8551 Background: Immune checkpoint inhibitors (ICIs) have improved survival outcomes in patients (pts) with advanced non–small cell lung cancer (NSCLC). Although pivotal trials stipulated ICI therapy for up to 2 years, many pts in clinical practice continue therapy beyond 2 years. Data on long-term toxicities developing after more than 2 years of ICI treatment are very limited. Methods: A single-institution retrospective analysis including pts who received greater than 2 years of ICI therapy for recurrent/metastatic NSCLC from 01/2012-01/2023 was performed. Patient demographics, disease characteristics, treatment history, toxicities, and outcomes were abstracted from the electronic medical record. Late IRAEs were defined as those occurring more than 2 years after initiation of ICI therapy. IRAEs were graded using CTCAE v5.0 by chart review. Independent sample t-tests and chi-square analyses were used for univariate comparisons. Association of late IRAE with OS and PFS was assessed using extended Cox regression with late IRAE modeled as a time varying covariate. Results: Our cohort consisted of 76 pts who received > 2 years of ICI therapy. Median age at the 2-year time point was 64 (range 40-83); 51 (67.1%) were White; 48 (63.2%) were Female; 66 (86.8%) had ECOG PS 0-1; 66 (88.0%) had non squamous histology. More than half (42/76, 55.3%) had PD-L1 TPS ≥50%, while 18/76 (23.7%) had PD-L1 TPS 1-49%. Median duration of ICI treatment was 41.1 mos (range 26.5- 84.4 mos). Before the 2-year time point, 44 (57.8%) patients had an IRAE; most (38/44, 86.3%) were Grade 1-2; only 11 (26.2%) of these pre-2-year IRAEs required steroids, with the remainder managed with symptomatic support. After 2 years on ICI therapy, 38/76 (50%) pts had a late IRAE, most of whom (23/38, 60.5%) had no evidence of IRAE prior to 2 years. Of the 15/38 pts (39.5%) with a prior IRAE before 2 years, only 3 late IRAEs (20%) involved the same organ system as the pre-2 year IRAE. Late IRAEs were most commonly dermatologic (26.3%), gastrointestinal (21.1%) and musculoskeletal (18.4%). Of the 38 late IRAEs, 7 (18.4%) were grade 1; 21 (55.3%) grade 2; and 10 (26.3%) grade 3. Steroids for late IRAEs were required in 14 (36.8%) pts while 24 pts (63.2%) were managed with supportive care. ICI therapy was discontinued due to late IRAE in 15/38 pts (39.4%). Higher rates of late IRAEs were seen in females (p=0.032), non-white pts (p=0.041), and in pts who had experienced grade > 2 or higher IRAE before 2 years (p=0.020). Late IRAE occurrence was not significantly associated with mPFS (p=0.965) or mOS (p=0.691). Conclusions: Late IRAEs are common in pts with NSCLC treated with ICI therapy beyond 2 years, including in patients with no prior history of IRAE. Females, non-Caucasian patients, and patients with previous high grade IRAEs may be at higher risk for developing late IRAEs. Late IRAEs do not appear to be associated with longer PFS and OS.

75 Background: Financial hardship leads to treatment nonadherence including missed medical appointments. This study describes a financial advocacy intervention, developed in response to Medicare’s Oncology Care Model’s price estimate requirement, and measures its effect on patient-initiated missed appointments (no-shows/cancellations). Methods: Logistic regression was used to analyze the clinical records of 2523 chemotherapy patients, before and after program implementation, at one semi-rural cancer center (2015 – 2018). Cluster-correlated covariance was used to correct variances for repeated measures. Results: Patients were assigned an advocate who provided out-of-pocket price estimates, health insurance education, and followed-up with the patient every 30 days for 6 months. 1,271 patients (50.4%) received the intervention. Average age was 64.1 and the average number of missed appointments was 8.7. After accounting for age, gender, marital status, insurance type, and number of chemotherapy infusions, receiving financial advocacy was associated with a 9% reduction in the odds of a missed appointment (OR = 0.91; 95% CI= 0.84 – 0.98; p = 0.013). Medicaid (OR = 1.36; 95% CI = 1.17 – 1.58; p = 0.000) and divorce (OR = 1.19; 95% CI = 1.03 – 1.37; p = 0.015) increased the odds of missed appointments, older age (OR = 0.99; 95% CI = 0.99 – 1.00, p = 0.000) and male gender (OR = 0.90; 95%CI = 0.84 – 0.98; p = 0.011) reduced the odds of missed appointments. Conclusions: The intervention provided some protection against missed appointments. However, marital status, age, gender, and insurance type remained significantly associated with missed appointments, even after accounting for financial advocacy. These findings indicate the need for more research on financial advocacy to identify the effective core components, target at-risk populations, and understand the effect of implementation on intervention effectiveness.[Table: see text]

e24165 Background: Head and neck cancer (HNC) treatment with chemoradiation therapy (CRT) is highly morbid. Patients develop significant symptoms including severe pain, dysphagia, and malnutrition resulting in poor quality of life (QOL), frequent hospitalizations, and high rates of depression. Although integration of palliative care (PC) teams into the care of incurable cancers has resulted in improved QOL, less is known about its integration into the care of patients receiving curative-intent treatments. We sought to characterize the use of PC during HNC CRT at a single cancer center. Methods: We performed a retrospective chart review of patients who received curative-intent CRT for HNC from 1/2008 to 1/2019 at Fox Chase Cancer Center. We excluded patients with metastatic disease or who participated in a PC intervention study. Data extracted included: demographics, comorbidities, cancer subtypes, human papilloma virus (HPV) status, inpatient admissions, feeding tube use, pain medication usage, and PC referrals throughout CRT. We used parametric and non-parametric statistics for hypothesis testing. Results: A total of 332 patients were included; median age was 60, 73% were male, 57.8% had oropharynx HNC, 10.2% had a history of chronic pain, and 18.9% had a psychiatric diagnosis. At the start of treatment, 38.5% of patients were on opioids for cancer pain or chronic pain, which increased to 90.7% of patients by the end of CRT. Six months after CRT, 30% of patients remained on opioids. 44.7% of patients required a feeding tube and 51.4% were admitted to the hospital. Despite the significant symptomatology, only 64 patients (19.2%) were referred to PC during treatment. The patients referred to PC were more likely to be younger (p=0.002), have an inpatient admission during treatment (p=0.001), a history of a psychiatric disorder (p=0.001) or substance abuse (p=0.003), and less likely to have HPV-related HNC. Conclusions: A large majority of patients receiving CRT for HNC faced significant pain with a high hospitalization rate, yet few had the expertise of PC involved in their care. Those who were referred to PC were more likely to have pre-existing substance use or chronic pain. These data provide strong rationale for the development of standardized, PC-based interventions to help with the symptomatology of HNC treatment to better serve this high needs population. [Table: see text]

1578 Background: Clinical pathways have emerged as a strategy to provide value-based care by reducing treatment variation across oncology practice. Despite evidence for cost reduction and improved patient outcomes, adoption of clinical pathways has met resistance in the oncology community given concerns regarding the impact on workflow, and cost. Flatiron Assist, an EMR based pathways tool, is a promising solution to minimizing workflow friction by proving a novel, streamlined user interface. Methods: The Abramson Cancer Center at the University of Pennsylvania and Flatiron Health conducted a pilot study from August 2021-December 2022 to test the use of Flatiron Assist within its thoracic oncology practices. The goal of the pilot was to increase utilization as well as usability of the pathways tool over time through clinician and developer iterative collaboration. Penn’s clinical pathways were incorporated into the tool and integrated into the EMR. An education program trained clinicians on the tool’s functionality. Based on clinician feedback, the Penn team identified two areas of improvement for the tool: speed and organization of the tool keys and prompts. Changes made over time included: 1) increasing the speed of the tool 2) restructuring the organization of keys and prompts, and 3) creation of a “hot button” for three common clinical scenarios in order to increase efficiency. Speed of the application (session length in seconds), utilization of the hot button (number of sessions/overall sessions) and overall physician adoption (treatment ordering with Flatiron Assist/total thoracic orders) were calculated during the pilot. Results: Provider utilization of the pathways tool increased over time with a baseline adoption rate of 29% in August 2021 and a peak of 74% in April 2022. Median session length decreased over time from 106.4 seconds (August 2021) to 39.9 seconds (December 2022). Peak performance for session length was 38.2 seconds (November 2022). Usage of the hot button introduced in August 2022 has increased over time from 27.6% of sessions to 41.8 % in November 2022. Thoracic providers ordered, on average, 1.7 new treatment plans per week, resulting in less than 1.5 extra minutes of work per week with optimal use of the pathways tool. Conclusions: Collaboration between Penn and Flatiron Health effectively reduced specific areas of friction associated with the use of Flatiron Assist. Changes made to the speed, button organization and functionality of the tool contributed to increased clinician utilization over time. Future development of EMR based pathway tools should incorporate clinician engagement and feedback.

8523 Background: The PACIFIC trial demonstrated a 10% improvement in 5-year survival with the addition of consolidation durvalumab versus placebo after chemoradiation (CRT) in good performance status patients (pts) with stage III non-small cell lung cancer (NSCLC). However, not all patients who complete CRT go on to receive consolidation durvalumab. We sought to describe real-world use of consolidation durvalumab or other immune checkpoint inhibitors (ICI) in this setting within a single academic health system. Methods: We retrospectively identified pts with unresectable stage III NSCLC treated with definitive CRT between October 2017 and October 2020 within the University of Pennsylvania Health System, including two urban hospitals and two satellite centers. Pts either received consolidation ICI (ICI group) or did not (no ICI group). Baseline characteristics of the groups were compared with the Chi-squared, Fisher exact, or Wilcoxon rank-sum test as appropriate. Overall survival (OS), measured from the last day of CRT, was compared using the Kaplan-Meier method and log-rank test. Results: Of the 148 consecutively treated pts who completed CRT, 108 (73%) received consolidation ICI; 40 (27%) did not. Within the ICI group, 42% completed 1 year (yr) of treatment. Within the no ICI group, reasons for non-receipt included disease progression (n = 14, 35%), CRT toxicity (n = 7, 18%), comorbidity or decline unrelated to CRT (n = 7, 18%), provider choice (n = 6, 15%) due to EGFR mutation (n = 5) or atypical histology (n = 1), pt refusal (n = 3, 8%), and death without progression (n = 3, 8%). The ICI group had better performance status (ECOG 0/1/2, 46%/49%/5% ICI vs 25%/48%/28% no ICI, p < 0.001) lower Charlson Comorbidity Index (median, 5 [IQR 4-6] ICI vs 6 [IQR 5-8] no ICI, p = 0.02), and lower rates of active autoimmune disease or immunosuppression (5% ICI vs 15% no ICI, p = 0.03). There were no differences between groups in age (median, 68 yrs [IQR 63-73] ICI vs 71 yrs [IQR 65-73] no ICI, p = 0.25), sex (female, 60% ICI vs 50% no ICI, p = 0.27), race (Black, 19% ICI vs 20% no ICI, p = 0.82), stage (IIIA/B/C, 42%/48%/11% ICI vs 40%/50%/10% no ICI, p = 0.96), and PD-L1 expression ( < 1%/1-50%/ > 50%/unknown, 36%/25%/29%/10% ICI vs 40%/25%/28%/8% no ICI, p = 0.97). 1- and 2-yr OS were 83% and 61% in the ICI group versus 52% and 34% in the no ICI group, respectively (p < 0.001). Within the no ICI group, OS was worse among those with versus those without disease progression (PD) post-CRT (1-yr OS 24% vs 74%, p = 0.03). Conclusions: In this retrospective study within a large academic health system, we found that over one-quarter of pts who completed chemoradiation for stage III NSCLC did not receive consolidation ICI, most commonly due to disease progression, CRT toxicity, or comorbidity. Survival amongst these pts is particularly poor, especially for those who experience PD shortly after CRT.

381 Background: With the growing complexity and cost of cancer care, adoption of oncology pathway tools as clinical decision support (CDS) has increased at the point of care. These tools have been shown to improve care quality, reduce variations in care and reduce healthcare costs. Flatiron Assist (FA) is a customizable CDS tool that is embedded in the electronic health record (EHR) to facilitate selection and documentation of National Comprehensive Cancer Network (NCCN) guideline concordant cancer treatment regimens. The time burden of engaging with CDS tools is a top concern for clinicians and often limits uptake of these tools. However, data on the actual time spent utilizing these tools in clinical practice is limited. Our study reports on the time spent by clinicians in various oncology practices, utilizing and ordering cancer treatment regimens with a CDS tool. Methods: We reviewed all completed sessions in FA from July 23, 2021 to May 23, 2022 by 878 prescribers at 24 academic and community oncology centers. Sessions that were part of an electronic prior authorization pilot program were excluded for purposes of standardization. Session length was defined as the time from the completion of application launch to application exit. For prescribers who completed over 20 sessions, we compared the mean sessions lengths of user’s initial sessions (#1-10) to subsequent sessions (#11-20) using paired t tests. We also determined the total number of potential CDS tool sessions per user per week. Results: All 14,394 CDS tool sessions were analyzed during the 10 month observation period. Median session length was 42 seconds (95th percentile, 253 seconds or 4.2 minutes). The average number of treatment orders placed was 5 per week per provider. Table shows the median session length across the 9 most commonly treated cancers. For the 31 prescribers who started using FA during the study and completed more than 20 sessions, the mean (SD) session length decreased from 95.2 (11.8) seconds in sessions #1-10 to 71.0 (6.0) seconds in sessions #11-20 (two-sided P = 0.034). Conclusions: Utilization of FA as a CDS tool to provide guideline concordant care is feasible and adds less than 4 minutes/week on average to a provider’s workflow. When used effectively, it can integrate seamlessly into provider workflows across academic and community practices without disruptive increases in time or effort of cancer treatment regimen ordering. Future studies are needed to evaluate the impact of product improvements, content changes, and electronic prior authorization integrations on time in this workflow.[Table: see text]

29 Background: Hospice has been associated with improved quality of life for patients, cost savings, and reduction in caregiver-grief-related depression. While cancer patients make up a plurality of hospice utilizers nationally, many patients are only on hospice for a limited period (in Medicare patients, a median of 18 days). Studies suggest engaging cancer patients to discuss goals and priorities using the Serious Illness Conversation (SIC) Guide has a positive impact on prognostic understanding and end-of-life planning. More frequent utilization of SICs may prompt earlier enrollment of oncology patients in hospice when appropriate. Methods: We identified cancer patients enrolled in hospice at the Abramson Cancer Center at Penn Presbyterian Medical Center from 2019-2020 after all providers received SIC training. Patient demographics, cancer diagnosis, type of hospice (home versus inpatient), SIC usage, palliative care referral patterns and time on hospice were abstracted. Results: 104 patients were enrolled in hospice during the study period. The majority of patients were female (51%). 45% were Caucasian, and 31% were African American. The most common cancer diagnoses were thoracic (52%) and gastrointestinal (32%) malignancies. 85 patients (82%) were enrolled on home hospice and 19 patients (18%) inpatient hospice. Palliative care usage included 50 inpatient and 24 outpatient consultations; 30 patients (29%) in the cohort never utilized palliative care. 52 (50%) of patients did not have a SIC. 47% (40 patients) enrolled in home hospice had an SIC while 63% (12 patients) on inpatient hospice had an SIC. The median time interval between a patient’s SIC conversation and hospice enrollment was longer in home hospice patients (74 days) compared to inpatient hospice (33 days). Patients on home hospice spent an average of 44 days on hospice versus 2 days in the inpatient setting. Conclusions: Half of the patients at Penn Presbyterian Medical Center enrolled in hospice during the study period did not have an SIC, and 29% did not see palliative care prior to starting hospice. The median time from SIC initiation to hospice enrollment was significantly longer for patients on home hospice compared to inpatient hospice suggesting a need for earlier SIC interventions. Patients enrolled in inpatient hospice spent a considerably shorter period of time on hospice also underscoring the importance of earlier end of life planning. Our findings indicate a need for additional interventions to facilitate earlier SIC conversations in the outpatient setting and a demand for increased palliative care access.

207 Background: Prostate cancer accounts for approximately 10% of cancer deaths in men worldwide. Clinical trials for metastatic castrate resistant prostate cancer (mCRPC) have established survival benefit with use of abiraterone (ABI) with prednisone or enzalutamide (ENZ). Despite their wide utilization, little is known about patient quality-of-life (QOL) outcomes for these agents. Our study evaluates patient reported QOL while taking ENZ or ABI/prednisone. Methods: 22 mCRPC patients were enrolled in an open label, nonrandomized manner to receive oral ENZ (n=12) or ABI/prednisone (n=6) per oncologist’s discretion. Patients completed multiple QOL validated questionnaires, including EPIC-26, FACT-P, and FACT-COG at baseline, 1,2,3,6,9 and 12 months or until progression/change of therapy. Surveys were scored by treatment group using mean, median, range, and standard deviations. QOL parameters were compared between the two groups with two-sided, two-sample T test and linear mixed models. Results: Surveys discontinued prior to 1 year secondary to disease progression/change of therapy were 58% and 33% for ENZ and ABI, respectively. By month 3, 50% of surveys were returned for ENZ and 33% for ABI. Month-to-month comparisons of QOL parameters including urinary irritation, incontinence, bowel, sexual, hormonal function, and overall well-being showed no significant differences between treatment groups or different rates of change. Perceived Cognitive Impairment was significantly lower for patients on ABI in month 3, yet Perceived Cognitive Ability favored ENZ in months 2 and 3. All other data points for cognition showed no significant differences. Conclusions: Data from FACT-COG shows discordance in perceived Cognitive Impairment and Abilities between ENZ and ABI in months 2-3. Other QOL domains indicated no difference between the two groups. The study was limited by a significant portion of patients with disease progression/change of therapy. For those on therapy, survey compliance remained high. Thus, the use of questionnaires is a feasible means of assessing patient outcomes and can be adapted to larger studies.