Underlying analysis for the paper: ‘Rainfall variability and adverse birth outcomes in Amazonia’


Erick A. Chacón-Montalván, Benjamin M. Taylor, Marcelo G. Cunha, Gemma Davies, Jesem D. Y. Orellana & Luke Parry



Paper’s abstract

Amazonian populations are increasingly exposed to climatic shocks, yet knowledge of related health impacts is limited. Understanding how health risks are coproduced by local climatic variability, place and social inequities is vital for improving decision-making, particularly in decentralized contexts. We assess the impacts of rainfall variability and multiscale vulnerabilities on birth weight, which has lifelong health consequences. We focus on highly river-dependent areas in Amazonia, using urban and rural birth registrations during \(2006-2017\). We find a strong but spatially differentiated relationship between local rainfall and subsequent river-level anomalies. Using Bayesian models, we disentangle the impacts of rainfall shocks of different magnitudes, municipal characteristics, social inequities and seasonality. Prenatal exposure to extremely intense rainfall is associated with preterm birth, restricted intra-uterine growth and lower mean birth weight (\(\leq -183\) g). Adverse birth outcomes also follow non-extreme intense rainfall (\(40%\) higher odds of low birth weight), drier conditions than seasonal averages (\(-39\) g mean birth weight) and conception in the rising-water season (\(-13\) g mean birth weight). Babies experience penalties totalling \(646\) g when born to adolescent, Amerindian, unmarried mothers that received no formal education or antenatal or obstetric health care. Rainfall variability confers intergenerational disadvantage, especially for socially marginalized Amazonians in forgotten places. Structural changes are required to reduce inequities, foster citizen empowerment and improve the social accountability of public institutions.

About this web page

This web renders the collection of scripts used for the underlying analysis done for the paper Rainfall variability and adverse birth outcomes in Amazonia published at Nature Sustainability. The analysis was done using the software R, so each page includes the r code, output and description of tasks we performed. You can navigate the scripts through the left sidebar, they are organized in the following sections:

  • Data processing: Due to the complexity of the processing, this project does not include all the processing scripts.
  • Data exploration: It includes some simple models for birth weight (BW), low birth weight (LBW), and preterm birth (PTB).
  • Data modelling: It includes all the models we used for BW, LBW and PTB with the corresponding MCMC diagnostics.
  • Summarisation: It includes tables and figures created using processed data and models.

How to use?

You can simply explore this web page to check some of the scripts we did for particular tasks (e.g. modelling, visualization). However, you could also use it to reproduce the results with the same data we used or to perform a similar analysis in other areas of study. In case you would like to reproduce it, contact us to share the data and a docker image that will facilitate the execution of the code. The source code to render this web page can be found at https://gitlab.com/ErickChacon/birthweight.