Academic Health System Research Articles

Poster 374: Patient knowledge of and willingness to pay for platelet-rich plasma injections: A prospective survey study on orthopedic sports medicine patients

Objectives: The efficacy of platelet-rich plasma (PRP) injections for sports-related injuries is a hotly- debated topic. PRP is a cash-pay procedure that is not covered by insurance, and it is actively advertised for a number of different indications, some without supporting evidence. Yet, an increasing number of patients are asking their physicians for PRP injections. The aim of this study is to better understand patients’ knowledge of PRP and willingness to pay out of pocket. Methods: This ongoing Institutional Review Board-approved prospective study included patients presenting with a new chief complaint at a sports medicine orthopedic office in a large urban academic health system. Patients completed an anonymous survey prior to evaluation by an orthopedic surgeon. The survey collected demographic information, clinical information, knowledge of PRP, willingness to pay for PRP based on expected percentage pain reduction, and motivators behind willingness to pay. Two cohorts included: 1) patients who had heard of PRP and 2) patients who had not. The latter group was given a short educational excerpt on PRP. Two-tailed Mann-Whitney U and Kruskal-Wallis tests evaluated pain scores and willingness to pay for PRP between the cohorts. Spearman rank correlation tests evaluated correlations between household income and willingness to pay. Results: To date, we have enrolled 32 patients of which 37.5% had previous orthopedic surgery, 43.75% had seen another medical provider for their current symptoms and 37.5% had previously heard of PRP. Patients who had heard of PRP (versus those that had not) reported lower median pain scores (median 4.75 versus 5.0 out of 10), the same hindrance of daily activities (median 3.0 out of 5), and longer duration of symptoms (median 5.0 months versus 1.0 month), although not statistically significant. Patients were willing to pay more for a greater anticipated reduction in pain: median of $50, $250, and $500 for 25%, 50% and 100% reduction in pain, respectively (Table 1) (p<0.05). Patients who had heard of PRP were willing to pay more for PRP (median $500 versus $200; Table 1); this was not statistically significant. There was no correlation between patients’ income and willingness to pay. Finally, regarding willingness to pay for a given percent reduction in pain, patients reported their top three motivators to be symptoms affecting their day-to-day life, desire to try non-surgical interventions, and desire to return to normal activity (Table 1). For patients who had not heard of PRP, after being informed of PRP these patients reported physician trust as the greatest motivator behind their willingness to pay (Table 1). Conclusions: This ongoing study demonstrates the correlation between willingness to pay and perceived effectiveness of PRP while also noting motivators to pay. Moreover, we were also able to demonstrate some differences between patients who had heard of PRP versus those that had not. These data may be further refined with ongoing enrollment and will provide important insights into patients’ knowledge and how this impacts decision-making in the context of PRP injections. [Table: see text]

Evaluation of a Primary Care-Based Multidisciplinary Transition Clinic for Patients Newly Initiated on Buprenorphine in the Emergency Department.

Care transitions represent vulnerable events for patients newly initiating medications for opioid use disorder (MOUD). Multidisciplinary primary care-based transition clinics may improve care linkage and retention in MOUD treatment. Additionally, these interventions may help primary care clinicians (PCPs) overcome barriers to adopting MOUD into practice. In this evaluation, we assessed the impact of a primary care-based transition clinic for patients newly initiating buprenorphine for opioid use disorder (OUD) in the emergency department. We conducted a retrospective program evaluation within a single academic health system involving adults who newly initiated buprenorphine for OUD through an emergency department-based program and were referred to follow up in either a dedicated multidisciplinary primary care-based transition clinic (SPARC) vs referral to usual primary care (UPC). We performed descriptive analyses comparing patient demographics, referral volume, linkage to care, treatment retention, and markers of high-quality care between the 2 groups. A log-rank test was used to determine the difference in probabilities of retention between SPARC and UPC over 6 months. Over 12 months, the number of referrals to SPARC was greater than to UPC (N = 64 vs N = 26). About 58% of patients referred to SPARC attended an initial visit vs 38% referred to UPC. Treatment retention was consistently greater in SPARC than UPC (1 m: 90% vs 60%; 3 m: 76% vs 40%; 6 m: 60% vs 30%). Markers of care quality including naloxone provision (100% vs 80%) and infectious screening (81% vs 40%) were greater in SPARC clinic. SPARC was associated with a statistically significant increased probability of retention in treatment as compared to UPC (P < .01). In this observational evaluation, a primary care-based multidisciplinary transition clinic for patients initiating buprenorphine MOUD was associated with expanded access to longitudinal OUD treatment and superior linkage to care, retention in care, and quality of care compared to referral to usual primary care. Further research using a more rigorous research design is required to further evaluate these findings.

Study of Patient and Physician Attitudes Toward Automated Prognostic Models for Patients With Metastatic Cancer.

For patients with cancer and their doctors, prognosis is important for choosing treatments and supportive care. Oncologists' life expectancy estimates are often inaccurate, and many patients are not aware of their general prognosis. Machine learning (ML) survival models could be useful in the clinic, but there are potential concerns involving accuracy, provider training, and patient involvement. We conducted a qualitative study to learn about patient and oncologist views on potentially using a ML model for patient care. Patients with metastatic cancer (n = 15) and their family members (n = 5), radiation oncologists (n = 5), and medical oncologists (n = 5) were recruited from a single academic health system. Participants were shown an anonymized report from a validated ML survival model for another patient, which included a predicted survival curve and a list of variables influencing predicted survival. Semistructured interviews were conducted using a script. Every physician and patient who completed their interview said that they would want the option for the model to be used in their practice or care. Physicians stated that they would use an AI prognosis model for patient triage and increasing patient understanding, but had concerns about accuracy and explainability. Patients generally said that they would trust model results completely if presented by their physician but wanted to know if the model was being used in their care. Some reacted negatively to being shown a median survival prediction. Patients and physicians were supportive of use of the model in the clinic, but had various concerns, which should be addressed as predictive models are increasingly deployed in practice.

Individual- and Neighborhood-Level Disparities in Audio-Only Telemedicine Utilization Across a Large Academic Health System.

Introduction: The objective of this study was to understand whether use of audio-only telemedicine visits differed by individual- and neighborhood-level patient characteristics during the COVID-19 pandemic. Methods: We conducted a retrospective cross-sectional study of telemedicine encounter data from a large academic health system. The primary outcome was rate of audio-only versus video visits. The exposures of interest were individual- (age, race, insurance, preferred language) and neighborhood-level (Social Deprivation Index [SDI]) patient characteristics. Results: Our study included 1,054,465 patient encounters from January 1, 2020 to December 31, 2021, of which 18.33% were completed via audio-only. Encounters among adults 75 years or older, Black patients, Spanish-speakers, and those with public insurance were more frequently conducted by audio-only (p < 0.001). Overall, populations showed decreasing rates of audio-only visits over time. We also observed an increase in the rate of audio-only encounters as SDI scores increased. Discussion: We found that audio-only disparities exist in telemedicine utilization by individual and zip code level characteristics. Though these disparities have improved over time as seen by our temporal analysis, marginalized and minority groups still showed the lowest rates of video utilization. In conclusion, access to audio-only care is a critical component to ensure that telemedicine is accessible to all populations. State and federal policy should support continued reimbursement of audio-only care to ensure equitable access to care while the implications of different care modalities are further studied.

Representation of Women Among Grand Rounds Speakers at an Academic Health System Early in the COVID-19 Pandemic.

Introduction An invitation to speak at grand rounds (GR) is considered an honor and an activity important for academic promotion. The aim of this study was to determine the representation of women among invited speakers at departmental GR and assess the impact of the COVID-19 pandemic on it. Methods We conducted a retrospective descriptive study on gender differences in all GR speakers between January 2019 and June 2021 at an academic health system in Western Massachusetts. We calculated the overall percentage of women presenters and compared it with the percentage of women faculty at our institution and nationally.To evaluate the impact of COVID-19 on this association, we calculated the absolute percentage difference between women and men speakers using the date of March 13, 2020, as the cut-off and conducted a sensitivity analysis using June 15, 2020, as the cut-off. Results During the study period, 46% (276/607) of GR speakers at our institution were women. This percentage reflected the percentage of the women faculty overall at our institution and was similar to women's representation among faculty nationally. Departments with high percentages of women faculty (Obstetrics and Gynecology, 76%; Pediatrics, 65%) had high percentages of women GR speakers (Obstetrics and Gynecology, 70%;Pediatrics, 51%; Psychiatry, 62%).COVID-19 did not appear to significantly influence gender representation among speakers. Conclusion At our institution, less than half of the GR speakers were women.However, this percentage appears to reflect the overall percentage of women faculty. Potential barriers and opportunities resulting from the COVID-19 pandemic did not appear to impact this finding.

Health research in academic health systems: time for a new model.

Research, along with patient care and education, is a core element of the academic health system's tripartite mission; it is essential to the academic health system's societal commitment to advancing the public's health. Research at academic health systems in the United States is increasingly resource-constrained and, in important ways, the underlying financial model supporting it has reached a point of unsustainability. This commentary reviews the roles that health research at academic health systems plays in society, describes the ways in which the current model of health research is under strain, and proposes an evolved model and series of organizational and operational steps to consider in moving health research forward.

1691-P: Weight Status, Longitudinal Weight Change, and Weight Loss Treatment Utilization among Primary Care Patients with Obesity

Background: Effective weight management treatments (WMTs) include nutrition counseling, very low-calorie (VLC) meal replacement, anti-obesity medications (AOMs), and bariatric surgery. Little is known about how these WMTs influence weight change among individual primary care patients and populations. Methods: This was a retrospective cohort study using medical record data from one academic health system. Participants were adults with primary care between October 2015 and March 2020. We used serial cross-sections to examine obesity prevalence and WMT use. We used a multistate model to investigate weight trajectories for patients with obesity and no WMT use prior to baseline. This is a time-to-event model with events representing transitions between 5 successive categories of weight change relative to baseline (i.e., ≥10% weight loss, 5-10% weight loss, baseline weight +/-5%, 5-10% weight gain, or ≥10% weight gain). The primary exposures were WMTs; controls were baseline age, BMI, and prior year primary care. Patients with WMT use were propensity matched 1:1 to controls. The primary outcome was the model-based probability of ≥5% weight loss at 1 year. Results: Among primary patients, the prevalence of obesity increased from 39.2% in 2017 to 40.7% in 2019 (n=138,632); WMT use increased from 2.8% to 8.9% (difference: 6.1%, 95% CI: 5.8-6.4%). In the multistate model (n=10,296), the one-year probability of ≥5% weight loss among patients without WMT exposure was 15.5% [95% CI: 14.8-16.1%]. In contrast, ≥5% weight loss was more likely for patients with year-long exposures to any WMT (nutrition counseling: 23.4%, 95% CI: 21.5-25.3%; VLC meal replacement: 56.4%, 95% CI: 48.7-62.8%; AOMs: 27.8%, 95% CI: 25.1-30.5%; bariatric surgery: 94.4%, 95% CI: 91.1-96.4%). Conclusions: Use of all WMTs increased the probability of achieving ≥5% weight loss at the individual-patient level but underutilization limits their ability to reduce weight at the population-level. Disclosure D.H.Griauzde: None. C.R.Richardson: Other Relationship; Annals of Family Medicine, JMIR Diabetes, Research Support; Dexcom, Inc., Blue Cross Blue Shield of Michigan, Neilson Foundation. J.M.Lee: Advisory Panel; GoodRx, Consultant; Tandem Diabetes Care, Inc. J.Henderson: None. A.P.Ehlers: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases (K23DK123416)

Peripheral Nerve Block Delays Mobility and Increases Length of Stay in Patients With Geriatric Hip Fracture.

Peripheral nerve blocks (PNB) has been increasingly used in the care of patients with geriatric hip fracture to reduce perioperative opiate use and the need for general anesthesia. However, the associated motor palsy may impair patients' ability to mobilize effectively after surgery and subsequently may increase latency to key mobility milestones postoperatively, as well as increase inpatient length of stay (LOS). The aim of this study was to investigate time-to-mobility milestones and length of hospital stay between peripheral, epidural, and general anesthesia. A retrospective review identified 1,351 patients aged 65 years or older who underwent surgery for hip fracture between 2012 and 2018 at a single academic health system. Patients were excluded if baseline nonambulatory, restricted weight-bearing postoperatively, or sustained concomitant injuries precluding mobilization, with a final cohort of 1,013 patients. Time-to-event analyses for discharge and mobility milestones were assessed using univariate Kaplan-Meier and multivariate Cox proportional hazard regression analyses. PNB was associated with delayed postoperative time to ambulation ( P < 0.001) and time to out-of-bed ( P = 0.029), along with increased LOS ( P < 0.001). Epidural anesthesia was associated with less delay to first out-of-bed ( P = 0.002), less delay to ambulation ( P = 0.001), and overall reduced length of stay ( P < 0.001). PNB was associated with slower mobilization and longer hospitalization while epidural anesthesia was associated with quicker mobilization and shorter hospital stays. Epidural anesthesia may be a preferable anesthesia choice in patients with geriatric hip fracture when possible. Level III.

Use of nonstigmatizing language is associated with improved outcomes in hospitalized people who inject drugs.

Stigma surrounding opioid use disorder (OUD) is a barrier to treatment. The use of stigmatizing language may be evidence of negative views toward patients. We aimed to identify associations between language and clinical outcomes in patients admitted for infectious complications of OUD. We performed a retrospective medical record review. Four U.S. academic health systems. Participantswere patients with OUD admitted for infectious complications of injection opioid use from January 1, 2018, to December 31, 2018, identified through international classification of diseases, 10th revisioncodes consistent with OUD and acute bacterial/fungal infection. Discharge summaries were reviewed for language, specifically: abuse, addiction, dependence, misuse, use disorder, intravenous drug use, and others. Binary outcomes including medication for OUD, planned discharge, naloxone provision, and an OUD treatment plan were evaluated using logistic regressions and admission duration was evaluated using Gamma regression. A total of 1285 records were reviewed and 328 met inclusion criteria. Of those, 191 (58%) were male, with a median age of 38 years. The most common term was "abuse" (219, 67%), whereas "use disorder" was recorded in 75 (23%) records. Having "use disorder" in the discharge summary was associated with increased odds of having a documented plan for ongoing OUD treatment (adjusted odds ratio[AOR]: 4.11, 95% confidence interval [CI]: 1.89-8.93) and having a documented plan for addiction-specific follow-up care (AOR: 2.31, 95% CI: 1.30-4.09). Stigmatizing language was common in this study of patients hospitalized for infectious complications of OUD. Best-practice language was uncommon, but when used was associated with increased odds of addiction treatment and specialty care referrals.

Current communication practices for biomarker testing in non-small cell lung cancer: Exploring patient and clinician perspectives

ObjectivesWe qualitatively explored patient and clinician experiences with biomarker testing in one academic health system to identify current communication practices and unmet testing information needs. MethodsWe conducted 1:1 in-depth interviews with 15 clinicians (i.e., nurses, oncologists, pathologists) and 12 patients diagnosed with non-small cell lung cancer between January and May 2022. Participants described experiences with biomarker testing as well as associated communication practices and needs. Interviews were audio-recorded and transcribed. Analysis was informed by the Framework Method. ResultsPatients described challenges retaining information early in their patient journey. While patients were generally aware of biomarkers and their effect on treatment options, they expressed limited knowledge of expected time delays between testing and receiving results. Additionally, many did not know their testing results. Clinicians and patients both noted no standard education material on biomarker testing is currently available. They suggested such materials could support patient knowledge and decision-making. ConclusionsCommunication between patients and clinicians about biomarker testing is largely delivered through verbal counseling at a time when patients may be cognitively compromised. All participants supported the idea of delivering standard, tangible education materials on biomarker testing to patients. Practice implicationsEducation materials may enhance counseling efforts and patient knowledge.

The effect of the COVID-19 pandemic on the prescribing of opioid and opioid use disorder medications within an academic medical center in California.

The COVID-19 pandemic impacted healthcare operations affecting many patients with chronic pain and substance use disorder. Our study aimed to evaluate the effects of the COVID-19 pandemic on opioid and opioid use disorder (OUD) medication prescribing practices within a large academic health system in southern California. This retrospective cohort study included patients who received a prescription for chronic opioids or therapy for OUD between November 1, 2019 and September 1, 2020. The date range was divided into five specific time periods during the pandemic: November through December 2019 (pre-COVID and reference period), January through February 2020 (early COVID), March through April 2020 (policy/guidance change period), May through June 2020 (early post-guidance period), and July through August 2020 (late post-guidance period). The primary outcome was change in morphine milligram equivalents (MME) prescribed. Secondary outcomes included encounter type, mode of prescription ordering, naloxone prescriptions, and urine drug screen obtainment. The cohort included 100 patients divided among the designated time periods. Seventy-percent of patients received opioids for chronic non-malignant pain and 10% received therapy for OUD. Although there were numerical increases in MMEs prescribed, no significant changes were seen in the MMEs prescribed at any timepoint relative to the pre-COVID timeframe despite reduced in-person visits, increased video and telephone encounters and increased electronic prescription utilization. Subgroup analyses of those with chronic pain only or OUD had similar findings. It appears that, generally, prescribing practices were sustained despite the various phases of the pandemic including transitions to and from telemedicine.

Laboratory-Developed Tests Account for a Small Minority of Tests Ordered in an Academic Hospital System.

To determine the frequency of use of laboratory-developed tests (LDTs) in an academic medical center system. Retrospective analysis of 2021 test order data from an academic medical center (hospital, outpatient clinics, and cancer center) was done. Measures included assay type, assay methodology, regulatory status, test order volume, inpatient vs outpatient setting, and provider medical specialty. Of the 3,016,928 tests ordered in 2021, 2,831,489 (93.9%) were tests cleared, approved, and/or authorized by the US Food and Drug Administration (FDA); 116,583 (3.9%) were LDTs; and 68,856 (2.3%) were standard methods. These test orders were performed using a total of 1,954 distinct assays. Of these, 983 (50.3%) were FDA assays, 880 (45.0%) were LDTs, and 91 (4.7%) were standard methods. Laboratory-developed tests were more commonly ordered in the outpatient vs inpatient setting and represented a higher proportion of the test volume at the cancer center compared with the university hospital (5.6% vs 3.6%, respectively). The top 167 LDT assays accounted for 90% of the LDT volume (104,996 orders). Among the 20 most frequently ordered LDTs were mass spectrometry assays and tests used in the care of immunocompromised patients. Internal/family medicine placed the greatest number of orders (1,044,642) and ordered one of the lowest proportions of LDTs (3.2%). Laboratory-developed tests made up a small percentage of the total laboratory tests ordered within the academic health system studied.

Patterns of diversity in cancer clinical trial enrollment at a New York City academic health system, 2018-2022.

e18630 Background: Mitigation of cancer clinical trial (CT) enrollment disparities is increasingly prioritized in healthcare, but disparity metrics are inconsistently published by individual health systems. We aimed to assess health-system-level trends in CT enrollment by race, ethnicity, sex, and treatment location (network vs main hospital) during the COVID-19 pandemic, as well as to evaluate county-level disparities in CT accrual fraction (AF) over a five-year period. Methods: This was a cross-sectional study of the population enrolled on therapeutic cancer CTs in a New York City academic health system (AHS) from 2018-2022. Demographics were descriptively analyzed and compared with most recent public NCI-reported therapeutic trial accrual data. New York County 5-year cancer incidence was estimated for each demographic subgroup using age-adjusted cancer incidence rates from public SEER/NPCR data and US Census data. Primary endpoint was AF, defined as proportion of patients accrued relative to cancer incidence. Results: Of 2,533 patients enrolled on CTs over the 5-year period, 51% were White, 18% Black, 17% Hispanic, 9.8% Asian, 0.08% American Indian/AK Native, 0% Pacific Islander, and 3.9% Multiple/unknown race; 47% were female. Minority accrual proportion was higher than that of NCI accrual data (45% vs 25%, p =&lt; 0.001). Annual enrollment fell from maximum 571 patients in 2019 to 457 in 2020, then increased to 499 by 2022. Meanwhile, the proportion of minority enrollment increased to a peak of 49.2% in 2020 and fell to 45% in 2022. Female enrollment of 49% in 2018 decreased to 42% in 2021, increasing sharply to 51% in 2022. On multivariate analysis, Black and Asian patients were more likely to be enrolled after the start of the pandemic than White patients (ORs 1.3, 95% CI 1.1-1.6; 1.4, 1.03-1.8). Black, Asian, and female patients were more likely to be enrolled at network hospital locations than White or male patients, respectively (ORs 1.9, 95% CI 1.3-2.7; 2, 1.2-3.1; 14, 8.6-21). Relative to county-level estimated cancer incidence, Black patients had a higher AF than White patients (0.96% vs 0.77%, OR 1.3, 95% CI 1.1-1.4), while Hispanic patients had a lower AF (0.55%, OR 0.72, 95% CI 0.64-0.8). Women had a lower AF than men (0.66% vs 0.74%, OR 0.9, 95% CI 0.83-0.97). Conclusions: Minority accrual at this AHS was higher than nationally reported. Despite decreased total accrual numbers in 2020 due to the COVID-19 pandemic, high accrual diversity was maintained. County-level AF disparities among women and Hispanic patients suggest opportunities for improvement in community outreach, and greater diversity of clinical trial enrollees at network hospitals may reflect successful outreach to the communities they serve.

COVID-19 Pandemic Effects on Seasonal Influenza Patterns in West Virginia

Introduction The COVID-19 pandemic began in West Virginia (WV) in March 2020 at the end of the seasonal increase in influenza virus infections. Diagnostic and surveillance testing for influenza continued throughout the pandemic because these viruses cause similar diseases and the effect of SARS-CoV-2 on seasonal influenza patterns were unpredictable. Methods Positivity rates for SARS-CoV-2 and influenza viruses were studied using test results obtained from samples collected across an academic health system from July 2018 through June 2022. Trends in COVID-19 variants were assessed using data obtained from national and local databases. Data were analyzed to detect differences in and correlations between SARS-CoV-2 and influenza positivity rates. Results Test positivity rates were higher for the 602,109 SARS-CoV-2 tests than the 186,563 influenza tests performed during the study period (13.7% vs. 3.5%, p&lt;0.0001). SARS-CoV-2 test positivity rose and fell with the emergence of SARS-CoV-2 variants of differing infectivity, being highest when the omicron variant predominated. Significant negative correlations between the number of SARS-CoV-2 and influenza positive tests were seen throughout stages of the pandemic. The normally expected influenza season did not occur during winter 2020 through spring 2021, and there were essentially no influenza cases during the delta surge. A more typical pattern of seasonal influenza was observed in late 2021 to early 2022 during the omicron surge. Conclusion There was an inverse correlation between SARS-CoV-2 and influenza test positivity rates observed throughout several COVID-19 surges in WV. The reasons for these correlations are unclear.

Creation and execution of an effective critical care pharmacist collaborative in an academic health system

AbstractHospital systems continue to face significant pressure to mitigate medication costs, which may be achieved by effective pharmacy collaboration between medical centers. The purpose of this narrative is to discuss the essential elements, methods, obstacles, and solutions that were encountered in the formation and function of a successful multicenter pharmacy collaborative. The objective is to share the tools and examples that illustrate the framework and mechanics employed by a collaborative to execute initiatives that improved patient care and institutional processes. In 2018, five academic medical centers (AMCs) formed a critical care pharmacy collaborative team (PCT). The group's sustainable leadership structure allowed for equal voicing while avoiding individual burnout. Focused monthly working meetings allowed for comparison of each AMC's best practices to identify unexplored practice gaps and project planning. System goals with standardized metrics were set and parallel institutional approaches were implemented. Pooling use experience from five AMCs allowed for a larger sample size for data evaluation to assist in management of high impact medications. Additional benefits of group collaboration extended to research opportunities and incorporation of pharmacy learners for practice‐based learning. During this journey, barriers were uncovered including inconsistent access to a centralized secure platform for information sharing, data collection and analysis, overreliance of responsibilities on few individuals, and unpredictable dedicated project time for team leads and members. Overall, pharmacist led stewardship projects across five AMCs were leveraged to advance patient care through increased clinical guideline adherence, operational efficiency, and medication cost savings.

Association of immune checkpoint inhibitors (ICI) and venous thromboembolism (VTE) in non–small-cell lung cancer (NSCLC): A single healthcare system experience.

e14680 Background: NSCLC and chemotherapy are well-known predisposing factors for VTE. There are conflicting data regarding the association between ICI and VTE in patients with NSCLC. In a retrospective study by Khorana et al. in 2299 NSCLC patients, VTE incidence was numerically lower for patients receiving ICI than chemotherapy. In another study by Attia et al., in 514 patients with lung cancer, there was an increased incidence of VTE with ICI. We conducted a retrospective chart review to further assess the relationship between VTE and ICI in patients with NSCLC who were treated at our facility. Methods: All adult patients &gt;18 years of age with NSCLC who received cancer treatment in our academic health system between January 2011- January 2021 were extracted from the tumor registry. Information was obtained about the type of cancer treatment, deep venous thrombosis (DVT) or pulmonary embolism (PE) occurrence, and associated comorbidities. Patients were divided into two groups – chemotherapy and combined ICI-chemotherapy. Fisher exact tests were used to compare categorical covariates by VTE status, while Wilcoxon rank-sum tests were used for continuous covariates. Multivariable logistic regression was performed to adjust for potential confounding. Results: 370 patients with NSCLC were included in the study. No statistically significant difference was found between VTE incidence and race, histology, or gender. There was a decreased VTE rate among non-advanced cancer (6.2% vs. 18.1%, p=.015) and squamous cell histology (7.5% vs. 28.1%, p=.009). Although the rate of VTE increases slightly in ICI/chemotherapy (17.4%, CI: 11.2%-25.8%) when compared to chemotherapy (15.3% %, CI: 11.2%-20.4%), this was not statistically significant (p=.645). After adjusting for sex, histology, and race, ICI/chemotherapy group had a statistically insignificant increase in odds of VTE events (adjusted OR = 1.24 95% CI = 0.67-2.3, p=.498) compared to the chemotherapy group. The odds of death were significantly lower in patients with ICI/chemotherapy (aOR=.33, 95% CI = .2-.54, p=.001). The median time between ICI use and VTE was 21 days (CI: 12.5-65), while the median time between chemotherapy and VTE was 65.5 days (CI=36.5-146.5). Conclusions: In patients with NSCLC who were treated at our facility, there was no significant VTE difference in patients related to race or gender identity. Likewise, there was no statistical difference in the incidence of VTE in the ICI/chemotherapy and chemotherapy groups. Multivariable logistic regression of various groups to the vital status of the patients shows a statistically significant decrease in death in ICI/chemotherapy group, which is a potential censoring event for VTE. Further studies, including meta-analyses, are required to evaluate the association further, given conflicting results on prior retrospective studies.

Economics of a health system's direct-to-consumer telemedicine for its employees.

To compare the mean per-episode unit cost for a direct-to-consumer (DTC) telemedicine service for medical center employees (OnDemand) with that of in-person care and to estimate whether the offered service increased the use of care. Propensity score-matched retrospective cohort study of adult employees and dependents of a large academic health system between July 7, 2017, and December 31, 2019. To estimate differences in per-episode unit costs within 7 days, we compared costs between OnDemand encounters and conventional in-person encounters (primary care, urgent care, and emergency department) for any similar condition using a generalized linear model. We used interrupted time series analyses limited to the top 10 clinical conditions managed by OnDemand to estimate the effect of OnDemand's availability on the trends for overall employee per-month encounters. A total of 10,826 encounters among 7793 beneficiaries were included (mean [SD] age, 38.5 [10.9] years; 81.6% were women). The mean (SE) 7-day per-episode cost among employees and beneficiaries was lower for OnDemand encounters at $379.76 ($19.83) relative to non-OnDemand encounters at $493.49 ($25.53), a mean per-episode savings of $113.73 (95% CI, $50.36-$177.10; P < .001). After the introduction of OnDemand, among employees with encounters for the top 10 clinical conditions managed by OnDemand, the trend for encounter rates per 100 employees per month increased marginally (0.03; 95% CI, 0.00-0.05; P = .03). These results suggest that DTC telemedicine staffed by an academic health system and offered directly to employees reduced the per-episode unit costs and only marginally increased utilization, suggesting lower cost overall.

Using machine learning-based boosting algorithm to predict hepatotoxicity after local therapy and association with survival in patients undergoing liver directed therapy.

e16229 Background: Hepatocellular carcinoma typically arises in the setting of chronic liver disease. In carefully selected patients, liver directed therapy has been demonstrated to improve survival; however, this benefit from cancer specific survival may be offset by increased mortality from diminished hepatic functional reserve. Treatment-related hepatotoxicity also may vary between liver directed treatment modalities for a given tumor. Our aim was to determine if a machine learning approach could estimate hepatotoxicity prior to treatment. Methods: We identified all patients with hepatocellular carcinoma who underwent liver directed therapy at a large urban academic health system who also had post-treatment laboratory follow-up at 6 months. Patients were categorized as having worse liver function if they experienced a decline in albumin-bilirubin (ALBI) grade or more than 0.5 absolute decline in ALBI score at post treatment follow-up. A machine learning, extreme-gradient boosting (XGBoost) approach was employed to identify predictive features for hepatotoxicity using fine-tuned hyperparameters. Results: The study cohort consisted of 269 patients. 48 (18%) were categorized as having worsened hepatic function after liver directed therapy, with an average decline of -0.72 in ALBI score. Median survival after liver directed therapy was 5.8 years, with worsening hepatic function associated with a 5 month decline in overall survival. 158 (59%) patients received transarterial chemoembolization, 99 (37%) patients received stereotactic body radiotherapy and 12 (7%) patients received Y-90. A machine learning model was developed resulting in an Area Under the ROC Curve (AUC) of 0.894 in a 10-point decision tree model. On SHAP analysis, the features with the largest impact on hepatotoxicity prior to treatment was pretreatment T-stage followed by Child-Pugh Score. Conclusions: A machine learning model was developed to accurately classify patients who sustained treatment related hepatotoxicity. Significant decline in ALBI score or grade was associated with a worsening of overall survival. Future studies in larger prospective clinical trials utilizing this machine learning-based methodology of assessing liver dysfunction after focal therapy are required to validate these findings.

Rates of surgical site infections and length of stay following major gynecologic oncology surgery for patients on federally funded insurance: A 6-month retrospective cohort.

e18627 Background: An estimated 94,000 women have newly diagnosed gynecologic cancer annually–with rates increasing over the last 20 years. The incidence of gynecologic cancers is known to vary by insurance status; more specifically, uninsured or federally insured women are more likely to have advanced-stage disease at the time of diagnosis and worse overall survival compared to privately insured women. Surgical site infections (SSIs) complicate up to 4% of hysterectomies in the United States. Other post-operative sequelae, such as wound dehiscence and 30-day readmission, further complicate the post-surgical course. We sought to determine differences in postoperative outcomes for patients with private and federally-funded insurance plans, such as SSI, length of hospital stay, and 30-day wound dehiscence, and 30 all-cause readmission following gynecologic oncologic surgery. Methods: We conducted a retrospective cohort study from a large metropolitan academic health system between January 1st, 2022, to June 30th, 2022. The study population included adult women who received gynecologic oncologic surgery–surgical staging or interval cytoreduction. Postoperative complications were compared to national rates from the National Surgical Quality Improvement Project (NSQIP) database. Trends were analyzed using chi-squared analysis and a two-sample t-test. Results: Over our 6-month study period, 233 patients underwent gynecologic oncologic surgery. Robotic-assisted hysterectomies were the most common (27.89%, n = 65), followed by total abdominal hysterectomies (21%, n = 49), laparoscopic hysterectomies (7.72%, n = 18), and supracervical hysterectomies (3%, n = 7). More patients required bilateral salpingo-oophorectomy (63.51%, n = 148) compared to unilateral procedures (9.87%, n = 23). Although most patients had private insurance (73.39%, n = 171), 26.1% had federally-funded health plans (n = 62)–Medicaid and Medicare. Rates of wound dehiscence were higher than the NSQIP rate of 1% among our cohort (3%, n = 7). Although there was a high rate of 30-day readmission (9.44%, n = 22) compared to NSQIP rate of 6.1%, the incidence of SSI was fewer than the NSQIP rate of 4% (3.86%, n = 9). Upon intersectional analysis, patients with federal insurance plans had higher rates of SSI (n = 5, p &lt; 0.05). Specifically, Medicare patients had a longer average length of stay at 2.64 days (p &lt; 0.05) compared to a study average of 1.76 days. Conclusions: Patients with federal insurance had longer hospital stays and higher rates of SSI compared to those with private plans. There was no difference in 30-day wound dehiscence or 30-day all-cause readmission between study populations. More studies are needed to risk-stratify patients by insurance type before gynecologic oncology surgery to improve postoperative outcomes further.

Social Determinants of Health Screening and Management: Lessons at a Large, Urban Academic Health System.

In October 2022 a multisite social determinants of health screening initiative was expanded across seven emergency departments of a large, urban hospital system. The aim of the initiative was to identify and address those underlying social needs that frequently interfere with a patient's health and well-being, often resulting in increased preventable system utilization. Building on an established Patient Navigator Program, an existing screening process, and long-standing community-based partnerships, an interdisciplinary workgroup was formed to develop and implement the initiative. Technical and operational workflows were developed and implemented, and new staff members were hired and trained to screen and support patients with identified social needs. In addition, a community-based organization network was formed to explore and test social service referral strategies. Within the first five months of implementation, more than 8,000 patients were screened across seven emergency departments (EDs), of which 17.3% demonstrated a social need. Patient Navigators see between 5% and 10% of total nonadmitted ED patients. Among the three social needs of focus, housing presented as the greatest need (10.2%), followed by food (9.6%) and transportation (8.0%). Among patients identified as rising/high risk (728), 50.0% accepted support and are actively working with a Patient Navigator. There is growing evidence to support the link between unmet social needs and poor health outcomes. Health care systems are uniquely positioned to provide whole person care by identifying unresolved social needs and by building capacity within local community-based organizations to support those needs.