COVID-19 outbreaks have affected restaurants throughout Los Angeles County, from a Panda Express in Sun Valley to the University of California’s Bruin Cafe. If you live in Los Angeles, you can access health department reports about these outbreaks online.

But in most of the country, diners are left in the dark about which restaurants have been linked to outbreaks of the virus.

Restaurants appear to be among the most common places to get infected with the COVID-19 virus, but contact tracing in most areas has been so lackluster that few health departments have been able to link disease clusters to in-person dining.

When KHN contacted the health departments serving the 25 most populous counties in the U.S., only nine could confirm they were collecting and reporting data on potential links between restaurants and COVID cases.

As of Monday, 13 of the 25 counties hadn’t announced changes to their indoor restaurant dining policies, despite record-setting numbers of new COVID infections in the U.S.

While public health researchers are convinced indoor dining is a risky activity in areas where COVID-19 is spreading, getting solid data to justify restaurant restrictions has been difficult. It takes in-depth, resource-heavy disease investigations to determine where people were exposed to the coronavirus, and those contact-tracing efforts have never gotten off the ground in most of the country.

This has made it hard to develop more specific information about risky restaurants and bars, and may have contributed to an overall feeling of powerlessness in the face of the pandemic among people and officials.

It didn’t have to be this way, said Dr. Bill Miller, a senior associate dean of research at the Ohio State University College of Public Health.

“We’ve really missed an opportunity” to use contact tracing systematically to provide “useful information to give us ideas of where we might need to be intervening,” he said.

For contact tracing of other infectious diseases, such as HIV/AIDS, investigators usually ask patients to think through all the contacts with whom they might have shared a virus. They also dive further into the past to try to determine who might have infected the person in the first place.

But U.S. contact tracing for COVID-19 hasn’t taken this approach, in part because of a lack of resources and public trust. Contact-tracing departments are stretched thin, gathering minimum data and facing a suspicious and often uncooperative population.

Contact tracers in Maricopa County, Arizona, prioritize learning the names of individuals over the locations where the coronavirus may be spreading. With the exception of long-term care facilities and a few other locations, investigators don’t consider something an outbreak until they can trace 10 potential cases to a location, said Ron Coleman, a county spokesperson.

As winter looms and people increasingly gather indoors, many local governments are flying blind, lacking the data to create and adjust COVID restriction policies that could make a meaningful dent in rising case rates.

“Imagine there’s some major sporting event,” Miller said. “You might miss an entire cluster that came out of a social situation” if you didn’t check whether, for example, a COVID-positive person had gone to a crowded bar to watch it.

The COVID virus spreads mainly through respiratory droplets that an infected person can release by sneezing, coughing or talking, and a restaurant meal combines several high-risk activities in a single setting: going maskless to eat and drink, meeting up with people outside your household “bubble,” and chatting over a leisurely meal. If the meal takes place indoors, poor ventilation aggravates these risks because of the virus’s potential to linger in still air.

Published research on the role restaurants play in the pandemic is highly suggestive. Taken altogether, the studies paint a scary picture of how potent restaurants can be in spreading COVID-19.

A Centers for Disease Control and Prevention study across 10 states found that those who had tested positive for COVID-19 were more than twice as likely to say they had dined at a restaurant in the two weeks before their illness began, compared with those who tested negative. Dining at a restaurant was the only activity that differed significantly between those who tested positive and those who tested negative for the coronavirus.

For example, that study seemed to show no increased risk of infection linked to shopping, gathering with 10 or fewer people or spending time in an office, said Kiva Fisher, a CDC epidemiologist and lead author of the study.

Not surprisingly, restaurant restrictions appear effective at slowing viral spread in a community. Out of the many social distancing restrictions states chose to implement at the beginning of the pandemic, shutting down restaurants had the strongest correlation to reducing the spread of the disease, according to researchers at the University of Vermont.

A recent Stanford University-led study that used mobile phone data from different cities to create a simulation of viral spread suggests that restaurants operating at full capacity spread four times as many additional COVID-19 infections as the next-worst location, indoor gyms.

The model predicts that only about 10% of “points of interest” — public places where people gather — account for over 80% of infections that occurred in public places, said Jure Leskovec of Stanford University, lead author of the mobile phone data study.

“There are a small number of these superspreader sites that account for a large majority of infections,” he said. One characteristic of superspreader sites is that “people are packed and stay there a long time.”

Still, none of these studies can definitively prove that restaurant dining causes infections, the researchers said. Identifying any individual restaurant case or cluster requires the kind of shoe-leather investigation that few communities in the U.S. have been able to conduct.

“You’d have to follow the person and have a lot more detail and information to be able to make that claim,” said CDC epidemiologist Fisher.

Many countries have succeeded in following individual trails of virus. In China, for instance, contact tracing revealed how a restaurant’s air conditioning unit may have carried a positive patient’s viral droplets from one table to two others, infecting nine other people.

In Japan, investigators use contact tracing to identify clusters of disease where people live or congregate. Out of about 3,000 cases confirmed from January to April in that country, investigators could identify 61 clusters, 16% of which were in restaurants or bars.

The failure to achieve comprehensive contact tracing means that decisions about whether to close restaurants, or how many customers to allow at a time, have relied heavily on the local political climate. Because the data from contact tracing is sketchy, it’s not always easy to correlate a community’s restaurant restrictions with case rates.

In San Diego, where indoor dining had been permitted with restrictions since the debut of the state’s tiered reopening system in August, 9.2% of COVID-infected residents reported visiting a bar or restaurant up to two weeks before their symptoms appeared. All indoor dining ended in the county Nov. 14 because the county reached a threshold of case reports that led to state-required closings.

[Editor’s note: On December 3, California governor Gavin Newsom announced a new stay-at-home order that shuts down both indoor and outdoor dining when hospital intensive-care capacity falls below 15%.]

In Houston, meanwhile, 8.7% of COVID-positive people interviewed for contact tracing listed a restaurant, cafe or diner as a potential source of exposure since June 1. Restaurants there have been allowed to operate at 75% of indoor capacity since mid-September.

Other local governments have contact tracing completion rates so low that the data gleaned may not be meaningful.

For example, in Philadelphia, only about 2% of the COVID patients interviewed by contact tracers reported going to a restaurant, and the city allowed restaurants to reopen for indoor dining on Sept. 8. But it’s not clear how representative the city’s figures are. In one recent week, Philadelphia investigators were able to reach only 29% of the 2,110 positive cases they sought to contact. Despite this, indoor dining was stopped on Nov. 20 to combat a surge of cases.

In California, the state restricts the operation of establishments based on overall case and positivity rates in each county. But counties with more robust contact-tracing programs, like Los Angeles, have been able to glean striking insights from interviewing positive patients.

In Los Angeles, about 6% of COVID infections have occurred among restaurant customers, according to the public health department, though only outdoor dining has been allowed there since the state debuted its current tiered system in August.

That data suggests that even outdoor dining may spread the virus, said Shira Shafir, an associate professor of community health sciences and epidemiology at UCLA.

She gets takeout regularly to support the restaurants in her neighborhood but hasn’t eaten out since February, having concluded it isn’t worth the risk to herself and other patrons, or to the restaurant workers.

“I don’t want to ask someone else to take a risk that I’m unwilling to take,” she said.