Identification of the first COVID-19 infections in the US using a retrospective analysis

David García‐García, Enrique Morales, César de la Fuente‐Núñez, Isabel Vigo, Eva S. Fonfría, César BordehoreResearch Square 2021 · DOI

Abstract

Accurate detection of early COVID-19 cases is crucial to drastically reduce infection, hospitalization, and death rates. However, it remains a challenge and methods for identifying initial COVID-19 cases are urgently needed. Here, we used the results from a seroprevalence study in 50 US states to apply our Retrospective Methodology to Estimate Daily Infections from Deaths (REMEDID) with the aim of analyzing the initial stages and spread of SARS-CoV-2 infections across the United States (US). Our retrospective data analysis revealed that the virus likely entered the country through California on December 28, 2019, which corresponds to 16 days before the officially recognized entry date established by the CDC. Thus, REMEDID provides evidence that SARS-CoV-2 entered the U.S. earlier than previously reflected in official data. Collectively, our mathematical modeling more accurately estimates the initial COVID-19 cases in the US, may be extrapolated to other countries, and may be used to retrospectively track the progress of the pandemic. Approaches such as REMEDID may enable health authorities to accelerate preventative measures aimed at controlling pandemics within weeks of their onset.

Introduction

Identifying the "Patient Zero" or the true start date of a pandemic is notoriously difficult. During the early stages of COVID-19, limited testing capacity and a high proportion of asymptomatic or mild cases meant that many infections went undetected. Official records usually rely on positive PCR tests, which are "lagging indicators." This study addresses the gap between when the virus actually arrived and when public health systems finally noticed it. Understanding this "silent spread" is crucial for improving future pandemic preparedness and response times.

Methodology: The REMEDID Approach

The researchers utilized a methodology called REMEDID (Retrospective Methodology to Estimate Daily Infections from Deaths). Instead of relying on unreliable early case counts, REMEDID works backward from the most stable data point available: deaths.

The process involves:

  1. Estimating the Case Fatality Ratio (CFR): The team used state-specific seroprevalence studies (blood antibody tests) to estimate the total number of people actually infected in each state. By comparing this to the number of deaths, they calculated a localized CFR.
  2. Probabilistic Back-calculation: Using known statistical distributions for the incubation period and the time from symptom onset to death, the model estimates when the original infections must have occurred to result in the observed deaths.
  3. Determining the Start Date: The first positive integer result in the resulting time series is identified as the date of the first infection.

The study integrated several key data sources, including seroprevalence data from 50 US states (Bajema et al., 2020), official COVID-19 statistics from USAFacts/CDC, international flight data, and clinical parameters derived from early Wuhan data.

Key Results and Insights

The most striking finding is the existence of a "Silent Month." On average, the virus was circulating in a state for over a month (32 days) before the first case was officially documented.

The Timeline of Viral Entry

The model identified the first US infection as occurring on December 28, 2019, in California. This is 29 days before California's first official case and 16 days before the CDC's recognized entry date for the country. The predicted entry points—California (Dec 28), Washington (Jan 9), and New York (Jan 19)—align with US airports receiving the highest volume of flights from China in late 2019.

By the time the first case was officially reported in Washington (Jan 22, 2020), the model suggests the virus was already present in several other states, including New York and Illinois.

Comparison of REMEDID estimated first cases vs. documented cases (Alabama to Nebraska)

State-Level Variations

The delay in detection was a generalized phenomenon across the entire country. However, Illinois was an outlier, showing only a 4-day gap. This is likely because the first Illinois case was a traveler from Wuhan who was quickly identified, suggesting the model can distinguish between sporadic imported cases and sustained community transmission.

The model also reordered the "ranking" of when states were hit. While West Virginia was officially the last state to report a case (March 17), REMEDID suggests Wyoming was actually the last state to see its first infection (Feb 28).

Discussion and Limitations

The accuracy of these results depends on the quality of reported death data and the representativeness of antibody surveys. The model assumes that the ratio of asymptomatic to symptomatic cases remained relatively stable and relies on clinical parameters (incubation and death timing) derived from early Wuhan data.

Despite these limitations, REMEDID offers a powerful tool for health authorities. By monitoring deaths and conducting early serological surveys, officials could use this mathematical "hindsight" to implement preventative measures weeks earlier than they would by relying on positive test results alone.

Summary updated

Abstract

Accurate detection of early COVID-19 cases is crucial to drastically reduce infection, hospitalization, and death rates. However, it remains a challenge and methods for identifying initial COVID-19 cases are urgently needed. Here, we used the results from a seroprevalence study in 50 US states to apply our Retrospective Methodology to Estimate Daily Infections from Deaths (REMEDID) with the aim of analyzing the initial stages and spread of SARS-CoV-2 infections across the United States (US). Our retrospective data analysis revealed that the virus likely entered the country through California on December 28, 2019, which corresponds to 16 days before the officially recognized entry date established by the CDC. Thus, REMEDID provides evidence that SARS-CoV-2 entered the U.S. earlier than previously reflected in official data. Collectively, our mathematical modeling more accurately estimates the initial COVID-19 cases in the US, may be extrapolated to other countries, and may be used to retrospectively track the progress of the pandemic. Approaches such as REMEDID may enable health authorities to accelerate preventative measures aimed at controlling pandemics within weeks of their onset.
BibTeX
@inproceedings{garciagarcia2021identification,
  author = {David García‐García and Enrique Morales and César de la Fuente‐Núñez and Isabel Vigo and Eva S. Fonfría and César Bordehore},
  title = {{Identification of the first COVID-19 infections in the US using a retrospective analysis}},
  booktitle = {Research Square},
  year = {2021},
  doi = {10.21203/rs.3.rs-707353/v1},
}
Figure