Grant Investigator Research Articles

1272. A Routine, HIV and HCV Testing and Treatment Program in a Large, Multi-Campus Emergency Department Finds High Prevalence of Acute HIV and Chronic Hepatitis C

BackgroundOne in seven people living with HIV in the United States are unaware of their serostatus. Approximately 11,280 annual HIV infections (30.2%) are caused by this subset of individuals (CDC, 2017). In addition, acute HCV infections have nearly tripled since 2011, with many states seeing a dramatic increase in incidence among younger people outside the birth cohort (CDC, 2017). Because many individuals still use emergency departments (EDs) for their healthcare needs, these institutions play an increasingly important role in screening patients for HIV and HCV and linking them to medical services. Routine, opt-out testing initiatives are particularly effective at identifying new cases of HIV and HCV that could have otherwise been missed by a risk-based approach to screening.MethodsIn early May 2017, physicians and advanced practice nurses from Sutter Health’s Alta Bates Summit Medical Center (ABSMC) and a nearby outpatient HIV clinic implemented a routine HIV and HCV screening program at the hospital’s large, two-campus ED system in Oakland, CA. ED medical directors created a Nursing Standardized Procedure (NSP) to allow registered nurses (RNs) to independently order both blood tests using an automated, best practice authority (BPA) screen in the electronic health record (EHR) of any patient who met CDC-defined age criteria for testing.ResultsOf the 6,315 people screened for HIV between May 1, 2017 and March 31, 2018, 43 (0.7%) patients tested positive. Twelve (57%) of the 21 patients found to have a new HIV diagnosis also had symptomatic, acute HIV infection (AHI). All 12 patients with AHI initiated anti-retroviral therapy (ART) within five to 96 hours of their preliminary positive test result. Of the 5,820 patients screened for HCV, 424 (7.3%) were anti-HCV positive, while 185 (3.2%) patients had chronic infection. Thirty-nine percent of chronic HCV cases were among younger patients born before 1965. All patients with HIV or chronic HCV were referred to medical care at East Bay Advanced Care (EBAC).ConclusionAn automated, routine HIV-HCV testing program integrated into standard nursing workflow at a community ED resulted in the timely screening, diagnosis, and treatment of many patients with acute HIV, and identified a high prevalence of chronic HCV infections among younger patients.Disclosures R. Anson, Frontline of Communities in the United States (Gilead, Inc.): Grant Investigator, Grant recipient. C. Hall, Frontline of Communities in the United States (Gilead, Inc.): Grant Investigator, Grant recipient.

2226. Immune Checkpoint Inhibitors in Solid Tumor Patients with Chronic Hepatitis C Virus Infection: A Prospective Case-Series

BackgroundImmune checkpoint inhibitors are a novel class of targeted therapy that activates T cell-mediated tumor cell death. Controversies exist about the safety and efficacy of immunotherapy in patients with chronic viral infections affecting T cells, such as hepatitis C virus (HCV). Herein, we analyzed the effect of immune checkpoint inhibitors on HCV viremia and HCV-related hepatic outcome.MethodsHCV-infected patients with solid tumors seen at MD Anderson Cancer Center (November 2012–April 2018) were enrolled in a prospective observational study. Patients were monitored for the development of HCV reactivation (HCV-RNA ≥1 log10IU/mL over baseline), hepatitis flare (alanine transaminase increase to ≥3 times upper limit of normal) and HCV-associated hepatitis (HCV reactivation and hepatitis flare) while on cancer treatment.ResultsOut of 205 chronically infected patients with solid tumors, 12 (6%) received immunotherapy and were seen in the HCV clinic, but only four (2%) returned for regular monitoring (Table 1). They were followed for 9 months. None of the four patients received concomitant chemotherapy or steroids. Hepatitis flare occurred in three patients, but HCV reactivation or HCV-associated hepatitis was not detected in any study patient. Immune checkpoint inhibitors were discontinued in one patient (25%) due to hepatitis flare unrelated to HCV.ConclusionThe use of immune checkpoint inhibitors appears to be safe in solid tumor patients with HCV infection.Table 1.Demographics, Types of Cancer, Immunotherapy Received, and Changes in Serum HCV-RNAPatientAge, yearsSexHCV GenotypeCirrhosisType of CancerImmunotherapyBaseline HCV-RNA (log 10IU/mL)Highest HCV-RNA After Baseline (log 10IU/mL)Hepatitis Flare 1 67Male1aYesHepatocellular carcinomaNivolumab + ipilimumab5.455.08Yes a 2 52Male1aNoMelanomaNivolumab + ipilimumab6.976.95Yes b 3 57Male3NoMelanomaPembrolizumab5.726.39No 4 55Male1aNoMelanomaIpilimumab5.757.18 cYes c a After partial hepatectomy. b Negative infectious work-up including hepatitis A, B, E, cytomegalovirus, and herpes simplex virus. c Six months after completing immunotherapy, HCV-associated hepatitis occurred after starting high-dose steroids for brain metastasis.Disclosures H. Torres, Gilead Sciences, Merck & Co., Inc.: Grant Investigator, Grant recipient. Vertex Pharmaceuticals: Grant Investigator, Grant recipient.

2301. Streptococcus pneumoniae Serotyping: Assessing the Performance of a PCR- and Sequencing-Based Testing Algorithm

Background Streptococcus pneumoniae is a bacterium that causes significant morbidity and mortality worldwide. Its capsular polysaccharides have been used successfully as vaccine antigens, and to characterize S. pneumoniae into 92 different serotypes. Phenotypic (Quellung reaction) or genotopic (PCR or sequencing) methods can be used for serotype assignment, but the performance may vary between methods. This study compared the performance of the Quellung reaction, to an algorithm using PCR- and sequence-based serotyping technologies for vaccine-preventable or closely related serotypes.MethodsA panel of geographically diverse isolates of S. pneumoniae spanning 92 different serotypes was provided by various references laboratories worldwide. Each isolate was subjected to conventional multiplex PCR methods, using previously established methods. Sanger sequencing was performed using genetic signatures defined in the PneumoCaT database. When discrepant, Quellung reaction were repeated, and next-generation sequencing and comparative genomics was used to evaluate the sequence composition of the cps loci.ResultsAs expected, PCR was unable to assign serotype in some cases, and some serotype results were insufficiently discriminatory. Following sequencing, 86.3% (404/468) of isolates were concordant with the Quellung serotyping. Discrepant analyses are underway.ConclusionAn algorithm based on PCR and sequencing, or next-generation sequencing alone, shows much promise for serotyping of S. pneumoniae. However, discrepant results were noted, suggesting either our current understanding of genetic signatures conferring serotype-specificity might not be complete, or the Quellung reaction results were incorrect. Accurate methods for serotyping are essential to monitor the impact of pneumococcal vaccines, and understand the epidemiology of S. pneumoniae diseases.Disclosures S. A. McNeil, GSK: Grant Investigator, Research grant. Pfizer: Grant Investigator, Research grant. Merck: Collaborator and Consultant, Contract clinical trials and Speaker honorarium. Novartis: Collaborator, Contract clinical trials. Sanofi Pasteur: Collaborator, Contract clinical trials.

528. Lab Stewardship for Clostridium difficile Testing Improves Appropriate Testing While Decreases Unnecessary Testing and Saves Laboratory Resources While Dramatically Helping to Reduce C diff Standardized Infection Ratios (SIR)

BackgroundUnnecessary testing for Clostridium difficile infection (CDI) can be both wasteful and contra productive—retesting the same positive patient after transfer to a new nursing unit will only to confirm the patient has CDI (already known) and likely be classified as a new case of hospital-onset (HO) CDI. Yet, it is also important to recognize community-onset (CO) CDI in hospital, not only because it prevents late recognition of CO CDI as being classified as an HO event, will also to afford appropriate contact precautions and therapeutic measures are instituted in a timely fashion. Laboratory stewardship (LS) can be helpful in improving appropriateness of C. difficile testing.MethodsWe developed 2 CDI testing algorithms. One focused on hospital days 1–3, the other for all C. difficile testing after hospital day 3 (AHD3). The LS quality improvement (QI) project was rolled out in 2 stages. During the first 6 months we focused on improving early detection of CO-CDI, while during the next 6 months a mandatory review of all C. difficile testing orders AHD3 was conducted by a 10 person team. Testing that concurred with the algorithm was approved. Nonapproval was communicated to the care teams. Appeals could be made on a case-by-case basis to the medical director of infection control. Validation audits of nonapproved cases were performed to determine whether testing algorithms were sound.ResultsCO-CDI detection steadily increased over the yearlong LS QI period (average of 6 cases/week at start vs. 12 cases/week at year’s end). During the 6 months of the AHD3 mandatory order review 678 C. difficile orders were placed, 428 (63.1%) were approved, 250 (36.9%) were rejected. Reduced use of laboratory resources is estimated to have saved $14,950. LS and frequent communication with care teams contributed better recognition of CO-CDI, decreased inappropriate repeat testing, avoidance of diagnosing colonized patients as HO-CDI and was associated with a significantly drop our CDI SIR (Figure 1).ConclusionAn algorithm-based guideline for a 2-step LS QI program focused on reviews of all C. difficile orders AFHD3 as well as improving early detection of CO-CDI and was associated with better laboratory resource utilization and markedly decreased C. difficile SIR. Efforts are currently underway to automate much of the review process. Disclosures J. P. Parada, Merck: Speaker’s Bureau, Speaker honorarium. A. Harrington, Biofire: Grant Investigator and Scientific Advisor, Consulting fee, Research grant and Speaker honorarium. Cepheid: Grant Investigator and Speaker’s Bureau, Research grant.

1803. Clinical Impact of Rapid Blood Culture Diagnostics Differs by Time of Day and Gram Stain Type: Lessons From Implementation of Verigene® Blood Culture Testing in a Children’s Hospital

BackgroundRapid blood culture diagnostics paired with antimicrobial stewardship (AS) enhances appropriate antimicrobial treatment for bloodstream infections (BSI). Ideal implementation strategies for blood culture diagnostics are not clear, including whether to perform molecular testing during off-hours or for all organism types.MethodsTo determine whether the clinical impact of the Verigene® Blood Culture Nucleic Acid Tests (VG) is influenced by time of day and Gram-positive (GP) or Gram-negative (GN) organism, we performed a single-center, retrospective evaluation of children with BSIs and VG testing April 2017–March 2018. VG testing was performed on all Gram stain positive blood cultures 24/7. AS providers were notified of VG results at all hours, but AS interventions occurred on weekdays, during office hours. Wilcoxon rank-sum and chi-squared tests were used for analyses.ResultsTwo hundred fifty-seven isolates (GP:184, 72%; GN:73, 28%) were identified from 224 cultures by standard of care (SOC) conventional culture. VG and SOC results were concordant in 173/224 (77%) cultures overall, 168/197 (85%) monomicrobial cultures, and 5/27 (19%) polymicrobial cultures. Thirty-eight of 257 isolates (15%) were not targets on VG. Among on-panel organisms, discordance was similar for GN (4/48, 8.3%) and GP isolates (16/171, 9.4%). Among 95 opportunities for antibiotic optimization based on VG results, antibiotic changes occurred in 80 (84%), with 48 de-escalations, 11 escalations, and 21 averted antibiotic starts. More modifications were made for patients with GP vs. GN BSI (75 vs. 5, P < 0.001). For GP BSI, mean time from VG result to antibiotic modification was 8.92 hours overall, and faster during day shift than night shift, although not statistically significant (P = 0.49) (Figure 1). Among patients with GP BSI, 4 were not admitted and 21 had antibiotics discontinued within 24 hours.ConclusionAt our children’s hospital, VG testing implemented with AS resulted in antibiotic optimization, but not as promptly as expected. Antibiotic changes occurred more frequently for GP than GN BSI and occurred more quickly when VG testing occurred during the day vs. night. There is a need for strategies that improve the impact of rapid blood culture diagnostics, especially during off-hours and for GN BSI. Disclosures R. Banerjee, Accelerate Diagnostics, Biomerieux, BioFire: Grant Investigator, Research grant and Research support. J. Schmitz, Luminex: Grant Investigator, Research grant.

LB17. Age-Related Differences in Influenza Type/Subtype Among Patients Hospitalized with Influenza, FluSurv-NET—2017–2018

BackgroundThe 2017–2018 influenza season had the highest rates of influenza hospitalizations since the 2009 H1N1 pandemic. We used data from the Influenza Hospitalization Surveillance Network (FluSurv-NET) to identify unique characteristics of the 2017–2018 season.MethodsWe included all patients residing within a FluSurv-NET catchment area, and hospitalized with laboratory-confirmed influenza during 2017–2018. We used multiple imputation, including age, surveillance site, and month of hospital admission as predictors, to impute influenza A subtype for 40–64% of cases across seasons with an unknown subtype. We calculated influenza hospitalization rates by type/subtype per 100,000 population. We compared 2017–2018 rates to rates during 4 prior seasons: 2016–2017, 2015–2016, 2014–2015, and 2013–2014.ResultsThe overall unadjusted hospitalization rates per 100,000 population varied from 31.5 during 2015–2016 to 105.1 during 2017–2018. After imputing A subtype, the 2017–2018 season had the highest rates observed for H3N2 (62.8) and B (28.5) than in any previous season, and the third highest rate of H1N1 (13.5) (Figure 1A). During 2017–2018, rates in adult ≥65 years peaked 3 weeks before they peaked in children 0–4 years. In contrast, during the four prior seasons, rates in adults ≥65 years peaked during the same week or 1 week after they peaked in children 0–4 years. During 2017–2018, the distribution of influenza type/subtypes varied significantly by age group (P < 0.0001); for example, the proportion of cases with H1N1 ranged from 19 to 29% in those <65 years to only 7% in those ≥65 years. During 2017–2018, H1N1 (the nonpredominant A virus) contributed >25% of A cases across all age groups (except ≥65 years) vs. all prior seasons where the nonpredominant A virus contributed <10% of A cases across all age groups (except ≥65 years) (Figure 1B–F).ConclusionsSeveral unique characteristics may have contributed to the high hospitalization rates observed during 2017–2018. Rates in older adults, who were predominantly infected with H3N2, peaked several weeks prior to children in contrast to prior seasons. Higher overall rates of H3N2 and B were observed in 2017–2018 compared with these prior seasons and substantial H1N1 co-circulation also occurred with marked variability by age group.Disclosures E. J. Anderson, NovaVax: Grant Investigator, Research grant. Pfizer: Grant Investigator, Research grant. AbbVie: Consultant, Consulting fee. MedImmune: Investigator, Research support. PaxVax: Investigator, Research support. Micron: Investigator, Research support. H. K. Talbot, Sanofi Pasteur: Investigator, Research grant. Gilead: Investigator, Research grant. MedImmune: Investigator, Research grant. Vaxinnate: Safety Board, none. Seqirus: Safety Board, none.

2454. Pertussis Vaccine Effectiveness and Waning Immunity in Alberta, Canada: 2004–2015

BackgroundDespite childhood vaccination coverage rates exceeding 75%, pertussis is still frequently reported in Canada. In Alberta, pertussis incidence ranged from 1.8 to 20.5 cases per 100,000 persons for 2004–2015. Most cases occurred in those aged < 15 years. We investigated pertussis vaccine effectiveness (VE) using a test-negative designed (TND) study.MethodsAll individuals who had undergone a real-time PCR laboratory test for Bordetella pertussis between January 1, 2004 and August 31, 2015, in the province of Alberta, Canada were included. Vaccination history was obtained from Alberta’s immunization repository. Vaccination status was classified as complete, incomplete, or not vaccinated, based on the province’s vaccination schedule. Multivariable logistic regression models were used to estimate adjusted odds ratios (aOR) and 95% confidence intervals (95% CI) for pertussis infection by time since last vaccination, comparing those with complete or incomplete vaccination to those not vaccinated. We adjusted for age, sex, income, urban/rural status, and the presence of a co-morbid condition. Vaccine effectiveness (VE) was calculated as [(1-aOR)*100].ResultsOf 28,154 individuals tested, 2,297 (12.3%) tested positive for B. pertussis. Among those with complete vs. no vaccination, VE was 88% (95% CI 85–91%) at 1 year, 83% (95% CI 79–86%) at 1 to 3 years, 70% (95% CI 63–76%) at 4 to 6 years, 28% (95% CI 12–42%) at 7 to 9 years, and -4% (95% CI -53 to 29%) at 10 or more years since a last dose of a pertussis vaccine (Figure 1). VE was similar but attenuated in the incompletely vaccinated group, with a comparable waning of immunity.ConclusionPertussis VE was high in the first year after vaccination, then declined noticeably after 5 years. Our results suggest there is a large number of adolescents and adults susceptible to pertussis. Regular boosters throughout childhood, adolescence, and during pregnancy are critical to protect those at greatest risk of infection and complications. Further validation of the strengths and weaknesses of the TND for assessing pertussis VE is needed.Figure 1.Pertussis VE by year since last vaccination in those with complete vaccination status.Disclosures M. L. Russell, Novartis Pharmaceuticals Canada Inc.: Grant Investigator and Unconditional Research Grant, Grant recipient. Merck Frosst Canada Inc.: Grant Investigator and Unconditional Research Grant, Grant recipient.

530. The Perfect Storm for Improved Standardized Infection Ratio (SIR)—Recognizing More Community-onset Clostridium difficile Infections Increases the Expected Number of C. difficile Cases While also Helping to Decrease the Actual Observed Number of Hospital Onset C. difficile Cases

BackgroundIt is essential to recognize the true burden of community-onset (CO) Clostridium difficile infection (CDI) in hospital, not only because it prevents late recognition of CO CDI as being classified as a hospital-onset (HO) event, but also to assure appropriate contact precautions and therapeutic measures are deployed in a timely fashion. We recognized that our timely diagnosis of CO-CDI was suboptimal and sought to improve early recognition of CO-CDI.MethodsWe developed an automated daily report of all patients noted to have loose stools documented in the nursing flow sheets during the first 3 days of hospitalization. This report was automatically forwarded to the nurse manager of the unit, as well as was reviewed daily, Monday–Friday, by the infection preventionists (IP) to determine whether stool testing had been sent on these symptomatic patients. If not, then the IP would call the nurse caring for the patient and encourage that a stool sample be sent ASAP and before the third hospital day was completed.ResultsWith this intervention, we increased early appropriate stool testing for patients with documented loose stools during the first 3 days of hospitalization leading to a marked increase in CO-CDI, as well as a notable decrease in HO-CDI lab ID events (Figure 1). Together, the increased recognition of CO-CDI increased our expected cases/SIR denominator and decreased observed cases/SIR numerator and substantially dropped our CDI SIR from a 2 years preintervention median SIR of 1.47 to 0.95 during the five quarters since the intervention has been in effect.ConclusionAfter several years of our CDI SIR remaining stubbornly around 1.5, we developed a system of enhanced recognition of patients who had loose stools early in their admission. This practice aided better recognition of CDI present on admission, substantially increasing our detection of CO-CDI. We also noted decreases in HO-CDI, in part secondary to no longer diagnosing patients who actually had CO-CDI later in their hospitalization and classifying CO-CDI as HO-CDI cases. In turn, we noted a remarkable decrease in our CDI SIR. Disclosures J. P. Parada, Merck: Speaker’s Bureau, Speaker honorarium. A. Harrington, Biofire: Grant Investigator and Scientific Advisor, Consulting fee, Research grant and Speaker honorarium. Cepheid: Grant Investigator and Speaker’s Bureau, Research grant.

326. Malaria vs. Bacterial Meningitis in Children With Spinal Tap in the Luanda Children’s Hospital, Angola

BackgroundIn Sub-Saharan Africa, both malaria (M), and bacterial meningitis (BM) cause fever and central nervous system (CNS) disturbance. We studied their prevalence, characteristics, outcome, and risk factors for poor outcome to better understand the clinical impact of suspected CNS infection in children.MethodsWe conducted a prospective study in the Children’s Hospital (HPDB) in the capital of Angola which attends 300 new patients daily. Spinal tap (ST) was performed for children presenting with altered consciousness, convulsions, prostration, or meningism. The analysis included children aged 3 month to 15 years with confirmed discharge diagnosis in 2016–2017.ResultsOf 941 children, the diagnosis was M in 56% (525), BM in 12% (116), epilepsy/convulsions in 9% (88), and other infections in 6% (60). Of all children, 16% (150/941) died, 6% (45/733) had severe, 14% (93/655) any neurological sequelae, and 27% (243/897) either died or had neurological sequelae. In children with M, the corresponding figures were 7% (35/525), 1.5% (7/476), 4% (19/443), and 11% (54/514). In children with BM, the figures were 41% (47/116), 15% (8/54), 33% (11/33), and 55% (58/105), respectively. Comparing with M, children with BM were younger (median age (IQR) 28 (61) vs. 60 (68) months, P < 0.0001), had an underlying illness (23/97 vs. 19/374, P < 0.0001), like sickle-cell disease (18/96 vs. 9/372, P < 0.0001), longer duration of illness (4 (4) vs. 3 (3) days, P < 0.0001, dyspnea (70/119 vs. 210/463, P = 0.009), were dehydrated (36/113 vs. 67/441, P < 0.0001), or malnourished (38/115 vs. 75/447, P = 0.0001). Multivariate analysis revealed as independent risk factors for death or neurological sequelae age <12 months (OR 1.71, 95% CI 1.02–2.88, P < 0.0001), duration of illness >3 days (2.48, 1.68–3.64, P < 0.0001), malnutrition (1.92, 1.20–3.05, P = 0.006), and dehydration (1.92, 1.16–3.14, P = 0.01). When BM vs. M was included in the analysis, BM appeared as the most important risk factor (OR 8.06, 4.44–14.65, P < 0.0001) and age lost its significance.ConclusionIn suspected CNS infection, M was the final diagnosis of most children. However, BM caused more deaths and neurological sequelae. Amendable factors, such as delay in treatment, dehydration, and malnutrition, appeared as risk factors for poor outcome.Disclosures T. Pelkonen, sanofi pasteur: Investigator, Sanofi Paster funded this study. Paediatric Research Foundation, Helsinki, Finland: Grant Investigator, Grant recipient. Päivikki and Sakari Sohlberg Foundation, Helsinki, Finland: Grant Investigator, Grant recipient

1790. Clinical and Economic Outcome Evaluation with Penicillin Skin Testing as an Antimicrobial Stewardship Initiative in a Not-for-Profit Community Health System

BackgroundPenicillin skin testing (PST) is a novel way to reduce the use of broad-spectrum agents, potentially resulting in unnecessary overuse and cost savings. This study evaluated clinical and economic outcomes of antimicrobials prescribed with and without PST in a community health system.MethodsThis quasi-experimental study compared an experimental group of 100 adult patients who completed PST for a self-reported penicillin allergy over an open enrollment period beginning January 2016 to a matched control group of 100 patients over the same time frame that had a listed penicillin allergy as well as consultation with infectious diseases. Patients in the control group were matched to the infection diagnosis codes of the members of experimental group and then randomly selected and matched on a 1:1 basis. The primary outcome was β-lactam days of therapy (DOT) defined as either a penicillin or cephalosporin (not carbapenem). The secondary outcome assessed the average cost of antimicrobial therapy before and after PST.ResultsThe control group consisted of 436 patients who met inclusion criteria with 100 patients from that group matched to the 100 patients in the PST group by diagnosis code. The most common self-reported allergy consisted of IgE-mediated (52%) and unknown (30%) in the PST group and IgE-mediated (33%), unknown (20%), and rash (32%) in the control group. Ninety-eight of 100 patients who underwent PST tested negative, with 71 out of 98 (73%) having changes directly made to their antimicrobial regimens immediately after PST. Β-lactam DOT for the PST group were 666 out of 1,094 (60.88%, with 34.82% being a penicillin specifically). Β-lactam DOT for the control group consisted of 386 out of 984 (39.64%, with 6.4% being a penicillin specifically). Chi-square test of homogeneity for β-lactam DOT between the two groups was significant (P < 0.00001). Changes to the antimicrobial regimen after PST saved the average patient $353.03 compared with no change in pre-PST regimen (P = 0.045).ConclusionPST led to immediate antimicrobial de-escalation in the majority of patients who tested negative. This led to a significant increase in β-lactam usage, specifically penicillins. These benefits were also associated with significant cost savings to patients, justifying the cost of performing PST.Disclosures B. M. Jones, ALK: Consultant, Grant Investigator and Speaker’s Bureau, Consulting fee, Grant recipient and Speaker honorarium. C. Bland, ALK: Grant Investigator and Speaker’s Bureau, Grant recipient and Speaker honorarium.

1871. Identifying Time Periods of High and Low Vancomycin Use

BackgroundA national goal has been set to decrease inappropriate antibiotic use by 2020. To quantify decreases in use, consumption metrics and benchmarking strategies are implicit. However, while tracking and reporting antimicrobial use is widely recommended, these data do not address appropriateness. Accordingly, we developed a methodology to identify and report high and low vancomycin use periods which may represent inappropriate or unsafe antimicrobial use.MethodsThis is an observational, retrospective study of facility-wide vancomycin consumption data, aggregated, and examined on a hospital level from three academic medical centers: Northwestern Medicine (NM), Michigan Medicine (UM), and Henry Ford (HF) Hospital. Utilization was quantified as antimicrobial days (AD) per 1,000 days present (DP) on a monthly basis, recorded over 46 consecutive months (January 2014 through October 2017) for NM and HF, and 40 consecutive months (July 2014 through October 2017) for UM. Linear regression models and prediction intervals were generated to identify high-usage months. Use exceeding the upper bound of a prediction interval of 80 percent in a given month was used to define increased use, and the lower bound was used to define decreased use.ResultsVancomycin use averaged 70.3 AD per 1,000 DP at NM, 89 at UM, and 153.8 at HF. Regression models indicated HF and UM consumption decreased at a monthly rate of 1.2 AD per 1,000 DP and 0.1 AD per 1,000 DP, respectively, whereas NM use increased at a rate of 0.1 AD per 1,000 DP over the study period. Overall, we identified n = 6, n = 5 and n = 6 vancomycin increased use months and n = 7, n = 6 and n = 5 decreased use months at NM, UM and HF, respectively.ConclusionOur methodology identified a total of 17 potential instances of increased and 18 decreased use periods for vancomycin. Patient-specific and/or hospital-level factors may contribute to inappropriate vancomycin use and requires further study. The relationship between increased or decreased antibiotic use and appropriateness should be a focus in future efforts. Once the link between use and appropriateness is known, interventions can target specific use periods to maximize benefit of the intervention.Disclosures J. Liu, Merck: Grant fund from Merck, Research grant. S. Davis, Achaogen: Consultant and Scientific Advisor, Consulting fee. Allergan: Consultant and Scientific Advisor, Consulting fee. Melinta: Consultant and Scientific Advisor, Consulting fee. Nabriva: Consultant and Scientific Advisor, Consulting fee. Zavante: Consultant and Scientific Advisor, Consulting fee. T. S. Patel, Merck: Grant Investigator, Research grant. K. Kaye, Zavante Therapeutics, Inc.: Scientific Advisor, Consulting fee. M. H. Scheetz, Merck & Co., Inc.: Grant Investigator, Grant recipient. Bayer: Consultant, Consulting fee.

720. Respiratory and Nonrespiratory Complications Among Patients Hospitalized with Influenza, FluSurv-NET, 2016–2017

BackgroundInfluenza is most commonly associated with respiratory complications; however, nonrespiratory complications occur frequently among patients hospitalized with influenza. We used data from the Influenza Hospitalization Surveillance Network (FluSurv-NET) to describe complications recorded on discharge summaries of patients hospitalized with influenza.MethodsWe included children (0–17 years) and adults (≥18 years), who resided within a FluSurv-NET catchment area and were hospitalized with laboratory-confirmed influenza during 2016–2017. We abstracted data on underlying conditions and discharge diagnoses from medical charts. We calculated the frequency of respiratory and nonrespiratory complications in all age groups and used univariate and multivariable logistic regression to examine factors associated with select complications among adults.ResultsAmong 17,489 patients, the most common respiratory complications were pneumonia (26%) and acute respiratory failure (23%) and the most common nonrespiratory complications were sepsis (16%) and acute renal failure (ARF) (12%). Complications varied by age group (figure). Pneumonia was the most common respiratory complication in all age groups except 0–4 years; among children aged 0–4 years bronchiolitis was most common (104/712; 15%). Among 97 children aged 0–4 years with bronchiolitis who underwent testing for respiratory syncytial virus (RSV), 37% had RSV. The most common nonrespiratory complication was seizures in children aged 0–17 years (17% had a history of prior seizures) and sepsis in adults. Among adults (n = 16,057), factors most strongly associated with ARF included chronic renal disease (adjusted odds ratio (AOR) 2.5; 95% confidence interval (95% CI) 2.2–2.8), male sex (AOR 1.5 95% CI 1.4–1.7) and age ≥65 years (AOR 1.4 95% CI 1.2–1.7); the factor most strongly associated with sepsis was chronic neuromuscular disease (AOR 1.5 95% CI 1.3–1.8).ConclusionInfluenza hospitalizations are associated with a broad spectrum of complications including pneumonia, respiratory failure, sepsis, ARF and seizures. During the influenza season, astute clinicians should keep influenza in the differential diagnosis for patients with a wide range of presentations. :Disclosures. H. K. Talbot, sanofi pasteur: Investigator, Research grant. Gilead: Investigator, Research grant. MedImmune: Investigator, Research grant. Vaxinnate: Safety Board, none. Seqirus: Safety Board, none. E. J. Anderson, NovaVax: Grant Investigator, Research grant. Pfizer: Grant Investigator, Research grant. AbbVie: Consultant, Consulting fee. MedImmune: Investigator, Research support. PaxVax: Investigator, Research support. Micron: Investigator, Research support.

1146. Effectiveness of Ultraviolet Irradiation on Candida auris: A Laboratory Study

Background Candida auris is a multidrug-resistant yeast which persists on healthcare surfaces for prolonged periods of time and is an emerging pathogen in hospitals. It has been linked to healthcare-associated infection (HAI) through surface transmission. Mobile ultraviolet (UV) light emitting devices from mercury sources have been shown to be effective in reducing C. auris bioburden but require prolonged exposure. In this study, we demonstrate the efficacy of an UV emitting device used in our hospital for terminal disinfection on C. auris.MethodsTwo C. auris strains (AR-381-CAU-01 and CAU-02) isolates obtained from Centers for Disease Prevention and Control (CDC) were used along with a Candida albicans (C. albicans) strain. An organism load of 10 μL containing 106 colony forming unit (CFU) was spread on a 20-mm diameter stainless steel coupon and exposed to the UV source from a pulsed xenon device at 5 feet distance and 4 feet height for 5, 10, and 30 minutes. Killing efficacy in terms of log reduction was calculated in comparison to untreated control coupons.ResultsMean CFU log10 reduction for C. albicans, CAU-01, and CAU-02 was 0.547, 1.051, and 0.952 at 5 minutes; 1.412, 1.975, and 1.879 at 10 minutes; and 2.639, 3.971, and 4.145 at 30 minutes, respectively. Figure 1 describes the mean log10 reduction as well as the minimum and maximum log reduction by isolates.ConclusionOur study demonstrates the UV from a pulsed xenon device is effective in reducing the C. auris on stainless steel coupons. Similar to previously published data on reduction of C. auris by other UV sources, extended exposure is required to achieve a higher log reduction of C. auris. We did not have any C. auris clinical infections to assess efficacy of UV on HAI reduction. Disclosures C. Jinadatha, Xenex Healthcare Service: Grant Investigator, Research grant.

LB5. Safety of In Utero Antiretroviral (ARV) Exposure: Neurologic Outcomes in HIV-Exposed, Uninfected Children

BackgroundAntiretroviral therapy for pregnant women with HIV has dramatically decreased perinatal transmission of HIV, but concerns remain regarding adverse neurologic outcomes from possible mitochondrial dysfunction or other mechanisms in children exposed in utero to antiretroviral (ARV) medications.MethodWe evaluated HIV-exposed uninfected (HEU) children enrolled in the Surveillance Monitoring for ART Toxicities (SMARTT) study, a longitudinal observational cohort study conducted by the Pediatric HIV/AIDS Cohort Study (PHACS) network. The primary outcome of interest was a “neurologic case” (microcephaly, febrile seizures, seizure disorders, ophthalmologic disorders, other neurologic conditions) as determined by clinical review blinded to ARV exposure. Log-binomial regression analysis was used to obtain adjusted relative risks (aRRs) for associations between in utero ARV exposure and neurologic case status, accounting for potential confounders including Hispanic ethnicity, tobacco use during pregnancy, and birth cohort (2011–2014 and 2015–2017 vs. <2011). To account for variable person-time follow-up within the cohort, Poisson regression models for adjusted incidence rate ratios (aIRRs) were also fitted.ResultAmong 3,747 eligible HEU children enrolled in SMARTT (52% male, 68% Black and 31% Hispanic), 237 were diagnosed with neurologic conditions, yielding an event rate of 6.3% (95% CI: 5.6%, 7.2%). Tobacco and alcohol use during pregnancy were common (17% and 8%, respectively). The majority of children had in utero ARV exposure (87%); 60% to PI-based regimens, 16% to NNRTI-based regimens and 7% to PI + NNRTI-based regimens. In adjusted models, there was a trend towards an association between efavirenz exposure (EFV) and neurologic case status (aRR: 1.60, 95% CI: 0.99, 2.58). This association was statistically significant in sensitivity analyses restricted to children enrolled prior to or shortly after birth (aRR: 1.80, 95% CI: 1.06, 3.05), excluding children with confirmed congenital anomalies (aRR: 1.66, 95% CI: 1.02, 2.64), and accounting for person-time follow-up (aIRR: 1.55, 95% CI: 1.00, 2.76).ConclusionEFV exposure during pregnancy was associated with a higher risk of neurologic abnormalities in infancy and childhood.Disclosures R. Van Dyke, Giliad Sciences: Grant Investigator, Research grant. E. G. Chadwick, Abbott Labs: Shareholder, stock dividends. AbbVie: Shareholder, stock dividends.

1273. Routine Opt-out HIV Screening and Detection of HIV Infection Among Emergency Department Patients

BackgroundThe Southeastern United States bears a disproportionate burden of HIV infection, accounting for nearly half of all new cases. The Centers for Disease Control and Prevention released routine opt-out testing recommendations in 2006. Our emergency department collaborated with our infectious diseases clinic (ECU-ID) to implement suggested guidelines among adults since March 2017.MethodsOur primary aim was to implement routine, opt-out HIV testing in the Vidant Medical Center Emergency Department (ED) for patients between 18 and 65 years of age who have blood work completed, and have not had a test documented in the electronic medical record (EMR) in the last year. A secondary aim was to successfully link HIV-positive patients to care at ECU-ID or preferred clinic. Methods defining programmatic success included developing nurse directed opt-out ordering protocol, integrating testing into normal ED workflow, utilizing the existing EMR to prompt testing, and hiring a linkage coordinator to initiate post-test counseling and linkage-to-care.ResultsSince March 2, 2017, a total of 7,109 HIV tests were performed; an average of 592 monthly tests conducted compared with a previous average of 10 stat tests. Testing increased 5,820% compared with 2015. Of the 21 HIV-positive patients found, 16 were newly diagnosed. Among those newly diagnosed, 14 (87.5%) were linked to care; and among the five known positives, two (40%) were linked to care. Reasons why patients could not be linked included incarceration, refusal to link to care, and re-location.ConclusionJoined with the implementation of a routinized ED HIV testing program, a seamless process was developed to link persons found to be positive in the ED to HIV care services; therefore, establishing a systems-level prevention model. Future plans include expanding testing to adolescents and utilizing similar methods to integrate Hepatitis C testing.Disclosures All Authors. Gilead Sciences, Inc.: Grant Investigator, Grant recipient and Salary.

2008. Effective and Early Diagnosis of Pneumonia in Patients With Acute Leukemia in a Comprehensive Cancer Center: How Can We Improve the Microbiological Diagnosis?

BackgroundPneumonia is one of the main causes of morbi-mortality in acute leukemia (AL) patients. The positive yield of microbiology diagnosis is still significantly low. The aim of the study was to evaluate the possible impact of use of diagnostic methods (within first 48 hours of diagnosis) in AL patients with pneumonia during chemotherapy.MethodsRetrospective study (January 2017–December 2017) at MD Anderson Cancer Center. The medical records of adult patients with AML, MDS, or ALL who developed CT-confirmed pneumonia after induction or second-line chemotherapy were reviewed, including demographic, clinical, microbiology data, and outcomes.ResultsDuring 2017, 174 patients with AL developed pneumonia confirmed by CT chest. Fifty (29%) of them during induction/second-line chemotherapy: 42 (84%) AML, five (10%) MDS, and three (6%) ALL. Thirty-one (62%) showed consolidation in CT, 14 (28%) nodules, and five (10%) both findings. Mean age was 65 (SD: 11.5, range: 24–87) years with 46% males. Thirty-three (66%) patients had neutropenia (ANC<500) at the time of pneumonia. ID was consulted in 38 (76%) and pulmonary in 37 (74%) patients. Bronchoscopy/BAL (bronch) was performed in only 24 (48%) patients, still with the highest diagnostic yield (13/24, 54%) compared with other diagnostic methods (sputum and blood cultures; and galactomannan, β-glucan, and cryptococcal antigen in serum). Twelve of 24 (50%) patients had an early bronch (within 48 hours), with higher identification of bacteria (3/12, 25%), fungi (2/12, 16.7%), and virus (3/12, 25%) compared with those 12 performed later. A trend of more viral infection (6/12, 50%), including CMV, was found in late-performed bronch (>48 hours after diagnosis). The patients with early bronch were sicker, with higher rate of ICU admission (42% vs. 0% in late group) and in-hospital mortality (25% vs. 8% in late group). However, those patients who underwent bronch later had a higher rate of 30-day re-admission (33% vs. 22% in early group).ConclusionBronchoscopy/BAL was the best diagnostic test in patients with AL and CT-confirmed pneumonia, even though it was only performed in 48% of patients. Early bronchoscopy (first 48 hs) has better diagnostic yield than late bronchoscopy (>48 hs), directing the antimicrobial therapy on these patients (based on the identification of bacteria, fungus or viruses), and decreasing the 30-day re-admission rate. Disclosures J. Adachi, Merck: Grant Investigator, Research grant.

2571. Higher Rates of Hospitalization and Infection-Related Hospitalization Among HIV-Exposed Uninfected Infants Compared with HIV Unexposed Uninfected Infants in the United States

BackgroundStudies from multiple countries have suggested impaired immunity in perinatally HIV-exposed uninfected (HEU) children, with elevated rates of all-cause hospitalization and infections. We estimated the incidence of all-cause hospitalization and infection-related hospitalization in the first 2 years of life among HEU children and compared this with HIV-unexposed uninfected (HUU) children in the US Among HEU children, we evaluated associations of maternal HIV disease-related factors during pregnancy with risk of infant hospitalization.MethodsWe evaluated HEU children enrolled in the Surveillance Monitoring for ART Toxicities (SMARTT) Study dynamic cohort of the Pediatric HIV/AIDS Cohort Study (PHACS) network who were born 2006–2017 and followed from birth. Data on HUU children were obtained from the Medicaid Analytic Extract database, restricted to states participating in SMARTT. We compared rates of first hospitalization, total hospitalizations, first infection-related hospitalization, total infection-related hospitalizations, and mortality between HEU and HUU children using Poisson regression. Among HEU children, multivariable Poisson regression models were fit to evaluate associations of maternal HIV factors with risk of hospitalization.ResultsOur analysis included 2,404 HEU and 3,605,864 HUU children. HEU children had approximately 2 times greater rates of first hospitalization, total hospitalizations, first infection-related hospitalization, and total infection-related hospitalizations compared with HUU children (figure). There was no significant difference in mortality. Among HEU children, maternal HIV disease factors, including viral load, CD4 count, antiretroviral regimen, and mode of HIV acquisition, were not associated with hospitalization rates.ConclusionCompared with HUU, HEU children in the United States have nearly twice the rate of hospitalization and infection-related hospitalization in the first 2 years of life, consistent with studies in other countries. Closer monitoring of HEU infants for infection and further elucidation of immune mechanisms is needed.Figure:Rates of Hospitalization and Mortality in First 2 Years of Life (95% CI) among SMARTT HEU versus Medicaid HUU in the United States, 2007–2016Disclosures E. G. Chadwick, Abbott Labs: Shareholder, stock dividends. AbbVie: Shareholder, stock dividends. R. Van Dyke, Giliad Sciences: Grant Investigator, Research grant.

1968. Procalcitonin-Guided Antibiotic Therapy for Lower Respiratory Tract Infections in a US Academic Medical Center

BackgroundEuropean trials using procalcitonin (PCT)-guided antibiotic therapy for patients with lower respiratory tract infections (LRTI) have resulted in significant reductions in antibiotic use without increasing adverse outcomes. Few prospective studies have examined PCT-guided antibiotic therapy for LRTI in the United States. Our objective was to examine whether an PCT algorithm compared with standard practice would reduce antibiotic exposure in patients with LRTI [pneumonia and acute exacerbations of chronic obstructive pulmonary disease (AECOPD)] in an American urban academic hospital.MethodsFrom April 17, 2017 until November 1, 2017, consecutive patients admitted to a medicine service were enrolled in the PCT intervention if they were receiving antibiotics for LRTI and gave consent. Providers were encouraged to discontinue antibiotics using a PCT algorithm with predefined cutoffs. Serum PCT was measured in the hospital laboratory once daily. Results and recommendations were communicated to providers by study team and in the medical record. Control patients were selected by reviewing charts for patients admitted to a medicine service for LRTI from December 1, 2016 to April 16, 2017. The primary endpoint was median antibiotic duration. Overall adverse outcomes at 30 days comprised death, transfer to an intensive care unit, antibiotic side effects, Clostridium difficile infection, disease- specific complications, and new antibiotic prescription for LRTI after discharge.Results174 patients were enrolled in the intervention group and 200 patients in the control group. Intervention group providers complied with the PCT algorithm in 75% of encounters. The rate of overall adverse outcomes was similar in PCT and control groups (21.8% vs. 23.5%; difference, –0.02; 95% CI, –0.10 to 0.07). PCT-guided therapy reduced the median antibiotic duration for pneumonia from 7 days to 6 (P = 0.05), and AECOPD from 4 days to 3 (P = 0.01). Noncompliance with the PCT algorithm resulted in 260 excess antibiotic days in 44 patients.ConclusionIn our center, 75% adherence to a PCT-guided algorithm safely reduced the duration of antibiotics for treating LRTI. Incentivizing providers to comply with PCT-guided algorithms could lead to further reductions in antibiotic use.Disclosures J. Townsend, BRAHMS: Grant Investigator, Research grant.

1570. Infectious Disease (ID) Complications in Immunocompromised (IC) Patients with Cancer Post-Hurricane Harvey at a Comprehensive Cancer Center in 2017

BackgroundDuring 2017, Houston had the most destructive flood-related disaster in recent history due to Hurricane Harvey. Afterward, educational material with information of possible ID problems was provided to all healthcare workers.MethodsProspective surveillance of flood-related ID complications in IC cancer patients. During the 60 days post-Harvey, we monitored referrals to the ID service at MDA Cancer Center. We used the following definitions: Type of exposure: direct to flood water, direct to flooded structures, and others indirect (i.e., prophylaxis). Association risk: “Yes” (direct exposure), “No” (asymptomatic, no exposure, or infection noted prior) and “Probable” (lack of records to establish correlation). Types of infections were classified as soft tissue, gastrointestinal, respiratory, IV line associated or fever. Recommendations were noted including types of antibiotics, vaccinations, or imaging.ResultsA total of 36 cases were referred to our department. Fifty-six percent had exposure to flood-water with/without exposure to structures, 33% to structures only and 11% were other (Figure 1). Regarding the association of an ID problem to flood-exposure, we found an equal distribution of 39% with an association and 39% with a probable association, and the remaining 22% with no association (Figure 2). Of the infections, the majority of infections were respiratory (42%) or soft tissue (31%) (Figure 3). There was a trend of broader antimicrobial coverage for water associated bacteria and mold infections. Only six immunizations recommendations were attained.Figure 2.Figure 3ConclusionTo our knowledge this is the first and largest study of ID complications in IC cancer patients following a natural disaster in medical literature. Our active surveillance showed a lower number of disaster related ID complications than anticipated, possibly because of difficulty determining exposure and underreporting of infections despite active education. Due to individual immunosuppression and exposure, there was variety of recommendations (antimicrobials, studies, or vaccinations). In the event of a weather disaster, we are developing a standard triage survey regarding type of exposure and impact, and also a process for effective immunizations.Disclosures J. Adachi, Merck: Grant Investigator, Research grant

2319. Molecular Epidemiology of Staphyloccoccus aureus Isolated From Korean Children

BackgroundMajor community acquired-methicillin-resistant Staphylococcus aureus (CA-MRSA) clones have been reported worldwide including ST1 in Asia, Europe, United States, ST8 in Europe and United States and ST30 in Australia, Europe and South America. Panton–Valentine leukocidin (PVL) positive ST30-SCCmec IV has been reported as an important CA-MRSA in Singapore, Japan and Latin America, however reports in Korean children are limited. Therefore we investigated the antimicrobial resistance and molecular characteristics of S. aureus among children in Korea.Methods S. aureus previously isolated from children at Seoul National University Bundang Hospital (2010–2016) were analyzed for multilocus sequence type, SCCmec typing, and PCR of PVL, qac A/B, smr and mupA genes. Electronic medical records were reviewed for clinical data and antibiotic susceptibility results.ResultsA total of 79 isolates from skin and soft-tissue infection (SSTI) (N = 41, 51.9%), bone and joint infection (N = 26, 32.9%) and staphylococcal scalded skin syndrome (SSSS) (N = 12, 15.2%) were included. Among these, 32 (40.5%) were MRSA. Among children with underlying diseases (20.3%, 16/79), 56.3% (9/16) were MRSA. After excluding these cases, among children ≤1 month of age, 84.6% (11/13) were MRSA, whereas in children ≥2 months of age, 95.2% (20/21) of SSTI, and 90.0% (18/20) of bone and joint infection were MSSA. All SSSS cases were MRSA. Among MSSA strains, ST30 (N = 28, 59.6%) was the predominant clone and among ST30, 96.6% (28/29) were MSSA. MRSA strains included ST72-SCCmec IV (N = 15, 46.9%), ST89-SCCmec IV (N = 10, 31.3%), ST 5-SCCmec II (N = 3, 9.4%) and ST1-SCCmec IV (N = 3, 9.4%). ST30 was the most common clone in SSTI and bone and joint infection whereas ST89-SCCmec IV was most common in SSSS. PVL was detected in 3 strains (3.8%, ST30-SCCmec IV N = 1, MSSA ST30 N = 2) and qac A/B in 3 strains (MRSA = 3), smr in 3 strains (MSSA = 1, MRSA=2) and mupA in 7 strains (MRSA = 5, MSSA = 2).ConclusionThe molecular epidemiology of S. aureus in Korean children differed from other countries. Among children with SSTI and bone and joint infection, ST30 was the predominant strain, and the majority was MSSA. Among MRSA isolates, ST72-SCCmec type IV was the most common clone in SSTI and bone and joint infection, and ST89-SCCmec type IV in SSSS.Disclosures H. Lee, National Research Foundation of Korea: Grant Investigator, Grant recipient