Bayesian clustering of 109 worldwide countries according to the trend of prostate cancer mortality rates from 1990 to 2019.

Neglecting possible growth heterogeneity and focusing only on the overall pattern of prostate cancer mortality rates can result in misunderstandings and incorrect conclusions about the growth process of the outcome. The main goal of this study was to capture the heterogeneity of prostate cancer mortality rates among countries from 1990 to 2019. To accomplish this aim, we performed the Bayesian latent Growth Mixture Model (GMM). In this longitudinal study, Prostate cancer mortality rates data from 1990 to 2019, as well as the Human Development Index (HDI), the Gross National Income (GNI), and the Life Expectancy at Birth (LEB), were obtained from WHO and UNDP platforms. The Bayesian GMM was used to discover homogeneous subgroups and estimate the pattern of prostate cancer mortality rates in each subgroup. The HDI, GNI and life expectancy effects were estimated using a Bayesian conditional Latent Growth Model (LGM). Globally, the intercept and the slope of the Bayesian LGM were equal to 8.77 and 0.19, respectively. The Bayesian GMM classified the 109 countries into four groups, which had significant positive growth patterns with different slopes except for the first class. The effect of HDI was not significant on the overall prostate cancer death rates, but GNI and LEB had a significantly positive effect on the model's intercepts and a significantly negative effect on the slope. Although the prostate cancer mortality rate increased globally, it has four distinct latent subgroups with various patterns. Additionally, the effects of HDI, GNI, and LEB on prostate cancer mortality rates varied significantly among the four subgroups, indicating a need for further investigation to identify causal factors.