Texas A&M Study Points To Four Major Predictors Of COVID-19 Spread

An analysis of New York City neighborhoods showed a significant association between detected COVID-19 cases and dependent children, population density, median household income and race.
By Caitlin Clark, Texas A&M University Division of Marketing & Communications November 11, 2020

New York City skyline
An analysis of the early stages of the pandemic in New York City could inform future pandemic response.

Getty Images


Socioeconomic factors identified in a Texas A&M University study as significant predictors of the spread of the novel coronavirus in New York City provide important insight into public health behaviors in the early stages of current and future pandemics.

Data collected from across New York City between March 1 and April 5 in the early stages of the COVID-19 pandemic found a significant association between detected cases of the disease and dependent children, population density, median household income and race.

The ecological study was conducted by Texas A&M aerospace engineering doctoral student Rich Whittle and Aerospace Engineering Assistant Professor Ana Diaz Artiles. The aim of the study was to identify potential neighborhood-level socioeconomic determinates of the positivity rate and to explain variation between neighborhoods during the beginning of the virus’ spread in New York City.

Data was collected from 177 Zip Code Tabulation Areas in the city. Whittle said their models focused on three sets of socioeconomic parameters – demographics, economic and health – using positive COVID-19 cases as the outcome. The results identified four consistent predictors of COVID-19 cases in neighborhoods:

  • An increase of just 5 percent in the under 18 population (dependent children) is associated with a 2.3 percent increase in the positivity rate;
  • An increase of 10,000 people per square kilometer is associated with a 2.4 percent increase in the positivity rate;
  • A decrease of $10,000 median household income is associated with a 1.6 percent increase in the positivity rate;
  • And with respect to race, a decrease of 10 percent in the white population is associated with a 1.8 percent increase in the positivity rate, while a 10 percent increase in the Black population is associated with a 1.1 percent increase in positivity.

“The study highlights the importance of public health management during and after the current COVID-19 pandemic,” Whittle and Diaz Artiles write in the study. “Further work is warranted to fully understand the mechanisms by which these factors may have affected the positivity rate, either in terms of the true number of cases or access to testing.”

Their work was published in BMC Medicine.

New York City quickly became the epicenter of the COVID-19 pandemic in the United States, with 345 confirmed cases by the time the World Health Organization declared a global pandemic on March 11. With a transmission rate five times higher than the rest of the country, the city’s number of cases grew to just 18,000 two weeks later.

The available dataset used in the study included 64,512 positive cases. The Zip Code Tabulation Areas examined were where patients reported their home address. The researchers note that during the time covered by the study, the New York City Department of Health and Mental Hygiene discouraged testing for people with mild symptoms. As a result, the data likely represent those people with more severe symptoms.

Whittle said other potential predictors covered in the study that were not found to be statistically significant included the 65 and older population, males per 100 females, percentage of unemployed residents, percentage living below the poverty line, percentage of uninsured residents, and the total number of hospital beds per 1,000 people within five kilometers.

The percentage of Hispanic, Asian and other races also were not found to have significant associations with the COVID-19 positivity rate.

Studies of previous pandemics suggest that socioeconomic factors can affect detection rates and medical outcomes, the researchers note in the study, which could account for the difference in reported cases between New York City neighborhoods. Whittle said disease mapping identified several high-risk areas, including Corona, Queens, the East Bronx, and the orthodox Jewish community around Borough Park, Brooklyn.

“If you go back and look at the Spanish flu, there were three waves that hit the U.S., of which the second wave was the worst. So having an earlier understanding of the socioeconomic risk factors allows policymakers to better target where they want to focus their efforts,” Whittle said. “If you understand the landscape, you can allocate your resources better.”

The study uses spatial modeling, which in the context of COVID-19, tries to account for cases that are geographically clustered in certain areas.

“Spatial model techniques capture how a variable of interest might be affected by its surroundings,” Diaz Artiles said. “Thus, when the spatial factor is taken into account, the behavior of that variable can be very different just based on location.”

One surprising finding was the lack of association in positive cases with the percentage of a neighborhood’s population of senior citizens. Whittle said this is likely due to the inclination of the older community to adhere to public health measures after the widely-reported risks of COVID-19 for this age group. At the same time, he said, this could mean that the younger population in New York City decided they were not at risk, leading them to engage in behaviors that led to a higher positivity rate among that age group.

When it comes to race as a predictor, Whittle and Diaz Artiles note the established sociological relationship between race and economic affluence. Black populations are more likely to live in densely populated, low-income neighborhoods, leading to increased contact patterns. Combined with a higher incidence of comorbidities, this could have led to an increase in symptomatic COVID-19 cases among Black residents.

The researchers said that their study has some limitations. The dependent variable was the number of detected cases, which could be much different from the true number. They also note the ecological fallacy of making individual inferences from aggregate data. Too, the significant predictors they found likely aren’t the only explanations for different positivity rates between neighborhoods.

Whittle said further studies are needed to determine underlying causes related to these predictors, paying particular attention to willingness to engage in public health behaviors and asymptomatic carrier transmission. Their methodology could also be applied at the county or state level, he said.

“This is an unprecedented situation in the U.S., so it’s important to get as much understanding as quickly as we can,” Whittle said.

Media contact: Caitlin Clark,

Related Stories

Recent Stories