Beyond Birth Weight: Identifying Early Life Exposure to Air Pollution through the Lens of Prenatal Ultrasound Scans

Evidence quantifying air pollution on birth outcomes in the developing context has started to accumulate. However, there is little evidence on how a fetus develops when exposed to air pollution during this critical stage of life and how the impact persists to determine intrauterine growth trajectories and birth outcomes that may shape long-term human capital.

Merging sources of daily air pollution data with hospital electronic records of prenatal ultrasound and birth outcome data as well as a longitudinal survey of parents’ health behavior in western China, this paper investigates the impact of prenatal exposures to outdoor air pollution on a rich set of fetal growth and birth outcomes as well as their trajectories.

We exploit variations in air pollutants that overlap with pregnancy. Pollution exposures are measured by days during pregnancy that fall into each of the unhealthy categories of air pollution regulated by the U.S. Environmental Protection Agency (EPA). We control for a very rich set of covariates, including family socioeconomic characteristics, parental health status and health behavior during pregnancy, fixed effects (year, month, district), non-linear time trend, weather conditions in flexible bin specifications (temperature, dew point, wind speed, and precipitation).

First, the birth outcome model indicate adverse impacts of PM10 (particulates with diameter of 10 micrometers or less) on a wide range of birth outcomes, while no such effect is found for NO2 or SO2 exposure. Specifically, being exposed to the top 20 percent pollution level for one more day will on average reduce birth weight by 5 grams, placenta weight by .75 gram, birth length by .3 millimeters, and abdominal circumference by .25 millimeters. The impact is generally larger for exposures in earlier months of pregnancy. Boys are more vulnerable than girls, and more operate through prematurity and less through Intrauterine Growth Restriction (IGR).

Second, prenatal ultrasound scans and birth outcomes are linked to capture intrauterine growth trajectories. Our main empirical strategy makes use of within-fetus variation between adjacent prenatal ultrasound scans, a short time window of only a few weeks, to mitigate the concern about unobserved time-invariant factors that may confound our identification. We find sizable negative effects of air pollution on key fetal developmental indicators measured in the prenatal ultrasound scans, such as head circumference (HC), femur length (FL), and abdominal circumference (AC). Specifically, being exposed to the top 10 percent pollution level for one more day will on average reduce prenatal head circumference by .15 millimeters, abdominal circumference by .05 millimeters, and femur length by .15 millimeters.

Among many paths to a particular birth outcome, some might be healthy while others are not. More generally, early childhood development may be subject to latent or immediate effects, which may not be sensitively reflected in birth outcomes. We show that the immediate impact on HC becomes muted at birth, while the latent impact on AC seems manifested at birth. The impact on FL, however, largely persists over time. All these evidence suggests the urgency to go beyond just using birth outcomes to proxy early life human capital.