July 2020: Adults With COVID-19 Were About Twice as Likely to Have Dined Out

In July 2020, CDC compared the recent lives of two groups of adults who had come in sick to be tested for the virus that causes COVID-19: 154 who tested positive and 160 who tested negative. Most everyday activities looked the same in both groups — shopping, gatherings at home, offices, salons, gyms, public transport, church. Two things stood out: close contact with someone known to have COVID-19, and places where people eat and drink on site.

The main findings

• Restaurants. People who tested positive were about twice as likely to have dined at a restaurant in the two weeks before they fell ill — 40.9% against 27.7%, an adjusted odds ratio (aOR) of 2.4 (95% CI 1.5–3.8). "Dining at a restaurant" meant any seating the restaurant offered — indoors, patio or outdoors.
• Among people with no known contact with a COVID-19 case (89 positive, 136 negative), the gap was sharper: restaurants aOR 2.8 (1.9–4.3), and bars and coffee shops aOR 3.9 (1.5–10.1).
• Close contact. 42% of those who tested positive reported close contact — within 6 feet for 15 minutes or more — with someone known to have COVID-19, against 14% of those who tested negative. For the positive group, 51% of those contacts were family members, consistent with spread inside households.
• Masks. 71% of the positive group and 74% of the negative group said they always wore a mask or cloth face covering in public in the two weeks before illness.
• Others around them. Among people who dined out, 19.0% of the positive group said none or only a few other people there were masking or distancing, against 2.3% of the negative group.

Adjusted odds of a positive test for each activity, for everyone (dark) and for those with no known COVID-19 contact (light). Image from CDC's report.

Why eating and drinking?

A mask can't be worn while eating or drinking, but shopping and many other indoor activities don't prevent it. Earlier reports had linked restaurant exposures to air circulation, and the authors note that the direction, ventilation and intensity of airflow may affect spread even when distancing and masking follow guidance.

Who took part

Of those not included, 470 were ineligible (no symptoms, or several tests) and 163 refused. CDC staff interviewed people by phone in English, Spanish, Arabic, Vietnamese, Portuguese or Russian. The sites were in California, Colorado, Maryland, Massachusetts, Minnesota, North Carolina, Ohio, Tennessee, Utah and Washington.

The groups differed: those who tested negative were more likely to be non-Hispanic White (77.5% vs 59.7%), to have a college degree (67.7% vs 50.3%) and to have at least one chronic condition (61.2% vs 48.7%). The analysis adjusted for age, sex, race and ethnicity and chronic conditions.

Limits

1. Everyone was sick and sought testing at one of 11 sites; people who tested negative may have had other respiratory infections from similar exposures, and those who didn't respond may differ.
2. Unmeasured confounding is possible. The restaurant question didn't separate indoor from outdoor dining, and the bar question didn't separate venues or how drinks were served.
3. The sample may not represent the U.S. population.
4. People knew their results, which may have shaped their answers.
5. Imperfect PCR accuracy may have put some people in the wrong group.

What CDC advised

Stay home after exposure to someone with COVID-19; wash hands, wear masks and keep distance. If someone at home is ill, clean and disinfect, cut down on shared meals and items, and wear gloves and masks. And as communities reopen, reduce exposures where people eat and drink on site, to protect customers, employees and communities.

Sources

Based on Fisher KA, Tenforde MW, Feldstein LR, et al., "Community and Close Contact Exposures Associated with COVID-19 Among Symptomatic Adults ≥18 Years in 11 Outpatient Health Care Facilities — United States, July 2020," Morbidity and Mortality Weekly Report 69(36), CDC (with its erratum, which corrected only a contact address); a work of the United States government in the public domain. The results paragraph and table missing from the imported copy were taken from the report's full text in PubMed Central (PMC7499837).