Association of platinum-based chemotherapy with live birth and infertility in female survivors of adolescent and young adult cancer

OBJECTIVE

To estimate the effect of platinum-based chemotherapy on live birth (LB) and infertility after cancer, in order to address a lack of treatment-specific fertility risks for female survivors of adolescent and young adult cancer, which limits counseling on fertility preservation decisions.

DESIGN

Retrospective cohort study.

SETTING

US administrative database.

PATIENTS

We identified incident breast, colorectal, and ovarian cancer cases in females aged 15-39 years who received platinum-based chemotherapy or no chemotherapy and matched them to females without cancer.

INTERVENTION

Platinum-based chemotherapy.

MAIN OUTCOME MEASURES

We estimated the effect of chemotherapy on the incidence of LB and infertility after cancer, overall, and after accounting for competing events (recurrence, death, and sterilizing surgeries).

RESULTS

There were 1,287 survivors in the chemotherapy group, 3,192 in the no chemotherapy group, and 34,147 women in the no cancer group, with a mean age of 33 years. Accounting for competing events, the overall 5-year LB incidence was lower in the chemotherapy group (3.9%) vs. the no chemotherapy group (6.4%). Adjusted relative risks vs. no chemotherapy and no cancer groups were 0.61 (95% confidence interval [CI] 0.42-0.82) and 0.70 (95% CI 0.51-0.93), respectively. The overall 5-year infertility incidence was similar in the chemotherapy group (21.8%) compared with the no chemotherapy group (20.7%). The adjusted relative risks vs. no chemotherapy and no cancer groups were 1.05 (95% CI 0.97-1.15) and 1.42 (95% CI 1.31-1.53), respectively.

CONCLUSIONS

Cancer survivors treated with platinum-based chemotherapy experienced modestly increased adverse fertility outcomes. The estimated effects of platinum-based chemotherapy were affected by competing events, suggesting the importance of this analytic approach for interpretations that ultimately inform clinical fertility preservation decisions.

The disproportionate burden of Alzheimer's disease and related dementias (ADRD) in diverse older adults diagnosed with cancer

INTRODUCTION

Older adults living with Alzheimer's disease and related dementias (ADRD) who are then diagnosed with cancer are an understudied population. While the role of cognitive impairment during and after cancer treatment have been well-studied, less is understood about patients who are living with ADRD and then develop cancer. The purpose of this study is to contribute evidence about our understanding of this vulnerable population.

MATERIALS AND METHODS

This was a retrospective cohort study of a linked, representative family of databases of cancer registries and Medicare administrative claims that make up the SEER-Medicare database. Older adults ages 68 and older with a first primary cancer type: breast, cervical, colorectal, lung, oral, or prostate were eligible for inclusion (N = 337,932). Prevalence estimates of ADRD across cancer types and a 5% non-cancer comparison sample were compared by patient factors.

RESULTS

The overall prevalence of patients who had an ADRD diagnosis anytime in the three years prior to their cancer diagnosis was 5.6%. Patients with ADRD were more likely to be female, older (over age 75), a racial/ethnic minority, single, with multiple chronic conditions, and a tumor diagnosed early (stage I) or were unstaged. Black patients with colorectal and oral cancer had the highest and second highest prevalence of ADRD compared to White patients (13.46% vs 7.95% and 12.64% vs 7.82% respectively, p < .0001). We observed the highest prevalence of ADRD among Black patients for breast (11.85%), cervical (11.98%), lung (8.41%), prostate (4.83), and the 5% sample (9.50%, p > .0001).

DISCUSSION

The higher prevalence of ADRD among Black and Latine older adults with cancer not only aligns with the trend observed in our non-cancer comparison sample, but also, these findings demonstrate the compounded risk experienced by minoritized older adults over the life course. The greater than expected prevalence of patients with ADRD who go on to develop cancer demonstrates better assessment of cognition is urgently needed. Accurate identification of these vulnerable populations is critical to improve assessment, care coordination, and address inequities in screening and treatment planning.

Associations Between Intraductal Prostate Cancer and Metastases Following Radical Prostatectomy in Men With Prostate Cancer in the Veterans Affairs Database

PURPOSE

Intraductal carcinoma of the prostate (IDC-P) is a relatively unstudied feature present in some prostate cancer (PC) diagnoses with several studies suggesting associations with higher Gleason scores (GS) and earlier time to biochemical recurrence (BCR) after definitive treatment. We looked to identify cases of IDC-P in the Veterans Health Administration (VHA) database and measure associations between IDC-P and pathological stage, BCR, and metastases.

METHODS

Patients in the VHA database diagnosed with PC from 2000 to 2017, treated with radical prostatectomy (RP) at the VHA were included in the cohort. BCR was defined as post-RP PSA >0.2 or administration of androgen deprivation therapy (ADT). Time to event was defined as time from RP to event or censor. Differences in cumulative incidences were assessed through Gray's test. Associations with IDC-P and pathologic features at RP, BCR and metastases were assessed through multivariable logistic and Cox regression models.

RESULTS

Of 13,913 patients meeting inclusion criteria, 45 patients had IDC-P. Median follow up was 8.8 years from RP. Multivariable logistic regressions showed patients with IDC-P were more likely to have GS ≥8 (Odds Ratio (OR) 1.14, P = .009) and higher T stages (T3 or 4 vs. T1 or 2 OR 1.14, P < .001). In total, 4,318 patients experienced a BCR, and 1,252 patients developed metastases of whom 26 and 12, respectively, had IDC-P. On multivariable regression IDC-P was associated with higher risk of BCR (IDC-P Hazard Ratio (HR) 1.71, P = .006) and metastases (HR 2.84, P < .001). Cumulative incidence of metastases at 4 years for IDC-P and non-IDC-P were 15.9% and 5.5% (P < .001) respectively.

CONCLUSIONS

In this analysis, IDC-P was associated with higher Gleason score at RP, shorter time to BCR, and higher rates of metastases. Further studies are warranted to investigate the molecular underpinnings of IDC-P to better guide treatment strategies for this aggressive disease entity.

Maternal comorbidity and adverse perinatal outcomes in survivors of adolescent and young adult cancer: A cohort study

OBJECTIVE

To evaluate risks of preterm birth (PTB) and severe maternal morbidity (SMM) in female survivors of adolescent and young adult cancer and assess maternal comorbidity as a potential mechanism. To determine whether associations differ by use of assisted reproductive technology (ART).

DESIGN

Retrospective cohort.

SETTING

Commercially insured females in the USA.

SAMPLE

Females with live births from 2000-2019 within a de-identified US administrative health claims data set.

METHODS

Log-binomial regression models estimated relative risks of PTB and SMM by cancer status and tested for effect modification. Causal mediation analysis evaluated the proportions explained by maternal comorbidity.

MAIN OUTCOME MEASURES

PTB and SMM.

RESULTS

Among 46 064 cancer survivors, 2440 singleton births, 214 multiple births and 2590 linked newborns occurred after cancer diagnosis. In singleton births, the incidence of PTB was 14.8% in cancer survivors versus 12.4% in females without cancer (aRR 1.19, 95% CI 1.06-1.34); the incidence of SMM was 3.9% in cancer survivors versus 2.4% in females without cancer (aRR 1.44, 95% CI 1.13-1.83). Cancer survivors had more maternal comorbidities before and during pregnancy; 26% of the association between cancer and PTB and 30% of the association between cancer and SMM was mediated by maternal comorbidities. Tests for effect modification of cancer status on perinatal outcomes by ART were non-significant.

CONCLUSIONS

Preterm birth and SMM risks were modestly increased after cancer. Significant proportions of elevated risks may result from increased comorbidities. ART did not significantly modify the association between adolescent and young adult cancer and adverse perinatal outcomes. The prevention and treatment of comorbidities provides an opportunity to improve perinatal outcomes among cancer survivors.

RACIAL, ETHNIC, AND AGE-RELATED DISPARITIES IN COMORBID CANCER AND DEMENTIA

Older adults make up over 60% of new cancer cases and will be 73% of survivors by 2040. Dementia, which is cognitive impairments that includes Alzheimer’s, affects 10.3% of this age group with a greater burden on Black and Latino/a older adults. It is essential to accurately characterize the prevalence of comorbid dementia and cancer (ADRD+Ca) to ensure guideline concordant care. The goal of our study was to characterize pre-existing ADRD+Ca across cancer sites (cervical, breast, prostate, colorectal, head and neck, and lung). Using a validated CMS algorithm, we found 5.6% ADRD+Ca in our sample. Female patients were more likely to report ADRD+CA vs men (62.5% vs 37.5%, p&lt;0.0001); similarly, age differences were observed with the biggest difference among 85+ ages 33.8% vs 11.7% respectively. The proportion of Black patients with ADRD+Ca was nearly double compared to Ca only groups. Patients with additional multiple comorbidities, unstaged site, and single/other marital status were also more likely to report ADRD+Cancer (p&lt;0.0001). Among cancer sites, we observed the highest prevalence in colorectal and cervical cancer both over 8%, the lowest was in prostate cancer (2.64%). Our findings show inequities in social factors and cancer subtypes among fee-for-service Medicare beneficiaries that warrants investigation.

ALZHEIMER’S DISEASE AND COMORBIDITIES: A COMPLEX INTERPLAY IN THE CONTEXT OF AGING

There is evidence on high prevalence of comorbidity in people with dementia and on associations between comorbidities and progression of Alzheimer’s disease (AD). Comorbidities accumulate with age and age is also a major risk factor for AD. Repeated measurements of comorbidity provide possibilities for gaining more knowledge about dynamic interconnection between comorbidities and AD development in the context of aging. We constructed the comorbidity index (CMI) for participants of the Health and Retirement Study aged 66+ years using data on onset of diseases from linked Medicare service use files (6,830 participants, 3,829 females, 3,001 males). We performed the joint analysis of longitudinal measurements of CMI and data on onset of AD and survival since onset of AD using the approach (the stochastic process model) that allows decomposing the overall association of trajectories of CMI with respective time-to-event outcomes into several aging-related characteristics represented by the model’s components and evaluated indirectly from the data. We found that, overall, CMI is significantly (p< 0.0001) associated with increased risk of onset of AD and decreased survival chances for persons with AD and that this association can be decomposed into associations of AD outcomes with different aging-related components with differentiated impact of genetic and non-genetic factors (such as APOE, polygenic scores, sex, birth cohort). In particular, age patterns and time trends in such components contribute to trends in AD prevalence so that taking into account the age dynamics and time trends in comorbidities (represented by CMI) is essential for forecasting future trends in AD prevalence.

Evaluating High-Dimensional Machine Learning Models to Predict Hospital Mortality Among Older Patients With Cancer

PURPOSE

Older hospitalized cancer patients face high risks of hospital mortality. Improved risk stratification could help identify high-risk patients who may benefit from future interventions, although we lack validated tools to predict in-hospital mortality for patients with cancer. We evaluated the ability of a high-dimensional machine learning prediction model to predict inpatient mortality and compared the performance of this model to existing prediction indices.

METHODS

We identified patients with cancer older than 75 years from the National Emergency Department Sample between 2016 and 2018. We constructed a high-dimensional predictive model called Cancer Frailty Assessment Tool (cFAST), which used an extreme gradient boosting algorithm to predict in-hospital mortality. cFAST model inputs included patient demographic, hospital variables, and diagnosis codes. Model performance was assessed with an area under the curve (AUC) from receiver operating characteristic curves, with an AUC of 1.0 indicating perfect prediction. We compared model performance to existing indices including the Modified 5-Item Frailty Index, Charlson comorbidity index, and Hospital Frailty Risk Score.

RESULTS

We identified 2,723,330 weighted emergency department visits among older patients with cancer, of whom 144,653 (5.3%) died in the hospital. Our cFAST model included 240 features and demonstrated an AUC of 0.92. Comparator models including the Modified 5-Item Frailty Index, Charlson comorbidity index, and Hospital Frailty Risk Score achieved AUCs of 0.58, 0.62, and 0.71, respectively. Predictive features of the cFAST model included acute conditions (respiratory failure and shock), chronic conditions (lipidemia and hypertension), patient demographics (age and sex), and cancer and treatment characteristics (metastasis and palliative care).

CONCLUSION

High-dimensional machine learning models enabled accurate prediction of in-hospital mortality among older patients with cancer, outperforming existing prediction indices. These models show promise in identifying patients at risk of severe adverse outcomes, although additional validation and research studying clinical implementation of these tools is needed.

Disparities in Telemedicine Utilization for Urology Patients During the COVID-19 Pandemic

OBJECTIVE

To determine the odds of accessing telemedicine either by phone or by video during the COVID-19 pandemic.

METHODS

We performed a retrospective study of patients who were seen at a single academic institution for a urologic condition between March 15, 2020 and September 30, 2020. The primary outcome was to determine characteristics associated with participating in a telemedicine appointment (video or telephone) using logistic regression multivariable analysis. We used a backward model selection and variables that were least significant were removed. We adjusted for reason for visit, patient characteristics such as age, sex, ethnicity, race, reason for visit, preferred language, and insurance. Variables that were not significant that were removed from our final model included median income estimated by zip code, clinic location, provider age, provider sex, and provider training.

RESULTS

We reviewed 4234 visits: 1567 (37%) were telemedicine in the form of video 1402 (33.1%) or telephone 164 (3.8%). The cohort consisted of 2516 patients, Non-Hispanic White (n = 1789, 71.1%) and Hispanic (n = 417, 16.6%). We performed multivariable logistic regression analysis and demonstrated that patients who were Hispanic, older, or had Medicaid insurance were significantly less likely to access telemedicine during the pandemic. We did not identify differences in telemedicine utilization when stratifying providers by their age, sex, or training type (physician or advanced practice provider).

CONCLUSION

We conclude that there are differences in the use of telemedicine and that this difference may compound existing disparities in care. Additionally, we identified that these differences were not associated with provider attributes. Further study is needed to overcome barriers in access to telemedicine.

Long-term antimüllerian hormone patterns differ by cancer treatment exposures in young breast cancer survivors

OBJECTIVE

To compare antimüllerian hormone (AMH) patterns by cancer status and treatment exposures across 6 years after incident breast cancer using administrative data.

DESIGN

In a cross-sectional design, AMH levels in patients who developed incident breast cancer between ages 15-39 years during 2005-2019 were matched 1:10 to levels in females without cancer in the OptumLabs Data Warehouse. Modeled AMH patterns were compared among cyclophosphamide-based chemotherapy, non-cyclophosphamide-based chemotherapy, no chemotherapy, and no breast cancer groups.

SETTING

Commercially insured females in the United States.

PATIENT(S)

Females with and without breast cancer.

EXPOSURE(S)

Breast cancer, cyclophosphamide- and non-cyclophosphamide-based chemotherapy.

MAIN OUTCOME MEASURE(S)

AMH levels.

RESULT(S)

A total of 233 patients with breast cancer (mean age, 34 years; standard deviation, 3.7 years) contributed 278 AMH levels over a median of 2 years (range, 0-6.7 years) after diagnosis; 52% received cyclophosphamide-based chemotherapy, 17% received non-cyclophosphamide-based chemotherapy (80% platinum-based), and 31% received no chemotherapy. A total of 2,777 matched females without cancer contributed 2,780 AMH levels. The pattern of AMH levels differed among the 4 groups. Among females without cancer and breast cancer survivors who did not undergo chemotherapy, AMH declined linearly over time. In contrast, among those who received cyclophosphamide-based and noncyclophosphamide-based chemotherapy, a nonlinear pattern of AMH level of initial fall during chemotherapy, followed by an increase over 2-4 years, and then by a plateau over 1-2 years before a decline was observed.

CONCLUSION(S)

In breast cancer survivors, AMH levels from administrative data supported ovarian toxicity of non-cyclophosphamide-based chemotherapy in breast cancer and efficiently depicted the timing and duration of changes in ovarian reserve to reflect the residual reproductive lifespan.

Association of Physician Referrals with Timely Cancer Care Using Tumor Registry and Claims Data

More Americans are being screened for and more are surviving colorectal cancer due to advanced treatments and better quality of care; however, these benefits are not equitably distributed among diverse or older populations. Differential care delivery outcomes are driven by multiple factors, including access to timely treatment that comes from high-quality care coordination. Providers help ensure such coordinated care, which includes timely referrals to specialists. Variation in referrals between providers can also result in differences in treatment plans and outcomes. Patients who are more often referred between the same diagnosing and treating providers may benefit from more timely care compared to those who are not. Our objective is to examine patterns of referral, or patient-sharing networks (PSNs), and our outcome, treatment delay of 30-days (yes/no). We hypothesize that if a patient is in a PSN they will have lower odds of a 30-day treatment initiation delay. Our observational population-based analysis using the National Cancer Institute (NCI)-linked tumor registry and Medicare claims database includes records for 27,689 patients diagnosed with colorectal cancer from 2001 to 2013, and treated with either chemotherapy, radiotherapy, or surgery. We modeled the adjusted odds of a delay and found 17.04% of patients experienced a 30-day delay in initial treatment. Factors that increased odds of a delay were lack of membership in a PSN (adjusted odds ratio [AOR]: 2.20; 95% confidence interval [CI]: 1.71-2.84), racial/ethnic minority status, and having multiple comorbidities. Provider characteristics significantly associated with greater odds of a delay were if dyads were not in the same facility (AOR: 1.95; 95% CI: 1.81-2.10), if providers were different genders, most notably male (diagnosing) and female (treating) [AOR: 1.23; 95% CI: 1.08-1.40, p = 0.0015]. PSNs appear to be associated with reduced of a care delay. The associations observed in our study address the demand for developing multilevel interventions to improve the delivery and coordination of high-quality of care for older cancer patients.

Metastasis and Mortality in Men With Low- and Intermediate-Risk Prostate Cancer on Active Surveillance

BACKGROUND

Active surveillance (AS) is a safe treatment option for men with low-risk, localized prostate cancer. However, the safety of AS for patients with intermediate-risk prostate cancer remains unclear.

PATIENTS AND METHODS

We identified men with NCCN-classified low-risk and favorable and unfavorable intermediate-risk prostate cancer diagnosed between 2001 and 2015 and initially managed with AS in the Veterans Health Administration. We analyzed progression to definitive treatment, metastasis, prostate cancer-specific mortality (PCSM), and all-cause mortality using cumulative incidences and multivariable competing-risks regression.

RESULTS

The cohort included 9,733 men, of whom 1,007 (10.3%) had intermediate-risk disease (773 [76.8%] favorable, 234 [23.2%] unfavorable), followed for a median of 7.6 years. The 10-year cumulative incidence of metastasis was significantly higher for patients with favorable (9.6%; 95% CI, 7.1%-12.5%; P<.001) and unfavorable intermediate-risk disease (19.2%; 95% CI, 13.4%-25.9%; P<.001) than for those with low-risk disease (1.5%; 95% CI, 1.2%-1.9%). The 10-year cumulative incidence of PCSM was also significantly higher for patients with favorable (3.7%; 95% CI, 2.3%-5.7%; P<.001) and unfavorable intermediate-risk disease (11.8%; 95% CI, 6.8%-18.4%; P<.001) than for those with low-risk disease (1.1%; 95% CI, 0.8%-1.4%). In multivariable competing-risks regression, favorable and unfavorable intermediate-risk patients had significantly increased risks of metastasis and PCSM compared with low-risk patients (all P<.001).

CONCLUSIONS

Compared with low-risk patients, those with favorable and unfavorable intermediate-risk prostate cancer managed with AS are at increased risk of metastasis and PCSM. AS may be an appropriate option for carefully selected patients with favorable intermediate-risk prostate cancer, though identification of appropriate candidates and AS protocols should be tested in future prospective studies.

Temporal Trends and Predictors of Opioid Use Among Older Patients With Cancer

OBJECTIVES

While opioids represent a cornerstone of cancer pain management, the timing and patterns of opioid use in the cancer population have not been well studied. This study sought to explore longitudinal trends in opioid use among Medicare beneficiaries with nonmetastatic cancer.

MATERIALS AND METHODS

Within a cohort of 16,072 Medicare beneficiaries ≥66 years old diagnosed with nonmetastatic cancer between 2007 and 2013, we determined the likelihood of receiving a short-term (0 to 6 mo postdiagnosis), intermediate-term (6 to 12 mo postdiagnosis), long-term (1 to 2 y postdiagnosis), and high-risk (morphine equivalent dose ≥90 mg/day) opioid prescription after cancer diagnosis. Multivariable logistic regression models were used to identify patient and cancer risk factors associated with these opioid use endpoints.

RESULTS

During the study period, 74.6% of patients received an opioid prescription, while only 2.66% of patients received a high-risk prescription. Factors associated with use varied somewhat between short-term, intermediate-term, and long-term use, though in general, patients at higher risk of receiving an opioid prescription after their cancer diagnosis were younger, had higher stage disease, lived in regions of higher poverty, and had a history of prior opioid use. Prescriptions for high-risk opioids were associated with individuals living in regions with lower poverty.

CONCLUSIONS

Temporal trends in opioid use in cancer patients depend on patient, demographic, and tumor characteristics. Overall, understanding these correlations may help physicians better identify patient-specific risks of opioid use and could help better inform future evidence-based, cancer-specific opioid prescription guidelines.

The impact of Medicaid expansion on colorectal cancer incidence among vulnerable populations

68

Background: With the Affordable Care Act (ACA), the number of uninsured patients in states that expanded Medicaid decreased more among racial/ethnic minorities and lower income adults. Increased access to care could influence colorectal cancer (CRC) incidence through increased screening. However, we lack research on whether Medicaid expansion differentially influenced CRC incidence among vulnerable patient subgroups. This population-based study examines whether Medicaid expansion with the ACA was associated with decreased CRC incidence among racial/ethnic minorities, and adults with lower income. Methods: We queried the Surveillance, Epidemiology, and End Results Program (SEER) database to calculate the age-adjusted incidence rates of CRC among patients under 65 years of age diagnosed between 2010 and 2018. We categorized states into two groups: states that expanded Medicaid on January 1, 2014, and states that did not expand Medicaid over the study period. We determined the change in CRC incidence before Medicaid expansion (2010-2013) and after Medicaid expansion (2014-2018). We used a difference-in-difference approach to determine whether changes in CRC incidence differed by expansion status among all patients and among subgroups stratified by race/ethnicity and other socioeconomic indicators. Results: Among the entire study cohort, from 2010-2013 to 2014-2018, rates of CRC (per 100,000) increased from 26.6 to 28.3, and this increasing rate did not differ by ACA expansion status (p=0.48). We found that the impact of ACA expansion on CRC incidence varied by race/ethnicity. The increase in CRC rates was higher among non-ACA expansion states compared to ACA expansion states for Hispanics (5.4 vs. 1.6 increase per 100,000; p=0.002), and Asian or Pacific Islanders (4.3 vs. 0.4 increase per 100,000; p=0.02), but not with Black (p=0.94), or non-Hispanic white patients (p=0.91). The change in CRC incidence between 2010-2013 and 2014-2018 did not differ by county-level household income, fraction under the federal poverty level, or education level (all p>0.05). Conclusions: This study found that Medicaid expansion through the ACA might differentially benefit Hispanic and Asian patients with respect to decreases in CRC incidence. This study reports on the first 5 years after the ACA, though the true benefits of increased access to care may take longer to manifest. Additional research with longer follow-up is required to fully understand the influence of Medicaid expansion.

Disparities in telemedicine during COVID-19

BACKGROUND

Oncology rapidly shifted to telemedicine in response to the COVID-19 pandemic. Telemedicine can increase access to healthcare, but recent research has shown disparities exist with telemedicine use during the pandemic. This study evaluated health disparities associated with telemedicine uptake during the COVID-19 pandemic among cancer patients in a tertiary care academic medical center.

METHODS

This retrospective cohort study evaluated telemedicine use among adult cancer patients who received outpatient medical oncology care within a tertiary care academic healthcare system between January and September 2020. We used multivariable mixed-effects logistic regression models to determine how telemedicine use varied by patient race/ethnicity, primary language, insurance status, and income level. We assessed geospatial links between zip-code level COVID-19 infection rates and telemedicine use.

RESULTS

Among 29,421 patient encounters over the study period, 8,541 (29%) were delivered via telemedicine. Several groups of patients were less likely to use telemedicine, including Hispanic (adjusted odds ratio [aOR] 0.86, p = 0.03), Asian (aOR 0.79, p = 0.002), Spanish-speaking (aOR 0.71, p = 0.0006), low-income (aOR 0.67, p < 0.0001), and those with Medicaid (aOR 0.66, p < 0.0001). Lower rates of telemedicine use were found in zip codes with higher rates of COVID-19 infection. Each 10% increase in COVID-19 infection rates was associated with an 8.3% decrease in telemedicine use (p = 0.002).

CONCLUSIONS

This study demonstrates racial/ethnic, language, and income-level disparities with telemedicine use, which ultimately led patients with the highest risk of COVID-19 infection to use telemedicine the least. Additional research to better understand actionable barriers will help improve telemedicine access among our underserved populations.

Impacts of an Opioid Safety Initiative on US Veterans Undergoing Cancer Treatment

BACKGROUND

There is limited research on how the opioid epidemic and consequent risk reduction policies have affected pain management among cancer patients. The purpose of this study was to analyze how the Opioid Safety Initiative (OSI) implemented at the Veterans Health Administration affected opioid prescribing patterns and opioid-related toxicity.

METHODS

We performed an interrupted time series analysis of 42 064 opioid-naïve patients treated at the Veterans Health Administration for prostate, lung, breast, and colorectal cancer from 2011 to 2016. Segmented regression was used to evaluate the impact of the OSI on the incidence of any new opioid prescriptions, high-risk prescriptions, persistent use, and pain-related emergency department (ED) visits. We compared the cumulative incidence of adverse opioid events including an opioid-related admission or diagnosis of misuse before and after the OSI. All statistical tests were 2-sided.

RESULTS

The incidence of new opioid prescriptions was 26.7% (95% confidence interval [CI] = 25.0% to 28.4%) in 2011 and increased to 50.6% (95% CI = 48.3% to 53.0%) by 2013 before OSI implementation (monthly rate of change: +3.3%, 95% CI = 1.3% to 4.2%, P < .001). After the OSI, there was a decrease in the monthly rate of change for new prescriptions (-3.4%, 95% CI = -3.9 to -2.9%, P < .001). The implementation of the OSI was associated with a decrease in the monthly rate of change of concomitant benzodiazepines and opioid prescriptions (-2.5%, 95% CI = -3.2% to -1.8%, P < .001), no statistically significant change in high-dose opioids (-1.2%, 95% CI = -3.2% to 0.9%, P = .26), a decrease in persistent opioid use (-5.7%, 95% CI = -6.8% to -4.7%, P < .001), and an increase in pain-related ED visits (+3.0%, 95% CI = 1.0% to 5.0%, P = .003). The OSI was associated with a decreased incidence of opioid-related admissions (3-year cumulative incidence: 0.9% [95% CI = 0.7% to 1.0%] vs 0.5% [95% CI = 0.4% to 0.6%], P < .001) and no statistically significant change in the incidence of opioid misuse (3-year cumulative incidence: 1.2% [95% CI = 1.0% to 1.3%] vs 1.2% [95% CI = 1.1% to 1.4%], P = .77).

CONCLUSIONS

The OSI was associated with a relative decline in the rate of new, persistent, and certain high-risk opioid prescribing as well as a slight increase in the rate of pain-related ED visits. Further research on patient-centered outcomes is required to optimize opioid prescribing policies for patients with cancer.

Comparison of methods to classify ADRD in claims and cancer registry data

Alzheimer’s disease and related dementias(ADRD) affects 10.3% of older Americans (65+), among these 15-30% go on to be diagnosed with cancer. The highest burden of ADRD is experienced by Latino/a (12.2%) and African-American (13.8%) older adults. Older patients with pre-existing ADRD are less likely to receive guideline-concordant cancer care due to lack of consideration of cognitive status, and underestimation of ADRD diagnosis is an issue in secondary data. Our study compares two validated algorithms for classifying ADRD in a sample of cancer patients, the NCI-Charlson and CMS-Chronic Conditions Warehouse (CCW) index. We used existing claims from NCI’s SEER-Medicare linked database (2004-2013, N=37,932). Patients were selected based on cancer diagnosis at any stage with at least 36 months of data prior to diagnosis to identify ADRD. We analyzed breast, lung, prostate, cervix, head & neck(HNC), and colorectal(CRC) cancers(CA). We found a prevalence of 2.8% (9549 cases of ADRD+CA) using the NCI-index compared with a prevalence of 5.6% (18989 cases) with the CCW-index. ADRD+CA numbers differed significantly in all cancers for all races, however, we observed the greatest magnitude of difference among Latino/a and African-American patients. The NCI index significantly underestimated prevalence compared with the CCW: 1.21% vs 3.28% Breast; 2.29% vs 4.60% CRC; 2.88% vs 6.44% Lung; 1.36% vs 8.62% Prostate, and 4.21% vs 11.61% HNC. Our findings suggest a need to develop validated algorithms for classification, using an evidence-base generated by incorporating information and decision-making theories from the expertise of clinicians currently diagnosing ADRD using clinical assessments in diverse populations.

Association of Health-Care System and Survival in African American and Non-Hispanic White Patients With Bladder Cancer

BACKGROUND

African American patients with bladder cancer have inferior outcomes compared with non-Hispanic White (White) patients. We hypothesize that access to health care is a primary determinant of this disparity. We compared outcomes by race for patients with bladder cancer receiving care within the predominant hybrid-payer health-care model of the United States captured in the Surveillance, Epidemiology, and End Results (SEER) database with those receiving care within the equal-access model of the Veterans' Health Administration (VHA).

METHODS

African American and White patients diagnosed with bladder cancer were identified in SEER and VHA. Stage at presentation, bladder cancer-specific mortality (BCM), and overall survival (OS) were compared by race within each health-care system.

RESULTS

The SEER cohort included 122 449 patients (93.7% White, 6.3% African American). The VHA cohort included 36 322 patients (91.0% White, 9.0% African American). In both cohorts, African American patients were more likely to present with muscle-invasive disease and metastases, but the differences between races were statistically significantly smaller in VHA. In SEER multivariable models, African American patients had worse BCM (hazard ratio [HR] = 1.22, 95% confidence interval [CI] = 1.15 to 1.29) and OS (HR = 1.26, 95% CI = 1.20 to 1.31). In contrast within the VHA, African American patients had similar BCM (HR = 0.97, 95% CI = 0.88 to 1.07) and OS (HR = 0.99, 95% CI = 0.93 to 1.05).

CONCLUSIONS

In this study of contrasting health-care models, receiving medical care in an equal-access system was associated with reduced differences in stage at presentation and eliminated disparities in survival outcomes for African American patients with bladder cancer. Our findings highlight the importance of reducing financial barriers to care to notably improve health equity and oncologic outcomes for African American patients.

Evaluating Prostate-Specific Antigen Screening for Young African American Men With Cancer

BACKGROUND

Despite higher risks associated with prostate cancer, young African American men are poorly represented in prostate-specific antigen (PSA) trials, which limits proper evidence-based guidance. We evaluated the impact of PSA screening, alongside primary care provider utilization, on prostate cancer outcomes for these patients.

METHODS

We identified African American men aged 40-55 years, diagnosed with prostate cancer between 2004 and 2017 within the Veterans Health Administration. Inverse probability of treatment-weighted propensity scores were used in multivariable models to assess PSA screening on PSA levels higher than 20, Gleason score of 8 or higher, and metastatic disease at diagnosis. Lead-time adjusted Fine-Gray regression evaluated PSA screening on prostate cancer-specific mortality (PCSM), with noncancer death as competing events. All statistical tests were 2-sided.

RESULTS

The cohort included 4726 patients. Mean age was 51.8 years, with 84-month median follow-up. There were 1057 (22.4%) with no PSA screening prior to diagnosis. Compared with no screening, PSA screening was associated with statistically significantly reduced odds of PSA levels higher than 20 (odds ratio [OR] = 0.56, 95% confidence interval [CI] = 0.49 to 0.63; P < .001), Gleason score of 8 or higher (OR = 0.78, 95% CI = 0.69 to 0.88; P < .001), and metastatic disease at diagnosis (OR = 0.50, 95% CI = 0.39 to 0.64; P < .001), and decreased PCSM (subdistribution hazard ratio = 0.52, 95% CI = 0.36 to 0.76; P  < .001). Primary care provider visits displayed similar effects.

CONCLUSIONS

Among young African American men diagnosed with prostate cancer, PSA screening was associated with statistically significantly lower risk of PSA levels higher than 20, Gleason score of 8 or higher, and metastatic disease at diagnosis and statistically significantly reduced risk of PCSM. However, the retrospective design limits precise estimation of screening effects. Prospective studies are needed to validate these findings.

Impact of underlying malignancy on emergency department utilization and outcomes

PURPOSE

Cancer patients frequently utilize the emergency department (ED) for a variety of diagnoses both related to and unrelated to their cancer, yet ED outcomes for cancer patients are not well documented. This study sought to define risks and identify predictors for inpatient admission and hospital mortality among cancer patients presenting to the ED.

PATIENTS AND METHODS

We utilized the National Emergency Department Sample to identify patients with and without a diagnosis of cancer presenting to the ED between January 2016 and December 2018. We used multivariable mixed-effects logistic regression models to assess the influence of cancer on outcomes of hospital admission after the ED visit and hospital mortality for the whole patient cohort and individual presenting diagnoses.

RESULTS

There were 340 million weighted ED visits, of which 8.3 million (2.3%) were associated with a cancer diagnosis. Compared to non-cancer patients, patients with cancer had an increased risk of inpatient admission (64.7% vs. 14.8%; p < 0.0001) and hospital mortality (4.6% vs. 0.5%; p < 0.0001). For each of the top 15 presenting diagnoses, cancer patients had increased risks of hospitalization (odds ratio [OR] range 2.0-13.2) or death (OR range 2.1-14.4). Although our dataset does not contain reliable estimation of stage, cancer site was the most robust individual predictor associated with the risk of hospitalization or death compared to other clinical or system-related factors.

CONCLUSIONS

Cancer patients in the ED have high risks for hospital admission and death when compared to patients without cancer. Cancer patients represent a distinct population and may benefit from cancer-specific risk stratification or focused interventions to improve outcomes.

Evaluating the clinical trends and benefits of low-dose computed tomography in lung cancer patients

Background

Despite guideline recommendations, utilization of low‐dose computed tomography (LDCT) for lung cancer screening remains low. The driving factors behind these low rates and the real‐world effect of LDCT utilization on lung cancer outcomes remain limited.

Methods

We identified patients diagnosed with non‐small cell lung cancer (NSCLC) from 2015 to 2017 within the Veterans Health Administration. Multivariable logistic regression assessed the influence of LDCT screening on stage at diagnosis. Lead time correction using published LDCT lead times was performed. Cancer‐specific mortality (CSM) was evaluated using Fine–Gray regression with non‐cancer death as a competing risk. A lasso machine learning model identified important predictors for receiving LDCT screening.

Results

Among 4664 patients, mean age was 67.8 with 58‐month median follow‐up, 95% CI = [7–71], and 118 patients received ≥1 screening LDCT before NSCLC diagnosis. From 2015 to 2017, LDCT screening increased (0.1%–6.6%, mean = 1.3%). Compared with no screening, patients with ≥1 LDCT were more than twice as likely to present with stage I disease at diagnosis (odds ratio [OR] 2.16 [95% CI 1.46–3.20]) and less than half as likely to present with stage IV (OR 0.38 [CI 0.21–0.70]). Screened patients had lower risk of CSM even after adjusting for LDCT lead time (subdistribution hazard ratio 0.60 [CI 0.42–0.85]). The machine learning model achieved an area under curve of 0.87 and identified diagnosis year and region as the most important predictors for receiving LDCT. White, non‐Hispanic patients were more likely to receive LDCT screening, whereas minority, older, female, and unemployed patients were less likely.

Conclusions

Utilization of LDCT screening is increasing, although remains low. Consistent with randomized data, LDCT‐screened patients were diagnosed at earlier stages and had lower CSM. LDCT availability appeared to be the main predictor of utilization. Providing access to more patients, including those in diverse racial and socioeconomic groups, should be a priority.

Clinical Data Prediction Model to Identify Patients With Early-Stage Pancreatic Cancer

PURPOSE

Pancreatic cancer is an aggressive malignancy with patients often experiencing nonspecific symptoms before diagnosis. This study evaluates a machine learning approach to help identify patients with early-stage pancreatic cancer from clinical data within electronic health records (EHRs).

MATERIALS AND METHODS

From the Optum deidentified EHR data set, we identified early-stage (n = 3,322) and late-stage (n = 25,908) pancreatic cancer cases over 40 years of age diagnosed between 2009 and 2017. Patients with early-stage pancreatic cancer were matched to noncancer controls (1:16 match). We constructed a prediction model using eXtreme Gradient Boosting (XGBoost) to identify early-stage patients on the basis of 18,220 features within the EHR including diagnoses, procedures, information within clinical notes, and medications. Model accuracy was assessed with sensitivity, specificity, positive predictive value, and the area under the curve.

RESULTS

The final predictive model included 582 predictive features from the EHR, including 248 (42.5%) physician note elements, 146 (25.0%) procedure codes, 91 (15.6%) diagnosis codes, 89 (15.3%) medications, and 9 (1.5%) demographic features. The final model area under the curve was 0.84. Choosing a model cut point with a sensitivity of 60% and specificity of 90% would enable early detection of 58% late-stage patients with a median of 24 months before their actual diagnosis.

CONCLUSION

Prediction models using EHR data show promise in the early detection of pancreatic cancer. Although widespread use of this approach on an unselected population would produce high rates of false-positive tests, this technique may be rapidly impactful if deployed among high-risk patients or paired with other imaging or biomarker screening tools.

Active surveillance for intermediate-risk prostate cancer in African American and non-Hispanic White men

BACKGROUND

The safety of active surveillance (AS) for African American men compared with non-Hispanic White (White) men with intermediate-risk prostate cancer is unclear.

METHODS

The authors identified patients with modified National Comprehensive Cancer Network favorable ("low-intermediate") and unfavorable ("high-intermediate") intermediate-risk prostate cancer diagnosed between 2001 and 2015 and initially managed with AS in the Veterans Health Administration database. They analyzed definitive treatment, disease progression, metastases, prostate cancer-specific mortality (PCSM), and all-cause mortality by using cumulative incidences and multivariable competing-risks (disease progression, metastasis, and PCSM) or Cox (all-cause mortality) regression.

RESULTS

The cohort included 1007 men (African Americans, 330 [32.8%]; Whites, 677 [67.2%]) followed for a median of 7.7 years; 773 (76.8%) had low-intermediate-risk disease, and 234 (23.2%) had high-intermediate-risk disease. The 10-year cumulative incidences of definitive treatment were not significantly different (African Americans, 83.5%; 95% confidence interval [CI], 78.5%-88.7%; Whites, 80.6%; 95% CI, 76.6%-84.4%; P = .17). Among those with low-intermediate-risk disease, there were no significant differences in the 10-year cumulative incidences of disease progression (African Americans, 46.8%; 95% CI, 40.0%-53.3%; Whites, 46.9%; 95% CI, 42.1%-51.5%; P = .91), metastasis (African Americans, 7.1%; 95% CI, 3.7%-11.8%; Whites, 10.8%; 95% CI, 7.6%-14.6%; P = .17), or PCSM (African Americans, 3.8%; 95% CI, 1.6%-7.5%; Whites, 3.8%; 95% CI, 2.0%-6.3%; P = .69). In a multivariable regression including the entire cohort, African American race was not associated with increased risks of definitive treatment, disease progression, metastasis, PCSM, or all-cause mortality (all P > .30).

CONCLUSIONS

Outcomes in the Veterans Affairs Health System were similar for African American and White men treated for low-intermediate-risk prostate cancer with AS.

Outcomes for Muscle-invasive Bladder Cancer with Radical Cystectomy or Trimodal Therapy in US Veterans

Background

Muscle-invasive bladder cancer (MIBC) remains undertreated despite multiple potentially curative options. Both radical cystectomy (RC) with or without neoadjuvant chemotherapy and trimodal therapy (TMT), including transurethral resection of bladder tumor followed by chemoradiotherapy, are standard treatments.

Objective

To evaluate real-world clinical outcomes of RC with neoadjuvant chemotherapy (RC-NAC), RC without NAC, TMT with National Comprehensive Cancer Network guideline–preferred radiosensitizing chemotherapy including cisplatin or mitomycin-C and 5-fluorouracil (pTMT), and TMT with nonpreferred chemotherapy (npTMT).

Design, setting, and participants

US veterans with nonmetastatic MIBC (T2-4aN0-3M0) were studied.

Outcome measurements and statistical analysis

Overall mortality (OM) was evaluated with multivariable Cox proportional hazard model. Bladder cancer-specific mortality (BCSM) was evaluated with multivariable Fine-Gray regression. Salvage cystectomy rates were obtained by chart review.

Results and limitations

Overall 2306 patients were included: 1472 (64%) with RC without NAC, 506 (22%) with RC-NAC, 163 (7%) with pTMT, and 165 (7%) with npTMT. On multivariable analysis, pTMT was associated with similar OM (hazard ratio [HR] 1.19; 95% confidence interval [CI] 0.94–1.50; p = 0.15) and BCSM (HR 1.34; 95% CI 0.99–1.83; p = 0.06) to RC-NAC; npTMT was associated with worse OM (HR 1.30; 95% CI 1.04–1.61; p = 0.02) and BCSM (HR 1.45; 95% CI 1.09–1.94; p = 0.01). RC without NAC was associated with similar OM (HR 1.08; 95% CI 0.95–1.24; p = 0.24) and BCSM (HR 1.02; 95% CI 0.86–1.21; p = 0.79). When stratified by age, among patients ≥65 yr of age, treatment with pTMT was associated with similar OM (HR 1.14; 95% CI 0.87–1.50; p = 0.35) and BCSM (HR 1.11; 95% CI 0.76–1.62; p = 0.60). Among patients <65 yr of age, pTMT was associated with worse OM (HR 1.82; 95% CI 1.14–2.91; p = 0.01) and BCSM (HR 2.51; 95% CI 1.52–4.13; p < 0.01). The 5-yr cumulative incidence of salvage cystectomy in the TMT group was 3.6%.

Conclusions

In MIBC, patients receiving pTMT have comparable survival in RC-NAC patients ≥65 yr and inferior survival in RC-NAC patients <65 yr. Salvage cystectomy rates were low.

Patient summary

Management of muscle-invasive bladder cancer is a multidisciplinary effort requiring thoughtful discussions with patients about treatment options, including trimodal therapy, which is an effective treatment option.

Predictors of inpatient admission for pediatric cancer patients visiting the emergency department

e22019

Background: Pediatric cancer patients represent a vulnerable cohort at risk of adverse outcomes after presenting to the emergency department (ED). Given the severity of cancer-related complications and uniqueness of this population, approaches to better risk stratify this cohort could potentially help define future interventions geared towards improving outcomes. We used a high-dimensional machine learning approach to help define the risk of hospitalization after an ED visit among pediatric patients with cancer. Methods: A cohort of cancer patients under 18 was identified from the Nationwide Emergency Department Sample (NEDS) between 2016-2018. We used a lasso regression model to predict inpatient admission after an ED visit. Model covariates included patient demographics, hospital characteristics, and International Classification of Diseases, version 10 (ICD-10) diagnosis codes. The data were split 75%/25% into training/testing data. The model was constructed with training data, and performance assessed on the test data using the area under the curve (AUC), with an AUC of 1.0 indicating perfect prediction. Results: We identified 129,631 pediatric cancer patients who visited the ED, of which 54.5% were subsequently admitted. The final predictive model included 150 variables, including 9 demographic, 6 hospital, and 135 unique ICD-10 codes. The model demonstrated excellent ability to predict hospitalization with an AUC of 0.96. The top 5 most important variables associated with inpatient admission were diagnoses of paralytic ileus/intestinal obstruction, neutropenia, sepsis, aplastic anemia/bone marrow failure, and bacterial infection. Conclusions: Pediatric cancer patients frequently present to the ED with complications of their cancer or their treatment, and over half of these patients are admitted. This study demonstrates the capacity of high-dimensional prediction models to help identify pediatric patients at risk of hospitalization. Additional research is needed to determine how to implement these predictive models in clinical practice.

Identifying pre-existing dementia in older adults diagnosed with cancer using a national claims database

e18678

Background: Older adults over the age of 65 represent the majority of patients diagnosed with (60%), among them, 15-30% have a pre-existing Alzheimer’s disease or related dementia (ADRD) that puts them at higher risk for over and under treatment. Studying the role of pre-existing ADRD in cancer patients is vital to understanding treatment planning behavior, patterns of health care utilization, and adverse treatment outcomes. Massive administrative datasets, or “big data” represent the information rich environment that is useful for this endeavor. Methods: Our study utilized a clinically validated algorithm to assess the prevalence of pre-existing ADRD and cancer across six cancer types. We utilized the SEER-Medicare dataset for analyzing the study years 2004-2015 (N = 337 932). We extracted ICD-9 codes to identify ADRD using the Centers for Medicaid Services Chronic Conditions Warehouse (CCW) algorithm. In sensitivity analysis we compared the prevalence of ADRD+Cancer using the NCI (2014) and CCW algorithms. Results: We found a significant difference between the two algorithms (p < .0001) and a higher overall prevalence of comorbid ADRD+Cancer using the CCW (6.6%). Additionally, we found ADRD+Cancer prevalence was significantly higher among racial and ethnic subgroups compared to White and unstaged tumors compared with any numbered American Joint Committee on Cancer (AJCC) stages (p < .0001). Conclusions: Using a clinically validated algorithm we were able to identify more cases of ADRD+Cancer in big data. This figure remains underestimated for ADRD+cancer compared to clinically-validated studies. Further research into the validation approach and codes that are used for ADRD classification can improve how we identify ADRD in massive administrative data. This is critical given the growing population of diverse older adults in the U.S.

Impact of cancer on the risk of unplanned 30-day readmissions

6581

Background: Hospital readmission are associated with unfavorable patient outcomes and increased costs to the healthcare system. Devising interventions to reduce risks of readmission requires understanding patients at highest risk. Cancer patients represent a unique population with distinct risk factors. The purpose of this study was to define the impact of a cancer diagnosis on the risks of unplanned 30-day readmissions. Methods: We identified non-procedural hospital admissions between January through November 2017 from the National Readmission Database (NRD). We included patients with and without a cancer diagnosis who were admitted for non-procedural causes. We evaluated the impact of cancer on the risk of 30-day unplanned readmissions using multivariable mixed-effects logistic regression models. Results: Out of 18,996,625 weighted admissions, 1,685,099 (8.9%) had record of a cancer diagnosis. A cancer diagnosis was associated with an increased risk of readmission compared to non-cancer patients (23.5% vs. 13.6%, p < 0.001). However, among readmissions, cancer patients were less likely to have a preventable readmission (6.5% vs. 12.1%, p < 0.001). When considering the 10 most common causes of initial hospitalization, cancer was associated with an increased risk of readmission for each of these 10 causes (OR range 1.1-2.7, all p < 0.05) compared to non-cancer patients admitted for the same causes. Compared to patients aged 45-64, a younger age was associated with increased risk for cancer patients (OR 1.29, 95%CI [1.24-1.34]) but decreased risk for non-cancer patients (OR 0.65, 95%CI [0.64-0.66]). Among cancer patients, cancer site was the most robust individual predictor for readmission with liver (OR 1.47, 95%CI [1.39-1.55]), pancreas (OR 1.36, 95%CI [1.29-1.44]), and non-Hodgkin’s lymphoma (OR 1.35, 95%CI [1.29-1.42]) having the highest risk compared to the reference group of prostate cancer patients. Conclusions: Cancer patients have a higher risk of 30-day readmission, with increased risks among younger cancer patients, and with individual risks varying by cancer type. Future risk stratification approaches should consider cancer patients as an independent group with unique risks of readmission.

Impact of cancer on emergency department outcomes

e18618

Background: Cancer patients frequently utilize the Emergency Department (ED) for a variety of diagnoses, both related and unrelated to their cancer. Patients with cancer have unique risks related to their cancer and treatment which could influence ED-related outcomes. A better understanding of these risks could help improve risk-stratification for these patients and help inform future interventions. This study sought to define the increased risks cancer patients face for inpatient admission and hospital mortality among cancer patients presenting to the ED. Methods: From the National Emergency Department Sample (NEDS) we identified patients with and without a diagnosis of cancer presenting to the ED between 2016 and 2018. We used International Classification of Diseases, version 10 (ICD10-CM) codes to identify patients with cancer, and to identify patient’s presenting diagnosis. Multivariable mixed-effects logistic regression models assessed the influence of cancer diagnoses on two endpoints: hospital admission from the ED, and inpatient hospital mortality. Results: There were 340 million weighted ED visits, of which 8.3 million (2.3%) occurred in patients with a cancer diagnosis. Compared to non-cancer patients, patients with cancer had an increased risk of inpatient admission (64.7% vs. 14.8%; p < 0.0001) and hospital mortality (4.6% vs. 0.5%; p < 0.0001). Factors associated with both an increased risk of hospitalization and death included older age, male gender, lower income level, discharge quarter, and receipt of care in a teaching hospital. We identified the top 15 most common presenting diagnoses among cancer patients, and among each of these diagnoses, cancer patients had increased risks of hospitalization (odds ratio [OR] range 2.0-13.2; all p < 0.05) and death (OR range 2.1-14.4; all p < 0.05) compared to non-cancer patients with the same diagnosis. Within the cancer patient cohort, cancer site was the most robust individual predictor associated with risk of hospitalization or death, with highest risk among patients with metastatic cancer, liver and lung cancers compared to the reference group of prostate cancer patients. Conclusions: Cancer patients presenting to the ED have high risks for hospital admission and death when compared to patients without cancer. Cancer patients represent a distinct population and may benefit from cancer-specific risk stratification or focused interventions tailored to improve outcomes in the ED setting.

Impact of the VA opioid safety initiative on pain management for cancer patients

102

Background: Limited research exists on how risk reduction policies in response to the opioid epidemic have impacted pain management among cancer patients. This study investigated the impact of the Veteran’s Health Administration (VHA) Opioid Safety Initiative (OSI) on opioid prescribing patterns and opioid-related toxicity among patients undergoing definitive cancer treatment. Methods: This retrospective cohort study included 42,064 opioid-naïve patients receiving definitive local therapy for prostate, lung, breast, and colorectal cancer at the VHA from 2011-2016. Interrupted time series analysis with segmented regression was used to evaluate the impact of the OSI, which launched October 2013. The primary outcome was the incidence of new opioid prescriptions with diagnosis or treatment. Secondary outcomes included rates of high daily dose opioid (≥ 100 morphine milligram equivalent) and concomitant benzodiazepine prescriptions. Additional long-term outcomes included persistent opioid use, opioid abuse diagnoses, pain-related ED visits, and opioid-related admissions. Results: Prior to OSI implementation, the incidence of opioid prescriptions among new cancer patients increased from 26.7% (95% CI 25.0 – 28.4) in the first quarter (Q1) of 2011 to 50.6% (95% CI 48.3 – 53.0) in Q3 2013. There was a monthly increase in opioid prescription rate pre-OSI followed by a monthly decrease post-OSI (Table). High-dose opioid prescriptions were rare, and the monthly rate was stable before and after the OSI. Monthly incidence of concomitant benzodiazepine prescriptions was stable pre-OSI and decreased post-OSI. Persistent opioid use increased pre-OSI and decreased post-OSI. Pain-related ED visits had an incidence of 0.8% (95% CI 0.4 – 1.0) in Q1 2011, 0.3% (95% CI 0.1 – 0.6) in Q3 2013, and 1.8% (95% CI 0.9 – 2.7) in Q4 2016, with an increasing monthly rate after the OSI. At three years, the cumulative incidence of opioid abuse was 1.2% for both the pre- and post-OSI groups but opioid-related admissions were greater in the pre-OSI cohort than the post-OSI cohort (0.9% vs. 0.5%, p < 0.001). Conclusions: The OSI was associated with a decrease in new, persistent, and certain high-risk opioid prescribing as well as an increase in pain-related ED visits. Further research on patient-centered outcomes is required to optimize opioid prescribing policies for patients with cancer.[Table: see text]

Evaluating a high-dimensional machine-learning model to predict hospital mortality among elderly cancer patients

1512

Background: Elderly hospitalized cancer patients face high risks of inpatient hospital mortality. Identifying patients at high risk of hospital mortality could help with risk stratification, and potentially help inform future interventions aimed at improving outcomes. We evaluated the predictive capacity of a high-dimensional machine-learning prediction tool to predict inpatient mortality, and compared the performance of this new tool to existing prediction indices. Methods: We identified cancer patients 75 and over who presented to an emergency department (ED) and were subsequently hospitalized from the National Emergency Department Sample (NEDS) between 2016 and 2018. We used an extreme gradient boosting approach to predict the risk of death during hospitalization. Model covariates included patient demographics, hospital characteristics, and International Classification of Diseases, version 10 (ICD-10) diagnosis codes recorded during the ED visit. The data were split 75%/25% into training/testing datasets. We constructed the model with training data and evaluated performance within the test data using area under the curve (AUC), with an AUC of 1.0 indicating perfect prediction. We compared the performance of this risk prediction model to standard prediction indices including the Hospital Frailty Risk Score, modified 5-item frailty index, and Charlson comorbidity index. Results: We identified 1,892,690 weighted-hospitalizations among elderly cancer patients, of which 133,379 (7.0%) who died in the inpatient setting. Our final predictive model included 238 features, which contained 5 demographic variables, 3 hospital characteristics, and 230 ICD-10 diagnosis codes. The predictive model achieved an AUC of 0.92. Our comparator models including the Hospital Frailty Risk Score, modified 5-item frailty index, and Charlson comorbidity index achieved AUCs of 0.67, 0.56, and 0.60, respectively. Conclusions: Using a high-dimensional machine-learning model enabled a high level of precision in predicting hospital mortality among elderly cancer patients, substantially out-performing existing prediction indices. High-dimensional prediction models show promise in helping to identify patients at risk of severe adverse outcomes, though additional validation is needed as well as research studying how to implement these tools into practice.

Association of increased intensity of prostate-specific antigen screening in younger African American men with improved prostate cancer outcomes

5004

Background: African-American (AA) men are substantially more likely to present with lethal prostate cancer (PCa) at younger ages than non-Hispanic White men. Despite this disparity, AA men are poorly represented in the prostate-specific antigen (PSA) screening studies on which evidence-based PCa screening guidelines are based. This limits proper PSA screening guidance for AA men, especially for those younger than 55. We examined associations of PSA screening intensity with disease severity at diagnosis and prostate cancer-specific mortality (PCSM) in AA men < 55 years of age. Methods: The earliest recommended age to begin discussion of PSA screening is 40 years. We identified AA men aged 40-55 years, diagnosed with PCa from 2004 to 2017 within the Veterans Health Administration. PSA screening was identified using procedural codes. Screening intensity was defined as percentage of years screened within the pre-diagnostic observation period. This included up to 5 years prior to diagnosis. Multivariable logistic regression assessed the influence of PSA screening intensity on metastatic disease at diagnosis. Lead-time correction using published screening-dependent lead times was performed. PCSM was evaluated using Fine-Gray regression and non-cancer death as a competing event. Additional analysis was performed stratifying PSA screening into ‘High’ and ‘Low’ groups centered on the mean. Results: The cohort included 4,654 AA men at a mean age of 51.8 years with mean PSA screening rate of 53.2%. The pre-diagnostic observation period ranged from 1 to 5 years (median = 5 years). Median follow-up was 7 years. At diagnosis, there was a higher prevalence of Gleason sum ≥ 8 (Grade Group ≥ 4) and metastatic disease in the ‘Low’ group compared with the ‘High’ group ([Gleason sum ≥ 8 (Grade Group ≥ 4)]: 18.6% vs. 14.4%, p < 0.01; Metastatic disease at diagnosis: 3.7% vs. 1.4%, p < 0.01). Increased PSA screening intensity was associated with significantly reduced odds of metastatic disease at diagnosis (odds ratio: 0.61, 95% confidence interval (CI) = [0.47-0.81], p < 0.01) and decreased risk of PCSM (sub-distribution hazard ratio: 0.75, 95% CI = [0.59-0.95], p = 0.02). Conclusions: In this large national cohort of AA men aged 40 to 55 years, PSA screening increased intensity was associated with decreased risk of lethal disease and metastases at time of diagnosis and decreased PCSM. These data support the hypothesis that PSA screening and early prostate cancer detection may improve outcomes in younger AA men.

Association of Prostate-Specific Antigen Velocity With Clinical Progression Among African American and Non-Hispanic White Men Treated for Low-Risk Prostate Cancer With Active Surveillance

IMPORTANCE

The association of prostate-specific antigen velocity (PSAV) with clinical progression in patients with localized prostate cancer managed with active surveillance remains unclear and, to our knowledge, has not been studied in African American patients.

OBJECTIVES

To test the hypothesis that PSAV is associated with clinical progression in patients with low-risk prostate cancer treated with active surveillance and to identify differences between African American and non-Hispanic White patients.

DESIGN, SETTING, AND PARTICIPANTS

This was a retrospective population-based cohort study using patient records from the Veterans Heath Administration Informatics and Computing Infrastructure on 5296 patients with a diagnosis of localized prostate cancer from January 1, 2001, to December 31, 2015, who were managed with active surveillance. Follow-up extended through March 31, 2020. Low-risk prostate cancer was defined as International Society of Urologic Pathology grade group (GG) 1 clinical tumor stage 2A or lower, PSA level of 10 ng/dL or lower, active surveillance, and no definitive treatment within the first year after diagnosis with at least 1 additional staging biopsy after diagnostic biopsy.

EXPOSURES

Prostate-specific antigen testing.

MAIN OUTCOMES AND MEASURES

The primary outcome was GG progression detected after repeated biopsy or prostatectomy, defined as GG2 or higher or GG3 or higher. The secondary outcome was incident metastases. Cumulative incidence functions and multivariable Cox proportional hazards regression models were used to test associations between PSAV and outcomes.

RESULTS

The final cohort (n = 5296) included 3919 non-Hispanic White men (74.0%; mean [SD] age, 65.7 [5.8] years) and 1377 African American men (26.0%; mean [SD] age, 62.8 [6.6] years). Compared with African American patients, non-Hispanic White patients were older (mean [SD] age, 65.7 [5.8] years vs 62.8 [6.6] years; P < .001), presented with higher cT stage (stage T2, 608 [15.5%] vs 111 [8.1%]; P < .001), had a higher Charlson Comorbidity Index score (1 and ≥2, 912 [23.3%] vs 273 [19.8%]; P = .002), had higher median income ($60 000 to ≥$100 000, 1223 [31.2%] vs 282 [20.5%]; P < .001), and had a higher median level of education (20% to ≥30% with college degree, 1192 [30.4%] vs 333 [24.2%]; P < .001). Progression to GG2 or higher occurred in 2062 patients (38.9%), with a cumulative incidence of 43.2%, and progression to GG3 or higher occurred in 728 patients (13.7%). Fifty-four patients (1.0%) developed metastases. On multivariable analysis, PSAV was significantly associated with progression to GG2 (hazard ratio, 1.32 [95% CI, 1.26-1.39]), GG3 (hazard ratio, 1.51 [95% CI, 1.41-1.62]), and metastases (hazard ratio, 1.38 [95% CI, 1.10-1.74]). Optimal PSAV thresholds that were associated with progression were significantly lower for African American patients (0.44 ng/mL/y) compared with non-Hispanic White patients (1.18 ng/mL/y).

CONCLUSIONS AND RELEVANCE

This study suggests that PSAV is significantly associated with grade progression among patients with low-risk prostate cancer managed with active surveillance, but at lower values for African American patients compared with non-Hispanic White patients. These data suggest that serial PSA measures may potentially substitute for multiple prostate biopsies and that African American patients may merit increased frequency of PSA testing.

Predicting Persistent Opioid Use, Abuse, and Toxicity Among Cancer Survivors

BACKGROUND

Although opioids play a critical role in the management of cancer pain, the ongoing opioid epidemic has raised concerns regarding their persistent use and abuse. We lack data-driven tools in oncology to understand the risk of adverse opioid-related outcomes. This project seeks to identify clinical risk factors and create a risk score to help identify patients at risk of persistent opioid use and abuse.

METHODS

Within a cohort of 106 732 military veteran cancer survivors diagnosed between 2000 and 2015, we determined rates of persistent posttreatment opioid use, diagnoses of opioid abuse or dependence, and admissions for opioid toxicity. A multivariable logistic regression model was used to identify patient, cancer, and treatment risk factors associated with adverse opioid-related outcomes. Predictive risk models were developed and validated using a least absolute shrinkage and selection operator regression technique.

RESULTS

The rate of persistent opioid use in cancer survivors was 8.3% (95% CI = 8.1% to 8.4%); the rate of opioid abuse or dependence was 2.9% (95% CI = 2.8% to 3.0%); and the rate of opioid-related admissions was 2.1% (95% CI = 2.0% to 2.2%). On multivariable analysis, several patient, demographic, and cancer and treatment factors were associated with risk of persistent opioid use. Predictive models showed a high level of discrimination when identifying individuals at risk of adverse opioid-related outcomes including persistent opioid use (area under the curve [AUC] = 0.85), future diagnoses of opioid abuse or dependence (AUC = 0.87), and admission for opioid abuse or toxicity (AUC = 0.78).

CONCLUSION

This study demonstrates the potential to predict adverse opioid-related outcomes among cancer survivors. With further validation, personalized risk-stratification approaches could guide management when prescribing opioids in cancer patients.

Prognostic utility of pretreatment neutrophil-lymphocyte ratio in survival outcomes in localized non-small cell lung cancer patients treated with stereotactic body radiotherapy: Selection of an ideal clinical cutoff point

•

Neutrophil-lymphocyte ratio is a promising prognostic marker for several cancers.

•

NLR is not useful as a marker of lung cancer survival in localized lung cancer.

•

NLR has potential as a marker of competing mortality risk in localized lung cancer.

•

NLR cutoff of 4.0 is proposed as a clinically useful cutoff point.

Background and purpose

Neutrophil-lymphocyte ratio (NLR) has been associated with overall survival (OS) in non-small cell lung cancer (NSCLC). We aimed to assess the utility of NLR as a predictor of lung cancer-specific survival (LCS) and identify an optimal, pretreatment cutoff point in patients with localized NSCLC treated with stereotactic body radiotherapy (SBRT) within the Veterans Affairs’ (VA) national database.

Materials and methods

In the VA database, we identified patients with biopsy-proven, clinical stage I NSCLC treated with SBRT between 2006 and 2015. Cutoff points for NLR were calculated using Contal/O’Quigley’s and Cox Wald methods. Primary outcomes of OS, LCS, and non-lung cancer survival (NCS) were evaluated in Cox and Fine-Gray models.

Results

In 389 patients, optimal NLR cutoff was identified as 4.0. In multivariable models, NLR > 4.0 was associated with decreased OS (HR 1.44, p = 0.01) and NCS (HR 1.68, p = 0.01) but not with LCS (HR 1.32, p = 0.09). In a subset analysis of 229 patients with pulmonary function tests, NLR > 4.0 remained associated with worse OS (HR 1.51, p = 0.02) and NCS (HR 2.18, p = 0.01) while the association with LCS decreased further (HR 1.22, p = 0.39).

Conclusion

NLR was associated with worse OS in patients with localized NSCLC treated with SBRT; however, NLR was only associated with NCS and not with LCS. Pretreatment NLR, with a cutoff of 4.0, offers potential as a marker of competing mortality risk which can aid in risk stratification in this typically frail and comorbid population. Further studies are needed to validate pretreatment NLR as a clinical tool in this setting.

PSA velocity to predict clinical progression in African American and non-Hispanic White patients with low-risk prostate cancer undergoing active surveillance

196

Background: The utility of prostate-specific antigen velocity (PSAV) to predict clinical progression in patients with localized prostate cancer (PC) on active surveillance remains unclear, and in African American (AA) patients on active surveillance remains undefined. Methods: We performed a cohort analysis using the national US Veterans Affairs Informatics and Computing Infrastructure (VINCI). We identified 5296 patients diagnosed with localized prostate cancer (PC) from 2001 to 2015 managed with active surveillance. Follow-up extended through March 31, 2020. We defined low-risk PC as ISUP grade group 1 (GG1) clinical tumor stage ≤ 2A, and PSA level ≤ 10 ng/dl; and active surveillance as no definitive treatment within the first year after diagnosis with at least one additional staging biopsy after diagnostic biopsy. The primary outcome was grade progression on repeat biopsy/prostatectomy defined as GG2 or GG3. The secondary outcome was incident metastases. Cumulative incidence functions and multivariable Cox proportional hazards models were used to test associations between PSAV and outcomes. Results: The final cohort included 3919 Non-Hispanic White patients (NHW) (74%) and 1377 AA (26%) patients. GG2 progression on repeat biopsy occurred in 2062 (38.9%) patients, with a cumulative incidence (CI) of 43.2 % and GG3 progression occurred in 728 (13.7%) patients, with a CI of 18% at seven years. Fifty-four (0.8%) patients developed metastases, with a CI of 1.4% at ten years. In unadjusted analyses, compared to NHW patients, AA patients were significantly more likely to progress to GG2 (52.8% vs 39.8%, p<0.001), and GG3 (22.2% vs 16.8%, p=0.01). On MV analysis, PSAV was a significant predictor of GG2 (HR 1.32 [1.26-1.39]), GG3 (1.5 [1.40-1.62]), and metastases (1.38 [1.10-1.74]). A significant interaction term between race and PSAV indicated the need for different PSAV thresholds for AA and NHW men. Based on maximally selected rank statistics, optimal thresholds for separating outcomes were different in AA vs NHW men. (0.44 vs. 1.18). Conclusions: In this analysis of PSAV in low-risk prostate cancer patients on AS—to our knowledge, the largest to date for AA patients—we observed that PSAV is a robust predictor of upgrading on restaging biopsy as well as metastasis. Compared to NHW patients, AA patients were more likely to progress at lower values of PSAV and therefore merit close follow-up on active surveillance protocols.

Impact of equal access healthcare on race disparities in bladder cancer

399

Background: Outcomes in bladder cancer are disproportionately worse for black patients compared to white patients. We hypothesize these disparities arise in part due to differences in access to healthcare and therefore may be mitigated in an equal access healthcare system, such as the Veterans Affairs’ (VA) system. Here, we examine outcomes by race for patients with bladder cancer within the VA system and then compare these outcomes to those in the Surveillance, Epidemiology, and End Results (SEER) database. Methods: We performed a retrospective cohort study using VA Informatics and Computing Infrastructure (VINCI) and SEER. We included all patients diagnosed with bladder cancer, American Joint Committee on Cancer (AJCC) stage 0-4 diagnosed between 2000 and 2018. Endpoints of overall survival (OS), bladder cancer-specific survival (BCS), and non-bladder cancer-specific survival (NCS) were evaluated in multivariable Cox and Fine-Gray models. Results: Using the VA dataset, we identified 36322 veterans (9.0% black, 91.0% white) with bladder cancer. Black veterans were more likely to have more comorbidities, reside in zip codes with lower median income and education levels, and present with higher stage disease (AJCC stages 2-4) than white veterans (23.3% vs 19%). In multivariable models accounting for disease stage among other covariables, there were no statistically significant differences in any survival endpoint (Table). Using the SEER dataset, we identified 130998 patients (5.9% black, 94.1% white) with bladder cancer. In similar multivariable models, SEER’s black patients had statistically significant inferior outcomes in all survival endpoints compared to SEER’s white patients (Table). Conclusions: While racial disparities for patients with bladder cancer in the SEER database were observed, no differences in survival outcomes between black and white patients were observed in the VA healthcare system. Of note, black veterans presented with more advanced stage, suggesting a delay in diagnosis or a more aggressive cancer phenotype compared to white patients. Our findings underscore the need to bridge healthcare disparities across diverse racial groups. Our study highlights the beneficial impact of an equal access healthcare system in reducing financial and social barriers to healthcare to counteract racial health disparities. Further research is required to delineate these disparities and guide appropriate screening strategies. [Table: see text]

Racial, Ethnic, and Socioeconomic Discrepancies in Opioid Prescriptions Among Older Patients With Cancer

PURPOSE

Minority race and lower socioeconomic status are associated with lower rates of opioid prescription and undertreatment of pain in multiple noncancer healthcare settings. It is not known whether these differences in opioid prescribing exist among patients undergoing cancer treatment.

METHODS AND MATERIALS

This observational cohort study involved 33,872 opioid-naive patients of age > 65 years undergoing definitive cancer treatment. We compared rates of new opioid prescriptions by race or ethnicity and socioeconomic status controlling for differences in baseline patient, cancer, and treatment factors. To evaluate downstream impacts of opioid prescribing and pain management, we also compared rates of persistent opioid use and pain-related emergency department (ED) visits.

RESULTS

Compared with non-Hispanic White patients, the covariate-adjusted odds of receiving an opioid prescription were 24.9% (95% CI, 16.0 to 33.9, P < .001) lower for non-Hispanic Blacks, 115.0% (84.7 to 150.3, P < .001) higher for Asian-Pacific Islanders, and not statistically different for Hispanics (-1.0 to 14.0, P = .06). There was no significant association between race or ethnicity and persistent opioid use or pain-related ED visits. Patients living in a high-poverty area had higher odds (53.9% [25.4 to 88.8, P < .001]) of developing persistent use and having a pain-related ED visit (39.4% [16.4 to 66.9, P < .001]).

CONCLUSION

For older patients with cancer, rates of opioid prescriptions and pain-related outcomes significantly differed by race and area-level poverty. Non-Hispanic Black patients were associated with a significantly decreased likelihood of receiving an opioid prescription. Patients from high-poverty areas were more likely to develop persistent opioid use and have a pain-related ED visit.

Validation of an oncology-specific opioid risk calculator in cancer survivors

BACKGROUND

Clinical guidelines recommend that providers risk-stratify patients with cancer before prescribing opioids. Prior research has demonstrated that a simple cancer opioid risk score might help identify to patients with cancer at the time of diagnosis with a high likelihood of long-term posttreatment opioid use. This current project validates this cancer opioid risk score in a generalizable, population-based cohort of elderly cancer survivors.

METHODS

This study identified 44,932 Medicare beneficiaries with cancer who had received local therapy. Longitudinal opioid use was ascertained from Medicare Part D data. A risk score was calculated for each patient, and patients were categorized into low-, moderate-, and high-risk groups on the basis of the predicted probability of persistent opioid use. Model discrimination was assessed with receiver operating characteristic curves.

RESULTS

In the study cohort, 5.2% of the patients were chronic opioid users 1 to 2 years after the initiation of cancer treatment. The majority of the patients (64%) were at low risk and had a 1.2% probability of long-term opioid use. Moderate-risk patients (33% of the cohort) had a 5.6% probability of long-term opioid use. High-risk patients (3.5% of the cohort) had a 75% probability of long-term opioid use. The opioid risk score had an area under the receiver operating characteristic curve of 0.869.

CONCLUSIONS

This study found that a cancer opioid risk score could accurately identify individuals with a high likelihood of long-term opioid use in a large, generalizable cohort of cancer survivors. Future research should focus on the implementation of these scores into clinical practice and how this could affect prescriber behavior and patient outcomes.

LAY SUMMARY

A novel 5-question clinical decision tool allows physicians treating patients with cancer to accurately predict which patients will persistently be using opioid medications after completing therapy.

Association Between African American Race and Clinical Outcomes in Men Treated for Low-Risk Prostate Cancer With Active Surveillance

IMPORTANCE

There is concern that African American men with low-risk prostate cancer may harbor more aggressive disease than non-Hispanic White men. Therefore, it is unclear whether active surveillance is a safe option for African American men.

OBJECTIVE

To compare clinical outcomes of African American and non-Hispanic White men with low-risk prostate cancer managed with active surveillance.

DESIGN, SETTING, AND PARTICIPANTS

Retrospective cohort study in the US Veterans Health Administration Health Care System of African American and non-Hispanic White men diagnosed with low-risk prostate cancer between January 1, 2001, and December 31, 2015, and managed with active surveillance. The date of final follow-up was March 31, 2020.

EXPOSURES

Active surveillance was defined as no definitive treatment within the first year of diagnosis and at least 1 additional surveillance biopsy.

MAIN OUTCOMES AND MEASURES

Progression to at least intermediate-risk, definitive treatment, metastasis, prostate cancer-specific mortality, and all-cause mortality.

RESULTS

The cohort included 8726 men, including 2280 African American men (26.1%) (median age, 63.2 years) and 6446 non-Hispanic White men (73.9%) (median age, 65.5 years), and the median follow-up was 7.6 years (interquartile range, 5.7-9.9; range, 0.2-19.2). Among African American men and non-Hispanic White men, respectively, the 10-year cumulative incidence of disease progression was 59.9% vs 48.3% (difference, 11.6% [95% CI, 9.2% to 13.9%); P < .001); of receipt of definitive treatment, 54.8% vs 41.4% (difference, 13.4% [95% CI, 11.0% to 15.7%]; P < .001); of metastasis, 1.5% vs 1.4% (difference, 0.1% [95% CI, -0.4% to 0.6%]; P = .49); of prostate cancer-specific mortality, 1.1% vs 1.0% (difference, 0.1% [95% CI, -0.4% to 0.6%]; P = .82); and of all-cause mortality, 22.4% vs 23.5% (difference, 1.1% [95% CI, -0.9% to 3.1%]; P = 0.09).

CONCLUSIONS AND RELEVANCE

In this retrospective cohort study of men with low-risk prostate cancer followed up for a median of 7.6 years, African American men, compared with non-Hispanic White men, had a statistically significant increased 10-year cumulative incidence of disease progression and definitive treatment, but not metastasis or prostate cancer-specific mortality. Longer-term follow-up is needed to better assess the mortality risk.

Trends in short-term, long-term, and high-risk opioid use among older cancer patients

187

Background: While opioids represent a cornerstone of cancer pain management, the timing and patterns of persistent and high-risk opioid use in the context of the ongoing opioid epidemic are not well studied. This study sought to explore longitudinal trends in short-term, long-term, and high-risk opioid use among older cancer patients. Methods: Within a cohort of 84,994 Medicare beneficiaries ≥ 65 years old diagnosed with cancer between 2007 and 2013, we determined the likelihood of being prescribed an opioid after cancer diagnosis (0-6 months, 6-12 months, and 1-2 years post-diagnosis) and of receiving a daily morphine equivalent dose (MED) ≥ 200 mg, a dose associated with higher opioid-related mortality. Multivariable logistic regression models were used to identify patient and cancer risk factors associated with outcomes. Results: The rates of opioid prescription at 0 to 6 months, 6 to 12 months, and 1 to 2 years after diagnosis were 60.7%, 32.7%, and 38.2% respectively. Among patients who were prescribed an opioid, 4.0% received a MED ≥ 200 mg within 2 years of diagnosis. The likelihood of opioid prescription 0 to 6 months after diagnosis increased over the study period (OR = 1.05 per year, CI = 1.04 to 1.06), while the likelihood of opioid prescription 6 to 12 months (OR = 0.98 per year, CI = 0.97 to 0.99) and 1 to 2 years (OR = 0.98 per year, CI = 0.97 to 0.99) after diagnosis decreased over the study period. The probability of receiving a MED ≥ 200 mg decreased over the study period (OR = 0.98 per year, CI = 0.96 to 0.99). On multivariable analysis, Black (OR = 1.14, CI = 1.06 to 1.23) and American Indian/Native Alaskan (OR = 1.46, CI = 1.06 to 2.02) patients were more likely to receive opioids 1 to 2 years after diagnosis. Patients living in areas with higher rates of poverty (OR = 1.40, CI = 1.32 to 1.48) were also more likely to receive an opioid 1 to 2 years after diagnosis. Black (OR = 1.72, CI = 1.59 to 1.86), Asian (OR = 1.46, CI = 1.19 to 1.80), and Hispanic (OR = 1.36, CI = 1.23 to 1.51) patients were more likely to receive a MED ≥ 200 mg. Conclusions: While the rate of short-term opioid use is rising, rates of persistent and high-risk opioid use are declining among cancer patients. This may reflect growing awareness of the dangers of opioid misuse among clinicians treating patients with cancer. Nevertheless, racial and socioeconomic disparities in patterns of opioid use are persistent in the cancer population.

Evaluation of the Use of Cancer Registry Data for Comparative Effectiveness Research

Importance

Researchers often analyze cancer registry data to assess for differences in survival among cancer treatments. However, the retrospective, nonrandomized design of these analyses raises questions about study validity.

Objective

To examine the extent to which comparative effectiveness analyses using observational cancer registry data produce results concordant with those of randomized clinical trials.

Design, Setting, and Participants

In this comparative effectiveness study, a total of 141 randomized clinical trials referenced in the National Comprehensive Cancer Network Clinical Practice Guidelines for 8 common solid tumor types were identified. Data on participants within the National Cancer Database (NCDB) diagnosed between 2004 and 2014, matching the eligibility criteria of the randomized clinical trial, were obtained. The present study was conducted from August 1, 2017, to September 10, 2019. The trials included 85 118 patients, and the corresponding NCDB analyses included 1 344 536 patients. Three Cox proportional hazards regression models were used to determine hazard ratios (HRs) for overall survival, including univariable, multivariable, and propensity score-adjusted models. Multivariable and propensity score analyses controlled for potential confounders, including demographic, comorbidity, clinical, treatment, and tumor-related variables.

Main Outcomes and Measures

The main outcome was concordance between the results of randomized clinical trials and observational cancer registry data. Hazard ratios with an NCDB analysis were considered concordant if the NDCB HR fell within the 95% CI of the randomized clinical trial HR. An NCDB analysis was considered concordant if both the NCDB and clinical trial P values for survival were nonsignificant (P ≥ .05) or if they were both significant (P < .05) with survival favoring the same treatment arm in the NCDB and in the randomized clinical trial.

Results

Analyses using the NCDB-produced HRs for survival were concordant with those of 141 randomized clinical trials in 79 univariable analyses (56%), 98 multivariable analyses (70%), and 90 propensity score models (64%). The NCDB analyses produced P values concordant with randomized clinical trials in 58 univariable analyses (41%), 65 multivariable analyses (46%), and 63 propensity score models (45%). No clinical trial characteristics were associated with concordance between NCDB analyses and randomized clinical trials, including disease site, type of clinical intervention, or severity of cancer.

Conclusions and Relevance

The findings of this study suggest that comparative effectiveness research using cancer registry data often produces survival outcomes discordant with those of randomized clinical trial data. These findings may help provide context for clinicians and policy makers interpreting observational comparative effectiveness research in oncology.

Claims-Based Approach to Predict Cause-Specific Survival in Men With Prostate Cancer

PURPOSE

Treatment decisions about localized prostate cancer depend on accurate estimation of the patient's life expectancy. Current cancer and noncancer survival models use a limited number of predefined variables, which could restrict their predictive capability. We explored a technique to create more comprehensive survival prediction models using insurance claims data from a large administrative data set. These data contain substantial information about medical diagnoses and procedures, and thus may provide a broader reflection of each patient's health.

METHODS

We identified 57,011 Medicare beneficiaries with localized prostate cancer diagnosed between 2004 and 2009. We constructed separate cancer survival and noncancer survival prediction models using a training data set and assessed performance on a test data set. Potential model inputs included clinical and demographic covariates, and 8,971 distinct insurance claim codes describing comorbid diseases, procedures, surgeries, and diagnostic tests. We used a least absolute shrinkage and selection operator technique to identify predictive variables in the final survival models. Each model's predictive capacity was compared with existing survival models with a metric of explained randomness (ρ²) ranging from 0 to 1, with 1 indicating an ideal prediction.

RESULTS

Our noncancer survival model included 143 covariates and had improved survival prediction (ρ² = 0.60) compared with the Charlson comorbidity index (ρ² = 0.26) and Elixhauser comorbidity index (ρ² = 0.26). Our cancer-specific survival model included nine covariates, and had similar survival predictions (ρ² = 0.71) to the Memorial Sloan Kettering prediction model (ρ² = 0.68).

CONCLUSION

Survival prediction models using high-dimensional variable selection techniques applied to claims data show promise, particularly with noncancer survival prediction. After further validation, these analyses could inform clinical decisions for men with prostate cancer.

Abstract A035: The influence of patient-provider language concordance in cancer care: Results of the Hispanic Outcomes by Language Approach (HOLA) randomized trial

Purpose: Delivering linguistically competent care is critical to serving limited English proficiency patients, and represents a key national strategy to reduce health disparities. Current acceptable standards of communication with non-English speaking patients include providers communicating through professional interpretive services, or bilingual providers speaking patients’ non-English languages directly. This study tests the impact of patient-provider language concordance on patient satisfaction through the conduct of a randomized clinical trial. Methods and Materials: Eighty-three adult Spanish-speaking cancer patients were randomized to receive care from either 1) a bilingual physician speaking to a patient directly in Spanish or 2) from the same physician speaking English and using a professional interpreter service. Validated questionnaires were administered to assess patient-reported satisfaction with both provider communication and overall care. Audio recordings of initial consultations with oncologists were transcribed and analyzed for content variations. Results: Compared to using professional interpretive services, patients cared for in direct Spanish reported significantly improved general satisfaction, technical quality of care, care team interpersonal manner, communication, and time spent with patient. Specific to physician communication, patients rated direct Spanish care more highly in perceived opportunity to disclose concerns, physician empathy, confidence in physician abilities, and general satisfaction with their physician. Analyzing the content of consultation encounters revealed differences between study arms, with the direct Spanish arm having more physician speech related to patient history verification and partnering activities. Additionally, patients in the direct-Spanish arm were more likely to initiate unprompted speech, and ask their providers questions. Conclusions: This study demonstrates improved patient-reported satisfaction among Spanish-speaking cancer patients cared for in direct Spanish compared to patients cared for with interpreter-based communication. Further research into interventions to mitigate this patient-provider language barrier is necessary to optimize care for this minority population.

Citation Format: Daniel M Seible, Souma Kundu, Alexa Azuara, Daniel Cherry, Steven Arias, Vinit Nalawade, Jonathan Cruz, Rolando Arreola, Elena M Martinez, Jesse Nodora, Douglas A Rahn, James D Murphy. The influence of patient-provider language concordance in cancer care: Results of the Hispanic Outcomes by Language Approach (HOLA) randomized trial [abstract]. In: Proceedings of the Twelfth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2019 Sep 20-23; San Francisco, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl_2):Abstract nr A035.