The comparative burden of brain and central nervous system cancers from 1990 to 2019 between China and the United States and predicting the future burden

Abstract

BackgroundBrain and central nervous system (CNS) cancers represent a major source of cancer burden in China and the United States. Comparing the two countries' epidemiological features for brain and CNS cancers can help plan interventions and draw lessons.MethodsData were extracted from the Global Burden of Disease repository. The average annual percentage change (AAPC) and relative risks of cancer burdens were calculated using joinpoint regression analysis and age-period-cohort (APC) models, respectively. Moreover, a Bayesian APC model was employed to predict the disease burden over the next decade.ResultsFrom 1990 to 2019, the number of incidences, deaths, and disability-adjusted life-years (DALYs) increased in China and the US, with a larger increase in China. Age-standardized incidence rates in China and the United States have shown an increasing trend over the past three decades, with AAPCs of 0.84 and 0.16%, respectively. However, the rates of age-standardized mortality and age-standardized DALYs decreased in both countries, with a greater decrease in China. Overall, age trends in cancer burden were similar for males and females, with two peaks in the childhood and elderly groups, respectively. The period and cohort effects on incidence showed an overall increasing trend in China and limited change in the US. However, the period effects for mortality and DALY were decreasing in both countries, while the cohort effects tended to increase and then decrease. Moreover, we predicted that the cancer burdens would continue to rise in China over the next decade.ConclusionThe burden of brain and CNS cancers is substantial and will continue to increase in China. Comprehensive policy and control measures need to be implemented to reduce the burden.

Increasing Burden of Osteoarthritis in China: Trends and Projections from the Global Burden of Disease Study 2019

“…BAPC and Nordpred models are commonly used in projecting disease burden [ 26 , 27 ]. We first validated the accuracy of the 2 models using data from 1990 to 2019.…”

“…Then, data from 1990 to 2009 were used to project the osteoarthritis burden from 2010 to 2019, and the observed values from 2010 to 2019 were used to validate the accuracy of the projected values. Error rate, calculated by ( predicted values - observed values ) ÷ (observed values), was used to assess the performance of the Nordpred and BAPC models [ 27 ]. Additionally, the number of events was also calculated under the assumptions that the rates remained the same (the optimistic reference), decreased by 1%, or increased by 1% per year, based on the observed rate in 2019, to facilitate comparison with projected results [ 28 , 29 ].…”

Type 1 and type 2 diabetes mortality burden: Predictions for 2030 based on Bayesian age‐period‐cohort analysis of China and global mortality burden from 1990 to 2019

“…The BAPC model is based on integrated nested Laplace approximations to approximate the marginal posterior distribution, mitigating some of the mixing and convergence issues traditionally associated with Markov Chain Monte Carlo sampling techniques used in Bayesian methods. This approach aids in estimating latent parameters and inferring critical insights into the complex interplay of age, time periods, and generational factors on diabetes mortality rates 23 , 24 . The BAPC method not only endows us with retrospective capabilities to review historical data but also provides foresight in predicting future trends, offering invaluable insights for the formulation of public health strategies.…”

Global spatiotemporal distributions of lymphoma from 1990 to 2019: A Joinpoint regression analysis based on the global burden of disease study 2019, and projections until 2044

“…Within the BAPC model, all unknown parameters are considered random, with suitable prior distributions. Bayesian inference utilizes a second-order random walk to smooth the prior effects of age, period, and cohort [ [39] , [40] , [41] ]. It has been observed to provide better coverage and accuracy compared to traditional projection models such as the Nordpred model, smooth spline model, and Poisson regression [ 24 , 38 , [42] , [43] , [44] ].…”