Associations of hurricane exposure and forecasting with impaired birth outcomes

The increasing frequency and intensity of unpredictable tropical storms, coupled with diminishing funding for forecasting, present a significant societal challenge. While accurate forecasts are crucial for mitigating infrastructure damage and protecting human lives, the impact of forecast accuracy on human health, particularly during pregnancy, remains understudied. This study investigates the relationship between in utero exposure to Hurricane Irene and birth outcomes, considering the role of forecast accuracy in influencing healthcare access and maternal stress. The release of uncertain forecasts poses a public health risk due to increased psychological stress and potential disruption of prenatal care. Existing literature links disaster exposure during pregnancy to impaired birth outcomes, including reduced birth weight, gestation length, and increased incidence of preterm and low birth weight infants. These adverse birth outcomes are associated with long-term implications for the child's health, development, and future well-being. This study addresses this gap by examining the impact of Hurricane Irene exposure on birth outcomes, while exploring the mediating role of forecast accuracy and the disruption of prenatal care services.

Prior research has established the detrimental effects of disaster exposure during pregnancy on birth outcomes, citing stress, environmental toxins, and disrupted healthcare access as key factors. Studies have linked *in utero* exposure to various stressors to reduced birth weight, shorter gestation lengths, and increased risks of preterm birth and low birth weight. These negative impacts have been further linked to long-term health consequences, including increased susceptibility to diseases, mental health challenges, reduced cognitive abilities, lower educational attainment, and reduced earning potential. However, a critical gap in the existing literature is the lack of large-scale studies examining the interplay between forecast accuracy and the resulting health impacts on pregnant populations exposed to hurricanes. The present study directly addresses this knowledge gap by investigating the impact of both direct hurricane exposure and the uncertainty associated with hurricane forecasts.

This study analyzed North Carolina Department of Health and Human Services vital statistics data, encompassing over 700,000 births from 2006 to 2012. The sample focused on births in communities affected by Hurricane Irene. A natural experiment approach was used, comparing birth outcomes before and after the hurricane's landfall within the same zip codes. The primary exposure variable was rainfall intensity at the residential address of pregnant women. The study incorporated zip code and monthly fixed effects to control for local factors and seasonal trends, with standard errors clustered at the county level. The impact of Hurricane Irene exposure on birth outcomes was assessed for various measures, including birth weight, gestation length, and the incidence of low birth weight, very low birth weight, preterm, and extremely preterm births. Potential mediating factors such as groundwater contamination and disruptions to prenatal care were also investigated. Groundwater contamination analysis utilized over 17,000 private well water samples. Prenatal care disruption was measured using the month prenatal care began and the total number of prenatal care visits. The study also investigated the impact of the length of time a pregnant woman resided within the National Hurricane Center's "Cone of Uncertainty" during Hurricane Irene, differentiating between those who ultimately experienced severe impacts and those who did not, to explore the role of type I and type II forecasting errors. Econometric analysis involved estimating an equation that included a binary variable indicating treatment (post-hurricane births), the natural logarithm of 24-hour maximum rainfall, and their interaction. This allowed for the estimation of nonlinear impacts of hurricane exposure across rainfall intensities. The study followed STROBE guidelines for observational studies, and ethical approval was obtained from East Carolina University's IRB.

In utero exposure to Hurricane Irene was associated with statistically significant negative impacts on birth outcomes. Birth weight was reduced by an average of 12.7g (0.17-0.61% reduction), with the largest effect observed in populations experiencing hurricane-force winds and heavy rainfall. Gestation lengths were shortened by an average of 0.10 weeks (0.18-0.36% reduction). There were statistically significant increases in the likelihood of low birth weight (0.56 percentage points, 2.52-10.34% increase), very low birth weight (0.38 percentage points, 15.33-34.67% increase), preterm birth (0.96 percentage points, 5.20-13.53% increase), and extremely preterm birth (0.56 percentage points, 12.07-26.90% increase). The magnitudes of these effects did not significantly vary across different wind and rainfall intensities, suggesting a mechanism other than direct physical storm exposures. Analysis of groundwater contamination did not reveal a relationship between storm exposure intensity and private well water contamination rates. Similarly, no relationship was found between exposure intensity and pre-pregnancy hypertension, previous poor pregnancy outcomes, gestational hypertension, or eclampsia. However, significant disruptions to prenatal care were observed. Hurricane exposure was linked to a delay in the first prenatal care appointment (0.24 months, 6.92-11.54% delay) and a reduction in the total number of prenatal care visits (0.63 appointments, 3.03-7.29% reduction). These disruptions in prenatal care were consistent across storm exposure intensities. Examining the additional advisories within the "Cone of Uncertainty", the study found that for heavily exposed populations (rain > 2 inches), additional advisories were associated with a statistically significant decrease in preterm births, while lightly exposed populations (rain ≤1 inch) experienced a marginally significant increase in the incidence of low birth weight, very low birth weight, preterm, and extremely preterm births.

The study's findings demonstrate widespread and detrimental impacts of in utero Hurricane Irene exposure on birth outcomes, consistent with previous research on disaster exposure and pregnancy. The consistent birth impacts across storm exposure intensities suggest that the primary mechanism is not direct physical exposures, but rather the anticipation of hurricane exposures and subsequent disruptions to healthcare services. The absence of a correlation between groundwater contamination and birth outcomes further supports this conclusion. The study's findings on prenatal care disruptions highlight a critical role of institutional responses to the anticipated hurricane exposure, independent of the actual storm's intensity. The study provides novel evidence linking uncertain hurricane forecasts to disruptions in healthcare access, thereby impacting birth outcomes. The observed magnitude of these impacts is comparable to those observed in response to other traumatic events during pregnancy. The results suggest that the timing and accuracy of hurricane forecasts are crucial in mitigating their negative impacts on maternal health and birth outcomes.

This study provides compelling evidence that in utero exposure to Hurricane Irene negatively affected birth outcomes, and that these effects are not solely attributable to direct physical exposure but are strongly mediated by forecast-driven disruptions to prenatal care. Future research should explore the heterogeneity in treatment effects, investigate the role of hurricane "experience," and consider the trimester of exposure. Behavioral research is needed to understand how populations respond to delayed, yet potentially more accurate, forecasts. Further investigation is also necessary to understand how different types of storms and their associated physical exposures might impact both population responses and birth outcomes.

While the study used a large dataset of North Carolina births, it focused on Hurricane Irene specifically and may not be generalizable to all hurricane events. The analysis relies on rainfall intensity as a proxy for exposure intensity and does not directly measure the physical or psychological stress experienced by pregnant women. The study does not explicitly model the psychological impacts of delayed forecasts. The assumption that pregnant women did not select into treatment based on their anticipated exposure to Hurricane Irene is a key limitation.