HistoEM: A Pathologist-Guided and Explainable Workflow Using Histogram Embedding for Gland Classification.

Pathologists have, over several decades, developed criteria for diagnosing and grading prostate cancer. However, this knowledge has not, so far, been included in the design of convolutional neural networks (CNN) for prostate cancer detection and grading. Further, it is not known whether the features learned by machine-learning algorithms coincide with diagnostic features used by pathologists. We propose a framework that enforces algorithms to learn the cellular and subcellular differences between benign and cancerous prostate glands in digital slides from hematoxylin and eosin-stained tissue sections. After accurate gland segmentation and exclusion of the stroma, the central component of the pipeline, named HistoEM, utilizes a histogram embedding of features from the latent space of the CNN encoder. Each gland is represented by 128 feature-wise histograms that provide the input into a second network for benign vs cancer classification of the whole gland. Cancer glands are further processed by a U-Net structured network to separate low-grade from high-grade cancer. Our model demonstrates similar performance compared with other state-of-the-art prostate cancer grading models with gland-level resolution. To understand the features learned by HistoEM, we first rank features based on the distance between benign and cancer histograms and visualize the tissue origins of the 2 most important features. A heatmap of pixel activation by each feature is generated using Grad-CAM and overlaid on nuclear segmentation outlines. We conclude that HistoEM, similar to pathologists, uses nuclear features for the detection of prostate cancer. Altogether, this novel approach can be broadly deployed to visualize computer-learned features in histopathology images.

A randomized, double-blinded, placebo-controlled clinical trial of sterile filtered human amniotic fluid for treatment of COVID-19.

IMPORTANCE: Acellular human amniotic fluid (hAF) is an antimicrobial and anti-inflammatory fluid that has been used to treat various pro-inflammatory conditions. In a feasibility study, we have previously demonstrated that hAF could be safely administered to severely ill patients with coronavirus disease-19 (COVID-19). The impact of acellular hAF on markers of systemic inflammation and clinical outcomes during COVID-19 infection remain unknown. OBJECTIVE: To determine the safety and efficacy of acellular, sterile processed intravenously administered hAF on markers of systemic inflammation during COVID-19. DESIGN, SETTINGS AND PARTICIPANTS: This single-center Phase I/II randomized, placebo controlled clinical trial enrolled adult (age ≥ 18 years) patients hospitalized for respiratory symptoms of COVID-19, including hypoxemia, tachypnea or dyspnea. The study was powered for outcomes with an anticipated enrollment of 60 patients. From 09/28/2020 to 02/04/2022 we enrolled and randomized 47 (of an anticipated 60) patients hospitalized due to COVID-19. One patient withdrew consent after randomization but prior to treatment. Safety outcomes to 30 days were collected through hospital discharge and were complete by the end of screening on 6/30/2022. INTERVENTIONS: Intravenous administration of 10 cc sterile processed acellular hAF once daily for up to 5 days vs placebo. MAIN OUTCOME AND MEASURES: Blood biomarkers of inflammation, including C-Reactive protein (CRP), lactate dehydrogenase, D-dimer, and interleukin-6 (IL-6), as well as safety outcomes. RESULTS: Patients who were randomized to hAF (n = 23) were no more likely to have improvements in CRP from baseline to Day 6 than patients who were randomized to placebo (n = 24) hAF: -5.9 [IQR -8.2, -0.6] vs placebo: -5.9 [-9.4, -2.05]; p = 0.6077). There were no significant differences in safety outcomes or adverse events. Secondary measures of inflammation including lactate dehydrogenase, D-dimer and IL-6 were not statistically different from baseline to day 6. CONCLUSIONS AND RELEVANCE: In this randomized clinical trial involving hospitalized patients with COVID-19, the intravenous administration of 10 cc of hAF daily for 5 days did not result in statistically significant differences in either safety or markers of systemic inflammation compared to placebo, though we did not achieve our enrollment target of 60 patients. TRIAL REGISTRATION: This trial was registered at ClinicalTrials.gov as #NCT04497389 on 04/08/2020.

Measuring the unmet needs of American military Veterans and their caregivers: Survey protocol of the HERO CARE survey.

BACKGROUND: Empowering Veterans to age in place is a Department of Veterans Affairs priority. Family or unpaid caregivers play an important role in supporting Veterans to achieve this goal. Effectively meeting the needs of Veterans and caregivers requires identifying unmet needs and relevant gaps in resources to address those needs. METHODS: Using a modified Socio-Ecological Model, we developed a prospective longitudinal panel design survey. We randomly selected 20,000 community-dwelling Veterans enrolled in the Veterans Health Administration (VHA), across five VHA sites. We oversampled Veterans with a higher predicted 2-year long-term institutional care (LTIC) risk. Veterans were mailed a packet containing a Veteran survey and a caregiver survey, to be answered by their caregiver if they had one. The Veteran survey assessed the following health-related domains: physical, mental, social determinants of health, and caregiver assistance. Caregivers completed questions regarding their demographic factors, caregiving activities, impact of caregiving, use of VA and non-VA services, and caregiver support resources. Follow-up surveys will be repeated twice at 12-month intervals for the same respondents. This article describes the HERO CARE survey protocol, content, and response rates. RESULTS: We received responses from 8,056 Veterans and 3,579 caregivers between July 2021 and January 2022, with 95.6% being received via mail. Veteran respondents were mostly males (96.5%), over 65 years of age (94.9%), married (55.0%), Non-Hispanic White (75.2%), and residing in urban areas (80.7%). CONCLUSIONS: This longitudinal survey is unique in its comprehensive assessment of domains relevant to older Veterans stratified by LTIC risk and their caregivers, focusing on social determinants, caregiver support, and the use of caregiver support resources. Survey data will be linked to Centers for Medicare & Medicaid Services and VA data. The results of this study will inform better planning of non-institutional care services and policy for Veterans and their caregivers.

Mifepristone for miscarriage treatment in Utah: A survey of clinician knowledge and assessment of an educational video on future use.

Objectives: We aimed to: (a) describe current use of mifepristone for early pregnancy loss (EPL) management in Utah, (b) identify predictors of knowledge pre- and posteducational video, and (c) explore postvideo impacts on the likelihood to use mifepristone. Mifepristone is subject to the Food and Drug Administration's (FDA) Risk Evaluation and Mitigation Strategy (REMS) requirements. Methods: Between September 2020 and March 2021 we surveyed Utah clinicians from six specialties caring for people experiencing EPL, recruited through professional organizations and hospital listservs. Participants viewed a 3.5-minute educational video on mifepristone for EPL and completed pre- and postvideo questionnaires. We evaluated predictors of high prevideo and improved postvideo knowledge using random forest regression conditional importance measures and partial dependency plots. We described current mifepristone use and video effects on likelihood to use mifepristone. Results: Of 506 participants, most specialize in emergency medicine (172, 34%) and practice in private settings (253, 51%). Two-thirds had heard of mifepristone (328/471, 70%). Of 176/471 (37%) attempting provision of mifepristone, actual provision occurred for 59% (104/176). Baseline knowledge scores were low (mean 4.81/13 [37%] correct). Predictors of high prevideo knowledge include provision or attempted provision of mifepristone, having heard of mifepristone, providing EPL management expectantly or via medication, and specialty type. Mean postvideo knowledge scores improved by 3.27 points (68% improvement, paired t-test; 95% confidence interval 2.82-3.72, p < 0.0001). Postvideo, 66% (242/364) stated they are much more or somewhat more likely to use mifepristone, with compliance with FDA requirements cited as a barrier to utilization. Conclusions: Among Utah providers, baseline mifepristone knowledge and use for EPL management are low. An educational video improved knowledge and likelihood of use, but FDA REMS requirements continue to be a barrier to including mifepristone in medication management of EPL.

Quantitative and qualitative impact of One Key Question on primary care providers' contraceptive counseling at routine preventive health visits.

OBJECTIVES: One Key Question (OKQ) is a clinical screening tool to assess pregnancy desire in the next year. We aimed to 1) describe the effect of OKQ implementation on contraceptive counseling rates at preventive health visits and 2) evaluate primary care providers' perception of OKQ implementation on their contraceptive counseling practices. STUDY DESIGN: We performed a quantitative retrospective chart review of preventive health visits at eight federally qualified health centers in Utah between 2014 and 2017. Implementation of OKQ included a brief training and inclusion of OKQ in the electronic medical record. Providers received OKQ training in August 2015 and re-training in March 2017. We assessed OKQ and contraceptive counseling documentation rates using interrupted-time-series analysis. We then conducted semi-structured interviews with providers and queried them about the impact of OKQ. We identified dominant themes using modified grounded theory to create an explanatory framework. RESULTS: Abstracting 6634 charts yielded 9840 visits with 56 unique providers (51% physician assistant, 34% physician, 14% nurse practitioner). Interrupted-time-series analysis showed a documentation increase of OKQ in late 2015 (2.6%) and again in spring 2017 (9%), however rates remained low. Contraceptive counseling rates (39.7%) did not change after OKQ implementation. Charts with evidence of a current contraceptive method were less likely to have a OKQ response documented. Interviewees reported OKQ's algorithm did not alter their contraceptive counseling. CONCLUSIONS: OKQ did not change documented rates of contraceptive counseling and uptake was low in quantitative and qualitative analyses. Our study suggests limited usefulness of OKQ in the primary care setting. IMPLICATIONS: Implementation of the One Key Question tool through training and optional EHR field did not increase documented rates of contraceptive counseling in a large federally qualified health center or affect provider contraceptive counseling. Our study suggests limited usefulness of OKQ as a robust screening tool in this primary care setting.

Determining the impact of the Zika pandemic on primary care providers' contraceptive counseling of non-pregnant patients in the US: a mixed methods study.

BACKGROUND: Global pandemics like Zika (ZIKV) factor into pregnancy planning and avoidance, yet little is known about how primary care providers (PCPs) incorporate public health guidance into contraceptive counseling. Study objectives include: 1) determining the impact of the ZIKV pandemic on contraceptive counseling changes; and 2) assessing PCP knowledge and practice regarding contraception, ZIKV, and CDC ZIKV guidelines. METHODS: Study components included: (1) a retrospective review of electronic health records of non-pregnant, reproductive age women presenting for preventive health visits between 2014 and 2017 assessed using interrupted time series analyses (ITSA) to identify changes in documentation of ZIKV risk assessment and contraceptive counseling; and (2) a sequential, cross-sectional study with quantitative surveys and qualitative, semi-structured interviews of PCPs providing preventive care to non-pregnant patients at eight federally qualified health centers in Utah. We performed descriptive analyses on survey data and analyzed qualitative data for dominant themes using a modified Health Belief Model. RESULTS: We conducted 6634 chart reviews yielding 9840 visits. The ITSA did not reveal changes in ZIKV risk assessment or contraceptive counseling. Twenty-two out of 40 (55%) eligible providers participated in the provider component. Participants averaged 69 and 81% correct on contraceptive and ZIKV knowledge questions, respectively. Sixty-five percent reported counseling consistent with CDC ZIKV guidelines. Qualitative analysis found providers unlikely to prioritize ZIKV risk assessment in contraceptive counseling for non-pregnant patients. CONCLUSIONS: PCPs who care for non-pregnant women are knowledgeable about contraception and ZIKV; however, there was no change in ZIKV risk assessment or contraceptive counseling. This stresses the importance of developing strategies to improve guideline uptake.

Robust Testing in Outpatient Settings to Explore COVID-19 Epidemiology: Disparities in Race/Ethnicity and Age, Salt Lake County, Utah, 2020.

OBJECTIVE: US-based descriptions of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection have focused on patients with severe disease. Our objective was to describe characteristics of a predominantly outpatient population tested for SARS-CoV-2 in an area receiving comprehensive testing. METHODS: We extracted data on demographic characteristics and clinical data for all patients (91% outpatient) tested for SARS-CoV-2 at University of Utah Health clinics in Salt Lake County, Utah, from March 10 through April 24, 2020. We manually extracted data on symptoms and exposures from a subset of patients, and we calculated the adjusted odds of receiving a positive test result by demographic characteristics and clinical risk factors. RESULTS: Of 17 662 people tested, 1006 (5.7%) received a positive test result for SARS-CoV-2. Hispanic/Latinx people were twice as likely as non-Hispanic White people to receive a positive test result (adjusted odds ratio [aOR] = 2.0; 95% CI, 1.3-3.1), although the severity at presentation did not explain this discrepancy. Young people aged 0-19 years had the lowest rates of receiving a positive test result for SARS-CoV-2 (<4 cases per 10 000 population), and adults aged 70-79 and 40-49 had the highest rates of hospitalization per 100 000 population among people who received a positive test result (16 and 11, respectively). CONCLUSIONS: We found disparities by race/ethnicity and age in access to testing and in receiving a positive test result among outpatients tested for SARS-CoV-2. Further research and public health outreach on addressing racial/ethnic and age disparities will be needed to effectively combat the coronavirus disease 2019 pandemic in the United States.

High value care education in general surgery residency programs: A multi-institutional needs assessment.

BACKGROUND: The ACGME mandates that residency programs provide training related to high value care (HVC). The purpose of this study was to explore HVC education in general surgery residency programs. METHODS: An electronic survey was distributed to general surgery residents in geographically diverse programs. RESULTS: The response rate was 29% (181/619). Residents reported various HVC components in their curricula. Less than half felt HVC is very important for their future practice (44%) and only 15% felt confident they could lead a QI initiative in practice. Only 20% of residents reported participating in a root cause analysis and less than one-third of residents (30%) were frequently exposed to cost considerations. CONCLUSION: Few residents feel prepared to lead quality improvement initiatives, have participated in patient safety processes, or are aware of patients' costs of care. This underscores the need for improved scope and quality of HVC education and establishment of formal curricula.

Clinical and laboratory predictors of 30-day mortality in severe acute malnourished children with severe pneumonia.

OBJECTIVE: To determine the predictors of mortality within 30 days of hospital admission in a diarrhoeal disease hospital in Bangladesh. METHODS: Cohort study of hospitalised children aged 0-59 months with severe acute malnutrition (SAM) and severe pneumonia in Dhaka Hospital, icddr,b, Bangladesh from April 2015 to March 2017. Those discharged were followed up, and survival status at 30 days from admission was determined. Children who died were compared with the survivors in terms of clinical and laboratory biomarkers. Multivariable logistic regression analysis was used for calculating adjusted odds ratio for death within 30 days of hospital admission. RESULTS: We enrolled 191 children. Mortality within 30 days of admission was 6% (14/191). After adjusting for potential confounders (hypoxia, CRP and haematocrit) in logistic regression analysis, independent factors associated with death were female sex (aOR = 5.80, 95% CI: 1.34-25.19), LAZ <-4 (aOR = 6.51, 95% CI: 1.49-28.44) and Polymorphonuclear Leucocytes (PMNL) (>6.0 × 109 /L) (aOR = 1.06, 95% CI: 1.01-1.11). Using sex, Z-score for length for age (LAZ), and PMNL percentage, we used random forest and linear regression models to achieve a cross-validated AUC of 0.83 (95% CI: 0.82, 0.84) for prediction of 30-day mortality. CONCLUSIONS: The results of our data suggest that female sex, severe malnutrition (<-4 LAZ) and higher PMNL percentage were prone to be associated with 30-day mortality in children with severe pneumonia. Association of these factors may be used in clinical decision support for prompt identification and appropriate management for prevention of mortality in this population.

OBJECTIF: Déterminer les prédicteurs de mortalité dans les 30 jours suivant l'admission à l'hôpital dans un hôpital pour maladies diarrhéiques au Bangladesh. MÉTHODES: Etude de cohorte d'enfants hospitalisés âgés de 0 à 59 mois atteints de malnutrition aiguë sévère (MAS) et de pneumonie sévère à l'hôpital de Dhaka, icddr,b, au Bangladesh d'avril 2015 à mars 2017. Ceux qui ont été libérés ont été suivis et leur état de survie à 30 jours de l'admission a été déterminé. Les enfants décédés ont été comparés aux survivants en termes de biomarqueurs cliniques et de laboratoire. Une analyse de régression logistique multivariée a été utilisée pour calculer le rapport de cotes ajusté pour le décès dans les 30 jours suivant l'admission à l'hôpital. RÉSULTATS: Nous avons inscrit 191 enfants. La mortalité dans les 30 jours suivant l'admission était de 6% (14/191). Après ajustement pour les facteurs confusionnels potentiels (hypoxie, CRP et hématocrite) dans l'analyse de régression logistique, les facteurs indépendants associés au décès étaient le sexe féminin (aOR = 5,80 ; IC95%: 1,34 à 25,19), LAZ <-4 (aOR = 6,51 ; IC95%: 1,49-28,44) et leucocytes polymorphonucléaires (LPMN) (>6,0 x 109 /L) (aOR = 1,06 ; IC95%: 1,01-1,11). En utilisant le sexe, le score Z de la taille pour l'âge (LAZ) et le pourcentage de LPMN, nous avons utilisé des modèles de régression linéaire et de forest aléatoires pour obtenir une AUC validée croisée de 0,83 (IC95%: 0,82-0,84) pour la prédiction de la mortalité à 30 jours. CONCLUSIONS: Les résultats de nos données suggèrent que le sexe féminin, la malnutrition sévère (LAZ <-4) et un pourcentage plus élevé de LPMN étaient susceptibles d'être associés à la mortalité à 30 jours chez les enfants atteints de pneumonie sévère. L'association de ces facteurs peut être utilisée dans l'aide à la décision clinique pour une identification rapide et une prise en charge appropriée pour la prévention de la mortalité dans cette population.

Modeling reductions in SARS-CoV-2 transmission and hospital burden achieved by prioritizing testing using a clinical prediction rule.

Prompt identification of cases is critical for slowing the spread of COVID-19. However, many areas have faced diagnostic testing shortages, requiring difficult decisions to be made regarding who receives a test, without knowing the implications of those decisions on population-level transmission dynamics. Clinical prediction rules (CPRs) are commonly used tools to guide clinical decisions. We used data from electronic health records to develop a parsimonious 5-variable CPR to identify those who are most likely to test positive, and found that its application to prioritize testing increases the proportion of those testing positive in settings of limited testing capacity. To consider the implications of these gains in daily case detection on the population level, we incorporated testing using the CPR into a compartmentalized disease transmission model. We found that prioritized testing led to a delayed and lowered infection peak (i.e. 'flattens the curve'), with the greatest impact at lower values of the effective reproductive number (such as with concurrent social distancing measures), and when higher proportions of infectious persons seek testing. Additionally, prioritized testing resulted in reductions in overall infections as well as hospital and intensive care unit (ICU) burden. In conclusion, we present a novel approach to evidence-based allocation of limited diagnostic capacity, to achieve public health goals for COVID-19.