Essential facts and figures -- COVID-19

post by Ardenlk · 2020-04-20T18:33:50.565Z · score: 19 (4 votes) · EA · GW · 0 comments


  Symptoms and severity
  Who is most at risk?
  How is the virus spread?
  How many will be infected?
  Reducing the risk of infection
No comments

This is a link post for 'Essential Facts and Figures -- COVID-19'.

We wrote this as part of 80,000 Hours' package of content on the outbreak [EA · GW], inspired by Ethan Alley's 'COVID-19 in brief' [EA · GW] post, which we thought was was a useful summary.

Our article is in the same spirit as Ethan's post, but covers somewhat different ground and is a bit more up-to date (last updated April 3rd v.s. March 13th). Since Ethan's post seemed to be useful to people on the Forum, we thought we might as well post our summary as well.

Symptoms and severity

Who is most at risk?

How is the virus spread?

How many will be infected?

Reducing the risk of infection


  1. A great deal of research has been published on COVID-19 symptoms. Our figures are based on the following (which as far as we can tell are roughly in line with most other studies): a 28 February paper, "Clinical Characteristics of Coronavirus Disease 2019 in China" found fever in 43.8% of patients upon admission to a hospital and 88.7% during hospitalization. A 30 January paper "Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study" found that upon admission to the hospital, 83% of patients had a fever, 82% had a cough, and 31% had shortness of breath. The CDC lists fever, cough, and shortness of breath as symptoms to watch for. ↩︎

  2. According to investigators from the Wuhan Medical Treatment Expert Group for COVID-19 as reported here, 84% of COVID-19 patients experienced loss of appetite, 29% had diarrhea, 0.8% experienced vomiting, and 0.4% had abdominal pain. It's hard to draw a conclusion here, because some of this contradicts earlier evidence from the 28 February paper cited in the previous footnote, which found diarrhea in only 2% of cases at admission. ↩︎

  3. The 30 January paper from footnote 1 found sore throat at admission in 5% of patients and runny nose (rhinorrhoea) in 4%. ↩︎

  4. See for instance this New York Times article from 22 March ↩︎

  5. A 26 March paper "Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus–Infected Pneumonia" and a 10 March paper, "The Incubation Period of Coronavirus Disease 2019 (COVID-19) From Publicly Reported Confirmed Cases: Estimation and Application" found a mean and median (respectively) incubation period of around 5 days. The World Health Organization reports similar figures. However, a 18 March preprint "Is a 14-day quarantine period optimal for effectively controlling coronavirus disease 2019 (COVID-19)?" suggests longer incubation periods. Based on a wider range of cases (2015 compared to 181 for the March 10 paper and 425 for the March 26 paper), the authors found incubation periods ranging from 0 to 33 days with a median of 7 days. It's not fully clear to me what accounts for the difference here, since all the above studies were based on confirmed cases, in China, in January and February. However, the last study's large sample size and explicit attempt to include less severe cases make its finding seem more reliable. The authors of that study also report longer incubation periods for children and people who seem to have been infected via food rather than breathing in droplets (both with a median of 9 days). ↩︎

  6. The Chinese CDC found that 80.9% of cases reported in China up to 11 February were 'mild', meaning without pneumonia or with only mild pneumonia (so you can still feel quite sick with a mild case), 13.8% were 'severe', involving shortness of breath and lower blood oxygen levels, and 4.7% were 'critical', meaning they had respiratory failure, septic shock or organ failure. See Table 1 for the figures and 'Variables' for the definitions. ↩︎

  7. As reported by the Centers for Disease Control, testing on the Diamond Princess cruise ship found 46.5% of passengers who tested positive were asymptomatic at the time of testing. However, COVID-19 has a long incubation period, and these people may have become symptomatic later. A 26 March follow up suggested based on modeling that 17.9% of those who tested positive never developed symptoms. ↩︎

  8. The influential 6 March paper "Impact of non-pharmaceutical interventions (NPIs) to reduce COVID19 mortality and healthcare demand" uses a 0.9% (95% credible interval 0.4%-1.4%) infection fatality rate for the UK in its analysis. Their estimate comes from making adjustments to the 13 March preprint "Estimates of the severity of COVID-19 disease", which finds a 0.66% infection fatality rate for China with a 95% credible interval of 0.39% to 1.33%. A 6 March preprint, "Adjusted age-specific case fatality ratio during the COVID-19 epidemic in Hubei, China, January and February 2020", found a rate of 1.6% (1.4%-1.8%). The report "2019-nCoV: preliminary estimates of the confirmed-case-fatality-ratio and infection-fatality-ratio, and initial pandemic risk assessment" (updated 19 February) found an infection fatality rate for China of 0.94% (0.37% to 2.9%). A very recent (29 March) estimate from Oxford's Center for Evidence Based Medicine puts the worldwide infection fatality rate considerably lower: "Taking account of historical experience, trends in the data, increased number of infections in the population at largest, and potential impact of misclassification of deaths gives a presumed estimate for the COVID-19 IFR between 0.1% and 0.26%."

    Even if the worldwide IFR is relatively low right now, it would not be surprising if infection fatality rates were considerably higher in places where the medical system is overwhelmed, such as Italy, or in the future if the infection spreads to a greater proportion of the population and hospitals are overwhelmed in more places. ↩︎

  9. There have been some early reports of reduced lung function after recovery from COVID-19, but no strong evidence. Speculation based on other diseases is inconclusive. COVID-19 is more severe than the common cold and less severe than SARS1 from 2003. The common cold doesn't result in long-term effects, but according to one study some survivors of SARS1 reported suffering from ongoing lung problems or chronic fatigue. ↩︎

  10. Some studies seem to indicate that children are not only less likely to be symptomatic, they are less likely to catch the virus. For instance, the "16-24 February Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19)" found that "Within Wuhan, among testing of ILI samples, no children were positive in November and December of 2019 and in the first two weeks of January 2020." However, that report also emphasizes that we cannot draw conclusions from the data available -- perhaps because children were less likely to be tested at all. And a 27 March preprint "Epidemiology and Transmission of COVID-19 in Shenzhen China: Analysis of 391 cases and 1,286 of their close contacts" suggests children are at a similar risk of infecting others as adults. We haven't had time to look into this topic. What does seem clear is that children are in much less danger themselves. But it seems likely that mild and asymptomatic cases can still be infectious (see footnote 15), so even then children can likely still infect others. ↩︎

  11. The joint mission report cited above reports a case fatality rate for the studied patients higher among men vs. women (4.7% vs. 2.8%), and a 21 February report from the Chinese CDC found a case fatality rate for men of 2.8% v.s. 1.7% for women. It's not clear, however, whether this difference is attributable to sex or something else that correlates with sex. ↩︎

  12. See guidelines from The Center for Disease Control and the World Health Organization. ↩︎

  13. A 17 March "correspondence," in the New England Journal of Medicine, "Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1", found that the novel coronavirus survives for a shorter time on copper and cardboard, and longer on stainless steel and plastic. The results for cardboard are more uncertain than for the other surfaces. ↩︎

  14. It seems very likely there is at least some infectious period before symptoms develop. An 18 March preprint "Temporal dynamics in viral shedding and transmissibility of COVID-19" estimates that infectiousness peaks on or before the onset of symptoms, based on finding the highest amount of virus in patients' throats at the time of symptom onset. They further suggested that a substantial amount of transmission could occur before the onset of symptoms. The Imperial College London paper cited above assumed that transmission was possible 12 hours before the onset of symptoms. Suggestively, the common cold is transmissible up to a few days before symptoms begin. ↩︎

  15. It's difficult to know how much transmission is due to mild and asymptomatic cases, as they usually go unrecognized. A 16 March paper "Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus"(SARS-CoV2)" tried to find how much of COVID-19's spread was due to these unrecognized cases in China, and reported that "Per person, the transmission rate of undocumented infections was 55% of documented infections ([46%–62%])." However, we don't know how mild the symptoms were in these undocumented infections. In general, there is controversy about the relationship between symptoms and infectiousness. ↩︎

  16. Up To Date’s guidelines on home care cite the CDC and WHO recommendations. There is some reason to think infectiousness peaks at the beginning of symptoms (see footnote 14), though even if that's right we don't know how long it takes to subside after that. ↩︎

  17. According to data collected by Our World in Data, by 14 March the US (population 327 million) had done 37,646 tests; Japan (127 million) had done 14,901 by 19 March, and India (1.3 billion) had done 14,514 by 20 March. On the other hand, by 20 March South Korea (51 million) had done 316,664 tests and Taiwan (24 million) had done 21,376. This suggests large variation in the number of cases being captured in each country. ↩︎

  18. The average number of people each infected person infects (the 'R0'), is uncertain and changes with circumstance. The Imperial College London paper cited above assumes that each person will on average infect 2.4 people, based on early transmission data from Wuhan. A newer (30 March) paper from Imperial, "Estimating the number of infections and the impact of nonpharmaceutical interventions on COVID-19 in 11 European countries" estimates that without interventions in Europe the average R0 across 11 countries was 3.57 (95% credible interval 3.01-4.66). A 7 April "early release" paper published on the CDC's website, "High Contagiousness and Rapid Spread of Severe Acute Respiratory Syndrome Coronavirus 2" found a much higher R0 of 5.7 for early transmission in China, with a credible interval of between 3.8 and a shockingly high 8.9. The 30 March paper from Imperial found that interventions implemented in the studied countries caused a reduction of 64% in the R0 on average. ↩︎

  19. "Herd immunity" is a phenomenon whereby a large enough portion of a population is immune to a virus, whether because they have already been infected or from vaccination, that the virus is mostly unable to spread in the population, even to those not immune. The threshold of infection for a population to have herd immunity depends on how many people the average infected person infects (the 'R0'), and how common and long lasting immunity is (which is unknown for those who have been infected with COVID-19, but thought to be reasonably common and long-lasting, as with other coronaviruses), among other things. The 50-80% figure comes from the 7 April "early release" cited in the previous footnote, according to which for an R0 of 2.2 the threshold for herd immunity is 55%, and for and R0 of 5.7 it is 82%" The threshold for herd immunity could be even higher if, as the authors of that paper say might have been the case in the early days of the epidemic in Wuhan, the R0 is even higher than 5.7. The UK's chief scientific advisor Patrick Vallance gave a figure of 60% for the UK in mid-march, as reported by Sky News and repeated in the Financial Times. ↩︎

  20. This is the main advice we've seen from many sources -- though different institutions have slightly different recommendations. See e.g., guidelines from the NHS, the CDC, and the Hong Kong government. ↩︎

  21. You can see a review of the evidence that surgical masks reduce infection for the mask wearer here. It's worth noting that in an interview with the American Association for the Advancement of Science George Gao, director-general of the Chinese Center for Disease Control and Prevention, says he thinks not using masks to help control the pandemic in the West is a mistake. ↩︎ ↩︎

  22. Some worry that homemade face masks could cause or worsen infections by capturing viruses and bringing them closer to the face, or if they are shared or not washed. They might also give people a false sense of security, causing people to be more cavalier about going out and being near other people, and even proponents agree they are not highly effective at preventing infection, either for the wearer or those around them. All that said, they may be better than nothing. This post makes the case for using homemade masks to help decrease transmission. ↩︎

  23. On 3 April the FDA announced plans to develop blood therapies, including convalescent plasma, for COVID-19, while allowing their simultaneous use for critically ill cases without other options. The sense of promise for these therapies is based on past successes treating other respiratory viruses with these methods as well as preliminary reports of improvement in COVID-19 patients given plasma (1, 2). These reports are not from controlled trials, which have not been conducted, and are based on use in 10 and 5 cases, respectively. ↩︎

  24. This 22 March article from the American Association for the Advancement of Science gives a helpful overview of different treatments being trialed and what we know about timelines for development. Here is another, similar article in STAT from 24 March. ↩︎

  25. We did not draw this estimate from a single source, but rather from a variety of popular articles discussing the necessary steps and typical obstacles in vaccine development. One such article from The Guardian (updated on 1 April) is here. ↩︎


Comments sorted by top scores.