Faculty Bibliography

BACKGROUND:School-based gardening and nutrition education interventions report improvements in dietary intake, notably through fruit and vegetables. However, gardening, cooking, and nutrition randomized controlled trials (RCT) are limited in evaluating dietary quality, and none have examined processed food consumption to date. OBJECTIVE:The objective of the study was to examine the effects of Texas Sprouts (TX Sprouts), a gardening, cooking, and nutrition education intervention vs. control on unprocessed and ultra-processed food (UPF) consumption in predominately low-income, Hispanic children. DESIGN/METHODS:TX Sprouts was a school-based cluster RCT that consisted of 16 elementary schools randomly assigned to either the TX Sprouts intervention (n=8 schools) or control (delayed intervention; n=8 schools) over three years (2016-2019). TX Sprouts schools received an outdoor teaching garden and 18 one-hour lessons taught by trained educators throughout the school year. Dietary intake data via two 24-hour dietary recalls were collected on a random subsample (n=468) at baseline and post-intervention. All foods and beverages were categorized using the NOVA food classification system (e.g., unprocessed, processed, ultra-processed). Generalized linear mixed effects modeling tested changes in percent calories and grams of NOVA groups between the intervention and control estimates with schools as random clusters. RESULTS:Of the sample, 63% participated in the free and reduced-price lunch program and 57% were Hispanic, followed by non-Hispanic White (21%) and non-Hispanic Black (12%). The intervention compared to control resulted in an increase in consumption of unprocessed foods (2.3% vs. -1.8% grams; p < 0.01) and a decrease in UPF (-2.4% vs. 1.4% grams; p=0.04). Hispanic children in the intervention group had an increase in unprocessed food consumption and a decrease in UPF consumption compared to non-Hispanic children (-3.4% vs. 1.5% grams; p < 0.05). CONCLUSIONS:Study results suggest school-based gardening, cooking, and nutrition education interventions can improve dietary intake, specifically increasing unprocessed food consumption and decreasing UPF consumption. CLINICAL TRIAL REGISTRATION/BACKGROUND:U.S. National Institutes of Health, U.S. National Library of Medicine, NCT02668744, https://clinicaltrials.gov/ct2/show/NCT02668744.

Physicians make critical time-constrained decisions every day. Clinical predictive models can help physicians and administrators make decisions by forecasting clinical and operational events. Existing structured data-based clinical predictive models have limited use in everyday practice owing to complexity in data processing, as well as model development and deployment^1-3. Here we show that unstructured clinical notes from the electronic health record can enable the training of clinical language models, which can be used as all-purpose clinical predictive engines with low-resistance development and deployment. Our approach leverages recent advances in natural language processing^4,5 to train a large language model for medical language (NYUTron) and subsequently fine-tune it across a wide range of clinical and operational predictive tasks. We evaluated our approach within our health system for five such tasks: 30-day all-cause readmission prediction, in-hospital mortality prediction, comorbidity index prediction, length of stay prediction, and insurance denial prediction. We show that NYUTron has an area under the curve (AUC) of 78.7-94.9%, with an improvement of 5.36-14.7% in the AUC compared with traditional models. We additionally demonstrate the benefits of pretraining with clinical text, the potential for increasing generalizability to different sites through fine-tuning and the full deployment of our system in a prospective, single-arm trial. These results show the potential for using clinical language models in medicine to read alongside physicians and provide guidance at the point of care.

Patients with heart failure with preserved ejection fraction (HFpEF) often experience dyspnea, decreased exercise tolerance, and decreased quality of life (QOL). Exercise training is a promising non-pharmacological treatment, with some improvement in exercise tolerance and QOL in HFpEF patients in randomized controlled trials (RCTs). We conducted a systematic review and meta-analysis of RCTs examining the effect of exercise therapy on exercise tolerance, QOL, and echocardiographic parameters in patients with HFpEF. Article database search of PubMed, Embase, and Cochrane Central Register of Controlled Trials identified 15 publications representing 579 unique patients. Results are presented as weighted mean difference (WMD) with 95% confidence intervals (CI). Exercise training (compared to control) demonstrated a significant improvement in exercise tolerance as measured by peak absolute VO2 (WMD [95% CI] = 164.67 [65.54, 263.79] mL/min), peak relative VO2 (WMD [95% CI] = 1.85 [0.98, 2.73] mL/min/kg), workload (WMD [95% CI] = 12.92 [4.67, 21.17] W), exercise time (WMD [95% CI] = 2.05 [1.57, 2.53] min), anaerobic threshold (WMD [95% CI] = 170.31 [35.40, 305.22] mL/min/kg), and 6-min walk test distance (WMD [95% CI] = 32.77 [20.72, 44.83] m); in QOL as measured by Short Form (SF-36) physical functioning domain (WMD [95% CI] = 9.95 [2.85, 17.05]) and SF-36 vitality domain (WMD [95% CI] = 6.24 [0.15, 12.34]); and in the echocardiographic measure of LVESD (WMD [95% CI] =  - 0.16 [- 0.28, - 0.04] cm). In conclusion, we found after systematic review and meta-analysis of RCTs that exercise therapy improves exercise tolerance and physical-related quality of life measures.

BACKGROUND:Clostridioides difficile (C. difficile) is a hospital-acquired infection. Overtesting for C. difficile leads to false positive results due to a high rate of asymptomatic colonization, resulting in unnecessary and harmful treatment for patients. METHODS:This was a quality improvement initiative to decrease the rate of inappropriate C. difficile testing across 11 hospitals in an urban, safety-net setting. Three best practice advisories were created, alerting providers of recent laxative administration within 48 hours, a recent positive test within 14 days, and a recent negative test within 7 days. The outcome measures were the number of C. difficile tests per 1000 patient days, as well as the rate of hospital onset C. difficile infection was compared pre- and post-intervention. The process measures included the rate of removal of the C. difficile test from the best practice advisory, as well as the subsequent 24-hour re-order rate. RESULTS:The number of C. difficile tests decreased by 27.3% from 1.1 per 1000 patient days pre-intervention (May 25, 2020-May 24, 2021) to 0.8 per 1000 patient days post-intervention, (May 25, 2021 to March 25, 2022), p<0.001. When stratified by hospital, changes in testing ranged from an increase of 12.5% to a decrease of 60.0%. Analysis among provider type showed higher behavior change among attendings than compared to trainees or advanced practice providers. There was a 12.1%, non-significant decrease in C. difficile rates from pre-intervention, 0.33 per 1000 patient days compared to post-intervention, 0.29 per 1000 patient days, p=0.32. CONCLUSION/CONCLUSIONS:Using only an electronic health record intervention, we successfully decreased C. difficile orders after 72 hours of admission in a large, safety-net system. Variation existed among hospitals and by provider type.

BACKGROUND:There were seven opioid overdoses in this New York City (NYC) federally qualified health center from December 2018 through February 2019, reflecting the rising rate of overdose deaths in NYC overall at the time. In response to these overdoses, we sought to increase the readiness of health center staff to recognize and respond to opioid overdoses and decrease stigmatizing attitudes around opioid use disorder (OUD). METHODS:An hour-long training focusing on opioid overdose response was administered to clinical and non-clinical staff of all levels at the health center. This training included didactic education on topics such as the overdose epidemic, stigma around OUD, and opioid overdose response, as well as discussion. A structured assessment was administered immediately before and following the training to evaluate change in knowledge and attitudes. Additionally, participants completed a feedback survey immediately after the training to assess acceptability. Paired t-tests and analysis of variance tests were used to assess changes in pre- and post-test scores. RESULTS:Over 76% of the health center staff participated in the training (N = 310). There were large and significant increases in mean knowledge and attitudinal scores from pre- to post-test (p < .001 and p < .001, respectively). While there was no significant effect of profession on attitudinal change scores, profession did have a significant effect on knowledge change scores, with administrative staff, non-clinical support staff, other healthcare staff, and therapists learning significantly more than providers (p < .001). The training had high acceptability among participants from diverse departments and levels. CONCLUSIONS:An interactive educational training increased staff's knowledge and readiness to respond to an overdose as well as improved attitudes toward individuals living with OUD. TRIAL REGISTRATION/BACKGROUND:This project was undertaken as a quality improvement initiative at the health center and as such was not formally supervised by the Institutional Review Board per their policies. Further, per the guidelines of the International Committee of Medical Journal Editors, registration is not necessary for clinical trials whose sole purpose is to assess an intervention's effect on providers.

Lifestyle medicine interventions that emphasize healthy behavior changes are growing in popularity in U.S. health systems. Safety-net healthcare settings that serve low-income and uninsured populations most at risk for lifestyle-related disease are ideal venues for lifestyle medicine interventions. Patient-reported outcomes are important indicators of the efficacy of lifestyle medicine interventions. Past research on patient-reported outcomes of lifestyle medicine interventions has occurred outside of traditional healthcare care settings. In this study, we aimed to assess patient-reported outcomes on nutrition knowledge, barriers to adopting a plant-based diet, food and beverage consumption, lifestyle behaviors, self-rated health, and quality-of-life of participants in a pilot plant-based lifestyle medicine program in an urban safety-net healthcare system. We surveyed participants at three time points (baseline, 3 months, 6 months) to measure change over time. After 6 months of participation in the program, nutrition knowledge increased by 7.2 percentage points, participants reported an average of 2.4 fewer barriers to adopting a plant-based diet, the score on a modified healthful plant-based diet index increased by 5.3 points, physical activity increased by 0.7 days per week while hours of media consumption declined by 0.7 h per day, and the percentage of participants who reported that their quality of sleep was "good" or "very good" increased by 12.2 percentage points. Our findings demonstrate that a lifestyle medicine intervention in a safety-net healthcare setting can achieve significant improvements in patient-reported outcomes. Key lessons for other lifestyle medicine interventions include using a multidisciplinary team; addressing all pillars of lifestyle medicine; and the ability for patients to improve knowledge, barriers, skills, and behaviors with adequate support.

In the spring of 2022, mpox spread to non-endemic countries, including the United States. In New York City (NYC), vaccine demand grew as quickly as case counts. With the leadership of the Regional Emerging Special Pathogens Treatment Center (RESPTC) at NYC Health and Hospitals/Bellevue (NYC H+H)-part of the largest public hospital system in the United States-an innovative vaccination model was established that overcame challenges involving health inequities, inadequate access, and lack of vaccine uptake, to successfully administer JYNNEOS vaccines to over 12,000 patients. Transmission has slowed since its peak in August 2022, which has been attributed to successful vaccination campaigns, infection-induced immunity, and behavioral changes among those at highest risk; however, a Centers for Disease Control and Prevention (CDC) assessment released on 4 April 2023 suggests jurisdictions with low vaccination levels (<35%) remain at risk for an mpox resurgence. Here, we summarize the critical aspects of our mpox vaccination model in NYC, which include integration into routine clinical care, prioritization of health equity, and reutilization of COVID-19 vaccination systems, to provide valuable insights for healthcare institutions as we move into the next stage of this ongoing outbreak.

PURPOSE/UNASSIGNED:Financial toxicity of cancer treatment is well described in the literature, including characterizations of its risk factors, manifestations, and consequences. There is, however, limited research on interventions, particularly those at the hospital level, to address the issue. METHODS/UNASSIGNED:From March 1, 2019, to February 28, 2022, a multidisciplinary team conducted a three-cycle Plan-Do-Study-Act (PDSA) process to develop, test, and implement an electronic medical record (EMR) order set to directly refer patients to a hospital-based financial assistance program. The cycles included an assessment of the efficacy of our current practice in connecting patients experiencing financial hardship with assistance, the development and piloting of the EMR referral order, and the broad implementation of the order set across our institution. RESULTS/UNASSIGNED:In PDSA cycle 1, we found that approximately 25% of patients at our institution experienced some form of financial hardship, but most patients were not connected to available resources because of our referral mechanism. In PDSA cycle 2, the pilot referral order set was deemed feasible and received positive feedback. Over the 12-month study period (March 1, 2021-February 28, 2022) of PDSA cycle 3, 718 orders were placed for 670 unique patients across interdisciplinary providers from 55 treatment areas. These referrals resulted in at least $850,000 in US dollars (USD) in financial aid in 38 patients (mean = $22,368 USD). CONCLUSION/UNASSIGNED:The findings from our three-cycle PDSA quality improvement project demonstrate the feasibility and efficacy of interdisciplinary efforts to develop a hospital-level financial toxicity intervention. A simple referral mechanism can empower providers to connect patients in need with available resources.