Blacks, Hispanics and Asians are still dying from covid-19 at higher rates than Whites


Dennis Bannister’s daughter, Demi, was the first to die.

She was only 28, a beloved third-grade teacher who likely caught the virus during a training at her Columbia, S.C., school district. Doctors diagnosed her with a bladder infection, and by the time they realized their mistake, it was too late. Not long after, the family’s matriarch, Shirley, 57, complained of difficulty breathing. She was twice sent home from the emergency room before returning by ambulance and being put on a ventilator. She died soon after.

Which left Dennis Bannister, childless and a widower, sitting on his porch last month, staring at the last of the green leaves and mourning. Why, he pondered, had the virus hit his family so hard, and not just them, but so many African Americans? Was there something that made them particularly vulnerable? Had they gotten the right care?

“Folks think maybe they saw an African American coming in, and they didn’t take them seriously,” reflected Bannister, who was also infected but asymptomatic. “I don’t know. I just pray God will help me find a way to deal with the situation.”

It’s not just grieving relatives who are demanding answers. Nearly nine months after the virus exploded in the United States, and amid big treatment strides, the disease continues to ravage African American and other minority communities with a particular vengeance. Black, Asian, Native American and Hispanic patients still die far more frequently than White patients, even as death rates have plummeted for all races and age groups, according to a Washington Post analysis of records from 5.8 million people who tested positive for the virus from early March through mid-October.

Image without caption
Image without caption

Dennis Bannister, seen at home in Columbia, S.C., lost his daughter, Demi, and wife, Shirley — seen with him — 20 days apart. (Photos by Sean Rayford for The Washington Post)

Death rates overall have fallen more than 80 percent from the pandemic’s peak in the spring, when refrigerator trucks were parked outside New York City hospitals and ice rinks were converted into morgues, according to an analysis of anonymized data collected by the Centers for Disease Control and Prevention.

But as another wave of infections sweeps across the country this fall, losses among racial and ethnic minorities remain disproportionately large. Black Americans were 37 percent more likely to die than Whites, after controlling for age, sex and mortality rates over time. Asians were 53 percent more likely to die; Native Americans and Alaskan Natives, 26 percent more likely to die; Hispanics, 16 percent more likely to die. Those higher case fatality rates for diagnosed people of color are on top of the increased infection rates for those unable to isolate at home because they are essential workers.

These patterns have devastated communities of color across the country: multigenerational Latino households in Texas, Pacific Islander families in Washington state, African American families in South Carolina.

Advocacy groups, researchers and other experts say many of these deaths are preventable, and they blame federal, state and local leaders for failing to take the disparities seriously and take steps to address them.

The shortage of testing in communities of color, which made headlines in the beginning, persists to this day. Despite their symptoms, for instance, neither Demi Bannister, nor her mother, Shirley, were tested for the virus until they were close to death in late August and September.

Critics also point to spotty race data, which has made disparities harder to identify and solve; weak enforcement of protocols like mask-wearing and social distancing at essential workplaces; delays in translating critical health alerts into other languages; conflicting guidance from health agencies that deepened distrust in some communities; economic and cultural factors that lead more families to live in multigenerational homes; and immigration policies that exacerbate crowded housing and discourage people from seeking medical care.

“It is a perfect storm that has been created that led to the deaths of groups of people,” said Elena Rios, head of the National Hispanic Medical Association.

Juliet Choi, chief executive of the Asian & Pacific Islander American Health Forum, said many of the measures sought by minority groups to mitigate the effects of the virus on their communities are easy to implement and inexpensive, but they have been mostly ignored.

“It comes down to political will and commitment,” Choi said. “We’re not asking that decades of systemic barriers get eliminated overnight, but there are many simple things we should be doing that we are not doing.”

There is growing evidence that such changes do make a difference. Faced with extreme disparities in covid-19 deaths, Michigan officials undertook a series of steps, from boosting testing to connecting people of color with primary care doctors. The state’s rapid progress proves the issues are neither intractable, nor rooted somehow in biology.

Death rate for men, 60s

[U.S. coronavirus cases and state maps: Tracking cases, deaths]

Garlin Gilchrist II, a Detroit native as well as the state’s lieutenant governor, formed one of the nation’s first state racial disparities task forces on covid-19 back in April. Made up of 23 community organizers, doctors and other experts, the group focused not only on boosting testing and contact tracing, but also tailoring messages on mask-wearing and other public health precautions to African American communities. It also addressed broader systemic issues, such as access to primary care, and helping those in rural areas access telemedicine.

When state epidemiologists ran the numbers again in September, they found a huge change: Black residents who in April accounted for 29.4 percent of cases and 40.7 percent of deaths now made up only 8 percent of cases and 10 percent of deaths — very similar to their percentage in the population.

Gilchrist emphasized the state’s efforts have not been complicated. “I think the reason we have been able to make progress is we chose to focus on it,” he said.

In April, Lt. Gov. Garlin Gilchrist II (D-Mich.) created a task force on racial disparities in the pandemic. Six months later, Black Michigan residents accounted for only 8 percent of the state's coronavirus caseload, a sharp decrease from April, when they were nearly 30 percent of all cases. (Don Campbell/Herald-Palladium/Associated Press)
In April, Lt. Gov. Garlin Gilchrist II (D-Mich.) created a task force on racial disparities in the pandemic. Six months later, Black Michigan residents accounted for only 8 percent of the state’s coronavirus caseload, a sharp decrease from April, when they were nearly 30 percent of all cases. (Don Campbell/Herald-Palladium/Associated Press)
Theresa McGarity of Michigan looks at an art installation representing Macomb County residents who died in the pandemic. She lost her mother to covid-19. (David Goldman/Associated Press)
Theresa McGarity of Michigan looks at an art installation representing Macomb County residents who died in the pandemic. She lost her mother to covid-19. (David Goldman/Associated Press)

Holes in the data

When epidemiologist Peter J. Fos first saw the coronavirus case data from his own state of Louisiana in the spring, he recalls being baffled. Respiratory diseases, like the flu, tend to move randomly through a population. “Rich, poor, male, female,” he said. “Usually, everyone is at risk together.”

Coronavirus cases, however, he noticed, seemed to be clustered in mostly poor Black neighborhoods. He called up friends in Mississippi for that state’s data and found similar patterns. The same thing was true in Michigan.

“We don’t think of infectious diseases as being a health disparity,” he recalled, “but here it was in front of me.”

Comparable data for many other states simply did not exist. Then and now, testing and data collection came under the aegis of the White House — which congressional Democrats, public health experts and civil rights leaders have criticized as being slow to respond to the disparities.

Public health experts say that good data is the cornerstone of an effective pandemic response. Without it, health officials and others fly blind. But U.S. data on the pandemic, especially about race and ethnicity, has been incomplete, or lacking altogether, in part because of the different ways state and local jurisdictions report it, and the federal government’s delay in mandating it.

Death rate for women, ages 0 to 49

Native American advocacy groups, for instance, have expressed concern about the exclusion of some of their communities from analyses, and inconsistencies in data from tribes, as compared with what states and local jurisdictions are collecting.

The CDC said in a statement that it is “working hard to address the threat of covid-19 among racial and ethnic minority groups” by expanding testing for at-risk groups, increasing engagement with “trusted racial and ethnic minority servicing organizations” including Black colleges and universities and churches, and working to get better data.

Yet some early efforts to address inequities were quashed: The CDC had internal discussions for example, about ways to increase testing in African American and other hard-hit communities. On May 3, a document posted on its website recommended that people from “racial and ethnic minority groups disproportionately affected by adverse covid-19 outcomes” get priority for testing. Three days later, the language was removed.

In response to questions from The Post, the CDC’s press office said the document was “a draft posted without proper vetting through all CDC channels.” An official speaking on the condition of anonymity because they were not authorized to discuss the issue said it was actually higher-ups at the Department of Health and Human Services and the White House who had asked for its removal. No follow-up guidance was ever issued.

Leon McDougle, a family medicine specialist at Ohio State University who heads the National Medical Association, the country’s oldest Black physicians group, said he had been optimistic the Trump administration would take steps to address the disparities. But after that document disappeared without any public explanation, he began to have doubts.

“I have real concerns about the … political influence that appears to play a factor in decisions,” he said.

Death rate for men, ages 0 to 49

In early June, CDC Director Robert Redfield apologized to lawmakers for “the inadequacy of our response” in documenting the nation’s high rates of infection and deaths among Black and Hispanic Americans. That same day, Adm. Brett Giroir, an HHS assistant secretary running the government’s coronavirus testing response, said that as of Aug. 1, the federal government would begin requiring laboratories to report racial and ethnic information for people who were tested for the virus.

The national coronavirus database available today is much more robust, but holes remain. Some records are missing race, gender or age, while others lack information on the patient’s other medical conditions, or even whether the person lived or died. Information is more likely to be blank for non-Whites. Among hospitalized patients, for instance, there was no information on whether the patient survived for 36 percent of Hispanics, 29 percent of Blacks, 26 percent of Asians and 24 percent of Whites.

The Post analyzed the data in numerous ways: with and without the missing information, with all infected patients, and with only patients who had been hospitalized. No matter which model it used, the results showed significantly higher death rates for Blacks, Hispanics, Native Americans and Asians than for Whites.

While minorities tend to get infected at higher rates, some studies have suggested that delays in diagnosis and their experiences in hospitals appear to contribute to poorer outcomes. A Boston-based data firm said African Americans with symptoms such as coughing may be less likely to get access to scarce coronavirus tests. An academic analysis of testing in New York City found that more testing existed in White neighborhoods although the highest positivity rates were in communities with a higher share of Black and Hispanic residents.

In the spring, scholars puzzling over such disparities overlaid maps of covid-19 deaths with maps of communities where heart disease, diabetes, obesity and other conditions were highest in the United States. They found that the hot zones matched up.

[These are the top coronavirus vaccines to watch]

Numerous studies have since borne out the relationship between these health conditions and high covid-19 death rates. When looking at the role of race, smaller studies — based on data from single hospital systems or regions — found that differences in mortality narrowed when controlling for ailments such as high blood pressure, diabetes, lung conditions and obesity.

In the CDC data, however, the difference in death rates grows even larger when controlled for those other health conditions — although information is missing for many patients.

Fos, a member of Louisiana’s task force on race and covid-19, argues the initial observation that high covid-19 death rates coincided with hot spots for other health problems led some public health officials to conclude the unequal burden was somehow inevitable. A few went so far as to suggest that those most affected were somehow to blame for their illnesses.

Rep. Maxine Waters (D-Calif.) called out U.S. Surgeon General Jerome Adams in April, for instance, for singling out African Americans with advice to avoid alcohol, tobacco, and drugs to lower their risk for severe disease from the virus. Although Adams had noted his own health struggles as part of “that legacy of growing up poor and Black in America,” Waters described the comments as “a backhanded attack on African Americans and communities of color.” Adams subsequently apologized, saying he “used the language that is used in my family” and that the comments were not meant to be offensive.

Death rate for women, 70s

Fos and other experts argue the clustering of covid-19 cases is more of a social and economic phenomenon.

He noted the high proportion of people in such communities who interact daily with the public as essential workers, who often live in more densely packed neighborhoods and multigenerational households, and who might not have regular doctors because they lack health insurance.

In addition, minorities face a long history of unequal access to medical care — which may have impacted treatment decisions and outcomes. A study using data from the Society for Critical Care Medicine, posted this month before peer review, found that African American patients were more likely than Whites to receive an older, less-expensive and riskier blood thinner linked to higher mortality from covid-19. Blood thinners have become a critical weapon in the arsenal used by doctors against the disease because many patients with severe disease develop clots.

It was not clear whether the administration of that medicine related to insurance coverage, physician preference, or something else. One of the study’s authors, Venky Soundararajan, chief science officer of data firm Nference, wondered whether some doctors chose the older, more established product for minority patients because the newer drugs were overwhelmingly tested on Whites.

“If you distill it down to the root causes,” Fos said, “they are ones we have known in our country for years and we’ve just done a very bad job of addressing it. It’s some of the same reasons people are protesting in the streets — police brutality, job discrimination, environmental justice. The coronavirus shows how much racism there is in health as well.”

Hidalgo County, Tex.

If New York City was the epicenter of the first wave of the pandemic, the stretch of Texas along the Rio Grande Valley near the U.S.-Mexico border is the epicenter of the second.

As of this week, four counties in this region had tallied a combined total of 70,000 cases and more than 3,300 deaths. Norma Ramirez, the Democratic chair for Hidalgo County, has said there is not one person in her region “who hasn’t been affected by this horrible virus.”

Vanessa Alvarado, outreach coordinator for LUPE, a community group in Texas founded by labor activists César Chávez and Dolores Huerta, said immigration policy and economic challenges have exacerbated the Latino communities’ vulnerability. A Trump administration rule that took effect Feb. 24 makes it more difficult for people to get green cards if they seek government help such as food stamps and some types of health and housing assistance.

That has left many fearful to seek care even if they are very sick, she said. And as layoffs have accelerated and people have sought cheap living arrangements, it has left families and friends crammed into small homes with no space for a “sick room” to isolate patients. About 27 percent of Hispanic households span three or more generations, according to a Pew Research report. That compares with 16 percent of White households.

In addition, Alvarado said, so many are struggling to make enough money for basic needs like food and shelter that the coronavirus has sometimes seemed a secondary concern. Even the most vulnerable in her community — men and women over the age of 60 with known health issues — have to go to work to make ends meet. In addition to everyday expenses, some families are now being hit with huge bills for funerals.

“’Whatever God wants. We leave it in God’s hands,’” Alvarado said. “That’s the general attitude because that’s all you have left.”

Alvarado and other community activists said it’s common to see the virus strike whole households in the Latino community.

Raquel Chavez, who is originally from Mexico, lost her father and a brother to the virus in September. It had started with her mother’s stomachache. After her mother arrived at the hospital and they took her vital signs, she was wheeled into intensive care and put on a ventilator. She tested positive for the coronavirus. Then her father came down with flu-like symptoms. His doctor sent him home with some cold medicine, and as a precaution, tested him for the coronavirus.

“They couldn’t give him extra attention because his test hadn’t come back,” she recalled. “But he didn’t last very long.”

Image without caption
Image without caption
Image without caption

Raquel Chavez lost her father, Ángel Chavez, and a brother, Juan Francisco Chavez, to the coronavirus. Other relatives have fallen ill, too. The family lives in Texas’s Rio Grande Valley, hit especially hard by the virus. (Photos by Brenda Bazán for The Washington Post)

Ángel Chavez, 69, died a few days later of a heart attack. At the funeral, his youngest son had such a bad coughing fit, he was unable to deliver the eulogy. Rushed to the hospital and put on oxygen, Juan Francisco Chavez died soon after at the age of 42.

Her father’s positive coronavirus test did not come back until after his body had been cremated, Raquel Chavez said. Numerous other relatives also were infected by then — her older brother, a nephew, two cousins — though everyone had worn masks at the funeral. She said she has been struck by how the virus can spread faster than test results can be processed.

“My mother cannot overcome what happened,” she said. “It was so quick.”

Spokane, Wash.

In Spokane, Wash., Jeffrey Yoshikawa said much of the information he has about the coronavirus comes from a Facebook group where younger, bilingual people translate critical guidelines and alerts into his native language, Marshallese, for their families. One of roughly 3,000 Pacific Islanders in his county — most from the Marshall Islands in the vast expanse of ocean between Hawaii and Southeast Asia — Yoshikawa said this online group has been critical to his community’s struggle against the virus.

While the Marshallese, who are allowed to live and work in the U.S. but who are not citizens, make up less than 1 percent of the county’s population, they represent an eye-popping 30 percent of covid-19 cases. The community has been vulnerable, he said, not only because of language barriers, but because many work outside the home in essential jobs and are not eligible for most federal health programs. The Marshallese also have startlingly high rates of cancer, likely linked to the 67 nuclear tests the United States conducted on or over the volcanic island chain between 1946 and 1958.

Despite a Federal Emergency Management Agency playbook that emphasizes the importance of communications during a national emergency — it recommends alerts and other key guidance be translated into 17 languages to reach as much of the United States population as possible — the Marshallese and many immigrant communities still rely on their young to tell them what they need to know.

When the CDC began posting information on the coronavirus in January and February, it was in English only. It wasn’t until March 11 that sites in Spanish, Chinese, Korean and Vietnamese appeared. Translations of the covid-19 content into other languages, such as Marshallese, are still unavailable.

Jeffrey Yoshikawa is a member of the Marshallese community in Spokane, Wash., which makes up 1 percent of the county's population but 30 percent of the county's coronavirus case count. (Rajah Bose for The Washington Post)
Jeffrey Yoshikawa is a member of the Marshallese community in Spokane, Wash., which makes up 1 percent of the county’s population but 30 percent of the county’s coronavirus case count. (Rajah Bose for The Washington Post)

Frustrated members of minority groups said many communities created their own communication channels — through cultural organizations, church networks, and other forums — to share health information. Nationally, the Asian & Pacific Islander American Health Forum set up a website to collect translations but access to timely information is hit or miss.

In the Marshallese community, Yoshikawa explained, “we have a few who work in the hospital and schools.”

“They are not doctors, but they can talk to the doctors and they can translate and get the information to the households so they can understand,” he said.

While it has been impossible to get a comprehensive picture of Asian Americans and the coronavirus because of the lack of detailed data, reports from Hawaii, San Francisco and Washington state suggest Pacific Islanders, especially the Marshallese, are being impacted disproportionately among the different Asian subgroups.

Yoshikawa, who works in customer service at Walmart, and his mother-in-law had such serious covid cases they were hospitalized.

“It was the worst experience of my life,” he said, describing how he struggled to breathe as doctors pumped experimental medicines into him to try to save his life. “I had to fight it and get myself strong every day.”

At least three in his community have died, he said; more Marshallese have passed away in other states, including Arkansas and Hawaii.

New Orleans

The struggle to combat the virus has focused on different issues in Black communities.

Reynold Verret, president of Xavier University of Louisiana, and Walter M. Kimbrough, president of Dillard University, set off waves of surprise and indignation in September when they announced they had volunteered as test subjects for an experimental vaccine for the coronavirus. They also appealed to students, faculty, staff and alumni of the historically Black schools to consider the same.

The blowback was immediate and loud.

“You are putting our children at risk with these fast-track vaccines. They are not lab rats!!!” one parent fired back on social media. Another commented, “How dare you suggest they use our community to experiment on.”

But Verret, a scientist with a background in immunology, said in an interview that it is important for him to participate precisely because he is a 66-year-old African American man. Diverse participation, he said, is critical to ensuring that vaccine testing is rigorous and fair so that all populations can benefit.

“The communities we serve look to us as examples,” Verret said.

Reynold Verret, president of Xavier University of Louisiana, volunteered for a test of an experimental coronavirus vaccine. He has a background in immunology. (Irving Johnson III/Xavier University of Louisiana)
Reynold Verret, president of Xavier University of Louisiana, volunteered for a test of an experimental coronavirus vaccine. He has a background in immunology. (Irving Johnson III/Xavier University of Louisiana) (Irving Johnson III/Xavier University of Louisiana)

The men’s decision to share their participation in the trials is part of an effort among African American leaders to address the mistrust they worry puts communities of color in greater danger. That mistrust is grounded in historical abuses such as the Tuskegee syphilis trials but has been exacerbated by missteps in the pandemic response.

As part of that endeavor, Gilchrist, Michigan’s lieutenant governor, tweeted about his recent flu shot. And the Rev. Jesse Jackson said at a public celebration of his 79th birthday last month in Chicago that his wish was that people of color protect themselves from the virus by following public health protocols.

On Sept. 21, the NMA, the Black physicians group, announced it would convene its own panel of experts outside of federal health agencies, to vet data on clinical trials related to vaccines —- essentially a parallel effort to address what it says are gaps in the government’s response and to counter doubts in African American communities about whether the vaccines will be safe and effective.

“Trust remains a huge problem,” said Terri Laws, an assistant professor at the University of Michigan at Dearborn whose research focuses on health and African American studies. “But we’re not where we were 100 years ago. Black professionals are stepping up to fill the gap.”

Columbia, S.C.

Dennis Bannister, his wife Shirley and their daughter Demetria — Demi — had been extremely cautious about the coronavirus. Shirley, chair of the nursing department at Midlands Technical College, had tried to work from home most days. Demi diligently sprayed the doorknobs with disinfectant whenever they came back from an outing and was constantly reminding everyone to be careful.

Halfway through a master’s program in education, Demi had been trying to save money for her own place while living with her parents. Dennis, who had retired from his job as an electrical engineer at a local factory, was delighted to have her.

Demi Bannister, 28, was initially diagnosed with a bladder infection before the coronavirus was confirmed. She died Sept. 7, and her mother died less than three weeks later. (Family photo)
Demi Bannister, 28, was initially diagnosed with a bladder infection before the coronavirus was confirmed. She died Sept. 7, and her mother died less than three weeks later. (Family photo)

She was their only child, and the three of them were close. Demi had been a math and science whiz with a penchant for the performing arts and had talked about becoming a doctor or actor. Her father was surprised when she told him she wanted to be a teacher. But when he volunteered to teach her third-grade students chess, he saw how much she loved it.

“When it was Black History Month, she put this choir together and it seemed like every child wanted to be in the choir,” he recalled. “It gave her joy, and she gave joy.”

Demi came down with a fever shortly after an in-person meeting before Labor Day with other teachers and staff. At her family medical practice, they guessed she had a bladder infection and gave her some medication. They did not test her for the coronavirus, Dennis said. A few days later, she was admitted to the hospital, and her infection with the coronavirus was confirmed. She died Sept. 7 without ever seeing her parents again.

Shirley, who had a history of diabetes and asthma, became ill shortly after her daughter’s death. Dennis said she had been despondent and told him, “All I worked for is nothing. My baby’s gone.”

“She was a fighter, a strong woman, my wife,” Dennis, who turned 65 this month, recalled. “I don’t know if she gave up hope or what.”

Shirley went to the emergency room twice because she was having trouble breathing. The first time, he said, doctors declined to admit her.

The second time, a few days later, Shirley insisted on an X-ray, which found pneumonia in her lungs. To his surprise, Dennis Bannister said, they sent her home again without performing a coronavirus test. “All that delay time. She was getting sicker,” he recalled. A few days later, he had to call 911 when her breathing deteriorated further. Bannister was able to see her in the hospital briefly before she was intubated.

She died Sept. 27.

Lena H. Sun, Chris Alcantara and Magda Jean-Louis contributed to this report.


The Post used a multivariate logistic regression model to test for survival or death of diagnosed covid-19 patients. Factors in the model were date, patient age group, gender, combined race and ethnicity, and presence of relevant preexisting conditions (co-morbidities). Each of those factors was found to contribute to the model beyond the 95 percent confidence limit. Odds ratios were also calculated comparing change over time, older age groups to the youngest age group, men to women, and other race-ethnicity groups to White non-Hispanics.

The limitation of the study was missing data on some of the 5.8 million covid-19 patients collected by local health departments and assembled by the Centers for Disease Control and Prevention. To address that, models were tested including and omitting co-morbidities, the factor that was most often missing. Because information tended to be more complete on the sickest patients, models were also tested that assumed missing information on patient survival meant that the patient had not died, that is, a missing value was treated as “alive” in the classification of “dead” or “alive.” Because limits on testing impacted the denominator of how many patients were confirmed to have covid-19, models were also used for hospitalized patients rather than all patients.

For combined race ethnicity, all White, Black and Asian patients are non-Hispanic. Hispanic patients can be of any race. Patients are placed exclusively in one group or the other. There is no double-counting.

This shows the parameters and results for the primary analysis model and supplemental models that tested for issues with missing data. All findings are significant beyond p<.0001 except for Native Americans/Alaskan natives, which are always p<.005.

Primary model

  • Treat missing survival as patient lived
  • Co-morbidities not included
  • Exclude cases missing date, age group, gender or race-ethnicity

Alaskan/Native American


Primary model including co-morbidities

  • Treat missing survival as patient lived
  • Co-morbidities included in model
  • Exclude cases missing date, age group, gender, co-morbidities or race-ethnicity

Native American/Alaskan


Model with unchanged survival information

  • No modification for missing survival, exclude cases with missing information
  • Co-morbidities not included
  • Exclude cases missing date, age group, gender, survival information or race-ethnicity

Native American/Alaskan


Model with unchanged survival information with co-morbidities

  • No modification for missing survival, exclude cases with missing information
  • Co-morbidities included
  • Exclude cases missing date, age group, gender, co-morbidities, survival information or race-ethnicity

Native American/Alaskan


Primary model for hospitalized patients

  • Include patients who were hospitalized
  • Treat missing survival as patient lived
  • Co-morbidities not included
  • Exclude cases missing date, age group, gender, race-ethnicity or hospitalization

Native American/Alaskan




Leave a comment