How COVID-19 First Spread in the United States, January–February 2020

This page summarizes a CDC report published in May 2020, early in the COVID-19 pandemic.

Between January 21 and February 23, 2020, public health agencies detected 14 U.S. cases of COVID-19, all linked to travel from China. The first case with no travel link was confirmed on February 26 in a California resident who had fallen ill on February 13, and a second on February 28 in Washington State — signs that the virus was spreading in communities. CDC and its partners examined four lines of evidence to work out when that spread had really begun.

1. Emergency department visits

The National Syndromic Surveillance Program receives real-time data from emergency departments at about 4,000 health care facilities in 47 states and the District of Columbia. In 14 counties with early community-acquired cases, there was no substantial rise in visits for COVID-like illness — fever with cough or breathing difficulty, or a coronavirus diagnostic code — before February 28.

CDC chart: emergency department visits for COVID-19-like illness in counties with early cases.

2. Testing stored specimens

Retesting about 11,000 respiratory specimens collected from January 1 onward found no positive results before February 20.

• Seattle Flu Study. Monitoring respiratory illness in the Seattle area since November 2018, the study began testing for SARS-CoV-2 in late February 2020 and got its first positive on February 28, from a sample taken February 24. None of 5,270 earlier samples from January 1 to February 20 was positive. The first positive among the retested samples was collected February 21; that week 8 of 1,255 samples (0.6%) were positive, and the next week 29 of 1,862 (1.6%).
• Influenza vaccine effectiveness networks. At a Washington site, none of 497 samples from January 19 to February 24 was positive; the first positive was collected February 25. At sites in Michigan, Pennsylvania, Texas, Wisconsin and Tennessee, none of 2,620 samples from January 19 to February 29 was positive.
• Children. As of May 22, 2020, 4 of about 3,000 samples from children in the New Vaccine Surveillance Network, collected January 1 to March 31, had tested positive, the earliest on March 20 in Seattle.

3. The virus's family tree

Most viruses from early Seattle-area cases belonged to one lineage, the Washington State clade, whose most recent common ancestor was estimated to date from about January 18 to February 9 (point estimate February 1). Its predicted genetic sequence matched the virus from the first U.S. imported case — a man who arrived in Seattle from Wuhan, China, on January 15 and fell ill four days later — though a similar virus from someone else could also have started it. In California and the Northeast, viruses from February to mid-March pointed to several introductions, mainly from Europe, followed by spread within the United States.

4. Early cases without travel

• In Santa Clara County, California, a woman who fell ill on January 31 died on February 6, and an unrelated man died at home between February 13 and 17. Neither had traveled abroad, and CDC found SARS-CoV-2 in their postmortem tissue.
• Outbreaks on two consecutive voyages of a Grand Princess cruise ship involved viruses from the Washington State clade, suggesting an infected passenger or crew member was aboard when it left San Francisco on February 11. Who that was is unknown.

What it adds up to

Together, the evidence suggests limited community transmission began between late January and early February, after an importation from China started the Washington State clade, which spread through the Seattle area and possibly beyond. Several importations from Europe followed in February and March. Before February 28, infections were too few to show up in emergency department data.

Whether the first U.S. case started the Washington clade is unclear. The contact investigation around him was thorough and found no secondary cases, which argues against it. But because infections are often asymptomatic and can spread before symptoms, other scenarios are possible: undetected infections among his contacts; contacts infected before he had symptoms, whom the standard investigation then would not have traced; or another infected passenger on the same flight.

Why no antibody results

The report did not include serologic (antibody) testing, which is a relatively insensitive way to detect a newly arrived virus in randomly collected samples, and whose specificity rarely approaches 100% without confirmatory tests. In a hypothetical survey of the Seattle area's 3.5 million people after the first 3,500 infections, true prevalence would be 0.1%, while a test with 99% specificity would produce ten times as many false positives. Antibody surveys are still useful for tracking an established pandemic.

Limitations

• The data are retrospective and cannot give as clear a picture as widespread testing from the start would have.
• Ongoing retesting may yet find earlier cases.
• The virus varied little worldwide in January and early February, limiting what genetic analysis could show.

Sources

• CDC COVID-19 Response Team, Michelle A. Jorden and colleagues, "Evidence for Limited Early Spread of COVID-19 Within the United States, January–February 2020," Morbidity and Mortality Weekly Report, CDC: https://www.cdc.gov/mmwr/volumes/69/wr/mm6922e1.htm