Although the epidemiology of malignant bone tumours in children and young adults has been explored, no definitive causation of any specific tumour has yet been identified. We performed a literature review (1970-2008) to find all papers covering possible aetiological factors involved in the development of bone tumours in children and young adults. Several associations have been reported with some consistency: the presence of hernias and Ewing sarcoma; high fluoride exposure and osteosarcoma; and parental farming and residence on a farm, younger age at puberty and family history of cancer for all bone tumours, especially osteosarcoma. Clearly further research is needed to confirm or refute these putative risk factors. It is likely that studies of gene-environment interactions may prove to be the most fruitful of future research.

There is a paucity of population-based studies examining incidence and survival trends in childhood bone tumours. We used high quality data from four population-based registries in England. Incidence patterns and trends were described using Poisson regression. Survival trends were analysed using Cox regression. There were 374 cases of childhood (ages 0 -14 years) bone tumours (206 osteosarcomas, 144 Ewing sarcomas, 16 chondrosarcomas, 8 other bone tumours) registered in the period 1981 -2002. Overall incidence (per million person years) rates were 2.63 (95% confidence interval (CI) 2.27 -2.99) for osteosarcoma, 1.90 (1.58 -2.21) for Ewing sarcoma and 0.21 (0.11 -0.31) for chondrosarcoma. Incidence of Ewing sarcoma declined at an average rate of 3.1% (95% CI 0.6 -5.6) per annum (P ¼ 0.04), which may be due to tumour reclassification, but there was no change in osteosarcoma incidence. Survival showed marked improvement over the 20 years (1981 -2000) for Ewing sarcoma (hazard ratio (HR) per annum ¼ 0.95 95% CI 0.91 -0.99; P ¼ 0.02). However, no improvement was seen for osteosarcoma patients (HR per annum ¼ 1.02 95% CI 0.98 -1.05; P ¼ 0.35) over this time period. Reasons for failure to improve survival including potential delays in diagnosis, accrual to trials, adherence to therapy and lack of improvement in treatment strategies all need to be considered.

BackgroundThere is a paucity of recent epidemiological data on bone cancers. The aim of this study was to describe incidence and survival patterns for bone cancers diagnosed during 1981 - 2002.MethodsCases aged 0 - 39 years (236 osteosarcomas, 166 Ewing sarcomas and 73 chondrosarcomas) were analysed using Poisson and Cox regressions.ResultsIncidence rates (per million persons per year) for osteosarcoma were 2.5 at age 0 - 14 years; 4.5 at age 15 - 29 years and 1.0 at age 30 - 39 years. Similarly, for Ewing sarcoma the incidence rates were 2.2; 2.9; 0.4 and for chondrosarcoma rates were 0.1; 1.2; 1.8 respectively. Incidence of osteosarcoma increased at an average annual rate of 2.5% (95% CI 0.4 - 4.7; P = 0.02), but there was no change in incidence of Ewing sarcoma or chondrosarcoma. There was a marginally statistically significant improvement in survival for Ewing sarcoma (hazard ratio (HR) per annum = 0.97; 95% CI 0.94 - 1.00; P = 0.06), although patients aged 15 - 39 years (n = 93) had worse overall survival than those aged 0 - 14 (n = 73; HR = 1.46; 95% CI 0.98 - 2.17; P = 0.06). There was no significant improvement in osteosarcoma survival (HR per annum = 0.98; 95% CI 0.95 - 1.01; P = 0.18).ConclusionsReasons for poorer survival in Ewing sarcoma patients aged 15 - 39 years and failure to significantly improve survival for osteosarcoma patients requires further investigation.

We address the task of choosing prior weights for models that are to be used for weighted model averaging. Models that are very similar should usually be given smaller weights than models that are quite distinct. Otherwise, the importance of a model in the weighted average could be increased by augmenting the set of models with duplicates of the model or virtual duplicates of it. Similarly, the importance of a particular model feature (a certain covariate, say) could be exaggerated by including many models with that feature. Ways of forming a correlation matrix that reflects the similarity between models are suggested. Then, weighting schemes are proposed that assign prior weights to models on the basis of this matrix. The weighting schemes give smaller weights to models that are more highly correlated. Other desirable properties of a weighting scheme are identified, and we examine the extent to which these properties are held by the proposed methods. The weighting schemes are applied to real data, and prior weights, posterior weights and Bayesian model averages are determined. For these data, empirical Bayes methods were used to form the correlation matrices that yield the prior weights. Predictive variances are examined, as empirical Bayes methods can result in unrealistically small variances.