Background: Candidemia is a leading cause of bloodstream infections (BSIs), and community-onset candidemia is being recognized as a public health problem. In the era of electronic health records (EHRs), we can use machine learning to detect patterns in patient data that may predict infections. Objective: We aimed to predict community-onset candidemia in patients admitted to the University of Iowa Hospital & Clinics (UIHC) using machine-learning algorithms. Methods: We retrospectively reviewed data for patients admitted to UIHC during 2015–2018. All adult inpatients who had a requested blood culture were included. Candidemia was defined as a blood culture positive for Candida within 48 hours after admission. Variables of interest were extracted from the EHR: age, sex, body mass index, and month of admission. We also included comorbidities upon admission defined by the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM): cardiovascular diseases, neurological disorders, chronic pulmonary disease, dementia, rheumatoid disease, peptic ulcer disease, liver disease, diabetes mellitus, hypothyroidism, renal failure, coagulopathy, obesity, weight loss, fluid and electrolyte disorders, anemia, alcohol abuse, drug abuse, psychiatric diseases, malignancy, and HIV/AIDS. We calculated Charlson and Elixhauser scores based on ICD-10-CM codes. We also included prehospitalization conditions (90 days before admission): Candida-positive cultures from sites other than blood, antibiotics/antifungals, hemodialysis, central lines, corticosteroids, surgeries, and intensive care unit (ICU) admissions. Mode and median imputation were used for missing information. Random forests with resampled training sets were used for prediction, and results were evaluated using 10-fold cross validation. Results: In total, 30,528 adult admissions were extracted; 73 admissions had an episode of candidemia (<1%). Median admission age was 61 years, and nearly half of admissions were female patients (44.7%). Mean BMI was 27.67. The most admissions occurred during the months of March, August, and November. The 3 most common ICD-10-CM codes were diabetes mellitus, hypertension, and cancer. Median Charlson and Elixhauser scores were 1 and 2, respectively. The model used 103 variables. The 3 most predictive variables were Elixhauser score on admission, and characteristics in the 90 days prior to admission were Candida from sites other than blood, use of a central line, and recent use of antibiotics/antifungals. The model’s area under the receiver operating characteristic curve was 0.72. Conclusions: Preadmission patient characteristics predicted community-onset candidemia. Machine-learning models may help detect patients eligible for screening for candidemia and prompt empiric antifungal therapy in high-risk patients in the first 48 hours of their admission.

Funding: None

Disclosures: None

Background: In 5 hospitals in Belo Horizonte (population, 3 million) between July 2016 and June 2018, a survey was performed regarding surgical site infection (SSI). We statistically evaluated SSI incidents and optimized the power to predict SSI through pattern recognition algorithms based on support vector machines (SVMs). Methods: Data were collected on SSIs at 5 different hospitals. The hospital infection control committees (CCIHs) of the hospitals collected all data used in the analysis during their routine SSI surveillance procedures; these data were sent to the NOIS (Nosocomial Infection Study) Project. NOIS uses SACIH software (an automated hospital infection control system) to collect data from hospitals that participate voluntarily in the project. In the NOIS, 3 procedures were performed: (1) a treatment of the database collected for use of intact samples; (2) a statistical analysis on the profile of the hospitals collected; and (3) an assessment of the predictive power of SVM with a nonlinear separation process varying in configurations including kernel function (Laplace, Radial Basis, Hyperbolic Tangent and Bessel) and the k-fold cross-validation–based resampling process (ie, the use of data varied according to the amount of folders that cross and combine the evaluated data, being k = 3, 5, 6, 7, and 10). The data were compared by measuring the area under the curve (AUC; range, 0–1) for each of the configurations. Results: From 13,383 records, 7,565 were usable, and SSI incidence was 2.0%. Most patients were aged 35–62 years; the average duration of surgery was 101 minutes, but 76% of surgeries lasted >2 hours. The mean hospital length of stay without SSI was 4 days versus 17 days for the SSI cases. The survey data showed that even with a low number of SSI cases, the prediction rate for this specific surgery was 0.74, which was 14% higher than the rate reported in the literature. Conclusions: Despite the high noise index of the database, it was possible to sample relevant data for the evaluation of general surgery patients. For the predictive process, our results were >0.50 and were 14% better than those reported in the literature. However, the database requires more SSI case samples because only 2% of positive samples unbalanced the database. To optimize data collection and to enable other hospitals to use the SSI prediction tool, a mobile application was developed (available at www.sacihweb.com).

Funding: None

Disclosures: None

Background: The inappropriate and irrational use of antibiotics both in humans and animals causes bacterial resistance. Bacterial resistance is common in low- and middle-income countries, including Bangladesh. Bangladesh has very limited information on antibiotic use and associated resistance. We sought to better understand antibiotic use in low-resource settings for the development of effective strategies to address inappropriate antibiotic use. Methods: We conducted a cross-sectional study among hospitalized children <5 years of age in a tertiary-care hospital in Barishal, Bangladesh, to collect data on antibiotic use. We collected data from 400 children during February–April 2019. Results: Among these 400 children, >50% were aged <1 year, and >60% of these children were boys. The average hospital stay was 3 days (range, 1–14). Most of the children had history of diarrhea and 18% had pneumonia. Most children (82%) were prescribed antibiotics. A combined form of antibiotics was prescribed for 17% of these children. In total, 14 different antibiotics were used. The most commonly used antibiotic was ceftriaxone (57%) followed by azithromycin (14%). The parental route was mostly preferred (75%) for antibiotic administration. Conclusions: Antibiotic prescription was common in children aged <5 years visiting a tertiary-care hospital. Most of the prescribed antibiotics were broad spectrum, which can promote bacterial resistance. Further studies are needed to identify the factors associated with overuse of antibiotics and bacterial resistance in low-resource settings.

Funding: None

Disclosures: None

Background:Acinetobacter spp are gram-negative bacteria that have emerged as a leading cause of hospital-associated infections, most often in the intensive care unit (ICU) setting. This is particularly important in Poland, where the prevalence of A. baumannii in various types of infections, including bloodstream infection (BSI), pneumonia, skin and soft-tissue infection (SSTI), and urinary tract infection (UTI) is higher than in neighboring countries. Recently, other Acinetobacter spp, including A. lwoffii or A. ursingii, have been found to be clinically relevant. In Poland, we have also observed a very rapid increase in antimicrobial resistance, significantly faster for A. baumannii than for other nosocomial pathogens. Methods: A study was conducted in 12 southern Polish hospitals, including 3 ICUs, from January 1 to December 31, 2018. Only adult hospitalized patients were included. Strains were identified using the MALDI-TOF method. Carbapenem resistance was determined using the minimum inhibitory concentration (MIC). Results: During the study, 194 strains belonging to the Acinetobacter genus were isolated. A. baumannii was the dominant species, 88.1% (n = 171), and 23 isolates (11.9%) were other Acinetobacter spp: A. ursingii (n = 5), A. lwofii (n = 4), A. haemolyticus (n = 4), A. junii (n = 3), A. radioresistens (n = 2), A. bereziniae (n = 2), and A. johnsonii (n = 2). Moreover, 15 Acinetobacter strains were collected from ICUs. The most Acinetobacter strains were isolated from SSTIs (n = 115) from non-ICU settings. Non–A. baumannii strains were also most frequently isolated from SSTIs; they constituted 11.3% of all Acinetobacter strains from this type of infection (n = 13). The total Acinetobacter prevalence was 2.6%, whereas the prevalence in the ICU setting was 7%. Acinetobacter prevalence in SSTIs was 10.4%. In pneumonia, Acinetobacter prevalence was 18.6% for ICUs (n = 13) and 2.7% for non-ICUs (n = 46). Strains from UTIs were isolated only with the non-ICU setting, and their prevalence was 0.7% (n = 14). More than half of the tested strains (52.1%) were resistant to carbapenems, but all non–A. baumannii strains were susceptible. The highest resistance to carbapenems was among strains from pneumonia cases in ICUs (58.3%) and resistance among all strains isolated from ICU was 50%. However, even higher resistance was noted among SSTI strains from non-ICUs (61.7%). Conclusions: Increasingly, more than A. baumannii, other species among Acinetobacter strains are isolated from patients hospitalized in Polish hospitals. To assess the significance of non–A. baumannii spp in clinical settings, precise species identification is needed. Therefore, the diagnostic methods used must be improved. Carbapenem-resistant A. baumannii infections are the biggest problem in pneumonia patients in ICUs and in SSTI patients in other hospital departments. Carbapenem resistance occurs in a very high percentage of A. baumannii strains; among non–A. baumannii strains it is not yet a therapeutic problem.

Funding: None

Disclosures: None

Background: Allogeneic bone marrow transplant (BMT) as well as liver, heart, and lung transplant patients have high reported incidence rates of Clostridioides difficile infection (CDI). The prevalence and incidence of asymptomatic colonization with Clostridioides difficile (ACCD) in this group is not known. Methods: ACCD was defined as the presence of C. difficile on screening cultures without positive clinical testing for CDI ±1 week from the date of sampling. Patients undergoing BMT as well as liver, heart, and lung transplants at MUSC between October 2017 and October 2019 were cultured for C. difficile at admission for transplant then once weekly during inpatient admissions and at each outpatient follow-up for 90 days after transplantation. Testing for CDI occurred at the discretion of treating physicians and was done by PCR. Transient ACCD was defined as a positive culture from samples collected <7 days apart, and persistent ACCD was defined as having 2 or more positive cultures collected a minimum of 7 days apart. Results: The baseline prevalences of ACCD were 1 of 5 (20%), 0 of 2 (0%), 1 of 40 (3%), and 2 of 16(13%) for lung, heart, liver and BMT patients, respectively. Of 63 patients, 3 had a pretransplant history of CDI, 2 of whom had baseline ACCD. Incident ACCD occurred in 23 of 63 patients (37%) (Table 1). Overall, ACCD was observed in 30 of 63 patients (48%). Of the 30 patients with ACCD, 14 displayed persistent asymptomatic colonization, whereas 16 displayed transient asymptomatic colonization. Also, 5 patients in the cohort were diagnosed with CDI after transplantation, of whom 3 had ACCD prior to or following CDI. Conclusions: The baseline prevalence of C. difficile colonization in transplant patients (6.3%) was not substantially greater than those observed in recent studies of hospitalized inpatients, but the incidence of new colonization events (37%) was high in this patient population with numerous pretransplant risk factors for CDI.

Funding: None

Disclosures: None

Background: In 2018, the CDC reported that isoniazid (INH)-resistant and multidrug-resistant Mycobacterium tuberculosis (MDR-TB, ie, resistant to at least INH and rifampin) represented 9.3% and 1.5% of TB cases, respectively, in the United States. Objective: We analyzed the prevalence of drug-resistant TB within the Veterans Health Administration (VHA) to determine factors associated with hospitalization. Method: Patients were identified using Council of State and Territorial Epidemiologists case definition for laboratory-confirmed TB by querying VHA data sources from January 1, 2010, to June 30, 2019. Susceptibility results were included for isolates from all body sites. Using a 2-proportion z test, we compared the following demographic factors for susceptible versus drug-resistant TB: age, gender, race/ethnicity, location of residence and birth. We also assessed the following clinical and hospitalization factors: pulmonary versus extrapulmonary disease, latent TB infection (LTBI) screening and treatment, length of stay (LOS), intensive care unit (ICU) stay, and death. Results: In total, 878 patients had lab-confirmed TB, and 812 (92%) had electronic drug-susceptibility results available. Of 812 patients, 49 (6%) showed anti-TB drug resistance (Fig. 1), which was less than that reported nationally by the CDC (Fig. 2). No patients had extensively drug-resistant TB. Only 18 of 49 patients (37%) with resistant TB had LTBI screening ≥3 months prior to diagnosis. Among 6 patients with LTBI, 3 (50%) received treatment. Patient state of residence was the only demographic factor associated with resistant TB. Arizona, Iowa, Massachusetts, Montana, Nevada, South Dakota, and Utah were significantly associated with drug-resistant TB cases (P < .05); however, overall numbers of VHA TB cases in these states were low. Patients with resistant TB were more likely to have extrapulmonary TB (13 of 49, 26.5%) than patients with susceptible TB (111 of 763, 14.5%; P < .05) (Table 1). Mortality at 30 and 90 days for resistant versus susceptible TB cases did not differ significantly: 6.1% versus 5.1% at 30 days (P = .75) and 10.2% versus 10.6% at 90 days (P = 0.92), respectively. The proportion of drug-resistant TB cases hospitalized did not differ from susceptible cases (59% vs 63%), respectively (P = .62) nor did the proportion with ICU stay (14.3% vs 12.3%), respectively (P = .40). Median LOS for drug-resistant TB cases and susceptible cases were similar: 5 days (range, 0–303 versus 4 days (range, 0–111), respectively. Conclusions: Rates of drug-resistant TB are lower in the VHA than in the general US population. However, improvement is needed in LTBI screening and treatment rates. Little has been published on drug resistance in extrapulmonary TB; however, our findings should alert clinicians to the possibility of resistance in these challenging infections.

Funding: None

Disclosures: None

Background: Previously, many infections could be treated effectively based on the clinician’s past clinical experience. The development of resistance to essentially all of the antimicrobial agents currently in use in clinical practice has made this scenario more of the exception than the norm. Selecting an appropriate antimicrobial agent has become increasingly more challenging; the clinician has to navigate through the variety of available agents in the face of increasing antimicrobial resistance. The diagnostic laboratory now plays very important role in clinical practice. To ensure safe and effective empirical treatment, a surveillance study of the susceptibility pattern of common pathogens and appropriate use of antibiotics is imperative. Objective: We report on the prevalence, distribution, and antibiotic susceptibility patterns of nosocomial pathogens isolated at the University of Port Harcourt Teaching Hospital (UPTH) and the effectiveness of the antibiotics commonly prescribed at the hospital in treating these infections. Methods: A retrospective cross-sectional study of specimens received at the microbiology laboratory was conducted over a 6-month period, from October 2015 to March 2016, using urine, blood, and semen specimens. In total, 5,160 samples received and analyzed at the laboratory within the study period were assessed. Results: Of the 5,160 specimens analyzed, 881(17.07%) were positive for bacteria: 691(78.43%) from urine, 86 (9.76%) from blood, and 104 (11. 81%) from semen. Escherichia coli (35.74%), Klebsiella pneumoniae (52.33%), and Staphylococcus aureus (65.4%) were the most frequently isolated pathogens from urine, blood, and semen, respectively. Widespread multidrug resistance was observed among the organisms. Klebsiella pneumoniae, S. aureus, and E. coli isolated from urine were resistant to amoxicillin/clavulanate, cefuroxime, ceftazidime, ciprofloxacin, ampicillin, gentamycin, and ceftriaxone. A review of the pattern of prescribing antibiotics revealed that in the emergency unit, ceftriaxone (34.09%) and metronidazole (30.09%) were most frequently prescribed, whereas in the general outpatient department, metronidazole (19.09%), amoxicillin (16.61%), amoxicillin/clavulanate (9.39%), and ofloxacin (9.39%) were often prescribed. S. aureus was susceptible to only ceftriaxone, whereas K. pneumoniae and E. coli were susceptible only to ofloxacin. Conclusions: Most of the isolated pathogens were not susceptible to the frequently prescribed antibiotics. Empirical prescribing of antibiotics without current epidemiological data of pathogens in the hospital can only further exacerbate the problem of antimicrobial resistance. The need for periodic epidemiological surveillance and rational use of antibiotics anchored on a good antibiotic stewardship program is therefore strongly recommended.

Funding: None

Disclosures: None

Background: The transmissibility of vaccine-strain viruses from immunocompromised patients, such as those with severe combined immune deficiency (SCID) is unknown. The infection control management of a patient diagnosed with SCID and infected with vaccine-strain varicella zoster virus (VZV) and measles virus is described below. A previously healthy, full-term boy was vaccinated at 14 months with measles mumps rubella varicella (MMR) vaccine. He had received prior vaccinations, including rotavirus, without adverse effects. During the 6 weeks after vaccination, the patient developed signs and symptoms clinically consistent with chicken pox and measles. An immune work-up revealed SCID. Methods: The Alberta Health Services (AHS) SCID protocol was followed to manage the patient upon admission at 17 months of age. Multiple meetings with various stakeholders were held to ensure appropriate precautions were followed to minimize the risk of pathogen transmission. Results: The patient was placed on airborne and contact precautions in a negative-pressure room. The pressure differential of the room to the corridor was continually monitored and displayed at the entry to the room. Staff known to be immune to VZV or measles were not required to wear an N95 respirator. All intrahospital movement of the patient was coordinated with the respective care teams and departments, including infection prevention and control, facilities maintenance and engineering, respiratory therapy, and diagnostic imaging. A mask was placed on the patient when movement outside the room was required. VZV testing was positive for the Oka/vaccine strain on all samples tested (ie, nasopharyngeal, skin, blood, and cerebrospinal fluid). Nasopharyngeal swabs and blood were PCR positive for measles genotype A/vaccine strain virus. Both viruses were persistently positive in spite of treatment with acyclovir, valganciclovir, varicella zoster immune globulin, and intravenous immune globulin. Conclusions: There is currently no documented transmission of measles vaccine-strain virus, and transmission of VZV vaccine-strain virus is rare. According to the AHS SCID protocol, the use of airborne and contact precautions for a patient identified with measles and/or VZV supersedes the use of a positive-pressure room for patients identified with SCID. Newborn screening for SCID was implemented in Alberta in June 2019. As a result, more SCID patients will be diagnosed earlier in their course, and therefore prior to most routine vaccinations. However, newborn screening will not pick up some types of combined immune deficiencies. Some children may still be at risk of vaccine-associated illnesses due to undiagnosed underlying immune deficiencies.

Funding: None

Disclosures: None

Background: Appropriate testing of blood procalcitonin (PCT) can potentially inform antibiotic de-escalation in patients with severe infections. When used along with observed clinical improvements, PCT testing can support antimicrobial stewardship. However, this testing must be used optimally to ensure that it is actionable, cost-effective, and provides patient benefit. Although this test is widely used, little is known about the appropriateness of this testing in select populations. Methods: In this retrospective review, we evaluated PCT monitoring patterns and appropriateness of use and relationship to antibiotic days of therapy in a system of community hospitals. We evaluated the use of PCT testing in patients with known confounders, namely pregnancy, chronic kidney disease, or neutropenia, which we classified as “inappropriate use” because these conditions can affect the interpretation of PCT results. We also evaluated the relationship between PCT testing and antibiotic days of therapy for patients with sepsis, pneumonia, or lower respiratory tract infections. Results: In a 1-year period, ∼206,302 PCT tests were performed at 146 facilities, an average of ∼1,413 per facility per year. Approximately 27.7% of these tests were given to patients who were pregnant or had a confounding comorbidity such as chronic kidney disease or neutropenia. Of these “inappropriate” tests, >90% were given to patients with chronic kidney disease. Older patients (aged 60–80 years, n = 93,021) were more likely to receive a PCT test while also having a confounding comorbidities; 24% of older patients with a PCT test also had chronic kidney disease. Of all patients with a PCT test and chronic kidney disease, ∼76% were also diagnosed with either sepsis, pneumonia, or lower respiratory tract infections. Conclusions: Confounding conditions can affect PCT levels independently of infection. Additionally, some clinicians use PCT tests as probes for other physiological maladies. This analysis demonstrated that there is opportunity for education about the appropriate use of this test, how to interpret results in the presence of confounding conditions, and how to transform PCT test results into actions that facilitate antimicrobial stewardship and better patient care.

Funding: None

Disclosures: None

Background: Effective medical device reprocessing (MDR) is essential in preventing the spread of microorganisms and maintaining patient safety. Alberta Health Services (AHS) is an Alberta-wide, integrated health system, responsible for delivering health services to >4.3 million people living in the province. In 2010, periodic province-wide MDR reviews were initiated by the provincial health system to verify that the cleaning, disinfection, and sterilization of reusable critical and semicritical medical devices met established standards. To date, there have been 3 review cycles; in cycle 3, a follow-up process for tracking and reporting corrective actions was initiated. Methods: As in previous MDR review cycles, cycle 3 included the use of a standardized suite of tools to measure compliance with standards set by Accreditation Canada, the Canadian Standards Association, and the Government of Alberta. Each cycle involved a review of MDR areas completed by trained reviewers. Interrater reliability among reviewers was maintained through training and debriefings following reviews to ensure agreement. Following reviews, reports were generated for areas, zones, and AHS. As part of the corrective actions and follow-up process, identified deficiencies were categorized into 5 themes. Corrective actions were tracked and periodic reports were generated showing the progress of deficiency resolution. Resolution rates (number of resolved deficiencies divided by total number of of deficiencies) were calculated for each of the identified themes as well as overall for cycle 3. Results: Overall compliance for cycle 3 was 93%. Cycle 3 reviews revealed that more than half of the deficiencies (58%) were identified previously in cycle 2. The resolution rates ranged from 78% to 95% for identified deficiencies for 4 of the 5 themes: documentation, technique, PPE/attire/hand hygiene, and other. The theme related to physical infrastructure showed a considerably lower resolution rate of 49%. The corrective action follow-up process showed increased overall resolution rate from 59% at the start of the follow-up process to 82% at its completion. When this resolution rate was applied to the initial survey compliance rate for cycle 3, overall compliance increased to 99%. Conclusions: Monitoring quality of MDR practices is essential in maintaining and improving patient safety. The standardized provincial review process identified common themes and a coordinated approach to support the resolution of many identified deficiencies. Most of those deficiencies were resolved; however, those deficiencies related to physical infrastructure of the MDR department continue to be seen across review cycles. This review process with follow up of these deficiencies can help bring attention to organization leadership and Funding: authorities during budget cycles.

Funding: None

Disclosures: None

Background: The WHO recommends the establishment of sustainable and evidence-based surveillance systems are recommended for the prevention of microbial resistance. For these surveillance systems, all medical microbiology laboratories are required to participate an external quality assessment (EQA) program that covers antimicrobial susceptibility testing (AST). Clinical microbiology EQA panels with 3 isolates have been provided 3 times per year to antimicrobial resistance (AMR) sentinel laboratories in Cambodia since 2012. We evaluated the performance of laboratory testing implemented between 2016 and 2019, based on 4 years EQA results to highlight the main sources of unsatisfactory analytical processes and to suggest areas for improvement. Methods: We analyzed the results of microbiology EQA in 7 AMR surveillance sentinel laboratories from 2016 to 2019, which were coordinated by the National Institute of Public Health (NIPH) under the program of Pacific Paramedical Training Centre (PPTC) from New Zealand. All participating laboratories were required to identify bacteria to the species level, to verify AST results, and to answer a case study question on parasitology. Feedback results and appropriate corrective actions were reviewed to identify the root cause of nonconformity and to suggest areas for improvement. Results: Proficiency test results of participating laboratories from 9 cycles with 27 isolates were analyzed. The overall average of EQA result was 94.0%. The laboratories failed to identify the isolated pathogens in 7.0% of the tests and failed to interpret the inhibition zone of AST (ie, resistant, intermediate or susceptible) in 6.0% of tested strains. The main causes of erroneous of PT results were either preanalytical (ie, handling of the samples, timing of analysis, equipment and reagent management), analytical (ie, quality control, unsuitable methods, confusion of samples, or errors of confirmation), or postanalytical mistakes (eg, interpretation guideline, cross-checking of results, or the information management system). Followed by the root causes, internal quality control and inventory management were the highest-priority suggestions for improvement. Conclusions: All participating laboratories showed good performance on EQA for evidence-based AMR surveillance. The national antimicrobial resistance data quality is sufficiently good and the data should be shared on national and international platforms. However, the regular monitoring of national AMR surveillance system should be conducted for continued quality improvement.

Funding: None

Disclosures: None

Background: More than 450,000 patients receive outpatient hemodialysis in the United States. Patients on hemodialysis are at high risk of bloodstream infections (BSIs), which are associated with significant morbidity and mortality. National prevention efforts targeting hemodialysis facilities have resulted in widespread changes in practice, including modifications to central venous catheter (CVC) maintenance procedures. We analyzed dialysis event surveillance data submitted to the CDC NHSN to describe changes in BSI rates among hemodialysis outpatients from 2014 to 2018. Methods: Outpatient hemodialysis facilities report BSIs (ie, positive blood cultures collected in the outpatient setting or within 1 calendar day after hospital admission) and the number of hemodialysis outpatients treated during the first 2 working days of each month to the NHSN. For each BSI, the suspected source (ie, vascular access, another site, contamination, or uncertain) and vascular access type are indicated: CVC, arteriovenous fistula (AVF) or arteriovenous graft (AVG). Pooled mean rates (per 100 patient months) were calculated for BSIs, access-related BSIs (ARBSIs), and BSIs and ARBSIs were stratified by vascular access type. Annual BSI rate trends were evaluated using a negative binomial regression model, which treated patient months as an offset variable and included access type, year, and an access-year interaction variable. Results: More than 6,000 outpatient hemodialysis facilities reported 134,961 BSIs from 2014 to 2018. Of these BSIs, 102,505 (76%) were categorized as access related. CVCs were present in 63% of BSIs and 70% of ARBSIs. Pooled mean BSI rates decreased 27% from 0.64 to 0.47 per 100 patient months; rates of ARBSIs decreased 27% from 0.49 to 0.36 per 100 patient months. Significant decreases in event rates occurred across vascular access strata (Fig. 1). The reduction in BSI and ARBSI burden was most pronounced among patients with CVCs. BSI rates in patients with CVCs decreased 32% from 2.16 per 100 patient months to 1.46 (annual average decrease, 9.5%), and ARBSI rates in patients with CVCs decreased 32% from 1.83 per 100 patient months to 1.24 (annual average decrease, 9.4%). Conclusions: Substantial reductions in BSI and ARBSI rates among hemodialysis outpatients occurred during this 5-year period, and these reductions appear to be most prominent among CVC and AVF patients. Improvements in infection prevention and control practices, including CVC care, have likely contributed to these reductions. Additional efforts to increase the uptake of known prevention practices and to identify new strategies for prevention might contribute to continued decreases in infections among this highly vulnerable population.

Funding: None

Disclosures: None

Background: Hepatitis A cases have been regularly increasing in Florida since 2016, after remaining relatively stable for several years. Between January 1, 2019, and October 19, 2019, 2,847 cases of hepatitis were reported to the Florida Department of Health, including 2,209 hospitalizations and 41 deaths due to infection with hepatitis A1. At least 93% of these cases was acquired within the state of Florida.1 During this outbreak, Pasco County clearly emerged as a focal point for new cases. As of week 42 of 2019, Pasco County was the second-highest county for new cases (n = 484) and the leading county for deaths due to hepatitis A infection (n = 8). Risk factors identified from hepatitis A cases reported between January 1, 2019, and October 19, 2019, included any drug use (57%), injection drug use (37%), noninjection drug use (35%), homelessness (22%), and being a man who had sexual intercourse with other men (MSM, 4%).1 Moreover, 33% of newly reported cases were interviewed and denied risk factors; 1 (20%) had recently been incarcerated or was currently incarcerated.1 Individuals with these risk factors are often individuals who also require emergency medical services (EMS). On August 1, 2019, a public health emergency (PHE) was issued for the entire state of Florida. The existence of a PHE allows for paramedics to administer vaccine (with the approval of the fire rescue medical director). Pasco County Fire Rescue routinely operates 28 advanced life support (ALS) ambulances with either a crew of 2 paramedics or a paramedic/EMT pair. On October 7, 2019, the Pasco County Board of County Commissioners (BCC) approved a Memorandum of Understanding between the Florida Department of Health in Pasco County, the Pasco County Fire Rescue Medical Director, and the BCC for ALS crews to provide hepatitis A vaccination to noncritical (ie, basic life support) patients during routine contact. This effort appears to be the first within the of Florida. Pasco County Fire Rescue (PCFR) began stocking hepatitis A vaccine on January 1, 2020 and ceased participation in the program on March 1, 2020. During the window of participation, PCFR responded to approximately 6,570 calls in which the patient would have been non-critical and eligible for hepatitis A vaccination. Twelve hepatitis A vaccinations (0.18%) were administered. Crews verbalized concerns about adding vaccination to their job description, feelings of being overwhelmed with existing emergency medical services (EMS) responsibilities, apathy regarding the potential benefit of provision of vaccine, and perception that successful participation in a vaccine campaign would lead to additional community paramedicine responsibilities. The vaccine program was terminated early due to the significant demands on EMS as SARS-CoV2 impacted Pasco County. This trial highlights the need for crew investments and buy-in for future partial or full community paramedicine initiatives.

1. Florida Department of Health, Hepatitis A Surveillance State Report, Week 42.

Funding: None

Disclosures: None

Background: Central-line–associated bloodstream infections (CLABSIs) are a significant contributor to morbidity and mortality for neonates; they also increased healthcare costs and duration of hospitalization. This population is susceptible to infections because of their undeveloped immune systems, and they require intravenous access until they can tolerate enteral feedings, which for extremely premature infants can take several weeks (if not months) to achieve. Our hospital is a regional-referral teaching hospital with 772 licensed beds. The neonatal intensive care unit (NICU) is a level 3, 35-bed unit where the most critically ill neonates receive care. After a sustained 3-year period of zero CLABSIs, we identified 10 infections between September 2016 through April 2018. Methods: A multidisciplinary team known as the neonatal infection prevention team (NIPT) was reinstated. This team included members from nursing and infection prevention (IP) and from NICU Shared Governance, as well as a neonatal nurse practitioner (NNP) and a neonatologist to review these CLABSIs. Evidence-based practices, policies, and procedures were implemented to help reduce CLABSIs. Nurse educators provided education and training. The infection prevention team reinstated and modified the central-line maintenance and insertion tools to document compliance and to identify any gaps in care. Nurses were expected to document line maintenance once per shift (a.m. and p.m.). All CLABSIs were entered into the CDC NHSN and the hospital’s safety event reporting system, which required follow-up by a clinical manager. The infection prevention team monitored NHSN standardized infection ratios (SIRs) monthly. The SIR is the number of observed events divided by the number predicted (calculated based on national aggregate data). Results: The highest reported quarterly SIR was 1.423, which occurred in the third quarter of 2018 (Fig. 1). Overall compliance with line maintenance protocols was 86% on the morning shift and 89% on the afternoon shift. With implementation of an evidence-based bundle, the NICU had a rolling 12-month SIR of 0.00 as of October 2019. Conclusions: Multidisciplinary team development, implementation of evidence-based bundle elements, and education on catheter care contributed to the long-term success in decreasing CLABSI rates in our NICU. Although this implementation achieved a zero CLABSI rate, we experienced some barriers, including compliance issues with staff not completing the audit tools, staff turnover, and high patient census.

Funding: None

Disclosures: None

Background: One of the strategies to reduce healthcare-associated infections (HAIs) and promote the quality of disease prevention and control actions is the creation of a hospital infection control program. This program is a set of deliberately and systematically developed actions aimed toward reducing the incidence and severity of infections to the maximum extent possible. In Brazil, studies on the subject still need to be improved; they focus on structural and process assessments, especially the survey of continuing education indicators as a quality requirement for the prevention of HAIs. The organizational context does not contribute to the success of the program, and difficulties remain in implementing recommendations and in implementing patient safety policies. Objective: To analyze hospital infection control programs in relation to quality components. Methods: This cross-sectional epidemiological study was conducted in health services located in the 5 official regions of Brazil: Midwest, Northeast, North, Southeast, and South. To select the study sites, nonprobabilistic sampling using the snowball technique was used. The potential study population consisted of 114 hospital infection control services. Health professionals responded to the structured instrument sent electronically via e-mail, and other health services near their locality, until reaching a national proportion. We used the “Hospital Infection Control Program Evaluation Questionnaire”; it consists of 36 multiple-choice questions. This tool was validated by 96 expert judges using the Cronbach’s alpha test (0.82) and the content validity index (0.88). A data analysis was performed using the multivariate principal component analysis technique (PCA). Results: Overall, 13 PCA components (Fig. 1) were used to build a score for measuring the performance of the hospital infection control program (ie, IQPC score). The Southern region had the best performance of the hospital infection control program (mi = 1.50; P = .02) (Fig. 2), private administration (mi = 0.45; P = .05), of hospitals that contained 300 beds or (mi = 1.38; P < .01), hospitals that used the NHSN criterion for HAI surveillance (mi = 2.12; P < .01), and those who searched prospective activity as a surveillance method (mi = 0.51; P < .01). Conclusions: The quality of nosocomial infection control programs still needs to be improved among health services, highlighting the need to invest in small, publicly managed hospitals that use retrospective active surveillance methods.

Funding: None

Disclosures: None

Background: Healthcare personnel (HCP) acquire MRSA on their gown and gloves during routine care activities for patients who are colonized or infected with MRSA at a rate of ∼15%. Certain care activities (eg, physical exam, care of endotracheal tube, wound care and bathing/hygiene) have been associated with a higher frequency of transmission from the patient to HCP gown and gloves than other activities (ie, administration of oral medicines, glucose monitoring, and manipulation of IV tubing/medication delivery). However, quantification of MRSA contamination and risk to subsequent patients is poorly defined. Objective: We sought to determine the mean MRSA colony-forming units (CFU) found on the gloves and gowns of HCP who acquire MRSA after various care activities involving patients with MRSA. Methods: We conducted a prospective cohort study at the University of Maryland Medical Center from December 2018 to October 2019. We identified patients colonized or infected with MRSA based on culture data from the prior 7 days. HCP performing prespecified care activities on eligible patients were observed. To isolate the risk of each care activity, HCP donned new gloves and gown prior to a specific care activity. Once that care activity was performed, HCP gloves and gown were swabbed prior to the any further care activities. HCP gloves were cultured with an E-swab by swabbing each digit up and down 3 times followed by 2 circles on the palm of their hands. HCP gowns were sampled by swabbing a 15 × 30-cm area along the beltline of the gown and along each inner forearm twice. E-swab liquid was then serially diluted and plated in triplicate on CHROMagar MRSA II (BD, Sparks, MD) to obtain CFU. We calculated the median CFUs and the interquartile range (IQR) for each specific care activity stratified by gown and gloves. Results: In total, 604 HCP–patient care interactions were observed. Table 1 displays the mean MRSA CFUs stratified by gown and gloves for each patient care activity of interest. Conclusions: The quantity of MRSA found on gowns and gloves varies depending on patient care activities. Recognition of differential transmission rates between various activities may allow different approaches to infection prevention, such as the use of personal protective equipment in high- versus low-risk activities and/or the use of more aggressive interventions for high-risk activities.

Funding: None

Disclosures: None

Background: The healthcare environment can serve as a reservoir for many microorganisms and, in the absence of appropriate cleaning and disinfection, can contribute to pathogen transmission. Identification of high-touch surfaces (HTS) in hospital patient rooms has allowed the recognition of surfaces that represent the greatest transmission risk and prioritization of cleaning and disinfection resources for infection prevention. HTS in other healthcare settings, including high-volume and high-risk settings such as emergency departments (EDs) and hemodialysis facilities (HDFs), have not been well studied or defined. Methods: Observations were conducted in 2 EDs and 3 HDFs using structured observation tools. All touch episodes, defined as hand-to-surface contact regardless of hand hygiene and/or glove use, were recorded. Touches by healthcare personnel, patients, and visitors were included. Surfaces were classified as being allocated to individual patients or shared among multiple patients. The number of touch episodes per hour was calculated for each surface to rank surfaces by frequency of touch. Results: In total, 28 hours of observation (14 hours each in EDs and HDFs) were conducted. 1,976 touch episodes were observed among 62 surfaces. On average, more touch episodes were observed per hour in HDFs than in EDs (89 vs 52, respectively). The most frequently touched surfaces in EDs included stretcher rails, privacy curtains, visitor chair arm rests and seats, and patient bedside tables, which together accounted for 68.8% of all touch episodes in EDs (Fig. 1). Frequently touched surfaces in HDFs included both shared and single-patient surfaces: 27.8% and 72.2% of HDF touch episodes, respectively. The most frequently touched surfaces in HDFs were supply cart drawers, dialysis machine control panels and keyboards, handwashing faucet handles, bedside work tables, and bed rail or dialysis chair armrests, which accounted for 68.4% of all touch-episodes recorded. Conclusions: To our knowledge, this is the first quantitative study to identify HTSs in EDs and HDFs. Our observations reveal that certain surfaces within these environments are subject to a substantially greater frequency of hand contact than others and that a relatively small number of surfaces account for most touch episodes. Notably, whereas HTSs in EDs were primarily single-patient surfaces, HTSs in HDFs included surfaces shared in the care of multiple patients, which may represent an even greater risk of patient-to-patient pathogen transmission than single-patient surfaces. The identification of HTSs in EDs and HDFs contributes to a better understanding of the risk of environment-related pathogen transmission in these settings and may allow prioritization and optimization of cleaning and disinfection resources within facilities.

Funding: None

Disclosures: None

Background: The clinical picture of influenza-like illness can mimic bacterial pneumonia, and empiric treatment is often initiated with antibacterial agents. Molecular testing such as polymerase chain reaction (PCR) is often used to diagnose influenza. However, traditional PCR tests have a slow turnaround time and cannot deliver results soon enough to influence the clinical decision making. The Detroit Medical Center (DMC) implemented the Xpert Flu test for all patients presenting with influenza-like illness (ILI). We evaluated antibacterial use after implementation of rapid influenza PCR Xpert Flu. Methods: We conducted a retrospective study comparing all pediatric and adult patients tested using traditional RT PCR during the 2017–2018 flu season to patients tested using the rapid influenza Xpert Flu during the 2018–2019 flu season in a tertiary-care hospital in Detroit, Michigan. These patients were further divided into 3 groups: not admitted (NA), admitted to acute-care floor (ACF), or admitted to intensive care unit (ICU). The groups were then compared with respect to percentage of antibacterial use after traditional RT PCR versus rapid influenza Xpert Flu testing during their hospital visit for ILI. The χ2 test was used for statistical analyses. Results: In total, 20,923 patients presented with influenza-like illness during the study period: 26% (n = 5,569) had the rapid influenza Xpert Flu and 73.4% (n= 15,354) had traditional RT PCR. For a comparison of the number of patients in 3 groups (NA, ACF, and ICU) and type of influenza PCR performed among these patients, please refer to Table 1. When comparing antibacterial use in the NA group, the proportions of patients who received antibacterial agents in the traditional RT PCR group versus the rapid influenza Xpert Flu group were 24.4% (n = 695) versus 3.9% (n = 450), respectively (P < .0001). In the ACF group, the proportions of patients who received antibacterial agents in the traditional RT PCR group versus the rapid influenza Xpert Flu group was 62.3% (n = 1,406) versus 27.7% (n = 994), respectively (P < .001). In the ICU group, the proportions of patients who received antibacterials in the traditional RT PCR group versus the rapid influenza Xpert Flu group were 80.3% (n = 382) versus 38.3% (n = 204), respectively (P < .0001). Conclusions: With rising antimicrobial resistance and increasing influenza morbidity and mortality, rapid diagnostics not only can help diagnose influenza faster but also can reduce inappropriate antimicrobial use.

Funding: None

Disclosures: None

Background:C. difficile infection has been a significant cause of morbidity and mortality over the past decade. Our hospital had rates of hospital-onset, laboratory-identified, C. difficile infection (HO-CDI) that were significantly higher than our state and national benchmarks. HO-CDI is defined as a test positive for C. difficile occurring on or after day 4 of hospitalization, regardless of the presence of symptoms. New leadership at the hospital sought a creative way to engage staff in finding solutions to our high rates of HO-CDI. Objective: The purpose of this intervention was to engage frontline staff in reporting and solving patient care situations that may increase infection risk to decrease HO-CDI rates. Methods: Starting in July 2015, real-time bedside RCAs were performed weekly for any HO-CDI on the unit to which the infection was attributed and on any unit from which the patient had been recently transferred. Top clinical leadership of the hospital, and all services and departments, physicians, nurses, and others involved with the patient’s care were expected to attend and identify factors that may have contributed to the infection. The findings were documented, and changes to care were made based on the findings. The rate of incident hospital onset HO-CDI per 10,000 patient days was used to measure outcome because standardized infection ratios for the period before 2016 were not available. Results: Staff members suggested 6 specific actions that were undertaken to decrease HO-CDI risk (Table 1). The HO-CDI rate during the preintervention period (2012–2014) was 6.85 per 10,000 patient days (275 cases). In the postintervention period (2016–2018) the HO-CDI rate was 3.13 per 10,000 patient days (101 cases). There was a 54% reduction in the HO-CDI rate in the postintervention period (P < .001). Conclusions: The multidisciplinary bedside RCA process resulted in staff providing recommendations for actions to reduce HO-CDI risk. Implementation of staff suggestions resulted in a sustained, significant decrease in HO-CDI.

Funding: None

Disclosures: None

Background: The Bundled Payment Care Improvement Program is a CMS initiative designed to encourage greater collaboration across settings of care, especially as it relates to an initial set of targeted clinical episodes, which include sepsis and pneumonia. As with many CMS incentive programs, performance evaluation is retrospective in nature, resulting in after-the-fact changes in operational processes to improve both efficiency and quality. Although retrospective performance evaluation is informative, care providers would ideally identify a patient’s potential clinical cohort during the index stay and implement care management procedures as necessary to prevent or reduce the severity of the condition. The primary challenges for real-time identification of a patient’s clinical cohort are CMS-targeted cohorts are based on either MS-DRG (grouping of ICD-10 codes) or HCPCS coding—coding that occurs after discharge by clinical abstractors. Additionally, many informative data elements in the EHR lack standardization and no simple and reliable heuristic rules can be employed to meaningfully identify those cohorts without human review. Objective: To share the results of an ensemble statistical model to predict patient risks of sepsis and pneumonia during their hospital (ie, index) stay. Methods: The predictive model uses a combination of Bernoulli Naïve Bayes natural language processing (NLP) classifiers, to reduce text dimensionality into a single probability value, and an eXtreme Gradient Boosting (XGBoost) algorithm as a meta-model to collectively evaluate both standardized clinical elements alongside the NLP-based text probabilities. Results: Bernoulli Naïve Bayes classifiers have proven to perform well on short text strings and allow for highly explanatory unstructured or semistructured text fields (eg, reason for visit, culture results), to be used in a both comparative and generalizable way within the larger XGBoost model. Conclusions: The choice of XGBoost as the meta-model has the benefits of mitigating concerns of nonlinearity among clinical features, reducing potential of overfitting, while allowing missing values to exist within the data. Both the Bayesian classifier and meta-model were trained using a patient-level integrated dataset extracted from both a patient-billing and EHR data warehouse maintained by Premier. The data set, joined by patient admission-date, medical record number, date of birth, and hospital entity code, allows the presence of both the coded clinical cohort (derived from the MS-DRG) and the explanatory features in the EHR to exist within a single patient encounter record. The resulting model produced F1 performance scores of .65 for the sepsis population and .61 for the pneumonia population.

Funding: None

Disclosures: None