Search

PURPOSE The COVID-19 pandemic–related disruptions in health care delivery might have affected end-of-life care in patients with cancer. We examined changes in place of death and hospice support for Medicaid and commercially insured patients during the pandemic. PATIENTS AND METHODS We linked Washington State cancer registry records with claims from Medicaid and two commercial insurers for patients with solid tumor age 18-64 years. The study included 322 Medicaid and 162 commercial patients who died between March 2017 and June 2019 (pre–COVID-19), along with 90 Medicaid and 47 commercial patients who died between March and June 2020 (COVID-19). Place of death was categorized as hospital, hospice (home or nonhospital facility), and home without hospice. Place of death was compared using adjusted multinomial logistic regressions stratified by payer and time period (pre–COVID-19 v COVID-19). The clinical and sociodemographic factors associated with dying at home without hospice were examined, and adjusted marginal effects (ME) are reported. RESULTS In the adjusted pre–COVID-19 analysis, Medicaid patients were more likely than commercially insured patients to die in hospital (48% v 36%; adjusted ME, 11%; P = .02). In the pre–COVID-19/COVID-19 analysis, Medicaid patients' place of death shifted from hospital (48% v 32%; ME, −16%; P < .01) to home without hospice (19.9% v 38.0%; ME, 16.5%; P < .01). However, there were no statistically significant changes pre–COVID-19/COVID-19 for commercial patients. As a result, during COVID-19, Medicaid patients were more likely than commercial patients to die at home without hospice (38% v 22%; ME, 16%; P = .04) as were male versus female patients (ME, 16%; P < .01). CONCLUSION The pandemic might have disproportionately worsened the end-of-life experience for Medicaid enrollees with cancer. Attention should be paid to societal and health system factors that decrease access to care for Medicaid patients.

Background: The purpose of this study was to determine factors associated with receipt of screening mammography by insured women before breast cancer diagnosis, and subsequent outcomes. Patients and Methods: Using claims data from commercial and federal payers linked to a regional SEER registry, we identified women diagnosed with breast cancer from 2007 to 2017 and determined receipt of screening mammography within 1 year before diagnosis. We obtained patient and tumor characteristics from the SEER registry and assigned each woman a socioeconomic deprivation score based on residential address. Multivariable logistic regression models were used to evaluate associations of patient and tumor characteristics with late-stage disease and nonreceipt of mammography. We used multivariable Cox proportional hazards models to identify predictors of subsequent mortality. Results: Among 7,047 women, 69% (n=4,853) received screening mammography before breast cancer diagnosis. Compared with women who received mammography, those with no mammography had a higher proportion of late-stage disease (34% vs 10%) and higher 5-year mortality (18% vs 6%). In multivariable modeling, late-stage disease was most associated with nonreceipt of mammography (odds ratio [OR], 4.35; 95% CI, 3.80–4.98). The Cox model indicated that nonreceipt of mammography predicted increased risk of mortality (hazard ratio [HR], 2.00; 95% CI, 1.64–2.43), independent of late-stage disease at diagnosis (HR, 5.00; 95% CI, 4.10–6.10), Charlson comorbidity index score ≥1 (HR, 2.75; 95% CI, 2.26–3.34), and negative estrogen receptor/progesterone receptor status (HR, 2.09; 95% CI, 1.67–2.61). Nonreceipt of mammography was associated with younger age (40–49 vs 50–59 years; OR, 1.69; 95% CI, 1.45–1.96) and increased socioeconomic deprivation (OR, 1.05 per decile increase; 95% CI, 1.03–1.07). Conclusions: In a cohort of insured women diagnosed with breast cancer, nonreceipt of screening mammography was significantly associated with late-stage disease and mortality, suggesting that interventions to further increase uptake of screening mammography may improve breast cancer outcomes.

Introduction We assessed the proportion of patients with advanced epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK) positive non–small-cell lung cancer (NSCLC) who receive tyrosine kinase inhibitors (TKIs) in the real-world, predictors of TKI use, and impact of TKI therapy on overall survival (OS). Materials and Methods We identified patients diagnosed with stage IV EGFR+ and ALK+ positive NSCLC from January 1, 2010 to December 31, 2018, in the Cancer Surveillance System registry and linked their records to Medicare and commercial insurance claims. We reported the proportions of patients with 1 or more TKI claims versus no TKI claims and used logistic regression to identify predictors of TKI use. We evaluated the effect of TKI use on OS by applying extended Cox proportional hazard models with TKI use as a time-dependent exposure and landmark analysis in a subcohort (N = 105). We adjusted Cox models for confounding patient characteristics. Results Of 117 eligible patients (median age = 69; 62% women; 88% EGFR+), 21 (17.9%) had no TKI claims. Diagnosis in 2015 to 2018 was independently associated with lower likelihood of TKI therapy compared with 2010 to 2014 (adjusted odds ratio, 0.29; P = .020). TKI use was associated with longer OS in a multivariate extended Cox model and in the landmark analysis (adjusted hazard ratio [HR], 0.58; 95% confidence interval [CI], 0.33; 0.99; P = .048; adjusted HR, 0.55; 95% CI, 0.30; 1.00; P = .050). Conclusion Approximately 18% of patients with advanced EGFR+ and ALK+ positive NSCLC do not receive TKIs and have inferior survival. Further studies need to investigate barriers of access to TKIs in biomarker-selected patients.

Background Polypharmacy raises the risk of drug-drug interactions and adverse events among patients with cancer. Most polypharmacy research has focused on adults age 65 or older enrolled in Medicare insurance. To better inform pharmacy practice and cancer care delivery, data are needed on polypharmacy among commercially insured patients with cancer and those younger than 65. Methods We performed a retrospective analysis of insurance enrollment and claims files linked to the Puget Sound Cancer Surveillance System for adults age 18 and older who were commercially insured, diagnosed with stage IV cancer, survived 30+ days after diagnosis, and did not enroll in hospice. We describe the prevalence of polypharmacy, chemotherapy use, and medication-related out-of-pocket (OOP) costs in the last month of life. Results Of 606 patients, 390 (64%) experienced polypharmacy (i.e. 5+ medications) in the last 30 days of life. Almost half (n = 297, 49%) received chemotherapy or targeted agents; chemotherapy was associated with significantly higher odds of polypharmacy (odds ratio (OR) 2.93, 95% confidence interval (CI) 2.04–4.20). The most commonly prescribed medications at end of life were opioids, benzodiazepines and anti-emetics. Among 484 patients (80%) incurring medication-related costs in the last month of life, median total OOP cost was $82 (interquartile range $30–$200). Seven patients (1%) had total costs above $5,000. The median chemotherapy-related OOP cost was $446 (IQR $150–$1896); 32 patients (7%) had chemotherapy-related OOP costs between $1,000 and $5,000. Conclusion Most patients with advanced cancer experienced polypharmacy at end of life, although most medications observed herein are commonly used for supportive care. Patients receiving chemotherapy had higher medication-related OOP costs, and chemotherapy was significantly associated with polypharmacy at end of life. Evaluation of polypharmacy at end of life may represent an important opportunity to improve quality of life and reduce costs for patients and families.

PURPOSE: We investigated the association of out-of-pocket (OOP) costs for tyrosine kinase inhibitors (TKIs) with overall survival (OS) in epidermal growth factor receptor ( EGFR)- and anaplastic lymphoma kinase ( ALK)-positive advanced non–small-cell lung cancer (NSCLC). We secondarily investigated associations of TKI OOP costs with TKI adherence, duration of therapy (DOT), and TKI discontinuation. METHODS: We used the Hutchinson Institute for Cancer Outcomes Research registry-claims database to identify patients with stage IV EGFR- or ALK-positive NSCLC; ≥ 1 claims for EGFR or ALK TKIs; and ≥ 3-month survival from TKI initiation. We estimated the average monthly TKI OOP costs per patient up to 3 months from TKI initiation, categorizing patients into quartiles of TKI OOP costs (Q1 < Q2 < Q3 < Q4). We conducted landmark analysis at 3 months from TKI initiation to compare Q1-3 v Q4 TKI OOP costs with respect to OS, TKI DOT, TKI adherence, and TKI discontinuation. RESULTS: Seventy-eight and twenty-seven patients comprised the Q1-3 and Q4 groups, respectively. Median monthly TKI OOP costs were $1,431 (Q1-3) v $2,888 (Q4). Compared with Q1-3, Q4 patients had inferior OS (adjusted hazard ratio [HR], 1.85; [95% CI, 1.11 to 3.10], similar TKI DOT (adjusted HR, 1.06; 95% CI, 0.53 to 2.15), decreased TKI adherence (adjusted odds ratio [OR], 0.28; 95% CI, 0.10 to 0.76), and higher TKI discontinuation rate (adjusted OR, 8.75; 95% CI, 2.59 to 29.52). CONCLUSION: Among patients with advanced EGFR- and ALK-positive NSCLC, higher TKI OOP costs are associated with decreased TKI adherence, a higher likelihood of TKI discontinuation, and inferior survival.

PURPOSE: Aggressive care at the end of life (EOL) can lead to unnecessary suffering and health care costs for patients with cancer. Despite geographic proximity and cultural similarities, we hypothesize that EOL care is more intense in the United States multipayer system versus the Canadian single-payer system. We compared health care utilization at EOL among patients with cancer in Alberta, Canada, with those in Washington state in the United States. METHODS: Adult patients with American Joint Committee on Cancer stage II to IV solid tumors who died between 2014 and 2016 in Alberta and between 2015 and 2017 in Washington were identified from regional population-based cancer registries linked to treatment and hospitalization records (Alberta) and health claims from major regional insurance plans (Washington). The proportion of patients receiving chemotherapy and having multiple emergency department (ED) visits, or intensive care unit (ICU) admissions in the last 30, 60, and 90 days of life (DOL) in Alberta and Washington were determined and compared using two-sample z-test and multivariable logistic regression (α = .006 after Bonferroni correction). RESULTS: Of patients, 11,177 in Alberta and 12,807 in Washington were included. Patients were similar in age (median, 71 v 72 year), with more patients in Washington with no comorbidities. More patients in Washington were treated with chemotherapy (12.6% v 6.6%; adjusted OR [aOR], 2.74), had multiple ED visits (16.2% v 12.1%; aOR, 1.40), and ICU admissions (23.7% v 3.9%; aOR, 14.27) in the last 30 DOL. Utilization was also higher in Washington in the last 60 and 90 DOL and among those with stage IV disease and those age 65 years and older. CONCLUSION: Utilization of chemotherapy, ED visits, and ICU admissions near EOL was higher in Washington versus Alberta. Future studies to characterize drivers of aggressive EOL care may help improve cancer care for patients in the United States and Canada.

PURPOSE: Systemic therapy use in the last 30 days of life (DOL) for patients with advanced cancer is a low-value medical practice. We hypothesized that systemic therapy use in the last 30 DOL increased after approval of antiprogrammed cell death protein 1 immune checkpoint inhibitors (ICIs) and has contributed to increased health care utilization and spending. METHODS: We investigated the change in prevalence of any systemic therapy use in the last 30 DOL among patients with advanced solid tumors in the 4 years before and after antiprogrammed cell death protein 1 ICI approval in 2014. We used cases from the Western Washington Cancer Surveillance System linked to commercial and Medicare insurance. We calculated the difference in prevalence between the pre- and post-ICI periods. We also calculated the annual prevalence of any systemic therapy and ICI use in the last 30 DOL and measured health care utilization (emergency department visits and hospitalizations) and costs during the last 30 DOL. RESULTS: Eight thousand eight hundred seventy-one patients (median age 73 years) were included; 34% and 66% in the pre-and post-ICI period, respectively. Systemic therapy use in the last 30 DOL was lower in the post-ICI versus pre-ICI period (12.4% v 14.4%; difference −2.0% [95% CI, −3.5 to −0.5]). The annual prevalence of systemic therapy use in the last 30 DOL also declined, although ICI use rose. Patients treated with ICIs in last 30 DOL had more emergency department visits, hospitalizations, and higher costs. CONCLUSION: Systemic therapy use in the last 30 DOL was lower in the period after ICI approval. However, ICI use rose over time and had higher utilization and costs in the last 30 DOL. Systemic therapy use in the last 30 DOL warrants monitoring, especially as more ICI indications are approved.

e13640 Background: Evidence-based, national guidelines for the diagnosis of recurrent metastatic breast cancer (MBC) recommend confirmation of recurrence with biopsy and reassessment of biomarker status. Real world practice patterns may demonstrate disparities in adherence to guidelines with implications for patients and health systems. Methods: We utilized the Hutchinson Institute for Cancer Outcomes Research (HICOR) data repository that links Washington State cancer registry data to enrollment and claims from the major insurance payers in the state. We identified women > 18 years old diagnosed with recurrent MBC between 2008 and 2017 with evidence of enrollment in a commercial plan (Premera or Regence), Medicare, or Medicaid. Recurrence in Stage I-III patients was detected through identification of ICD 9/10 codes for metastatic disease or resumption of breast cancer systemic therapy (after minimum of 4 months from completion of early breast cancer therapy). Using claims, we identified receipt of and factors associated with biopsy, biomarker re-assessment, and treatment administered at recurrence. Results: We identified 715 patients with recurrent MBC (any ER or HER2 status) with median age of 62 (range 52-73). The majority of the cohort were Caucasian (89%) with the rest comprising of Asian, Black, American Indian, and Hispanic patients. Based on the Rural Urban Commuting Area (RUCA) classification, patients predominantly lived in metropolitan neighborhoods (97%). Approximately 13% of patients lived in areas of high deprivation (area of deprivation index, ADI, 8-10). Patients had either Commercial (53.1%), Medicaid (4.2%), Medicare (29.7%) or multiple (13.0%) insurance. Patients were primarily treated at high volume centers (70.9%), though 23% of patients were seen at low volume centers (2) were more likely to undergo biopsy confirmation (20.3% vs 13.0%, p = 0.02). Biopsy was more often performed in patients receiving care at a high-volume center compared to low-volume center (74.3% vs 18.6%, p = 0.03). First line treatment selection was directly associated with receipt of biopsy and biomarker testing. Hormone therapy only was more common in patients who did not undergo biopsy (62.3% vs 37.7%, p

Background There is a need for automated approaches to incorporate information on cancer recurrence events into population-based cancer registries. Objective The aim of this study is to determine the accuracy of a novel data mining algorithm to extract information from linked registry and medical claims data on the occurrence and timing of second breast cancer events (SBCE). Methods We used supervised data from 3092 stage I and II breast cancer cases (with 394 recurrences), diagnosed between 1993 and 2006 inclusive, of patients at Kaiser Permanente Washington and cases in the Puget Sound Cancer Surveillance System. Our goal was to classify each month after primary treatment as pre- versus post-SBCE. The prediction feature set for a given month consisted of registry variables on disease and patient characteristics related to the primary breast cancer event, as well as features based on monthly counts of diagnosis and procedure codes for the current, prior, and future months. A month was classified as post-SBCE if the predicted probability exceeded a probability threshold (PT); the predicted time of the SBCE was taken to be the month of maximum increase in the predicted probability between adjacent months. Results The Kaplan-Meier net probability of SBCE was 0.25 at 14 years. The month-level receiver operating characteristic curve on test data (20% of the data set) had an area under the curve of 0.986. The person-level predictions (at a monthly PT of 0.5) had a sensitivity of 0.89, a specificity of 0.98, a positive predictive value of 0.85, and a negative predictive value of 0.98. The corresponding median difference between the observed and predicted months of recurrence was 0 and the mean difference was 0.04 months. Conclusions Data mining of medical claims holds promise for the streamlining of cancer registry operations to feasibly collect information about second breast cancer events.

Objectives Studies of local stage prostate cancer survivors suggest that treatments carry risk of persistent impotence, incontinence, and bowel dysfunction. To examine impacts of cancer type and side effects on health-related quality of life (HRQoL) in long-term cancer survivorship, we evaluated 5-year follow-up of patients with prostate cancer and compared results with a matched group of male long-term survivors of other local-stage cancers. Materials and Methods We examined genitourinary, bowel and sexual symptoms, and general quality of life. Matched survivors of colorectal, lung, and bladder cancers were recruited via registries in 3 different regions in the United States. Patients were surveyed 3–5 years after diagnosis with the SF-12 and EPIC to evaluate general mental and physical health-related quality of life (HRQoL) and patient function and bother. Results We analyzed responses from long-term prostate (n = 77) and bladder, colorectal, and lung cancer (n = 124) patients. In multivariate analysis, long-term local stage prostate cancer survivors had significantly higher SF-12 physical component scores but did not differ from long-term survivors of other cancers in terms of their SF-12 mental summary scores. Prostate survivors had similar mental, urinary, bowel, and sexual HRQoL compared to long-term survivors of other local stage cancers. Conclusion Long-term general and prostate-specific HRQoL was similar between local stage prostate and bladder, colorectal, and lung patients with cancer. Future research focusing on factors other than initial treatment and the cancer type per se may provide more meaningful information regarding factors that predict disparities on HRQoL among longer-term survivors of early stage male cancers.

Regional public reporting of performance measures in oncology can facilitate local decision making across stakeholders, but small numbers of patients and clinics pose a challenge to creating statistically robust measures. In this article, we describe our development of the Community Cancer Care in Washington State: Quality and Cost Report, the first publicly available report showing clinic-level quality and cost measures at the regional level. We learned key lessons in how to adapt national performance reporting to our regional setting using a registry-linked multipayer claims database. In short, limited numbers of eligible patients for some nationally recognized metrics led us to group metrics and use a 3-year performance window. After completing clinic attribution and other requirements of metric construction, the final metrics included between 62.9% and 88.4% of the eligible patients. To link total costs to some quality measures, we had to define a treatment and surveillance episode of care. Risk adjustment was challenged by the ability to include a limited number of risk adjustors and their potential concentration in a few clinics. We used a different quality score than national performance reporting to account for variation in the range of risk-standardized rates. Current methodology does not permit us to determine whether clinically meaningful differences in quality or costs exist, which inhibits value comparisons. Stakeholder engagement was critical for providing methodologic feedback. In conclusion, we found that refining national metrics was necessary to facilitate public reporting in a regional setting. Further methodologic development can strengthen public reporting and future applications.

BACKGROUND The American Society for Radiation Oncology (ASTRO) issued a Choosing Wisely recommendation (CWr) against use of intensity modulated radiotherapy (IMRT) as part of breast conservation therapy in 2013, noting that “its routine use has not been demonstrated to provide significant clinical advantage”. Also, as IMRT is more expensive vs conventional radiotherapy (cRT), use in this setting may represent wasteful spending. The aims of this study were to characterize: 1) IMRT use immediately preceding the ASTRO CWr, 2) the cost-consequences of IMRT vs. cRT in the year after breast cancer diagnosis, and 3) excess annual U.S. national expenditure on IMRT in this setting. METHODS Surveillance Epidemiology and End Results (SEER) records for women age ≥66 years with a first primary diagnosis of Stage I/II breast cancer (2007-2012) were linked with Medicare claims (2007-2013). All cases had coverage for ≥12-months before/after diagnosis (or until death), and to be consistent with the CWr, had lumpectomy 1 month before to 6 months after diagnosis, and radiation therapy within 6 months of lumpectomy. We evaluated receipt of IMRT vs cRT within 1 year of diagnosis in 12 SEER registries. We also calculated mean direct medical expenditure in the year after diagnosis by summing all allowable charges from Medicare claims. Costs were inflated to 2017 USD. We evaluated differences in means by registry using ANOVA and compared mean costs for cases treated with IMRT vs cRT using t-test. We projected excess annual U.S. national expenditure on IMRT (vs. cRT) using the findings from aims 1 & 2 and assuming 209000 incident Stage I/II cases and that 55% receive lumpectomy. RESULTS Among 13037 women meeting all inclusion criteria, mean age was 74, 89.0% were white, and 19.8% received IMRT. Table 1 shows the proportions receiving IMRT by SEER registry and mean cost for cases treated with IMRT vs cRT. The proportion with IMRT varied significantly between registries (p Table 1: Results for Selected RegistriesRegistry% with IMRTMean Cost in Year After Dx ($) IMRTcRTpAll 12 Registries19.84314533501 DISCUSSION From 2007-2013, use of IMRT in Stage I/II breast cancer was substantial (19.8%) and varied significantly across SEER registries. Cases treated with IMRT incurred significantly higher cost ($9644) in the year following diagnosis vs. those treated with cRT. If patterns of care remain similar today, there is potentially as much as $219 million in annual U.S. national expenditure on low-value radiotherapy following lumpectomy. Our findings suggest an opportunity to improve quality in cancer care while reducing expenditure by curbing use of IMRT. Future studies should develop interventions to align practice with ASTRO CW recommendations and reduce variation in practice between regions. Citation Format: Roth JA, McDougall J, Halasz L, Fedorenko C, Sun Q, Patel S. IMRT use after lumpectomy in early-Stage breast cancer: Patterns of care and cost-consequences [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-11-14.

145 Background: The COVID-19 pandemic disrupted medical care, including routine cancer screening for breast, colorectal, lung and cervical cancers. We aimed to investigate the impact of the pandemic on stage at diagnosis for cancer patients. Methods: Using data from the Washington State SEER records we compared AJCC stage for patients diagnosed with cancer in 2017-2019 to 2020 for two time periods, March to June (initial pandemic months) and July to December (later pandemic months). Patients were included if they were age 18+, diagnosed with a solid tumor, and not diagnosed at autopsy. Results: In the early phase of the pandemic, March – June 2020, there was a shift to cancers being diagnosed at a later stage compared to the same time period in 2017-2019 (Stage III: 13.5% to 14.9%, Stage IV: 16.2% to 19.7%). There was also a decrease in cancer diagnoses for cancers that are often detected through routine screening. As a percentage of all cancer diagnoses, both melanoma (13.2% to 9.8%) and colon cancer diagnoses (7.2% to. 6.7%) decreased during the early pandemic. In the later phase of the pandemic, July to December 2020, the stage at diagnosis showed an indication of returning to pre-pandemic levels with an increase in the proportion of early stage cancers (In situ: 16.6% to 19.3%, Stage I: 38.8% to 41.1%). Stage at diagnosis trends varied by tumor type. For colorectal cancer, the overall number of diagnoses decreased during the initial pandemic months. Stage I diagnoses decreased and Stage IV cancer diagnoses increased in both early and late stages of the pandemic. Conclusions: In Washington State, the COVID-19 pandemic had an impact on stage at diagnosis potentially caused by delays or interruptions in medical care. Additional studies are needed to understand how this shift in stage at diagnosis impacted treatment and outcomes for patients.

Purpose: Despite guidelines emphasizing symptom management over aggressive treatment, end-of-life care for persons with cancer in the United States is highly variable. In consultation with a regional collaboration of patients, providers, and payers, we investigated indicators of high-quality end-of-life care to describe patterns of care, identify areas for improvement, and inform future interventions to enhance end-of-life care for patients with cancer. Methods: We linked insurance claims to clinical information from the western Washington SEER database. We included persons ≥ 18 years of age who had been diagnosed with an invasive solid tumor between January 1, 2007, and December 31, 2015, and who had a recorded death date, were enrolled in a commercial plan for the last month of life, and made at least one insurance claim in the last 90 days of life. Results: In the last month of life, among 6,568 commercially insured patients, 56.3% were hospitalized and 48.6% underwent at least one imaging scan. Among patients younger than 65 years of age, 31.4% were enrolled in hospice; of those younger than 65 years of age who were not enrolled in hospice, 40.5% had received an opioid prescription. Over time, opioid use in the last 30 days of life among young adults not enrolled in hospice dropped from 44.7% in the period 2007 to 2009 to 42.5% in the period 2010 to 2012 and to 36.7% in the period 2013 to 2015. Conclusion: Hospitalization and high-cost imaging scans are burdensome to patients and caregivers at the end of life. Our findings suggest that policies that facilitate appropriate imaging, opioid, and hospice use and that encourage supportive care may improve end-of-life care and quality of life.

Information sources about prostate cancer treatment and outcomes are typically designed for patients. Little is known about the availability and utility of information for partners. The objectives of our study were to evaluate information sources used by partners to understand prostate cancer management options, their perceived usefulness, and the relationship between sources used and satisfaction with treatment experience. A longitudinal survey of female partners of men newly diagnosed with local-stage prostate cancer was conducted in three different geographic regions. Partners and associated patients were surveyed at baseline (after patient diagnosis but prior to receiving therapy) and at 12 months following diagnosis. Information sources included provider, literature, friends or family members, Internet websites, books, traditional media, and support groups. Utility of an information source was defined as whether the partner would recommend it to caregivers of other patients with local-stage prostate cancer. Our study cohort included 179 partner-patient pairs. At diagnosis, partners consulted an average of 4.6 information sources. Non-Hispanic white partners were more likely than others to use friends and family as an information source (OR = 2.44, 95% CI (1.04, 5.56)). More educated partners were less likely to use support groups (OR = 0.31, 95% CI (0.14, 0.71)). At 12-month follow-up, partners were less likely to recommend books (OR = 0.23, 95% CI (0.11, 0.49)) compared to baseline. Partners consulted a large number of information sources in researching treatment options for local-stage prostate cancer and the types of sources accessed varied by race/ethnicity and educational attainment. Additional resources to promote selection of high-quality non-provider information sources are warranted to enable partners to better aid patients in their treatment decision-making process.

Purpose In 2013, the American Society for Radiation Oncology (ASTRO) issued a Choosing Wisely recommendation against the routine use of intensity modulated radiotherapy (IMRT) for whole breast irradiation. We evaluated IMRT use and subsequent impact on Medicare expenditure in the period immediately preceding this recommendation to provide a baseline measure of IMRT use and associated cost consequences. Methods and materials SEER records for women ≥66 years with first primary diagnosis of Stage I/II breast cancer (2008–2011) were linked with Medicare claims (2007–2012). Eligibility criteria included lumpectomy within 6 months of diagnosis and radiotherapy within 6 months of lumpectomy. We evaluated IMRT versus conventional radiotherapy (cRT) use overall and by SEER registry (12 sites). We used generalized estimating equations logit models to explore adjusted odds ratios (OR) for associations between clinical, sociodemographic, and health services characteristics and IMRT use. Mean costs were calculated from Medicare allowable costs in the year after diagnosis. Results Among 13,037 women, mean age was 74.4, 50.5% had left-sided breast cancer, and 19.8% received IMRT. IMRT use varied from 0% to 52% across SEER registries. In multivariable analysis, left-sided breast cancer (OR 1.75), living in a big metropolitan area (OR 2.39), living in a census tract with ≤$90,000 median income (OR 1.75), neutral or favorable local coverage determination (OR 3.86, 1.72, respectively), and free-standing treatment facility (OR 3.49) were associated with receipt of IMRT (p

Background: The purpose of this study was to assess advanced imaging (bone scan, CT, or PET/CT) and serum tumor biomarker use in asymptomatic breast cancer survivors during the surveillance period. Patients and Methods: Cancer registry records for 2,923 women diagnosed with primary breast cancer in Washington State between January 1, 2007, and December 31, 2014, were linked with claims data from 2 regional commercial insurance plans. Clinical data including demographic and tumor characteristics were collected. Evaluation and management codes from claims data were used to determine advanced imaging and serum tumor biomarker testing during the peridiagnostic and surveillance phases of care. Multivariable logistic regression models were used to identify clinical factors and patterns of peridiagnostic imaging and biomarker testing associated with surveillance advanced imaging. Results: Of 2,923 eligible women, 16.5% (n=480) underwent surveillance advanced imaging and 31.8% (n=930) received surveillance serum tumor biomarker testing. Compared with women diagnosed before the launch of the Choosing Wisely campaign in 2012, later diagnosis was associated with lower use of surveillance advanced imaging (odds ratio [OR], 0.68; 95% CI, 0.52–0.89). Factors significantly associated with use of surveillance advanced imaging included increasing disease stage (stage III: OR, 3.65; 95% CI, 2.48–5.38), peridiagnostic advanced imaging use (OR, 1.76; 95% CI, 1.33–2.31), and peridiagnostic serum tumor biomarker testing (OR, 1.35; 95% CI, 1.01–1.80). Conclusions: Although use of surveillance advanced imaging in asymptomatic breast cancer survivors has declined since the launch of the Choosing Wisely campaign, frequent use of surveillance serum tumor biomarker testing remains prevalent, representing a potential target for further efforts to reduce low-value practices.

e13038 Background: Evidence-based, national guidelines for the management of metastatic breast cancer (MBC) recommend numerous treatment options that do not capture the nuances of real-world practice. Disparities may exist across Washington State with financial implications for patients and health systems. The objective of this study was to assess practice patterns around treatment of ER+/HER2- MBC in actual clinical practice. Methods: We collaborated with Hutchinson Institute for Cancer Outcomes Research (HICOR) to link enrollment and insurance claims records with Washington State cancer registries from 2008-2017. Our cohort comprised of women >18 years old with de novo ER+/HER2- MBC who met enrollment criteria in one of four payors (Premera, Regence, Medicare, or Medicaid). We identified receipt of first line treatment, categorized as CDK4/6 inhibitors plus endocrine therapy (CDKi+ET), chemotherapy (CT), or endocrine therapy alone (ET). We examined factors influencing treatment selection using Fisher's and Kruskal-Wallis tests. Total costs (defined as costs from inpatient and outpatient claims one year after diagnosis) was estimated for patients and payors. Results: We identified 140 patients with de novo ER+/HER2- MBC with median age of 64 (range 28-95). The majority of the cohort were Caucasian (90%) with the rest comprising of Asian, Black, American Indian, and Hispanic patients. Based on the Rural Urban Commuting Area (RUCA) classification, patients predominantly lived in metropolitan neighborhoods (96%). Over 20% of patients lived in areas of high deprivation (area of deprivation index, ADI, 8-10). Patients had either Commercial (40.7%), Medicaid/Medicare (43.6%) or multiple (15.7%) insurance. Our data show that 17 patients (12%) received first line therapy with CDKi + ET, 64 patients (46%) with CT, and 59 patients (42%) with ET alone. Factors influencing treatment selection include age, co-morbidity score, and payor type. Older patients (>65 years old) were more likely to receive ET alone compared to younger patients (56% vs 44%, p value

e18590 Background: The use of platinum agents (PT) in neoadjuvant chemotherapy (CTX) for triple negative breast cancer (TNBC) has been debated in research and practice guidelines. In high risk patients, the addition of PT to standard anthracycline and taxane regimens improves pathological complete response without definitive evidence of survival benefit (Korde JCO 2021). Due to PT’s toxicities and unclear benefit in early stage breast cancer, this study aimed to elucidate the real world practice patterns of PT use and overuse in breast cancer patients. Methods: The cohort was defined as women with stage I-III TNBC (ER, PR, and Her2 negative) who received curative intent surgery between 2011-2018. We utilized the Hutchinson Institute for Cancer Outcomes Research (HICOR) database, which links enrollment and claims records from commercial and public health insurance plans with clinical information from Washington state cancer registries (Manohar JACR 2020). We hypothesized there was overuse of PT CTX in this cohort. Overuse was defined as PT use in stage I, in the adjuvant period (NCCN 2020), and in nonstandard CTX regimens (PT use without taxane and anthracycline (NCCN 2020)). Results: Of the 910 women in the cohort, average age was 63, 90% were white, 682 (75%) received CTX, and 36% had commercial insurance, versus 53% with public insurance (Medicare 45%, Medicaid 8%). Of those receiving CTX, 39% received neoadjuvant and 67% adjuvant. Of those receiving CTX, 85 (13%) received PT, and 28 (4%) received PT without anthracycline. Of those receiving PT, 20% had stage I disease, and of those receiving adjuvant CTX, 43 (9%) received PT in the adjuvant period. Conclusions: We found there was overuse and nonstandard use of PT in this real world data. This overuse was demonstrated by the fact that 20% of women receiving PT had low risk stage I disease, and 9% of women receiving adjuvant CTX received PT in the adjuvant period. Additionally, there was nonstandard use, with 4% of women getting CTX receiving PT without anthracycline. Next steps in this research are to evaluate factors influencing overuse and nonstandard use, as well as complication rates with PT use and overuse.[Table: see text]

6546 Background: Washington was the first US state to experience the COVID-19 pandemic. Transmission risks and patient fears of visiting oncology practices during its onset resulted in rapid adoption of telehealth services. We hypothesized that the pandemic would widen disparities in oncology practice visits between Medicaid and commercially insured patients, resulting higher rates of emergency department (ED) visits during initial treatment. Methods: Linking Washington State SEER records with Medicaid and commercial insurance enrollment and claims records, we compared adults age

6504 Background: Increasing evidence shows that cancer patients (pts) experience financial hardships after diagnosis. Few studies, however, have used objective financial data to estimate the relative risk of adverse financial events (AFEs) in cancer pts versus individuals without cancer. Using a retrospective case-control design, we investigated whether cancer pts are at increased risk of new AFEs, as measured by their credit reports. Methods: Western Washington Surveillance Epidemiology and End Results (SEER) cancer registry (cases) and voter registry (controls) records from 2013 to 2018 were linked to quarterly credit records from TransUnion (2012-2020), one of the 3 largest national credit agencies. Controls were age and sex matched to cases and assigned an index date corresponding to the diagnosis (dx) date of the matched case. Individuals with evidence of any AFE in the credit report closest to index/dx date or did not survive to 24 months were excluded. Cases and controls experiencing any of the following AFEs within 24 months were compared, using two-sample z tests: severe (3rd party collections, charge-offs), more severe (tax liens, delinquent mortgage payments), and most severe (foreclosures, repossessions). Multivariate logistic regression models were used to evaluate the association between cancer dx and AFE, adjusting for age, sex, dx year, and available credit 6 months before the index/dx date. Results: A total of 332,825 individuals (84,185 cases and 248,640 controls, mean age 66 (SD 13), 52.7% female) were included. The mean available line of credit in the year before index/dx date was $12,303. AFEs were more common in cases versus controls (Table). After adjusting for age, sex, available credit above or below $12,303, and dx year, cancer dx was significantly associated with any AFE (OR 1.77, 95% CI 1.7-1.85, p65 and higher available baseline credit were associated with decreased risk of any and each category of AFE. Conclusions: Within 24 months from dx, significantly higher proportions of cancer pts experienced AFEs relative to controls. Such events on credit reports have serious and long-lasting consequences on financial status. Studies that link clinical and financial data to investigate the impacts of these events on treatment decisions, quality of life, and clinical outcomes are needed.[Table: see text]

6502 Background: The COVID-19 pandemic dramatically reduced family access to hospitals and created new barriers to home hospice care, raising concerns about how the pandemic has impacted cancer patients’ place of death and end of life home hospice support. Hypothesizing that Medicaid-enrolled cancer patients may be at greater risk of disruptions in end-of-life care compared to commercially insured patients, we examined changes in place of death and home hospice support for Medicaid and Commercial enrollees following the pandemic. Methods: We linked WA State cancer registry records with claims from Medicaid and approximately 75% of commercially insured cancer patients in the state. Patients ages 18-64 with solid-tumor malignancies who died March-June 2020 (COVID) were compared to those who died March-June 2017-2019 (Pre-COVID). Place of death was categorized as hospital, home with hospice, and home without hospice; nursing home deaths were excluded. Given our sample size, we examined differences in the likelihood of place of death with Fisher’s exact tests and multinomial logistic regressions stratified by payer and by COVID period, controlling for age, gender, race, stage, cancer type, and census tract-level neighborhood deprivation. We report marginal effects. Results: In Fisher’s exact analyses, Medicaid but not commercial patients were significantly less like to die in hospital and more likely to die at home without hospice during COVID (Table). In pre-post adjusted analysis of Medicaid patients, the probability of dying in the hospital was 12.3% (p=0.03) percentage points lower during the pandemic versus before, while the probability of dying at home without hospice was 11.1% (p=0.04) greater. Place of death did not change significantly pre-post for commercial patients. In addition, Pre-COVID, the probability of dying in the hospital was 10.7% (p=0.03) greater for Medicaid than commercial patients. During COVID, the probability of dying at home without hospice was 15.8% (p=0.04) greater for Medicaid versus commercial patients but lower for women (ME=20.2%; p=0.01) and colorectal versus breast cancer patients (ME=39.2%; p=0.01). Conclusions: Following COVID, Medicaid patients place of death shifted from hospital to homes, but without an increase in the use of home hospice services. In contrast, place of death and hospice use among commercial patients did not significantly change. This widening disparity in home deaths without hospice services raises concerns that the pandemic disproportionately worsened end of life experience for low income patients with cancer.[Table: see text]

There is a need for automated approaches to incorporate information on cancer recurrence events into population-based cancer registries.The aim of this study is to determine the accuracy of a novel data mining algorithm to extract information from linked registry and medical claims data on the occurrence and timing of second breast cancer events (SBCE).We used supervised data from 3092 stage I and II breast cancer cases (with 394 recurrences), diagnosed between 1993 and 2006 inclusive, of patients at Kaiser Permanente Washington and cases in the Puget Sound Cancer Surveillance System. Our goal was to classify each month after primary treatment as pre- versus post-SBCE. The prediction feature set for a given month consisted of registry variables on disease and patient characteristics related to the primary breast cancer event, as well as features based on monthly counts of diagnosis and procedure codes for the current, prior, and future months. A month was classified as post-SBCE if the predicted probability exceeded a probability threshold (PT); the predicted time of the SBCE was taken to be the month of maximum increase in the predicted probability between adjacent months.The Kaplan-Meier net probability of SBCE was 0.25 at 14 years. The month-level receiver operating characteristic curve on test data (20% of the data set) had an area under the curve of 0.986. The person-level predictions (at a monthly PT of 0.5) had a sensitivity of 0.89, a specificity of 0.98, a positive predictive value of 0.85, and a negative predictive value of 0.98. The corresponding median difference between the observed and predicted months of recurrence was 0 and the mean difference was 0.04 months.Data mining of medical claims holds promise for the streamlining of cancer registry operations to feasibly collect information about second breast cancer events.

Purpose: As new quality metrics and interventions for potentially preventable emergency department (ED) visits are implemented, we sought to compare methods for evaluating the prevalence and costs of potentially preventable ED visits that were related to cancer and chronic disease among a commercially insured oncology population in the year after treatment initiation. Methods: We linked SEER records in western Washington from 2011 to 2016 with claims from two commercial insurers. The study included patients who were diagnosed with a solid tumor and tracked ED utilization for 1 year after the start of chemotherapy or radiation. Cancer symptoms from the Centers for Medicare & Medicaid Services metric and a patient-reported outcome intervention were labeled potentially preventable (PpCancer). Prevention Quality Indicators of the Agency for Healthcare Research and Quality were labeled potentially preventable–chronic disease (PpChronic). We reported the primary diagnosis, all diagnosis field coding (1 to 10), and 2016 adjusted reimbursements. Results: Of 5,853 eligible patients, 27% had at least one ED visit, which yielded 2,400 total visits. Using primary diagnosis coding, 49.8% of ED visits had a PpCancer diagnosis, whereas 3.2% had a PpChronic diagnosis. Considering all diagnosis fields, 45.0%, 9.4%, and 18.5% included a PpCancer only, a PpChronic only, and both a PpCancer and a PpChronic diagnosis, respectively. The median reimbursement per visit was $735 (interquartile ratio, $194 to $1,549). Conclusion: The prevalence of potentially preventable ED visits was generally high, but varied depending on the diagnosis code fields and the group of codes considered. Future research is needed to understand the complex landscape of potentially preventable ED visits and measures to improve value in cancer care delivery.

The American Society of Clinical Oncology (ASCO)/American Board of Internal Medicine (ABIM) Choosing Wisely (CW) measures aim to reduce the use of interventions that lack evidence of benefit in cancer care. The study presented here characterized adherence to the 2012 ASCO/ABIM CW recommendations by linking health plan claims data with a regional cancer registry and sought to identify areas for research interventions to improve adherence.SEER records for patients diagnosed with cancer in Western Washington State between 2007 and 2014 were linked with enrollment and claims from a large regional commercial insurance plan. Using claims and SEER records, algorithms were developed to characterize adherence to each CW measure. In addition, we calculated differences in total reimbursements and procedure-specific reimbursements for patients receiving adherent and nonadherent care.A total of 22,359 unique individuals with cancer were linked with insurance enrollment records and met basic eligibility criteria. Overall adherence varied from 53% (breast surveillance) to 78% (breast staging). Within each measure, adherence varied substantially by stage at diagnosis and by cancer site in situations in which the CW measure affected multiple types of cancer. The difference in reimbursements between adherent and nonadherent populations across all five measures was approximately $29 million.Adherence to the ASCO/ABIM CW measures varies widely, as does the cost implication of nonadherence. A structured approach to evaluating adherence and cost impact is needed before developing programs aimed at improving adherence to the ASCO/ABIM CW measures.

3 Background: Patients with EGFR+ or ALK+ NSCLC benefit from oral TKIs, but high patient OOP TKI costs could negatively impact survival by reducing likelihood of continuing TKI therapy. We assessed the association of high OOP TKI costs with overall survival (OS), medication possession ratio (MPR) and duration of TKI therapy (DOT) in patients with metastatic EGFR+ and ALK+ NSCLC. Methods: We identified patients with EGFR+ and ALK+ NSCLC diagnosed between 01/01/2010 and 12/31/2016 in the Washington State SEER registry using natural language processing, followed by manual confirmation of molecular tests. We linked registry records to commercial and Medicare (including part D) claims. Eligible patients had stage IV NSCLC, sensitizing EGFR mutations or ALK+ by FISH, ≥1 pharmacy claims for EGFR or ALK TKIs, ≥3 months survival from TKI start, and ≥12 months of insurance enrollment post diagnosis. We estimated OOP TKI costs by subtracting the amount paid from the amount allowed in pharmacy claims. We categorized patients by quartiles of monthly OOP costs calculated for the first 3 months of TKI therapy (Q1 < Q2 < Q3 < Q4). We used landmark survival analysis with multivariate Cox regression to test the association of monthly OOP costs in the first 3 months of TKIs and OS starting at 3 months of TKI therapy between Q1-Q3 vs Q4. We used t-tests to compare MPR and DOT for the 1st TKI between Q1-3 and Q4. Results: For 106 eligible patients (median age 69; 67% female; 73% White; 35% Medicare, 85% EGFR+), the median monthly OOP TKI costs at 3 months were $0 (Q1); $1,432 (Q2); $1,798 (Q3); $2,888 (Q4). Mean MPR was 1.20 vs. 1.06 (P = 0.02), and median DOT was 8 vs. 4 months (P < 0.01) for Q1-3 vs. Q4. Median OS was 22 vs. 9 months for Q1-3 vs 4. Compared with Q1-3, Q4 patients had a hazard ratio for death of 1.76 (95% CI = 1.05; 2.94; P = 0.03), adjusted for age, sex, insurance, mutation, income, chemotherapy use, and time to TKI start. There were no statistically significant associations comparing Q1 to Q2-4, or Q1-2 to Q3-4. Conclusions: Higher patient OOP TKI costs are associated with lower number of TKI prescriptions, shorter duration of TKI therapy and inferior survival in advanced EGFR+ and ALK+ NSCLC.

270 Background: Approximately half of cancer patients undergoing outpatient chemotherapy experience unplanned Emergency Department (ED) visits and Inpatient (IP) stays. Current machine learning algorithms that identify high-risk patients are based on pre-treatment variables which can not detect changes in risk over time. Deep learning recurrent neural networks can model complex longitudinal patient histories. This study tests the feasibility of using an interpretable recurrent neural network to predict a patient’s daily likelihood of ED and unplanned IP stays in the six months following chemotherapy initiation. Methods: Medicare and commercial claims data were linked with cancer registry records for patients in Washington State from 2011 to 2017. The study included patients diagnosed with any primary tumor site, excluding leukemia, and treated with chemotherapy. We used the Reverse Time Attention model (RETAIN) with a 1:10 case-control match and included registry elements; diagnoses, procedures, medication, and utilization pre-and post-chemotherapy initiation. Patients were randomly divided into internal training, validation, and test sets (75%, 10%, 15%). Model accuracy was measured by the areas under the receiver operating curve (ROC) and precision-recall curve (PRC), and the Youden sensitivity and specificity. Results: Of the 15,400 eligible patients; 4,037 (26.2%) visited the ED a median of 1 time (6,080 total visits); 5,116 (33.2%) had a median of 1 IP stay (7,839 total stays). Both models had good predictive accuracy: The top 20 predictors for ED visits included 5 chemotherapy regimes, 12 procedures, and 2 tumor characteristics; IP stays included all chemotherapy regimes. Conclusions: The promising performance of RETAIN supports the possibility of building a tool capable of estimating daily hospitalization risk. However, future research, particularly with alternative data sources, may be required to predict hospitalization in a real time clinical setting. [Table: see text]

134 Background: Molecular testing practices and cost-sharing policies may result in delayed initiation of TKI therapy. We assessed predictors of delayed initiation of TKIs in metastatic EGFR+ or ALK+ NSCLC. Methods: We identified patients with EGFR+ and ALK+ NSCLC diagnosed between 01/01/2010 and 12/31/2016 in the Washington State SEER registry using validated natural language processing methods. We linked registry records to commercial and Medicare (including part D) claims. Eligible patients had stage IV NSCLC, sensitizing EGFR mutations or ALK+ by FISH, ≥ 1 pharmacy claims for EGFR or ALK TKIs, and ≥12 months of insurance enrollment post-diagnosis. Potential predictors included age, sex, race, Census-level median household income, urban status, insurance type, comorbidity, histology, mutation type, and receipt of chemotherapy prior to first TKI claim (pre-TKI chemo). We defined time to TKI initiation as the interval from diagnosis to first pharmacy claim for EGFR or ALK TKIs. We fitted Cox regression models to identify predictors of delays in TKI initiation, defining covariates with a P < 0.05 in a final multivariate model as independently associated with delays. Results: For 122 patients (median age 70; 65% female; 74% White; median income $66,580; 98% metropolitan; 35% Medicare; 80% EGFR+; 12% using pre-TKI chemo), the median time to TKI initiation was 6.7 weeks (IQR = 3.9 to 14.0). Independent predictors of TKI delays included male sex (HR = 0.51; 95%CI = 0.34; 0.76); Medicare insurance (HR = 0.32; 95% CI = 0.20; 0.53) and pre-TKI chemo (HR = 0.37; 95% CI = 0.20; 0.66). Median time to TKI initiation was 9.7 vs. 5.8; 7.8 vs. 4.1; and 16.0 vs. 6.3 weeks in male vs. female, Medicare vs. commercial insurance, and pre-TKI chemo (yes vs no), respectively. Conclusions: Male sex, Medicare insurance, and chemotherapy prior to TKI are associated with delays in TKI initiation for EGFR+ and ALK+ stage IV NSCLC patients. Possible explanations include higher prevalence of smoking in males resulting in lower priority for molecular testing, high cost-sharing policies for TKIs in Medicare patients, and prolonged time to obtain molecular test results leading patients to start chemotherapy first.

166 Background: Rural residents are diagnosed at later stages of cancer compared to urban residents, have poorer survival, and face distinct barriers to receiving quality cancer care. ASCO has developed policy initiatives to address rural cancer care; however, little is known about quality of cancer care among patients residing in rural areas. This study examined the impact of rurality on performance metrics, controlling for socioeconomic status and insurance type. Methods: We linked Washington state cancer registry records from 2015-2017 with claims records for two large commercial insurers, Medicare, and Medicaid. Using claims from this database, we generated eight nationally recognized quality measures. Rurality was measured by the Rural-Urban Commuting Area Codes (RUCAs) categorized into 4 levels (Metro, Metro with commute, Micropolitan, Small Town/Rural). Process and outcome measures were adjusted for age, sex, race, comorbidity score, stage, cancer type, marital status, the Area Deprivation Index, and treatment factors where appropriate. Results were stratified by payer type. Results: The table below lists the effect of a patient’s rurality on the quality metric where significant (p

158 Background: Evidence that rural residents compared to urban residents are more likely to be diagnosed at later stages and have worse survival outcomes has prompted recent policy initiatives by ASCO to address the rural cancer care gap. However, rural residents are generally poorer, potentially confounding the cause of these disparities. This study examined the impact of rurality, travel time to oncologist (TTO), and neighborhood deprivation (ND) on stage of diagnosis and 3-year survival in a regional setting. Methods: Cancer registry records for patients in Western Washington were linked with claims from regional commercial insurers, Medicare, and Medicaid at time of diagnosis. The study included adult patients with solid tumors diagnosed between 2012-2014. Rurality was sourced from the Rural-Urban Commuting Area Codes (RUCAs): Metro, Metro with Commute, Micropolitan, Small Town/Rural. ND was measured by the census block Area Deprivation Index from 1 (least) to 10 (most) deprived; TTO from Google Maps (minutes). Multinomial logistic regression measured stage at diagnosis (local, regional, distant). Cox survival models were stratified by insurance type. We adjusted all models for age, gender, race, marital status, cancer type and grade, and hierarchical condition categories. Results: The table below lists the effect size of rurality, TTO, and ND on the outcomes where significant (p

Little is known about how referrals to different cancer specialists influence cancer care for non-small-cell lung cancer (NSCLC). Among Medicare enrollees, we identified factors of patients and their primary care physician that were associated with referrals to cancer specialists, and how the types of cancer specialists seen correlated with delivery of guideline-based therapies (GBTs).Data from patients with stages III and IV NSCLC included in the SEER-Medicare database were linked to their physicians in the American Medical Association Masterfile database. Using logistic regression, we (1) identified patient and physician factors that were associated with referrals to cancer specialists (medical oncologists, radiation oncologists, and surgeons); (2) identified the types of referral to cancer specialists that predicted greater likelihood of receiving GBT (per National Comprehensive Cancer Network guidelines).A total of 28,977 patients with NSCLC diagnosed from January 1, 2000 to December 31, 2005 met eligibility criteria. Younger age, white race, higher income, and primary physician specialty other than family practice predicted higher likelihood of referrals to medical oncologists (P.01 for all predictors). Seeing the three types of cancer specialists predicted higher likelihood of GBT (stage IIIA: odds ratio [OR] = 20.6; P.001; IIIB: OR = 77.2; P.001; and IV: OR = 1.2; P = .011), compared with seeing a medical oncologist only. Use of GBTs increased over the study period (42% to 48% from 2000 to 2005; P.001).Referrals to all types of cancer specialists increased the likelihood of treatment with standard therapies, particularly in stage III patients. However, racial and income disparities still prevent optimal referrals to cancer specialists.

65 Background: Polypharmacy, or the concurrent use of multiple medications, may expose patients to drug-drug interactions and excessive costs. There are limited data on polypharmacy for commercially insured older adults, primarily Medicare Advantage patients, that may have better access to medication management services than Medicare fee-for-service patients. We characterized medication use and out-of-pocket (OOP) medication costs in the last month of life among patients age 65+ who did not enroll in hospice to examine medication use in this population. Methods: We linked enrollment and claims records from two regional commercial insurers to Surveillance, Epidemiology, and End Results (SEER) Cancer Surveillance System and Washington State Cancer Registry records for patients diagnosed with a stage IV malignancy in Washington State between January 1, 2007-December 31, 2016. We calculated OOP costs as the difference between allowed and paid claim amounts and adjusted OOP costs (aOOP) for inflation to 2017 dollars. Results: Among 345 patients with medication claims in their last month, 156 (45%) had a chemotherapy claim. Average age was 74 years (range 65-95), 55% (n = 190) were male, and 150 had lung cancer (44%). Patients averaged 7.7 medication claims (range 1-50); 151 (44%) had 1-4, 99 (29%) had 5-9, and 95 (27%) had 10+. Common symptom-related medications were opioids, benzodiazepines, anti-emetics. Chemotherapy was associated with higher odds of 10+ prescriptions (OR 1.38, 95% CI 1.26-1.51). Excluding four patients with aOOP chemotherapy costs > $14,000, average aOOP costs were $101 for chemotherapy claims (range $0-$8957) and $33 for non-chemotherapy claims ($0-$1993). Costs for those in the highest quartile ranged from $320-$8957 for chemotherapy claims and $100-1993 for non-chemotherapy claims. Conclusions: Most subjects had at least five medication claims in their last 30 days. One-quarter had 10+ claims, which was associated with chemotherapy receipt. Hospice enrollment could reduce OOP costs, as hospice provides symptom-related medications. Interventions facilitating hospice enrollment and reducing chemotherapy use may minimize polypharmacy and cost burden for patients and families.

We surveyed prostate cancer patients about complementary and alternative medicine (CAM) use and evaluated patient factors that correlated with CAM use 6 months following diagnosis. The Prostate CAncer Therapy Selection study was a prospective, observational multi-site study of men’s treatment decision-making process after a diagnosis of local stage prostate cancer. Recruitment occurred in community urology practices in Washington State, hospital-based urology clinics affiliated with the University of Southern California, and Kaiser Permanente in Northern California. Eligible study participants included men over age 21 diagnosed with local stage prostate cancer between May 1, 2005 and December 31, 2006. Fifty-two percent of survey respondents (379) reported using one or more types of CAM. Of the patients, 51% used one CAM method, 26% used two methods, and 23% used three or more methods. The most commonly reported category was mind–body therapies (65%). Only 43% of patients discussed their CAM use with a health professional; of those, 20% informed their primary care physician and 30% told the doctor managing their prostate cancer care. Less than half thought the CAM they used was “very helpful”, but a majority thought it was somewhat helpful for their condition. Further research is needed to characterize the goals prostate cancer patients have for CAM, whether the treatments met those goals, and how this translates into the perceived helpfulness of these therapies. The implications of patients not discussing CAM use with health professionals at the time of prostate cancer treatment need further studies.

105 Background: While many quality metrics are generated and reported at a national level, regionally reported metrics may be more meaningful to stakeholders and more amenable to locally-generated interventions aimed at improving quality. We present the lessons learned from adapting national performance reporting standards to our regional setting. Methods: We linked 2014-2016 Washington State cancer registry records with Medicare and commercial insurance claims, capturing about 70% of the state’s cancer patients. A consortium of payers, oncology providers, and patients selected metrics from national initiatives (QOPI, MACRA, OCM, and the Choosing Wisely Campaign). We then searched the National Quality Forum and the National Quality Measures Clearinghouse for published specifications on those metrics. If none were available or there was a lack of consensus, we constructed algorithms with clinical and technical expert review. We calculated clinic-level risk standardized rates (RSRs) using hierarchical regression modeling to adjust for variation in clinic size and constructed quality composites. Results: We generated 13 quality metrics for 36,900 cancer patients cared for at 27 clinics. Our adaptation of national performance measures and implications include: 1) Many treatment process measures had too few numbers to individually report annually requiring us to group quality metrics and cancer types, and report over a three year period. 2) Although we were able to include tumor characteristics (e.g. AJCC stage) from registry records, risk adjustment was challenged by limitations in the number of risk adjustors due to the smaller number of patients. 3) After applying the hierarchical models, risk adjustment had minimal effect on clinic rankings. 4) The small number of clinics and limited range of clinic RSRs were more accurately captured by a quality composite that reflected clinic level differences to the regional average rather than national quintile groupings. Conclusions: Refinement of national metrics is necessary for public reporting in a regional setting. Further methodological development is critical for robust reporting and applications to value based purchasing.

31 Background: Although many quality measures exist in oncology, there have been few efforts to prioritize, measure, and report quality and costs for an entire region. Here, we report the results of a multi-year, multi-stakeholder effort aimed at public reporting of nationally recognized quality metrics for oncology practices in Washington State. Methods: Quality metrics were selected from nationally-recognized measures through a structured process involving oncology providers, health insurance leaders, patient advocates, and policy experts. Cancer registry records from 2014 to 2016 were linked with claims data from two commercial insurers and Medicare, representing approximately 70% of cancer patients in the state. Patients were assigned to oncology clinics using claims data; 27 clinics were large enough for inclusion in at least one measure in the report. Thirteen metrics were combined to produce one clinic-level quality score per measure. Each quality score was mapped to total episode cost per patient. Results were adjusted to account for clinic size, cancer characteristics (e.g. stage), demographics, and comorbidity where appropriate. Results: Conclusions: The Community Cancer Care in Washington State: Quality and Cost Report 2018 is the first publicly accessible report showing clinic-level quality measures linked to cost in oncology. Its ultimate goal is to improve care and lower costs by spurring collaboration, research and innovation.[Table: see text]

104 Background: As quality reports are released for regions and individual clinics, an important question is whether those reports can be used for value based purchasing, particularly for community oncology clinics. We evaluated the reliability of select quality measures and estimated the likelihood that a clinic’s performance would be incorrectly categorized (misclassified) in the top quartile. Methods: We linked 2014-2016 cancer registry records for patients with enrollment and claims from Medicare and two major commercial insurers in Washington State. We calculated risk standardized rates (RSRs) for ED and hospital use during treatment and 3 quality measures for end of life care. Reliability (0-1 scale: 0-unreliabile, > 0.7 good reliability, 1 perfectly reliable) was calculated as signal/(signal + statistical noise) from hierarchical logistic regression modeling for each metric. Misclassification was characterized as the probability of false negative and false positive assignment of clinics to the top quartile of performers in the region. We generated results for 3 and 1 year performance periods. Results: Over the 3 year period, the hospitalization metrics included 7,373 patients, 25 clinics; end of life metrics included 8,165 patients; 24 clinics. Conclusions: Although these metrics had fairly high levels of reliability, approximately one-third of clinics could be incorrectly identified as a top quartile performer. Use of these metrics in value based purchasing should account for potential misclassification to minimize unintended consequences.[Table: see text]

75 Background: Uniformity in receipt of high quality cancer care is imperative to reduce health care disparities. In order to help prioritize efforts aimed at reducing disparities in care, we compared several quality metrics for Washington State cancer patients enrolled in Medicaid and commercial insurance plans. Methods: We linked 2014-2016 Washington state cancer registry records for cancer patients under the age of 65 with enrollment and claims records for the two largest commercial insurers in the state and Medicaid. We then generated thirteen nationally recognized quality measures. Outcome measures were adjusted for age, sex, comorbidity score, stage, cancer site, and treatment factors where appropriate. Process measures were not adjusted. Results: 6,868 commercially insured and 2,379 Medicaid patients are represented in the reported quality measures. Conclusions: Care of Washington state cancer patients enrolled in Medicaid is comparable or superior to commercially insured patients, except for significantly higher ED use during chemotherapy. Further research is needed to understand why Medicaid-enrolled cancer patients utilize the ED at more than double the rate of commercially insured patients.[Table: see text]

823 Background: Few studies have examined how patterns of care relate to clinical consensus guidelines for metastatic colorectal cancer (mCRC) in real-world settings. This study investigated treatment patterns around systemic therapies for older mCRC patients using SEER-Medicare data. Methods: We utilized data from a linkage of national Medicare claims data with records from the 12 Surveillance, Epidemiology, and End Results (SEER) regions for patients diagnosed with stage IV CRC between 2011 and 2014. Other inclusion criteria included adenocarcinoma histology, being > 65 years of age, and continuous Part A/B enrollment 12 months prior to diagnosis. We conducted bivariate analysis and multivariate modeling to evaluate the association between age, race, gender, SEER region, and comorbidity score with receipt of systemic therapy. We examined first and subsequent lines of therapy, and classified them as adherent or non-adherent to NCCN guidelines. Results: Of 102,461 CRC patients, 6,360 (6.2%) patients had evidence of metastatic disease and met all inclusion criteria. 3,155 (49.6%) received no systemic therapy. Of the 2,976 (46.8%) who received systemic therapy with identifiable regimens, FOLFOX/CAPOX was the most common 1L regimen (32.9%), followed by 5-FU or capecitabine alone (26.4%) – both identified as guideline-adherent therapies. Non-guideline adherent regimens (such as use of single-agent oxaliplatin or bevacizumab) were uncommon as 1L therapies (1.4%). 1,553 patients (52.2%) progressed to a 2L therapy, 715 (24.0%) progressed to a 3L therapy, and 320 (10.8%) to a 4L therapy. A large proportion of 2L (26.9%) and 3L (24.2%) treatments contained bevacizumab. 96 patients (3.2%) received regorafenib, representing 6.7% of 3L and 10.6% of 4L therapies. Patients who did not receive systemic therapy were older (mean age of 80.8 vs 74.9) and had multiple comorbidities (63.1% vs 48.4% with Klabunde score > 1). Conclusions: In a Medicare-insured mCRC cohort, patients generally received systemic therapy adhering to consensus guidelines. However, many patients did not receive systemic therapy – suggesting there may be delays in diagnosis and barriers to treatment which lower systemic therapy utilization.

Purpose Patients with cancer are more likely to file for bankruptcy than the general population, but the impact of severe financial distress on health outcomes among patients with cancer is not known. Methods We linked Western Washington SEER Cancer Registry records with federal bankruptcy records for the region. By using propensity score matching to account for differences in several demographic and clinical factors between patients who did and did not file for bankruptcy, we then fit Cox proportional hazards models to examine the relationship between bankruptcy filing and survival. Results Between 1995 and 2009, 231,596 persons were diagnosed with cancer. Patients who filed for bankruptcy (n = 4,728) were more likely to be younger, female, and nonwhite, to have local- or regional- (v distant-) stage disease at diagnosis, and have received treatment. After propensity score matching, 3,841 patients remained in each group (bankruptcy v no bankruptcy). In the matched sample, mean age was 53.0 years, 54% were men, mean income was $49,000, and majorities were white (86%), married (60%), and urban (91%) and had local- or regional-stage disease at diagnosis (84%). Both groups received similar initial treatments. The adjusted hazard ratio for mortality among patients with cancer who filed for bankruptcy versus those who did not was 1.79 (95% CI, 1.64 to 1.96). Hazard ratios varied by cancer type: colorectal, prostate, and thyroid cancers had the highest hazard ratios. Excluding patients with distant-stage disease from the models did not have an effect on results. Conclusion Severe financial distress requiring bankruptcy protection after cancer diagnosis appears to be a risk factor for mortality. Further research is needed to understand the process by which extreme financial distress influences survival after cancer diagnosis and to find strategies that could mitigate this risk.

86 Background: National bodies have proposed a number of metrics to measure quality of care at the end of life (EOL). MACRA legislation allows clinics to select the metrics they report to CMS. The self-selection of reported metrics leaves open questions about how representative certain measures, particularly in isolation, may be of overall EOL care in community settings. We examined the consistency of clinical-level performance across three common EOL metrics. Methods: We linked cancer registry records for solid tumor cancer patients diagnosed in Washington State from 2013-2016 with claims from two regional commercial insurers. Representing national recommendations, we compiled 3 EOL quality metrics for each clinic: 1) Chemotherapy in last 14 days of life (DoL); 2) More than 1 ED visit in last 30 DoL; and 3) Admission to ICU in last 30 DoL. Consistency was measured by comparing performance in the top and bottom 3rd across metrics. We compared consistency based on unadjusted rates and risk-standardized rates calculated following CMS methods. Results: The study included 1,535 patients across 12 clinics (median 110 [IQR: 54 – 199] patients/clinic). The clinic rates are below. (See Table.) According to both unadjusted and adjusted rankings, no clinics ranked in the top 3rd across all metrics. Half of clinics (6 of 12) simultaneously ranked in the top 3rd and bottom 3rd of a metric, i.e. high/low performers. The number of high/low performers varied when examining discrete pairs of metrics. The overall pattern was mainly driven by inconsistency between performance in the chemotherapy and ED metrics. Of the other discrete pairs, ICU/Chemo and ICU/ED, clinics were more consistent in performance. Conclusions: We found that clinic performance was not consistently in the highest or lowest tertile across common EOL metrics, suggesting that requiring clinics to report a standard set of metrics may provide a more accurate indication of quality. Furthermore, different population management strategies may be required to improve care targeted by each measure. Future work should focus on the development of multi-dimensional EOL quality performance measures. [Table: see text]

79 Background: Polypharmacy, defined as the concurrent use of multiple medications, may expose patients to drug-drug interactions and excessive medication costs. Patients with advanced cancer may benefit from medication discontinuation when faced with life-limiting illness. While polypharmacy prevalence has been explored in older patients, there are limited data for younger patients with advanced malignancy. To better understand medication use in this population, we characterized medication use and associated out-of-pocket (OOP) costs in the last month of life among commercially insured subjects under age 65 who did not enroll in hospice. Methods: We linked enrollment and claims records from two regional commercial insurers to Surveillance, Epidemiology, and End Results (SEER) Cancer Surveillance System and Washington State Cancer Registry records for patients diagnosed with a stage IV solid tumor malignancy in Washington state between January 1, 2007-December 31, 2016. We excluded patients who did not incur at least 1 claim in their last month. We calculated OOP costs as the difference between allowed and paid claim amounts, adjusted to 2016 dollars. Results: See Table. Among 369 patients, the average age was 56 years (range 23-64) and 51% (n=189) were male. Patients most frequently had lung (n=149, 40%), pancreatic (n=36, 10%) or colorectal cancer (n=32, 9%). Patients had an average of 7 medication claims (range 1-26) in the last month of life; the most commonly observed non-chemotherapy claims were for opioids, benzodiazepines and anti-emetics. Conclusions: A majority of subjects incurred at least 5 medication claims in their last 30 days; almost one-third had 10 or more. Frequent claims were for medication for pain or other symptoms. OOP costs ranged considerably. Future research will focus on developing interventions to assure effective medication use congruent with patient and family preferences and goals of care. [Table: see text]

e15011 Background: Few studies have examined patterns of care relating to guideline-recommended therapies for metastatic colorectal cancer (mCRC) in real-world settings. The goal of this study was to investigate treatment patterns for initial and subsequent therapies for mCRC patients in two large commercial plans in Washington State, USA. Methods: We utilized the Hutchinson Institute for Cancer Outcomes Research linkage of Western Washington Cancer Surveillance System (CSS) SEER data with claims from Regence BlueShield and Premera BlueCross (representing 20% of the WA insured population) for patients diagnosed with mCRC between 2012 and 2016. We identified all patients with Stage IV CRC at diagnosis or stage I-III CRC with subsequent claims suggesting metastatic recurrence. We conducted descriptive analysis and multivariate modeling to evaluate the association between age, race, gender, and rurality with receipt of systemic therapy. We examined 1L (first line) regimens received in the first 15 days of treatment, and classified them as adherent or not to NCCN guidelines. In light of recent FDA approval for regorafenib (9/2012) and TAS-102 (9/2015), we examined use of these therapies for refractory mCRC. Results: Among 4,536 CRC cases identified in the database, 426 patients had evidence of metastatic disease and met all inclusion criteria. 87 (20.4%) received no systemic therapy. Of the 339 (79.5%) who received systemic therapy, FOLFOX was the most common 1L regimen (29.5%), followed by FOLFOX+bevacizumab (12.3%). There was evidence of 1L treatment outside of NCCN guidelines – 8.5% of patients received single-agent bevacizumab, and 8.0% received single-agent oxaliplatin. Patients who did not receive systemic therapy were more likely to be older and live in a large metropolitan area. For subsequent therapies, 24 (7.1%) patients received regorafenib, and 5 (1.5%) received TAS-102. Conclusions: In a commercially-insured mCRC cohort, a majority received 1L guideline-based therapy. Many patients did not receive systemic therapy despite current recommendations - further evaluation of factors contributing to these decisions are warranted. Use of regorafenib and TAS-102 was uncommon.

6582 Background: Although ASCO Choosing Wisely guidelines recommend against routine surveillance testing or imaging for asymptomatic individuals with early-stage breast cancer (ESBC) treated with curative intent, they are frequently performed. Physician specialty and costs associated with surveillance testing and imaging were examined in ESBC patients. Methods: Cancer registry patient records in Western Washington from 2007 to 2015 were linked with claims from two regional commercial insurers. Selected patients had been diagnosed with stage I/II breast cancer and treated with mastectomy or lumpectomy + radiation. Surveillance was considered from the first 4 month gap in treatment (surgery, chemo, radiation) through 13 months or restart of treatment. Evaluation and Management (E&M) and procedure codes for tumor marker (CEA, CA 15-3, CA 27.29) and advanced imaging (PET, CT, bone scan) were identified. Specialty codes were used to determine provider type. Physician visits were matched to tests using E&M codes in the +/- 7 days around each test. Cost included total reimbursed amount from insurers during the surveillance period. Results: During surveillance, 2,193 patients averaged 13.3 physician visits [median: 11, IQR: 8-17]. Oncologists (91%) and PCPs (83%) were the most common specialties with an average of 3.7 visits each. Overall, 37% of patients received tumor marker tests (avg = 2.8 tests/patient) and 17% received advanced imaging (avg = 1.5 images/patient). The mean total cost during the surveillance period was $18,403 (SD $26,640). Costs were higher for those patients who received tumor marker testing or advanced imaging. Conclusions: Patients frequently see oncologists and PCPs during early surveillance. Targeting oncologists to improve appropriate tumor marker testing could have the largest impact on aligning practice with Choosing Wisely recommendations and potentially reducing the financial burden on patients. [Table: see text]

e18337 Background: Our objectives were to determine advanced imaging use (whole body imaging with bone scan, CT, or PET-CT and breast MRI) during breast cancer surveillance, and to identify drivers of potential imaging overuse as outlined by ASCO’s Choosing Wisely initiative. Methods: Cancer registry records for 2923 women diagnosed with primary breast cancer in Washington State from January 1, 2007 to December 31, 2014 were linked with claims data from two regional commercial insurance plans. Inclusion criteria included women with AJCC stage 0-3 disease treated with curative intent. Women without continuous insurance enrollment from 3 months prior to diagnosis until 14 months after diagnosis were excluded. Surveillance began 4 months after the end of primary therapy and lasted for 15 months or until restart of treatment. Women’s (age, race, family history) and tumor (grade, receptor status, stage) characteristics were collected. Evaluation and management codes from claims data were used to determine mammography, advanced imaging, and tumor biomarker use during the peri-diagnostic and surveillance periods. Multivariate logistic regression models were used to identify factors associated with advanced imaging use during surveillance. Results: Of eligible women, 80.0% (2332/2923) received mammography, 16.5% (483/2932) received whole body imaging, and 21.5% (670/2932) received breast MRI during the surveillance period. Whole body imaging was significantly associated with increasing stage of disease (stage 3: OR = 3.39, 95% CI: 2.30-5.02), peri-diagnostic whole body imaging (OR = 1.80, 95% CI: 1.36-2.38), and surveillance tumor biomarker use (OR = 1.83, 95% CI: 1.46-2.31). Significant predictors of surveillance breast MRI included young age ( < 45 years: OR = 2.40, 95% CI:1.78- 3.25), family history (OR = 1.58, 95% CI:1.26-1.98), peri-diagnostic breast MRI (OR = 2.01, 95% CI: 1.56-2.59), and surveillance tumor biomarker use (OR = 1.74, 95% CI:1.41-2.17). Conclusions: Peri-diagnostic use of advanced imaging and surveillance use of tumor biomarkers are associated with advanced imaging use during surveillance, and may represent targets for interventions to increase adherence to Choosing Wisely clinical guidelines.

6505 Background: The Centers for Medicare and Medicaid Services (CMS) released a quality metric for potentially preventable chemotherapy-associated emergency department (ED) use, effective in 2020. This metric excludes diagnoses with emerging evidence for outpatient management, such as proactive symptom management (PSM) and those for ambulatory care sensitive chronic conditions. Little is known about the intersection between potentially preventable ED visits due to cancer vs. other chronic disease. This study characterized the number and costs of ED visits during treatment. Methods: Western Washington cancer registry records from 2011- 2015 were linked with claims from two commercial insurers. Patients with newly diagnosed solid tumors undergoing initial treatment with chemotherapy or radiation were eligible. ED use was tracked one year post treatment initiation. ED diagnosis codes for fields 1-10 from the CMS metric and the PSM literature were labeled “Potentially Preventable” (Pp). Codes from the Agency for Healthcare Research and Quality’s Prevention Quality Indicators (PQI) for Chronic Conditions were labeled “Potentially Preventable-Chronic Disease” (PpChronic). Costs were adjusted to $2016. Results: Of the 7,053 eligible patients, 2,543 (36.1%) visited the ED (median # visits [IQR]: 1 [1-2]). The most commonly listed codes included Pain (1,054 visits) and Dyspnea (279 visits) for Pp, Hypertension-PQI (652 visits) and COPD-PQI (206 visits) for PpChronic, and Diabetes (247 visits) and Hyperlipidemia (181 visits) for the other codes. Spending on ED visits including both potentially preventable cancer and chronic disease diagnoses totalled $706,552 (20% of ED costs). Conclusions: One fifth of ED costs potentially resulted from simultaneous poor cancer symptom and chronic disease management. Future research should explore the role of chronic illness in categorizing which ED visits are potentially preventable during cancer treatment. [Table: see text]

159 Background: Moving cancer care towards a value framework requires patients, providers, and payers to weigh cost, quality, and outcomes in decision-making. Many efforts are underway to help providers and payers make value decisions, but little has been developed for patients. Our regional value in cancer care effort used a claims-registry database to develop quality and cost reports aiming to provide actionable data to all stakeholders. Methods: Reports were generated using cancer registry records for Western WA from 2007-2015 linked with claims from two regional commercial insurers. Patients were presented quality reports on regionally prioritized metrics and the 2012 ASCO Choosing Wisely guidelines on breast cancer surveillance and EOL care. Patients also reviewed cost reports for episodes of care (diagnosis, treatment, end-of-life (EOL)) and out-of-pocket (OOP) cost estimates. Feedback stemmed from 1) stakeholder meetings over a 2-year period, 2) working groups of patients, payers and providers, and 3) an annual regional meeting on value in cancer care. Results: In total, 13 patients provided feedback at one or more outreach event. See table. Conclusions: Reports from a claims-registry database may not support the information needs of patients for care decision-making or representing “value”. Patients desired understanding more about patients “like them” for decision-making. [Table: see text]

226 Background: Early studies of the oncology medical home suggest that intensive outpatient care (e.g. 24-hour phone triage, same-day infusion) reduces emergency department (ED) and inpatient (IP) use during cancer treatment. Little is known about which services are most cost-effective. One strategy is to measure observed variation in ED and IP rates to pinpoint care features associated with low-use clinics. This study examined clinic-level variation in ED and IP use in a community setting. Methods: Cancer registry records for Western Washington from 2011 to 2015 were linked with claims from two regional commercial insurers. Included patients were diagnosed with breast, lung, colorectal, or prostate cancer and treated with chemotherapy or radiation. All ED and IP use was tracked 1 year after treatment start using claims data. Observed clinic rates were measured as the percentage of patients with 1 or more visits. Expected clinic rates were determined from regional average rates weighted by clinic’s cancer-specific stage mix. Observed-to-expected clinic ratios were calculated and the Wilson Score test (95% CI) was used to determine statistically different rates. Results: The 18 clinics included 4,558 eligible patients (median 196 pts/clinic; range: 35-859). Unstaged lung patients had the highest ED rates (38.5%); unstaged breast had the lowest (13.3%). The highest IP rate was among unstaged colorectal (66.7%); the lowest in local breast (11.1%). One clinic had an observed rate that was significantly above its expected rate in both ED only and ED to IP. One clinic was significantly below its expected rate in both ED to IP and IP only. Conclusions: Even after adjusting for cancer-specific stage, there was sizable clinic-level variation in the percentage of patients visiting the ED or IP. Investigation into care delivery features and practice characteristics, along with further risk adjustment, may yield insights into best practices and identify clinics for intervention. [Table: see text]