Potential Cost-Effectiveness of Risk-Based Pancreatic Cancer Screening in Patients With New-Onset Diabetes

View More View Less
  • 1 CHOICE Institute, Department of Pharmacy, University of Washington, Seattle, Washington;
  • | 2 Pancreatic Cancer Action Network, Manhattan Beach, California;
  • | 3 Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, Washington; and
  • | 4 Department of Gastroenterology and Nutrition, Division of Internal Medicine, The University of Texas MD Anderson Cancer Center, Houston, Texas

Background: There are no established methods for pancreatic cancer (PAC) screening, but the NCI and the Pancreatic Cancer Action Network (PanCAN) are investigating risk-based screening strategies in patients with new-onset diabetes (NOD), a group with elevated PAC risk. Preliminary estimates of the cost-effectiveness of these strategies can provide insights about potential value and inform supplemental data collection. Using data from the Enriching New-Onset Diabetes for Pancreatic Cancer (END-PAC) risk model validation study, we assessed the potential value of CT screening for PAC in those determined to be at elevated risk, as is being done in a planned PanCAN Early Detection Initiative trial. Methods: We created an integrated decision tree and Markov state-transition model to assess the cost-effectiveness of PAC screening in patients aged ≥50 years with NOD using CT imaging versus no screening. PAC prevalence, sensitivity, and specificity were derived from the END-PAC validation study. PAC stage distribution in the no-screening strategy and PAC survival were derived from the SEER program. Background mortality for patients with diabetes, screening and cancer care expenditure, and health state utilities were derived from the literature. Life-years (LYs), quality-adjusted LYs (QALYs), and costs were tracked over a lifetime horizon and discounted at 3% per year. Results are presented in 2020 US dollars, and we took a limited US healthcare perspective. Results: In the base case, screening resulted in 0.0055 more LYs, 0.0045 more QALYs, and $293 in additional expenditures for a cost per QALY gained of $65,076. In probabilistic analyses, screening resulted in a cost per QALY gained of <$50,000 and <$100,000 in 34% and 99% of simulations, respectively. In the threshold analysis, >25% of screen-detected PAC cases needed to be resectable for the cost per QALY gained with screening to be <$100,000. Conclusions: We found that risk-based PAC screening in patients with NOD is likely to be cost-effective in the United States if even a modest fraction (>25%) of screen-detected patients with PAC are resectable. Future studies should reassess the value of this intervention once clinical trial data become available.

Background

Pancreatic cancer (PAC) is the third most common cause of death due to cancer in the United States,1 with a 5-year overall survival of approximately 10%.2 PAC survival prognosis is poor irrespective of stage at diagnosis and whether the extent of disease allows for surgical treatment (hereafter, “resectable”), but patients diagnosed with stage I resectable disease do fare significantly better (26-month median cancer-specific survival) versus, for example, those diagnosed with unresectable stage IV disease (4.8-month median cancer-specific survival).3 These large differences in survival motivate the need for effective PAC early-detection strategies that can identify cases before disease spread.

The primary challenge surrounding the development and implementation of effective screening for idiopathic PAC is the low incidence of disease. Annualized PAC incidence is approximately 13.1 patients per 100,000 individuals across the entire US population, but 90% of patients are aged ≥50 years (∼37 patients per 100,000 individuals).2 Due to the low incidence of PAC, even highly specific tests could lead to a substantial number of false-positive screening results. In turn, this large group would potentially experience distress about a potential PAC diagnosis, receive unnecessary and potentially harmful subsequent diagnostic evaluation, and incur substantial medical costs. Because broad age-based PAC screening criteria may result in net harm and/or low-value medical expenditure, it is likely necessary to optimize candidate PAC screening strategies by focusing initial eligibility on the readily identifiable highest-risk population subgroups.4

In recent years, there has been an increased effort to identify those individuals at sufficient risk of developing PAC to possibly experience a net benefit from screening.5 One early-detection strategy under investigation is focusing on patients with new-onset diabetes (NOD), a factor that has shown an association with PAC risk across a range of clinical, epidemiologic, and laboratory research. Hyperglycemia occurs in approximately 85% of patients with PAC, and diabetes is present in 47% of patients with PAC.6,7 In one study, among most patients (75%) with PAC and diabetes, the diabetes was new-onset (<3 years since diagnosis).6 Furthermore, approximately 25% of patients with PAC are diagnosed with diabetes 6 to 36 months before the PAC diagnosis.6

The Enriching New-Onset Diabetes for Pancreatic Cancer (END-PAC) model is a statistical framework that uses age, change in blood glucose, and change in weight to predict PAC risk in patients with NOD. Using an END-PAC cutoff score ≥3 has shown good sensitivity (78%) and specificity (92%) for identifying PAC with CT imaging in a validation study.8 The outcomes of a screening strategy utilizing the END-PAC model are now being evaluated in an ongoing prospective trial (Early Detection Initiative [EDI]), sponsored by the Pancreatic Cancer Action Network [PanCAN]). The EDI study randomizes individuals aged 50 to 85 years with glycemically defined NOD to observation or intervention arms, with intervention defined as applying END-PAC scoring and pancreas-protocol CT imaging to those with an END-PAC score >0. The primary endpoint is a reduction in the percentage of stage III/IV PAC diagnoses. If the risk-based screening strategy using END-PAC proves effective in the EDI trial, then economic evidence will be needed to show its value to a range of policy, payer, and clinical guideline/recommendation bodies. Early-stage cost-effectiveness analysis—an evaluation of the potential value of a medical intervention preceding or concurrent to assessment of clinical effectiveness—is a recommended approach that allows stakeholders to ascertain the potential value of medical interventions, understand key drivers of value, and develop/refine value propositions so that they are ready concurrent to pivotal trial evidence.9

The objective of this study was to develop an early-stage cost-effectiveness model to assess the potential value of risk-based CT PAC screening using the END-PAC model in patients with NOD. This study establishes a decision model framework that can be iteratively refined. Additional data from the EDI trial, when available, will provide initial insights into the potential value of the EDI trial screening strategy and can help identify the major drivers of screening value, thereby informing future research and data collection efforts.

Methods

Methods Overview

We created an integrated decision tree and Markov state-transition model framework in Microsoft Excel (version 16.16.12, Microsoft Corporation) to assess the cost-effectiveness of risk-based PAC screening of patients with NOD aged ≥50 years using standard-contrast CT (pancreas-protocol) imaging versus usual care (no screening) (Figure 1). As in the PanCAN EDI trial, patients diagnosed with diabetes who also met the following criteria were considered to have NOD: (1) fasting glucose or hemoglobin A1c level meeting glycemic criteria within 90 days before randomization, (2) ≥1 fasting glucose or hemoglobin A1c values measured in the past 18 months, (3) did not previously meet glycemic criteria for diabetes, and (4) were not previously treated with antidiabetes medications. The date of the first abnormal glycemic parameter was considered as the date of onset of diabetes. The decision tree (Figure 1A) shows 2 different strategies applied to cohorts of patients meeting these inclusion criteria. In the screening strategy, patients with END-PAC scores >0 received CT screening for PAC, and those with END-PAC scores ≤0 did not receive screening. Because patients in the usual care strategy did not undergo screening, those with PAC in this cohort were detected at standard community rates (where there was no screening).2 After being stratified as a PAC case or noncase, patients in both strategies entered respective Markov state-transition models for tracking long-term clinical and economic outcomes (Figure 1B and C). These state-transition models track patient time in monthly cycles across a set of predefined health states using subgroup-specific transition probabilities.

Figure 1.
Figure 1.

Simplified schematic of (A) model decision tree, (B) PAC case, and (C) noncase Markov state–transition frameworks.

Abbreviations: END-PAC, Enriching New-Onset Diabetes for Pancreatic Cancer; NOD, new-onset diabetes; PAC, pancreatic cancer.

Citation: Journal of the National Comprehensive Cancer Network 2021; 10.6004/jnccn.2020.7798

The model framework was used to calculate the deterministic and probabilistic estimates of life-years (LYs), quality-adjusted LYs (QALYs), and direct medical expenditures over a lifetime horizon in each of the screening strategies. Costs were analyzed in 2020 US dollars, and both costs and outcomes were discounted at 3% per year following the recommendation of the Panel on Cost-Effectiveness in Health and Medicine.9 The analysis was performed from a limited US healthcare payer perspective.

Clinical Characteristics and Overall Survival

The prevalence of PAC and the sensitivity and specificity of the END-PAC score were calculated using results from the END-PAC validation study,8 which included a retrospective cohort of 1,561 patients diagnosed with NOD between 2000 and 2015 in Olmstead County, Minnesota, and the surrounding 27 county areas,10,11 in addition to a prospective cohort of 100 patients recruited into a pilot screening study for PAC-NOD.12 Stage distribution for patients detected in the no-screening strategy and PAC survival groups were derived from SEER statistics for PAC cases diagnosed 2000 to 2007.2,13 Stage distribution for patients detected via screening was based on assumptions derived from expert opinion due to the lack of data. Survival in patients with NOD without PAC was derived from a report of life expectancy and cause-specific mortality ratios for patients with diabetes (vs general population) from the peer-reviewed literature14 applied to life tables from the US Life Tables.15

Health-Related Quality of Life

The model framework incorporated health state utility values specific to events before and after PAC diagnosis. The baseline health state utility (before screening) was derived from a cost-effectiveness analysis of PAC surveillance in the United States that used Audit of Diabetes Dependent Quality of Life data from a study of 99 patients who had undergone total pancreatectomy at the Mayo Clinic from 1985 through 200216 and was applied for 2 weeks.17 Because no PAC screening–specific references were available, utilities for true-positive, false-positive, and true-negative screening results were based on a study of routine mammography screening and applied for a period of 2 weeks.18 Preprogression and postprogression health state utilities for PAC cases were derived from a clinical trial of neoadjuvant FOLFIRINOX (fluorouracil/leucovorin/irinotecan/oxaliplatin) for patients with borderline resectable or locally advanced PAC19 and a clinical trial of lanreotide in patients with metastatic gastroenteropancreatic neuroendocrine tumors,20 respectively.

Direct Medical Expenditures

Estimated screening and diagnostic procedure costs, PAC treatment costs, end-of-life costs, and diabetes management costs were derived from the 2020 CMS Physician Fee Schedule21 and peer-reviewed literature2224 (Table 1). Treatment costs included surgery, radiation, and chemotherapy.22 Diabetes care costs included treatments, comorbidity management, and hospitalizations as documented in an analysis of 7,109 people with type 2 diabetes participating in the Translating Research Into Action for Diabetes study.23,25 The National Health Expenditures Deflator was developed by the Centers for Medicare & Medicaid Services.26

Table 1.

Model Inputs

Table 1.

Outcomes

The model framework was used to estimate LYs, QALYs, direct medical expenditures for screening, and usual care strategies. These outcomes were used to calculate the cost per LY and QALY gained for screening versus usual care. The QALY is considered the gold standard outcome in cost-effectiveness research9 because it adjusts life expectancy by a quality-of-life utility score.27 A utility score of 0 represents the value for death, and 1 represents the value for perfect health. The incremental cost-effectiveness ratio (ICER)—the ratio of the difference in costs between strategies and the difference in effects (QALYs) between strategies—was calculated using the following equation:

ICER=Total costscreeningTotal costnoscreeningQALYscreeningQALYno screening

Sensitivity Analyses

We conducted a one-way sensitivity analysis to identify the most influential model inputs on the cost per QALY gained. We also conducted a probabilistic analysis with 10,000 simulation runs to evaluate the cost per QALY gained and the range of incremental outcomes based on uncertainty in all input parameters. The probabilistic analysis drew input values from distributions defined by 95% confidence intervals and standard uncertainty ranges (probabilities = beta; counts and costs = normal; hazard ratios = log-normal), and the results were used to develop a cost-effectiveness acceptability curve.

Scenario Analyses

We also explored a series of scenarios intended to highlight the effects of uncertainty in key model inputs on the potential value of risk-based PAC screening in patients with NOD. Specifically, we explored scenarios with varied inputs for the fraction of resectable screen-detected PAC cases (25%–55%), the fraction of patients with NOD with an END-PAC score >0, and the prevalence of PAC among patients with NOD with an END-PAC score >0 (Figure 2). We also explored scenarios with 1 or 2 screening CT scans among patients with an END-PAC score >0. This approach reflected differences in the design of the END-PAC validation study,8 which involved a single CT scan, and the EDI trial, which involves a second CT scan approximately 6 months after the initial screening CT. A second CT scan may identify additional cases, but it would do so at substantial additional cost. Thus, it was important to assess the value of an additional CT scan for this screening strategy.

Figure 2.
Figure 2.

Deterministic cost per QALY gained under alternative PAC screening strategy performance scenarios. We assessed this range of scenarios to explore the value implications given the uncertainty about the performance of the risk-based PAC screening strategy before the large-scale EDI randomized trial.

Abbreviations: EDI, Early Detection Initiative; END-PAC, Enriching New-Onset Diabetes for Pancreatic Cancer; NOD, new-onset diabetes; PAC, pancreatic cancer; QALY, quality-adjusted life-year.

Citation: Journal of the National Comprehensive Cancer Network 2021; 10.6004/jnccn.2020.7798

Threshold Analysis

Because the difference in the proportion of patients with resectable PAC between strategies was a key driver of the value of the risk-based screening strategy in NOD, we conducted threshold analyses to assess the proportion of PAC cases that would need to be detected at a resectable stage to achieve a cost per QALY gained of <$100,000. These findings provide important screening performance benchmarks that have implications for the degree of value provided by screening.

In addition, because the prevalence of PAC in screening cohorts is uncertain and influential in cost-effectiveness outcomes, we explored the minimal prevalence of PAC required for the risk-based screening strategy to achieve a cost per QALY gained of <$100,000. These findings will provide stakeholders with insights about the potential value of risk-based screening in NOD if the PAC risk is lower than expected based on initial data.

Additional analyses explored how much PAC care costs would need to increase to exceed a cost per QALY gained of $100,000 with screening. This analysis applied a uniform inflation factor to cancer care costs for all PAC stages and provided insights about how future changes in costs of PAC care may impact the value of risk-based screening in NOD.

We also assessed what additional proportion of screen-detected PAC cases would be necessary among screening participants with an END-PAC score >0 to maintain the same cost per QALY gained as in the base case with 1 screening CT scan. This analysis provides insights about the potential value of implementing a second screening CT scan in those with an END-PAC score >0, as is being done in the EDI trial.

Results

Base Case Deterministic Results in Overall NOD Cohorts

In the base case, screening resulted in 0.0055 more LYs, 0.0045 more QALYs, and $293 in additional expenditures, for a cost per QALY gained of $65,076 (Table 2). When results were restricted to patients diagnosed with PAC in each strategy, screening resulted in 0.67 more LYs, 0.55 more QALYs, and $22,687 in additional expenditures (Table 3).

Table 2.

Base Case Deterministic Model Results for the Overall NOD Cohorts by Screening Strategy

Table 2.
Table 3.

Base Case Deterministic Model Results for PAC Cases Among NOD Cohorts by Screening Strategy

Table 3.

Sensitivity Analyses

In a one-way sensitivity analysis, the most influential inputs on the cost per QALY gained were the proportion of screen-detected PAC cases that are resectable, the health state utility for resectable PAC from 6 months after surgery to progression, and the proportion of clinically detected PAC cases with distant-stage disease (Figure 3).

Figure 3.
Figure 3.

Tornado diagram displaying the 10 most influential inputs on the cost per QALY gained.

Abbreviations: HR, hazard ratio; OS, overall survival; PAC, pancreatic cancer; QALY, quality-adjusted life-year.

Citation: Journal of the National Comprehensive Cancer Network 2021; 10.6004/jnccn.2020.7798

In probabilistic analyses, screening resulted in increased LYs, increased QALYs, and increased cost in >99% of simulations. The cost per QALY gained with screening was <$50,000 and <$100,000 in 11% and 99% of simulations, respectively (Figure 4).

Figure 4.
Figure 4.

Cost-effectiveness acceptability curve displaying the probability of cost-effectiveness of risk-based pancreatic cancer screening in patients with new-onset diabetes and END-PAC score >0 at WTP thresholds of $0 to $300,000 per QALY gained.

Abbreviations: END-PAC, Enriching New-Onset Diabetes for Pancreatic Cancer; QALY, quality-adjusted life-year; WTP, willingness-to-pay.

Citation: Journal of the National Comprehensive Cancer Network 2021; 10.6004/jnccn.2020.7798

Scenario Analyses

In scenario analyses, when 25% of screen-detected PAC cases were resectable, the cost per QALY gained was $99,841, compared with $49,977 when 55% of screen-detected PAC cases were resectable (Figure 2). The addition of a second screening CT scan resulted in a cost per QALY gained of $88,350. The cost per QALY gained was $60,586 when 42% of patients with NOD had an END-PAC score >0 and $69,900 when that fraction was 62%. When the prevalence of PAC among patients with NOD with an END-PAC score >0 was 1.3%, the cost per QALY gained was $70,229 versus $61,252 when the prevalence was 1.9%.

Threshold Analyses

In threshold analyses, at least 25% of screen-detected PAC cases needed to be resectable for the cost per QALY gained with screening to be <$100,000. We also found that PAC care costs would need to increase by 106%, the mortality rates of patients with PAC would need to increase by 45% (hazard ratio, 1.45), and the overall 3-year incidence of PAC in the screening cohort would need to be <0.3% for the cost per QALY gained to exceed $100,000 with screening. Finally, to maintain the base case cost per QALY gained, a second screening CT scan would need to detect an additional 1.4% of the END-PAC >0 population with PAC (1.5 × PAC cases detected with 1 CT scan).

Discussion

We developed a simulation model framework to assess the early-stage cost-effectiveness of risk-based CT screening for PAC in patients with NOD. Although the use of health economic modeling has historically been confined to late stages in medical technology development (eg, after phase III clinical trials), the utility of early-stage assessments is becoming increasingly evident. Preliminary economic evaluations aim to predict economic value and serve to guide decision-making and priority-setting.28 Early-stage economic models provide insight into key drivers of diagnostic value and help steer evidence development and innovation.29 In this early-stage value assessment, we learned that this risk-based screening strategy is likely to be considered an acceptable value across a range of analyses. At a base case cost of $63,063 per QALY gained, this risk-based PAC screening strategy has a high probability of being cost-effective at standard willingness-to-pay thresholds in the United States. Our base case findings assume that 40% of screen-detected patients with PAC would be resectable (vs ∼10%–15% resectable without screening). This increase in the detection of resectable PAC with screening seems highly plausible, but it should be noted that we find that risk-based PAC screening in patients with NOD is likely cost-effective even with more modest fractions (>25%).

Among PAC cases detected in the screening and usual care strategies, screening strategy cases were expected to have average gains of approximately 0.67 LYs and 0.54 QALYs due to an earlier stage at diagnosis. This screening stage shift was accompanied by higher lifetime direct medical expenditures, because patients with resectable PAC live longer and thus accrue higher treatment and monitoring costs. As new high-cost therapies are developed for PAC in the future, they are likely to first enter the market indicated for stage IV disease and may thereby further increase the value of PAC screening if survival gains are not substantial. This will be an important factor to continuously monitor if risk-based PAC screening is determined effective in the EDI trial and moves toward clinical implementation.

The primary limitation of this model is that we sought to project the potential value of a screening strategy that is still under evaluation in the ongoing EDI trial. Although we do not have the benefit of modeling based on the outcomes of the targeted 12,500 EDI study participants, we do know key information about the predictive value of the END-PAC risk stratification model based on findings in 2 retrospective cohort studies.8,30 If CT screening according to END-PAC score does not result in a significant stage shift compared with observation alone, or false-positive diagnoses result in substantial additional cost or harm to patients, then this strategy is less likely to be cost-effective. As EDI trial data become available, we will iteratively update the model framework to reassess the value of this screening strategy. It should also be noted that the overall strategy of screening the NOD population for early detection of PAC has limitations, including the dependence on regular screening for blood sugar and weight in the US population. In addition, we recognize that not all PAC is diabetogenic. Furthermore, we did not consider healthcare expenditures unrelated to early detection, diagnosis, and treatment of PAC or diabetes, and therefore took a limited healthcare perspective. The inclusion of such costs would increase model complexity but have a limited impact on cost-effectiveness outcomes given that the vast majority of patients in both strategies do not develop PAC. Finally, our analysis does not consider incidental findings from CT scans in the screening strategy. We did not model these effects due to a lack of evidence about the likelihood of these events and their health and economic consequences. Future analyses should incorporate incidental findings as data emerge from the EDI trial and other sources.

The EDI trial is playing a pivotal role in assessing the feasibility of PAC screening in NOD. The END-PAC score is based on 3 readily available clinical factors, making it easy to identify patients with an elevated PAC risk that warrants clinical workup. In addition, the strategy uses standard CT imaging, a highly accessible screening modality.31 Thus, risk-based screening using the END-PAC score is both straightforward to implement and likely to be cost-effective. Results from the trial will inform our understanding of the true feasibility of this strategy.

Conclusions

This early-stage cost-effectiveness study provides initial insights into the potential value of the EDI trial screening strategy and can help identify the major drivers of screening value, thereby informing future research and data collection efforts. Future studies should reassess the value of this intervention once PanCAN EDI data become available.

Acknowledgments

The authors thank Winona Wright for study coordination and technical assistance with this article; the NIH-supported Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer for helpful suggestions and ancillary study status; and the PanCAN Early Detection Initiative participants and investigators for their important contributions to the development of effective pancreatic cancer screening strategies.

References

  • 1.

    Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin 2019;69:734.

  • 2.

    Howlader N, Noone A, Krapcho M, et al. , eds. SEER cancer statistics review, 1975–2017. Accessed December 23, 2020. Available at: https://seer.cancer.gov/csr/1975_2017/

    • Search Google Scholar
    • Export Citation
  • 3.

    Katz MHG, Hu CY, Fleming JB, et al. Clinical calculator of conditional survival estimates for resected and unresected survivors of pancreatic cancer. Arch Surg 2012;147:513519.

    • Search Google Scholar
    • Export Citation
  • 4.

    Poruk KE, Firpo MA, Adler DG, et al. Screening for pancreatic cancer: why, how, and who? Ann Surg 2013;257:1726.

  • 5.

    Lucas AL, Kastrinos F. Screening for pancreatic cancer. JAMA 2019;322:407408.

  • 6.

    Pannala R, Basu A, Petersen GM, et al. New-onset diabetes: a potential clue to the early diagnosis of pancreatic cancer. Lancet Oncol 2009;10:8895.

    • Search Google Scholar
    • Export Citation
  • 7.

    Chari ST, Leibson CL, Rabe KG, et al. Probability of pancreatic cancer following diabetes: a population-based study. Gastroenterology 2005;129:504511.

    • Search Google Scholar
    • Export Citation
  • 8.

    Sharma A, Kandlakunta H, Nagpal SJS, et al. Model to determine risk of pancreatic cancer in patients with new-onset diabetes. Gastroenterology 2018;155:730739.e3.

    • Search Google Scholar
    • Export Citation
  • 9.

    Sanders GD, Neumann PJ, Basu A, et al. Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: second panel on cost-effectiveness in health and medicine. JAMA 2016;316:10931103.

    • Search Google Scholar
    • Export Citation
  • 10.

    Melton LJ III. History of the Rochester Epidemiology Project. Mayo Clin Proc 1996;71:266274.

  • 11.

    Rocca WA, Yawn BP, St Sauver JL, et al. History of the Rochester Epidemiology Project: half a century of medical records linkage in a US population. Mayo Clin Proc 2012;87:12021213.

    • Search Google Scholar
    • Export Citation
  • 12.

    Maitra A, Sharma A, Brand RE, et al. A prospective study to establish a new-onset diabetes cohort: from the Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer. Pancreas 2018;47:12441248.

    • Search Google Scholar
    • Export Citation
  • 13.

    O’Neill CB, Atoria CL, O’Reilly EM, et al. Costs and trends in pancreatic cancer treatment. Cancer 2012;118:51325139.

  • 14.

    Wright AK, Kontopantelis E, Emsley R, et al. Life expectancy and cause-specific mortality in type 2 diabetes: a population-based cohort study quantifying relationships in ethnic subgroups. Diabetes Care 2017;40:338345.

    • Search Google Scholar
    • Export Citation
  • 15.

    Arias E, Heron M, Xu J. United States life tables, 2012. Natl Vital Stat Rep 2016;65:165.

  • 16.

    Billings BJ, Christein JD, Harmsen WS, et al. Quality-of-life after total pancreatectomy: is it really that bad on long-term follow-up? J Gastrointest Surg 2005;9:10591067.

    • Search Google Scholar
    • Export Citation
  • 17.

    Corral JE, Das A, Bruno MJ, et al. Cost-effectiveness of pancreatic cancer surveillance in high-risk individuals: an economic analysis. Pancreas 2019;48:526536.

    • Search Google Scholar
    • Export Citation
  • 18.

    Bonomi AE, Boudreau DM, Fishman PA, et al. Quality of life valuations of mammography screening. Qual Life Res 2008;17:801814.

  • 19.

    Choi JG, Nipp RD, Tramontano A, et al. Neoadjuvant FOLFIRINOX for patients with borderline resectable or locally advanced pancreatic cancer: results of a decision analysis. Oncologist 2019;24:945954.

    • Search Google Scholar
    • Export Citation
  • 20.

    Meng Y, McCarthy G, Berthon A, et al. Patient-reported health state utilities in metastatic gastroenteropancreatic neuroendocrine tumours—an analysis based on the CLARINET study. Health Qual Life Outcomes 2017;15:131.

    • Search Google Scholar
    • Export Citation
  • 21.

    Centers for Medicare & Medicaid Services. Physician fee schedule search. Accessed December 23, 2020. Available at: https://www.cms.gov/apps/physician-fee-schedule/search/search-criteria.aspx

  • 22.

    Tramontano AC, Chen Y, Watson TR, et al. Pancreatic cancer treatment costs, including patient liability, by phase of care and treatment modality, 2000-2013. Medicine (Baltimore) 2019;98:e18082.

    • Search Google Scholar
    • Export Citation
  • 23.

    Li R, Bilik D, Brown MB, et al. Medical costs associated with type 2 diabetes complications and comorbidities. Am J Manag Care 2013;19:421430.

    • Search Google Scholar
    • Export Citation
  • 24.

    Bhulani N, Gupta A, Gao A, et al. Palliative care and end-of-life health care utilization in elderly patients with pancreatic cancer. J Gastrointest Oncol 2018;9:495502.

    • Search Google Scholar
    • Export Citation
  • 25.

    TRIAD Study Group. The Translating Research Into Action for Diabetes (TRIAD) study: a multicenter study of diabetes in managed care. Diabetes Care 2002;25:386389.

    • Search Google Scholar
    • Export Citation
  • 26.

    Keehan SP, Cuckler GA, Poisal JA, et al. National health expenditure projections, 2019-28: expected rebound in prices drives rising spending growth. Health Aff (Millwood) 2020;39:704714.

    • Search Google Scholar
    • Export Citation
  • 27.

    Torrance GW, Feeny D. Utilities and quality-adjusted life years. Int J Technol Assess Health Care 1989;5:559575.

  • 28.

    Clemens K, Garrison LP Jr, Jones A, et al. Strategic use of pharmacoeconomic research in early drug development and global pricing. Pharmacoeconomics 1993;4:315322.

    • Search Google Scholar
    • Export Citation
  • 29.

    Steuten LM, Ramsey SD. Improving early cycle economic evaluation of diagnostic technologies. Expert Rev Pharmacoecon Outcomes Res 2014;14:491498.

    • Search Google Scholar
    • Export Citation
  • 30.

    Chen W, Butler RK, Lustigova E, et al. Validation of the enriching new-onset diabetes for pancreatic cancer model in a diverse and integrated healthcare setting. Dig Dis Sci 2021;66:7887.

    • Search Google Scholar
    • Export Citation
  • 31.

    Ginde AA, Foianini A, Renner DM, et al. Availability and quality of computed tomography and magnetic resonance imaging equipment in U.S. emergency departments. Acad Emerg Med 2008;15:,780783.

    • Search Google Scholar
    • Export Citation

Submitted July 10, 2020; final revision received December 14, 2020; accepted for publication December 14, 2020. Published online June 21, 2021.

Contributions: Study concept: All authors. Data curation: Schwartz, Matrisian, Sharder, Chari, Roth. Data analysis: Schwartz, Roth. Manuscript preparation: All authors. Manuscript revision: All authors.

Disclosures: The authors have disclosed that they have not received any financial consideration from any person or organization to support the preparation, analysis, results, or discussion of this article.

Funding: This study was funded by a grant from the Pancreatic Cancer Action Network (J.A. Roth).

Correspondence: Joshua A. Roth, PhD, MHA, Public Health Sciences Division, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, Seattle, WA 98109. Email: jroth@fredhutch.org
  • View in gallery

    Simplified schematic of (A) model decision tree, (B) PAC case, and (C) noncase Markov state–transition frameworks.

    Abbreviations: END-PAC, Enriching New-Onset Diabetes for Pancreatic Cancer; NOD, new-onset diabetes; PAC, pancreatic cancer.

  • View in gallery

    Deterministic cost per QALY gained under alternative PAC screening strategy performance scenarios. We assessed this range of scenarios to explore the value implications given the uncertainty about the performance of the risk-based PAC screening strategy before the large-scale EDI randomized trial.

    Abbreviations: EDI, Early Detection Initiative; END-PAC, Enriching New-Onset Diabetes for Pancreatic Cancer; NOD, new-onset diabetes; PAC, pancreatic cancer; QALY, quality-adjusted life-year.

  • View in gallery

    Tornado diagram displaying the 10 most influential inputs on the cost per QALY gained.

    Abbreviations: HR, hazard ratio; OS, overall survival; PAC, pancreatic cancer; QALY, quality-adjusted life-year.

  • View in gallery

    Cost-effectiveness acceptability curve displaying the probability of cost-effectiveness of risk-based pancreatic cancer screening in patients with new-onset diabetes and END-PAC score >0 at WTP thresholds of $0 to $300,000 per QALY gained.

    Abbreviations: END-PAC, Enriching New-Onset Diabetes for Pancreatic Cancer; QALY, quality-adjusted life-year; WTP, willingness-to-pay.

  • 1.

    Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin 2019;69:734.

  • 2.

    Howlader N, Noone A, Krapcho M, et al. , eds. SEER cancer statistics review, 1975–2017. Accessed December 23, 2020. Available at: https://seer.cancer.gov/csr/1975_2017/

    • Search Google Scholar
    • Export Citation
  • 3.

    Katz MHG, Hu CY, Fleming JB, et al. Clinical calculator of conditional survival estimates for resected and unresected survivors of pancreatic cancer. Arch Surg 2012;147:513519.

    • Search Google Scholar
    • Export Citation
  • 4.

    Poruk KE, Firpo MA, Adler DG, et al. Screening for pancreatic cancer: why, how, and who? Ann Surg 2013;257:1726.

  • 5.

    Lucas AL, Kastrinos F. Screening for pancreatic cancer. JAMA 2019;322:407408.

  • 6.

    Pannala R, Basu A, Petersen GM, et al. New-onset diabetes: a potential clue to the early diagnosis of pancreatic cancer. Lancet Oncol 2009;10:8895.

    • Search Google Scholar
    • Export Citation
  • 7.

    Chari ST, Leibson CL, Rabe KG, et al. Probability of pancreatic cancer following diabetes: a population-based study. Gastroenterology 2005;129:504511.

    • Search Google Scholar
    • Export Citation
  • 8.

    Sharma A, Kandlakunta H, Nagpal SJS, et al. Model to determine risk of pancreatic cancer in patients with new-onset diabetes. Gastroenterology 2018;155:730739.e3.

    • Search Google Scholar
    • Export Citation
  • 9.

    Sanders GD, Neumann PJ, Basu A, et al. Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: second panel on cost-effectiveness in health and medicine. JAMA 2016;316:10931103.

    • Search Google Scholar
    • Export Citation
  • 10.

    Melton LJ III. History of the Rochester Epidemiology Project. Mayo Clin Proc 1996;71:266274.

  • 11.

    Rocca WA, Yawn BP, St Sauver JL, et al. History of the Rochester Epidemiology Project: half a century of medical records linkage in a US population. Mayo Clin Proc 2012;87:12021213.

    • Search Google Scholar
    • Export Citation
  • 12.

    Maitra A, Sharma A, Brand RE, et al. A prospective study to establish a new-onset diabetes cohort: from the Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer. Pancreas 2018;47:12441248.

    • Search Google Scholar
    • Export Citation
  • 13.

    O’Neill CB, Atoria CL, O’Reilly EM, et al. Costs and trends in pancreatic cancer treatment. Cancer 2012;118:51325139.

  • 14.

    Wright AK, Kontopantelis E, Emsley R, et al. Life expectancy and cause-specific mortality in type 2 diabetes: a population-based cohort study quantifying relationships in ethnic subgroups. Diabetes Care 2017;40:338345.

    • Search Google Scholar
    • Export Citation
  • 15.

    Arias E, Heron M, Xu J. United States life tables, 2012. Natl Vital Stat Rep 2016;65:165.

  • 16.

    Billings BJ, Christein JD, Harmsen WS, et al. Quality-of-life after total pancreatectomy: is it really that bad on long-term follow-up? J Gastrointest Surg 2005;9:10591067.

    • Search Google Scholar
    • Export Citation
  • 17.

    Corral JE, Das A, Bruno MJ, et al. Cost-effectiveness of pancreatic cancer surveillance in high-risk individuals: an economic analysis. Pancreas 2019;48:526536.

    • Search Google Scholar
    • Export Citation
  • 18.

    Bonomi AE, Boudreau DM, Fishman PA, et al. Quality of life valuations of mammography screening. Qual Life Res 2008;17:801814.

  • 19.

    Choi JG, Nipp RD, Tramontano A, et al. Neoadjuvant FOLFIRINOX for patients with borderline resectable or locally advanced pancreatic cancer: results of a decision analysis. Oncologist 2019;24:945954.

    • Search Google Scholar
    • Export Citation
  • 20.

    Meng Y, McCarthy G, Berthon A, et al. Patient-reported health state utilities in metastatic gastroenteropancreatic neuroendocrine tumours—an analysis based on the CLARINET study. Health Qual Life Outcomes 2017;15:131.

    • Search Google Scholar
    • Export Citation
  • 21.

    Centers for Medicare & Medicaid Services. Physician fee schedule search. Accessed December 23, 2020. Available at: https://www.cms.gov/apps/physician-fee-schedule/search/search-criteria.aspx

  • 22.

    Tramontano AC, Chen Y, Watson TR, et al. Pancreatic cancer treatment costs, including patient liability, by phase of care and treatment modality, 2000-2013. Medicine (Baltimore) 2019;98:e18082.

    • Search Google Scholar
    • Export Citation
  • 23.

    Li R, Bilik D, Brown MB, et al. Medical costs associated with type 2 diabetes complications and comorbidities. Am J Manag Care 2013;19:421430.

    • Search Google Scholar
    • Export Citation
  • 24.

    Bhulani N, Gupta A, Gao A, et al. Palliative care and end-of-life health care utilization in elderly patients with pancreatic cancer. J Gastrointest Oncol 2018;9:495502.

    • Search Google Scholar
    • Export Citation
  • 25.

    TRIAD Study Group. The Translating Research Into Action for Diabetes (TRIAD) study: a multicenter study of diabetes in managed care. Diabetes Care 2002;25:386389.

    • Search Google Scholar
    • Export Citation
  • 26.

    Keehan SP, Cuckler GA, Poisal JA, et al. National health expenditure projections, 2019-28: expected rebound in prices drives rising spending growth. Health Aff (Millwood) 2020;39:704714.

    • Search Google Scholar
    • Export Citation
  • 27.

    Torrance GW, Feeny D. Utilities and quality-adjusted life years. Int J Technol Assess Health Care 1989;5:559575.

  • 28.

    Clemens K, Garrison LP Jr, Jones A, et al. Strategic use of pharmacoeconomic research in early drug development and global pricing. Pharmacoeconomics 1993;4:315322.

    • Search Google Scholar
    • Export Citation
  • 29.

    Steuten LM, Ramsey SD. Improving early cycle economic evaluation of diagnostic technologies. Expert Rev Pharmacoecon Outcomes Res 2014;14:491498.

    • Search Google Scholar
    • Export Citation
  • 30.

    Chen W, Butler RK, Lustigova E, et al. Validation of the enriching new-onset diabetes for pancreatic cancer model in a diverse and integrated healthcare setting. Dig Dis Sci 2021;66:7887.

    • Search Google Scholar
    • Export Citation
  • 31.

    Ginde AA, Foianini A, Renner DM, et al. Availability and quality of computed tomography and magnetic resonance imaging equipment in U.S. emergency departments. Acad Emerg Med 2008;15:,780783.

    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 451 451 375
PDF Downloads 297 297 246
EPUB Downloads 0 0 0