University Of Minnesota Medical Center Research Articles

Applying Machine Learning Across Sites: External Validation of a Surgical Site Infection Detection Algorithm

Surgical complications have tremendous consequences and costs. Complication detection is important for quality improvement, but traditional manual chart review is burdensome. Automated mechanisms are needed to make this more efficient. To understand the generalizability of a machine learning algorithm between sites, automated surgical site infection (SSI) detection algorithms developed at one center were tested at another distinct center. NSQIP patients had electronic health record (EHR) data extracted at one center (University of Minnesota Medical Center, Site A) over a 4-year period for model development and internal validation, and at a second center (University of California San Francisco, Site B) over a subsequent 2-year period for external validation. Models for automated NSQIP SSI detection of superficial, organ space, and total SSI within 30 days postoperatively were validated using area under the curve (AUC) scores and corresponding 95% confidence intervals. For the 8,883 patients (Site A) and 1,473 patients (Site B), AUC scores were not statistically different for any outcome including superficial (external 0.804, internal [0.784, 0.874] AUC); organ/space (external 0.905, internal [0.867, 0.941] AUC); and total (external 0.855, internal [0.854, 0.908] AUC) SSI. False negative rates decreased with increasing case review volume and would be amenable to a strategy in which cases with low predicted probabilities of SSI could be excluded from chart review. Our findings demonstrated that SSI detection machine learning algorithms developed at 1 site were generalizable to another institution. SSI detection models are practically applicable to accelerate and focus chart review.

Hypoglycemia: Comparison of Health Status Outcomes Between Patients After Allogeneic Hematopoietic Cell Transplantation.

Patients who have undergone hematopoietic cell transplantation (HCT) often face lengthy hospital stays. Hospitalized patients' compromised health status puts them at risk for complications to recovery when glucose is not controlled. This study aimed to investigate differences in outcomes in patients who experienced hypoglycemia compared to patients who did not experience hypoglycemia post-allogeneic HCT. A retrospective chart review and secondary data analysis were conducted. The sample consisted of 198 adult patients hospitalized for their first allogeneic HCT at the University of Minnesota Medical Center between August 2015 and December 2017. Hypoglycemic patients were compared with nonhypoglycemic patients until discharge or 100 hospitalization days post-transplantation. A total of 20 patients (10%) experienced hypoglycemic events during the study time frame. There were significant differences between the two groups. Hypoglycemia may be a marker for higher acuity illness in this population. Nurses should increase vigilance in managing the blood glucose levels of patients undergoing HCT with known comorbidities and complications.

202. Implementation of an Outpatient Parenteral Antimicrobial Therapy (OPAT) Collaborative for Patients with Staphylococcus aureus or Gram-Negative Bacilli Bacteremia Requiring Home Infusion: The PANTHIR Program

Background Staphylococcus aureus (SA) and Gram-negative bacilli (GNB) bacteremia often require prolonged treatment courses due to high morbidity and mortality risk. Outpatient parenteral antibiotic therapy (OPAT) has emerged as a preferred delivery method. Few data have been published regarding the follow-up and adverse event rates among OPAT patients. We describe outcomes in patients with SA or GNB bacteremia transitioning from an academic medical center to home infusion, prompting the implementation of the Parenteral ANtimicrobial therapy Transitions to Home Infusion Review (PANTHIR) program.MethodsA retrospective chart review of adult patients with SA or GNB bacteremia at the University of Minnesota Medical Center requiring home infusion represent a 26-month period. Baseline outcomes, including 30-day hospital readmissions and adverse drug events (ADEs), were calculated. The PANTHIR program was launched as an interdisciplinary collaborative with an infectious diseases (ID) provider, pharmacists, and home infusion specialists. Core program elements include inpatient identification, ID pharmacist review, care plan documentation and communication, and OPAT program measures.ResultsThe retrospective cohort included 69 patients. 23.2% experienced a hospitalization within 30 days of discharge and 26.1% experienced an ADE (Table 1). The mean duration of therapy was 22 days. No patient received aminoglycosides and one required vancomycin. A primary goal was to improve the continuity of care for potentially life-threatening bacteremia during the vulnerable inpatient to outpatient transition. Electronic health record functionality allowed for creation of an OPAT navigator for infectious diseases (ID) pharmacist transition plan documentation, electronic communication with designated provider and home infusion pharmacist, and retrieval of focal data points for ongoing program evaluation. 28 patients have been enrolled in the PANTHIR program with outcomes data collection underway.Table 1. Retrospective data among University of Minnesota Medical Center patients hospitalized with SA or GNB bacteremia requiring home infusion on discharge. ConclusionHospital readmission rates and ADEs are frequent among patients with SA or GNB bacteremia requiring OPAT via home infusion. An ID pharmacist-directed program in collaboration with an ID provider is feasible for OPAT transitions and may serve as a roadmap for other institutions.Disclosures Dana Simonson, PharmD, BCPS, Janssen (Advisor or Review Panel member, Other Financial or Material Support, Webinar Series Speaker Fall 2019)

1108. Transplantation and Immigration: Comparing Infectious Complications Between Foreign-born vs. U.S.-born Kidney Transplant Recipients in Minnesota

BackgroundImmigrant patients face barriers to kidney transplantation due to language, cultural, and economic issues. Unprepared health systems and providers further contribute to health disparities in transplantation. Foreign-born patients are also at risk for reactivation of latent infections which differ from U.S.-born population. Stratifying transplant recipients according to country of birth could guide clinicians in the prevention, anticipation, diagnostics, and treatment of post-transplant infections.MethodsA retrospective, observational, multicenter study of patients that underwent kidney transplantation from 1/2014-12/2018 at the University of Minnesota Medical Center and Hennepin Healthcare is being conducted. Sociodemographic, clinical, and laboratory data are collected, including infectious episodes during the first year post-transplant.ResultsOne-hundred patients are included in this preliminary analysis (recruitment goal is 800 patients). Sixty-five patients were males (65%), with median age 56 years (range 20 – 77). The majority were Caucasians (64%), followed by Asians (12%) and Africans (9%). Living donation was 59%. Seventy-eight patients developed infectious complications during the first year after transplantation, for a total of 175 infectious episodes: viral etiology (51%), followed by bacterial (42%) and fungal (7%). No tropical diseases were found. Comparing foreign-born (30%) vs. U.S.-born (70%), foreign-born recipients had a higher frequency of latent tuberculosis infection (LTBI) (37% vs 1%, p< 0.001), hepatitis B core antibody positive (20% vs 0, p< 0.001), and deceased donor transplant (67% vs 30%, p= 0.001). CMV mismatch (3% vs 36%, p= 0.002) was more frequent in U.S.-born recipients; CMV reactivation was similar in both groups. While not statistically significant, more foreign-born recipients had an infection in the first year post-transplant (90% vs 73%, p= 0.1), and higher median infectious episodes (2 vs 1, p= 0.6).Comparison of foreign-born vs U.S.-born kidney transplant recipients (n=100, preliminary data) ConclusionPer this preliminary data, foreign-born transplant recipients had a higher frequency of LTBI, hepatitis B core antibody and infectious complications, but lower frequency of CMV mismatch. No cases of tuberculosis, hepatitis B reactivation or tropical diseases were observed.Disclosures Patricia F. Walker, MD, DTM&H, FASTMH, UpToDate (Other Financial or Material Support, Royalties from UpToDate for chapter on refugee screening)

133. Creation of an Emergency Department (ED)-Specific Urine Antibiogram and Evaluation of Urinary Tract Infection (UTI) Prescribing Practices at a Tertiary Academic Medical Center

BackgroundInpatient settings have been the focus of most antimicrobial stewardship (AS) practices; however, the Joint Commission has now published requirements for accredited ambulatory healthcare organizations to implement AS. This analysis aimed to create a urine isolate antibiogram and to evaluate microbiological data and prescribing practices for UTI patients in the ED.MethodsThis retrospective cohort study included adults admitted to either of two EDs at University of Minnesota Medical Center and diagnosed with a UTI between 1/1/2018 and 12/31/2018. Patients were excluded if subsequently admitted to an inpatient unit or if they had a repeat culture growing the same organism as the index isolate. Diagnosis of cystitis versus pyelonephritis was based on ICD-10 coding. Data including urinalysis and culture results, susceptibilities, empiric antibiotic selection, and readmissions were collected.ResultsData from 350 isolates were collected for inclusion in the antibiogram. Patient characteristics corresponding to this isolate collection included 78.9% female, median age of 41 years, and 64.6% diagnosed with pyelonephritis. Escherichia coli was the most common organism (70%), followed by Klebsiella pneumoniae (5.4%) and Proteus mirabilis (4%). Combined susceptibilities of E. coli isolates from both EDs were: 49.4% ampicillin, 55.7% ampicillin/sulbactam, 69.8% sulfamethoxazole/trimethoprim, 83.7% ciprofloxacin (CIP), 85.7% cefazolin, 93.9% ceftriaxone, and 94.3% nitrofurantoin (NIT). The most common discharge antibiotics prescribed for cystitis patients were NIT (29.8%) and cephalexin (25%). Pyelonephritis patients were most frequently prescribed CIP (32.3%) and cefdinir (14.2%). Drug-bug mismatches occurred in 19.1% of patients (10.5% cystitis vs. 23.9% pyelonephritis). The rates of ED readmission within 96-hours and inpatient admission within 30 days, for any reason, were 4.3% and 9.1% respectively.ConclusionBased on the ED-specific urine antibiogram generated, NIT (cystitis) and CIP (pyelonephritis) could be considered first-line agents for empiric treatment of UTI at our institution. Drug-bug mismatches were more common in pyelonephritis patients. These data will be used to develop a treatment algorithm aimed at improving treatment of UTI in the ED.Disclosures Elizabeth B. Hirsch, PharmD, Merck (Grant/Research Support)Nabriva Therapeutics (Advisor or Review Panel member)

Advanced reperfusion strategies for patients with out-of-hospital cardiac arrest and refractory ventricular fibrillation (ARREST): a phase 2, single centre, open-label, randomised controlled trial

Among patients with out-of-hospital cardiac arrest (OHCA) and ventricular fibrillation, more than half present with refractory ventricular fibrillation unresponsive to initial standard advanced cardiac life support (ACLS) treatment. We did the first randomised clinical trial in the USA of extracorporeal membrane oxygenation (ECMO)-facilitated resuscitation versus standard ACLS treatment in patients with OHCA and refractory ventricular fibrillation. For this phase 2, single centre, open-label, adaptive, safety and efficacy randomised clinical trial, we included adults aged 18-75 years presenting to the University of Minnesota Medical Center (MN, USA) with OHCA and refractory ventricular fibrillation, no return of spontaneous circulation after three shocks, automated cardiopulmonary resuscitation with a Lund University Cardiac Arrest System, and estimated transfer time shorter than 30 min. Patients were randomly assigned to early ECMO-facilitated resuscitation or standard ACLS treatment on hospital arrival by use of a secure schedule generated with permuted blocks of randomly varying block sizes. Allocation concealment was achieved by use of a randomisation schedule that required scratching off an opaque layer to reveal assignment. The primary outcome was survival to hospital discharge. Secondary outcomes were safety, survival, and functional assessment at hospital discharge and at 3 months and 6 months after discharge. All analyses were done on an intention-to-treat basis. The study qualified for exception from informed consent (21 Code of Federal Regulations 50.24). The ARREST trial is registered with ClinicalTrials.gov, NCT03880565. Between Aug 8, 2019, and June 14, 2020, 36 patients were assessed for inclusion. After exclusion of six patients, 30 were randomly assigned to standard ACLS treatment (n=15) or to early ECMO-facilitated resuscitation (n=15). One patient in the ECMO-facilitated resuscitation group withdrew from the study before discharge. The mean age was 59 years (range 36-73), and 25 (83%) of 30 patients were men. Survival to hospital discharge was observed in one (7%) of 15 patients (95% credible interval 1·6-30·2) in the standard ACLS treatment group versus six (43%) of 14 patients (21·3-67·7) in the early ECMO-facilitated resuscitation group (risk difference 36·2%, 3·7-59·2; posterior probability of ECMO superiority 0·9861). The study was terminated at the first preplanned interim analysis by the National Heart, Lung, and Blood Institute after unanimous recommendation from the Data Safety Monitoring Board after enrolling 30 patients because the posterior probability of ECMO superiority exceeded the prespecified monitoring boundary. Cumulative 6-month survival was significantly better in the early ECMO group than in the standard ACLS group. No unanticipated serious adverse events were observed. Early ECMO-facilitated resuscitation for patients with OHCA and refractory ventricular fibrillation significantly improved survival to hospital discharge compared with standard ACLS treatment. National Heart, Lung, and Blood Institute.

Comparison of continuous versus intermittent infusion cyclosporine and impact on transplant related outcomes in allogeneic transplant recipients.

Continuous infusion (CIVI) cyclosporine (CsA) is an alternative for allograft recipients intolerant of twice daily infusions (TDI). The importance of achieving therapeutic levels of CsA early after allogeneic HCT has been demonstrated in previous studies. Our study evaluated the incidence of acute graft versus host disease (GVHD) and survival among patients receiving CIVI vs. TDI CsA during their first allogeneic HCT. A retrospective study of adult patients undergoing first allogeneic HCT at the University of Minnesota Medical Center between 2011 and 2017. Patients were grouped according to the administration method. The primary outcome was the occurrence of acute grade II-IV GVHD by day +180. Secondary outcomes included the 1-year incidence of chronic GVHD, relapse, and overall survival. 42 patients intolerant of TDI CsA received CsA via CIVI for >48 hours for a median of 9 days (range, 3-32 days). CsA concentrations were similar between groups. We found no difference between the rates of grade II-IV acute (45% vs 53%, p = 0.59) or chronic (17% vs 30%, p = 0.20) GVHD or overall survival (57% vs 67%, p = 0.10). Subgroup analysis of patients that received myeloablative conditioning or umbilical cord blood did not reveal significant differences in GVHD or overall survival. Cumulative incidence of relapse was higher among the continuous infusion group (39% vs. 23%, p < 0.01). Due to the finding of increased risk of relapse, cyclosporine should be administered as traditional twice daily infusion unless necessary. A prospective clinical trial is needed to confirm these results.

S0072 Necrotizing Pancreatitis: Long-Term Follow-Up and Outpatient Intervention

INTRODUCTION: Acute necrotizing pancreatitis (ANP) is a severe form of pancreatitis with significant morbidity and mortality. More patients are surviving index hospitalization with minimally invasive treatment approaches. The long-term outcomes in patients with ANP have not been well described. We investigated the success with outpatient interventions including cholecystectomy and chemical dependency treatment in relation to long-term outcomes of repeated hospitalization, alcohol use and mortality. METHODS: Patients who were admitted to the University of Minnesota Medical Center from 2010 to 2019 and included in a prospectively collected database were screened to identify those with biliary or alcohol related ANP. Data on index hospitalization and patient characteristics was extracted using retrospective chart review. Long-term follow up and outpatient interventions were assessed with written and/or telephone surveys. RESULTS: A total of 211 patients with alcohol (103/211, 48.8%) or biliary (108/211, 51.2%) ANP were identified after initial screening. Average age was 54.3 ± 15.2 years, 71% were male (150/211) and 14.2% deceased (30/211). There were a total of 67 responses to written or telephone surveys and the average length of follow up was 56 months. Cholecystectomy was reported by 95.6% (43/45) with biliary ANP, the remaining 4.4% (2/45) had met with a surgeon and were not surgical candidates. Only 11.6% (5/44) reported hospitalization in the prior year related to pancreatitis or related complications and 20.9% (9/43) reported frequent pain. Only 72.7% (16/22) with alcohol related ANP recalled having any provider discuss alcohol abstinence, 31.8% (7/22) reported completing any type of chemical dependency treatment and 40.9% (9/22) continued to drink alcohol. Although 9.1% (2/22) reported hospitalization in the past year, 31.8% (7/22) had ongoing pain related to their pancreas and only 50% (11/22) have followed up with a GI provider at the University of Minnesota. CONCLUSION: Patients with biliary ANP are successful in following up for cholecystectomy. More attention needs to be drawn to those with alcohol related ANP acknowledging that up to 40% continue to drink with only a small proportion who have followed through with chemical dependency treatment. Particular emphasis should be placed prior to hospital discharge to educate patients on alcohol abstinence, encourage participation in chemical dependency treatment and schedule close follow up with a GI provider.Table 1.: Baseline characteristics of patients who alcohol or biliary necrotizing pancreatitis.Table 2.: Survey responses from patients with biliary pancreatitis.Table 3.: Patients who responded to survey have alcohol related pancreatitis.

Left Ventricular Assist Device Outcomes in Non English Speaking Patients

Introduction Non-English speaking (NES) status has been identified as a risk factor for adverse outcomes in patients with cardiovascular disease. Given the complexity of left ventricular assist device (LVAD) care and the need for frequent provider/patient interactions, we aimed to review outcomes of NES LVAD recipients in a single center LVAD cohort. Methods The analysis included all adult patients who underwent first time, continuous flow LVAD implementation at the University of Minnesota Medical Center from June 2005 to July 2019 (n=413). Patients were identified as NES if they identified English as there second language regardless of interpreter use. To assess the impact of NES on post LVAD outcomes, 1-year readmissions and 2-year survival was compared between NES and English speaking LVAD recipients. Results The study cohort included 13 NES (3%) and 400 (97%) English speaking patients. Six of the NES patients (46%) needed an interpreter during clinical interactions. Detailed baseline characteristics and post LVAD complications of the NES cohort are presented (table). Between English Speaking vs. NES LVAD recipients, no statistically significant differences were detected between age (Mean [±SD]) (58 [±14] vs 56 [±15], P=0.65), male sex (80% vs 92%, P=0.28) or bridge to transplant (BTT) status (59% vs 54%, P=0.69). Observed complications among the NES patients over a median (IQR) follow up time of 348 (202, 647) days were the following; driveline infection: 6 (46%), GI bleed: 4 (30%), stroke: 2 (15%), pump thrombosis: 1(8%), and device replacement for driveline malfunction: 1 (8%). No statistically significant differences were detected between 1-year readmissions between English and NES LVAD patients, (2 [IQR 1, 3] vs 1 [IQR 0.5, 4], P=0.86) or 2-year survival (77% vs. 84%, P=0.89). Conclusion Non English Speaking LVAD recipients had a similar survival compared to the larger LVAD cohort in this single center analysis, however this sample size was small. High rates of driveline infections were observed among NES patients. Larger data will be needed to confirm these observations with a higher degree of confidence.

The Benefits of Massage Therapy with BMT Patients: Impact on Anxiety, Depression and Pain Symptoms

Abstract Summary This poster presentation will detail the implementation and data collection and analysis for massage therapy services provided to BMT patients during their hospital stay at the University of Minnesota Medical Center. The BMT Social Workers sought funding through the hospital foundation to hire an experienced massage therapist to provide massage therapy services for 9 months. A pre- and post-survey was provided to each patient at each massage session. The outcome data shows that patients experienced a decrease in pain, anxiety and depression symptoms. Patients also reported overwhelmingly high satisfaction scores for the massage therapy services they received. This presentation will include recommendations for implementing a massage therapy program in other oncology settings. Abstract Content Research shows that massage therapy can reduce symptoms of pain, nausea, fatigue, anxiety and depression in cancer patients. In light of this compelling research, in 2018 the Bone & Marrow Transplant Social Workers at the University of Minnesota Medical Center sought and were rewarded funding from the Fairview Foundation to incorporate massage therapy into the integrative services offered to patients in the BMT Program. In September 2018, a skilled massage therapist, Chandler Yorkhall, BCTMB, was hired and began providing massage therapy and energy work services two days per week to BMT patients on the inpatient unit. The BMT social workers designed a pre- and post-survey to be distributed to patients each time they receive a massage. The surveys consist of evidence-based scales used to measure pain, anxiety and depression symptoms, as well as a Likert scale measuring patient satisfaction and a section for comments. The pre-survey was provided to the patient during the massage therapist's initial assessment before each session. The post-survey was then completed by the patient and placed in a provided envelope in the patient's room and collected by the nurse. Over 400 massage therapy sessions were provided to BMT patients on Unit 5C from September 2018 through May 2019. Of those sessions, 266 pre and post-assessments were completed by 127 BMT patients. 95% of the time, patients were “very satisfied” or “satisfied” with the massage therapy services they received. Data showed an improvement of depression symptoms by 32.7%, pain reduction by 31% and anxiety reduction 25.5%. The feedback and outcomes for the massage therapy program were overwhelmingly positive. It has proven to be a greatly needed and appreciated service that can have positive impacts for patient experience as well as reduction in patients’ symptoms of pain, anxiety and depression. This presentation will detail the implementation, data collection and analysis for massage therapy services provided. The presentation will also provide recommendations for providing massage therapy services in oncology settings.

High Risk of Relapse with Intermediate Dose Cytarabine for Consolidation in Young Favorable Risk AML Patients Following Induction with 7+3

Introduction CALGB study in 1994 reported improved leukemia free survival (LFS) with sequential courses of high dose cytarabine (HIDAC- 3 gm/m2, q12h on days 1, 3 and 5) compared to lower doses of 100 mg/m2 and 400 mg/m2 per day. Following this study, HIDAC has been commonly used for chemotherapy consolidation in AML patients 60 years or younger. More recently, several groups investigated the role of intermediate dose cytarabine (IDAC-1.5 gm/m2 q12h D 1, 3 and 5 or D 1-3) in this population and reported equivalent outcomes for IDAC compared to HIDAC (Table 1). ELN guidelines published in January 2017 recommend IDAC for consolidation therapy in young (&amp;lt;60 years) favorable risk AML patients. Following these studies and ELN publication, consolidation protocols for AML patients 60 years or younger were changed from HIDAC to IDAC at Singapore General Hospital (SGH) in 2011 and at University of Minnesota Medical Center (UMMC) in February 2017. NCCN continues to recommend HIDAC as category 1 recommendation for consolidation of patients in this setting. We performed a combined analysis of outcomes in young favorable risk AML patients to assess impact of this change in practice on outcomes at our institutions as a quality initiative. Methods Patients 60 years or younger with ELN favorable risk AML between 2004 and 2015 at SGH and between September 2015 and March 2018 at UMMC, who underwent induction therapy with 7+3 regimen (idarubicin 12 mg/m2 D1-3 or daunorubicin 60 - 90 mg/m2 and cytarabine 100 mg/m2 continuous infusion from D1-7) and consolidation with cytarabine monotherapy were identified. These dates were chosen to allow reasonably equal time distribution before and after change in our practice. We extracted relevant outcomes data using chart review. We used Chi-square testing and applied Kaplan-Meier survival analysis to detect differences in relapse and survival outcomes. Results Of 67 ELN favorable risk patients 60 years or younger, 42 received HIDAC and 25 received IDAC. Median age was 39 years (range 16-59). 64 patients received 1 induction cycle and 3 patients received 2 induction cycles to achieve complete remission. Median LFS and overall survival (OS) with HIDAC vs IDAC were not-reached (NR, median time for follow up = 6.9 years) vs 1.35 years (p=0.004) and NR vs 2.27 years (p=0.001) respectively (Figure 1). Cumulative incidence of relapse at 2 years was 23.8% for HIDAC and 60% for IDAC groups (p=0.003). 3-year LFS and OS for HIDAC vs IDAC groups were 71% vs 36% (p=0.06) and 90% vs 48% (p=0.002). Discussion Historical data suggests chemotherapy consolidation with 3-4 cycles of HIDAC achieves long-term LFS of 60 - 70 % in young favorable risk AML patients. While the outcomes in HIDAC group in this study are comparable to historical data with good outcomes, a significantly larger proportion of patients in the IDAC group had early relapse (60%) and death. A major difference between the patients in this study and the other studies which showed similar outcomes with HIDAC and IDAC is the mode of induction. Data from large randomized studies in AML (Table 1) that suggest similar efficacy of IDAC compared to HIDAC used higher intensity induction regimens (frequently double induction therapy) compared to 7+3 and/or used other agents in combination with Ara-C for consolidation. These data are not broadly applicable in patients who receive a single cycle of 7+3 for induction, when selecting dose of Ara-C as monotherapy for consolidation. Furthermore, studies that used 7+3 for induction therapy showed benefit for HIDAC compared with multi-agent chemotherapy that included IDAC (Table 1). Conclusion Using single agent IDAC for post remission therapy is associated with higher rates of relapse and poor overall survival compared to HIDAC among young, ELN favorable risk AML patients who achieve complete remission following 7+3. The large randomized studies that suggested equivalency of IDAC to HIDAC for post remission therapy in AML patients, either used more than one cycle of intensive induction regimen and/or used cytarabine in combination with an anthracycline for consolidation. Hence, ELN guidelines should be cautiously interpreted given the above limitations in generalizability. Our study suggests that HIDAC, rather than IDAC, is the preferred chemotherapy consolidation regimen in young, favorable risk AML patients following standard 7+3 induction. Foot note: BK &amp; NAAH contributed equally. FH &amp; ZL contributed equally. Disclosures Kolla: Amgen: Equity Ownership. Weisdorf:Pharmacyclics: Consultancy; Incyte: Research Funding; Fate Therapeutics: Consultancy.

2416. Effect of Sequential Universal Bleach Cleaning and Best Practice Alerts on Clostridioides difficile Infection Rates

BackgroundA gap analysis prompted consideration of expanded bleach disinfection beyond rooms housing patients in isolation for Clostridioides difficile infection (CDI) and emphasis on CD testing stewardship at the University of Minnesota Medical Center (UMMC), a tertiary care center spanning two campuses in close proximity with adult patients on the East Bank (EB), adult and pediatric patients on West Bank (WB).MethodsAn electronic best practice advisory (BPA) went live in April 2018 on both the EB and WB (Figure 1). The BPA first discourages CD testing in the event of a prior positive within 10 days or a prior negative within 7 days. Second, the BPA discourages CD testing in patients with fewer than 3 loose stools in a 24 hour period, who have received laxatives in the last 48 hours, or who lack CDI symptoms (fever > 38C, abdominal pain, or leukocytosis > 11,000). Providers can bypass the BPA based on clinical judgment; those who override the BPA are provided just-in-time education via email.Following a successful pilot in three wards, the EB Environmental Services (ES) team expanded the use of bleach to include all terminal cleaning regardless of isolation status in June 2018 (Figure 1). Daily cleaning on the EB was excluded from universal bleach utilization, as were daily and terminal cleaning on the WB.CD testing throughout the study period occurred via polymerase chain reaction (PCR) of the toxin B gene. ES performance, assessed by adenosine triphosphate (ATP) bioluminescence testing, and hand hygiene rates were unchanged throughout the study period.ResultsAdult-only hospital-onset (HO)-CDI rates decreased during the study period on both hospital campuses, with the EB exhibiting a greater decrease, (Fig 1), while community-onset (CO) and community-onset healthcare facility associated (CO-HCFA) rates remained steady during the study period (Fig 3). Whole-house (adult and pediatric) CD testing was largely unchanged while the proportion of tests triggering the BPA decreased (Fig 2).ConclusionUniversal bleach utilization during terminal cleaning combined with an electronic BPA were associated with decreased adult HO-CDI rates. However, the BPA did not impact CD testing rates, suggesting that decreased HO-CDI rates may be unattributable to testing stewardship. Disclosures All authors: No reported disclosures.

1064. Antimicrobial Stewardship Program at a Large Academic Medical Center, Impact Over 12 Years

BackgroundThe University of Minnesota Medical Center (UMMC) is a tertiary care facility, which has had a comprehensive antimicrobial stewardship program (ASP) for 12 years.MethodsThe antimicrobial stewardship team is comprised of a full-time PharmD and ID staff physicians. Recommendations are placed in the electronic medical record as a progress note. Verbal recommendations may also be made.ResultsThere was a downward trend in Hospital-acquired (HA) C. difficile diarrhea from 2007 to 2014 from 1.2 to 0.5/1000 patient-days (pt day). Rates appear stable from 2014 to 2019 with adjustment for change to NHSN lab-based CDI surveillance (Figure 1). From 2009 to 2019 a decrease was seen in VRE hospital-acquired infections (HAI) from 0.53 to 0.21/1,000 patient-days and in MRSA HAIs from 0.2 to 0.14/1,000 patient-days. Newly acquired ESBL HAIs have remained relatively stable from 2009 to 2019 at 0.09 to 0.05/1,000 patient-days. CRE HAIs are low but stable rates at 0.02/1,000 patient-days (Figure 2). We track antimicrobial utilization for internal and national reporting (starting in July 2017). A SAAR for all Antibacterial agents (ICUs, wards, and oncology units) of 1.33 in 2018. Our top four agents average DOT; piperacillin/tazobactam (66.81), cefepime (34.40), oral levofloxacin (23.56) and intravenous meropenem (21.49). We demonstrate lower average DOT for our restricted antimicrobials (206.21) as compared with our nonrestricted antimicrobials (236.74) (Figure 3). Cost savings continued from year to year. After adjusting for inflation annually, our expected costs ($84.08) compared with actual costs ($40.12 ytd 2019), demonstrates effective cost management of antimicrobial agents. (Figure 4)ConclusionWe observed a decrease in HAIs VRE and C. difficile infections after 3 years of operation, and MRSA after 5 years. This downward trend has continued. ESBL HAIs remain relatively stable and CRE are stable at low rates but remain emerging HAIs of concern. We are now focusing efforts on limiting unneeded fluoroquinolone and carbapenem use. We continue to analyze our SAAR data and internal DOT data to identify areas of opportunity to improve antimicrobial use. The ASP outcomes have continued to cost justify ongoing efforts. The effects of the program and the Infection Prevention Department appear to be synergistic. Disclosures All authors: No reported disclosures.

The Utilization of JAK2 Molecular Testing at the University of Minnesota Medical Center

Abstract Objectives JAK2 gene mutation status is included in the 2017 WHO diagnostic criteria of a myeloproliferative neoplasm (MPN) and therefore indicated in the diagnostic workup. Overutilization and inappropriate testing, however, can cause harm and lead to medical errors, and most importantly, it can significantly drive up the costs of patients’ care. Here we report an attempt to evaluate the success rate and the associated costs of JAK2 testing as well as to correlate it with clinical indications and patient presentation. Methods We conducted the analysis of next-generation testing orders/results performed at out institution between 9/2/2015 and 7/13/2018. We reviewed the rate of positive results as well as the type of order submitted to laboratory and compared the costs of individual tests. Results We analyzed 3,474 NGS orders/results performed at our institution. Of those, 1,203 were submitted for JAK2 testing, making this the most commonly ordered NGS test. Out of 1,203 results, there were 820 unique patient specimens and 118 tested positive (15%). The JAK2 testing was primarily ordered as part of the NGS MPN panel, which includes JAK2, CALR, and MPL mutation hotspots (70%), the custom panel (23%), JAK2 single gene (5%), or as part of comprehensive myeloid panel (2%). The mutation detection rate was 15%, 12%, 11%, and 15%, respectively. Conclusion Testing for JAK2 gene mutation status is the most commonly ordered NGS test in our laboratory. The detection rate for pathogenic mutation, irrespective of the type of genetic panel ordered, was 11% to 15%. Our analysis of the JAK2 ordering and reporting suggests that the JAK2 testing may be somewhat overutilized at our institution. An ongoing analysis of clinical indications, patient presentation, and associated costs will help to fully elucidate the true utility of this testing and will be included in the poster presentation.

Medically integrated pharmacies make patient care more efficient

Medically integrated pharmacies make patient care more efficient

Providing Comprehensive Care to Patients with Epidermolysis Bullosa in the Perioperative Environment

The University of Minnesota Medical Center and Masonic Children’s Hospital EB Center offers comprehensive multidisciplinary care to children with Epidermolysis Bullosa (EB). This is the only center in the United States to offer Bone Marrow Transplantation (BMT) as a treatment for EB. To determine candidacy, these patients come to the Operating Room to have their dressings fully removed, labs drawn, photographs taken and skin biopsies collected under general anesthesia. The child is transported to PACU and remains under sedation while dressings are reapplied, often a 2-3 hour process. Since these patients are at high risk for severe blistering skin injuries, it was evident that modifications to equipment, practice changes, and staff education needed to occur to keep them safe during critical times in Perioperative Services when their usual expert care provider, often a parent, is not present.

Towards more Accessible Precision Medicine: Building a more Transferable Machine Learning Model to Support Prognostic Decisions for Micro- and Macrovascular Complications of Type 2 Diabetes Mellitus.

Although machine learning models are increasingly being developed for clinical decision support for patients with type 2 diabetes, the adoption of these models into clinical practice remains limited. Currently, machine learning (ML) models are being constructed on local healthcare systems and are validated internally with no expectation that they would validate externally and thus, are rarely transferrable to a different healthcare system. In this work, we aim to demonstrate that (1) even a complex ML model built on a national cohort can be transferred to two local healthcare systems, (2) while a model constructed on a local healthcare system's cohort is difficult to transfer; (3) we examine the impact of training cohort size on the transferability; and (4) we discuss criteria for external validity. We built a model using our previously published Multi-Task Learning-based methodology on a national cohort extracted from OptumLabs® Data Warehouse and transferred the model to two local healthcare systems (i.e., University of Minnesota Medical Center and Mayo Clinic) for external evaluation. The model remained valid when applied to the local patient populations and performed as well as locally constructed models (concordance: .73-.92), demonstrating transferability. The performance of the locally constructed models reduced substantially when applied to each other's healthcare system (concordance: .62-.90). We believe that our modeling approach, in which a model is learned from a national cohort and is externally validated, produces a transferable model, allowing patients at smaller healthcare systems to benefit from precision medicine.

456 Bacterial wound isolates and antibiotic susceptibility in Epidermolysis Bullosa

Epidermolysis bullosa (EB) causes mucocutaneous blisters predisposing the host to polymicrobial colonization with potentially resistant organisms. Limited data are available on the microbiology and antibiotic susceptibility in EB patients. Understanding the skin flora in EB may direct clinical care. We reviewed wound cultures from patients with EB to identify their isolates and susceptibilities patterns. Patients, aged 0-22 years were recruited through the EB clinic at the University of Minnesota between 2007 and 2017. We compared susceptibility patterns with the current antibiogram from the University of Minnesota Medical Center (2017). Three hundred and forty-two clinical samples from 68 patients were reviewed. The most common subtype of EB was recessive dystrophic EB (60%). Staphylococcus aureus (SA) was most common (26%), third of which were methicillin-resistant (MRSA), and none were resistant to vancomycin or linezolid. Methicillin-sensitive SA (MSSA) was 72% susceptibility to clindamycin, and 95% to trimethoprim/sulfamethoxazole (TMP/SMX), similar to the published antibiogram. In contrast, the susceptibility of MSSA to erythromycin, ciprofloxacin, and levofloxacin was decreased in the EB cohort, compared to our institution’s antibiogram. Decrease susceptibility was also noticed in EB patients for MRSA to clindamycin and TMP/SMX. Pseudomonas aeruginosa (PA) was the second most common organism (14%), followed by coagulase-negative staphylococci (CONS) 12%. PA isolates were 81% susceptible to ceftazidime and 82% to meropenem, similar to the published antibiogram. PA susceptibility to ciprofloxacin was decreased compared to general antibiogram. This study shows over-all decreased susceptibilities pattern in bacterial isolates from EB patients compared to the general patient population in our institution and may guide clinical care especially when choosing empiric antibiotic for a suspected invasive bacterial infection in EB patients.

Electronically Triggered Sepsis Alert in Non-Intensive Care Inpatients Using Modified Systemic Inflammatory Response Syndrome Criteria: A Retrospective Observational Study with In-Depth Analysis of Surgical Patients.

Background: Our center initiated an electronic Sepsis Best Practice Alert (sBPA) protocol to aid in early sepsis detection and treatment. However, surgery alters peri-operative physiology, which may trigger an sBPA for noninfectious causes. This study aimed to provide early evaluation of automated sBPA utility in surgical patients. Methods: This study was a retrospective review of the outcomes of patients admitted to the University of Minnesota Medical Center (but not to the intensive care unit) from August 2015-March 2016 and compared how the sBPA performed in those having and not having surgery. An sBPA prompted nursing to draw blood for an immediate lactate assay if two modified systemic inflammatory response syndrome (mSIRS) criteria or three mSIRS criteria within 24 hours after surgery were met. Physicians were notified if the lactate concentration was >2 mmol/L. Further review was performed of data collected prospectively on the surgical patients. Results: A total of 10,335 patients were admitted (2,158 surgery and 8,177 non-surgery). Of these, 33% of the surgery patients and 35% of the patients not having surgery triggered sBPAs. In surgery patients, 13% of lactate concentrations were >2 mmol/L versus 25% in patients not having surgery. An sBPA was triggered more frequently after procedures with a wound class of 4 (5% vs. 2%), emergency operation (23% vs. 10%), and longer operations (280 min vs. 222 min (p < 0.05 for all). Surgery patients triggering sBPAs had longer hospital stays (9.6 vs. 4.4 days; p < 0.05), more surgical site infections (7% vs. 2%; p < 0.05), and a similar mortality rate (3% vs. 4%; p = 0.15) than those who did not trigger an sBPA. Conclusion: An sBPA fired in a third of all inpatients, and an sBPA that prompted lactate measurements was less likely to be abnormal in surgery patients than in those not having surgery. There was no difference in the mortality rate in surgical patients who fired and those who did not; however, the sBPA did identify patients with a more complicated post-operative course. Further refinements of the electronic trigger should increase BPA specificity.

Cyclosporine Administration and Graft Versus Host Disease: Continuous Intravenous Infusion vs. Intravenous Bolus Dosing

Introduction Continuous infusion (CIVI) cyclosporine (CSA) is an alternative for allograft recipients intolerant of twice daily infusions (TDI). Trough concentrations are an accepted surrogate for monitoring therapy, however total exposure may be dependent on peak concentrations not obtained when given via CIVI. Therefore, maintaining the same trough goals for both routes of administration may lead to decreased CSA exposure and subsequent increased rates of graft versus host disease (GVHD). The purpose of this study was to evaluate the incidence of acute GVHD and overall survival among patients receiving CIVI vs. TDI CSA. Methods We performed a retrospective study of adult patients undergoing first allogeneic HCT at the University of Minnesota Medical Center between 2011 and 2017 that received > 48 hours of CSA. All patients initially received CSA via TDI per institutional standard and targeted whole blood trough concentrations of 200-400 ng/mL via HPLC. Any change to CIVI was made at the discretion of the treating physician. All patients were changed to oral CSA when able. Patients were grouped according to administration method; those that received cyclosporine via CIVI for > 48 hours at any time during initial transplant admission were classified as CIVI and those that received CSA exclusively as TDI were classified as TDI. Results 308 patients met inclusion criteria (CIVI 42, TDI 266). Donor type, GVHD prophylaxis, underlying diagnosis, and conditioning intensity were similar between groups. The majority of subjects received HLA-matched sibling or cord blood allografts. There were an equal proportion of myeloablative and reduced intensity conditioning regimens, and nearly all GVHD prophylaxis therapy consisted of cyclosporine combined with either mycophenolate or methotrexate. Forty two patients received CSA via CIVI for > 48 hours for a median of 9 days (range, 3-32 days). The average time of change from TDI to CIVI was 3 days after transplant. The TDI group received only TDI for a median 17 days (range 4-62 days) prior to conversion to oral CSA. CSA concentrations at day -1 and weeks 1, 2, and 3 were similar between the groups (Figure 1). We found no difference between the rates of grade II-IV acute (45% vs 53%, p = 0.59) or chronic (17% vs 30%, p = 0.20) GVHD or overall survival (57% vs 67%, p = 0.10) between groups. Subgroup analysis of patients that received myeloablative conditioning did not reveal significant differences in GVHD or overall survival. Conclusion In contrast to previous studies, we did not observe an association of CIVI CSA administration with acute GVHD. Among patients intolerant of bolus CSA infusions, a brief transition to CIVI CSA appears to be a safe and acceptable route of administration without increased risk of GVHD or adverse effects on overall survival.