Transcript of Interview with Robin Yabroff, Ph.D., of NCI’s Health Services and Economics Branch
As noted in other parts of the Cancer Trends Progress Report there’ve been a lot of changes in incidence, survival, and the types of treatment in the past several decades. But a couple of years ago, we realized that our cost assessments didn’t really reflect all of these changes in incidence, and survival and treatment, so we worked together with a bunch of colleagues– statisticians, epidemiologists, and economists to develop new methods, and also to use the most recent data to try and estimate–come up with better estimates of the cost of cancer care in the U.S. When we think about cost, one of the things we’ve looked at is the trajectory of cost over time. And in the initial period following diagnosis and also among patients that die from cancer, the period right before death, those are very expensive, compared to the period in between. And so we call those, those distinct periods– we call them phase of care. And so, it’s really important to try and estimate the cost of cancer in the U.S. to also understand the prevalence or the number of people that are alive in each one of those phase of care at any given point in time. And so, a lot of the work that we’ve been doing recently that’s reflected in these new cost estimates are based on new methods for estimating cancer prevalence by phase of care. These estimates that we’re presenting this year in 2010 are not directly comparable to other estimates that have been presented previously. What we do plan on doing in the future is using these same methods in reporting trends. The idea is this would allow, this would allow projection of the cost of cancer care to the future using current population trends. And we feel like those data may be very important for policy, for plant resource allocation, for planning, and for understanding what the future burden of cancer is likely to be. One of the things we’re currently developing is use is–is a public use projections website that would allow users to make their own assumptions about trends in cancer incidents and survival and cost and try and project what would happen for different cancer sites or for all cancers overtime.
We use data from a variety of sources including SEER which is Surveillance Epidemiology and End Results. It’s a tumor registry in the U.S. and it currently represents about 26 percent of the U.S. population. We use those data to estimate cancer prevalence by phase of care, so using both incidence and survival to estimate cancer prevalence. And then we use data from SEER-Medicare which is a linkage between the SEER tumor registry and Medicare claims which cover about 97 percent of the U.S. elderly population among other, there are also other populations that Medicare covers. But importantly, it covers almost all of the elderly population. And that, of course, is the population where cancer is most common, it’s in the elderly. So, we used the SEER-Medicare data to estimate cost by phase of care, then we used population data from the U.S. Census Bureau to look at population counts by age group. We’ve applied the cost estimates to the prevalence estimates to come up with estimates of cancer burden in 2006. And in 2006, we estimated the cancer expenditures to be approximately 104.1 billion dollars. Included in this report, we also have information from some studies looking at–trying to estimate loss productivity due to early death from cancer. And so we included estimates from those studies in this report as well. One of the limitations of our approach is that we have to estimate for under 65 population using data from the over 65 population. And that we used to try and make that estimate, we use information from managed care studies which included patients in both age groups. But in general, younger patients tend to use more care, and so, we estimate the cost to be a little higher for younger populations compared to older populations. So, one of the main things we did in this new, in this new report is we used methods developed specifically for estimating that burden of cancer. But we also use the most recent data from SEER tumor registries on incidence and survival. We used the most recent population data from the U.S. Census Bureau. And we also used the most recent data from SEER-Medicare which is SEER tumor registry information linked to Medicare claims. And currently, there are over 3 and half million cancer patients included in SEER-Medicare. So, we feel like we’ve got a good–we got a good estimates of what’s going on in cancer patients in the U.S.