Census undercount could leave healthcare underfunded
FOR EACH PERSON who went uncounted in the 2010 census, Vermont lost more than $2,300 in 2015 federal grants for Medicaid and the Children’s Health Insurance Program, foster care and adoption assistance, and child care for low-income families. It’s a similar story for 36 other states.
In 2017 alone, census data helped direct $1.1 trillion in Medicaid and CHIP spending, according to the Counting for Dollars initiative at the George Washington University Institute of Public Policy. Another $400 billion was allocated for housing choice vouchers, Temporary Assistance for Needy Families, social services block grants, Head Start, school lunch programs and the Women, Infants and Children nutrition program.
CMS also relies on census data to set formulas for Medicare Part B payments.
More than 300 federal programs use the census to help distribute all or some of their funds, according to George Washington University research professor Andrew Reamer, who leads the Counting for Dollars initiative.
The next decade of funding for these programs will rely heavily on the 2020 census, which is facing unique challenges due to the COVID-19 pandemic and political pressure. Relying on census data can pose some challenges to lower-income or minority communities, which have seen higher rates of COVID-19 and lean more on publicly funded social programs.
The 2010 census didn’t count at least 10 million people and double-counted up to 10 million people, according to a recent article in the New England Journal of Medicine. Even though the overall count seemed correct, the census significantly undercounted people of color and overcounted non-Hispanic whites.
“Demographic changes over the last decade suggest that the population in 2020 will be harder to enumerate. Hardto-count groups—including complex households, renters, young children, immigrants and people of color—will represent a larger share of the population in 2020 than they did in 2010,” according to a 2019 report from the Urban Institute.
According to a 2018 report by Reamer, most states could lose up to $1 in federal Medicaid funds for every $100 spent with a 1% undercount. Some areas and populations are more likely to be undercounted, leading other states and localities to score more funding than they merit while others don’t get enough.
Aug. 26, the self-response rate to the 2020 census was 64.6%, 1.9 percentage points lower than 2010’s final rate.
COVID-19 is making it tougher to count the population accurately because people are less willing or able to talk to census workers. But its impact on the census is uneven because the virus has affected regions and populations differently over time.
Some providers have taken it on themselves to help. Arizona-based Mountain Park Health Center started reaching out to its patients before COVID-19 struck, partnering with a not-for-profit to educate patients about the census when they visited its clinics. The community health center continues to reach out to people through email and social media, but it had to pause in-person events promoting the census and a partnership with Phoenix’s mobile census vans that allow people to fill out a census form on the spot.
It’s a familiar role for Mountain Park, which also registers people to vote and serves as an official polling place during elections, said Janey Pearl Starks, Mountain Park’s director of marketing and engagement, in an email. The organization believes that one way to determine the health of a community is by its level of community engagement and the degree to which community members feel their voices are heard.
“It is important that those in the communities we serve are counted in the census, both for the actual benefits of a community having an accurate count, but also for the simple yet important act
Relying on census data can pose some challenges to lower-income or minority communities, which have seen higher rates of COVID-19 and lean more on publicly funded social programs.
of civic engagement,” she said.
Every person living in the U.S., regardless of citizenship status, is required by law to be counted in the U.S. decennial census. The results decide who wields political power (how congressional seats are apportioned, for example); determine how much federal funding state governments and local communities receive; and inform a wide range of health-related policies and decisions.
Experts worry the pandemic and actions by the Trump administration might lead to a sizable undercount, which could put billions of taxpayer dollars at risk of being misspent, worsen public health and intensify health inequity.
“We can’t ensure a fair distribution of health-promoting resources if we’re using formulas built on the back of invalid, biased data,” said Philip Alberti, senior director of health equity research and policy for the Association of American Medical Colleges. “Making sure we get the census right is a preeminent concern for folks in public health and healthcare.”
The pandemic makes it harder to survey people living in institutional settings like nursing homes and prisons due to restrictions. And the current economic volatility makes it more likely that people will move, adding to the difficulty of tracking people down to complete their surveys.
While Americans can respond to the 2020 census online, a first for the survey, the change could skew the results since older adults, people of color and lower-income populations might be less likely to respond electronically than wealthier, white populations.
The Census Bureau may use statistical techniques to account for biased data samples, but the more inaccurate the data, the harder it is to do that well. It’s also controversial because the Constitution’s enumeration clause requires a real count of all “persons” in the nation.
“It would be without historical precedent,” said Doug Strane, health policy research program manager for PolicyLab at Children’s Hospital of Philadelphia.
The Census Bureau also collects more detailed data through its American Community Survey, which is sent to nearly 300,000 addresses each month to gather information about ancestry, citizenship, educational attainment, income, language, housing characteristics and other information. It’s by far the most comprehensive federal data collection on the social determinants of health, but an undercount would weaken experts’ ability to effectively use the information.
“I’m truly worried that our efforts to understand and address the structural determinants of health are potentially undercut by invalid census counts,” Alberti said. “If we get the population numbers wrong, our public health research and (the) evaluation of our results will become less reliable if we don’t know how we’re doing over time.”
Health systems and insurers could have trouble executing population health management strategies and performing well in risk-based contracts if social safety-net programs are underfunded due to inaccurate census information.
Health systems and insurers need to consider how they will address their populations’ social determinants of health if the 2020 census results in inadequate funding for critical programs during the coming decade, Strane and others said.
Earlier this month, the Census Bureau said that it would stop collecting information Sept. 30, even though in April it extended the deadline from Aug. 15 to Oct. 31. At the time, the bureau said it was worried the pandemic would make it harder to collect accurate information.
The agency said it decided to end data collection early to meet its Dec. 31 deadline to provide Congress with census figures. Experts say the Census Bureau is more likely to undercount people of color, immigrants, low-income people and other marginalized populations if it stops collecting data early because they’re harder to reach.
National Minority Quality Forum CEO Gary Puckrein said it’s a “tremendous disservice to society” when people “monkey around with (census) data for political purposes.”
The census gives public officials, researchers, providers, insurers, suppliers and others crucial data to prepare for and respond to pandemics, natural disasters and other emergencies. The information becomes more useful as it becomes more accurate, which has led the Census Bureau to refine how it conducts the census over time.
Part of the reason it’s difficult for public health experts to collect information about and understand how COVID-19 affects people of different races and ethnicities is that people have trouble identifying themselves through the census because of how the bureau collects information.
For example, the census asks people of Middle Eastern and North African descent to self-identify as white, according to an article in the American Journal of Public Health. The census also makes it tough for Latinx people to clearly describe themselves because it asks multiple related questions instead of one.
Even before the pandemic and recent moves by the Trump administration, there were representation issues with census data. Alberti said it created a “kind of vicious circle” because the Census Bureau consistently overcounts people with higher socio-economic status and undercounts those with lower socio-economic status.
“There’s a lot of blame, particularly on communities of color, for their ‘underlying comorbidities,’ ” he said. “The blame for them actually lies in the fact that we have (an unequal) distribution of re
● sources across our communities.”