Abstract Session C3: Cancer Research Moderator: Neeraja B. Peterson, - - PDF document

Abstract Session C3: Cancer Research Moderator: Neeraja B. Peterson, MD Discussant: Karla Kerlikowske, MD, SGIM 2014 Distinguished Professor in Cancer Research CARE FRAGMENTATION AND SURVIVAL FOR PATIENTS WITH STAGE III COLON CANCER Tanvir Hussain1,2; Hsien-Yen Chang2; Christine M. Veenstra3; Craig E. Pollack1,2. 1Johns Hopkins University School of Medicine, Baltimore, MD; 2Johns Hopkins Bloomberg School of Public Health, Baltimore, MD;

3University of Michigan Health Systems, Ann Arbor, MI. (Tracking ID #1939909)

BACKGROUND: Cancer care can be complex and fragmented, spanning many settings and providers. Though the Institute of Medicine cites fragmentation as a priority for improving cancer care, little is known about aspects of fragmentation that affect cancer outcomes. We examine two features which may have different implications for improving care delivery: receiving surgical and medical oncologic care from providers affiliated with different hospitals; and receiving care from a surgeon-oncologist pair who infrequently shares patients with one another. We focus on stage III colon cancer patients as guidelines recommend both surgery and chemotherapy. METHODS: Patients with stage III colon cancer diagnosed between 2000-2005 were identified from SEER- Medicare data. Patients were assigned to their operative surgeon and the medical oncologist who billed for the plurality of their visits in the year following diagnosis. Surgeons and oncologists were linked to the hospital where they billed most for inpatient care. Patients were classified as experiencing "hospital fragmentation" if their surgeon and oncologist were assigned to different hospitals (versus the same hospital). We determined the number of patients each surgeon-oncologist pair shared; patients were classified as having "high patient- sharing" physicians if their oncologist and surgeon shared many patients (top quartile of the shared patients distribution versus lower three). Patient-sharing has been validated as a measure of collaboration and information exchange between physicians. Our primary outcome was all cause mortality (censor date 12/31/2007). Secondary outcomes included timely receipt of chemotherapy (9 months of diagnosis) and cost of care [total claims using MEDPAR (Part A), NCH (Part B), and Outsaf files)] in the 12 months following

• diagnosis. We used Cox proportional hazard and regression models, adjusted for patient demographics,

socioeconomic status, comorbidities, SEER site; surgeon's yearly procedure volume, oncologist patient panel size; and hospital characteristics (volume; NCI, for-profit, and academic status). Generalized estimating equations and robust standard errors were used to account for hierarchical data and clustering. RESULTS: Our sample included 7443 patients. Median survival was 3.04 years. One-third (N=2471) received care from surgeons and oncologists associated with different hospitals. No difference in morality was associated with hospital fragmentation (adjusted HR=1.00, 95% CI: 0.93-1.07). We observed an increased risk of death among patients whose surgeons and medical oncologists shared few patients (lower three quartiles of shared patients) compared to those whose doctors were in the top quartile (HR=1.15, 1.06-1.25). No statistical interaction between the two predictors was noted in the final model (Wald's test, p=0.324). Neither hospital fragmentation nor patient-sharing predicted timely receipt of chemotherapy or 12 month costs of care. CONCLUSIONS: Receiving care from physicians associated with the same hospital did not improve survival for stage III colon cancer patients, whereas receiving care from physicians sharing many patients with another

• did. These results suggest that efforts to improve care fragmentation need to examine the informal relationships

between physicians that may be reflected by patient-sharing. Further, cancer survival, quality, and costs may not be improved by delivery redesign which addresses fragmentation solely by consolidating options for care to one institution.

PROSTATE CANCER PATIENTS: THE RELATIONSHIP WITH RACE/ETHNICITY AND HEALTH LITERACY Celia P. Kaplan1,2; Anna Napoles1,2; Steven Gregorich1; Tung T. Nguyen1,2; Jennifer Livaudais-Toman1; Eliseo J. Perez- Stable1,2; Eric J. Small1,2; Mack Roach2,3. 1University of California San Francisco, San Francisco, CA; 2University of California San Francisco, San Francisco, CA; 3University of California San Francisco, San Francisco, CA. (Tracking ID #1935608) BACKGROUND: Clinical trials provide the foundation for advances in cancer diagnostics and therapeutics and are the major channel for translating treatment-related discoveries in cancer care into clinical practice. To ensure that the benefits and burdens of this research are distributed fairly among all cancer patients and that research findings are generalizable, it is important that minorities participate in clinical trials. However, enrollment of African Americans and Latinos remains lower than that of non-Latino whites. Limited health literacy, more common among minorities than whites, is a key barrier to clinical trial recruitment. We sought to examine the association between race/ethnicity, health literacy and stated willingness to participate in clinical trials among prostate cancer patients. METHODS: Potential participants consisted of men diagnosed with prostate cancer in 2008 between the ages of 18 and 75, recruited through the California Cancer Registry, living within 60 miles of a site offering access to one or more clinical trials in eight California counties: Alameda, Contra Costa, Marin, Monterey, San Francisco, Santa Clara, Santa Cruz and Los Angeles Counties. Between November 2011 and November 2012, we identified all cases of prostate cancer in African American, Latino and Asian American patients and a random sample of prostate cancer cases in white men. Participants completed a 30-minute telephone interview in English, Spanish or Chinese (Mandarin or Cantonese) to assess their attitudes, knowledge and willingness to participate in prostate cancer clinical trials. Health literacy was assessed with a 3-item previously validated scale that asked about comfort reading hospital materials, filling out medical forms, and understanding written information (scale scores ranged from 1-5, dichotomized as <3.5 = low literacy, ≥3.5 = medium- high literacy). In multivariable analyses, logistic regression was used to estimate effects of ethnicity and health literacy on willingness to participate. We estimated odds ratios [OR] and 95% confidence intervals [CI]. RESULTS: A total of 1,869 men were contacted and the final sample included 855 men for a response rate of 46%. More than half (52%) were age 65 years or older. The ethnically diverse sample included white (42%), Latino (24%), African American (19%) and Asian American men (15%). More than 50% completed college. Nearly 70% were born in the U.S. The majority of men (98%) were insured and 81% had medium to high health literacy (composite score≥3.5). Only 17% believed that "clinical trials are beneficial" but 75% believed that "clinical trials are important", while only 38% expressed a willingness to participate in clinical trials. Bivariate and multivariable analysis results are presented in Table 1. CONCLUSIONS: Among men with prostate cancer, African American and Latino men expressed a greater willingness to participate in prostate cancer clinical trials than non-Latino white men. Health literacy was not associated with willingness to participate. Given lower rates of clinical trial participation in these racial/ethnic groups, our findings suggest that factors other than willingness to participate affect actual participation and need to be explored.

Table 1. Bivariate and multivariable analyses: Willingness to participate in prostate cancer clinical trials Willing to participate % OR (95% CI) aOR* (95% CI) Race/Ethnicity Non-Latino white 34% Ref Ref African American 44% 1.31 (0.89-1.92) 1.69 (1.01-2.82)° Asian 30% 0.85 (0.55-1.32) 0.86 (0.46-1.61) Latino 55% 2.21 (1.55-3.14)° 1.75 (1.01-3.05)° Health Literacy Low (score<3.5) 48% Ref Ref Medium to High (score≥3.5) 39% 0.75 (0.53-1.07) 1.34 (0.77-2.34)

*Analysis is also adjusted for age, marital status, education, nationality, employment, health insurance, prior research participation, knowledge of clinical trials, attitudes towards clinical trials, and Gleason score. °p<0.05

MAMMOGRAPHY RATES 3 YEARS AFTER THE USPSTF GUIDELINES CHANGES James F. Wharam1; Bruce E. Landon2; Fang Zhang1; Xin Xu1; Steve Soumerai1; Suzanne Fletcher1; Dennis Ross-Degnan1. 1Harvard Medical School, Boston, MA; 2Harvard Medical School, Boston, MA. (Tracking ID #1939213) BACKGROUND: In November 2009, the United States Preventive Services Task Force (USPSTF) changed from recommending mammography every 1-2 years among women age ≥40 to (1) recommending personalized screening decisions for women age 40-49 and (2) recommending screening every 2 years for women age 50-74. Our objective was to determine the impact of the updated USPSTF guidelines on mammography rates. METHODS: We performed a retrospective before-after time-series analysis, comparing mammography rates in 2012 to those predicted based on 2005-2009 trends. The sample included 1.71-2.08 million women annually age 40-64 from a large US health plan. We assessed annual and biennial mammography rates, examining for the relative change between observed and predicted rates in 2012. RESULTS: Among women age 40-49, adjusted annual mammography rates increased from 40.7% to 44.4% between 2005 and 2009 then declined to 42.3% by 2012 (Figure 1), an estimated 9.55% reduction (95% CI, - 10.10% to -8.99%) relative to the rate predicted by 2005-9 trends. Black women age 40-49 experienced a lesser decrease of 3.27% (-5.22% to -1.28%, Figure 2). Yearly mammography rates among women age 50-64 increased from 48.7% to 51.3% from 2005 to 2009 but then declined by 5.87% (-6.32% to -5.42%) to 49.7% by

• 2012. For biennial screening, adjusted rates in 2005-6, 2008-9, and 2011-12 among women age 40-49 were

66.4%, 68.1%, and 64.7%, respectively, a 7.86% relative reduction (-8.49% to -7.22%). White, Hispanic, and Asian women age 40-49 experienced similar relative reductions of -8.69% (-9.41% to -7.97%), -7.33% (-9.35% to -5.27%), and -9.24% (-12.40% to -5.98%), respectively, whereas black women experienced a nonsignificant reduction (-1.75%, [-3.89% to 0.44%]). Biennial mammography rates among women age 50-64 were 73.4%, 74.0%, 72.0% in 2005-6, 2008-9, and 2011-12, respectively, a 4.39% relative reduction (-4.89% to -3.88%) that was similar among white, Hispanic, and Asian women. Black women age 50-64 experienced a small increase in biennial mammography (2.90%, [1.24% to 4.58%]). CONCLUSIONS: After publication of the 2009 USPSTF guidelines, annual and biennial mammography rates declined by 4-17% among women who were not black. Black women generally experienced lesser decreases. Reductions in biennial mammography might be an unintended consequence of the updated guidelines.

Figure 1. Annual mammography rates by age group before and after publication of the 2009 USPSTF guidelines Figure 2. Annual mammography rates among black women by age group before and after publication of the 2009 USPSTF guidelines

SCREENING FOR HEPATOCELLULAR CARCINOMA IN CHRONIC LIVER DISEASE: A SYSTEMATIC REVIEW Amirala Pasha2; Devan Kansagara1,2; Joel Papak1,2; Maya O'Neil1; Michele Freeman1; Rose Relevo1; Ana Quinones2; Makalapua Motu'apuaka1; Janice H. Jou1,2. 1Portland VA Medical Center, Portland, OR; 2Oregon Health and Science University, Portland, OR. (Tracking ID #1939197) BACKGROUND: Hepatocellular carcinoma (HCC) incidence and mortality has increased over the past four decades abroad and in the United States, with localized tumors accounting for most of the increase guidelines recommend routine surveillance for hepatocellular carcinoma (HCC) in high-risk patients. However, recommendations for HCC screening remain controversial in part because of concerns over the quality and paucity of existing evidence, and because there have been concerns raised about overdiagnosis and patient harms in other cancer screening programs. We conducted a systematic review of the published literature to better understand the incremental benefits and harms of routine HCC screening compared to clinical diagnosis, and of treating HCC found as a result of screening. METHODS: We searched Medline, PsycInfo, and Cochrane databases to March 2013; clinical trial registries; reference lists; and technical advisors. We included clinical trials and observational studies comparing screening to no screening, and clinical trials comparing different screening intervals. We also included clinical trials and

• bservational studies comparing active to conservative treatment in patients with early-stage HCC. Mortality

and adverse events were the main outcomes of interest. Because there were few comparative studies on treatments other than transarterial chemoembolization (TACE), we also examined noncomparative

• bservational studies. The quality of studies and the overall strength of evidence were dual-reviewed using

published criteria. RESULTS: Of 11,321 citations, 264 were reviewed at the full-text level. Thirty-six studies contained primary data relevant to the efficacy of HCC screening or treatment of early-stage HCC and met our inclusion criteria. We also examined two systematic reviews of treatment modalities. Two trials and sixteen observational studies provide very low strength evidence on the effects of screening on mortality because of methodologic flaws and issues of lead- and length-time bias. Moderate strength evidence from two trials shows no advantage of shorter screening intervals (3-4 months compared to 6-12 months). None of the included screening studies reported harms of screening. Low strength evidence from a meta-analysis of 8 studies found a risk of needle-track seeding from liver biopsy of 2.7%. One recent systematic review of the diagnostic accuracy of imaging for HCC screening and diagnosis found very few studies reporting harms data: one study found contrast-enhanced CT was associated with adverse events in 13-15% of patients, while another found mild-moderate adverse events in 25% patients receiving gadoxetic acid-enhanced MRI. We found no studies evaluating the psychologic harms of screening. No studies specifically enrolled patients with screen-detected HCC, so we examined studies of patients with early-stage HCC as a way of approximating screen-detected disease. Overall, there is little evidence with which to draw conclusions about the net benefits of actively treating early-stage HCC compared to conservative treatment. Observational studies show that patients selected for treatment with OLT, resection, or RFA had good long-term survival that was substantially higher than patients not selected for such therapy, but it is unclear whether such effects reflect a true effect of treatment or reflect confounding by

• indication. Serious harms occurred in 3-20% of patients, depending on the intervention.

CONCLUSIONS: There is very low strength evidence from which to draw conclusions about the effects of HCC screening on mortality in high-risk patients with chronic liver disease. Screening tests can identify early stage HCC and patients who are selected for surgical treatment often have good long-term survival, but some treatments may be associated with substantial harms. There is very limited evidence to draw firm conclusions about the incremental benefits of using routine screening to identify HCC and treat HCC found as a result of routine screening. Trials examining the balance of benefits and harms of HCC screening in patients with chronic liver disease should be considered.

RACIAL/ETHNIC DIFFERENCES IN PALLIATIVE CARE CONSULTATION FOR INPATIENTS WITH ADVANCED CANCER Rashmi K. Sharma1; Kenzie A. Cameron1; Jamie Von Roenn1; Joan S. Chmiel1; Holly G. Prigerson2; Frank

• Penedo1. 1Northwestern University, Chicago, IL; 2Cornell Medical College, New York City, NY. (Tracking ID

#1926304) BACKGROUND: Racial/ethnic minorities with advanced cancer are less likely to receive palliative care and hospice services than non-Hispanic whites (NHW) and more likely to receive intensive, non-curative treatment at the end of life, yet little is known about racial/ethnic differences in rates of inpatient palliative care consultation (PCC). We sought to evaluate the association between race/ethnicity and PCC for advanced cancer patients. METHODS: We used the electronic medical record (EMR) to identify all patients with metastatic cancer hospitalized between January 1, 2009 and December 31, 2010 at an urban academic medical center. Patient- level multivariable logistic regression was used to evaluate the association between race/ethnicity and PCC. RESULTS: 6288 advanced cancer patients (69% NHW, 19% African American (AA), 6% Hispanic) were hospitalized during the study period of which 1086 patients had a PCC. More AA (22%) and Hispanic patients (20%) received an inpatient PCC compared to NHW patients (16%) (p < 0.001). In univariate analyses, AA race/ethnicity (OR 1.49, 95% CI 1.27-1.74), Hispanic ethnicity (OR 1.33, 95% CI 1.01-1.74), and insurance [Medicaid (OR 1.87, 95% CI 1.48-2.36) and Medicare (OR 1.24, 95% CI 1.07-1.42), each vs. private insurance] were associated with higher odds of PCC. Greater severity of illness (measured using the APR-DRG severity of illness score) was linearly associated with higher odds of PCC (trend test p < 0.001). After adjusting for insurance status, race/ethnicity, prior hospitalization, and severity of illness, AA and Medicaid patients had higher odds (OR 1.28, 95% CI 1.08-1.52, and OR 1.60, 95% CI 1.23-2.09) of inpatient PCC than NHW and private insurance patients, respectively. CONCLUSIONS: Hospitalized AA patients with advanced cancer were more likely to receive a PCC than NHW patients, even after adjusting for insurance status and severity of illness, although rates of PCC remained

• low. Our findings of increased receipt of PCC among AA patients in the hospital appear contrary to studies

reporting decreased access to palliative care services by racial/ethnic minorities in the outpatient setting. Better understanding of the factors that are driving higher rates of inpatient PCC rates for AA patients with advanced cancer may thus help to inform the development of interventions to improve access to palliative care for AA patients across other settings.

Factors associated with inpatient palliative care consultation for patients with advanced cancer

Variables Bivariate Analysis Multivariate Analysis* OR 95% CI OR 95% CI Race Non-Hispanic White (reference) African American 1.49 1.27-1.74 1.28 1.08-1.52 Hispanic 1.33 1.01-1.74 1.20 0.91-1.60 Insurance Private (reference) Medicaid 1.87 1.48-2.36 1.60 1.23-2.09 Medicare 1.24 1.07-1.42 1.03 0.89-1.20 Other 0.92 0.63-1.36 0.95 0.63-1.43 APR-DRG Severity of illness 1: Minor severity (reference) 2: Moderate severity 2.00 1.56-2.58 1.80 1.39-2.33 3: Major severity 3.97 3.10-5.08 3.38 2.62-4.37 4: Extreme severity 7.55 5.73-9.95 7.19 5.39-9.59