Health Reports
The cancer survival index: Measuring progress in cancer survival to help evaluate cancer control efforts in Canada

by Larry F. Ellison

DOI: https://www.doi.org/10.25318/82-003-x202100900002-eng

Abstract

Background

A comprehensive evaluation of progress in cancer survival for all cancer types combined has not previously been conducted for Canada. The cancer survival index (CSI) is superior to age standardization in measuring such progress.

Data and methods

Data are from the population-based Canadian Cancer Registry, record-linked to the Canadian Vital Statistics Death database. CSI estimates for both sexes combined were calculated as the weighted sum of the sex- and cancer-specific age-standardized net survival estimates. Sex-specific CSI estimates were calculated separately using sex-specific cancer type weights.

Results

From the 1992-to-1994 period to the 2015-to-2017 period, the five-year CSI increased 8.6 percentage points to 63.7%. It increased by 8.9 percentage points to 61.8% among males, and by 8.2 percentage points to 65.8% among females. The contribution of a cancer and sex combination to change in the CSI over time is a function of its assigned weight and changes in its age-standardized net survival. Female breast was the most influential cancer and sex combination, contributing 10.1% to the overall increase, followed by prostate (8.2%) and female lung (7.3%). The increase in the index since the 2005-to-2007 period was most impacted by lung cancer among both females (11.1%) and males (9.4%). While prostate cancer survival increased over the entire study period, it has recently decreased, resulting in a counterproductive 8.1% contribution since the 2005-to-2007 period.

Interpretation

Steady progress has been made in overall cancer survival in Canada since the early 1990s. Female breast cancer has contributed the most to this progress overall, but more recently female lung cancer has been the most influential.

Keywords

all cancers, cancer survival index, malignant neoplasms, population surveillance, registries, survival analysis

Author

Larry F. Ellison (larry.ellison@canada.ca) is with the Centre for Population Health Data at Statistics Canada, Ottawa, Ontario.

End of text box

Introduction

Population-based cancer survival estimates permit the monitoring of progress in cancer outcomes over time. When combined with cancer incidence and mortality time trends, they provide an indication of progress in overall cancer control.^{Note 1} Estimates of survival for all cancers combined may provide summary measures with which to compare survival across geographic areas or within a geographic area over time. These estimates have routinely been age-standardized to control for differences in the age distribution of cancer cases within the populations being compared.^{Note 2Note 3} However, cancer survival has also been shown to vary widely by cancer type and sometimes by sex within a particular cancer type.^{Note 4Note 5Note 6Note 7Note 8} The distribution of incident cancer cases by cancer type (case-mix) and sex can also vary over time within a population or between populations during a given time period. A relatively new concept, the cancer survival index (CSI), is a summary measure of cancer survival that accounts for all three of these potentially confounding factors (i.e., age, sex and cancer type).^{Note 9Note 10Note 11Note 12} In doing so, the CSI ensures more valid comparisons of cancer survival across populations for all cancers combined.

 The current study is the first comprehensive evaluation of progress in cancer survival for all cancer types combined in Canada. The results span the complete time period of the Canadian Cancer Registry^{Note 13} (CCR) and are unaffected by changes in the age, sex and case-mix of cancers over this time. Specifically, predicted Canadian net CSI estimates for the three-year period from 2015 to 2017 are presented and compared with corresponding actual estimates dating as far back as the 1992-to-1994 period. Comparisons are made for both sexes combined and for males and females separately. Further insight is provided by the determination of the most influential cancer and sex combinations and the leading cancer types within each sex, in regard to changes in the CSI since the periods of 1992 to 1994 and 2005 to 2007.

Data and methods

Data sources and definitions

Canadian Cancer Registry death-linked analytic file

The data source was a pre-existing analytic file created by linking CCR cases diagnosed from 1992 to 2017 to mortality information complete through December 31, 2017, via Statistics Canada’s Social Data Linkage Environment.^{Note 14} The CCR is a population-based database comprised of cases diagnosed among Canadian residents since 1992.^{Note 13} The mortality information was obtained from the CCR itself, the Canadian Vital Statistics Death (CVSD) database—whose current scope is all deaths in Canada^{Note 15}—and T1 personal master files (as reported on tax returns). The use of death information on tax returns facilitated the identification of additional death events of patients on the CCR that may not have been included in the CVSD, such as out-of-country deaths. It was also used to validate the date of death when discrepancies between dates on the CCR and the CVSD were encountered.

The analytic file followed the multiple primary coding rules of the International Agency for Research on Cancer (IARC).^{Note 16} Survival time was measured in days. Cases were defined based on the International Classification of Diseases for Oncology, Third Edition,^{Note 17} and classified using Surveillance, Epidemiology, and End Results (SEER) Program grouping definitions.^{Note 6} Non-melanoma skin cancer cases are not in the scope of the CCR.^{Note 13}

Expected survival

Expected survival probabilities, necessary for the calculation of net survival (NS), were mostly obtained from sex-specific complete annual provincial life tables.^{Note 18} As complete life tables were not available for Prince Edward Island or the territories, expected survival for these jurisdictions were derived, up to age 99, from abridged life tables for Canada and the affected jurisdictions^{Note 19} and from complete Canadian life tables^{Note 18} using a method suggested by Dickman et al.^{Note 20} For ages 100 to 109, where this wasn’t possible for these jurisdictions, complete Canadian life table values were directly used. Expected survival probabilities used in the calculation of NS for prostate and female breast cancer were adjusted for cancer-specific mortality rates in the general population.^{Note 21Note 22Note 23} More detail on the general approach for this adjustment can be found elsewhere.^{Note 5}

Inclusion and exclusions

All new primary cancers diagnosed in individuals aged 15 to 99 years were initially included. For cancers of the bones and joints, the age range was 20 to 99. Cases from the province of Quebec were excluded because cancer incidence data from this province had not been submitted to the CCR since the 2010 data year. Next, cases for which the diagnosis had been established through autopsy only or death certificate only, or for which a death had been established but the year of death was unknown, were excluded (1.5%). The dataset was then further restricted to first primary cancers per person, per individual cancer,^{Note 24Note 25Note 26Note 27} resulting in a further exclusion of 0.2% of cases.

Statistical analysis

NS is used as the underlying measure of cancer survival. NS estimates were derived using an algorithm^{Note 30} that has been augmented by Ron Dewar of the Nova Scotia Cancer Care Program (Dewar R, 2020, email communication, June 22) to include the Pohar Perme estimator of NS^{Note 31} using the hazard transformation approach. NS is the preferred method for comparing cancer survival in population-based cancer studies because it adjusts for the fact that different populations may have different levels of background risk of death.^{Note 28} The focus of this report is on five-year survival, though some short-term results in the form of one-year survival are also presented

Cancer survival index estimates

CSI estimates for the three-year period from 2015 to 2017 are compared with corresponding estimates dating as far back as the 1992-to-1994 period. To highlight recent changes, comparisons for the most recent 10-year period since the 2005-to-2007 period are also provided. CSI estimates for both sexes combined were calculated as the weighted sum of the unrounded sex- and cancer-specific age-standardized NS estimates.^{Note 11} Sex-specific CSI estimates were calculated separately as the weighted sum of the unrounded cancer-specific age-standardized NS estimates for each sex. Weights were derived from the proportionate distribution of incident cases diagnosed from 2010 to 2014 using the IARC version of the CCR tabulation file (excluding cases diagnosed in Quebec) released on January 29, 2018. This version of the CCR file has previously been used to construct both the Canadian cancer survival standard weights^{Note 5} and the weights used to derive a Canadian childhood cancer survival index.^{Note 29} Only cases diagnosed between the ages of 15 and 99 years (20 to 99 years for bone cancer) with malignant behaviour or in situ neoplasms of the urinary bladder were considered in the calculations. Non-melanoma skin cancer cases were excluded. The weights are a minor refinement on those used in the 2019 Canadian Cancer Statistics publication.^{Note 12}

Age-standardized NS estimates for individual cancers were calculated as the weighted sum of the age-specific estimates for a given cancer. The Canadian cancer survival standard weights were used for this process.^{Note 5} Fifty-five cancers were considered—the cancers traditionally reported on for cancer incidence, survival and prevalence by Statistics Canada, except the categories corresponding to the corpus uteri and uterus not otherwise specified were combined. Weights for this combined cancer were derived in the same manner as for the standard weights for individual cancers and are as follows: 15 to 44 (0.050), 45 to 54 (0.169), 55 to 64 (0.353), 65 to 74 (0.261) and 75 to 99 (0.167). A five-year age-specific NS estimate could not be derived for 15 of the 11,220 (0.13%) necessary cancer, sex, age group and overlapping three-year time period combinations. In these extremely rare instances, the corresponding five-year NS estimate for the opposite sex was used as a proxy.

Predicted survival

The period method was used to determine predicted NS estimates for 2015 to 2017. The cohort method was used to derive non-predictive (actual) estimates of five-year NS between the periods of 1992 to 1994 and 2010 to 2012, and one-year NS between the periods of 1992 to 1994 and 2014 to 2016. Empirical evaluations of period analysis have shown that this method provides estimates that closely predict the survival that is eventually observed for people diagnosed in the period of interest, particularly when survival is fairly constant.^{Note 32Note 33Note 34} When survival is generally increasing (or decreasing), a period estimate tends to be a conservative prediction of the survival that is eventually observed.^{Note 33Note 35} All survival estimates were expressed as percentages. Changes in CSI estimates over time were expressed as percentage point differences.

Measuring component contributions to changes in net cancer survival index over time

The component contribution of a sex- and cancer-specific combination to the overall change in the CSI for both sexes combined over time was calculated as the product of the change in its sex- and cancer-specific age-standardized NS estimate and its corresponding CSI weight. The component contribution to the overall sex-specific changes in the CSI for a particular cancer results similarly from the multiplication of the change in its sex-specific age-standardized NS estimate by its corresponding sex-specific CSI weight.

The component percentage contribution of each sex- and cancer-specific combination to the overall change in the CSI for both sexes combined over time was then determined as the absolute value of the quotient of the component sex- and cancer-specific combination contribution to the change divided by the total change. The component cancer-specific percentage contribution to overall sex-specific changes in the CSI over time was similarly determined using sex-specific values. Component percentage contributions must be non-negative and sum to 100%; however, if survival has decreased in the time period under consideration for a given component, its percentage contribution to the change in the CSI will be counterproductive to the goal of improved overall survival.

To aid in the interpretation of component contributions to changes in the CSI over time, sex-specific changes in the five-year age-standardized NS estimates from the 1992-to-1994 period to the 2015-to-2017 period for the more common and specific individual cancers are provided in Appendix Table A1.

Results

Weights

For both sexes combined, the cancer type and sex combination assigned the highest weight was female breast cancer, at 0.12772 (Table 1). Close behind was prostate cancer, at 0.12348, followed by male and female lung cancer, at 0.06658 and 0.06265, respectively; male and female colon cancer, at 0.04088 and 0.03899, respectively; and male bladder cancer (0.03639). These seven cancer type and sex combinations comprised virtually half of the total assigned weight, or very nearly as much as the remaining 95 combinations put together. Sex-specific weights were highest for prostate (0.24008), lung (0.12944) and colon (0.07949) cancer among males and breast (0.26298), lung (0.12900) and colon (0.08027) cancer among females.

Changes in one- and five-year net cancer survival index over time

From the 1992-to-1994 period to the 2015-to-2017 period, the five-year CSI increased by 8.6 percentage points to 63.7% (Figure 1). Over the entire time span, the one-year CSI increased by 6.4 percentage points to 79.1%. Over the most recent 10-year period between the 2005-to-2007 period and the 2015-to-2017 period, the one-year CSI increased by 3.1 percentage points, exceeding the 2.7 percentage point increase recorded for the corresponding five-year index over the same time span.

The CSI for all cancers combined was higher among females than among males for the entire study period, though the difference declined slightly over time. Among males, the five-year CSI increased by 8.9 percentage points to 61.8% and the one-year index increased by 6.8 percentage points to 77.6%. Among females, the five-year CSI increased by 8.2 percentage points to 65.8% and the one-year index increased by 5.9 percentage points to 80.6%. From the 2005-to-2007 period to the 2015-to-2017 period, the increases were by 3.4 (one-year) and 2.8 (five-year) percentage points among males, and 2.8 (one-year) and 2.5 (five-year) percentage points among females.

Contributions to change in net cancer survival index over time

Female breast cancer most influential to the increase in the five-year net cancer survival index since the 1992-to-1994 period

Of the 102 cancer type and sex combinations that contributed to the 8.6 percentage point increase in the five-year CSI since the 1992-to-1994 period, female breast cancer was found to be the most influential (Table 2). Because of a 0.12772 weight and a 7.1 percentage point increase in five-year age-standardized NS (Appendix Table A1), female breast cancer contributed 10.1% to the overall increase. The next leading combinations were prostate cancer (8.2%), female lung and bronchus (lung) cancer (7.3%), male non-Hodgkin lymphoma (6.2%), and male lung cancer (5.1%).

Lung cancer most influential to the increase in five-year net cancer survival index since the 2005-to-2007 period

The increase in the five-year CSI since the 2005-to-2007 period was most impacted by lung cancer among both females (11.1%) and males (9.4%). Among females, five-year lung cancer survival increased by 7.0 percentage points from the 2005-to-2007 period to the 2015-to-2017 period, while among males it increased by 5.6 percentage points. The third most influential contributor was prostate cancer, at 8.1%. However, unlike its positive contribution over the entire study period, the contribution of prostate cancer over the more recent time-span was counterproductive to the goal of improved overall survival, owing to a 2.6 percentage point decrease in its five-year age-standardized NS. The next leading cancer type and sex combinations were miscellaneous cancers (male 6.6%, female 5.9%) and bladder cancer among males (6.1%).

Female lung cancer also contributed the most to increases in the one-year CSI, whether over the entire study period (13.2%) or since the 2005-to-2007 period (female 15.6%). Male lung cancer was second, at 8.9% (13.7% since the 2005-to-2007 period). Male miscellaneous cancers were next, at 7.6% (10.5% since the 2005-to-2007 period), followed by female miscellaneous cancers at 7.2% (8.7% since the 2005-to-2007 period). The contributions of female breast cancer (10^th, at 2.9%) and prostate cancer (15^th, at 1.9%) were not nearly as impactful to increases in the one-year CSI over the study period as they were for the five-year duration.

Contributions to change in sex-specific net cancer survival index over time

Prostate cancer, at 15.5%, has had the largest impact on the 8.9 percentage point increase in the five-year CSI among males since the 1992-to-1994 period (Table 3). The next most influential cancers among males were non-Hodgkin lymphoma (11.7%), followed by lung (9.6%) and colon (9.1%) cancer. When the analysis was restricted to the most recent time span since the 2005-to-2007 period, lung cancer was first, at 15.7%. The second most impactful, albeit in a counterproductive manner caused by a decrease in survival, was prostate cancer, at 13.5%. Miscellaneous cancers (11.0%) and bladder cancer (10.2%) were the only others to exceed 7%.

At 21.5%, breast cancer has had the largest impact on the 8.2 percentage point increase in the five-year CSI among females since the 1992-to-1994 period. Lung cancer was second among females, at 15.6%, followed by miscellaneous cancers (9.2%), colon cancer (9.2%) and non-Hodgkin lymphoma (8.7%). From the 2005-to-2007 period to the 2015-to-2017 period, the leading contributing cancer types to the improvement in the CSI among females were lung cancer (27.5%), miscellaneous cancers (14.7%) and breast cancer (9.7%).

Discussion

Relatively steady progress has been made in five-year net cancer survival for all cancers combined in Canada since the early 1990s. The five-year CSI increased 8.6 percentage points from the 1992-to-1994 period to the 2015-to-2017 period. A slightly higher increase was observed among males than among females. The most influential cancer and sex combination in regard to the overall change was female breast cancer. However, since the 2005-to-2007 period, the change in the CSI has been most impacted by lung cancer among both females and males.

Particular features of the CSI have been described in detail elsewhere.^{Note 9Note 10Note 11} In short, the index summarizes the overall cancer survival in a population in one number and can be used to monitor progress over time. In conjunction with other metrics, the index may assist public health professionals and policy makers in evaluating the effectiveness of cancer control plans. Changes in the CSI over time better reflect the changes in survival because of diagnosis and treatment factors. This is because such changes are unaffected by temporal changes in the age, sex and cancer type distribution of cancer cases. Because of the incorporation of NS, they are also unaffected by any changes in the underlying risk of death in the population from other causes.

The contribution of a cancer and sex combination to a change in the CSI over time is a function of the fixed weight assigned to the component and changes in the age-standardized NS of the component over the given time span. Changes in survival among cancer and sex combinations that occur more frequently in the population under consideration are emphasized to a greater extent than changes in those that are rarer. In Canada, this means that breast cancer among females and prostate cancer are given the most emphasis. Changes in the CSI also reflect changes in survival among many individual cancer and sex combination components. In general, the largest increases in age-standardized NS were observed among blood cancers like chronic myeloid leukemia.^{Note 4}

The largest contributors to the increase in the five-year CSI were female breast cancer and prostate cancer. Gains in breast cancer survival have been attributed to the success of population-based screening programs, which have resulted in the earlier diagnosis of cases.^{Note 12} However, while gains in age-standardized NS were achieved for both of these cancers over the study period, their contributions to the CSI were more attributable to their assigned weights. To a lesser extent, the same was true for lung cancer among males, which was the fifth leading contributor to the index. By contrast, the contribution of non-Hodgkin lymphoma among males was more attributable to a relatively large increase in age-standardized NS, as opposed to the effect of its weight. The latter, while 11^th among more than 100 cancer and sex combinations, was about one-fifth that of female breast and of prostate cancer. The contribution of lung cancer among females to the CSI exceeded its assigned weight as well. It also surpassed the contribution of lung cancer among males as a direct result of faster-paced increases in survival among females. With the exception of non-Hodgkin lymphoma, the effect that large increases in survival among blood cancers have had on the index have been attenuated by their corresponding relatively small weights.

Since the 2005-to-2007 period, the leading contributors to the increase in the five-year CSI have been female and then male lung cancer. While the weights assigned to these cancer type and sex combinations were the third and fourth highest of all such combinations, their proportionate contributions to the index were still considerably higher than the weights. This indicates that substantive improvements in survival were instrumental to their contributions. Advances in treatment, including the increasing use of targeted therapy drugs, may be behind at least some of the more recent survival gains for lung cancer.^{Note 36} Given the high frequency of lung cancer among all diagnoses of cancer, and its relatively inferior current prognosis, continued improvements in NS for this disease appear to offer the best opportunity for future improvements in the CSI. One potential pathway for this may be through the introduction of recommended lung cancer screening programs.^{Note 37}

The third leading contributor to the CSI from the 2005-to-2007 period to the 2015-to-2017 period was prostate cancer. However, unlike its impact over the entire study period, the contribution of prostate cancer since the 2005-to-2007 period has been counterproductive because of a recent decrease in survival. Efforts to reduce the overdiagnosis (i.e., the detection of disease that would not progress to cause symptoms or death^{Note 38}) of this cancer likely played an indirect role in the survival reduction. Following changes to prostate cancer screening guidelines,^{Note 39Note 40Note 41} age-standardized prostate cancer incidence rates in Canada (excluding Quebec) decreased by 25% from 2011 to 2017, and have declined by 33% since 2007.^{Note 42} The decrease disproportionately affected cases that were less likely to result in death. Specifically, the proportion of stage I and II prostate cancer cases decreased by 5 percentage points from 2011 to 2017.^{Note 43}

The five-year CSI among females was 4 to 5 percentage points higher than among males throughout the study period. By comparison, Quaresma et al. reported an advantage among females twice as large among those diagnosed with cancer in England and Wales.^{Note 10} The Canadian difference can be predominantly attributed to survival being a little higher among females than males for several individual cancers. While this was also a factor in England and Wales, the majority of the difference there was attributed to the fact that the cancers that are most common in women (e.g., breast) generally have higher survival than the cancers that are most common in men (e.g., lung). In Canada, however, the most common cancers among females generally have similar weights and survival to the most common cancers among males.

Strengths and limitations

The current study has a number of strengths, including the use of incidence data from the CCR, one of the highest-quality national population-based cancer registries in the world.^{Note 44} Another major strength is the additional consideration of cancer type and sex as potential confounders in measuring changes in cancer survival over time for all cancers combined. Doing so ensured more valid comparisons of cancer survival, as did the use of such a large number of individual cancers in the derivation of the CSI. Additionally, the use of recent Canadian-specific weights in the adjustment process in this study enhanced the interpretability of CSI estimates in the Canadian context.

The CSI has several limitations. As a summary measure, it does not provide any information regarding survival for individual cancers. It also does not reflect any given individual’s prognosis and is therefore not useful in a clinical sense. In addition, it does not provide insight into the reasons behind changes in survival estimates over time. Changes in the CSI over time need to be interpreted in the context of public health interventions undertaken during the time period considered. A limitation of this study was the absence of complete incidence data from the province of Quebec. The CCR has not received data from this jurisdiction since the 2010 data year, and data that were submitted up to 2010 were based primarily on hospital data (i.e., hospitalizations or day surgeries). Thus, the results herein pertain to Canada, excluding Quebec.

Conclusion

This study represents the first comprehensive evaluation of progress in cancer survival for all cancer types combined in Canada. As measured by the CSI, steady improvement in overall NS has been made in Canada since at least the early 1990s. Continued monitoring of this index is recommended as a means to help measure ongoing progress in the diagnosis and management of cancer in Canada, including the effects of the COVID-19 pandemic. Future work will involve deriving provincial-level CSI estimates for the purposes of examining both interprovincial differences in the CSI and province-specific trends in the CSI over time.

Appendix