![]() |
![]() ![]() |
||||||||||||||||
![]() ![]() ![]() |
|
||||||||||||||||
|
Workshop Report National Initiative to Improve Cancer Survival Information
Workshop Overview This workshop (held June 19-20, 1997, in Ottawa, Ontario) brought together major stakeholders to establish a Canadian initiative to develop systematic collection of population-based information on cancer survival. The broad goals of the workshop were to delineate the current needs regarding such data and to recommend practical mechanisms to systematically generate this information. Three key areas for discussion were identified: collection of staging data, collection of treatment data and standardized approaches to analysis. Presentations (summarized below), designed to establish common ground for workshop discussions, provided background information on population-based collection mechanisms, data and methodology requirements for conducting meaningful cancer survival surveillance, and practical examples of using cancer survival information. History of National Initiative In December 1995, the newly established Cancer Bureau Steering Committee of the Laboratory Centre for Disease Control (LCDC) formally recognized that existing population-based cancer survival information was sparse, thereby limiting effective monitoring of progress against cancer. The Committee recommended developing an initiative to provide such information and formed a planning group representing key stakeholder organizations. This planning group identified three priority areas for discussion (listed above) and proposed a stakeholder workshop to start building a cancer survival information initiative. Concurrently, the Canadian Coalition on Cancer Surveillance (CCOCS) had evolved from recommendations made at a workshop of the National Cancer Institute of Canada (NCIC), "Surveillance Systems for Cancer Control in Canada" (November 1996; Kananaskis, Alberta). The mandate of this multi-stakeholder coalition (for which LCDC functions as secretariat) is to develop a comprehensive cancer surveillance system. One of the Coalition's five subgroups, the Patient Management Working Group, has been developing a patient management component of the surveillance system. Thus, the present workshop also provided an opportunity to advance the efforts of the Working Group by developing a systematic means to capture stage and treatment data on a population basis. Workshop Terms of Reference Workshop participants agreed on these specific goals.
Presentation Summaries Survival of Cancer Patients:
Basic Issues Arduino Verdecchia The EUROCARE project, the first pan-European study on survival in cancer patients, provided comparisons among 11 European countries by analyzing data on 80,000 cancer patients from 30 population-based cancer registries in a uniform framework. Major confounding factors (i.e. disease definition, date of diagnosis, proportion of death certificates only [DCO], follow-up procedures) were controlled by collecting data according to a common protocol. Statistical confounding factors (method of analysis, competing mortality, selection of cases with wrong or inconsistent data) were controlled by using appropriate methodology for estimating relative survival figures as well as criteria for analysis. Differences among countries, and sex and age groups were considered for interpretation as real effects (i.e. possibly due to effective treatment, early diagnosis, management and follow-up of patients, etc.). As health care systems may vary from country to country in type, effectiveness and available resources, cancer patient survival can be regarded as an indicator of the health care system's effectiveness, and differences in cancer patient survival, as indicative of situations that could be improved. Attempts to interpret levels and trends have been made both statistically and by ad hoc data collection within samples of patients in some registries. High-resolution studies on a sampling basis aim at characterizing differences in diagnostic and therapeutic standards or time-related improvements. Specific multivariate analyses have been performed to study differences expressed as relative risks, adjusted for age and period of diagnosis, and to study interactions between the main factors. Mixture models were proposed to distinguish between potentially cured patients and the proportion of patients bound to die from the disease. Based on this approach, the interpretation of differences is greatly improved. However, the interpretation of differences still remains the major goal for present and future work of the EUROCARE study. Comparability of cancer statistics, including patient survival data, can be reached by adopting common protocols and a unified framework of analysis, as in the EUROCARE study. Comparing cancer statistics between European countries and the US is the goal of specific projects now in place. Worldwide comparability may be possible in the foreseeable future. US Experience: Population-based Cancer Survival Monitoring Lynn Gloeckler Ries SEER (Surveillance, Epidemiology, and End Results) data cover about 9.5% of the US population, using both active and passive follow-up. Patient follow-up is generally the poorest for female patients under 20 years old, for whom marriage and high mobility impact on follow-up success. Data collected by SEER include patient demographics (age, sex, race, geographic location), primary tumour site and morphology, diagnostic methods, extent of disease (EOD) and first course of treatment. Stage captured by EOD has been collected since the program's inception. The detail and amount of information vary somewhat over time; the data set currently includes size, extension of tumour and lymph nodes (involvement, number positive, number examined). The EOD scheme (localized, regional, distant) can be used for long-term trends of historical staging and can generally be converted to the TNM (tumour, node, metastasis) classification scheme of the American Joint Committee on Cancer (AJCC). Often, extent-of-disease data yield a better evaluation of prognostic factors than stage alone. The interrelationship among survival, EOD (stage), treatment and demographic variables is complicated. Impacts of using both stage and treatment data to interpret survival must be considered carefully. Population-based treatment can be used to examine diffusion of therapy but should not be used to evaluate the efficacy of treatment, which is best left in the domain of clinical trials. In addition, SEER has implemented special studies for certain sites to collect information on treatment given in outpatient settings since routine methods tend to underreport these treatments. Several published reports about survival have examined the effects of extent of disease and other factors on survival. Recent reports examine lung, ovarian and breast cancer survival. A study on breast cancer explored conditional survival, that is, survival subject to predetermined periods of post-diagnosis survival. The study shows that survival differences vary substantially according to post-diagnosis period and suggests this approach may be well suited to providing more precise prognostic data. SEER data are available free of charge on CD-ROM and through Internet access. Both incidence data and a survival component are available, as is stage-specific survival according to AJCC staging standards. In addition, site-specific surgery (from 1983 onward) and radiation data are available. Cancer Survival in Ontario: Population-based Survival Analysis Margaret Sloan This presentation focused on analyses published in Cancer Survival in Ontario,1 a monograph publication of the OCTRF based on data from the Ontario Cancer Registry (OCR). Included were incident cancers diagnosed between January 1969 and December 1988, with outcome follow-up complete to December 1989. Relative survival rates by sex, age group and time period were tabulated for each of 25 site groups, as well as for all sites combined. Ms Sloan highlighted some interesting results and discussed issues regarding data quality and completeness, some of which derive from the passive cancer registration procedures used by the OCR, and some of which are inherent in the original data sources. The most obvious limitation is the absence of staging data for a large portion of the tumours, in fact, for virtually all of the earlier cases. Limitations related to the use of relative survival as the statistic of choice were discussed. These arise for the most part from the definition and calculation of expected survival. Because only population-based life tables are readily available, it is not possible to calculate relative survival for subgroups based on, for example, socio-economic status, ethnic or cultural background, or distribution of risk factors such as smoking. Cancer Data Collection Systems in Canada: National Data Collection Eric Holowaty The current registry system in Canada provides a comprehensive means of reporting cancer incidence and mortality, envied by other nations for its national scope. Despite this accomplishment, the system has limited capacity for cancer control surveillance. The vision of the CCOCS to develop a surveillance system that extends current databases to include other relevant data elements such as staging is appropriate. To become fully effective, the system will require concurrent development of standards and procedures to ensure compliance with standards for data collection. The development of such policies and procedures to ensure uniformity in data quality across registries is of paramount importance, particularly at a time when capture of new data elements is being discussed. In Ontario, some of the current barriers to data quality include lag time between diagnosis and registration, lack of microscopic confirmation (for 15% of cases), loss to follow-up and missing data elements. Unexplained variations in regional incidence-to-mortality ratios indicate problems related to completeness of registration. In general, about 85% of the problems appear in 20% of the cases; thus, follow-back may resolve this to some extent. Reallocation of funds to target problem areas appears wise. Currently, collection costs for data are approximately $25 per case. Andy Coldman The BC Cancer Agency maintains the provincial cancer registry and an information database on all patients referred to the Agency for treatment or follow-up (about 60% of total). The registry includes basic demographic information (name, sex, date of birth, date of death, cause of death) and information on each cancer diagnosis (ICDO code, date of disease, address of diagnosis). The Agency's patient database contains further information on stage of disease, detailed radiotherapy treatment information and follow-up data. The primary source of cancer registration is the pathology report, and 75% of subjects are registered within three months of diagnosis. Within two years of diagnosis, 62% have been referred, 21% are deceased and 26% are alive and non-referred. Staging is available for 48% of all cancer diagnoses and in 91% of those referred to the BC Cancer Agency. National Staging Initiative Report Brian O'Sullivan Interest in the systematic application of cancer staging and collection of such data have existed for decades. Recently, however, efforts to develop a comprehensive strategy for collection and capture of such data have received support through a series of workshops, opinion surveys and committee deliberations. In particular, the Canadian Committee on Cancer Staging, a subcommittee of the Advisory Committee on Cancer Control of the NCIC, was given further support for the ongoing Consultation on Cancer Staging at the 1996 NCIC workshop on cancer surveillance systems. The Consultation, involving leaders in oncology in the areas of administration, clinical research and cancer prevention, has culminated in a draft formulation of (i) an outline of the principles of cancer staging and its use; (ii) an assessment of the value of cancer staging for patient care, cancer program management and clinical/epidemiologic research; and (iii) recommendations about the use of staging for developing standards of care, training to enhance application and processes for capture, compilation and quality assurance in data handling. Draft recommendations
Other practical suggestions, enhancements or alternatives to these recommendations to strengthen and facilitate the consultation process and products would be welcome. Lessons Learned in Implementing the Capture of Stage Information in a Regional Cancer Centre Bill Evans It has been difficult to capture information on tumour stage in most acute-care institutions and regional cancer centres. Three years ago, the Ottawa Regional Cancer Centre (ORCC) initiated systematic capture of stage information on all newly diagnosed cancer patients. Many lessons were learned. First, leadership from an institution's administration (including the CEO and manager of health information services) is essential. In the case of the ORCC, additional commitment came from the physician head of the Health Records Committee and discipline heads. This ensured that the process worked effectively. Previous attempts to capture stage information, which were not fully embraced locally, have failed. Institutions attempting to capture stage information will receive multiple excuses for non-compliance. Physicians usually indicate that they already stage patients in order to determine appropriate care, but this information is not captured in a consistent fashion to allow for recording in information systems. Physicians tend to resist requirements to complete more forms and may see the process as an administrative exercise. Some claim that collection has little value because staging systems continually evolve and may be replaced by non-anatomical prognostic indicators, such as biomarkers. Others argue that stage is only one of several important prognostic factors, and they may be reluctant to capture any if it is not possible to capture them all. The ORCC process moved quickly once the organization clearly established its intent to capture stage information by introducing a policy requiring all new patients to be staged using the TMN classification. The policy stipulated that medical staff comply as a condition of their employment under the Centre's Medical Staff By-laws. Physicians were engaged in providing input for the process of capturing stage and the design of the staging forms. Forms were developed using desktop publishing and modelled after the AJCC staging forms. The ORCC Medical Advisory Committee's recommendation that staging forms be completed within three months required a system to flag incomplete charts. Non-compliance was reported to the individual physician and, subsequently, to the discipline head. The initial six months were difficult and demanding for Health Information Services staff, who had to follow up on many incomplete records. Gradually, compliance improved and physicians began using stage information for research purposes and program planning. Eventually, stage progression had to be addressed, an issue particularly problematic for patients referred to the Centre long after their original diagnosis. From a cancer surveillance perspective, capturing the original tumour stage is important; however, physicians need current stage information for treatment decisions. Thus, the staging forms were modified to capture both, and the policy was modified accordingly. The ORCC experience reveals that staging principles should be defined clearly in the policy on staging. Not all physicians know how to stage; therefore, training may be needed. Finally, audit of the data is essential to determine data quality. The Centre uses stage information to do retrospective chart reviews, to estimate the number of patients available for new studies, to define the Centre's clients for accreditation purposes, to facilitate clinic scheduling decisions, to target academic detailing to the counties in the Centre's catchment area and to estimate new drug usage and cost. Staff are still learning how to optimally display and use the available stage information. Questions arise as to what data to display, who should receive data and how often data should be summarized. The numerous purposes for which the data have already been used more than justify the Centre's effort to implement stage capture. The availability of these data provides new opportunities to be creative in educational initiatives, research studies and Centre management. Barriers to Using Registry Data to Evaluate the Outcomes of Cancer Treatment Bill MacKillop Through the experience of collecting cancer treatment and outcome data in Ontario over the past four years, the following limitations have been identified: lack of information about the quality of collected data and insufficient data on patients' prognostic characteristics (e.g. stage at diagnosis, comorbidity, functional status) and even on the specific treatment outcomes of primary interest (e.g. definitive outcomes such as death or recurrence vs more time-dependent, subjective outcomes such as quality of life or patient satisfaction with treatment symptom control). Currently, treatment data are not collected at the Ontario Cancer Registry level. However, various details of treatment are routinely recorded at time of care, and thus linkage to Registry-based (outcome) data can occur. Depending on the treatment data in question, this linkage may involve either primary or secondary data capture. One problem in capturing treatment data is that they may be collected through disease-based rather than person-based mechanisms, thus precluding linkage with other patient data (e.g. prescription data for hormone therapy). The need for population-based collection of treatment data is underscored by the findings of two recent studies.2,3 These studies linked radiotherapy data (collected from Ontario's cancer centres) to individual patient records at the Registry level and ascertained waiting periods for treatment. Access to radiotherapy treatment varied substantively across Ontario jurisdictions, and treatment delivery rates were lower and involved longer waiting periods than in the US. These results affirm that the system delivery of radiotherapy in Ontario does not provide equitable access or treatment within medically acceptable time lines. Such studies illustrate the utility of collecting treatment data, in that these data can be applied to care management and can provide indicators to assess the cost and quality of delivery. Along with developing a systematic, comprehensive means of collecting treatment data, it is important to emphasize collecting a broader spectrum of outcome data, for instance, the effect on quality of life over time, considering both curative treatments and palliative treatments for alleviating discomfort. More information about both hospital and home care (i.e. support and continuing care) is also required for effective surveillance of treatment delivery in Canada. In summary, the capture and recording of treatment information seems to be adequate; the problem arises in transferring this information to registries and users. Until adequate linking systems are developed, the information is available but not accessible. Methodology: Cancer Survival Analysis Methods and Applications Timo Hakulinen Observed survival rates of cancer patients can be obtained by using the life table method or the Kaplan-Meier method. The latter utilizes individual exact observation times, whereas the former considers these times in groups, for instance, by year or month of follow-up. However, observed survival rates give an incomplete, pessimistic picture of cancer patients' survival because various non-cancer causes of death contribute to lower rates. Moreover, the rates are not comparable as such between young and old patient groups since older patients have an increased mortality due to non-cancer competing risks of death. One way to correct for competing risk mortality is to regard all the deaths due to competing risks as censoring events for the patients' observation times. The corrected or cause-specific survival rates derived in this way presuppose knowledge of individual patients' causes of death. This information can be unavailable or unreliable.4,5 The second way of correcting for competing risk mortality is to compare the patients' mortality with that of a general population group, considered practically disease-free, with respect to important demographic factors such as sex, age and calendar time.6 Then the excess mortality of the patients is used to generate relative survival rates that describe patient survival under cancer-associated excess mortality only. Specialized software packages7-9 can calculate the relative survival rates and take into account important prognostic factors related to the patient and the tumour. The methods used may be viewed as generalizations of the Cox proportional hazards model and are particularly suited for non-proportional hazards. Non-proportionality of the mortality and the patients' excess mortality is more the rule than the exception as the importance of prognostic factors determining the patients' survival changes over time. For example, the stage is typically a determinant of early survival and its importance is practically non-existent after a few years of follow-up after diagnosis. Routine publications on survival analysis in the US and many other countries, and the European EUROCARE collection have been produced using relative survival rates.10-12 A further interesting application has been the use of patients' residence and social class as a determinant for the specific general population group, thereby making it possible to study the equity of cancer patient survival related to place of residence and social class. For example, an analysis of 12 common sites in the Nordic countries showed that about 2-3% of all excess deaths of these patients were associated with residence (Nordic countries) or social class (only Finland studied).13
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Breakout Sessions and Discussion
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Capturing Stage Data Nationally The mandate of the Canadian Coalition on Cancer Surveillance (CCOCS) expresses the ultimate goal of a system that captures all relevant data required to monitor the efforts and impact of cancer control. The existing surveillance system, limited to incidence and mortality data, is insufficient for this task. Ideally, an expanded system would include information on stage or extent of disease at diagnosis; initial and follow-up treatment; patient characteristics (e.g. sociodemographic); access to treatment and detection programs, care and support; and quality of outpatient or palliative care. Such a system, flexible and evolving according to needs, would take extensive effort to develop. Some components of such an ideal system are already in the developmental stages. As described in the presentation summaries, the national initiative to collect staging data for surveillance purposes is well under way. The majority of the practical steps and issues for this data collection identified by the Workshop breakout groups are already being considered by the Canadian Committee on Cancer Staging (CCCS). These discussion points (outlined below) will benefit future efforts to collect other data that may be needed within the comprehensive cancer surveillance system envisioned by the CCOCS. Purpose of collecting stage data Staging data is essential for interpreting trends in cancer survival, specifically to discriminate between the effects of various determinants of cancer survival (e.g. treatment, screening programs, age of the population, period of diagnosis). These data complement clinical trial data, currently the only means to properly evaluate cancer patients' survival, but generalizable only to highly select populations. At the patient level, optimum care is feasible only with proper staging. Thus, patient stage data benefit both immediate (direct patient care) and long-term (subsequent patient care guided by population data) decision making. Leadership The CCOCS can provide national leadership to guide the system development process by identifying sources of financial support, quality assurance methods and standards, and guidelines for access to a national database, for data analysis and for use of outcome information. Methods and standards developed by the North American Association of Cancer Registries could be adapted to the Canadian setting. Access to the database should be universal, portable and comprehensive, but with safeguards to protect the confidentiality of data and integrity of results. Administrative support Discussion is under way to include stage data as part of the minimum data requirements for CCHSA accreditation standards for cancer treatment institutions. Senior administrators can help ensure quality reporting through policies or by-laws stating, for example, that pathology reports be released only with clinical staging or that no fees be recovered without completed forms. The Royal College of Physicians and Surgeons can educate members about guidelines and processes for staging. Endorsement by the cancer agencies will further support the reporting of staging information. Lines of responsibility for data collection Clear lines of responsibility and accountability will ensure effective, standardized collection of staging (and treatment) information. Guidelines should clarify who owns, assembles and manages the data and when data collection should take place, i.e. with the pathology report or upon discharge. A framework or a system to enhance current reporting would need the co-operation of clinicians and pathologists to support mechanisms ensuring standardized and timely data collection. Management tools and mechanisms Hospitals will need standardized forms, training, policies and procedures, to be developed centrally. In addition, standardized software will be required for management, editing, conversion, record reconciliation and validation. A management group including practitioners should be responsible for setting benchmarks for data collection and developing standards for determining who can access the summary data. Strategies to safeguard personal data will need to be developed to include checks and balances and legislative safeguards of privacy (e.g. release of aggregate data only). Electronic reporting would ideally be incorporated within the existing systems (e.g. provincial cancer registries, Statistics Canada, Canadian Institute for Health Information). The data may not need to reside in one location: a "virtual" system of data collections may be feasible. However, reporting from multiple sources will require methods for resolution of multiple TNM codes at some level. Implementation options for collecting stage (or other) data
Promotion of uses for data set Although much data collection already occurs, it is not always clear who the users are and how they can utilize the data. Clients, including patients, advocacy groups, care providers, researchers and policy planners, may not know what data are available or how they can be used. To maximize use, applications for the data elements now available and potential ones (i.e. stage) should be demonstrated and promoted to appropriate users. Consolidation and publication of current regional and provincial systems could provide concrete evidence of the utility of data collection to physicians, underlining the usefulness of an expanded national cancer surveillance system. On the other hand, the limitations of analyses without stage data in existing reports on cancer survival status should be highlighted to encourage reporting of stage data. CCOCS recommendations for additional surveillance system core data collection should be published. Provincial offices or regional cancer centres should be approached to form an "agency of chairs" or to recommend "champions" to identify potential users and to promote data reporting. Various strategies would increase interest in collecting the data and encourage compliance in reporting: oncology groups could encourage education and physician compliance; workshops for cancer centre teams and pathologists could promote the usefulness of value-added data to those involved in collecting and managing the information; a process could be promoted to provide useful and timely information back to practitioners and registries; and a business plan could describe the usefulness of value-added information. Funding There are no technological barriers to collecting additional data; however, funding for collection and reporting of cancer data is largely decentralized. A national system of data collection would require reconciliation of data from multiple sources, edits and quality checks, thereby adding to registry workloads. Registry collaboration will thus be required both to ensure data quality and to lobby for sufficient funding to support health records and systems development. When seeking funding, it should be emphasized that the costs of data collection and management are minimal compared to the costs of treatment. Further, stage or treatment data are useful in monitoring the efficacy of control programs and in other cost-benefit evaluations. Capturing Treatment-related Data Many of the considerations for collecting population-based stage data also apply to treatment data. The applications for population-based treatment-related data, as for stage data, extend beyond cancer survival surveillance, for instance, to examining patient access to optimal care. Cancer survival surveillance that includes population-based treatment data will enhance the ability to interpret survival patterns observed. Only summary data (e.g. whether surgery, chemotherapy or radiation therapy was used in initial treatment, whether stage was diagnosed at surgery) would be required; more detailed treatment data (e.g. chemotherapy dosages, radiation fractions used) are more appropriate for use in clinical trials. Population-based collection of treatment data may also be useful in the following patient management surveillance applications.
Barriers to capturing treatment data are similar to those for stage data. A standardized system for data collection; mechanisms to monitor data quality; compliance with reporting guidelines; and, above all, the administrative will to initiate and maintain collection will address these barriers. Data capture options could be initiated by a sampling or population-based approach, or a mix of both. A proposed core data set for treatment and treatment-related information is provided as the Appendix. Analyses, Methodology and Applications for Cancer Survival Surveillance The following strategies will facilitate cancer survival surveillance at a national level. Analytic considerations Given the diversity of resources, data and data quality across registries, the most appropriate way to conduct large-scale population-based survival analyses is to establish an expert working group comprising representatives from the provincial cancer registries and LCDC. The Cancer Bureau of LCDC should facilitate this group's formation and initiate efforts to produce a national report. International experts should be consulted in the initial phases of the working group's efforts. The working group's mandate should include specifying methodology and computer application tools to be used and defining minimum data standard inclusion criteria. A basic national descriptive report should include national and provincial survival experiences (as available) by site, sex, age group and calendar period. Subsequent reports could focus on targeted projects such as explanation of regional discrepancies and impact of socio-economic status on cancer survival.
Although information on cancer stage and treatment would greatly facilitate the interpretation of population-based survival analyses, the process for capturing these data at a population level is anticipated to take some time. A national report should be prepared without delay to document the differences in survival across regions. Future work can provide more in-depth analyses and interpretation of discrepancies in outcomes once supplementary data on stage and treatment are available. Cancer survival reports should be prepared periodically. Data considerations It is absolutely essential that a national death clearance be conducted on all Canadian data from 1969 on. Although the death clearance is now performed in some provinces, there is no national death clearance mechanism (to identify deaths in provinces other than the province in which the cancer was diagnosed). Further, death clearance has not been conducted for an extended period in several provinces, and this has precluded even cursory monitoring of cancer survival among these populations. National death clearance, after completion of clearance of 1969-to-current data, should be systematic and periodic (at intervals to be recommended by the working group). Workshop participants with expertise in analysis strongly endorsed efforts to collect stage and treatment data on a population basis and identified additional data elements for cancer survival monitoring. Priorities identified included ethnicity and socio-economic status (SES). Although it is feasible to use aggregate SES data for ecological level analyses, individual data are preferable. The Workshop group strongly endorsed continued attention to quality assurance processes. Summary Mechanisms and technology already exist to collect survival data nationally. The major challenge is to develop the will and collaboration of those responsible to ensure the completeness, quality and timeliness of data. Administrative willpower is the key to success. To develop systematic national data collection, several key challenges must be addressed.
Acknowledgements The contributions of these corporate sponsors are gratefully acknowledged: Rhone-Poulenc Rorer, Eli Lilly Canada, Bristol-Myers Squibb and Glaxo-Wellcome Pharmaceuticals. References 1. McLaughlin JR, Sloan MR, Janovjak DP. Cancer survival in Ontario. Toronto: Ontario Cancer Treatment and Research Foundation, 1995. 2. MacKillop WJ, Groome PA, Zhang-Solomons J, Zhou Y, 3. 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 1995;32(2):531-9. 4. Hakulinen T. Cancer data collection and comparability in Europe. In: Duchêne J, Wunchch G, editors. Collecte et comparabilité des données démographiques et sociales en Europe. Louvain-laNeuve: Academia/L'Harmattan, 1995:487-503. 5. Hakulinen T, Teppo L. Causes of death among female patients with cancer of the breast and intestines. Ann Clin Res 1977;9:15-24. 6. Ederer F, Axtell LM, Cutler SJ. The relative survival rate: a statistical methodology. Natl Cancer Inst Monogr 1961;6:101-21. 7. Hakulinen T, Abeywickrama K. A computer program package for relative survival analysis. Comp Prog Biomed 1985;19:197-207. 8. Hakulinen T, Tenkanen L. Regression analysis of relative survival rates. Appl Stat 1987:36:309-17. 9. Estéve J, Benhamou E, Croasdale M, Raymond L. Relative survival and the estimation of net survival; elements for further discussion. Stat Med 1990;9:529-38. 10. Axtell LM, Asire AJ, Myers MH, editors. Cancer patient survival. Report number 5. Bethesda: US Department of Health, Education and Welfare, 1976. 11. Toms JR. Trends in cancer survival in Great Britain. Cases registered between 1960 and 1974. London: Cancer Research Campaign, 1982. 12. Berrino F, Sant M, Verdecchia A, Capocaccia R, Hakulinen T, Estéve J. Survival of cancer patients in Europe. The EUROCARE study. Lyon: International Agency for Research on Cancer, 1995; IARC Scientific Publications No 132. 13. Dickman P, Gibbert RW, Hakulinen T. Estimating potential savings in cancer deaths by eliminating regional and social class variation in cancer survival in the Nordic countries. J Epidemiol Community Health 1997;51;289-98. Participants Carol Acorn (Canadian Institute for Health Information);
Susan Aitken (Ontario Breast Screening Programme); Fred Alexander (Tom
Baker Cancer Centre); David Beatty (Sunnybrook Health Science Centre);
Lou Benedet (British Columbia Cancer Registry); Penny Brasher (Alberta
Cancer Board); Kathy Clarke (Laboratory Centre for Disease Control [LCDC]);
Andrew Coldman* (British Columbia Cancer Agency); Ron
Dewar (Nova Scotia Cancer Registry); Bill Evans** (Ottawa
Regional Cancer Centre); Gord Fehringer (Cancer Care Ontario [formerly
OCTRF]); Barbara Foster (LCDC); Suzana Fraser (LCDC); Timo Hakulinen*
(Finnish Cancer Registry); Juanita Hatcher (Alberta Cancer Board); Eric
Holowaty* (Cancer Care Ontario); Neill Iscoe (Sunnybrook
Regional Cancer Centre); Deborah Jordan (LCDC); Sherry Kennedy (Canadian
Institute for Health Information); Erich Kliewer (Manitoba Cancer Treatment
and Research Foundation); Tony Labillois (Statistics Canada); Karen Lemieux
(Cancer Care Ontario); Isra Levy (LCDC); Diane Logan (Ottawa Regional
Cancer Centre); William MacKillop* (Kingston Regional
Cancer Centre); John Mark Mahue (Glaxo-Wellcome Pharmaceuticals); Yang
Mao (LCDC); David McCready (Women's Hospital); Christina Mills (LCDC);
Jayanti Mukherjee (Bristol-Myers Squibb); Brian O'Sullivan*
(Princess Margaret Hospital); Joseph Pater (Queen's University); Gilles
Pelletier (Ministère de la Santé et des Services sociaux); Lynn Gloeckler
Ries* (US National Cancer Institute); Diane Robson (Saskatchewan
Cancer Agency); Dena Schanzer (LCDC); F Shepherd (The Toronto Hospital);
Margaret Sloan* (Cancer Care Ontario); Hartley Stern
(Ottawa Civic Hospital); Simon Sutcliffe (British Columbia Cancer Agency);
Jane Thomas (Riverside Hospital of Ottawa); Jon Tonita (Saskatchewan Cancer
Agency), Arduino Verdecchia* (Instituto Superiore di
Sanità); Stephen Walter (McMaster University); Heather Whittaker (Manitoba
Cancer Treatment and Research Foundation); Barbara Whylie**
(National Cancer Institute of Canada); Don Wigle* (LCDC); Jun Zhang
(LCDC) |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Author References |
![]() |
|||
![]() |
Last Updated: 2002-10-20 | ![]() |