in the previous 8 weeks. Recurrent CDI was defined as >1 incident CDI episode in 1 year. Clinical and treatment data were abstracted on a random 33% sample of adult (>17 years) cases. Definitive treatment categories were defined as the single antibiotic agent, metronidazole or vancomycin, used to complete a course. We examined the effect of time of infection, location of treatment, and number of CDI episodes on the use of metronidazole only. Results: We analyzed treatment information for 831 adult sampled cases. Overall, cases were treated at 29 hospitals (568 cases), 4 nursing homes (6 cases), and 101 outpatient providers (257 cases). The mean age was 60 (IQR, 34–86), and 111 (13.4%) had recurrent infection. Moreover, ∼28% of first-incident CDI episodes, 8% of second episodes, and 6% of third episodes were treated with metronidazole only. Compared to facility-based providers, outpatient providers were more likely to treat initial CDI episodes with metronidazole only (44% vs 21%; relative risk [RR], 2.1; 95% CI, 1.7–2.7). Treatment changed over time from 56% metronidazole only in January to 10% in December (Fig. 1). First-incident cases in the first quarter of 2018 were more likely to be treated with metronidazole only compared to those in the fourth quarter (RR, 2.76; 95% CI, 1.91–3.97). Conclusions: Preferential use of vancomycin for initial CDI episodes increased throughout 2018 but remained <100%. CDI episodes treated in the outpatient setting and nonrecurrent episodes were more likely to be treated with metronidazole only. Additional studies on persistent barriers to prescribing oral vancomycin, such as cost, are warranted.

Funding: None

Disclosures: Scott Fridkin reports that his spouse receives a consulting fee from the vaccine industry.

Background: Carbapenem-resistant Acinetobacter baumannii (CRAB) is a serious threat to patient safety due to limited treatment options and propensity to spread in healthcare settings. Using Emerging Infections Program (EIP) data, we describe changes in CRAB incidence and epidemiology. Methods: During January 2012 to December 2018, 9 sites (Colorado, Connecticut, Georgia, Maryland, Minnesota, New Mexico, New York, Oregon, and Tennessee) participated in active laboratory- and population-based surveillance. An incident case was defined as the first isolation of A. baumannii complex, in a 30-day period, resistant to ≥1 carbapenem (excluding ertapenem) from a normally sterile site or urine of a surveillance area resident. Cases were considered hospital-onset (HO) if the culture was collected >3 days after hospital admission; all others were community-onset (CO). Cases were classified as device-associated (DA) if the patient had 1 or more medical devices (ie, urinary catheter, central venous catheter (CVC), endotracheal/nasotracheal tube, tracheostomy, or another indwelling device) present in the 2 days prior to culture collection. Temporal trends were estimated using generalized linear models adjusted for age, race, sex, and EIP site. Results: Overall, 984 incident CRAB cases were identified, representing 849 patients. Among these patients, 291 (34%) were women, 510 (61%) were nonwhite, and the median age was 62 years (mean, 59; range, 0–102). Among the cases, 226 (23%) were HO; 758 (77%) were CO; and 793 (81%) were DA. Overall incidence rates in 2012 and 2018 were 1.58 (95% CI, 1.29–1.90) and 0.60 (95% CI, 0.40–0.67) per 100,000 population, respectively. There was a 15% annual decrease in incidence (adjusted rate ratio [aRR] 0.85; 95% CI: 0.82-0.88, P < .0001). Decreases were observed among sterile site (aRR 0.88; 95% CI, 0.84–0.93) and urine cases (aRR 0.83; 95% CI, 0.80–0.87). Annual decreases occurred for HO cases (aRR, 0.78; 95% CI, 0.73–0.85) and CO cases (aRR, 0.86; 95% CI, 0.83–0.9). The DA cases decreased 16% annually overall (aRR, 0.84; 95% CI, 0.81–0.88). Decreases among cases in patients with CVC (aRR, 0.85; 95% CI, 0.80–0.90) and urinary catheters (aRR, 0.84; 95% CI, 0.80–0.88) were smaller than what was seen in patients with other indwelling devices (aRR, 0.81; 95% CI, 0.77–0.86). Discussion: Overall, from 2012 to 2018, the incidence of CRAB decreased >60%. Decreases were observed in all case groups, regardless of source, infection onset location, or types of devices. Smaller annual decreases in rates of CO-CRAB than HO-CRAB suggest that there may be opportunities to accelerate prevention outside the hospital to further reduce the incidence of these difficult-to-treat infections.

Funding: None

Disclosures: None

Background: Carbapenem-resistant Enterobacteriaceae (CRE), particularly carbapenemase-producing (CP) CRE, pose a major public health threat. In 2016, the phenotypic definition of CRE expanded to include ertapenem resistance to improve sensitivity for detecting CP-CRE. We compared characteristics of CRE resistant to ertapenem only (CRE-EO) to CRE resistant to ≥1 other carbapenem (CRE-O). Methods: The Georgia Emerging Infections Program performs active, population-based CRE surveillance in metropolitan Atlanta. CRE cases were defined as any Escherichia coli, Klebsiella pneumoniae, K. oxytoca, K. variicola, Enterobacter cloacae complex, or Enterobacter aerogenes resistant to ≥1 carbapenem by the clinical laboratory and isolated from urine or a sterile site between 2016 and 2018. Data were extracted from retrospective chart review and 90-day mortality from Georgia vital statistics for 2016–2017. Polymerase chain reaction (PCR) for carbapenemase genes was performed on a convenience sample of isolates by the CDC or Georgia Public Health Laboratory. We compared characteristics of CRE-EO cases to CRE-O cases using χ2 tests or t tests. Results: Among 927 CRE isolates, 553 (60%) were CRE-EO. CRE-EO were less frequently isolated from blood (5% vs 12%; P < .01) and less commonly K. pneumoniae (21% vs 58%; P < .01) than CRE-O. CRE-EO cases were more often women (65% vs 50%; P < .01), had a lower Charlson comorbidity index (mean ± SD, 2.4±2.3 vs 3.0±2.6; P < .01), and were less commonly at a long-term care facility (24% vs 31%) or hospital (15% vs 21%; P < .01) in the 4 days prior to the CRE culture. CRE-EO were more susceptible to all antibiotics tested at the clinical laboratory (P < .01) except for tigecycline (P = 1.0) (Table 1). Of the 300 (32%) isolates tested for carbapenemase genes, 98 (33%) were positive (7% CRE-EO vs 62% CRE-O; P < .01). Of the CP isolates, we identified blaKPC in 93 cases (95%), blaNDM in 3 cases (3%), blaOXA-48-like in 2 cases (2%). CRE-EO cases had lower 90-day mortality (13% vs 21%; P < .01). Conclusions: CRE-EO are epidemiologically distinct from CRE-O and are less likely to harbor carbapenemase genes. CRE-EO may require less intensive infection prevention interventions and have more therapeutic options.

Funding: None

Disclosures: None

Background: Nosocomial central-line–associated bloodstream infections (CLABSIs) are an important cause of morbidity and mortality in hospitalized patients. CLABSI surveillance establishes rates for internal and external comparison, identifies risk factors, and allows assessment of interventions. Objectives: To determine the frequency of CLABSIs among adult patients admitted to intensive care units (ICUs) in CNISP hospitals and evaluate trends over time. Methods: CNISP is a collaborative effort of the Canadian Hospital Epidemiology Committee, the Association of Medical Microbiologists and Infectious Disease Canada and the Public Health Agency of Canada. Since 1995, CNISP has conducted hospital-based sentinel surveillance of healthcare-associated infections. Overall, 55 CNISP hospitals participated in ≥1 year of CLABSI surveillance. Adult ICUs are categorized as mixed ICUs or cardiovascular (CV) surgery ICUs. Data were collected using standardized definitions and collection forms. Line-day denominators for each participating ICU were collected. Negative-binomial regression was used to test for linear trends, with robust standard errors to account for clustering by hospital. We used the Fisher exact test to compare binary variables. Results: Each year, 28–42 adult ICUs participated in surveillance (27–37 mixed, 6–8 CV surgery). In both mixed ICUs and CV-ICUs, rates remained relatively stable between 2011 and 2018 (Fig. 1). In mixed ICUs, CLABSI rates were 1.0 per 1,000 line days in 2011, and 1.0 per 1,000 line days in 2018 (test for linear trend, P = .66). In CV-ICUs, CLABSI rates were 1.1 per 1,000 line days in 2011 and 0.8 per 1,000 line days in 2018 (P = .19). Case age and gender distributions were consistent across the surveillance period. The 30-day all-cause mortality rate was 29% in 2011 and in 2018 (annual range, 29%–35%). Between 2011 and 2018, the percentage of isolated microorganisms that were coagulase-negative staphylococci (CONS) decreased from 31% to 18% (P = .004). The percentage of other gram-positive organisms increased from 32% to 37% (P = .34); Bacillus increased from 0% to 4% of isolates and methicillin-susceptible Staphylococcus aureus from 2% to 6%). The gram-negative organisms increased from 21% to 27% (P = .19). Yeast represented 16% in 2011 and 18% in 2018; however, the percentage of yeast that were Candida albicans decreased over time (58% of yeast in 2011 and 30% in 2018; P = .04). Between 2011 and 2018, the most commonly identified species of microorganism in each year were CONS (18% in 2018) and Enterococcus spp (18% in 2018). Conclusions: Ongoing CLABSI surveillance has shown stable rates of CLABSI in adult ICUs from 2011 to 2018. The causative microorganisms have changed, with CONS decreasing from 31% to 18%.

Funding: CNISP is funded by the Public Health Agency of Canada.

Disclosures: Allison McGeer reports funds to her for studies, for which she is the principal investigator, from Pfizer and Merck, as well as consulting fees from Sanofi-Pasteur, Sunovion, GSK, Pfizer, and Cidara.

Background: The relationship between nursing staffing and healthcare-associated infections (HAIs) has been explored previously, with conflicting results. Intensive care units increasingly struggle to maintain trained staff. In May 2019, clinical coordinator (CC) roles changed to include 50% of time in direct patient care rather than supportive roles. In this study, we used shift records to explore the impact of staffing on HAI risk. Methods: Daily staffing records from December 2018 August 2019 for the medical-respiratory unit (MRICU) and the cardiac surgery unit (CSICU) were reviewed. Both units staff a fixed 2:1 patient:nurse ratio (1:1 for specific cardiac surgeries). Staff deficiency was defined as assignments filled by nurses pulled from other units/supplemental/or CC roles. Staff support comprised nursing assistants and unit secretaries. Census, admissions, and complexity score for number of devices were used to estimate care acuity. In CSICU, additional points were added for continuous renal replacement therapy, extracorporeal membrane oxygenation, ventricular assist devices, transplant, operative cases. NHSN definitions were used for central-line–associated bloodstream infections (CLABSIs) and catheter-associated urinary tract infections (CAUTIs). The Spearman correlation coefficient was used to determine relationship between staffing, acuity, and risk window for HAI (days 1–10 preinfection). Linear regression was used to determine whether staffing deficiencies and/or support associate with the risk window prior to HAI. The final model included census and complexity score as control variables. The statistical analysis was performed using SAS version 9.4 software (Cary, NC). Results: Overall, 8 HAIs occurred in the study period: medical-respiratory intensive care unit (MRICU: 3 CAUTIs and 1 CLABSI) and cardiac surgery intensive care unit (CSICU: 1 CAUTI and 3 CLABSIs). Staffing and census fluctuated daily (Table 1). Total number of nurses correlated with complexity scores (r = 0.35; P < .0001) and daily census (r = 0.31; P < .0001) in the CSICU, and the census (r = 0.12; P = .04) in the MRICU. Nursing deficiencies correlated with days 1–10 before infection (r = 0.20; P = .0013) in the CSICU. In the regression model for the CSICU, nursing deficiencies increased in the time prior to HAI (P = .004), and support staff decreased in the time prior to HAI (P = .034) while controlling for census and complexity. These relationships were not significant in the MRICU. Conclusion: The lack of core nurses to support the staffing structure in CSICU correlated with periods prior to CLABSI or CAUTI in this small, unit-based study. Failure to recruit and retain highly skilled core staff may produce HAI risks, particularly for CLABSI in specialized units.

Funding: None

Disclosures: Michelle Doll, Research Grant from Molnlycke Healthcare

Background: Carbapenem-resistant Enterobacteriaceae (CRE) are a major public health problem. Ceftazidime-avibactam (CZA) is a treatment option for CRE approved in 2015; however, it does not have activity against isolates with metallo-β-lactamases (MBLs). Emerging resistance to CZA is a cause for concern. Our objective was to describe the microbiologic and epidemiologic characteristics of CZA-resistant (CZA-R) CRE. Methods: From 2015 to 2017, 9 states participated in laboratory- and population-based surveillance for carbapenem-resistant Escherichia coli, Klebsiella pneumoniae, K. oxytoca, K. aerogenes, and Enterobacter cloacae complex isolates from a normally sterile site or urine. A convenience sample of isolates from this surveillance were sent to the CDC for antimicrobial susceptibility testing (AST) using reference broth microdilution (BMD) including an MBL screen, species confirmation with MALDI-TOF, and real-time PCR to detect blaKPC, blaNDM, and blaOXA-48–like genes. Additional AST by BMD was performed on CZA-R isolates using meropenem-vaborbactam (MEV), imipenem-relebactam (IMR), plazomicin (PLZ), and eravacycline (ERV). Epidemiologic data were obtained from a medical record review. Community-associated cases were defined as having no healthcare exposures in the year prior to culture, no devices in place 2 days prior to culture, and culture collected before calendar day 3 after hospital admission. Data were analyzed in 3 groups: CRE that were CZA-susceptible (CZA-S), CZA-R that were due to blaNDM, and CZA-R without blaNDM. Results: Among 606 confirmed CRE tested with CZA, 33 (5.4%) were CZA-R. Of the CZA-R isolates, 16 (48.5%) harbored a blaNDM gene, of which 2 coharbored blaNDM and blaOXA-48-like genes; 9 (27.3%) harbored only a blaKPC gene. Of the 17 CZA-R isolates without blaNDM, all were MBL screen negative. CZA-R due to blaNDM were more frequently community-associated (43.8%) than CZA-S or CZA-R without blaNDM (11.0% and 5.9%, respectively); a higher percentage of CZA-R cases due to blaNDM also had recent international travel (25%) compared to the other groups (1.8% and 5.9%, respectively). CZA-R without blaNDM were more susceptible to MEV (76%), IMR (71%), PLZ (88%), and ERV (65%) compared to CZA-R due to blaNDM (19%, 6%, 56%, and 44%, respectively). Conclusions: The emergence of CZA-R isolates without blaNDM are concerning; however, these isolates are more susceptible to newer antimicrobials than those with blaNDM. In addition to high rates of resistance to newer antimicrobials, isolates with blaNDM are more frequently community-associated than other CRE. This underscores the need for more aggressive measures to stop the spread of CRE.

Funding: None

Disclosures: None

Background: Carbapenemase-producing Enterobacteriales (CPE) outbreaks have been linked to contaminated wastewater drainage systems in hospitals. The optimal strategy for CPE decontamination of drains is unknown. In this randomized controlled trial, we aimed to determine whether combining chemical, mechanical, and heat cleaning was superior to routine cleaning for drain decontamination. Methods: We enrolled CPE-contaminated hospital drains at 2 geographic locations. Eligible drains were those initially found to be culture positive in a 2017 study and that remained positive (by RT-PCR) when retested twice in August 2018. Drains were stratified by type (sink versus shower) and randomized with a 1:1 allocation ratio (as per computer-generated randomization) to standard-of-care cleaning (comparator) or combined chemical, mechanical, and heat cleaning (intervention) on day 0. Drain tail pieces were swabbed on days 0 (before administration of the intervention), 1, 2, 3, 7, and 14, and at months 1, 2, 3, 4, 5, and 6. Swabs were placed into brain heart infusion with 10% Dey-Engley neutralizing broth and incubated overnight. Direct RT-PCR was performed to detect KPC, VIM, NDM, OXA-48–like, IMP, GES, and SME genes. The primary outcome was drain decontamination, defined as no detectable carbapenemase gene in the drain from day 1 to 7 (inclusive). Results: Overall, 33 CPE-contaminated drains were enrolled (7 sink and 26 shower); 17 and 16 drains were randomized to the intervention and comparator, respectively. Moreover, 12 (36%) drains met the primary outcome of decontamination, 18 (55%) remained contaminated, and 3 (9%) could not be assessed. Among drains that could be assessed, 11 of 15 (74%) in the intervention group met the primary outcome of decontamination compared to 1 of 15 (7%) in the comparator group (P = .0005). Of the 11 drains in the intervention group that were decontaminated, the carbapenemase gene present at enrollment was subsequently detected in 10 (91%): 1 (10%) at day 14, 3 (30%) at month 1, 4 (40%) at month 3, 1 (10%) at month 4, and 1 (10%) at month 6. The median time to a swab yielding CPE was 1 day in the comparator group versus 14 days in the intervention group (Fig. 1). Overall, 24 drains (73%) had a carbapenemase gene (that was not detectable at enrollment) appear in the follow-up. Of patients identified as CPE colonized or infected during this study, none occupied rooms with these drains. Conclusions: Chemical, mechanical, and heat cleaning were superior to standard cleaning for CPE decontamination of hospital drains at 7 days, but these trends were not sustained. Such cleaning may be useful if applied repeatedly.

Funding: None

Disclosures: Allison McGeer reports funds to her institution for studies for which she is the principal investigator from Pfizer and Merck as well as consulting fees from Sanofi-Pasteur, Sunovion, GSK, Pfizer, and Cidara.

Background: Pathogenic Clostridioides difficile is the most common cause of nosocomial infections in the United States. However, the prevalence of C. difficile colonization in the general population is poorly understood. Objective: In this study, we sought to determine the presence and nature of various strains of Clostridioides difficile colonizing a representative sample of 121 asymptomatic adult volunteers from around the globe, consisting of 110 healthy and 11 stable Crohn’s patients. Methods: Next-generation sequencing was performed on fecal samples from 121 study participants. Stool samples were collected by patients utilizing a Zymo collection kit, which preserves bacterial DNA and RNA. Following collection, DNA was extracted, quantitated, and then normalized for downstream library fabrication utilizing shotgun methodology. Prepared and indexed libraries were subsequently pooled and sequenced on the Illumina NextSeq 550 System. Results: All 121 of 121 subjects (100%) were found to possess the bacterium Clostridioides difficile as identified by the NGS bioinformatics metagenomic pipeline. To visualize comparative abundances of Clostridioides difficile present in study participants, normalized read counts were highlighted (Fig. 1). Conclusions: NGS provides a unique opportunity to increase the resolution and identification of Clostridioides difficile compared to traditional categorizations, such as PCR ribotypes (ie, RT027), restriction endonuclease groups (BI), and North American pulsotypes (ie, NAP1). This is accomplished by its ability to differentiate species based on a nucleotides, while targeting entire bacterial genomes. Our approach for this study was to utilize a bioinformatics pipeline that would provide Clostridioides difficile strain-specific resolution when aligning to genomes in the NCBI (National Center for Biotechnology Information) database. In our representative sample of 121 volunteers, all (100%) possessed at least 1 Clostridioides difficile strain in their gut. Although it is recognized that some Clostridioides difficile strains are pathogenic, our findings suggest that nonpathogenic Clostridioides difficile strains make up an important component of the commensal gut microbiome and may perhaps play a protective role. Although symptomatic toxigenic CDI is a clear indication for therapy, Clostridioides difficile colonization with nontoxigenic strains is not believed to be a direct precursor for CDI. These findings demonstrate the need to be aware of the existence of numerous strains of Clostridioides difficile, and the relevance of sequencing prior to hospitalization or antibiotic treatment to help predict those at risk of CDI, and after treatment to be aware of any loss of what appear to be protective components of our microbiome.

Funding: None

Disclosures: Dr. Sabine Hazan reports that she is the founder and CEO of Ventura Clinical Trials and that she and her spouse receive salaries from the company. She also receives a salary from ProgenaBiome.

Background: Blood culture contamination rates are frequently higher than the ≤3% standard in the emergency department (ED). Objective: We sought to determine whether the implementation of a blood diversion device that mechanically sequesters the initial aliquot of the blood culture sample decreased blood culture contamination rates. Methods: We performed a quasi-experimental study in two 500-bed hospitals. The blood-diversion device was implemented in the ED in hospital A, but not in hospital B, starting in January 2018. Preintervention data were collected over a 29-month baseline period, and postintervention data were collected for 20 months. Both hospitals provided ongoing feedback on contamination rates. Blood culture contamination was defined as presence of common skin microbiota (eg, coagulase-negative staphylococci) in only 1 of ≥2 blood culture sets collected within 24 hours. Preintervention and postintervention blood culture contamination rates were calculated based on total blood cultures collected and were compared within and between hospitals using the Wilcoxon rank-sum test. Changes in preintervention and postintervention total and ED contamination rates within hospitals were calculated as rate ratios (RRs) using interrupted time series (ITS) analysis with segmented Poisson regression. Results: Among 212,789 total blood cultures (hospital A, 70,005; hospital B, 142,784), 4,025 (1.8%) were contaminated. In hospital A, the intervention resulted in a decrease in overall median blood culture contamination rates (2.4% vs 1.4%; P < .001) and ED median blood culture contamination rates (4.7% vs 2.6%; P < .001), whereas in hospital B there was no significant change during the same period in overall (2.3% vs 2.0%) or ED (5.0% vs 5.0%) median blood culture contamination rates. In the ITS analysis, the intervention was associated with an immediate decrease in hospital A’s contamination rate by 21.3% (level change RR, 0.79; 95% CI, 0.63–0.98; P = .04) overall and 21.0% (level change RR, 0.79; 95% CI, 0.62–1.0; P =.06) in the ED. After the intervention, there was a continued decrease in hospital A’s overall (trend change RR, 0.95; 95% CI, 0.93–0.97; P < .001) and ED (trend RR, 0.94; 95% CI, 0.92–0.96; P < .001) blood culture contamination rates, but not in hospital B’s overall (trend change RR, 1.02; 95% CI, 1.00–1.02; P = .01) or ED (RR, 1.00; 95% CI, 0.99–1.02; P = .30) blood culture contamination rates during the same period. Conclusions: Implementation of the blood diversion device in the ED resulted in a >20% relative reduction from a baseline of 5% of ED blood culture contamination rates. Continued improvement after implementation suggests ongoing efforts to address the workflow and the culture of safety are needed to optimize the use of this device.

Funding: None

Disclosures: None

Background: Infection control programs (ICPs) are essential to reducing, in a progressive and sustained manner, healthcare-associated infections (HAIs). To achieve this goal these programs need not only resources (ie, trained human resources and financial resources) but also institutional leadership support. In our country, epidemiological surveillance is voluntary and is registered in the Argentina National Hospital Infection Surveillance System (VIHDA) where 114 hospitals and 78 adult intensive care unit (ICU) report their HAI rates. Many of these institutions do not have IPC or specific resources for this purpose. On the other hand, there are institutions with IPC, recognized by an international accreditation like Joint Commission International, that carry out an advanced and continuous program, with specific improvement goals for prevention and infection control. There is an imperative need in low- and middle-income countries to highlight the impact of ICPs in this setting and to promote regulations for mandatory surveillance and ICPs in all acute-care hospitals. Objective: To compare the rates for device-associated infections in ICUs of institutions with advanced ICPs versus national rates. Design: We conducted an observational, retrospective study, which includes device associated infection rates in adult critical care units from 2014 to 2018. We included all ICUs reporting to VIHDA and 3 surgical-medical teaching hospitals with an advanced ICP and Joint Commission International accreditation (Hospital Italiano de Buenos Aires, Hospital Universitario Austral, Hospital Aleman). The VIHDA definition was used to define central line-associated bloodstream infection (CLABSI), catheter-related urinary infection (CAUTI), and ventilator-associated pneumonia (VAP). The rates were compared as adjusted reasons for exposure time using openepi.com software provided by the CDC. Results: Device associated infection rates in hospitals with advanced ICPs and in hospitals in the national surveillance system in Argentina are shown in Table 1. Compliance with infection control measures and bundles for device-associated infections in the 3 hospitals with advanced ICPs was >80%. No data were available for the rest of hospitals included the national surveillance system. Conclusions: Lower infection-control rates, catheter-related bloodstream infection and VAP, are possible in a middle-income country like Argentina when resources are allocated for this purpose and hospital leadership reinforces the efforts. Notably, all 3 hospitals improved their rates over time. The differences in catheter-related bloodstream infection and VAP rates between these hospitals and the rest of the hospitals in our surveillance system was significant and highlights the need for support when it comes to implementing ICPs.

Funding: None

Disclosures: Wanda Cornistein reports fees for conferences not related to this topic from the following speaker’s bureaus: Pfizer, Merck, and Becton Dickinson.

Background: In this study, we assessed whether meropenem restriction led to delays in administration for patients in a medical intensive care unit (MICU) at a large tertiary-care urban teaching hospital. Methods: The antimicrobial stewardship program (ASP) at Virginia Commonwealth University Health System (VCUHS) requires approval for restricted antimicrobial orders placed between 8 a.m. and 9 p.m. Between 8 a.m. and 5 p.m. (daytime), authorized approvers include ASP and infectious diseases (ID) physicians. From 5 p.m. to 9 p.m. (evening) orders are approved by ID fellows. Orders were entered as Stat, Now, and Routine. Between 9 p.m. and 8 a.m. (night), patients receive doses without approval. Meropenem restriction began in mid-January 2018. Pre- and postmeropenem restriction periods were defined as February–December 2017 and February–December 2018. Meropenem use data were compared for adult patients in the MICU. A multivariable Cox regression model was implemented to compare (1) time from order entry to approval; (2) time from order approval to patient administration; (3) total time from order entry to patient administration, adjusting for order priority, approver (ASP, ID consult, ID fellow, pharmacy); and (4) time of day of order placement (day, eve, night). The analyses were performed using SAS version 9.4 software (SAS Institute, Cary, NC). Result: Time from order approval to patient administration was significantly decreased in the postrestriction period (HR, 1.840; P < .001) (Table 1). Stat orders were faster compared to routine orders for order entry to approval (HR, 1.735; P < .001), approval to administration (HR, 2.610; P < .001), and total time from order entry to administration (HR, 2.812; P < .001). No significant differences were found in time to approval by approving service. Time from order entry to approval was faster for nighttime orders than for daytime orders (HR, 1.399; P = .037). Conclusions: Our data indicate that the time from order entry to administration decreased following meropenem restriction in our MICU. More research is needed to identify the reason for this finding, but we postulate that this is due to an effect on drug administration prioritization within nursing workflow. These data will inform our local meropenem restriction efforts.

Funding: None

Disclosures: Michelle Doll reports a research grant from Molnlycke Healthcare.

Background: Methicillin-resistant Staphylococcus aureus (MRSA) nasal colonization has been a well-established risk for developing MRSA pneumonia. In previous studies, the MRSA nasal screening test has shown an excellent negative predictive value (NPV) for MRSA pneumonia in patients without exclusion criteria such as mechanical ventilation, hemodynamic instability, cavitary lesions, and underlying pulmonary disease. MRSA nasal screening can be used as a stewardship tool to de-escalate broad antibiotic coverage, such as vancomycin. Objective: The purpose of this study was to determine whether implementation of a MRSA nasal screening questionnaire improves de-escalation of vancomycin for patients with pneumonia. Methods: A retrospective review was performed on 250 patients from October 2018 to January 2019 who received MRSA nasal screening due to their prescriber choosing only “respiratory” on the vancomycin dosing consult form. Data obtained included demographics and clinical outcomes. Statistical analyses were performed, and P < .05 was considered significant. Results: Of the 250 patients screened, only 19 patients (8%) were positive for MRSA. Moreover, 40% of patients met exclusion criteria. In 149 patients without exclusion criteria, the MRSA nasal swab had a 98% NPV. Although not statistically significant, vancomycin days of therapy (DOT) based on MRSA nasal swab result was 1 day shorter in those with negative swabs (3.49 days negative vs 4.58 days positive; P = .22). Vancomycin DOT was significantly reduced in pneumonia patients without exclusion criteria (3.17 days “no” vs 4.17 days “yes”; P = .037). Conclusions: The implementation of an electronic MRSA nasal screening questionnaire resulted in reduced vancomycin DOT in pneumonia patients at UAB Hospital. The MRSA nasal swab is an effective screening tool for antibiotic de-escalation based on its 98% NPV for MRSA pneumonia if utilized in the correct patient population.

Funding: None

Disclosures: Rachael Anne Lee reports a speaker honoraria from Prime Education, LLC.

Background: In June 2019, 3 people were diagnosed with Ebola virus disease (EVD) in Kasese district, Uganda, all of whom had come from the Democratic Republic of Congo (DRC). Although no secondary transmission of Ebola occurred, an assessment of infection prevention and control (IPC) using the WHO basic IPC facility assessment checklist revealed significant gaps. Robust IPC systems are critical for the prevention of healthcare-associated infections like EVD. A rapid intervention was developed and implemented in Kasese to strengthen IPC capacity in high-risk facilities. Methods: Of 117 healthcare facilities, 50 were considered at high risk of receiving suspected EVD cases from DRC based on population movement assessments. In August 2019, IPC mentors were selected from 25 high-risk facilities and assigned to support their facility and a second high-risk facility. Mentors ensured formation of IPC committees and implemented the national mentorship strategy for IPC preparedness in non-EVD treatment facilities. This effort focused on screening, isolation, and notification of suspect cases: 4 mentorship visits were conducted (1 per week for 1 month). Middle and terminal assessments were conducted using the WHO IPC checklist 2 and 4 weeks after the intervention commenced. Results were evaluated against baseline data. Results: Overall, 39 facilities had data from baseline, middle, and end assessments. Median scores in facility IPC standard precautions increased from baseline 50% (IQR, 39%–62%) to 73% (IQR, 67%–76%) at the terminal assessments. Scores increased for all measured parameters except for water source (access to running water). Greatest improvements were seen in formation of IPC committees (41% to 75%), hand hygiene compliance (47% to 86%), waste management (51% to 83%), and availability of dedicated isolation areas (16% to 42%) for suspect cases. Limited improvement was noted for training on management of suspect isolated cases and availability of personal protective equipment (PPE) (Fig. 1). No differences were noted in scores for facilities with nonresident mentors versus those with resident mentors at baseline (48% vs 50%) and end assessments (72% vs 74%). Conclusions: This intervention improved IPC capacity in health facilities while avoiding the cost and service disruption associated with large-scale classroom-based training of health workers. The greatest improvements were seen in activities relying on behavior change, such as hand hygiene, IPC committee, and waste management. Smaller changes were seen in areas requiring significant investments such as isolation areas, steady water source, and availability of personal protective equipment (PPE). Mentorship is ongoing in moderate- and lower-risk facilities in Kasese district.

Funding: None

Disclosures: Mohammed Lamorde reports contract research for Janssen Pharmaceutica, ViiV, Mylan.

Background: Healthcare services are increasingly shifting from inpatient to outpatient settings. Outpatient settings such as emergency departments (EDs), oncology clinics, dialysis clinics, and day surgery often involve invasive procedures with the risk of acquiring healthcare-associated infections (HAIs). As a leading cause of HAI, Clostridioides difficile infection (CDI) in outpatient settings has not been sufficiently described in Canada. The Canadian Nosocomial Infection Surveillance Program (CNISP) aims to describe the epidemiology, molecular characterization, and antimicrobial susceptibility of outpatient CDI across Canada. Methods: Epidemiologic data were collected from patients diagnosed with CDI from a network of 47 adult and pediatric CNISP hospitals. Patients presenting to an outpatient setting such as the ED or outpatient clinics were considered as outpatient CDI. Cases were considered HAIs if the patient had had a healthcare intervention within the previous 4 weeks, and they were considered community-associated if there was no history of hospitalization within the previous 12 weeks. Clostridioides difficile isolates were submitted to the National Microbiology Laboratory for testing during an annual 2-month targeted surveillance period. National and regional rates of CDI were stratified by outpatient location. Results: Between January 1, 2015, and June 30, 2019, 2,691 cases of outpatient-CDI were reported, and 348 isolates were available for testing. Most cases (1,475 of 2,691, 54.8%) were identified in outpatient clinics, and 72.8% (1,960 of 2,691) were classified as community associated. CDI cases per 100,000 ED visits were highest in 2015, at 10.3, and decreased to 8.1 in 2018. Rates from outpatient clinics decreased from 3.5 in 2016 to 2.7 in 2018 (Fig. 1). Regionally, CDI rates in the ED declined in Central Canada and increased in the West after 2016. Rates in outpatient clinics were >2 times higher in the West compared to other regions. RT027 associated with NAP1 was most common among ED patients (26 of 195, 13.3%), whereas RT106 associated with NAP11 was predominant in outpatient clinics (22 of 189, 11.6%). Overall, 10.4% of isolates were resistant to moxifloxacin, 0.5% were resistant to rifampin, and 24.2% were resistant to clindamycin. No resistance was observed for metronidazole, vancomycin, or tigecycline. Compared to CNISP inpatient CDI data, outpatients with CDI were younger (51.8 ± 23.3 vs 64.2 ± 21.6; P < .001), included more females (56.4% vs 50.9%; P < .001), and were more often treated with metronidazole (63.0% vs 56.1%; P < .001). Conclusions: For the first time, CDI cases identified in outpatient settings were characterized in a Canadian context. Outpatient CDI rates are decreasing overall, but they vary by region. Predominant ribotypes vary based on outpatient location. Outpatients with CDI are younger and are more likely female than inpatients with CDI.

Funding: None

Disclosures: Susy Hota reports contract research for Finch Therapeutics.

Background: Multidrug-resistant organisms (MDROs) are a threat to public health. The objective of this study was to define risk factors and outcomes of patients with positive blood cultures due to MDROs in 2 rural community hospitals as compared to a tertiary-care academic center. Methods: Retrospective cohort study with IRB approval from 1 tertiary-care academic center and 2 rural community hospitals (Barnes-Jewish [BJH], Parkland Health Center, and Missouri Baptist Sullivan Hospital) from July 1, 2013, to August 1, 2018. Demographics, comorbidities, procedures, outcomes, and blood and urine culture data were collected from the BJH informatics database for hospitalized patients with positive blood cultures due to MDROs. MDROs were defined according to European and US CDC standards. Results: Of the patients with positive blood cultures growing organisms with the potential to be MDR, 1,065 (55%) blood cultures grew MDROs from the academic center and 157 (33%) grew MDROs from the 2 community hospitals (P < .0001). Among these, methicillin-resistant Staphylococcus aureus (35% at BJH and 37% at community hospitals) and MDR Enterobacteriaceae (29% at BJH and 36% at community hospitals) were the most common organisms grown from blood cultures at all hospitals. Among patients with positive MDRO blood cultures, 60% were males and 69% were white, with a mean age of 58 years at BJH. At the community hospitals, 47% were male and 99% were white, with a mean age of 66 years. The most common comorbidity in patients with MDRO bacteremia at BJH was cancer, compared to diabetes at the community hospitals. At all hospitals, >33% of patients with MDRO bacteremia required an ICU stay. Also, 17% of patients with MDRO bacteremia at BJH died during hospitalization compared to 4% at the community hospitals. Among individuals with positive MDRO blood cultures, 9% had a matching isolate from a urine culture at BJH and 46% had a matching urine isolate at the community hospitals. Conclusions: At an academic medical center, the most common organisms identified in MRDO-positive blood cultures included MRSA, MDR Enterobacteriaceae, and VRE. However, at the community hospitals, MRSA, MDR Enterobacteriaceae, and ESBL Enterobacteriaceae were most common. Patients with a positive MDRO blood culture were more likely to have a matching isolate from urine culture at a community hospital compared to the academic center. Further research is needed regarding risk factors and interventions to prevent, detect, and treat MDRO infections.

Funding: None

Disclosures: Margaret A. Olsen reports consulting fees for contract research from Pfizer, Merck, and Sanofi Pasteur.

Background: Wheelchairs can contribute to healthcare-associated infection transmission due to direct contact with patients and healthcare workers and due to wide spatial movement in facilities. Objective: We utilized location data of a wheelchair to inform an agent-based model for estimating the contribution of a single contaminated patient ride in a wheelchair to subsequent environmental contamination and to estimate the potential for wheelchair disinfection between patients to disrupt this spread. Methods: The destination and origin of wheelchairs were tracked in several facility locations: specialty care services, long-term care, radiology, acute care, common spaces, domiciliary, and outpatient clinics. An agent-based model was developed in which the probability of the wheelchair traveling directly from one location to another was informed by wheelchair origin and destination data. We assumed that the first patient’s hands were contaminated with methicillin-resistant Staphylococcus aureus (MRSA). For each patient trip, each simulated patient made contact with the wheelchair arm rests and a surface in the destination location. To evaluate potential exposures of uninfected patients, all patients riding in the wheelchair after the contaminated patient were assumed to be uncontaminated. In total, 50 patient rides were simulated. The concentration and number of contaminated surfaces in each hospital area were compared in addition to the average concentration of MRSA on patient hands over time. The intervention simulation involved a disinfection of wheelchair armrests with 90%, 70%, or 50% efficacy. Results: The 3 areas that had the largest estimated number of contaminated surfaces after 50 wheelchair trips following the first patient assumed to be infected were specialty care services, long-term care, and acute care. This finding was consistent with the paths that were most frequented by the wheelchair. Without cleaning between patients, the fiftieth patient to use the wheelchair had an average MRSA concentration of 41.5 CFU/cm2. With cleaning between patients, assuming a 50% cleaning efficacy, average MRSA concentration on the hands for the fiftieth patient was reduced to 7.4 ×10-14 CFU/cm2. Conclusions: We have demonstrated that cleaning, even with efficacies as low as 50%, may protect patients using contaminated wheelchairs from potential pathogen exposures. This study also demonstrates that tracking portable equipment can be useful not only for exposure modeling but also for predicting where the largest number of surfaces contaminated via portable equipment routes may be found. Future steps include performing a sensitivity analysis to evaluate the influence of spatial assumptions.

Funding: None

Disclosures: None

Background: Automated testing instruments (ATIs) are commonly used by clinical microbiology laboratories to perform antimicrobial susceptibility testing (AST), whereas public health laboratories may use established reference methods such as broth microdilution (BMD). We investigated discrepancies in carbapenem minimum inhibitory concentrations (MICs) among Enterobacteriaceae tested by clinical laboratory ATIs and by reference BMD at the CDC. Methods: During 2016–2018, we conducted laboratory- and population-based surveillance for carbapenem-resistant Enterobacteriaceae (CRE) through the CDC Emerging Infections Program (EIP) sites (10 sites by 2018). We defined an incident case as the first isolation of Enterobacter spp (E. cloacae complex or E. aerogenes), Escherichia coli, Klebsiella pneumoniae, K. oxytoca, or K. variicola resistant to doripenem, ertapenem, imipenem, or meropenem from normally sterile sites or urine identified from a resident of the EIP catchment area in a 30-day period. Cases had isolates that were determined to be carbapenem-resistant by clinical laboratory ATI MICs (MicroScan, BD Phoenix, or VITEK 2) or by other methods, using current Clinical and Laboratory Standards Institute (CLSI) criteria. A convenience sample of these isolates was tested by reference BMD at the CDC according to CLSI guidelines. Results: Overall, 1,787 isolates from 112 clinical laboratories were tested by BMD at the CDC. Of these, clinical laboratory ATI MIC results were available for 1,638 (91.7%); 855 (52.2%) from 71 clinical laboratories did not confirm as CRE at the CDC. Nonconfirming isolates were tested on either a MicroScan (235 of 462; 50.9%), BD Phoenix (249 of 411; 60.6%), or VITEK 2 (371 of 765; 48.5%). Lack of confirmation was most common among E. coli (62.2% of E. coli isolates tested) and Enterobacter spp (61.4% of Enterobacter isolates tested) (Fig. 1A), and among isolates testing resistant to ertapenem by the clinical laboratory ATI (52.1%, Fig. 1B). Of the 1,388 isolates resistant to ertapenem in the clinical laboratory, 1,006 (72.5%) were resistant only to ertapenem. Of the 855 nonconfirming isolates, 638 (74.6%) were resistant only to ertapenem based on clinical laboratory ATI MICs. Conclusions: Nonconfirming isolates were widespread across laboratories and ATIs. Lack of confirmation was most common among E. coli and Enterobacter spp. Among nonconfirming isolates, most were resistant only to ertapenem. These findings may suggest that ATIs overcall resistance to ertapenem or that isolate transport and storage conditions affect ertapenem resistance. Further investigation into this lack of confirmation is needed, and CRE case identification in public health surveillance may need to account for this phenomenon.

Funding: None

Disclosures: None