Paying For Value By Samir Soneji and JaeWon Yang 10.1377/hlthaff.2014.0174 HEALTH AFFAIRS 34, NO. 3 (2015): 390–399 ©2015 Project HOPE— The People-to-People Health Foundation, Inc.
doi:
Samir Soneji (samir.soneji@ dartmouth.edu) is an assistant professor at the Dartmouth Institute for Health Policy and Clinical Practice and a member of the Norris Cotton Cancer Center, both in Lebanon, New Hampshire. JaeWon Yang was an undergraduate at Dartmouth College, in Hanover, New Hampshire, at the time this article was written.
New Analysis Reexamines The Value Of Cancer Care In The United States Compared To Western Europe ABSTRACT Despite sharp increases in spending on cancer treatment since 1970 in the United States compared to Western Europe, US cancer mortality rates have decreased only modestly. This has raised questions about the additional value of US cancer care derived from this additional spending. We calculated the number of US cancer deaths averted, compared to the situation in Western Europe, between 1982 and 2010 for twelve cancer types. We also assessed the value of US cancer care, compared to that in Western Europe, by estimating the ratio of additional spending on cancer to the number of quality-adjusted lifeyears saved. Compared to Western Europe, for three of the four costliest US cancers—breast, colorectal, and prostate—there were approximately 67,000, 265,000, and 60,000 averted US deaths, respectively, and for lung cancer there were roughly 1,120,000 excess deaths in the study period. The ratio of incremental cost to quality-adjusted life-years saved equaled $402,000 for breast cancer, $110,000 for colorectal cancer, and $1,979,000 for prostate cancer—amounts that exceed most accepted thresholds for cost-effective medical care. The United States lost qualityadjusted life-years despite additional spending for lung cancer: −$19,000 per quality-adjusted life-year saved. Our results suggest that cancer care in the United States may provide less value than corresponding cancer care in Western Europe for many leading cancers.
S
ince 1970 the United States has experienced the steepest rise in health care spending per capita of all countries in the Organization for Economic Cooperation and Development (OECD).1 Cancer accounted for 5 percent of the spending rise during the period 1987– 2000: Only heart disease, pulmonary conditions, and mental disorders accounted for larger shares.2 Despite sharp increases in spending on cancer treatment, US cancer mortality rates decreased only modestly—by 12 percent since 1970, compared to a 62 percent decrease for heart disease.3,4 This has raised questions about the additional value of US cancer care derived from the 390
H e a lt h A f fai r s
March 2 015
3 4: 3
additional spending, compared to the situation in other high-income countries.5 The value of US cancer care has been assessed through comparisons of gains in survival time after a cancer diagnosis. The recent assessment by Tomas Philipson and coauthors,6 cited in the 2013 Economic Report of the President as evidence of cost-effective gains in life expectancy,7 concluded that greater gains in breast and prostate cancer survival in the United States, compared to certain European countries, had generated nearly $600 billion dollars in value.6 However, this study did not account for the stage of cancer at diagnosis. Consequently, it was subject to the well-known biases associated
with diagnosis and screening that inflate cancer survival time, including lead-time bias (shifting the date of diagnosis earlier without affecting the date of death) and overdiagnosis bias (detecting pseudo-disease that would not likely progress to clinically detectable disease).8 In contrast to the work of Philipson and coauthors, other cross-national studies that used the same European survival data and accounted for stage of cancer at diagnosis concluded that the differences in breast and colorectal cancer survival were largely explained by this clinically significant factor.9,10 Philipson and coauthors argued that greater improvements in US cancer survival could not be an artifact of lead-time or overdiagnosis biases because US cancer mortality rates fell faster than those in Europe. In a secondary analysis using population-level cancer mortality rates, Philipson and coauthors concluded that the United States averted 87,000 breast cancer and 222,000 prostate cancer deaths between 1982 and 2005, compared to nine selected European countries. These numbers of averted deaths would represent a substantial 9 percent and 30 percent of the observed number of US breast and prostate cancer deaths, respectively, during the study period. We do not yet know if this calculation is replicable and— perhaps more important—whether the United States averted cancer deaths and saved life-years when compared to a broader benchmark of European countries. Our article improves upon previous research in assessing the value of US cancer care by comparing the United States to the broad benchmark of all twenty countries in Western Europe between 1982 and 2010. We focused exclusively on cross-national comparisons based on cancer mortality rates derived from complete national vital statistics, which are not subject to the same biases inherent in cancer survival times. We included in our comparator group countries not previously considered: those with high health care spending (the Netherlands and Switzerland), with similar gains in life expectancy as the United States over the past quarter-century (Denmark), with nationally organized screening programs (England), and with large populations (Italy and Spain).11,12 Western Europe serves as a suitable comparator for the United States because it is a region with the same magnitude of population and commensurate living standards, but different prevailing approaches to organizing health care.12
Study Data And Methods Population-Level Cancer Mortality Rates We collected population-level cancer mortality
rates in Western European countries and the United States from the World Health Organization (WHO) Cancer Mortality Database. Rates were grouped by five-year age groups, beginning with people ages 40–44 years, between 1982 and 2010. We examined rates for breast, cervix uteri, colon and rectum (colorectal), Hodgkin’s lymphoma, leukemia, lung (including trachea and bronchus), melanoma, non-Hodgkin’s lymphoma, prostate, stomach, testis, and thyroid cancers. The twenty Western European countries were Austria, Belgium, Denmark, England, Finland, France, Germany, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Northern Ireland, Norway, Portugal, Scotland, Spain, Sweden, Switzerland, and Wales. We also collected cancer mortality rates for Slovakia and Slovenia, to replicate the results of Philipson and coauthors. We obtained population counts for adults ages forty and older between 1982 and 2010 in Western European countries and the United States, based on each country’s vital statistics.13 We replaced missing cancer mortality rates (2.6 percent) for Belgium (2000–03, all cancers; 1985 and 1987–91, non-Hodgkin’s lymphoma and thyroid), Finland (1988, non-Hodgkin’s lymphoma and thyroid), Germany (1982, non-Hodgkin’s lymphoma and thyroid), Iceland (2010, all cancers except colorectal), Italy (2004–05, all cancers), Portugal (2004–05, all cancers), and Switzerland (1995–2005, Hodgkin’s lymphoma and thyroid) by fitting an ordinary least squares regression line on the logarithm of the age group– specific mortality rate as a function of time. Cost Of Cancer Care We followed the general approach of Philipson and coauthors to calculate the cost of cancer care. First, we collected total health care expenditure data from the OECD Health Statistics 2014 database. Costs were reported in current US dollars and adjusted for purchasing power parity. We replaced missing total health care costs for France (1982–84, 1986–89), Germany (1991), and Italy (1982–87) by fitting an ordinary least squares regression line on the logarithm of cost as a function of time. We then inflation-adjusted the total health care expenditure to constant 2010 dollars using gross domestic product implicit deflator data from the US Bureau of Economic Analysis. To calculate the cost of cancer care for each country, we multiplied the total inflation-adjusted health care expenditures by the countryspecific percentage of total expenditure that was spent on cancer, which we obtained from the Swedish Karolinska Institute and the US National Cancer Institute (NCI). This method of estimating the annual cost of US cancer care Ma rc h 2 015
34 : 3
H e a lt h A f fai r s
391
Paying For Value produced costs that approximately equaled those calculated by the NCI using Medicare administrative claims from cancer patients in the Surveillance, Epidemiology, and End Results (SEER) Program registries (see online Appendix Exhibit 1 for a comparison).14 The OECD reports total health care spending for the United Kingdom instead of separately by constituent country.We apportioned total health care spending for the United Kingdom into country-specific spending by applying each country’s share of National Health Service spending for each year: 83 percent for England, 3 percent for Northern Ireland, 9 percent for Scotland, and 5 percent for Wales.15 To calculate the cost of country-specific cancer care separately for each cancer, we multiplied the country-specific cancer costs by the proportion that each of the twelve cancer types represented of all incident cases using the SEER registries for the United States and the EUROCARE cancer registry for Europe.16 We then created a weighted average cost of cancer-specific care for Western Europe, in which the weights equaled the proportions of the population of adults ages forty and older in each country by year. We adjusted this weighted annual cost of cancer-specific care for Western Europe by the annual ratio of the US population and the Western European population. Finally, we summed the annual cost of cancerspecific care for the United States and Western Europe from 1982 to 2010 at a 3 percent discount rate. For detailed numerical examples of the calculations of cancer care cost, cancer deaths averted, and life-years saved, see the Appendix.14 Cancer Deaths Averted We created weighted cancer mortality rates by five-year age groups, sex, and year for the Western European countries. The weights equaled the proportions of the population in each country by age group, sex, and year. To calculate the number of cancer deaths averted, we subtracted the aggregate weighted European cancer mortality rate from the US cancer mortality rate and multiplied this difference by the US population in each age group, sex, and year. In doing so, we accounted for country-specific differences in the distribution of the population across age groups. Life-Years Saved To calculate the number of life-years saved, we started with the number of US cancer deaths averted for each age group, sex, and year. Then we assigned to each stratum its expected number of life-years lived, based on US life tables. Next we adjusted life-years saved by age- and sex-specific health-related quality of life utilities, based on the Short Form Health Survey SF-6D.17 Finally, we summed the annual total and quality-adjusted life-years saved from 1982 to 392
H e a lt h A f fai r s
March 2 015
3 4: 3
2010 at a 3 percent discount rate. Value Of Cancer Care We estimated the value of US cancer care, relative to that of Western European cancer care, by calculating the ratio of incremental costs (US cancer care costs minus Western European cancer care costs) to the number of life-years saved, as well as to the number of quality-adjusted life-years (QALYs) saved.We calculated these ratios for each of the twelve cancer types. Recalculation Of Previous Estimates Of Cancer Deaths Averted We also attempted to replicate the number of cancer deaths averted that were estimated by Philipson and coauthors for all cancers but lung. Philipson and coauthors did not consider lung cancer—the leading cause of cancer death in the United States and Europe—so we had no basis for comparison.18 We considered the same set of nine European countries (Finland, France, Germany, Iceland, Norway, Scotland, Slovakia, Slovenia, and Sweden) and the same study period (1982–2005) used by Philipson and coauthors. In addition to calculating cancer deaths averted using mortality rates, we also statistically modeled the number of cancer deaths, closely following the approach of Philipson and coauthors. For a comparison across other cancers, see Appendix Exhibit 4.14 Sensitivity Analyses We conducted multiple sensitivity analyses to assess the robustness of our findings with respect to the time interval chosen and the handling of missing data. First, we calculated the number of US cancer deaths averted after the years of peak US mortality for breast, colorectal, and prostate cancer. By doing this, we were able to more directly attribute the cancer deaths averted after those peaks to the introduction of effective screening and treatment. Second, we considered two alternative approaches to missing cancer mortality rates: multiple imputation and fitting an ordinary least squares regression line on the observed agegroup-specific mortality rate—instead of on its logarithm, as in the main analysis—as a function of time (for an example of these alternative approaches, see Appendix Exhibit 2).14 Limitations We acknowledge several limitations of our study. First, our results may be subject to bias due to misclassification of the underlying cause of death on death certificates. Earlier studies found that the classification of cause of death on death certificates was more accurate for cancer than for other causes, however.19 Moreover, the agreements in the cause of death between medical records and death certificates for breast, lung, and prostate cancers were among the highest across all cancer types.20 Second, the validity of our results depends, in
part, on consistent coding of cancer deaths across countries and over time. Cross-national comparisons found breast, colorectal, lung, and prostate cancers to be among the mostly consistently coded cancers.21 Our study period included coding according to both the International Classification of Diseases, Ninth Revision (ICD-9), and the Tenth Revision (ICD-10). However, we did not observe any sharp changes in cancer mortality rates in the year after a country transitioned to ICD-10.
Study Results Cancer Deaths Averted Between 1982 and 1986, and again in 1990, US breast cancer mortality rates exceeded those of Western Europe (the comparator). This resulted in 2,592 excess breast cancer deaths, which partially offset the 69,389 averted deaths that occurred after 1986 (Exhibit 1). US colorectal cancer mortality rates were lower than those of Western European countries throughout the study period and led to 264,632 averted deaths (Exhibit 2). In contrast, US lung cancer mortality rates exceeded those of the comparator countries throughout the study period and led to 1,119,599 excess deaths (Appendix Exhibit 3).14 Finally, between 1982 and 1995 (except in 1985), higher US prostate cancer mortality rates led to 11,759 excess deaths, which partially offset the 71,641 averted deaths that occurred after 1995 (Exhibit 3). We present the number of cancer deaths averted for the additional eight cancers in Appendix Exhibit 3.14 Value Of Cancer Care The cost of breast cancer care in the United States exceeded that in Western Europe by $435 billion between 1982 and 2010 (Exhibit 4). During the same time period, the 66,797 averted breast cancer deaths overall led to 1,082,013 QALYs saved. Thus, the ratio of incremental cost to QALYs saved equaled $402,369. For colorectal cancer, the ratio equaled $110,009 ($326 billion and 2,962,177 QALYs saved). For lung cancer, the 1,119,599 excess deaths led to 21,571,483 QALYs lost, while the cost of lung cancer care in the United States exceeded that in Western Europe by $406 billion. Thus, the ratio of incremental cost to QALYs saved equaled −$18,815. For prostate cancer, the ratio equaled $1,978,542 ($435 billion and 219,678 QALYs saved). Recalculation Of Previous Estimates Of Cancer Deaths Averted We could not reproduce the number of averted US breast and prostate cancer deaths calculated by Philipson and coauthors. Using the same WHO cancer mortal-
ity rate data and same set of nine European countries, we calculated 19,518 averted breast cancer deaths and 87,832 averted prostate cancer deaths between 1982 and 2005, which represents 78 percent fewer breast cancer deaths and 60 percent fewer prostate cancer deaths averted than reported by Philipson and coauthors. For a comparison across other cancers, see Appendix Exhibit 4.14 We calculated more averted colorectal cancer deaths for the United States, compared to the nine European countries, than Philipson and coauthors did: 291,273 averted deaths versus 281,932. Finally, we calculated 482,191 averted stomach cancer deaths for the United States compared to the nine European countries, in contrast to the 224,212 excess stomach cancer deaths estimated by Philipson and coauthors. Sensitivity Analyses US cancer mortality rates peaked in 1985 for breast cancer, 1984 for colorectal cancer, and 1991 for prostate cancer. After these peak years, declining US mortality rates led to 68,872 breast cancer deaths averted, 261,859 colorectal cancer deaths averted, and 62,385 prostate cancer deaths averted. We reached nearly identical results for the number of US cancer deaths averted based on alternative approaches to missing cancer mortality rates, as shown in Appendix Exhibit 5.14 For example, using multiple imputation, we calculated that 65,781 breast cancer deaths and 59,129 prostate cancer deaths were averted.
Discussion Our study has two central findings. First, the number of deaths averted in the United States compared to all of Western Europe from 1982 to 2010 varied substantially by cancer type. The
Exhibit 1 Excess And Averted Breast Cancer Deaths In The United States, Compared To Western Europe, 1982–2010
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text).
March 2015
34:3
H e a lt h A f fai r s
393
Paying For Value Exhibit 2 Excess And Averted Colorectal Cancer Deaths In The United States, Compared To Western Europe, 1982–2010
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text).
United States averted the largest number of deaths for stomach and colorectal cancer and experienced the largest number of excess deaths for lung cancer and non-Hodgkin’s lymphoma (Exhibit 4). The United States averted a modest number of breast and prostate cancer deaths. Second, the additional value derived from costlier US cancer care also varied considerably by cancer type. For example, the ratio of incremental cost to QALYs saved equaled $402,369 for breast cancer, $110,009 for colorectal cancer, $1,978,542 for prostate cancer, and −$18,815 for lung cancer (Exhibit 4). Deaths Averted And Cancer Type The relatively small number of averted breast cancer deaths that we calculated resulted from similar levels of breast cancer mortality rates in the United States and Western Europe that declined at a comparable pace over time. This transnational
Exhibit 3 Excess And Averted Prostate Cancer Deaths In The United States, Compared to Western Europe,1982–2010
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text).
394
Health Affai rs
March 2 015
3 4: 3
pattern of mortality decline may be due, in part, to widely adopted advances in breast cancer treatment, such as the use of tamoxifen and breast-conserving surgery with adjuvant radiotherapy.22 The use of trastuzumab, which began earlier in the United States than in Western Europe, may have contributed to slightly steeper declines in US breast cancer mortality rates in the 2000s. In contrast to treatment, mammography screening rates varied widely among countries and likely did not lead to the transnational pattern of mortality decline.23 If screening led to declines in breast cancer mortality, we would expect increases in screening to precede decreases in mortality by five or more years, based on results of mammography screening trials. In addition, we would expect to see countries with comparatively higher screening rates experience faster declines in breast cancer mortality. However, we observed neither outcome. For example, US mammography screening rates peaked in 2000, fifteen years after the peak in breast cancer mortality rates.24 A recent study of three pairs of European countries with similar health care services and prevalence of risk factors for breast cancer mortality found similar declines in breast cancer mortality, despite substantial differences in the timing and level of the countries’ national screening programs.25 Finally, differences in breast cancer diagnosis could affect survival time from diagnosis through leadtime and overdiagnosis biases. But because those differences do not affect mortality rates, they cannot explain the transnational pattern of declining breast cancer mortality. US prostate cancer mortality rates declined faster over time than concurrently decreasing European rates. However, the causes of US and European declines remain largely unknown.26 Observational studies and clinical trials produce conflicting results on the potential protective effect against prostate cancer of statin drug use, which began earlier and remains at higher levels in the United States than in Europe.27 The role of screening in reducing prostate cancer mortality is also ambiguous. Indeed, prostate-specific antigen (PSA) testing began earlier in the United States than in Western Europe, and the rise in PSA testing preceded the decline in US prostate cancer mortality.28 Yet clinical trials do not demonstrate any short-term mortality benefits and produce conflicting conclusions regarding the long-term benefits of PSA testing.29 Compared to screening and prevention, the United States and Western Europe were more equal in their adoption of advances in neoadjuvant and adjuvant hormonal therapy, which may have contributed to their similar declines in
Exhibit 4 Ratio Of Incremental Costs To Total And Quality-Adjusted Life-Years Saved For Twelve Cancer Types In The United States, Compared To Western Europe Type of cancer Breast Cervix uteri Colorectal Hodgkin’s lymphoma Leukemia Lung Melanoma Non-Hodgkin’s lymphoma Prostate Stomach Testis Thyroid gland
Incremental cost ($ millions) 435,369 29,287
Incremental cost divided by life-years saved ($) 306,544 −712,751
Incremental cost divided by QALYs saved ($) 402,369 −855,019
Deaths averted 66,797 4,354
Life-years saved 1,420,249 −41,090
264,632 4,859
3,860,194 161,074
325,866 19,526
84,417 121,224
110,009 156,045
−64,530 −1,119,599
−1,370,884 −28,311,995
32,620 405,872
−23,795 −14,336
−30,790 −18,815
−39,144 −164,429
−1,025,066 −3,696,336
109,773 117,692
−107,089 −31,840
−136,592 −41,362
59,882 621,820
294,273 13,705,501
434,642 49,015
1,477,003 3,576
1,978,542 4,635
3,372 18,320
100,832 482,902
17,714 51,593
175,674 106,840
222,839 139,681
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text). NOTES Negative deaths averted represent excess cancer deaths. Incremental costs are expressed in 2010 dollars.
prostate cancer mortality rates.28 In contrast to breast and prostate cancer, the larger number of averted colorectal cancer deaths in the United States was likely due to screening, which began earlier and increased to higher levels in the United States, compared to Western Europe. Clinically, screening has been shown to reduce mortality in several large-scale randomized controlled trials of fecal occult blood testing and sigmoidoscopy and in colonoscopy case-control studies.30,31 Temporally, increases in colorectal cancer screening preceded the steepest declines in colorectal cancer mortality by at least five years in the United States.32 Screening was previously estimated to explain 53 percent of the observed decline in colorectal cancer mortality in the United States between 1975 and 2000, whereas changes in risk factors and treatment explained 35 percent and 12 percent, respectively.33 Countries with nationally organized screening programs experienced sharper declines in colorectal cancer mortality than countries without screening programs.34 In addition to higher screening rates, greater use of curative-intent surgery and adjuvant chemotherapy in the United States likely contributed to averted colorectal cancer deaths.35 Despite lower historical rates of cigarette smoking, lung cancer mortality rates in the United States exceeded those in Western Europe throughout the study period. After US lung cancer mortality rates peaked in 1993, they declined faster than Western European rates, which is almost certainly due to faster US reductions in cigarette smoking.36 If the current pace of decline persists, lung cancer mortality rates in the
United States may eventually fall below those in Western Europe and lead to averted lung cancer deaths. The large number of averted stomach cancer deaths may be due to lower rates of Helicobacter pylori infection and more substantial improvements in food preparation in the United States than in Western Europe. More than 60 percent of stomach cancer cases are attributed to H. pylori infection.37 The preparation of meats by salting or smoking increases the intake of carcinogenic nitrosamines. Surgery has long been the most common treatment for early-stage stomach cancer in the United States and Western Europe. However, clinical trials produce conflicting conclusions about whether more extensive lymph node dissection and the use of adjuvant therapy improve survival.38 Differences From Previous Estimates Of Cancer Deaths Averted Our findings, based on both the demographic and statistical approaches we used to calculate deaths averted, differ considerably from those of Philipson and coauthors, especially in the cases of breast and prostate cancer. For both of these types of cancer, we observed fewer deaths averted than Philipson and coauthors did, because of methodological differences. We argue that the interpretation of the statistical model by Philipson and coauthors produces several implausible results. First, Philipson and coauthors calculated that the prostate cancer mortality rate of the United States declined by 15.8 deaths more per 100,000 people than the corresponding decline in the prostate cancer mortality rate of nine European countries, which is far from what actually ocMarch 2015
34:3
H e a lt h A f fai r s
395
Paying For Value curred. The observed prostate cancer mortality rate of the United States declined by 14.3 deaths per 100,000 (from 54.9 deaths per 100,000 in 1982 to 40.6 deaths per 100,000 in 2005). During the same period, the observed prostate cancer mortality rate of the nine European countries declined by 5.2 deaths per 100,000 (from 57.1 to 51.9). Thus, the US rate declined by only 9.1 more deaths per 100,000 than the European rate. Second, Philipson and coauthors calculated that there were 224,212 excess stomach cancer deaths for the United States, compared to the nine European countries. But between 1982 and 2005, stomach cancer mortality rates in the United States were always substantially below those in the nine European countries. Thus, there could only be averted—not excess— stomach cancer deaths in the United States. Policy Implications Our study highlights two potentially actionable areas of focus for US policy makers. First, the greatest number of deaths averted occurred in cancers for which decreasing mortality rates were more likely to be the result of successful prevention and screening than advancements in treatment. The Affordable Care Act (ACA) led to full coverage of annual wellness visits for the elderly that address weight loss, smoking cessation, and nutrition. The ACA also eliminated copayments and deductibles for Medicare-covered preventive services that received a grade of A or B from the US Preventive Services Task Force. If greater access to wellness visits and preventive services translates to greater use of these types of medical care, the ACA may lead to more cancer deaths’ being averted and may do so in a cost-effective manner. Second, new payment models that incentivize Results from this study were presented in part at the thirty-eighth annual meeting of the American Society of Preventive Oncology, Arlington, Virginia, March 8–11, 2014. This research was supported by funding from the National Cancer Institute (Grant No. RC2CA148259), the National
high-value cancer care may lead to more cancer deaths being averted and reduce the cost of cancer care in the United States. A recent international commission identified three areas in which cancer care costs could be reduced: endof-life care, medical imaging, and chemotherapeutic agents.39 For example, palliative care provides higher value to patients with terminal cancer because it almost always is less expensive than chemotherapy, is aligned with patients’ preferences, and improves the quality of life.40 Health care systems and insurers that form accountable care organizations (for example, Medicare Pioneer ACOs) would be able to share in cost savings from the use of palliative care instead of intensive care.
Conclusion We calculated the number of cancer deaths averted for the United States, compared to Western Europe, and the value of US cancer care measured by the ratio of incremental costs to QALYs saved. The cost-effectiveness ratio was lower for colorectal cancer than for breast and prostate cancer, and it was negative for lung cancer. This indicates that prevention, screening, and treatment may be more cost-effective in the case of colorectal cancer than in the case of other leading cancers. Additionally, the number of breast and prostate cancer deaths averted was smaller than a previous study calculated. Our results suggest that cancer care in the United States did not always avert deaths compared to Western Europe, and when it did avert deaths, it often did so at substantial cost. ▪
Institute on Aging (Grant No. PO1AG19783), and the National Center for Advancing Translational Sciences (Award No. KL2TR001088), all at the National Institutes of Health. The funding agencies had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data;
or in the preparation, review, or approval of the manuscript. The authors thank Valerie Lewis, Jonathan Skinner, Shila Soneji, Harold Sox, and three anonymous reviewers for helpful comments and suggestions.
NOTES 1 Reinhardt UE, Hussey PS, Anderson GF. U.S. health care spending in an international context. Health Aff (Millwood). 2004; 23(3):10–25. 2 Thorpe KE, Florence CS, Joski P. Which medical conditions account for the rise in health care spending? Health Aff (Millwood). 2004;23(4): w4-437–45. 3 National Cancer Institute. Surveillance, Epidemiology, and End Results Program [home page on the Internet]. Bethesda (MD): NCI; [cited 2015 Jan 30]. Available from:
396
Health A ffairs
M a r c h 20 1 5
34:3
http://www.seer.cancer.gov 4 National Center for Health Statistics [home page on the Internet]. Hyattsville (MD): NCHS; [last updated 2015 Jan 30; cited 2015 Feb 10]. Available from: http:// www.cdc.gov/nchs/products/ vsus.htm 5 Cutler DM. Are we finally winning the war on cancer? J Econ Perspect. 2008;22(4):3–26. 6 Philipson T, Eber M, Lakdawalla D, Corral M, Conti R, Goldman D. An analysis of whether higher health
care spending in the United States versus Europe is “worth it” in the case of cancer. Health Aff (Millwood). 2012;31(4):667–75. 7 Council of Economic Advisers. Economic report of the president [Internet]. Washington (DC): Government Printing Office; 2013 Mar [cited 2015 Jan 22]. Available from: http://www.whitehouse.gov/sites/ default/files/docs/erp2013/full_ 2013_economic_report_of_the_ president.pdf 8 Autier P, Boniol M, Héry C, Masuyer
9
10
11
12
13
14
15
16
17
18
E, Ferlay J. Cancer survival statistics should be viewed with caution. Lancet Oncol. 2007;8(12):1050–2. Sant M, Allemani C, Capocaccia R, Hakulinen T, Aareleid T, Coebergh J, et al. Stage at diagnosis is a key explanation of differences in breast cancer survival across Europe. Int J Cancer. 2003;106(3):416–22. Ciccolallo L, Capocaccia R, Coleman MP, Berrino F, Coebergh JWW, Damhuis RAM, et al. Survival differences between European and US patients with colorectal cancer: role of stage at diagnosis and surgery. Gut. 2005;54(2):268–73. Squires DA. The U.S. health system in perspective: a comparison of twelve industrialized nations [Internet]. New York (NY): Commonwealth Fund; 2011 Jul [cited 2015 Jan 30]. (Issue Brief). Available from: http://www.commonwealth fund.org/~/media/files/ publications/issue-brief/2011/jul/ 1532_squires_us_hlt_sys_ comparison_12_nations_intl_brief_ v2.pdf National Research Council. Explaining divergent levels of longevity in high-income countries. Washington (DC): National Academies Press; 2011. Human Mortality Database [home page on the Internet]. Berkeley (CA): The Database; [cited 2015 Jan 22]. Available from: http://www .mortality.org To access the Appendix, click on the Appendix link in the box to the right of the article online. Crawford R, Emmerson C, Tetlow G. A survey of public spending in the UK [Internet]. London: Institute for Fiscal Studies; 2009 Sep [cited 2015 Jan 22]. (IFS Briefing Note No. BN43). Available from: http:// www.ifs.org.uk/bns/bn43.pdf De Angelis R, Sant M, Coleman MP, Francisci S, Baili P, Pierannunzio D, et al. Cancer survival in Europe 1999–2007 by country and age: results of EUROCARE-5—a population-based study. Lancet Oncol. 2014;15(1):23–34. Fryback D, Dunham NC, Palta M, Hanmer J, Buechner J, Cherepanov D, et al. US norms for six generic health-related quality-of-life indexes from the National Health Measurement study. Med Care. 2007;45(12): 1162–70. Jemal A, Center MM, DeSantis C,
19
20
21
22
23
24
25
26
27
28
29
Ward EM. Global patterns of cancer incidence and mortality rates and trends. Cancer Epidemiol Biomarkers Prev. 2010;19(8):1893–907. Sarfati D, Blakely T, Pearce N. Measuring cancer survival in populations: relative survival vs cancerspecific survival. Int J Epidemiol. 2010;39(2):598–610. German RR, Fink AK, Heron M, Stewart SL, Johnson CJ, Finch JL, et al. The accuracy of cancer mortality statistics based on death certificates in the United States. Cancer Epidemiol. 2011;35(2):126–31. Percy C, Muir C. The international comparability of cancer mortality data. Results of an international death certificate study. Am J Epidemiol. 1989;129(5):934–46. Allemani C, Sant M, Weir HK, Richardson LC, Baili P, Storm H, et al. Breast cancer survival in the US and Europe: a CONCORD highresolution study. Int J Cancer. 2013; 132(5):1170–81. Saika K, Sobue T. Time trends in breast cancer screening rates in the OECD countries. Jpn J Clin Oncol. 2011;41(4):591–2. Breen N, A Cronin K, Meissner HI, Taplin SH, Tangka FK, Tiro JA, et al. Reported drop in mammography: is this cause for concern? Cancer. 2007;109(12):2405–9. Autier P, Boniol M, Gavin A, Vatten LJ. Breast cancer mortality in neighbouring European countries with different levels of screening but similar access to treatment: trend analysis of WHO Mortality Database. BMJ. 2011;343:d4411. Collin SM, Martin RM, Metcalfe C, Gunnell D, Albertsen PC, Neal D, et al. Prostate-cancer mortality in the USA and UK in 1975–2004: an ecological study. Lancet Oncol. 2008; 9(5):445–52. Bonovas S, Filioussi K, Sitaras NM. Statin use and the risk of prostate cancer: a metaanalysis of 6 randomized clinical trials and 13 observational studies. Int J Cancer. 2008;123(4):899–904. Neppl-Huber C, Zappa M, Coebergh JW, Rapiti E, Rachtan J, Holleczek B, et al. Changes in incidence, survival, and mortality of prostate cancer in Europe and the United States in the PSA era: additional diagnoses and avoided deaths. Ann Oncol. 2011; 23(5):1325–34. Hayes JH, Barry MJ. Screening for
30
31
32
33
34
35
36
37
38
39
40
prostate cancer with the prostatespecific antigen test: a review of current evidence. JAMA. 2014; 311(11):1143–9. Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med. 1993;328(19):1365–71. Kahi CJ, Myers LJ, Slaven JE, Haggstrom D, Pohl H, Robertson DJ, et al. Lower endoscopy reduces colorectal cancer incidence in older individuals. Gastroenterology. 2014; 146(3):718–25. Hoff G, Dominitz JA. Contrasting US and European approaches to colorectal cancer screening: which is best? Gut. 2010;59(3):407–14. Edwards BK, Ward E, Kohler BA, Eheman C, Zauber AG, Anderson RN, et al. Annual report to the nation on the status of cancer, 1975–2006. Cancer. 2010;116(3):544–73. Bosetti C, Levi F, Rosato V, Bertuccio P, Lucchini F, Negri E, et al. Recent trends in colorectal cancer mortality in Europe. Int J Cancer. 2011; 129(1):180–9. Allemani C, Rachet B, Weir HK, Richardson LC, Lepage C, Faivre J, et al. Colorectal cancer survival in the USA and Europe: a CONCORD high-resolution study. BMJ Open. 2013;3(9):e003055. Devesa SS, Bray F, Vizcaino AP, Parkin DM. International lung cancer trends by histologic type: male: female differences diminishing and adenocarcinoma rates rising. Int J Cancer. 2005;117(2):294–9. Parkin DM. The global health burden of infection-associated cancers in the year 2002. Int J Cancer. 2006;118(12):3030–44. Russell MC, Mansfield PF. Surgical approaches to gastric cancer. J Surg Oncol. 2013;107(3):250–8. Kelly RJ, Smith TJ. Delivering maximum clinical benefit at an affordable price: engaging stakeholders in cancer care. Lancet Oncol. 2014; 15(3):e112–8. Schnipper LE, Smith TJ, Raghavan D, Blayney DW, Ganz PA, Mulvey TM, et al. American Society of Clinical Oncology identifies five key opportunities to improve care and reduce costs. J Clin Oncol. 2012; 30(14):1715–24.
March 2 015
3 4: 3
Health Affa irs
397
doi:
Samir Soneji (samir.soneji@ dartmouth.edu) is an assistant professor at the Dartmouth Institute for Health Policy and Clinical Practice and a member of the Norris Cotton Cancer Center, both in Lebanon, New Hampshire. JaeWon Yang was an undergraduate at Dartmouth College, in Hanover, New Hampshire, at the time this article was written.
New Analysis Reexamines The Value Of Cancer Care In The United States Compared To Western Europe ABSTRACT Despite sharp increases in spending on cancer treatment since 1970 in the United States compared to Western Europe, US cancer mortality rates have decreased only modestly. This has raised questions about the additional value of US cancer care derived from this additional spending. We calculated the number of US cancer deaths averted, compared to the situation in Western Europe, between 1982 and 2010 for twelve cancer types. We also assessed the value of US cancer care, compared to that in Western Europe, by estimating the ratio of additional spending on cancer to the number of quality-adjusted lifeyears saved. Compared to Western Europe, for three of the four costliest US cancers—breast, colorectal, and prostate—there were approximately 67,000, 265,000, and 60,000 averted US deaths, respectively, and for lung cancer there were roughly 1,120,000 excess deaths in the study period. The ratio of incremental cost to quality-adjusted life-years saved equaled $402,000 for breast cancer, $110,000 for colorectal cancer, and $1,979,000 for prostate cancer—amounts that exceed most accepted thresholds for cost-effective medical care. The United States lost qualityadjusted life-years despite additional spending for lung cancer: −$19,000 per quality-adjusted life-year saved. Our results suggest that cancer care in the United States may provide less value than corresponding cancer care in Western Europe for many leading cancers.
S
ince 1970 the United States has experienced the steepest rise in health care spending per capita of all countries in the Organization for Economic Cooperation and Development (OECD).1 Cancer accounted for 5 percent of the spending rise during the period 1987– 2000: Only heart disease, pulmonary conditions, and mental disorders accounted for larger shares.2 Despite sharp increases in spending on cancer treatment, US cancer mortality rates decreased only modestly—by 12 percent since 1970, compared to a 62 percent decrease for heart disease.3,4 This has raised questions about the additional value of US cancer care derived from the 390
H e a lt h A f fai r s
March 2 015
3 4: 3
additional spending, compared to the situation in other high-income countries.5 The value of US cancer care has been assessed through comparisons of gains in survival time after a cancer diagnosis. The recent assessment by Tomas Philipson and coauthors,6 cited in the 2013 Economic Report of the President as evidence of cost-effective gains in life expectancy,7 concluded that greater gains in breast and prostate cancer survival in the United States, compared to certain European countries, had generated nearly $600 billion dollars in value.6 However, this study did not account for the stage of cancer at diagnosis. Consequently, it was subject to the well-known biases associated
with diagnosis and screening that inflate cancer survival time, including lead-time bias (shifting the date of diagnosis earlier without affecting the date of death) and overdiagnosis bias (detecting pseudo-disease that would not likely progress to clinically detectable disease).8 In contrast to the work of Philipson and coauthors, other cross-national studies that used the same European survival data and accounted for stage of cancer at diagnosis concluded that the differences in breast and colorectal cancer survival were largely explained by this clinically significant factor.9,10 Philipson and coauthors argued that greater improvements in US cancer survival could not be an artifact of lead-time or overdiagnosis biases because US cancer mortality rates fell faster than those in Europe. In a secondary analysis using population-level cancer mortality rates, Philipson and coauthors concluded that the United States averted 87,000 breast cancer and 222,000 prostate cancer deaths between 1982 and 2005, compared to nine selected European countries. These numbers of averted deaths would represent a substantial 9 percent and 30 percent of the observed number of US breast and prostate cancer deaths, respectively, during the study period. We do not yet know if this calculation is replicable and— perhaps more important—whether the United States averted cancer deaths and saved life-years when compared to a broader benchmark of European countries. Our article improves upon previous research in assessing the value of US cancer care by comparing the United States to the broad benchmark of all twenty countries in Western Europe between 1982 and 2010. We focused exclusively on cross-national comparisons based on cancer mortality rates derived from complete national vital statistics, which are not subject to the same biases inherent in cancer survival times. We included in our comparator group countries not previously considered: those with high health care spending (the Netherlands and Switzerland), with similar gains in life expectancy as the United States over the past quarter-century (Denmark), with nationally organized screening programs (England), and with large populations (Italy and Spain).11,12 Western Europe serves as a suitable comparator for the United States because it is a region with the same magnitude of population and commensurate living standards, but different prevailing approaches to organizing health care.12
Study Data And Methods Population-Level Cancer Mortality Rates We collected population-level cancer mortality
rates in Western European countries and the United States from the World Health Organization (WHO) Cancer Mortality Database. Rates were grouped by five-year age groups, beginning with people ages 40–44 years, between 1982 and 2010. We examined rates for breast, cervix uteri, colon and rectum (colorectal), Hodgkin’s lymphoma, leukemia, lung (including trachea and bronchus), melanoma, non-Hodgkin’s lymphoma, prostate, stomach, testis, and thyroid cancers. The twenty Western European countries were Austria, Belgium, Denmark, England, Finland, France, Germany, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Northern Ireland, Norway, Portugal, Scotland, Spain, Sweden, Switzerland, and Wales. We also collected cancer mortality rates for Slovakia and Slovenia, to replicate the results of Philipson and coauthors. We obtained population counts for adults ages forty and older between 1982 and 2010 in Western European countries and the United States, based on each country’s vital statistics.13 We replaced missing cancer mortality rates (2.6 percent) for Belgium (2000–03, all cancers; 1985 and 1987–91, non-Hodgkin’s lymphoma and thyroid), Finland (1988, non-Hodgkin’s lymphoma and thyroid), Germany (1982, non-Hodgkin’s lymphoma and thyroid), Iceland (2010, all cancers except colorectal), Italy (2004–05, all cancers), Portugal (2004–05, all cancers), and Switzerland (1995–2005, Hodgkin’s lymphoma and thyroid) by fitting an ordinary least squares regression line on the logarithm of the age group– specific mortality rate as a function of time. Cost Of Cancer Care We followed the general approach of Philipson and coauthors to calculate the cost of cancer care. First, we collected total health care expenditure data from the OECD Health Statistics 2014 database. Costs were reported in current US dollars and adjusted for purchasing power parity. We replaced missing total health care costs for France (1982–84, 1986–89), Germany (1991), and Italy (1982–87) by fitting an ordinary least squares regression line on the logarithm of cost as a function of time. We then inflation-adjusted the total health care expenditure to constant 2010 dollars using gross domestic product implicit deflator data from the US Bureau of Economic Analysis. To calculate the cost of cancer care for each country, we multiplied the total inflation-adjusted health care expenditures by the countryspecific percentage of total expenditure that was spent on cancer, which we obtained from the Swedish Karolinska Institute and the US National Cancer Institute (NCI). This method of estimating the annual cost of US cancer care Ma rc h 2 015
34 : 3
H e a lt h A f fai r s
391
Paying For Value produced costs that approximately equaled those calculated by the NCI using Medicare administrative claims from cancer patients in the Surveillance, Epidemiology, and End Results (SEER) Program registries (see online Appendix Exhibit 1 for a comparison).14 The OECD reports total health care spending for the United Kingdom instead of separately by constituent country.We apportioned total health care spending for the United Kingdom into country-specific spending by applying each country’s share of National Health Service spending for each year: 83 percent for England, 3 percent for Northern Ireland, 9 percent for Scotland, and 5 percent for Wales.15 To calculate the cost of country-specific cancer care separately for each cancer, we multiplied the country-specific cancer costs by the proportion that each of the twelve cancer types represented of all incident cases using the SEER registries for the United States and the EUROCARE cancer registry for Europe.16 We then created a weighted average cost of cancer-specific care for Western Europe, in which the weights equaled the proportions of the population of adults ages forty and older in each country by year. We adjusted this weighted annual cost of cancer-specific care for Western Europe by the annual ratio of the US population and the Western European population. Finally, we summed the annual cost of cancerspecific care for the United States and Western Europe from 1982 to 2010 at a 3 percent discount rate. For detailed numerical examples of the calculations of cancer care cost, cancer deaths averted, and life-years saved, see the Appendix.14 Cancer Deaths Averted We created weighted cancer mortality rates by five-year age groups, sex, and year for the Western European countries. The weights equaled the proportions of the population in each country by age group, sex, and year. To calculate the number of cancer deaths averted, we subtracted the aggregate weighted European cancer mortality rate from the US cancer mortality rate and multiplied this difference by the US population in each age group, sex, and year. In doing so, we accounted for country-specific differences in the distribution of the population across age groups. Life-Years Saved To calculate the number of life-years saved, we started with the number of US cancer deaths averted for each age group, sex, and year. Then we assigned to each stratum its expected number of life-years lived, based on US life tables. Next we adjusted life-years saved by age- and sex-specific health-related quality of life utilities, based on the Short Form Health Survey SF-6D.17 Finally, we summed the annual total and quality-adjusted life-years saved from 1982 to 392
H e a lt h A f fai r s
March 2 015
3 4: 3
2010 at a 3 percent discount rate. Value Of Cancer Care We estimated the value of US cancer care, relative to that of Western European cancer care, by calculating the ratio of incremental costs (US cancer care costs minus Western European cancer care costs) to the number of life-years saved, as well as to the number of quality-adjusted life-years (QALYs) saved.We calculated these ratios for each of the twelve cancer types. Recalculation Of Previous Estimates Of Cancer Deaths Averted We also attempted to replicate the number of cancer deaths averted that were estimated by Philipson and coauthors for all cancers but lung. Philipson and coauthors did not consider lung cancer—the leading cause of cancer death in the United States and Europe—so we had no basis for comparison.18 We considered the same set of nine European countries (Finland, France, Germany, Iceland, Norway, Scotland, Slovakia, Slovenia, and Sweden) and the same study period (1982–2005) used by Philipson and coauthors. In addition to calculating cancer deaths averted using mortality rates, we also statistically modeled the number of cancer deaths, closely following the approach of Philipson and coauthors. For a comparison across other cancers, see Appendix Exhibit 4.14 Sensitivity Analyses We conducted multiple sensitivity analyses to assess the robustness of our findings with respect to the time interval chosen and the handling of missing data. First, we calculated the number of US cancer deaths averted after the years of peak US mortality for breast, colorectal, and prostate cancer. By doing this, we were able to more directly attribute the cancer deaths averted after those peaks to the introduction of effective screening and treatment. Second, we considered two alternative approaches to missing cancer mortality rates: multiple imputation and fitting an ordinary least squares regression line on the observed agegroup-specific mortality rate—instead of on its logarithm, as in the main analysis—as a function of time (for an example of these alternative approaches, see Appendix Exhibit 2).14 Limitations We acknowledge several limitations of our study. First, our results may be subject to bias due to misclassification of the underlying cause of death on death certificates. Earlier studies found that the classification of cause of death on death certificates was more accurate for cancer than for other causes, however.19 Moreover, the agreements in the cause of death between medical records and death certificates for breast, lung, and prostate cancers were among the highest across all cancer types.20 Second, the validity of our results depends, in
part, on consistent coding of cancer deaths across countries and over time. Cross-national comparisons found breast, colorectal, lung, and prostate cancers to be among the mostly consistently coded cancers.21 Our study period included coding according to both the International Classification of Diseases, Ninth Revision (ICD-9), and the Tenth Revision (ICD-10). However, we did not observe any sharp changes in cancer mortality rates in the year after a country transitioned to ICD-10.
Study Results Cancer Deaths Averted Between 1982 and 1986, and again in 1990, US breast cancer mortality rates exceeded those of Western Europe (the comparator). This resulted in 2,592 excess breast cancer deaths, which partially offset the 69,389 averted deaths that occurred after 1986 (Exhibit 1). US colorectal cancer mortality rates were lower than those of Western European countries throughout the study period and led to 264,632 averted deaths (Exhibit 2). In contrast, US lung cancer mortality rates exceeded those of the comparator countries throughout the study period and led to 1,119,599 excess deaths (Appendix Exhibit 3).14 Finally, between 1982 and 1995 (except in 1985), higher US prostate cancer mortality rates led to 11,759 excess deaths, which partially offset the 71,641 averted deaths that occurred after 1995 (Exhibit 3). We present the number of cancer deaths averted for the additional eight cancers in Appendix Exhibit 3.14 Value Of Cancer Care The cost of breast cancer care in the United States exceeded that in Western Europe by $435 billion between 1982 and 2010 (Exhibit 4). During the same time period, the 66,797 averted breast cancer deaths overall led to 1,082,013 QALYs saved. Thus, the ratio of incremental cost to QALYs saved equaled $402,369. For colorectal cancer, the ratio equaled $110,009 ($326 billion and 2,962,177 QALYs saved). For lung cancer, the 1,119,599 excess deaths led to 21,571,483 QALYs lost, while the cost of lung cancer care in the United States exceeded that in Western Europe by $406 billion. Thus, the ratio of incremental cost to QALYs saved equaled −$18,815. For prostate cancer, the ratio equaled $1,978,542 ($435 billion and 219,678 QALYs saved). Recalculation Of Previous Estimates Of Cancer Deaths Averted We could not reproduce the number of averted US breast and prostate cancer deaths calculated by Philipson and coauthors. Using the same WHO cancer mortal-
ity rate data and same set of nine European countries, we calculated 19,518 averted breast cancer deaths and 87,832 averted prostate cancer deaths between 1982 and 2005, which represents 78 percent fewer breast cancer deaths and 60 percent fewer prostate cancer deaths averted than reported by Philipson and coauthors. For a comparison across other cancers, see Appendix Exhibit 4.14 We calculated more averted colorectal cancer deaths for the United States, compared to the nine European countries, than Philipson and coauthors did: 291,273 averted deaths versus 281,932. Finally, we calculated 482,191 averted stomach cancer deaths for the United States compared to the nine European countries, in contrast to the 224,212 excess stomach cancer deaths estimated by Philipson and coauthors. Sensitivity Analyses US cancer mortality rates peaked in 1985 for breast cancer, 1984 for colorectal cancer, and 1991 for prostate cancer. After these peak years, declining US mortality rates led to 68,872 breast cancer deaths averted, 261,859 colorectal cancer deaths averted, and 62,385 prostate cancer deaths averted. We reached nearly identical results for the number of US cancer deaths averted based on alternative approaches to missing cancer mortality rates, as shown in Appendix Exhibit 5.14 For example, using multiple imputation, we calculated that 65,781 breast cancer deaths and 59,129 prostate cancer deaths were averted.
Discussion Our study has two central findings. First, the number of deaths averted in the United States compared to all of Western Europe from 1982 to 2010 varied substantially by cancer type. The
Exhibit 1 Excess And Averted Breast Cancer Deaths In The United States, Compared To Western Europe, 1982–2010
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text).
March 2015
34:3
H e a lt h A f fai r s
393
Paying For Value Exhibit 2 Excess And Averted Colorectal Cancer Deaths In The United States, Compared To Western Europe, 1982–2010
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text).
United States averted the largest number of deaths for stomach and colorectal cancer and experienced the largest number of excess deaths for lung cancer and non-Hodgkin’s lymphoma (Exhibit 4). The United States averted a modest number of breast and prostate cancer deaths. Second, the additional value derived from costlier US cancer care also varied considerably by cancer type. For example, the ratio of incremental cost to QALYs saved equaled $402,369 for breast cancer, $110,009 for colorectal cancer, $1,978,542 for prostate cancer, and −$18,815 for lung cancer (Exhibit 4). Deaths Averted And Cancer Type The relatively small number of averted breast cancer deaths that we calculated resulted from similar levels of breast cancer mortality rates in the United States and Western Europe that declined at a comparable pace over time. This transnational
Exhibit 3 Excess And Averted Prostate Cancer Deaths In The United States, Compared to Western Europe,1982–2010
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text).
394
Health Affai rs
March 2 015
3 4: 3
pattern of mortality decline may be due, in part, to widely adopted advances in breast cancer treatment, such as the use of tamoxifen and breast-conserving surgery with adjuvant radiotherapy.22 The use of trastuzumab, which began earlier in the United States than in Western Europe, may have contributed to slightly steeper declines in US breast cancer mortality rates in the 2000s. In contrast to treatment, mammography screening rates varied widely among countries and likely did not lead to the transnational pattern of mortality decline.23 If screening led to declines in breast cancer mortality, we would expect increases in screening to precede decreases in mortality by five or more years, based on results of mammography screening trials. In addition, we would expect to see countries with comparatively higher screening rates experience faster declines in breast cancer mortality. However, we observed neither outcome. For example, US mammography screening rates peaked in 2000, fifteen years after the peak in breast cancer mortality rates.24 A recent study of three pairs of European countries with similar health care services and prevalence of risk factors for breast cancer mortality found similar declines in breast cancer mortality, despite substantial differences in the timing and level of the countries’ national screening programs.25 Finally, differences in breast cancer diagnosis could affect survival time from diagnosis through leadtime and overdiagnosis biases. But because those differences do not affect mortality rates, they cannot explain the transnational pattern of declining breast cancer mortality. US prostate cancer mortality rates declined faster over time than concurrently decreasing European rates. However, the causes of US and European declines remain largely unknown.26 Observational studies and clinical trials produce conflicting results on the potential protective effect against prostate cancer of statin drug use, which began earlier and remains at higher levels in the United States than in Europe.27 The role of screening in reducing prostate cancer mortality is also ambiguous. Indeed, prostate-specific antigen (PSA) testing began earlier in the United States than in Western Europe, and the rise in PSA testing preceded the decline in US prostate cancer mortality.28 Yet clinical trials do not demonstrate any short-term mortality benefits and produce conflicting conclusions regarding the long-term benefits of PSA testing.29 Compared to screening and prevention, the United States and Western Europe were more equal in their adoption of advances in neoadjuvant and adjuvant hormonal therapy, which may have contributed to their similar declines in
Exhibit 4 Ratio Of Incremental Costs To Total And Quality-Adjusted Life-Years Saved For Twelve Cancer Types In The United States, Compared To Western Europe Type of cancer Breast Cervix uteri Colorectal Hodgkin’s lymphoma Leukemia Lung Melanoma Non-Hodgkin’s lymphoma Prostate Stomach Testis Thyroid gland
Incremental cost ($ millions) 435,369 29,287
Incremental cost divided by life-years saved ($) 306,544 −712,751
Incremental cost divided by QALYs saved ($) 402,369 −855,019
Deaths averted 66,797 4,354
Life-years saved 1,420,249 −41,090
264,632 4,859
3,860,194 161,074
325,866 19,526
84,417 121,224
110,009 156,045
−64,530 −1,119,599
−1,370,884 −28,311,995
32,620 405,872
−23,795 −14,336
−30,790 −18,815
−39,144 −164,429
−1,025,066 −3,696,336
109,773 117,692
−107,089 −31,840
−136,592 −41,362
59,882 621,820
294,273 13,705,501
434,642 49,015
1,477,003 3,576
1,978,542 4,635
3,372 18,320
100,832 482,902
17,714 51,593
175,674 106,840
222,839 139,681
SOURCE Authors’ analysis of data from the World Health Organization Cancer Mortality Database and the Human Mortality Database (see Note 13 in text). NOTES Negative deaths averted represent excess cancer deaths. Incremental costs are expressed in 2010 dollars.
prostate cancer mortality rates.28 In contrast to breast and prostate cancer, the larger number of averted colorectal cancer deaths in the United States was likely due to screening, which began earlier and increased to higher levels in the United States, compared to Western Europe. Clinically, screening has been shown to reduce mortality in several large-scale randomized controlled trials of fecal occult blood testing and sigmoidoscopy and in colonoscopy case-control studies.30,31 Temporally, increases in colorectal cancer screening preceded the steepest declines in colorectal cancer mortality by at least five years in the United States.32 Screening was previously estimated to explain 53 percent of the observed decline in colorectal cancer mortality in the United States between 1975 and 2000, whereas changes in risk factors and treatment explained 35 percent and 12 percent, respectively.33 Countries with nationally organized screening programs experienced sharper declines in colorectal cancer mortality than countries without screening programs.34 In addition to higher screening rates, greater use of curative-intent surgery and adjuvant chemotherapy in the United States likely contributed to averted colorectal cancer deaths.35 Despite lower historical rates of cigarette smoking, lung cancer mortality rates in the United States exceeded those in Western Europe throughout the study period. After US lung cancer mortality rates peaked in 1993, they declined faster than Western European rates, which is almost certainly due to faster US reductions in cigarette smoking.36 If the current pace of decline persists, lung cancer mortality rates in the
United States may eventually fall below those in Western Europe and lead to averted lung cancer deaths. The large number of averted stomach cancer deaths may be due to lower rates of Helicobacter pylori infection and more substantial improvements in food preparation in the United States than in Western Europe. More than 60 percent of stomach cancer cases are attributed to H. pylori infection.37 The preparation of meats by salting or smoking increases the intake of carcinogenic nitrosamines. Surgery has long been the most common treatment for early-stage stomach cancer in the United States and Western Europe. However, clinical trials produce conflicting conclusions about whether more extensive lymph node dissection and the use of adjuvant therapy improve survival.38 Differences From Previous Estimates Of Cancer Deaths Averted Our findings, based on both the demographic and statistical approaches we used to calculate deaths averted, differ considerably from those of Philipson and coauthors, especially in the cases of breast and prostate cancer. For both of these types of cancer, we observed fewer deaths averted than Philipson and coauthors did, because of methodological differences. We argue that the interpretation of the statistical model by Philipson and coauthors produces several implausible results. First, Philipson and coauthors calculated that the prostate cancer mortality rate of the United States declined by 15.8 deaths more per 100,000 people than the corresponding decline in the prostate cancer mortality rate of nine European countries, which is far from what actually ocMarch 2015
34:3
H e a lt h A f fai r s
395
Paying For Value curred. The observed prostate cancer mortality rate of the United States declined by 14.3 deaths per 100,000 (from 54.9 deaths per 100,000 in 1982 to 40.6 deaths per 100,000 in 2005). During the same period, the observed prostate cancer mortality rate of the nine European countries declined by 5.2 deaths per 100,000 (from 57.1 to 51.9). Thus, the US rate declined by only 9.1 more deaths per 100,000 than the European rate. Second, Philipson and coauthors calculated that there were 224,212 excess stomach cancer deaths for the United States, compared to the nine European countries. But between 1982 and 2005, stomach cancer mortality rates in the United States were always substantially below those in the nine European countries. Thus, there could only be averted—not excess— stomach cancer deaths in the United States. Policy Implications Our study highlights two potentially actionable areas of focus for US policy makers. First, the greatest number of deaths averted occurred in cancers for which decreasing mortality rates were more likely to be the result of successful prevention and screening than advancements in treatment. The Affordable Care Act (ACA) led to full coverage of annual wellness visits for the elderly that address weight loss, smoking cessation, and nutrition. The ACA also eliminated copayments and deductibles for Medicare-covered preventive services that received a grade of A or B from the US Preventive Services Task Force. If greater access to wellness visits and preventive services translates to greater use of these types of medical care, the ACA may lead to more cancer deaths’ being averted and may do so in a cost-effective manner. Second, new payment models that incentivize Results from this study were presented in part at the thirty-eighth annual meeting of the American Society of Preventive Oncology, Arlington, Virginia, March 8–11, 2014. This research was supported by funding from the National Cancer Institute (Grant No. RC2CA148259), the National
high-value cancer care may lead to more cancer deaths being averted and reduce the cost of cancer care in the United States. A recent international commission identified three areas in which cancer care costs could be reduced: endof-life care, medical imaging, and chemotherapeutic agents.39 For example, palliative care provides higher value to patients with terminal cancer because it almost always is less expensive than chemotherapy, is aligned with patients’ preferences, and improves the quality of life.40 Health care systems and insurers that form accountable care organizations (for example, Medicare Pioneer ACOs) would be able to share in cost savings from the use of palliative care instead of intensive care.
Conclusion We calculated the number of cancer deaths averted for the United States, compared to Western Europe, and the value of US cancer care measured by the ratio of incremental costs to QALYs saved. The cost-effectiveness ratio was lower for colorectal cancer than for breast and prostate cancer, and it was negative for lung cancer. This indicates that prevention, screening, and treatment may be more cost-effective in the case of colorectal cancer than in the case of other leading cancers. Additionally, the number of breast and prostate cancer deaths averted was smaller than a previous study calculated. Our results suggest that cancer care in the United States did not always avert deaths compared to Western Europe, and when it did avert deaths, it often did so at substantial cost. ▪
Institute on Aging (Grant No. PO1AG19783), and the National Center for Advancing Translational Sciences (Award No. KL2TR001088), all at the National Institutes of Health. The funding agencies had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data;
or in the preparation, review, or approval of the manuscript. The authors thank Valerie Lewis, Jonathan Skinner, Shila Soneji, Harold Sox, and three anonymous reviewers for helpful comments and suggestions.
NOTES 1 Reinhardt UE, Hussey PS, Anderson GF. U.S. health care spending in an international context. Health Aff (Millwood). 2004; 23(3):10–25. 2 Thorpe KE, Florence CS, Joski P. Which medical conditions account for the rise in health care spending? Health Aff (Millwood). 2004;23(4): w4-437–45. 3 National Cancer Institute. Surveillance, Epidemiology, and End Results Program [home page on the Internet]. Bethesda (MD): NCI; [cited 2015 Jan 30]. Available from:
396
Health A ffairs
M a r c h 20 1 5
34:3
http://www.seer.cancer.gov 4 National Center for Health Statistics [home page on the Internet]. Hyattsville (MD): NCHS; [last updated 2015 Jan 30; cited 2015 Feb 10]. Available from: http:// www.cdc.gov/nchs/products/ vsus.htm 5 Cutler DM. Are we finally winning the war on cancer? J Econ Perspect. 2008;22(4):3–26. 6 Philipson T, Eber M, Lakdawalla D, Corral M, Conti R, Goldman D. An analysis of whether higher health
care spending in the United States versus Europe is “worth it” in the case of cancer. Health Aff (Millwood). 2012;31(4):667–75. 7 Council of Economic Advisers. Economic report of the president [Internet]. Washington (DC): Government Printing Office; 2013 Mar [cited 2015 Jan 22]. Available from: http://www.whitehouse.gov/sites/ default/files/docs/erp2013/full_ 2013_economic_report_of_the_ president.pdf 8 Autier P, Boniol M, Héry C, Masuyer
9
10
11
12
13
14
15
16
17
18
E, Ferlay J. Cancer survival statistics should be viewed with caution. Lancet Oncol. 2007;8(12):1050–2. Sant M, Allemani C, Capocaccia R, Hakulinen T, Aareleid T, Coebergh J, et al. Stage at diagnosis is a key explanation of differences in breast cancer survival across Europe. Int J Cancer. 2003;106(3):416–22. Ciccolallo L, Capocaccia R, Coleman MP, Berrino F, Coebergh JWW, Damhuis RAM, et al. Survival differences between European and US patients with colorectal cancer: role of stage at diagnosis and surgery. Gut. 2005;54(2):268–73. Squires DA. The U.S. health system in perspective: a comparison of twelve industrialized nations [Internet]. New York (NY): Commonwealth Fund; 2011 Jul [cited 2015 Jan 30]. (Issue Brief). Available from: http://www.commonwealth fund.org/~/media/files/ publications/issue-brief/2011/jul/ 1532_squires_us_hlt_sys_ comparison_12_nations_intl_brief_ v2.pdf National Research Council. Explaining divergent levels of longevity in high-income countries. Washington (DC): National Academies Press; 2011. Human Mortality Database [home page on the Internet]. Berkeley (CA): The Database; [cited 2015 Jan 22]. Available from: http://www .mortality.org To access the Appendix, click on the Appendix link in the box to the right of the article online. Crawford R, Emmerson C, Tetlow G. A survey of public spending in the UK [Internet]. London: Institute for Fiscal Studies; 2009 Sep [cited 2015 Jan 22]. (IFS Briefing Note No. BN43). Available from: http:// www.ifs.org.uk/bns/bn43.pdf De Angelis R, Sant M, Coleman MP, Francisci S, Baili P, Pierannunzio D, et al. Cancer survival in Europe 1999–2007 by country and age: results of EUROCARE-5—a population-based study. Lancet Oncol. 2014;15(1):23–34. Fryback D, Dunham NC, Palta M, Hanmer J, Buechner J, Cherepanov D, et al. US norms for six generic health-related quality-of-life indexes from the National Health Measurement study. Med Care. 2007;45(12): 1162–70. Jemal A, Center MM, DeSantis C,
19
20
21
22
23
24
25
26
27
28
29
Ward EM. Global patterns of cancer incidence and mortality rates and trends. Cancer Epidemiol Biomarkers Prev. 2010;19(8):1893–907. Sarfati D, Blakely T, Pearce N. Measuring cancer survival in populations: relative survival vs cancerspecific survival. Int J Epidemiol. 2010;39(2):598–610. German RR, Fink AK, Heron M, Stewart SL, Johnson CJ, Finch JL, et al. The accuracy of cancer mortality statistics based on death certificates in the United States. Cancer Epidemiol. 2011;35(2):126–31. Percy C, Muir C. The international comparability of cancer mortality data. Results of an international death certificate study. Am J Epidemiol. 1989;129(5):934–46. Allemani C, Sant M, Weir HK, Richardson LC, Baili P, Storm H, et al. Breast cancer survival in the US and Europe: a CONCORD highresolution study. Int J Cancer. 2013; 132(5):1170–81. Saika K, Sobue T. Time trends in breast cancer screening rates in the OECD countries. Jpn J Clin Oncol. 2011;41(4):591–2. Breen N, A Cronin K, Meissner HI, Taplin SH, Tangka FK, Tiro JA, et al. Reported drop in mammography: is this cause for concern? Cancer. 2007;109(12):2405–9. Autier P, Boniol M, Gavin A, Vatten LJ. Breast cancer mortality in neighbouring European countries with different levels of screening but similar access to treatment: trend analysis of WHO Mortality Database. BMJ. 2011;343:d4411. Collin SM, Martin RM, Metcalfe C, Gunnell D, Albertsen PC, Neal D, et al. Prostate-cancer mortality in the USA and UK in 1975–2004: an ecological study. Lancet Oncol. 2008; 9(5):445–52. Bonovas S, Filioussi K, Sitaras NM. Statin use and the risk of prostate cancer: a metaanalysis of 6 randomized clinical trials and 13 observational studies. Int J Cancer. 2008;123(4):899–904. Neppl-Huber C, Zappa M, Coebergh JW, Rapiti E, Rachtan J, Holleczek B, et al. Changes in incidence, survival, and mortality of prostate cancer in Europe and the United States in the PSA era: additional diagnoses and avoided deaths. Ann Oncol. 2011; 23(5):1325–34. Hayes JH, Barry MJ. Screening for
30
31
32
33
34
35
36
37
38
39
40
prostate cancer with the prostatespecific antigen test: a review of current evidence. JAMA. 2014; 311(11):1143–9. Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med. 1993;328(19):1365–71. Kahi CJ, Myers LJ, Slaven JE, Haggstrom D, Pohl H, Robertson DJ, et al. Lower endoscopy reduces colorectal cancer incidence in older individuals. Gastroenterology. 2014; 146(3):718–25. Hoff G, Dominitz JA. Contrasting US and European approaches to colorectal cancer screening: which is best? Gut. 2010;59(3):407–14. Edwards BK, Ward E, Kohler BA, Eheman C, Zauber AG, Anderson RN, et al. Annual report to the nation on the status of cancer, 1975–2006. Cancer. 2010;116(3):544–73. Bosetti C, Levi F, Rosato V, Bertuccio P, Lucchini F, Negri E, et al. Recent trends in colorectal cancer mortality in Europe. Int J Cancer. 2011; 129(1):180–9. Allemani C, Rachet B, Weir HK, Richardson LC, Lepage C, Faivre J, et al. Colorectal cancer survival in the USA and Europe: a CONCORD high-resolution study. BMJ Open. 2013;3(9):e003055. Devesa SS, Bray F, Vizcaino AP, Parkin DM. International lung cancer trends by histologic type: male: female differences diminishing and adenocarcinoma rates rising. Int J Cancer. 2005;117(2):294–9. Parkin DM. The global health burden of infection-associated cancers in the year 2002. Int J Cancer. 2006;118(12):3030–44. Russell MC, Mansfield PF. Surgical approaches to gastric cancer. J Surg Oncol. 2013;107(3):250–8. Kelly RJ, Smith TJ. Delivering maximum clinical benefit at an affordable price: engaging stakeholders in cancer care. Lancet Oncol. 2014; 15(3):e112–8. Schnipper LE, Smith TJ, Raghavan D, Blayney DW, Ganz PA, Mulvey TM, et al. American Society of Clinical Oncology identifies five key opportunities to improve care and reduce costs. J Clin Oncol. 2012; 30(14):1715–24.
March 2 015
3 4: 3
Health Affa irs
397
