Twin peaks : a spatial and temporal study of twinning rates in Brazil

Twin births are an important public health issue due to health complications for both mother and children. While it is known that contemporary factors have drastically changed the epidemiology of twins in certain developed countries, in Brazil, relevant data are still scarce. Thus, we carried out a population-based study of live births in spatial and temporal dimensions using data from Brazil's Live Birth Information System, which covers the entire country. Over 41 million births registered between 2001 and 2014 were classified as singleton, twin or multiple. Twinning rates (TR) averaged 9.41 per 1,000 for the study period and a first-order autoregressive model of time-series analysis revealed a global upward trend over time; however, there were important regional differences. In fact, a Cluster and Outlier Analysis (Anselin Local Moran's I) was performed and identified clusters of high TR in an area stretching from the south of Brazil's Northeast Region to the South Region (Global Moran Index = 0.062, P < 0.001). Spearman's correlation coefficient and a Wilcoxon matched pairs test revealed a positive association between Human Development Index (HDI) and TRs in different scenarios, suggesting that the HDI might be an important indicator of childbearing age and assisted reproduction techniques in Brazil. Furthermore, there was a sharp increase of 26.42% in TR in women aged 45 and over during study period. The upward temporal trend in TRs is in line with recent observations from other countries, while the spatial analysis has revealed two very different realities within the same country. Our approach to TR using HDI as a proxy for underlying socioeconomic changes can be applied to other developing countries with regional inequalities resembling those found in Brazil