RESUMO
BACKGROUND: Cancer survival is a key component to assess the overall effectiveness of healthcare systems in their cancer management efforts. A key supporting tool for planning and decision making was introduced with the development of an index of cancer survival that summarises survival for all adults and cancer types into one single estimate, but the implementation details have not been previously described. METHODS: We detail the construction of the index, including the structure, the calculation of 'sex-age-cancer' specific weights and our proposed modelling strategy to estimate net survival. We provide some practical recommendations through an illustration using a synthetic dataset ('Replica') that we generated for this purpose. An example of R code usage to estimate the index using our approach is provided. RESULTS: The 'Replica' contains 500 000 artificial cancer records that mimic a cohort of adult cancer patients diagnosed with cancer in England between 1980 and 2004. Using this dataset, we estimated an index of cancer survival at one, five, and ten years after diagnosis for five selected periods of diagnosis, and provide an example of interpretation of these results. DISCUSSION: We propose a flexible penalised regression modelling strategy to estimate the index's 'sex-age-cancer' specific cancer survival components that minimises the estimation challenge of these components. This tutorial will support researchers in constructing an index of cancer survival for their own setting, facilitating the enrichment of existing toolkits of cancer indicators to more effectively measure progress against cancer in their respective regions/countries.
Assuntos
Neoplasias , Humanos , Neoplasias/mortalidade , Feminino , Masculino , Adulto , Taxa de Sobrevida , Inglaterra/epidemiologia , Pessoa de Meia-Idade , IdosoRESUMO
Relative survival represents the preferred framework for the analysis of population cancer survival data. The aim is to model the survival probability associated with cancer in the absence of information about the cause of death. Recent data linkage developments have allowed for incorporating the place of residence into the population cancer databases; however, modeling this spatial information has received little attention in the relative survival setting. We propose a flexible parametric class of spatial excess hazard models (along with inference tools), named "Relative Survival Spatial General Hazard," that allows for the inclusion of fixed and spatial effects in both time-level and hazard-level components. We illustrate the performance of the proposed model using an extensive simulation study, and provide guidelines about the interplay of sample size, censoring, and model misspecification. We present a case study using real data from colon cancer patients in England. This case study illustrates how a spatial model can be used to identify geographical areas with low cancer survival, as well as how to summarize such a model through marginal survival quantities and spatial effects.
Assuntos
Neoplasias do Colo , Humanos , Modelos de Riscos Proporcionais , Análise de Sobrevida , Simulação por Computador , Tamanho da Amostra , Modelos EstatísticosRESUMO
Background and Aims: Esophageal cancer claims more than 500,000 deaths worldwide, with half occurring in China. We aimed to synthesize existing evidence on stage-specific survival from this cancer in China to inform cancer control strategies. Methods: English and Chinese literature databases were systematically searched to identify original research published up to May 31, 2019 that reported stage-specific survival from esophageal cancer in China. Two meta-analyses were performed using random-effects models to summarize stage-specific survival differences on relative and absolute scales. The number of esophageal cancer deaths that might have been prevented by early detection in China, in 2018, was estimated assuming 2 different downstaging scenarios. Results: One hundred fifty eligible studies were identified, 97 had non-overlapping study populations (83,063 participants), 47 were included in the meta-analysis of hazard ratios, and 26 in the meta-analysis of survival probabilities. Late-stage (III-IV) was associated with 92% higher hazard of death compared with early-stage (0-II) (95% confidence interval 1.62-2.28), corresponding to an absolute 5-year survival difference of 31.2% (29.9%-32.4%). In all, 5.2% esophageal cancer deaths could have been prevented in China, in 2018, if the observed stage distribution at diagnosis (â¼50% early-stage) was shifted to the real-life conditions of a population-based endoscopic screening program (â¼60% early-stage) and 26.9% if shifted to that observed in the controlled setting of a randomized trial (â¼90% early-stage). Conclusion: Shifting downwards the stage distribution of esophageal cancer through screening would bring moderate reductions in mortality from the disease. Treatment improvements for early-stage patients are needed to reduce further mortality from this cancer.