Monday, October 24, 2016: 2:00 PM - 3:30 PM
Bayshore Ballroom Salon E, Lobby Level (Westin Bayshore Vancouver)

Karen M. Kuntz, ScD
University of Minnesota School of Public Health
Health Policy and Management

2:00 PM

Daniel S. Reuland, MD, MPH1, Alison Brenner, PhD2, Richard Hoffman, MD MPH3, Andrew McWilliams, MD MPH4, Robert Rhyne, MD MPH5, Hazel Tapp, PhD6, Mark Weaver, PhD7, Danelle Callan5, Khalil Harbi, MPH7, Brisa Hernandez, BUS6 and Michael Pignone, MD, MPH1, (1)University of North Carolina School of Medicine, Chapel Hill, NC, (2)Cecil G Sheps Center for Health Services Research, University of North Carolina at Chapel Hill, Chapel Hill, NC, (3)University of Iowa, Iowa City, IA, (4)Carolinas Health Care, Charlotte, NC, (5)University of New Mexico School of Medicine, Albuquerque, NM, (6)Carolinas Healthcare System, Charlotte, NC, (7)University of North Carolina at Chapel Hill, Chapel Hill, NC

Colorectal cancer (CRC) screening decision aids and patient navigation address multiple different screening barriers and are potentially complementary interventions. We conducted a randomized controlled trial to determine the effect of a CRC screening decision aid plus patient navigation on decision making and screening completion outcomes in a diverse patient population. 


We enrolled patients aged 50-75 who were not up-to-date with recommended CRC screening and were attending primary care visits at safety-net clinics in Charlotte, NC and Albuquerque, NM. After a baseline survey, participants were randomized to view either a multi-media CRC screening decision aid in English or Spanish that presented colonoscopy and stool testing (FOBT/FIT) as screening options or a food safety video.  After the provider encounter, intervention patients received support for screening completion from a bilingual/bicultural patient navigator. Decision making outcomes included screening-related knowledge, discussions with the provider, and test preferences.  Screening completion was assessed by blinded chart review at six months. We used generalized estimating equations, controlling for study site and clustering by provider, to test for differences between groups. We used logistic regression to examine relationships between decision making outcomes and screening completion. 


Participant characteristics (n=264): mean age 58 years, 65% female, 61% Latino (71% of whom preferred Spanish); 16% White non-Latino; 16% Black non-Latino; 77% household income under $20,000; 40% low literacy; 30% Medicaid; 33% uninsured. Intervention participants had greater CRC screening knowledge, adj-diff 1.8 points out of 6 (95%CI 1.6, 2.1) and reported more screening discussions with providers (84% vs 41%); adj-diff 27% (95%CI 16%, 38%), including more frequent discussions of both FOBT/FIT and colonscopy options (24% vs 8%). Intervention participants were more likely to indicate a specific screening test preference (93% vs 65%); adj-diff 27% (95%CI 16%,38%) and more likely to prefer FOBT/FIT (67% vs. 46%).  Intervention participants were also more likely to complete CRC screening within 6 months: (66% vs 28%); adj-diff 38% (95%CI 27%, 49%). Screening completion was higher among those having greater knowledge (OR=1.28; 95%CI 1.05, 1.58), CRC screening discussions (OR=2.97; 95%CI 1.21, 4.38), and a specific screening test preference OR=3.6; (95%CI 1.7, 7.6).


A combined decision aid plus patient navigation intervention substantially increases both CRC screening completion and the quality of the decision making process in linguistically diverse, vulnerable patients.

2:15 PM

Mathyn Vervaart1, Emily A. Burger, PhD2 and Eline Aas, PhD2, (1)University of Oslo, Oslo, Norway, (2)Department of Health Management and Health Economics, University of Oslo, Oslo, Norway
Purpose: Different sources of patient heterogeneity (e.g., comorbidities related or unrelated to disease development or intervention effectiveness) may contribute to differential cost-effectiveness profiles when evaluating new healthcare interventions. Using the stratified analysis framework to examine the value of subgroup policies, we demonstrate the importance of capturing patient heterogeneity in colorectal cancer (CRC) screening. 

Methods: We developed a Markov model to capture the potential impacts of patient heterogeneity, represented by comorbidities related and unrelated to CRC development and screening performance, on the cost-effectiveness of CRC screening involving once-only sigmoidoscopy compared to no screening. We simulated cohorts of Norwegian men and women aged 60 years-old until death or 100 years with one of six comorbidity subgroups that differentially influenced the risks of developing CRC, dying from CRC, dying from background mortality or screening-related adverse events. Chronic obstructive pulmonary disease (COPD), dementia, and chronic renal failure reflected patients with “unrelated” comorbidities, while diabetes mellitus, obesity and smoking reflected patients with “related” comorbidities. Input parameters reflected Norwegian life tables adjusted for overall colorectal cancer mortality, and national epidemiologic and economic data. Screening effectiveness, aimed at reducing the future incidence of cancer, was based on a randomized controlled trial in Norway. In addition we reviewed the literature to derive comorbidity-specific risks of colorectal cancer incidence, adverse events, overall- and cancer specific mortality and utility values. For each comorbidity subgroup, we calculated the discounted (4%) incremental cost-effectiveness ratio (ICER), defined as the cost per quality-adjusted life year (QALY) gained, and the resulting net monetary benefit (NMB) gained from stratification, using a societal perspective.

Results: Unrelated comorbidities generally led to less attractive cost-effectiveness ratios (i.e., increased the ICER), while related comorbidity improved the cost-effectiveness profile of screening for CRC. The subgroup-specific ICERs ranged from €17,131 per QALY gained for obesity to €52,150 per QALY gained for dementia, compared to €25,071 per QALY gained for the average general population. Consequently, there was a positive NMB gained by stratification for a range of willingness-to-pay thresholds, indicating the potential value of considering patient subgroups in the analysis.


Conclusions: Decision makers should consider patient heterogeneity when evaluating population-based screening programs and other healthcare interventions, as decisions in favor of providing or rejecting treatment for the entire population can result in inefficient use of resources.

2:30 PM

Mary Linton Peters, MD, MS1, Alvin Jeon, BS2, Curtis Heberle, BS2, Emily Dowling, MHS2, Florian Boulnois, MS3, Chung Yin Kong, PhD4, Daniel Chung, MD5, William Brugge, MD5, Chin Hur, MD, MPH2 and Pari Pandharipande, MD, MPH2, (1)Hematology/Oncology Division, Beth Israel Deaconess Medical Center, Boston, MA, (2)Institute for Technology Assessment, Massachusetts General Hospital, Boston, MA, (3)Massachusetts General Hospital - Institute for Technology Assessment, Boston, MA, (4)Harvard Medical School, Boston, MA, (5)Gastroenterology Division, Massachusetts General Hospital, Boston, MA

   Purpose:   First-degree relatives (FDRs) of patients with pancreatic adenocarcinoma (PAC) have elevated PAC risk, in part due to known genetic mutations.  While MRI screening is effective for early PAC detection in some high-risk populations, false-positive results – and associated, downstream mortality risks – attenuate its benefits.  Our goal was to evaluate the potential of genetic testing to improve screening decisions and life expectancy (LE) in the population of FDRs.

   Methods:   We developed a microsimulation model of PAC, calibrated to Surveillance, Epidemiology, and End Results program data.  We compared two strategies for FDRs: 1) test all FDRs with a panel of genetic mutations (e.g. BRCA2, CDKN2A), and screen with MRI for test+ FDRs if LE gains were achievable; 2) test no FDRs, and screen with MRI (all or none) if LE gains were achievable.  For test+ individuals, our model accounted for mutation-specific relative risks of PAC.  For test+, test-, and no-test scenarios, we simulated multiple screening approaches, defined by starting age and frequency, to determine if any resulted in an increase in LE.  Primary outcomes were LE gains by test outcome, and LE differences for the test vs. no-test decisions.     

Results: For test+ women, MRI screening yielded LE gains for 81% of mutation carriers.  Among those mutation carriers for whom screening could increase LE, maximum LE gains ranged from 1.2 days (BRCA2+ women screened once at age 55) to 156 days (STK11+ women screened annually from age 35). For test- women, MRI screening did not yield LE gains.  For no-test women, the LE gain from screening was minimal (0.18 days or less) regardless of the screening approach.  For test+ men, MRI screening also yielded LE gains for 81% of mutation carriers.  For test+ men, maximum LE gains ranged from 1.50 days (BRCA2+ men screened once at age 55) to 178 days (STK11+ men screened annually from age 35)  For men in the no-test and test- groups,  LE gains from screening were minimal (0.73 days and 0.44 days, respectively). 

   Conclusions:   Genetic testing of FDRs can target MRI screening for PAC to higher risk populations.  Timing and frequency of screening can be optimized by genetic test result, sparing morbidity and adding life expectancy.

2:45 PM

Hilary Whitham, PhD, MPH1, Stephen Hawes, PhD, MS2, Fernando Alarid-Escudero, MS, PhD Candidate3, Alan Lifson, MD, MPH1, J. Michael Oakes, PhD1, Haitao Chu, PhD4 and Shalini Kulasingam, PhD1, (1)Division of Epidemiology & Community Health, University of Minnesota, Minneapolis, MN, (2)Department of Epidemiology, University of Washington, Seattle, WA, (3)Division of Health Policy and Management, University of Minnesota, Minneapolis, MN, (4)Division of Biostatistics, University of Minnesota, Minneapolis, MN

   Purpose: The interaction between human immunodeficiency virus (HIV) and human papillomavirus (HPV) results in an increased burden of cervical cancer among HIV-positive women. However, optimal cervical cancer prevention strategies specific to the high-risk population of HIV-positive women remain unknown due, in part, to limited natural history data. Using data from a series of longitudinal natural history studies conducted in Senegal, West Africa among HIV-positive women, we quantify the costs and benefits of targeted screening strategies relevant to sub-Saharan Africa and identify optimal strategies across a wide range of willingness-to-pay (WTP) thresholds.

   Methods: Using a Markov cohort model, we quantified the relative cost-effectiveness of six screening strategies (Hybrid Capture 2 HPV testing, rapid HPV testing, cytology, visual inspection with acetic acid (VIA), HPV testing followed by cytology triage, and HPV testing followed by VIA triage) and five screening frequencies using projected life expectancy and incremental cost-effectiveness ratios (ICER) over a 15-year time horizon. Natural history inputs were estimated from data on over 600 HIV-positive women who were followed for an average of two years between 1994 and 2010. Cytology and HPV DNA testing were performed at approximately 4-month intervals. Competing risk modeling was used to estimate transition probabilities between clinically relevant precancerous natural history stages. Transitions from precancerous stages to cancer were calibrated using a Bayesian Markov chain Monte Carlo approach. One-way and probabilistic sensitivity analyses were conducted to explore the impact of uncertainty on results. Following WHO criterion, strategies with ICERs less than Senegal's 2013 per capita Gross Domestic Product (GDP) were considered ‘very cost-effective' (I$1,050) and strategies less than three times the per capita GDP were considered ‘cost-effective' (I$3,150).

   Results: In base case analyses, VIA was the most cost-effective strategy examined. Compared to no screening, annual VIA resulted in a discounted increased life expectancy of 1.9 months and a 38% reduction in cervical cancer incidence with an ICER of I$1,309 per life year saved. When accounting for uncertainty in all inputs (see Figure 1), VIA emerged as the most likely to be cost-effective at a WTP threshold greater than I$1,440. High underlying HPV prevalence among HIV-positive women substantially reduced the cost-effectiveness of HPV testing.

   Conclusion: Targeted cervical cancer screening of HIV-positive women using VIA represents an important prevention opportunity among this high-risk population.

3:00 PM

Jennifer Mason Lobo, PhD and Tracey Krupski, MD, University of Virginia, Charlottesville, VA

   To evaluate cost, radiation exposure, and cancer control of imaging surveillance strategies to screen for secondary renal cell carcinoma (RCC).


   We present a Monte Carlo simulation model to assess clinical outcomes and costs of post-surgery screening for recurrent RCC.  Imaging surveillance includes abdominal imaging (ultrasound, abdominal computerized tomography (CT)) and chest imaging (chest x-ray (CXR), chest CT).  Numerical results are presented for low risk patients who have undergone partial nephrectomy to manage the primary tumor. We compare hypothetical strategies to current guidelines from four clinical organizations: American Urological Association (AUA), Canadian Urological Association (CUA), European Association of Urology (EAU), and National Comprehensive Cancer Network (NCCN).  Hypothetical strategies vary in the number of years of follow-up (3, 4, or 5 years), modalities (ultrasound and CXR, or abdominal and chest CT), and frequency of imaging (1 or 2 studies in each location per year). Costs, recurrence information, radiation exposure estimates, and modality-specific detection rates were taken from secondary sources. Model outcomes are percentage of recurrent patients diagnosed, size of the recurrent tumor at diagnosis, radiation exposure due to imaging, and cost of screening.


   CUA guidelines have the lowest average cost for recurrent patients ($435.88), the lowest average radiation exposure (22.96mSv), and the highest percentage of recurrent patients diagnosed (95%, same as EAU); however, CUA has the largest tumor size at recurrence (2.13cm).  EAU has similar cancer control outcomes to CUA with increased costs ($1297.60) and radiation (65.06mSv).  AUA and NCCN diagnose the smallest percentage of recurrent patients, primarily due to follow-up duration, though the tumor size at diagnosis is improved (1.64cm and 1.49cm, respectively).  The percentage of patients diagnosed with the hypothetical strategies varies (45.9-95.1%); longer follow up and more precise CT imaging are associated with higher rates.  Strategies with CT imaging have smaller tumor size at diagnosis, increased costs, and higher radiation exposure compared to strategies with ultrasound and CXR.  Higher numbers of imaging studies per year provide little benefit.


   CUA provides a balance of a high percentage of diagnoses with low costs and radiation.  The analysis of hypothetical strategies shows that surveillance should be continued for 5 years, and a single imaging study in each location is sufficient.  Imaging modality should be chosen based on available budget and cancer control goals.