As a state, Texas’ infant mortality rate is below the national average, at 5.7 infant deaths per 1,000 live births. But within the state, some communities experience much higher rates, with stark differences between ZIP codes sitting only a few miles apart.

In a new study from the University of Texas System Office of Health Affairs and UT Health Northeast, researchers found “substantial variation” in infant mortality across ZIP codes, with some ZIP codes studied experiencing no infant deaths over four years, while others reported more than 1 percent of their newborns dying before their first birthdays.

The analysis is based on vital statistics data from the Texas Department of State Health Services between 2011 and 2014. Infant mortality rates were calculated for ZIP codes with at least 400 births during the study period — ultimately 34 percent of the state’s ZIP codes were included in the study, accounting for 93 percent of all Texas births in those four years.

Eileen Nehme, co-author of the report and an assistant professor at UT System Population Health and UT Health Northeast, said the study was conducted as part of larger efforts to produce community-level health indicators to help drive local decision-making and more precisely target public health interventions. She noted that a number of health systems and agencies — such as public health departments seeking national accreditation — have to conduct community health needs assessments and so the demand for localized data has only grown.

“Usually, the most granular you can get is county or city level,” Nehme told me. “So we wanted to contribute to both the public health practice world as well as help researchers looking into place-based disparities. Infant mortality, in particular, is a really important population health indicator.”

The study found that within Texas cities, infant mortality can vary significantly. For example, in Fort Worth, infant mortality was more than six times higher in the 76164 ZIP code, at 12.3 infant deaths per 1,000 deaths, than in the neighboring ZIP code of 76107, at 1.8 deaths per 1,000 births. Fifteen ZIP codes studied didn’t experience a single infant death in the four years studied, while 59 documented an infant mortality rate of more than 1 percent.

Racial and ethnic disparities in infant mortality varied by geography, though all racial and ethnic groups had ZIP codes with both very low infant mortality rates and very high ones, sometimes at more than 10 deaths per 1,000 births. Black mothers experienced the highest rates of infant mortality in Texas — as they do nationally, with black women experiencing twice the rate as their white counterparts — and face the greatest geographic variation. White women experienced a relatively lower risk of infant mortality statewide, but had some big spikes at the city level. In the east Texas city of Longview, for example, more than one ZIP code had a rate among white women that was twice the state average. Hispanic women have the lowest infant mortality rate, but they also experienced geographic pockets of particularly high death rates.

Nehme said it wasn’t too surprising to find such pockets of high infant mortality at such a granular level, but the variations were fairly striking. She pointed to the finding that within Houston, black infant mortality rates can vary a whopping eight-fold, from a low of 3.3 deaths per 1,000 births to more than 28 deaths only a few miles away.

“I don’t think we were surprised because we already know there’s a lot of racial and ethnic variation at the country level, so it’s not surprising that we’d see it even more so at a more granular level,” Nehme said. “But it does unmask the idea that we’re doing okay on infant mortality rates — there’s a variation that’s pretty stark.”

Nehme and co-authors Dorothy Mandell, Daniel Oppenheimer, Nagla Elerian and David Lakey write:

Zip codes with both low and high rates of infant mortality can be found for all groups. Thus, race/ethnicity cannot be the sole explanation for a community’s high or low rates.

There is substantial evidence that social, environmental, and economic factors at the community level partially explain this variation. Socio-economic status of the community, income inequality, and air pollution are three community-level factors that have been found to be related to infant mortality. Expanding this existing research to understand how these factors are playing out across communities in Texas will be an important next step.

That’s exactly what Nehme and her colleagues are doing — she said they’re now beginning work to understand what’s driving such stark variations by ZIP code. In addition to the social and economic factors contributing to infant mortality, Nehme suspects contributors such as smoking, obesity, access to family planning and pregnancy spacing.

She hopes local public health workers can use the new data to help direct their infant mortality interventions.

“We’re trying to straddle that place between practice and research — the more that can happen, the bigger public health impact I think we can have,” Nehme said. “We could be doing better for all of our infants and we need to understand why.”

In 2015, the U.S. infant mortality rate was 5.9 deaths per 1,000 live births, a higher rate than many other high-income nations.

Access a copy of the new study, “Infant Mortality in Communities Across Texas,” as well as an interactive map of findings here.