MESURE : Méthodes d’Estimation de la SUrvie RElative
Ce travail a bénéficié d'une aide de l'Agence Nationale de la Recherche dans le cadre de l'appel à projets programme Blanc 2009 (ANR-09-BLAN-0357-01).
Contact : Roch GIORGI
Cancer survival is a valuable indicator of progress in cancer control. Over the past decade, population-based cancer registry data have been used increasingly worldwide to evaluate and improve the quality of cancer care. At present, the methods that estimate cancer survival from population-based data rely mainly on the relative survival approach. However, the models used to estimate relative survival at different times since diagnosis and to incorporate prognostic covariates may vary between research centres. The conclusions from such studies may rely on the methods used to analyse the data and observed differences could be attributed to methodological and statistical approaches or to real effects.
Thus, the influence of different methodological options must be explored and assessed. In addition, some important problems are still waiting for appropriate solutions (i.e. for example, spatial variation in survival time, unobserved heterogeneity).
A research network whose members are in charge of the production of national estimates of cancer survival in their respective countries will discuss, develop, and monitor implementation of a common methodology that will make unbiased cross-national comparisons feasible. That collaboration between statistical research teams having methodological expertise in that field will reinforce the search for this statistical methodology and will provide the adequate tools to improve the quality and the comparability of cancer survival data in Europe. It will also permit the appropriate extension of that methodology to other chronic diseases and to “clinical research” data. Results will have direct application for public health decision makers.
The overall aim of this project is to improve the current methods for estimating cancer survival and to define the required specifications for the production of large-scale cancer survival statistics useful in Public Health in order to obtain i) more accurate estimates, and ii) comparable information on cancer survival and on its public health impact.
The main impact will concern the optimal use of registry data for cancer epidemiology. The research will produce a range of new clinical and public health insights from cancer survival data. The project is original in the cancer survival estimation framework because it proposes to explore and to address two complementary themes:
- Improvements in cancer relative survival analysis methodology;
- Creation of a “standardized” approach for estimating cancer survival in France and in European Regions using Registry data.
Both themes are challenging in a public health point of view. With this project, we hope to provide a more complete set of analytic tools to estimate cancer survival and to assess the public health implications of these estimates.
In this project, we will have to investigate several methods for analyzing cancer survival and to address several questions concerning the underlying hypothesis of these methods. First of all, in order to define the required specifications for the production of large-scale cancer survival statistics useful in Public Health (which indicators and which key prognostic covariates?), it is necessary to formalize the “state of the art” and to identify the current limits of the methods, focusing on the need for cross-national comparisons. Then, using the background methodology used in (non-relative) survival analysis, we will have to propose new extensions of the existing relative survival models (for example, to account for individual frailty).
At the end of this project, in order to optimize the use of registry data for cancer epidemiology, we will be able to publish guidelines for population-based cancer survival analysis. These will be directed to the cancer registration community.
With this project, we aim to provide a more complete set of analytic tools to estimate cancer survival and to assess the public health implications of these estimates. To achieve these objectives, we whish to create a new network between established research teams which already have substantial and complementary experiences, both in cancer survival analysis and in methodological development. Both competences are necessary in order to analyze the data with a relevant epidemiological expertise and to deal with technical (statistical) problems.
To achieve our objectives, the scientific program is divided into 3 work packages (WP), one of them divided into 3 sub-work packages:
- WP1: Comparison of the current methods used to estimate relative survival
- WP2: Extension and development of statistical methods used to estimate relative survival
- WP2.1: Statistical researches to deal with the important remaining issues
- WP2.2: Statistical programs
- WP3: Production of guidelines for cancer survival analysis with population-based data, answering to public health considerations
Partners of the project
- Laboratoire d’Enseignement et de Recherche sur le Traitement de l’Information Médicale, Marseille, France (EA 3283) - site web
- Service de Biostatistique des Hospices Civils de Lyon, Lyon – UMR CNRS 5558 et Université Claude Bernard Lyon I, Lyon, France
- Service de Biostatistique et Informatique Médicale, Dijon, France – INSERM Equipe 5 U866
- Laboratoire d’épidémiologie et de santé publique, Strasbourg, France (EA 1801)
- Réseau FRANCIM: Réseau des registres français des cancers, France
- Non communicable Disease Epidemiology Unit, London School of Hygiene and Tropical Medicine, London, United-Kingdom - site web
- Centro Nazionale di Epidemiologia, Instituto Superior di Sanita, Roma, Italy - site web
- Department of Epidemiology and Biostatistics, McGill university, Montreal, Canada - site web