Research, Rapid Research Review

Rapid Research Review: Epidemiology and COVID-19 - Details Matter

Social media has made everyone an epidemiologist in this era of COVID-19. But surprise! Details matter. Learn how to interpret the data and consider it in context.

Note: The COVID pandemic is currently ongoing and the numbers cited here were current as of the writing of this article on 4/20/20.

The COVID-19 pandemic has led to an incredible surge in interest in epidemiology. It warms my stats-nerd heart when my mom quotes case fatality rates and "flattening the curve" over Zoom. However, I think there remains some misconceptions around these metrics, leading to inaccurate interpretations within the media and by its consumers. Here, I provide an overview of the case fatality rate metric and how differences in case definitions, testing, and demographic characteristics / underlying health conditions may impact this metric.

What is a Case Fatality Rate (CFR)?
A case fatality rate is defined as the number of deaths due to a disease that occur over a given period of time (usually a year) divided by the size of the population at risk of dying from said disease (identified cases).1

The media frequently mentions differing estimates of the CFR of COVID-19 and for good reason. This estimate provides critical insight into the severity of a disease and, in regard to COVID19, has been a critical metric in the decision to develop and enforce Shelter in Place rules.

Variation in CFR
The case definition of COVID-19 varies across countries and can even change over time within the same jurisdictions.

Definition of a death
On 4/14/20 New York’s death count (numerator for case fatality) increased from 6,587 to 10,367 when the decision was made to include all patients who died with presumed COVID-19 without a confirmed positive test.2

Definition of a case
If we include presumed COVID-19 cases in the numerator, we also need to account for all COVID-19 cases in the denominator. The ideal way to do this would be to test every individual in the whole population.  The closest to this ideal is in South Korea where a global testing strategy and has led to a confirmed 8,320 cases and 81 deaths (CFR = 81/8320 = 0.97%) as of 4/17/2020.3,4

Compare this to New York City where the NYC Department of Health is reporting 132,467 cases and 11,683 deaths (CFR = 11683/123467 = 9.46%) as of 4/20/2020. The New York City Department of Health recommends (as of 4/11/2020) testing only hospitalized patients given the current shortage of testing kits.5 At the same time, the US Navy tested nearly all crewmembers of the USS Theodore Roosevelt, a ship where a COVID-19 outbreak occurred, and of 660 cases, over 60% were asymptomatic.2

This combination of limited testing and a signification number of asymptomatic cases is most likely leading to a significant underestimation of the number of cases causing an overestimation of the case fatality rate.

The test
The other thing to consider is that at this point there is also no gold standard for diagnosing COVID-19. Currently, testing involves acquiring a sample with a nasopharyngeal swab and running an RT-PCR test. One article identified the sensitivity of this test to be as low as 59% for patients with suspected COVID-19 in Wuhan, China.6

Confounders
CFRs are also not same because populations are the same. For example, in Italy, 23% of the population is over the age of 65 compared to 11% China. Age is the strongest predictor of death in COVID-19 patients, and therefore, this significant difference is a key driver of CFR.7 This point is well-illustrated in recent article by Onder, et al.8 The reported CFR from Italy and China was 7.2% and 2.3%, respectively. In Italy over 50% of deaths occurred among those 80 or older. Compare this to China where only 20% of deaths occurred in those 80 or older.

Details matter
The next time you hear the most updated case fatality rate, be an informed consumer. Think about context:

  • Who is the source?
  • What are key health policies in that population?
  • How is the population at risk defined?
  • What are the characteristics of the population?
  • How is a case defined?
  • Is testing limited?
  • How good is the test?

References

  1. Pagano MG, Kimberlee. Principles of Biostatistics. Second Edition ed. Pacific Grove, CA: Duxbury; 2001.
  2. Durkin E. NYC death toll jumps by 3,700 after uncounted fatalities are added. Politico. 04/14/2020, 2020.
  3. KCDC. Updates on COVID-19 in Republic of Korea(as of 17 March) [press release]. Division of Risk assessment and International cooperation. 03/17/2020.
  4. Normile D. Coronavirus cases have dropped sharply in South Korea. What's the secret to its success?  Science Magazine. 2020.
  5. NYC Health. 2020 Health Alert #10: COVID-19 Updates for New York City Face Mask Use Policy, Swab Shortage, Reporting COVID-19 Related Deaths and Crisis Communication Resources [press release]. 4/11/2020.
  6. Ai T, Yang Z, Hou H, et al. Correlation of Chest CT and RT-PCR Testing in Coronavirus Disease 2019 (COVID-19) in China: A Report of 1014 Cases. Radiology. 2020:200642.
  7. Jordan RE, Adab P, Cheng KK. Covid-19: risk factors for severe disease and death. BMJ. 2020;368:m1198.
  8. Onder G, Rezza G, Brusaferro S. Case-Fatality Rate and Characteristics of Patients Dying in Relation to COVID-19 in Italy. JAMA. 2020;323(18):1775-1776.

Related Articles

Getting Started in EM Research: Lessons from Rising Stars

The EMRA Research Committee interviewed 4 soon-to-be legends in the emergency medicine research community, from residents to established faculty, to understand their experiences. Our interviewees — Ma

Rapid Research Review: Power

EM Resident 10/02/2016
Rapid Research Review: Power Power Recall that Type I error  is the probability of falsely rejecting the null hypothesis (H0) based on our sample even though H0 is true in the population. Alternative
CHAT NOW
CHAT OFFLINE