Corrected March 18| While health insurance coverage has broadly expanded across the country since implementation of the Affordable Care Act of 2010, many impoverished communities continue to lag behind, according to census data released Thursday.

Overall, the nation’s uninsured population fell to 8.7 percent in 2020, from 15 percent in 2013.

In Texas, however, American Community Survey results released Thursday show the uninsured rate was twice the national average — 17.3 percent, which was 3 percentage points more than the state with the next-highest uninsured rate, Oklahoma.

Texas also had eight of the 10 congressional districts in the country with the highest uninsured populations. All but nine of the state’s 36 congressional districts were in the bottom 10 percent of insurance coverage rates nationwide, according to ACS data for 2020.

The state also housed the congressional district with the highest uninsured rate in the country, the 29th Congressional District. In the Houston district held by Democratic Rep. Sylvia R. Garcia, 31 percent of the population did not have health insurance, the 2020 data found.

Nationwide, the ACS found districts with higher uninsured rates also tended to have lower incomes and be in states where government officials chose against expanding Medicaid coverage under the Affordable Care Act, the 2010 health care overhaul signed into law by President Barack Obama.

ACA expanded Medicaid eligibility to people up to 138 percent of the federal poverty level, but Texas was one of the states that did not elect to broaden eligibility. A 2012 Supreme Court decision, NFIB v. Sebelius, tossed provisions of the law that made the expansion mandatory.

Texas also had the top 10 congressional districts in the country with the highest uninsured rate among people making less than that 138 percent threshold. All but two of the state’s congressional districts were among the top 10 percent of that category.

A 2020 report by the state’s comptroller estimated that more than 650,000 people in the state lost their employer-related insurance during the coronavirus pandemic.

Since the implementation of the Affordable Care Act, the nation’s overall uninsured rate fell from nearly 15 percent in 2013 to 8.7 percent in 2020. According to ACS data that looked at a five-year period ending in 2020, only 12 states had an uninsured rate of higher than 10 percent: Alaska, Arizona, Florida, Georgia, Idaho, Mississippi, Nevada, North Carolina, Oklahoma, South Carolina, Texas and Wyoming.

In similar five-year data ending in 2013, only 10 states had an uninsured rate lower than 10 percent of the population: Connecticut, Delaware, Hawaii, Iowa, Massachusetts, Minnesota, North Dakota, Pennsylvania, Vermont and Wisconsin.

Texas congressional districts also lagged behind the rest of the country in the 2013 ACS data. It had six of the 10 congressional districts with the highest uninsured population in the country. Nearly 40 percent of the population in Texas’ 33rd Congressional District, held by Rep. Marc Veasey, a Democrat, did not have health insurance in 2013.

The uninsured rate in that district dropped to 30.5 percent in 2020, but it remained one of the highest uninsured districts in the country.

Unlike the decennial census, the ACS is a sample survey that relies on data from a smaller number of people to represent the larger population and is released annually. The ACS normally provides data drilled down to geographic areas with as few as 65,000 people.

The data released Thursday reflects a snapshot of surveys taken from 2016 through 2020. The Census Bureau calculates rates based on responses taken over that five-year period. However, the data’s release comes after glitches overcome by the agency, which had to pull back on the one-year version release of the 2020 American Community Survey due to problems with the data.

The data for the 2020 one-year ACS did not meet agency standards, Census officials said, because too many people did not respond because of the pandemic. Instead the agency released an experimental dataset with the one-year data.

After pulling back on the one-year data, the agency changed the way it calculates the ACS results, which was intended to reduce the impact of the problematic 2020 responses.

“[W]e have worked tirelessly over the last few months to refine our methodology and reduce the impact of nonresponse bias in the 2016–2020 ACS 5-year data products,” Donna Daily, the ACS division chief, said in a Census Bureau release Thursday.

The agency created the ACS in 2010 as a replacement for the long-form census questionnaire that used to be sent every decade to a subset of American households. The ACS survey asks about 3 million households a year more extensive questions than the 10-question decennial census, rolling results into one- and five-year data releases.

This report was corrected to accurately reflect the Supreme Court case involved.