Skip to Main Content

Editors’ note (Dec. 12, 2019): The author of the study upon which this story is based has retracted the paper, published last week in Science Advances, citing data classification errors.

Police shootings of unarmed black Americans have escalated a national debate about racial injustice and the use of force. A new study says they could also be harming the health of infants not directly involved in the incidents.


Research published Wednesday in Science Advances finds that black mothers who were exposed to police shootings of unarmed black Americans during pregnancy were more likely to give birth to infants who were born prematurely or with low birth weight. Living within 1 kilometer of such incidents tended to have a greater effect on these indicators of infant health. The effect was also more pronounced among infants born to mothers who had previously had children without exposure to such events.

Black infants are already more likely to be born prematurely and at low birth weights — less than 5.5 pounds. In 2018, for instance, the rate of premature births among African American mothers was about 50% higher than the rate in white mothers. And black women are up to three times more likely to give birth to babies with low birth weight.

Premature birth is linked with other health problems later in life, including developmental delays and hearing problems. And a baby born with a low birth weight can have other complications, including trouble feeding or an increased risk of infection.


Many factors, including socioeconomic status and access to health care, influence the divide between black infants and infants of other races, but the new findings — which only represent an association and not a causal link — could have implications for how public health experts consider closing the developmental gap between black infants and infants of other races.

STAT spoke with Joscha Legewie, a sociologist at Harvard University and the author of the study, to learn more.

The interview has been lightly edited and condensed.

Why did you decide to look into this?

I’ve been working on the social consequences and spillover effects of police activity on minority youth, and that includes research on how policing affects education and health. Last year, the New York Times Magazine had an article on the stark racial disparities in infant health. That got me thinking more on long-term consequences of aggressive policing and how police violence might influence infant health.

What kind of data did you use? And why the focus on California?

I used population-level data on birth records from California between 2007 to 2016. It also included geographic information on where the mother lived at the time of birth. The other data component was police killings over the same time period. We started our own data collection process several years ago that covers fatal encounters. Combining these two data sources, I was able to compare women who are exposed to unarmed killings during pregnancy versus those who aren’t.

[The focus on California] was largely driven by the fact that California provides a standardized process for going through this data. And it has guidelines for protecting patient privacy. The process allows researchers like me to apply [to use this] data for research purposes. Such data is collected across the U.S., but California has a standardized process with considerations for privacy.

Optimally, I would like to study this on a national scale, but there are tradeoffs. There is [federal] information on birth records, but often doesn’t include the same, detailed information on geographic location of mothers’ residence. But California is a very large state, which makes it a good case to study. Compared to smaller states, the number of police killings there are higher simply because it is a larger state. At the same time, California has a smaller black population than other states, so it would be interesting to study this in states with larger black populations.

What did you find?

There’s a substantial effect of exposure to police killings during the first and second trimester on [low birth weight and gestational age]. This correlational link is very specific to police killings of unarmed black [people]. I did not find a similar pattern or similar effects on the gestational age and birth weight in infants of other races or ethnic groups. I didn’t find a similar effect on armed black Americans. And I did not find a similar effect of these killings on white or Hispanic Americans.

What do you think might be the underlying reason for these findings?

This very race-specific pattern helps me to, at least partially, rule out possible explanations, namely that this is driven by crime. If crime and violence were driving this effect, we would expect a larger effect from [exposure] to police killings of armed individuals.

So the fact that this effect is concentrated among police killings of unarmed black individuals tells me that the stress and the perception of injustice is the most plausible explanation of this. That’s something that I’m not measuring directly, so future research can explore more directly the level of stress and how these police killings contribute to or are an important stressor.

How do you separate this from what we already know about the gap between black infants and infants of other races?

The way I see it is that there are these stark racial and ethnic disparities, stark black-white gaps in birth weight and gestational age and other indicators of infant health. There’s an increasing tension to the role of experiences of systematic discrimination during pregnancy that contribute to these racial disparities. But very few studies have linked this to very particular incidences, such as experience of a police killing in your residential environment. In that sense, I think this work is very much part of the increasing research that looks at the role of experiences of discrimination, and advances it by tying it to very particular experiences that happen in your neighborhood.

Are there policy takeaways?

There aren’t concrete policy steps, but this encourages us to think more seriously about exposure to these types of events as contributing to racial/ethnic disparities. When we think about ways of closing the gap [for] infant weight and gestational age between black and white infants, we have to take very seriously what is happening in parents’ neighborhoods. It can start the thinking process about what’s happening in their residential surroundings and how that contributes to the health of the mother.

What are your next research steps?

One aspect I’m very interested in is further looking at what type of police killings are more consequential. [This study] looked at armed versus unarmed. But I’m very interested in looking at the extent to which incidents receive a lot of media coverage, whether they’re extensively covered on social media, and whether incidents that receive larger attention in this way have a larger effect on infant health.