Cover Image

PAPERBACK
$63.25



View/Hide Left Panel

Appendix I
Cost-Effectiveness Analyses of Colorectal Cancer Screening: Results from a Pre-conference Modeling Exercise Michael Pignone, M.D., M.P.H.

SLIDE 1

SLIDE 1 NOTES: I would like to thank the following people for their work and advice on this exercise or on a previous review conducted for the US Preventive Services Task Force: Judy Wagner, Louise Russell, Martin Brown, Somnath Saha, Jeanne Mandelblatt, Tom Hoerger, Steve Teutsch, and all of the modelers who participated in this exercise.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary Appendix I Cost-Effectiveness Analyses of Colorectal Cancer Screening: Results from a Pre-conference Modeling Exercise Michael Pignone, M.D., M.P.H. SLIDE 1 SLIDE 1 NOTES: I would like to thank the following people for their work and advice on this exercise or on a previous review conducted for the US Preventive Services Task Force: Judy Wagner, Louise Russell, Martin Brown, Somnath Saha, Jeanne Mandelblatt, Tom Hoerger, Steve Teutsch, and all of the modelers who participated in this exercise.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 2 SLIDE 2 NOTES: The aims of the pre-Workshop modeling exercise, as I see it, were two-fold: The first was to compare the several different cost-effectiveness analyses of colorectal cancer screening. Such a comparison has three motivations: to gain insight into reasons for different results; to determine areas for future research focus; and potentially to learn something about how different screening strategies stack up against one another. The last motivation, however, is less important than the first two. The second objective is to use insights from this exercise to better inform future modeling and direct future CRC research efforts. The exercise should lead to better understanding of our parameters, better models, and an identification of questions or issues that need to be incorporated into the models.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 3 SLIDE 3 NOTES: As with some other cancers, colorectal cancer is, of course, an important disease that is amenable to screening. However, unlike many conditions amenable to preventive interventions, there are several different screening tests available, each supported by its own body of evidence on effectiveness, risks and costs. In other areas of cancer screening there is usually one dominant modality. With CRC screening, several modalities are available. Also, unlike many other areas, there are several recent high-quality published cost-effectiveness analyses which reached different conclusions about the relative merits of alternative screening strategies. That fact provides an interesting opportunity, because it offers us a chance to explore how that variation might arise, and whether the variation is there for good reasons, or whether we should try to reduce the variation through standardization of methods and assumptions.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 4 SLIDE NOTES 4: A precursor to this exercise was the work my colleagues and I did for the US Preventive Services Task Force in 2000 (Pignone et al., 2002). At that time, we reviewed seven published models. All seven found that any of the main screening strategies for colorectal cancer were cost-effective compared with no screenings. The cost-effectiveness of any screening strategy compared with doing nothing was generally below $30,000 per year of life added across all models. The models, however, did reach some different results as to the most effective and most cost-effective strategies. Some of those results were surprising to us. We also concluded that the variations were likely due to differences and uncertainties in input parameters, but it was impossible to sort out these factors from the published studies. We called for an exercise similar to the one undertaken for this workshop.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 5 SLIDE 5 NOTES: Here is a brief description of the methods used in the pre-workshop modeling exercise. Each modeler was asked to analyze 5 screening strategies, as well as no screening, as listed above. The prototype radiological screening test was defined to have characteristics somewhere in between barium enema, which is relatively inexpensive, and virtual colonoscopy, which is more sensitive but more expensive.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 6 SLIDE 6 NOTES: We then specified standardized values for inputs in the four categories listed above. The modelers were asked to analyze each of the six strategies 10 times, with each run involving a different combination of original or standardized parameter values.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 7 SLIDE 7 NOTES: For each run, we did the following: We ordered the years of life saved for every strategy from lowest to highest. We identified the strongly dominated strategies—those which were both less effective and more costly than at least one other strategy. Strongly dominated strategies were eliminated. Of the remaining strategies, we identified those that were weakly dominated—they were both less effective and their costs per year of life added were higher than at least one other strategy. The remaining strategies constitute the undominated set. We then calculated the incremental cost-effectiveness ratio: the incremental costs per incremental year of life added by moving from the least (undominated) strategy to the next least (undominated) strategy, and so on. We designated a “preferred strategy” for any cost-effectiveness limit as the one with the highest effectiveness (years of life added) whose incremental cost-effectiveness ratio meets a given limit.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 8 SLIDE 8 NOTES: This and the next 5 slides review the standardized assumptions in each general area. Here are some basic assumptions that were common to all runs and all strategies.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 9 SLIDE 9 NOTES: No notes.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 10 SLIDE 10 NOTES: No notes.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 11 SLIDE 11 NOTES: Note that we did not model more complex assumptions regarding complications, such as the possibility of bleeding (short of perforation) with colonoscopy, or other complications, such as a patient who is falls and breaks a bone after colonoscopy.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 26 SLIDE 26 NOTES: This slide groups lifetime costs by model. There are now some differences across strategies for all models, but they are relatively small across different tests, with FOBT generally less costly in each model than the other screening strategies. The relative costs across strategies tend to follow pretty much the same pattern across the different models.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 27 SLIDE 27 NOTES: Here is the average cost-effectiveness under the standardized assumptions (Run 6). The results are quite similar to what was seen under the original assumptions for average cost-effectiveness. The results here vary from about $6,000 per life-year saved to about $25,000 per life-year saved. It appears from visual inspection that there is slightly less overall variation than we had under the original assumptions.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 28 SLIDE 28 NOTES: The rest of this presentation is about incremental cost-effectiveness ratios (as opposed to average cost-effectiveness ratios) and preferred strategies. Recall that the incremental cost-effectiveness ratio (ICER) is calculated by eliminating all strongly and weakly dominated strategies and then sorting the remaining strategies in ascending order according to years of life added compared with no screening. The incremental ratio is calculated for each strategy as the extra costs incurred per extra year of life added by moving from each strategy to the next most effective strategy.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 29 SLIDE 29 NOTES: Here are the ICERs under the original assumptions (Run 1). There are definite differences across models in which strategies are dominated and which are not. For example, flexible sigmoidoscopy every five years is dominated in all but the Miscan model. In four of the five models, colonoscopy is either weakly or strongly dominated by other tests. In the Vijan model it has a cost-effectiveness ratio of $38,000 per year of life added.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 30 SLIDE 30 NOTES: This chart shows the ICER’s under the standardized assumptions (Run 6). With assumptions standardized, the first four models get very similar results. Under the specific set of standardized assumptions made about each strategy, FOBT screening generally had ICER’s between $5,000 and $12,000 per year of life saved, while flexible sigmoidoscopy every five years was dominated in all models. Radiology and colonoscopy were dominated in most models. When they were not, they had a high incremental cost-effectiveness ratio. Finally, annual FOBT plus flexible sigmoidoscopy under these particular standardized assumptions produced additional life years at a fairly high additional cost. The highest estimate was from the Vanderbilt model, with over $350,000 per additional life year saved. Those results merit more discussion.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 31 SLIDE 31 NOTES: This slide shows—for the original assumptions—the most effective strategy (i.e., the strategy that produces the largest number of additional years of life among all non-dominated strategies) under given incremental cost-effectiveness thresholds. So, for example, the Harvard model predicts that annual FOBT is the most effective strategy among all strategies whose ICER is $20,000 or less. There is a great deal of variation across models in which strategy is preferred at any cost-effectiveness threshold.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 32 SLIDE 32 NOTES: This chart is the same as the previous chart, except that the assumptions are fully standardized (Run 6). Here, almost all of the differences across models disappear. The most effective strategy at any different threshold is the same. In fact, the only difference is the threshold level at which the FOBT with FSIG overtakes FOBT alone as the preferred strategy.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 33 SLIDE 33 NOTES: This exercise had several limitations. Some were a function of the limited time we had to design and conduct the exercise and the amount of effort that the modelers could realistically expect to make to support the exercise. The exercise did not examine the effect of different assumptions about the natural history of colorectal cancers. The modelers were provided with only one set of standard values for assumptions. Those standard values were not all realistic; many were selected because they would require the least amount of model redesign. True values of such parameters might be quite different, and another exercise would be warranted for such parameters. The effects of standardizing assumptions might differ with other sets of screening strategies. It might be useful, for example, to do an exercise that includes more complex screening strategies such as one that begins with one screening test and transitions over time to another as individuals age. Finally, there was no accounting for uncertainty in estimates. There is a method of doing sensitivity analysis, and we did not do Monte Carlo simulations to generate confidence intervals around some of our parameters. So we are dealing with point estimates here.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 34 SLIDE 34 NOTES: In this chart and the next, the same results are grouped by model instead of by strategy. You can see that there is some variation in terms of life years saved within each model by the different strategies, suggesting that the strategies have different levels of effectiveness.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary SLIDE 35 SLIDE 35 NOTES: Here are some preliminary thoughts about implications of this exercise. First, it would certainly be a good idea to establish some standard cost inputs, to eliminate this major source of variation across models. We also need additional research in modeling compliance. I believe we are still missing some of the key input parameters that would help us to more effectively and accurately model what actually happens in terms of compliance. Finally, I would like to see this process be applied in a number of different health areas. This kind of exercise can teach us a great deal, and I hope that this can be a model for future meetings or group projects. Not only can we understand better why results differ, but we may also advance the field of modeling itself.

OCR for page 118
Economic Models of Colorectal Cancer Screening in Average-Risk Adults: Workshop Summary REFERENCES Ness RM, Holmes A, Klein R, Greene J, Dittus R. 1998. Outcome states of colorectal cancer: Identification and description using patient focus groups. Am J Gastroenterol 93(9):1491–1497 . Pignone M, Saha S, Hoerger T, Mandelblatt J. 2002. Cost-effectiveness analyses of colorectal cancer screening: a systematic review for the U.S. Preventive Services Task Force. Ann Intern Med. 137(2):96–104.