Do Precipitation Levels Affect Preterm Birth Rates in Puerto Rico?
Name: Xue Yu Category: Engineering and Technology Abstract ID# 1750
Do Precipitation Levels Affect Preterm Birth Rates in Puerto Rico? Xue
1,* Yu ,
Akram
1 Alshawabkeh ,
Zlatan
1 Ferric ,
José F.
2 Cordero
1Department
of Civil and Environmental Engineering, Northeastern University, Boston, MA 02115 2University of Georgia College of Public Health, Department of Epidemiology and Biostatistics, Athens, GA 30602
Puerto Rico Testsite for Exploring Contamination Threats Approaching
20% 20% CLOSE to
of preterm births of Preterm Births
MORE than This program is supported by Award Number P42ES017198 from the National Institute of Environmental Health Sciences.
150 150
MORE than
Superfund Sites
Contaminated Sites
Puerto Rico Hypothesis
Results
• Environmental changes such as precipitation have some effects on preterm birth rates
General Linear Model • Limited modeling ability • Adjusted r2=0.03, p=0.002
Reasoning • • •
Puerto Rico receives a high amount of precipitation, about 1,550 mm a year There are clear seasonal rainfall patterns, with June to November as wet season and hurricane occurrences between August and November The large amount of precipitation may facilitate the introduction of pollutants to the water system
Approach • • •
Non-linear or semi-random walk pattern of preterm birth rate time series. Use the city of San Juan, Puerto Rico as the study location. Apply distributed lag non-linear models (DLNM) to explore the effects of precipitation on preterm birth rates, where the preterm birth rates were fitted with quasi-Poisson regression: log 𝐸 𝑌𝑡
Data • Preterm birth rates data were collected from the National Center for Health Statistics (NCHS). • Climate data (precipitation, temperature) were collected from National Oceanic and Atmospheric Administration (NOAA). • The following: time series plots of preterm birth rates (rate), monthly precipitation (rain), and maximum and minimum monthly temperatures at San Juan, Puerto Rico
= 𝛼 + 𝛽1𝑃𝑟𝑒𝑐𝑖𝑝𝑖𝑡𝑎𝑡𝑖𝑜𝑛 + 𝛽2𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 + 𝛽3𝑀𝑜𝑛𝑡ℎ𝑁𝑆(𝑀𝑜𝑛𝑡ℎ, 12)
•
Apply Bayesian State Space model to estimate the effect of precipitation on preterm birth rates -- Using Monte Carlo Markov Chain (MCMC) to update the posterior distributions of preterm birth rates based on precipitation inputs.
Acknowledgements The project described was supported by Award Number P42ES017198 from the National Institute Of Environmental Health Sciences.
Distributed Lag Non-linear model • High precipitation has more (negative) effect on the change of preterm birth rates
Non-linear relationship between precipitation and preterm birth rates • The precipitation effects for zero lag and twelve-month lag are similar, except the lag 12 months has deeper U-shape. • The 100 mm precipitation effects showed a U-shape, with higher risks of preterm birth rates at both low and high lags. • The 150 mm precipitation effects decreased with the increase of time lags.
Implications and future work Our Results showed that higher precipitation has more immediate effects (small time lags) on preterm birth rates, while the effects of lower precipitation on preterm birth rates are showed after a few months. The future work include: evaluate the precipitation effects to other areas of Puerto Rico; apply Bayesian State Space Model to estimate precipitation effects on preterm birth rates.
Do Precipitation Levels Affect Preterm Birth Rates in Puerto Rico? Xue
1,* Yu ,
Akram
1 Alshawabkeh ,
Zlatan
1 Ferric ,
José F.
2 Cordero
1Department
of Civil and Environmental Engineering, Northeastern University, Boston, MA 02115 2University of Georgia College of Public Health, Department of Epidemiology and Biostatistics, Athens, GA 30602
Puerto Rico Testsite for Exploring Contamination Threats Approaching
20% 20% CLOSE to
of preterm births of Preterm Births
MORE than This program is supported by Award Number P42ES017198 from the National Institute of Environmental Health Sciences.
150 150
MORE than
Superfund Sites
Contaminated Sites
Puerto Rico Hypothesis
Results
• Environmental changes such as precipitation have some effects on preterm birth rates
General Linear Model • Limited modeling ability • Adjusted r2=0.03, p=0.002
Reasoning • • •
Puerto Rico receives a high amount of precipitation, about 1,550 mm a year There are clear seasonal rainfall patterns, with June to November as wet season and hurricane occurrences between August and November The large amount of precipitation may facilitate the introduction of pollutants to the water system
Approach • • •
Non-linear or semi-random walk pattern of preterm birth rate time series. Use the city of San Juan, Puerto Rico as the study location. Apply distributed lag non-linear models (DLNM) to explore the effects of precipitation on preterm birth rates, where the preterm birth rates were fitted with quasi-Poisson regression: log 𝐸 𝑌𝑡
Data • Preterm birth rates data were collected from the National Center for Health Statistics (NCHS). • Climate data (precipitation, temperature) were collected from National Oceanic and Atmospheric Administration (NOAA). • The following: time series plots of preterm birth rates (rate), monthly precipitation (rain), and maximum and minimum monthly temperatures at San Juan, Puerto Rico
= 𝛼 + 𝛽1𝑃𝑟𝑒𝑐𝑖𝑝𝑖𝑡𝑎𝑡𝑖𝑜𝑛 + 𝛽2𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 + 𝛽3𝑀𝑜𝑛𝑡ℎ𝑁𝑆(𝑀𝑜𝑛𝑡ℎ, 12)
•
Apply Bayesian State Space model to estimate the effect of precipitation on preterm birth rates -- Using Monte Carlo Markov Chain (MCMC) to update the posterior distributions of preterm birth rates based on precipitation inputs.
Acknowledgements The project described was supported by Award Number P42ES017198 from the National Institute Of Environmental Health Sciences.
Distributed Lag Non-linear model • High precipitation has more (negative) effect on the change of preterm birth rates
Non-linear relationship between precipitation and preterm birth rates • The precipitation effects for zero lag and twelve-month lag are similar, except the lag 12 months has deeper U-shape. • The 100 mm precipitation effects showed a U-shape, with higher risks of preterm birth rates at both low and high lags. • The 150 mm precipitation effects decreased with the increase of time lags.
Implications and future work Our Results showed that higher precipitation has more immediate effects (small time lags) on preterm birth rates, while the effects of lower precipitation on preterm birth rates are showed after a few months. The future work include: evaluate the precipitation effects to other areas of Puerto Rico; apply Bayesian State Space Model to estimate precipitation effects on preterm birth rates.
