|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 1999 American Society for Clinical Oncology Associations Between Community Income and Cancer Survival in Ontario, Canada, and the United StatesFrom the The Radiation Oncology Research Unit and Departments of Oncology and Community Health and Epidemiology, Queen's University, Kingston Regional Cancer Centre, Kingston; and Kingston General Hospital, Kingston, Ontario, Canada. Address reprint requests to William J. Mackillop, MD, Radiation Oncology Research Unit, Kingston General Hospital, Apps Level 4, Kingston, Ontario K7L 2V7 Canada; email william.mackillop{at}cancercare.on.ca
PURPOSE: The objectives of this study were as follows: (1) to compare the magnitude of the association between socioeconomic status (SES) and cancer survival in the Canadian province of Ontario with that in the United States (U.S.), and (2) to compare cancer survival in communities with similar SES in Ontario and in the U.S. METHODS: The Ontario Cancer Registry provided information about all cases of invasive cancer diagnosed in Ontario from 1987 to 1992, and the Surveillance, Epidemiology and End Results Registry (SEER) provided information about all cases diagnosed in the SEER regions of the U.S. during the same time period. Census data provided information about SES at the community level. The product-limit method was used to describe cause-specific survival. Cox proportional hazards models were used to describe the association between SES and the risk of death from cancer. RESULTS: There were significant associations between SES and survival for most cancer sites in both the U.S. and Ontario, but the magnitude of the association was usually larger in the U.S. In the poorest communities, there were significant survival advantages in favor of cancer patients in Ontario for many disease groups, including cancers of the lung, head and neck region, cervix, and uterus. However, in upper- and middle-income communities, there were significant survival advantages in favor of the U.S. for all cases combined and for several individual diseases, including cancers of the breast, colon and rectum, prostate, and bladder. CONCLUSION: The association between SES and cancer survival is weaker in Ontario than it is in the U.S. This is due to a combination of better survival among patients in the poorest communities and worse survival among patients in the wealthier communities of Ontario relative to those in the U.S.
THE ASSOCIATION between socioeconomic status (SES) and cancer survival is well established.1-18 The phenomenon has been described in several different countries using many different measures of SES, including income,1-3,6,11,16-18 education,1,3,8,11,14,16 occupation,4,10,12,13,15 health insurance status,2,7 and composite indices derived from several individual indicators.3,6 Qualitatively similar results have been obtained, whether measures of the patient's SES or measures of the SES of the community in which the patient resides are used in the analysis. The magnitude of the association varies from one cancer site to another, but the outcome of cancer is almost always worse among lower SES groups.1-18 There are several potential reasons for the association between SES and survival. One explanation is that wealthier people have better access to care and/or obtain care that is of better quality than that of poorer people.1-3,10-12,19-22 It might, therefore, be expected that the magnitude of the association between SES and survival would be smaller in countries that have publicly funded health care systems than in those that do not, although significant associations between SES and survival have been demonstrated in Canada,1,2,8,11,17,18 the United Kingdom,3 and the Netherlands,4,5,10,12,13 all of which have more or less comprehensive government-sponsored health systems. It is not possible, based on published reports, to compare the strengths of the associations observed in different countries, because different methods were used in each national study. Therefore, we decided to compare, in a single study using the same methodology, the magnitude of the association between SES and cancer survival between two countries with different health systems. We chose to study Canada and the United States (U.S.), two neighboring countries with quite similar per capita incomes23 but with very different health care systems. In Canada, most health care for all residents is provided by universal, comprehensive, provincial health insurance schemes.24,25 Residents make no direct payment for medical services or hospitalization, and there is no parallel private sector. In the U.S., health care is delivered primarily by a fee-for-service system, with the main source of health insurance coverage for most of the population being voluntary employment-based private health care plans.26,27 Government-sponsored insurance programs such as Medicaid and Medicare cover the indigent and elderly, but more than 10% of the population remains uninsured at any given time.27-29 Our hypothesis was that the magnitude of the association between SES and survival would be weaker in Canada than in the U.S., because Ontario's system might mitigate the adverse impact of poverty on cancer outcome by removing barriers to care for the poor. In a previous study, we showed that there were clinically important and statistically significant differences in survival between wealthier and poorer communities in Ontario.17 However, a subsequent study showed that SES was more strongly associated with cancer survival in the U.S. city of Detroit, Michigan, than in the Canadian city of Toronto, Ontario,18 and the authors of that study concluded that there was "pronounced socioeconomic inequality" in cancer survival in the U.S. compared with "Canada's consistently egalitarian distribution." A study confined to two cities did not seem to us to provide a sufficient basis for generalization, and, therefore, we have extended these observations to the entire province of Ontario and to all nine regions of the Surveillance, Epidemiology and End Results (SEER) registries, which cover a large and fairly representative part of the U.S. population.30 The specific objectives of this study were as follows: (1) to describe the association between community median household income and cause-specific survival in the province of Ontario and in the SEER regions of the U.S., (2) to test the hypothesis that the SES-associated survival gradient is smaller in Ontario than in the U.S., and (3) to compare survival in similar SES groups in Ontario and the U.S.
Study Design All incident cases of cancer diagnosed between 1987 and 1992 in the Canadian province of Ontario and in the regions of the U.S. covered by the SEER program were included. Cases that were not confirmed microscopically, second primary tumors, and nonmelanoma skin cancers were excluded, as were tumors arising in patients who were younger than 20 years of age at the time of diagnosis. Information about the SES of the community in which the patient resided at the time of diagnosis was linked to the Ontario registry from the 1991 Canadian census31 and to the SEER registry from the 1990 U.S. census.32 The populations of Ontario and of the SEER regions of the U.S. were each divided into quintiles based on community median household income. Cancer outcomes were compared between countries, among socioeconomic groups within each country, and within comparable socioeconomic groups between countries.
Sources of Data Ontario Cancer Registry. The Ontario Cancer Registry (OCR) is a population-based tumor registry operated by Cancer Care Ontario.36 It is estimated that it captures 98% of all incident cases in the province of Ontario.37 The OCR provided records of all cases of invasive cancer diagnosed in Ontario between 1987 and 1992. The file included the following variables: ICD-9 code, date of diagnosis, date of birth, date of last contact, cause of death, sex, postal code, and Ministry of Health residence code. 1990 U.S. census. Socioeconomic data about the U.S. population were obtained from the 1990 U.S. census at census tract (median population of approximately 3,800) and county level (median population of approximately 19,000). The geographic codes used by the U.S. census were available in the SEER database and it was, therefore, straightforward to attribute socioeconomic descriptors to each cancer case based on the patient's place of residence at the time of diagnosis. Approximately 85% of cases had community SES data linked at the level of the census tract, 15% had SES data linked at the county level, and in only 0.008% of cases was neither linkage possible. The linked database contained 486,327 cases. 1991 Canadian census. The 1991 Canadian census contained socioeconomic data at the level of the census enumeration area (EA, median population of approximately 700) and at the level of the census subdivision (CSD, median population of approximately 2,000). Two sets of geographic codes were available in the cancer registry: postal code (mean population of approximately 50) and a "residence code" (mean population of approximately 9,000) used by the Ministry of Health of Ontario. Correspondences between postal codes and EA codes and between residence codes and CSD codes had been previously established and were encoded in conversion files made available to us by Statistics Canada and the Ontario Ministry of Health. Census data were linked to the cancer registry from EA to postal code when the postal code was available and from CSD to residence code when the postal code was not available, as described previously.17 Approximately 88% of cases had community SES data linked to them by postal code and 10.0% by residence code, but in 2.3% neither linkage was possible. The linked database contained 187,650 cases.
Creating SES Groups
Analysis of Survival and Cause of Death
Median Household Income as an Indicator of SES Figure 1 shows the frequency distribution of household incomes within the SES quintiles created based on median household incomes in Ontario and in the areas of the U.S. covered by the SEER registries. Canadian incomes are in 1991 Canadian dollars, whereas U.S. incomes are in 1990 U.S. dollars. The shapes of the distributions observed in Ontario and the U.S. were quite similar for all corresponding quintiles. Each quintile contains a broad distribution of incomes, but there were few households of extremely high income in the bottom quintile, and few households of extremely low income in the top quintile. Note that the income groupings used to describe the U.S. and Ontario distributions were not identical, because the choice of cutoff points was constrained by the available census data. Figure 2 shows the median income in each quintile relative to the median for the whole population. At the two extremes, the gradient is a little steeper in the U.S. than in Ontario.
Comparison of Survival Between Ontario and the U.S.
Associations Between Income and Cause-Specific Survival
Figure 3 illustrates the relationship between income group and cause-specific survival for three selected diseases. The income-specific survival curves for lung cancer and breast cancer both show an association with SES that was stronger in the U.S. than in Ontario. In both of these diseases, the trends observed seem relatively constant across the whole range of incomes in Ontario, but the differences get larger at the lower end of the income spectrum in the U.S. Similar trends were seen in several other diseases (Table 2). In some diseases, the relationships between community income and survival seemed to be more complex. For example, in cervical cancer, 5-year survival rates among patients in the lowest four quintiles in Ontario were similar, and only patients in the top quintile stood out as faring better, whereas in the U.S., 5-year survival rates among patients in the four highest quintiles were similar, and only patients in the bottom quintile stood out as faring worse.
Associations Between Income and the Risk of Death from Cancer We first carried out separate analyses in Ontario and the U.S., treating income quintile as a continuous variable. This approach provides an overall estimate of the income-associated trend in survival in each country.3 The results of this analysis are shown in the first column of Table 3. Both in Ontario and in the U.S., there were significant associations between income and the risk of cancer death in most disease groups. The trend was always toward a higher risk of cancer death among the poor. The trend was generally stronger in the U.S. than in Ontario; it was significantly stronger for several individual disease groups and for all diseases combined (excluding prostate cancer). There was no disease in which the SES-associated trend in survival was significantly stronger in Ontario.
In a second set of Cox models, the Ontario and U.S. cases were analyzed together, and the risk of cancer death in each socioeconomic group in each country was referenced to the risk of death in the third quintile of the U.S. population. These data are also shown in Table 3. Reading across the table permits comparison of the risk of cancer death across socioeconomic strata within each country. It confirms that there are significant gradients between wealthy and poor income groups in both Ontario and in the U.S., and also that those differences tend to be greater in the U.S. The trends in both countries were fairly even across the four wealthiest income groups, but the increase in risk between the second-poorest group and the poorest group was usually larger in the U.S. than in Ontario. Tables 2 and 3 also permit comparison of cancer outcomes between similar socioeconomic groups in Ontario and the U.S. In the majority of cancers, cause-specific survival was lower and the risk of death from cancer was higher in the poorest U.S. communities than it was in the poorest Ontario communities. In contrast, cause-specific survival was lower and the risk of death from cancer was higher in the wealthiest Ontario communities than it was in the wealthiest U.S. communities. The findings in the middle and upper middle-income groups were generally similar to those in the wealthiest group, although the differences between Ontario and the U.S. were smaller. Figure 4 displays these relative risks for selected sites. In head and neck cancer, the income-associated gradient across the top four quintiles was similar in Ontario and in the U.S., but in the fifth quintile, the relative risk of death climbed steeply in the U.S., whereas it increased only slightly in Ontario. In cancer of the stomach, there was only a weak association between community income and the risk of death in both countries. In breast cancer, the relative risk of death was lower in the U.S. than in Ontario in the top four quintiles, but the relative risks observed in the two countries converged in the bottom quintile. For all diseases combined (excluding prostate cancer), the relative risk of dying in the poorest group was higher in the U.S. than in Ontario, but the relative risk of death among the upper income groups was higher in Ontario than in the U.S.
The results presented in Table 3 and illustrated in Fig 4 are based on a Cox analysis that implicitly assumes that the difference in SES between the first and second quintile is the same as the difference between the second and the third quintile, and so on. Furthermore, the U.S./Ontario comparisons assume that the SES gradient across quintiles is the same in the two countries, although we know from Fig 2 that this is not so. We have, therefore, been comparing the impact of socioeconomic rank, rather than SES, on survival. Table 4 shows the results obtained when the median income in each quintile, relative to the median income of the entire population, is entered directly into the model as a continuous variable alongside the results obtained using SES rank. To facilitate comparison of the two sets of data, differences in the strength of the associations between the U.S. and Ontario are described as a ratio of the strength of the association in Canada to that of the U.S., expressed as a percentage. In all of those diseases in which the SES-associated survival gradient is steeper in the U.S. than in Canada, the difference between the two countries is smaller when relative income as opposed to rank is used in the model. In those diseases in which the SES-associated gradient is steeper in Ontario, the difference is larger. The differences reflect the steeper income gradient across quintiles in the U.S. compared with that in Ontario.
In this study, we confirmed that there was a significant association between SES and cause-specific survival in many disease groups both in the U.S. and Ontario. Consistent with our initial hypothesis, we found that the association between SES and survival was generally stronger in the U.S. than in Ontario. Unexpectedly, however, we found that the weaker association between SES and survival in Ontario owed as much to poorer survival among the upper- and middle-income groups as to better survival among the poor. The validity of these comparisons between residents of wealthier and poorer neighborhoods, and between the populations of Ontario and the SEER regions of the U.S., depends on whether there are any systematic errors that might bias the results, and several specific aspects of this study deserve discussion. First, the date of diagnosis is not assigned in exactly the same way in the two cancer registries. The date recorded in the Ontario registry is the date of the histologic or cytologic diagnosis, whereas the date in the SEER registries is the date when the diagnosis was first mentioned in the record; in actuality, this date may precede the microscopic diagnosis. This difference would not bias comparisons among SES groups but might give the U.S. an apparent advantage in survival over Ontario. It is improbable that this would have an important effect on survival at 5 years, but even small differences in the assigned date of diagnosis might affect the early part of the survival curve to an extent that would be statistically significant. We addressed this concern by carrying out secondary analyses that looked at cause-specific survival, conditional on survival for 1 month after diagnosis, to reduce the impact of imprecisions in the date of diagnosis. This did not significantly affect any of our results. Furthermore, this type of error would be expected to have a similar effect on all diseases, and we consider it improbable that it was responsible for the disease-specific differences in survival that we observed in this study. Second, there were differences between the Ontario and SEER registries in the procedures for ascertainment of deaths. Deaths of SEER cases are identified by linkage of the registry to U.S. national death records,44 whereas deaths of Ontario cases are identified by linkage to provincial death records only. Thus deaths among Ontario patients who have migrated out of Ontario are not necessarily identified. This may result in a small bias in survival in favor of Ontario. The impact of the problem has not been precisely quantified, but analysis of provincial migration statistics indicates that any effect on our survival data will be less than 1% (data not presented). Analysis of Canadian census data also showed that there is only a weak correlation between median household income and interprovincial migration (r = .08), which reduces concern that this may have affected the observed SES-associated survival gradient in Ontario. Third, the proportion of deaths attributed to causes other than cancer was slightly higher in SEER than in Ontario. We are aware of no difference in the process for assignment of cause of death between Ontario and the U.S., but if cancer deaths in the U.S. were more likely to be miscoded as deaths from other causes, or if deaths from other causes in Ontario were more likely to be miscoded as deaths from cancer, then this might result in an advantage in cause-specific survival in the U.S. However, the observed excess in the rate of death from causes other than cancer in the U.S. corresponds almost exactly to the excess in all-cause mortality predicted from the life-tables.38-41 There is no reason to believe that the recording of cause of death would be more accurate in one social group than in another, and it seems unlikely that the observed SES-associated gradients in survival could have been produced by this type of artifact. There was a higher proportion of deaths from unknown causes in the U.S. (3.6%) as compared with Ontario (1.1%). The majority of these deaths were probably due to cancer, and in the analyses presented above, they were, therefore, attributed to cancer. However, this must have produced a bias in cause-specific survival in favor of Ontario, the magnitude of which depends on the proportion of these deaths that were in fact due to causes other than cancer. We did a secondary analysis in which the deaths from unknown causes were censored. This increased the survival advantages in favor of the U.S., and decreased the advantages in favor of Ontario, but not to an extent that would affect our main conclusions. We found no association between SES and the frequency of missing information about cause of death, and, therefore, there was little potential for this to bias comparisons across socioeconomic strata. Fourth, our choice of SES indicator must have influenced the details of our results to some extent. However, median household income is strongly correlated with most other census indicators of SES,17 and similar results are obtained when other census indicators are used instead.45 Perhaps of greater concern than the choice of census indicator is the size of the census unit used to attribute community SES to the individual case. The average population of the census units from which socioeconomic information was linked to the registry in Ontario was smaller than that in the U.S. This might be expected to provide greater resolution and thus increase the power to find associations between SES and survival in Ontario as compared with the U.S. However, the dispersion of incomes in each corresponding quintile in Ontario and the U.S. was actually quite similar, so the results should be reasonably comparable. Fifth, in the main analysis, we only used the SES indicator to rank communities and to assign them to quintiles. We confined ourselves to comparing wealthier and poorer groups in each country without attempting to define wealth or poverty. Creating truly comparable SES groups would have required taking into consideration the cost of living, the burden of taxation, and the direct and indirect social benefits in a whole range of different communities in Canada and the U.S. It was beyond our ability to do this correctly, and, therefore, we confined ourselves to comparing survival between more and less privileged groups in the two societies. In effect, we have compared the effect of socioeconomic rank rather than SES on survival. This approach ignores the greater difference in income between the wealthiest and poorest quintiles in the U.S. as compared with Ontario. However, in a parallel analysis, we used the relative income in each quintile in the Cox model. Using this approach, the overall income-associated gradient in cancer survival in Canada was still considerably less steep than that of the U.S., although the difference between the two societies did diminish. Quite apart from the potential for artifact, great caution is required in interpreting the results of this type of population-based ecologic study. We have not tried to equate community income with the individual patient's income. It is assumed that the median household income in a neighborhood in some way reflects the living conditions and lifestyle of its residents, but it cannot be assumed that there is a causal relationship between the individual's SES and cancer survival without considering the potential for confounding by other factors that might independently affect cancer outcome. Race has often been invoked as a possible confounding factor in this type of study.1,7,9,11,14,16,18,27,46,47 Race-specific information was unavailable for the Ontario cohort, but we did a separate analysis of the U.S. cohort that was stratified by race. Race-specific income was linked from the U.S. census to each SEER case, and new income quintiles were created. As a group, African Americans had a much lower median income compared with that of other races ($23,190 v $35,895), and much lower cause-specific survival rates in many cancers. However, the income-associated gradient in the risk of cancer death among African Americans was very similar to that of other races. The SES-associated trend statistic, as reported in Table 3 above, was 1.023 in African Americans and 1.022 in other races for stomach cancer, 1.062 in African Americans and 1.046 in other races for lung cancer, and 1.035 in African Americans and 1.045 in other races for bladder cancer. The SES-associated trend statistic for all sites combined except prostate was 1.076 in African Americans and 1.055 in other races, indicating a slightly higher but generally comparable gradient for African Americans. We also acknowledge that survival, as it is defined in this study, may be increased either by delaying death or by making an earlier diagnosis. Earlier diagnosis may have a real effect on the outcome of cancer if earlier treatment leads to higher long-term cure rates. However, unless it is possible to control for stage at presentation, earlier diagnosis may also artificially increase survival by the phenomenon known as lead-time bias. It seems likely that the large difference in cause-specific survival in prostate cancer between Ontario and the U.S. reflects more intensive use of prostate-specific antigen screening in the U.S. population and the identification of more early cases. Consistent with this hypothesis, the incidence of the disease is much higher in the U.S. than in Canada, although mortality is quite similar.48 A similar phenomenon may partly explain the Ontario/U.S. difference in breast cancer survival, although there is less evidence to support this. Whereas differences in stage mix have the potential to bias survival results through lead-time bias, earlier stage at diagnosis is also an indicator of good medical care, so those observations cannot be entirely dismissed as artifact. Stage was not available in the Ontario registry, but "general summary stage" was available in SEER, and we did a separate analysis in which we controlled for stage to give us some estimate of how much of the SES-associated gradient in survival was attributable to differences in stage at presentation. In most sites, controlling for stage reduces the strength of the association between SES and survival. For lung cancer, the trend statistic (used in Table 3) was reduced from 1.051 to 1.039; for breast cancer, it was reduced from 1.119 to 1.080; and for all sites combined (except prostate), it was reduced from 1.078 to 1.052, indicating that approximately one third of the association between SES and survival was attributable to differences in stage mix between wealthier and poorer communities. However, general summary stage provides an incomplete picture of variations in the anatomic extent of cancer, and controlling for tumor-node-metastasis stage might have accounted for a larger part of the effect. Nonetheless, it seems likely that at least part of the association is due to factors that operate after the diagnosis, including differences in access to care and quality of care. In conclusion, our study confirms that SES-associated gradients in cancer survival are less steep in Ontario than they are in the U.S. The shallower SES-associated gradient in Ontario is partly due to better cancer outcomes in the poorest Ontario communities compared with the poorest American communities. This may reflect the success of the Canadian medical system in removing financial barriers to access to care. It may also reflect the relative success of Canadian government programs, such as unemployment insurance and welfare, in mitigating the impact of poverty. However, the shallower SES-associated gradients in cancer survival in Ontario are also partly a reflection of the poorer outcomes experienced by residents of middle- and upper-income communities in Ontario in comparison to their counterparts in the U.S. This observation would suggest that although the Canadian health system may provide equal access to equivalent standards of care for wealthy and poor patients, it does not necessarily assure optimal access to optimal care. There is evidence from other sources that access to cancer care in Canada is not as good as it might be. Waiting lists for investigations and treatment (which are essentially unknown in the U.S.) are widespread in Canada,49,50 and such delays may have an adverse effect on cancer outcomes.51,52 There is also evidence that the centralized networks of provincial cancer clinics, which provide all the radiotherapy and much of the chemotherapy in many Canadian provinces, do not make care uniformly accessible to Canada's dispersed population.53 Furthermore, the structures that support quality assurance in cancer programs in U.S. general hospitals may be less well developed in similar institutions in Canada.54 The cancer committees, multidisciplinary tumor boards, institutional cancer registries, and ongoing programs of audit, required for accreditation by the Commission on Cancer in the U.S., are often not available in Canada. There is no direct evidence that these structures and processes have an impact on the outcome of cancer, but that hypothesis may be worth investigating. There is only limited information about differences in actual patterns of cancer treatment between Canada and the U.S., but it is known that in some situations, U.S. doctors favor more aggressive approaches to treatment than their Canadian counterparts.55,56 However, the impact of these different medical philosophies on cancer outcomes has not yet been established. Variations in outcome—both within societies and between nations—are potentially of great importance, because they may hold the key to improving overall outcomes through identifying remediable problems in health care delivery. Further research is required to identify specific health system–related factors that may be responsible for the observed differences in outcome between wealthy and poor patients and between the U.S. and Ontario.
Supported by grants from Cancer Care Ontario and the National Cancer Institute of Canada (W.J.M.). P.A.G. is a Career Scientist of the Ministry of Health of Ontario.
1. Dayal HH, Polissar L, Dahlbert S: Race, socioeconomic status, and other prognostic factors for survival from prostate cancer. J Natl Cancer Inst 74:1001-1006, 1985 2. Bonett A, Roder D, Esterman A: Determinants of case survival for cancers of the lung, colon, breast, and cervix in South Australia. Med J Aust 141:705-709, 1984[Medline] 3. Schrijvers CTM, Mackenbach JP, Lutz J-M, et al: Deprivation, stage at diagnosis and cancer survival. Int J Cancer 63:738-743, 1995[Medline]
4.
Auvinen A, Karjalainen S, Pukkala E: Social class and cancer patient survival in Finland. Am J Epidemiol 142:1089-1102, 1995 5. Schrijvers CTM, Coebergh J-W, van der Heijden LH, et al: Socioeconomic variation in cancer survival in the Southeastern Netherlands, 1980-1989. Cancer 75:2946-2953, 1995[Medline] 6. Brenner H, Mielck A, Klein R, et al: The role of socioeconomic factors in the survival of patients with colorectal cancer in Saarland/Germany. J Clin Epidemiol 44:807-815, 1991[Medline] 7. Perkins P, Cooksley CD, Cox JD: Breast cancer: Is ethnicity an independent prognostic factor for survival? Cancer 78:1241-1247, 1996[Medline] 8. Stavraky KM, Skillings JR, Stitt LW, et al: The effect of socioeconomic status on the long-term outcome of cancer. J Clin Epidemiol 49:1155-1160, 1996[Medline] 9. Ansell D, Whitman W, Lipton R, et al: Race, income, and survival from breast cancer at two public hospitals. Cancer 72:2974-2978, 1993[Medline] 10. Auvinen A: Social class and colon cancer survival in Finland. Cancer 70:402-409, 1992[Medline] 11. Cella DF, Orav EJ, Kornblith AB, et al: Socioeconomic status and cancer survival. J Clin Oncol 9:1500-1509, 1991[Abstract] 12. Karjalainen S, Pukkala E: Social class as a prognostic factor in breast cancer survival. Cancer 66:819-826, 1990[Medline] 13. Vagero D, Persson G: Cancer survival and social class in Sweden. J Epidemiol Community Health 41:204-209, 1987[Abstract] 14. Dayal H, Polissar L, Yang CY, et al: Race, socioeconomic status, and other prognostic factors for survival from colo-rectal cancer. J Chronic Dis 40:857-864, 1987[Medline]
15.
Bassett MT, Krieger N: Social class and black-white differences in breast cancer survival. Am J Public Health 76:1400-1403, 1986
16.
Steinhorn SC, Myers MH, Hankey BF, et al: Factors associated with survival differences between black women and white women with cancer of the uterine corpus. Am J Epidemiol 124:85-93, 1986 17. Mackillop WJ, Zhang-Salomons J, Groome P, et al: Socioeconomic status, and cancer survival in Ontario. J Clin Oncol 15:1680-1689, 1997[Abstract] 18. Gorey KM, Holowaty EJ, Fehringer G, et al: An international comparison of cancer survival: Toronto, Ontario, and Detroit, Michigan metropolitan areas. Am J Public Health 82:1156-1163, 1997 19. Kogevinas M, Marmot MG, Fox AJ, et al: Socioeconomic differences in cancer survival. J Epidemiol Community Health 45:216-219, 1991[Abstract] 20. Powell IJ, Schwartz K, Hussain M: Removal of the financial barrier to health care: Does it impact on prostate cancer at presentation and survival? A comparative study between black and white men in a Veterans Affairs system. Urology 46:825-830, 1995[Medline] 21. Clark JY, Thompson IM: Military rank as a measure of socioeconomic status and survival from prostate cancer. South Med J 87:1141-1144, 1994[Medline] 22. Dansey RD, Hessel PA, Browde S, et al: Lack of a significant independent effect of race on survival in breast cancer. Cancer 61:1908-1912, 1988[Medline] 23. The World Bank: World Tables 1995. Baltimore, MD, The Johns Hopkins University Press, 1995 24. The Canada Health Act. Statutes of Canada, c.6, s.12, 1984 25. The Ontario Health Care Accessibility Act, Revised Statutes of Ontario, c.3 [as amended], 1990 26. Aukerman GF: Access to health care for the uninsured: The perspective of the American Academy of Family Physicians. JAMA 265:2856-2858, 1991[Medline] 27. Davis K: Expanding Medicare and employer plans to achieve universal health insurance. JAMA 265:2525-2528, 1991[Abstract] 28. Holahan J, Moon M, Welch WP, et al: An American approach to health system reform. JAMA 265:2537-2540, 1991[Abstract] 29. Thorpe KE, Siegel JE, Dailey T: Including the poor: The fiscal impacts of Medicaid expansion. JAMA 261:1003-1007, 1989[Abstract]
30.
Frey CM, McMillen MM, Cowan CD, et al: Representativeness of the Surveillance, Epidemiology, and End Results Program Data: Recent trends in cancer mortality rates. J Natl Cancer Inst 84:872-877, 1992 31. Statistics Canada: 1996 Census Handbook. Ottawa, Ontario, Canada, Industry Canada, publication 92-352-XPE, 1997 32. US Bureau of the Census: Census of Population and Housing, 1990: Summary Tape File 3 on CD-ROM Technical Documentation. Washington, DC, US Bureau of the Census, 1992 33. Gloeckler-Ries LA, Hankey BF, Edwards BK (eds): Cancer Statistics Review 1973-87. Bethesda, MD, National Cancer Institute, 1990 34. Zippin C, Lum D, Hankey BF: Completeness of hospital cancer case reporting from the SEER program of the National Cancer Institute. Cancer 76:2343-2350, 1995[Medline] 35. Puckett CD (ed): The Educational Annotation of ICD-9-CM. (ed 4). Reno, NV, Channel Publishing, 1995 36. Clarke EA, Marrett LD, Krieger N, et al: Cancer registration in Ontario: A computer approach in Jenson OM (ed):Cancer Registration: Principles and Methods246-257Lyon, France, International Agency for Research on Cancer, publication 95, 1991 37. Robles S, Marrett LD, Clarke EA, et al: An application of capture-recapture methods to the estimation of completeness of cancer registration. J Clin Epidemiol 41:495-501, 1988[Medline] 38. Statistics Canada/Canadian Center for Health Information: Life Tables, Canada and the Provinces, 1985-1987. Ottawa, Ontario, Canada, Minister of Supply and Services, publication 82-003S, 1991 39. Statistics Canada/Health Statistics Division: Life Tables, Canada and the Provinces, 1990-1992. Ottawa, Ontario, Canada, Minister of Industry, publication 84-537, 1995 40. US Bureau of the Census: Statistical Abstract of the United States: 1986 (ed 106). Washington, DC, US Bureau of the Census, 1986 41. US Bureau of the Census: Statistical Abstract of the United States: 1990 (ed 110). Washington, DC, US Bureau of the Census, 1990 42. Ederer F, Axtell LM, Cutler SJ: The relative survival rate: A statistical methodology. Natl Cancer Inst Monogr 6:101-126, 1961 43. Kalbfleisch JD, Prentice RL: The statistical analysis of failure time data. New York, NY, John Wiley and Sons, 1980 44. Seiffert JE (ed): Standards for Completeness, Quality, Analysis, and Management of Data (vol 3). Sacramento, CA, American Association of Central Cancer Registries, 1994 45. Lemieux K, Zhang-Salomons J, Holowaty EJ, et al: The association between census-based measures of socioeconomic status and cancer survival in Ontario. Clin Invest Med 21:34, 1998 (abstr) (suppl) 46. Dayal HH, Power RN, Chiu C: Race and socioeconomic status in survival from breast cancer. J Chronic Dis 35:675-683, 1982[Medline] 47. Wegner EL, Kolonel LN, Nomura AMY, et al: Racial and socioeconomic status differences in survival of colorectal cancer patients in Hawaii. Cancer 49:2208-2216, 1982[Medline] 48. National Cancer Institute of Canada: Canadian Cancer Statistics 1996, Toronto, Ontario, Canada, National Cancer Institute of Canada, 1996 49. Mackillop WJ, Fu H, Quirt CF, et al: Waiting for radiotherapy in Ontario. Int J Radiat Oncol Biol Phys 30:221-228, 1994[Medline] 50. Mackillop WJ, Zhou Y, Quirt CF: A comparison of delays in the treatment of cancer with radiation in Canada and the United States. Int J Radiat Oncol Biol Phys 32:531-539, 1995[Medline] 51. Mackillop WJ, Bates HT, O'Sullivan B, et al: The effect of delay in treatment on local control by radiotherapy. Int J Radiat Oncol Biol Phys 34:243-250, 1996[Medline] 52. O'Sullivan B, Mackillop WJ, Grice B, et al: The impact of waiting for definitive radiotherapy in carcinoma of the tonsillar region. The Royal College of Physicians and Surgeons of Canada Annual Meeting (CARO), Toronto, Ontario, Canada, September 24-27, 1998 (abstr)
53.
Mackillop WJ, Groome PA, Zhou Y, et al: Does a centralized radiotherapy system provide adequate access to care? J Clin Oncol 15:1261-1271, 1997 54. Mackillop WJ, Catton P, Ashbury F, et al: Issues in the implementation of cancer staging in Canada. Cancer Prev Control 2:299-303, 1998[Medline] 55. Palmer MJ, O'Sullivan B, Steels R, et al: Controversies in the management of non-small cell lung cancer: The results of an expert surrogate study. Radiother Oncol 19:17-28, 1990[Medline] 56. O'Sullivan B, Mackillop WJ, Gilbert R, et al: Controversies in the management of laryngeal cancer: Results of an international survey of patterns of care. Radiother Oncol 31:23-32, 1994[Medline] Submitted November 30, 1998; accepted March 18, 1999. This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
- -) and the SEER regions of the U.S. (–•–).
