Select Your Print Options

Coronavirus SARS-CoV2 infection (COVID-19)

Essential Evidence

Mark H. Ebell, MD, MS, Professor, College of Public Health, University of Georgia
Mindy A. Smith, MD, MS, Clinical Professor, Department of Family Medicine, Michigan State University
Henry C. Barry, MD, MD, MS, Professor Emeritus, Michigan State University
Pete Yunyongying, MD, FACP, Associate Professor, Carle-Illinois College of Medicine, University of Illinois
John Hickner MD, MS, Professor Emeritus, University of Illinois-Chicago

Mark H. Ebell, MD, MS, Professor, College of Public Health, University of Georgia

Last updated: 2022-01-07 © 2022 John Wiley & Sons, Inc.

Overall Bottom Line

  • EDITOR'S NOTE: This topic has been made freely available and is being updated regularly. We are at times linking to preprint servers and providing direct links to articles where possible, and most of this literature has been made freely available. As preprint work becomes peer reviewed and formally published, we will update the citations. Since preprint servers have not been peer-reviewed, and the data and conclusions may change, information from them should be used with great caution if at all.
  • Suspect COVID-19 when the virus is circulating in the population and a patient reports signs and symptoms of respiratory tract infection, or less commonly fever and neurologic symptoms or thrombosis. Most common symptoms are fever, cough, myalgias, and dyspnea; loss of taste and smell are also common. Approximately 40% of all patients are asymptomatic but appear to be as infectious as symptomatic patients. B
  • • The most appropriate diagnostic test is RT-PCR of multiple specimens carried out according to guidelines.9 Point-of-care PCR is highly specific, but sensitivity varies by manufacturer; rapid antigen testing has much lower sensitivity (56.2%) and should only be used in symptomatic patients with higher viral loads.188B
  • Preventive measures include hand washing, surface cleaning, face masks, case isolation, quarantine of contacts for 14 days, school and university closures, social distancing, and sheltering at home. The most effective available face mask should be used when in indoor public spaces to prevent spread. Modeling indicates that only by doing all of these measures can the number of severe cases requiring ventilation not overwhelm hospitals. B
  • Two mRNA vaccines from Pfizer/BioNTech and Moderna have approximately 95% efficacy at preventing symptomatic disease and good safety against SARS-CoV-2. The adenovirus vectored vaccine from Johnson and Johnson/Janssen is 67% effective overall, but 74.4% in the US population that was studied. Vaccine effectiveness for the delta variant is about 87% to 90% overall, but lower among the immunocompromised and elderly. Boosters increase protection about 10-fold and are recommended at least 6 months after the second dose of vaccine. B
  • In patients not requiring oxygen or only requiring low-flow oxygen, remdesivir shortens the duration of hospitalization (11 vs. 15 days) and may reduce mortality slightly.108B
  • Systemic corticosteroids are highly effective at reducing mortality in patients with COVID-19 who are mechanically ventilated (NNT = 7) or who are on oxygen (NNT = 20) but not in hospitalized patients not requiring oxygen.161171B
  • Two studies have found that use of inhaled budesonide in outpatients with early disease results in a shorter duration of symptoms and possibly a lower risk of hospitalization, death, and the need for urgent visits.B
  • A single RCT enrolling 1497 high risk outpatients with symptomatic COVID-19 compared fluvoxamine with placebo and reported a reduced likelihood of hospitalization (11% vs. 16%, NNT = 20, 95% CI 12-61).B281
  • The monoclonal antibody bamlanivimab and the combination of casirivimab and imdevimab (Regeneron) have been given emergency use authorization for treatment of outpatients not on supplemental oxygen but at high risk for severe disease. A systematic review found an NNT of 21 to 24 to prevent hospitalization.280B
  • In newly hospitalized patients not requiring mechanical ventilation, the Janus kinase inhibitor tofacitinib 10 mg twice daily reduced the composite of death or respiratory failure (18.1% vs. 29.0%, p = 0.04, NNT = 9).265B
  • Multiple randomized controlled trials have confirmed that hydroxychloroquine (HCQ) is not effective for severe disease, mild disease, early disease, or as post-exposure prophylaxis, and is associated with a higher risk of adverse events.173105A
  • Patients can be considered cured using a test-based strategy (recovery from fever without antipyretics and without respiratory symptoms plus 2 negative PCR tests 24 hours apart). For outpatients in settings where tests are not widely available, the CDC recommends that isolation be maintained for at least 10 days after illness onset and at least 3 days (72 hours) after recovery, defined as: at least 3 days free of fever without antipyretics, 3 days without respiratory symptoms, and at least 7 days after onset of symptoms.Data support that after 10 days, the likelihood of transmission appears negligible. C
  • The overall case fatality rate is estimated to be between 0.5% and 0.9% and is higher in older patients and those with comorbidities. This estimate, from early in the pandemic, is likely lower now due to better treatment and ventilator management.19865B


  • COVID-19 (coronavirus disease 2019) is a viral lower respiratory infection first reported in Wuhan City, China that has rapidly spread to become a pandemic. It is caused by novel coronavirus named 2019-nCoV and more recently SARS-CoV2.


  • Information about incidence and the case fatality rate are evolving. Peak incidence predictions nationally and by state for the US are provided by the modeling group at the University of Washington.
  • Several other Web sites and the Johns Hopkins Coronavirus Resource Center provide detailed information on new cases, total cases, and deaths that are updated daily.
  • In a study in Los Angeles County, California early in teh pandemic, researchers invited a random sample of 1952 adult residents, 863 of whom agreed to be tested. Thirty-five (4%) individuals tested positive; after adjusting for the sensitivity and specificity of the test and population weighting they estimated 4.6% of adults had been infected, roughly a 50-fold increase in the number actually diagnosed.113
  • In a study of 220 women admitted for delivery in New York in late March/early April 2020, all were tested and 15.4% were positive for SARS-CoV2, most of whom were asymptomatic.60
  • A Swiss study sampled 2766 participants who were demographically similar to the overall population of the Geneva Canton for antibodies to COVID-19. After accounting for test accuracy and other factors, the range of seroprevalence during the first 4 weeks of the study ranged from 4.8% to 10.9%. Prevalence was highest in those 20 to 49 years of age, and they estimate 11.6 infections in the community for each confirmed case.129
  • The CDC has begun a series of seroprevalence studies in 6 states, using blood specimens obtained for reasons other than COVID-19 testing (preprint server, not peer-reviewed). They used an antibody test that is 96% sensitive and 99.3% specific, and adjusted the results to account for false negatives and false positives. In the most recent report, age and sex-standardized positive rates were 1.1% in Washington, 1.9% in south Florida, 2.2% in Utah, 2.4% in Minneapolis-St. Paul, 2.7% in Missouri, 3.2% in Philadelphia, 4.9% in Connecticut, 5.8% in Louisiana, and 6.9% in the metro New York City region. Based on the number of confirmed cases, they estimate a range of 6 to 24 total cases per confirmed case, with 4 of 6 jurisdictions having 10.8 to 11.9 total cases per confirmed case. These estimates are based on data collected from mid-March to late May, so current conditions may reflect a higher seroprevalence and lower ratio of undetected to confirmed cases.152
  • The rate of infection and hospitalization is higher for Black, Hispanic/Latino, and American Indian/Alaska Native than would be expected from their proportion in the general population.168

Other Impact

  • Based on modeling by the COVID-19 research unit at Imperial College in London early in the pandemic, the case fatality ratio is estimated to be 0.9% overall (95% credible interval 0.4%-1.4%). It is lowest in children (0.002%) and is higher with increasing age (0.08% for 30-39, 0.60% for 50-59, 2.2% for 60-69, 5.1% for 70-79 and 9.3% for 80 and older).
  • Undercounting deaths from COVID is likely. A report updated on September 11, 2020 estimated over 263,000 excess deaths globally. This was based on comparing current overall mortality trends in multiple countries with historical data.
  • Using excess mortality data, in New York City COVID-19 was more deadly in March through May of 2020 than the 1918 flu epidemic was during its peak months.162
  • The CDC estimates that the US experienced 890,990 more deaths than would otherwise be expected during the period from January 26 to November 20, 2021. Surprisingly, the greatest percent increase occurred in those between 25 and 44 years of age, who had a 26.5% increase over expected deaths. For Americans between 25 and 44, 45 to 64, 65 to 74, 75 to 84, and 85 years of age or older, the percent increase in deaths were 27%, 14%, 24%, 22%, and 15%, respectively. The average percentage increase was 54% in Hispanic persons, 29% above average for non-Hispanic American Indian or Alaska Native persons, 33% above average for Black persons, 35% above average for those of other or unknown race or ethnicity, and 37% above average for Asian persons. Regularly updated data are available at:

Causes of the Condition

  • The cause of COVID-19 is the novel coronavirus SARS-CoV-2.
  • Coronaviruses are enveloped, single-stranded RNA viruses that include those that caused a large-scale epidemic of severe acute respiratory syndrome (SARS) in 2002-3 and the Middle-Eastern Respiratory Syndrome (MERS), a persistent epidemic in the Arabian Peninsula since 2012. There also several widely circulating species that cause mild respiratory tract infections in humans.74


  • COVID-19 infection ("COronaVirus Disease 2019") is caused by the novel coronarvirus SARS-CoV-2. It is the 7th coronavirus reported to cause disease in humans.
  • Like SARS-CoV, SARS-CoV-2 is in the subgenus sarbecovirus. The genome has been sequenced and it is more closely related genetically to SARS than to MERS (Middle Eastern Respiratory Syndrome).16
  • COVID-19 appears to cause a prothrombotic state with microthrombi identified in multiple organs including the lungs, kidneys, heart, and liver, based on autopsy studies. In addition, megakaryocytes (bone marrow cells responsible for producing platelets) were found in higher than usual numbers in the lungs and heart.146
  • Coronaviruses make use of a large envelope protein called the spike to engage host cell receptors and catalyze membrane fusion. In a recent study using llamas immunized with prefusion-stabilized betacoronavirus spike proteins, investigators identified neutralizing cross-reactive single-domain camelid antibodies that can attach to and neutralize the viruses’ spike protein and may serve as potential therapeutic candidate for human vaccines.83
  • Incubation period
  • The incubation period is a median of 5 days, and 97.5% who develop symptoms will do so within 11.5 days. About 1% may develop symptoms more than 14 days after exposure.36
  • Duration of viral shedding and infectiousness
  • The median duration of viral shedding in survivors is 20 days, based on early data from Wuhan City, China.18
  • A systematic review of 8 studies reporting on viral shedding found that 40.5% (significant heterogeneity) of patients with COVID-19 shed the virus in their stools.128
  • Based on a series of 94 patients, researchers estimated that 44% (95% CI 25%-69%) were infected by presymptomatic or asymptomatic index patients. Shedding declined over a median 21-day period. They estimate that patients were infectious 2.3 days prior to symptom onset (95% CI 0.8-3.0 days), with a peak infectiousness at 0.7 days prior to symptom onset (95% CI -0.2-2.0 days).115
  • A South Korean study of 193 symptomatic and 110 asymptomatic individuals found that they tested positive for 17 to 19 days, with little difference between symptomatic and asymptomatic persons.179
  • Among 96 consecutive patients hospitalized with COVID-19, 22 had mild disease and 74 had severe disease. The patients had daily PCR assays of sputum, saliva, blood, urine and stool. Among the 3497 samples, SARS-CoV-2 RNA was detected in 59% of the patients. The median duration of fecal shedding of SARS-CoV-2 RNA was 22 days compared with 18 days in the respiratory samples and 16 days in the serum samples. It was detected in only a single urine sample. Patients with severe disease shed virus for 1 week longer than those with mild disease (21 days vs. 14 days, respectively).68
  • In a case-control study, compared to patients without GI symptoms, the 107 patients who had presented with GI symptoms had a longer duration of viral shedding (41 vs. 32 days).135
  • In an Italian study that tested the entire town of Vo', persons were infectious for between 3.6 to 6.5 days, with infectiousness peaking on the day of symptom onset.145 A Taiwanese study similarly found that persons were infectious for about 5 to 6 days.78
  • The NBA “bubble” in Orlando, Florida that isolated all NBA players, staff and vendors during the second half of the season and playoffs provided a natural experiment for determining if those who persistently test positive with a sensitive PCR test are infectious. Of the 3,648 individuals who participated in the bubble, 36 individuals persistently tested positive after the 10-day isolation period, which began on the first day of symptoms of infection or the first positive test. The mean number of persistent positive days for these 36 individuals was 31 days. There were at least 1,480 person-days of close, direct contact among these individuals and others in the bubble and there were no documented cases of transmission. Thus, after 10 days, the likelihood of transmission appears negligible.258
  • Mode of transmission of virus
  • It is thought that the virus is spread primarily through respiratory droplets and via aerosols. However, the virus can be spread by asymptomatic persons.27 In fact, a modeling study estimated that about 59% of all infections are spread by asymptomatic persons, 35% of whom are pre-symptomatic and 24% of whom will never become symptomatic.227
  • A study of 128 bus riders in 2 busses in China found a very high attack rate in the one bus with a single infected person, suggesting a strong likelihood of airborne spread via aerosols as many persons not in close proximity were infected.284
  • In Singapore, a country where there was very careful contact tracing, researchers estimated that 10 of 157 (6.4%) locally-acquired cases were infected by someone who was presymptomatic (e.g., 1 to 3 days before symptom onset) based on contact with the presymptomatic individual and no contact with any symptomatic persons.114 A modeling study estimated even higher rates of infection from persons with mild or asymptomatic disease, due to greater activity levels.48
  • Researchers reported a cluster of 10 COVID-19 cases that were most likely spread from one infected person at a restaurant in Guangzhou, China. The investigators could find no other exposures to COVID-19 for the other two infected families. After careful contact tracing and assessment of the physical environment, the researchers hypothesize that the infection may have been spread by the air-conditioning unit. If so, this report has significant implications as countries open up restaurants and other business venues requiring close physical proximity.64
  • The CDC performed a case-control study that included 154 who tested positive and 160 who tested negative. Case-patients were more likely to have reported dining at a restaurant (adjusted odds ratio [aOR] = 2.4, 95% CI 1.5-3.8) in the 2 weeks before illness onset than controls. When the analysis was restricted to the 225 participants who did not report recent close contact with a person with known COVID-19, cases were more likely than were controls to have dined at a restaurant (aOR = 2.8, 95% CI 1.9-4.3) or gone to a bar/coffee shop (aOR = 3.9, 95% CI 1.5-10.1).175
  • A series of 116 patients with COVID-19 found that coinfection was fairly uncommon, most often with rhinovirus (6.9%), RSV (5.2%) and other coronaviruses (4.3%).58
  • Household transmission
  • A systematic review identified 44 studies of transmission within households and 10 studies of transmission within families (where not all members of the family lived in the same household). The overall estimated secondary infection rate was 16.4% in households and 17.4% among families, meaning that roughly one in six household or family contacts become infected. However, the secondary infection rates varied greatly, from a low of less than 1% in a South Korean study to a high of 45% in an Italian study. Combining the 5 US studies, 232 of 722 (32%) household contacts became infected. Spouses were at greatest risk of secondary infection, 38%. Adults were more likely to transmit infection than children, and the secondary infection rate was higher for adults (15.2%) than for children (7.9%). 234
  • In Taiwan, based on contact tracing of 2761 close contacts of 100 patients confirmed with COVID-19, there were 22 secondary infections, an attack rate of 0.7% (95% CI 0.4%-1.0%) The attack rate was highest among the 1,818 individuals exposed within 5 days of symptom onset (1.0%) compared to those with later exposure (0 cases from 852 contacts). The rate was also highest among household contacts (4.6%) and non-household family contacts (5.3%). Contacts with only pre-symptomatic exposure had a 0.9% incidence of infection. This study suggests that isolation of infected individuals is insufficient to halt transmission of COVID-19.78
  • A large South Korean contact tracing study found that the transmission rate was 11.8% for 10,592 household contacts of infected persons, compared to only 1.9% for non-household contacts. Transmission was lowest for young children age 0 to 9 years (5.3%), highest for older children age 10 to 19 years (18.6%), and was somewhere in between for adults, increasing with increasing age.155
  • In a large Indian study, the risk of transmission from an index case to an exposed contact was 10.7% for high-risk contacts, who had close social contact or direct physical contact with index cases without protective measures, and 4.7% for low-risk contacts, who were in the proximity of index cases but did not meet these criteria for high-risk exposure.189
  • Two studies examined household transmission in adults and children. In a Swiss study, adult household contacts were suspected or confirmed to have COVID-19 infection before the study child in 79% (31/39) of cases. The study child deveIoped symptoms before any other household contact in only 8% (3/39) of households.159 In a similar international study of COVID-19 transmission in 31 household clusters from China, Singapore, South Korea, Japan and Iran, investigators found that a child was the first (index) case in only 3 of the 31 (9.7%) household clusters. (preprint server, not peer reviewed)
  • A Scottish national study found that among patient-facing healthcare workers, while the numbers were small, the risk of hospitalization was much higher for them (hazard ratio 3.3, 95% CI 2.1-5.1) as was the risk for their household members (HR 1.8, 95% CI 1.1-2.9).197

Risk Factors

Risk Factor
Living in or traveling to an endemic area for SARS-CoV-2
Risk factors for severe COVID-19 infection include increasing age, comorbidities (diabetes, COPD, asthma, heart disease), elevated CRP, LDH, ALT or AST, decreased albumin, lymphopenia, and neutrophilia.
There is no independent association between vitamin D levels and the risk of infection with SARS-CoV-2, based on a large study that adjusted for multiple covariates.


Screening and Prevention

Bottom Line

  • Multiple interventions employed simultaneously (e.g., masks, distancing, contact tracing, quarantine, immunization) provide the greatest likelihood of reducing the spread of COVID-19 in a community (the so-called "Swiss Cheese" model of prevention).
  • Guidelines
  • The Johns Hopkins Center for Health Security has created guidance for state and local governments with regards to re-opening the economy.
  • CDC provides guidance for healthcare facilities and clinics.
  • The World Health Organization provides guidance regarding surveillance for SARS-CoV-2 infection in various settings.
  • Vaccines
  • A variety of approaches are being used, including inactivated or weakened coronavirus, viral vector vaccines, nucleic acid vaccines (DNA and RNA), mRNA, and protein-based vaccines. These approaches are explained in a graphical article in Nature.
  • Two mRNA vaccines from Pfizer/BioNTech and Moderna have approximately 95% efficacy against symptomatic disease and good safety. The adenovirus vectored vaccine from Johnson and Johnson/Janssen is 67% overall, but 74.4% in the US population that was studied. The Oxford -AstraZeneca vaccine was also approximately 67% effective overall. The efficacy should not be directly compared because the different vaccines were studied in different populations, at different times during the pandemic, and when different variant strains were or were not circulating.
  • Pfizer BioNTech vaccine
  • A total of 37,706 persons were randomized to the Pfizer BioNTech mRNA vaccine or placebo and had at least 2 months follow-up. There were a similar number of withdrawals between dose 1 and 2, about 300, and somewhat more in the placebo group after dose 2 (95 vs. 48). Only 2.6% were found to have antibodies due to a prior infection, and results presented below are for all patients who received 2 doses of vaccine regardless of prior infection. Participants were 49.4% female, with a mean age of 50 years. By age group, 58% were 16 to 55 years, 42% were older than 55 years, and 21% were older than 65 years. The population was fairly representative by race and ethnicity of the US as a whole, with 83% white, 9.3% Black, 4.3% Asian, and 28% Hispanic or Latinx. Efficacy was approximately 95% regardless of age, race, or ethnicity. With regards to safety, injection site pain, fatigue, and headache were common but generally mild.205 The vaccine is equally effective in adolescents age 12 to 15 years (efficacy 100%, 95% CI 75%-100%).263
  • An Israeli study matched vaccinated and unvaccinated patients, and found that the Pfizer-BioNTech vaccine was 92% effective at preventing any infection, 94% at preventing symptomatic infection, 87% for preventing hospitalization, and 92% for preventing severe COVID-19.210
  • A Kaiser Permanente study in California compared about 1 millionfully vaccinated members with 2 million unvaccinated members. While there was some waning of efficacy against infection, there was no waning in the protection against hospitalization. The authors report the vaccine’s overall effectiveness against hospitalization for COVID was 93% against delta variants and 95% against other strains. 278
  • An Israeli study compared fully vaccinated persons who had and had not received a 3rd "booster" dose of the Pfizer BioNTech vaccine. They found that the risk of any infection was much higher in the un-boosted group (adjusted risk ratio 11.3, 95% CI 10.4-12.3), as was the risk of severe infection (aRR 19.5, 95% CI 12.9-29.5). There was no added protection during the first week after the booster, and the protection rose to substantial levels about 12 to 16 days after the booster was given.276
  • Moderna vaccine
  • The mRNA-1273 SARS-CoV-2 vaccine developed by Moderna was the second of two messenger-RNA vaccines to be approved by the FDA for emergency use, approved in December of 2020. This randomized, placebo controlled trial, the basis for approval of this vaccine, enrolled 30,420 individuals. Participants were each given two injections 28 days apart of either the vaccine or placebo. Starting 14 days after the second injection, symptomatic COVID-19 infection occurred in 185 participants in the placebo group and in 11 participants in the vaccinated group. Therefore, the vaccine efficacy for preventing symptomatic COVID-19 infection was 94.1% (95% CI 89.3%- 96.8%).
  • The results were similar in subgroups such as those who had evidence of infection at baseline (2.2% of participants) and those 65 years and older. Results were similar 14 days after the first dose, indicating early protection even before the second dose was administered. There were 30 severe COVID infections, all in the placebo group, showing a high level of protection against severe COVID disease. About 84% of participants had local reactions including swelling, redness and pain, to the first injection and 89% did after the second injection.204
  • Johnson and Johnson (Janssen) vaccine
  • The vaccine was developed using a replication incompetent adenovirus encoded with a variant of the SARS-CoV-2 spike protein; this is a traditional vaccine development approach. The vaccine was evaluated in a racially and ethnically diverse group of 43,783 adults, with data for efficacy are reported for 39,058 adults. Efficacy was 66.9% (95% CI 59.0%-73.4%) overall for preventing confirmed moderate to severe COVID-19 occurring at least 14 days after vaccination. Efficacy was similar for older and younger patients. Efficacy did vary by region, though, with 74.4% efficacy seen in the US, 64.7% in Latin America, and only 52.0% in South Africa, likely due to the South African variant. For prevention of severe/critical COVID-19, the vaccine had 76.7% overall efficacy. Results were similar for prevention of cases that required hospitalization, ICU admission, mechanical ventilation, or ECMO (75.0%) although numbers are small and confidence intervals are broad. All 7 deaths attributed to COVID-19 occurred in the placebo group and were in South Africa; there were fewer all-cause deaths as well in the vaccine group in those vaccinated at least 14 days before (3 vs. 15). Like the Pfizer and Moderna vaccines, side effects included mild and transient injection site pain (49%), headache (39%), fatigue (38%), and myalgias (33%).285
  • Oxford-AstraZeneca vaccine
  • This vaccine was studied in 17,177 patients in the UK, Brazil, and South Africa and 67% effectiveness for symptomatic infection. The vaccine had 63% effectiveness in those receiving 2 standard doses, and 80.7% effective (95% CI 62.1%-90.2%) for those who received the low dose for the initial injection. Interestingly, efficacy was higher with a longer interval between the first and second vaccination: 82.4% when the second injection was given more than 12 weeks after the first injection compared to 54.9% when the second injection was given less than 6 weeks after the first injection. To find out if the initial dose provided some protection, they analyzed the effectiveness of the first dose starting 22 days after the initial injection. The first dose was 76% effective from day 22 to day 90 in preventing symptomatic infections, though it was not effective in preventing asymptomatic infections during this time period.203 A pooled analysis of randomized trial data from early 2021 found that the AstraZeneca vaccine was 67% effective for preventing symptomatic COVID, 100% effective at preventing hospitalizations, and that modeling suggested greater effectiveness with a 3 month interval between doses.203
  • Vaccine adverse effects
  • A study using a registry of 10,162,227 vaccine eligible persons compared adverse events that occurred 1 to 21 days after vaccination with adverse events that occurred in vaccinated individuals 22 to 42 days after vaccination. They studied 23 adverse events including acute myocardial infarction, Bell palsy, cerebral venous sinus thrombosis, Guillain-Barré syndrome, myocarditis/pericarditis, pulmonary embolism, stroke, and thrombosis with thrombocytopenia syndrome and found no significant difference for any of them between vaccinated and unvaccinated persons. 274
  • A study in US military personnel (largely young men) found that the incidence of myocarditis with immunization was approximately 4 to 5 cases per 100,000 immunizations.249
  • A linked dataset from 40 hospitals found that the incidence of myocarditis following mRNA vaccine administration was approximately 1 per 100,000 vaccinated persons, and of pericarditis was 1.8 per 100,000.272
  • An Israeli study found the risk of myocarditis was 2.7 per 100,000 vaccinated persons.273
  • Cerebral venous sinus thrombosis (CVST) with thrombocytopenia is a rare complication of the Janssen/Johnson & Johnson vaccine. Based on voluntarily reported data from VAERS, after 7 million doses 12 patients were identified with CVST. Most were age 18 to 39 years, and 7 had at least one risk factor for CVST (obesity, hypothyroidism, or combined oral contraceptive use). The patients became symptomatic between 6 and 15 days of the vaccine and were hospitalized between 2 and 14 days from onset of symptoms. Eleven of the 12 presented with headache and one developed a headache later. Eight of the 12 had additional venous thromboses (e.g., portal vein, internal jugular, etc.) and 7 had intracerebral hemorrhages. All of the patients were hospitalized, 10 to intensive care. At the time of publication, 5 patients were still in the hospital, 3 died, and 4 were discharged to home. The true incidence is likely higher due to the voluntary nature of VAERS reporting.259
  • CVST was also observed in patients who received the Oxford-AstraZeneca vaccine at a rate of 2.5/100,000 vaccinations in excess of that observed in the general population (RR 20, 95% CI 8.1-42). Thromboembolic events were also slightly more common (11 excess events/100,000 vaccinations).260
  • Based on voluntary reporting through the CDC VAERS system, anaphylaxis to the Pfizer-BioNTech vaccine occurs once in every 200,000 doses,207 and to the Moderna vaccine in 1 in 400,000 doses.206
  • However, a study that tracked 64,900 employees at Mass General Brigham hospital in Boston found a higher rate of 1 per 4,056 doses delivered.250
  • In a computational model, a vaccine must have high effectiveness (at least 60%) and high uptake (at least 75%) to single-handedly stop a pandemic.180
  • Vaccines and variants
  • A CDC presentation on the Delta variant was leaked to the press. It reports that the risk of infection among vaccinated persons compared to unvaccinated is 21 vs. 177 /100,000, for hospitalization was 0.1 vs. 2.52/100,000, and for mortality 0.04 vs. 0.96/100,000. Vaccine effectiveness for the Delta variant is 87% to 90% overall. However, for immunocompromised persons, the efficacy of vaccines is lower at 59% to 80%. There is also a lower effectiveness in the elderly; for example in long term care facilities the effectiveness was 70% to 75%.
  • The CDC also reported that the R0 (reproductive number, indicating the average number of new infections generated by an infectious person) for the delta variant is 5 to 10, compared with 1.5 to 3 for the ancestral strain. It causes a higher viral load and is detectable longer.
  • A case-control study in the UK found that efficacy of the Pfizer BioNTech and Oxford Astra Zeneca vaccines was good against the delta variant.270
  • Post-exposure and pre-exposure prophylaxis
  • Hydroxychloroquine (HCQ) is not effective for post or pre-exposure prophylaxis. In an RCT with 821 persons with a moderate-risk or high-risk exposure to someone with confirmed COVID-19, there was no difference between groups in the primary outcome of infection. The findings were the same at 5, 10 and 14 days.105
  • A small RCT randomized 132 healthcare workers to HCQ 600 mg daily or placebo. After 8 weeks, 4 persons in each group had confirmed cases of COVID-19, indicating no benefit. Adverse effects were more common in the HCQ group (NNTH = 6).186
  • Non-Pharmaceutical Inverventions (NPI)
  • A meta-analysis sponsored by WHO identified 172 observational studies of non-pharmaceutical interventions in healthcare and community settings. They found lower risk of infection with distancing (-10.2%, moderate certainty), facemask use (-14.3%, low certainty), and eye protection (-10.6%).121
  • Frequent, adequate handwashing for 20 seconds using warm, soapy water is strongly recommended. A simple online intervention to educate patients has been evaluated and was demonstrated to decrease transmission of respiratory tract infections in a large trial with 20,066 patients. The "GermDefence" site from this study has been adapted for COVID-19 and is freely available.47
  • A CDC study compared COVID-19 cases and deaths during the 1 to 20 days prior to implementation of bans on indoor dining and found steady increases in cases at 1 to 100 days after bans were relaxed and deaths at 61 to 100 days after relaxation in counties allowing on-premises dining. Those increases were not seen in counties that did not open up indoor dining.
  • Physical distancing has been broadly mandated around the world. A Cochrane review of 29 studies (10 modeling studies of COVID-19, 4 observational, and 15 modeling studies of SARS or MERS) has concluded that these measures are consistently effective in slowing the spread of an epidemic. It is most effective when implemented early and in conjunction with other measures like closing schools and restricting travel.55
  • A comprehensive analysis determined incidence rate ratios before and after various distancing measures (school closures, workplace closures, restrictions on mass gatherings, public transportation closure and lock-down orders) were implemented in 149 countries. They found that implementation resulted in a 13% reduction in the incidence rate ratio.156
  • A prospective cohort study of 144 geopolitical regions worldwide found little or no association with latitude, temperature, or relative humidity but strong associations with physical distancing, school closures, and banning mass gatherings.86
  • A study applying econometric approaches to the question of whether non-pharmaceutical interventions are effective studied 1717 administrative units in six countries. Their modeling concludes that implementation of all policy measures (such as school closures, social isolation guidelines, travel bans) resulted in reductions in the rate of growth of cases, with larger reductions in Iran, China, South Korea and Italy and smaller reductions in the US and France. They estimate that these measures reduced the number of confirmed cases in those 6 countries by over 60 million, with a reduction of confirmed cases in the US by 4.8 million and total cases by 60 million (assuming that many cases were asymptomatic or never confirmed).127
  • Contact tracing and isolation
  • Early detection and screening of contacts is important to prevent the spread of COVID-19 infection early during an outbreak. It is not feasible later.7
  • Contacts of COVID-19 exposed individuals should quarantine for 14 days with daily monitoring for fever and other symptoms. Persons living alone should have daily virtual contact with someone who can obtain help in the event of worsening
  • Some countries such as Taiwan and Singapore had success with early and aggressive implementation of contact tracing. A report from Taiwan on 100 index cases and 2761 close contacts found a 0.7% rate of secondary infections, all occurring with exposure during the first 5 days of symptoms. Of 852 close contacts after 5 days there were no secondary infections.78
  • An outbreak in Beijing was contained within one month using a combination of rapid response, extensive testing, quick results reporting, and comprehensive contact.166
  • Digital approaches to contact tracing using mobile phones have been developed.130
  • Facemasks
  • Mask mandates by governments
  • A CDC study compared COVID-19 cases and deaths during the 1 to 20 days prior to implementation of the policies, and periods after the policies were or were not enacted. They used multivariate analysis to adjust for other kinds of bans such as bans on gatherings and stay at home orders, and various combinations of mask and restaurant policies. They found consistent and steady reductions in cases and deaths in counties that implemented mask mandates, and steady increases in cases at 1 to 100 days after implementation and deaths at 61 to 100 days after implementation in counties allowing on-premises dining.
  • A CDC study reported on Kansas counties with and without a mask mandate. By August 17–23, 2020, the 7-day rolling average COVID-19 incidence had decreased by 6% to 16 cases per 100,000 among mask-mandated counties and increased by 100% to 12 per 100,000 among counties without a mask mandate.240
  • Surgical masks
  • During the 2003 SARS outbreak in Taiwan, wearing surgical face masks on entering the hospital, on hospital wards and in out-patient clinics prevented nearly all cases of SARS among healthcare workers.75
  • None of 41 healthcare workers who had exposure to aerosol-generating procedures for at least 10 minutes at a distance of less than 2 meters of a patient with COVID-19 became ill; most (85%) were wearing a surgical mask and the remainder were wearing N95 masks.76
  • A study of exhaled breath of patients infected with seasonal coronavirus found some evidence that surgical face masks reduce spread of both droplet particles greater than 5 microns and aerosol particles ≤5 microns in diameter.
  • A single RCT from 2015 comparing surgical masks with cloth masks or no mask found that cloth masks had the highest rates of respiratory infection.71
  • Cloth masks
  • The CDC recommended on April 2nd, 2020 that members of the public wear home-made face masks when in a public place where appropriate physical distancing may be difficult, such as a grocery store or pharmacy. Instructions for making masks are provided. However, it is not known whether this will provide a net benefit, and concern has been expressed that it might lead to abandonment of physical distancing measures by some persons. The National Academy of Science concluded that homemade masks have a filtration rate ranging from 0% to 50%.
  • A comparison of 2-layer cotton masks also found that a surgical mask was much more effective.63
  • A Chinese study found that rigorous use of face masks and daily use of disinfectants in the home reduced transmission within a household. This is particularly relevant for households where older or vulnerable family members may reside.104
  • A study of participants wearing 3-layer cloth masks found no evidence of oxygen desaturation: 96.1% before, 96.5% while wearing, and 96.3% after.199
  • A randomized trial in Denmark found that 3-layer surgical type masks were not effective in preventing COVID19 infection in the wearer. The infection rate, however, was very low in both groups and the study was underpowered: it could not exclude a relative effectiveness of up to 46%. Importantly, they also did not study if face masks prevented spread of COVID-19 from infected individuals to others which is the primary benefit. Therefore, based on other studies, the recommendation stands to use the most effective available face mask to prevent spread stands.202
  • School and university outbreaks
  • A cluster randomized trial in English schools and colleges compared two strategies: 1) 10 days of isolation after COVID-19 exposure, or 2) daily lateral flow testing with the exposed student remaining in class as long as they tested negative. During the 10-week period following randomization, the researchers tracked absences due to COVID and the rate of COVID transmission. There was no statistically significant difference in the rate of symptomatic lab-confirmed infections in the control group (59.1 per 100,000 per week) and intervention group (61.8 per 100,000 per week). Additionally, the rate of absenteeism by students and staff was low and not significantly different between the groups (1.6% vs. 1.3%). 282
  • During a heatwave, students and teachers were excused from facemasks, resulting in a large outbreak in the school. It was also clear from the physical layout that distancing was not possible. Overall, 13% of students and 17% of staff tested positive, and 43% of students and 75% of staff were symptomatic.151
  • The first large US university to return to in person instruction in early August (University of North Carolina at Chapel Hill) was forced to pivot to online instruction after 4 clusters with over 130 students in the first week of classes.
  • The CDC compared three kinds of counties: counties with a college or university holding in-person instruction, counties with a college or university holding online instruction only, and counties without a college or university. They compared several key metrics during the 21 days prior to the start of classes and the 21 days following the start of classes. Not surprisingly, incidence increased 56% in counties with in-person instruction, and decreased in non-university and remote instruction counties. Test positivity rates increased in in person instruction counties and decreased in remote instruction and non-university counties.229
  • Surface decontamination
  • The virus can also remain on surfaces like cardboard (up to 24 hours) and steel or plastic (up to 72 hours). However, it is not known how long before the level of viral pathogen on the surface is below the amount needed to cause human infection, nor the effect of environmental factors on duration.28
  • A Chinese study found that rigorous use of face masks and daily use of disinfectants in the home reduced transmission within a household. This is particularly relevant for households where older or vulnerable family members may reside.104
  • Surface decontamination using 60% or higher ethanol, 0.5% or higher hydrogen peroxide, or a dilute bleach (0.1% sodium hypochlorite) is recommended given persistence of the virus on surfaces.45
  • Personal Protective Equipment (PPE) for healthcare workers
  • Healthcare personnel doing procedures that may generate aerosols should use a fitted respirator mask (N95) as well as face shield, gloves, and gown.
  • N95 respirators can be decontaminated and re-used in times of shortage up to 3 times, based on one study, using either UV radiation or vaporized hydrogen peroxide. Use of 70% ethanol was found to be acceptable for decontamination once, but additional decontamination with ethanol resulted in a sharp drop in mask filtration performance.138
  • In a study of 420 healthcare workers sent to Wuhan for a 6-week period to care for COVID-19 patients (16 hours/week average in ICU), all were provided with PPE (protective suits, masks, gloves, goggles, face shields, and gowns) and none tested positive after two weeks of quarantine upon their return home.132
  • In a study of healthcare workers doing home visits in India, the attack rate was 12/62 (19%) prior to use of face shields and 0/50 (0%) after their use was initiated.163


Bottom Line

  • Suspect severe coronavirus infection in persons with fever, cough, and other respiratory symptoms, or less commonly fever and neurologic symptoms or thrombosis when the virus is circulating in the community. C
  • RT-PCR of multiple specimens is the most appropriate diagnostic testing method. Testing should be carried out according to guidelines from and in consultation with public health authorities.8B
  • Point-of-care PCR is highly specific, but sensitivity varies by manufacturer (76.8% for Abbott ID Now, 99.4% for Cepheid Xpert Xpress).188B
  • Rapid antigen testing has much lower sensitivity (56.2%) and should only be used in symptomatic patients with higher viral loads.188B

Differential Diagnosis

Community-acquired pneumonia caused by strep, influenza, legionella, mycoplasma, or other viruses
COPD exacerbation
Acute pulmonary edema
Interstitial lung disease
Hantavirus pulmonary syndrome
Opportunistic infections causing pneumonia-complicating HIV infection (pneumocystis carinii, histoplasmosis, or disseminated varicella pneumonia)

Diagnostic Criteria

  • The WHO R&D Blueprint Group recommends classifying patients with regards to disease severity as follows:
    • 1: Not hospitalized
    • 2: Hospitalized, not requiring supplemental oxygen
    • 3: Hospitalized, requiring supplemental oxygen
    • 4: Hospitalized requiring nasal high-flow oxygen, non-invasive mechanical ventilation, or both
    • 5: Hospitalized, requiring invasive mechanical ventilation, ECMO or both
    • 6: Death
  • The WHO defines several clinical syndromes:
    • Mild illness: uncomplicated upper RTI as well as non-specific symptoms such as fever, cough, fatigue, anorexia, myalgias, sore throat, dyspnea, nasal congestion, and headache. Rarely patients may present with diarrhea, nausea or vomiting.
    • Pneumonia (non-severe): Pneumonia but no need for supplemental oxygen and no signs of severe pneumonia. In child characterized by cough with tachypnea >60 bpm for <2 months; ≥50 bpm for 2 to 11 months; ≥40 bpm for 1 to 5 years.
    • Severe pneumonia: Adult or adolescent: fever or suspected RTI plus one of respiratory rate >30 bpm, severe respiratory distress, or O2 sat ≥93% on room air. Children with cough or difficulty breathing and at least one of O2 sat <90%, severe distress with grunting or severe retractions; inability to feed or drink, lethargy or unconsciousness, or seizures.
    • Acute respiratory distress syndrome: CXR or chest CT with bilateral opacities; respiratory failure not fully explained by cardiac failure or fluid overload; impaired oxygenation.
      • Oxygen impairment in adults: Mild ARDS: 200 mmHg <PaO2/FiO2 ≤300 mmHg (with PEEP or CPAP ≥5 cmH2O, or non-ventilated); Moderate ARDS: 100 mmHg <PaO2/FiO2 ≤200 mmHg (with PEEP ≥5 cmH2O, or non-ventilated); Severe ARDS: PaO2/FiO2 ≤100 mmHg (with PEEP ≥5 cmH2O, or non-ventilated); When PaO2 is not available, SpO2/FiO2 ≤315 suggests ARDS (including in non-ventilated patients)
      • Oxygen impairment in children (OI = oxygenation index and OSI = oxygenation Index using SpO2. Use PaO2-based metric when available. If PaO2 not available, wean FiO2 to maintain SpO2 ≤97% to calculate OSI or SpO2/FiO2 ratio): Bilevel ( NIV or CPAP) ≥5 cmH2O via full face mask: PaO2/FiO2 ≤300 mmHg or SpO2/FiO2 ≤264; Mild ARDS (invasively ventilated): 4 ≤OI <8 or 5 ≤OSI <7.5; Moderate ARDS (invasively ventilated): 8 ≤OI <16 or 7.5 ≤OSI <12.3; Severe ARDS (invasively ventilated): OI ≥16 or OSI ≥12.3.
    • Sepsis: In adults: life-threatening organ dysfunction caused by a dysregulated host response to suspected or proven infection. Signs of organ dysfunction include: altered mental status, difficult or fast breathing, low oxygen saturation, reduced urine output, fast heart rate, weak pulse, cold extremities or low blood pressure, skin mottling, or laboratory evidence of coagulopathy, thrombocytopenia, acidosis, high lactate or hyperbilirubinemia. In children: suspected or proven infection and ≥2 age based systemic inflammatory response syndrome criteria, of which one must be abnormal temperature or white blood cell count. Severe sepsis in children is defined as sepsis accompanied by cardiovascular dysfunction, ARDS, or 2 or more organ dysfunctions.181

Using the History and Physical

  • Suspect COVID-19 when the virus is circulating in the population and a patient reports signs and symptoms of respiratory tract infection. Most common symptoms are fever, cough, myalgias, and dyspnea.
  • A systematic review of 43 Chinese studies with 3600 patients reported the following frequency of symptoms: fever 83%, cough 60%, fatigue 38%, myalgias 28%, dyspnea 25%, and diarrhea 8.4% (see Table 1).102
  • Cough is typically dry; the WHO/China report found that 67% had a dry or nonproductive cough while the remainder had a productive cough with thick phlegm.
  • In a series of 202 Italian outpatients with mild to moderate COVID-9 infection, 64% reported alteration of taste or smell, and 101 of 130 with altered taste or smell described it as moderate, severe, or very severe.99 A Swiss study confirmed this, finding it in 62% of COVID-19 patients. Median day of onset was day 3, and it was most often moderate to severe. It was more common in younger and female patients.100 It is thought that anosmia occurs because the coronavirus attaches to ACE2 receptors which are very common in the nasal epithelium and olfactory bulb.
  • The increased blood clotting observed in cases of COVID-19 has caused strokes (see Prognosis | Complications) and small vessel lesions presenting as swollen, discolored toes or fingers. Lesions are similar to pernio, a vascular inflammatory disorder seen in cold, moist climates. Skin lesions are thought to be present in up to 20% of cases and have been seen in symptomatic and asymptomatic patients.
  • A systematic review of 12 studies found that GI manifestations occurred in 3.9% of patients.128
  • A study of 215 consecutive patients hospitalized for COVID-19 in China found CNS manifestations (dizziness, headache, impaired consciousness, stroke, ataxia or seizure) in 25%, while 9% had peripheral nervous system manifestations (altered taste or smell, visual impairment, neuropathic pain).122
  • In a study of 91 children with COVID-19 identified by contact tracing in South Korea, 2% remained asymptomatic, 60% had respiratory symptoms, and 55% had non-specific systemic symptoms. Viral shedding was 14 to 20 days, longer in those who were more symptomatic.170
  • Asymptomatic infections
  • A review identified 16 cohorts of COVID-19 positive individuals from Iceland, Italy, Greece, France, Japan, Argentina, and the United States, (including four ship outbreaks) ranging in size from 76 to 13,080 individuals. The percentage of people who tested positive but were asymptomatic at the time of testing ranged from a low of 6.3% in nursing home residents to highs 87.8% of occupants in a Boston homeless shelter, 87.9% on an obstetric service in New York City, and 96% of 3146 3,146 inmates in state prison systems in Arkansas, North Carolina, Ohio and Virginia. The three cohorts that came from representative samples of the population suggest the rate of asymptomatic infection is 40% to 45%. It appears that asymptomatic infection is more common in younger persons.126 A study that included over 85% of persons living in the Italian town of Vo' found that 42% of infections were asymptomatic. However, asymptomatic persons were shown to spread infection and had similar viral loads to symptomatic persons.145
  • An outbreak investigation on the USS Theodore Roosevelt found that of those who tested positive, 43% remained asymptomatic, 30% were presymptomatic, and 22% were symptomatic at the time of testing.200
  • Infants and children
  • Infants and children generally have milder illness and a milder clinical presentation. In one series of 171 children with confirmed infection with COVID-19, 41% had fever, and 16% had no signs or symptoms. Only 3 required ICU support, all of whom had serious comorbidities (hydronephrosis, leukemia, and intussusception).29 See the Managing Special Populations section for a description of characteristics of children with the multisystem inflammatory syndrome.

Selecting Diagnostic Tests

  • PCR testing
  • Infection should be confirmed by PCR testing using a nasopharyngeal swab. The WHO provides guidance regarding how to prioritize testing when resources or supplies are limited.
  • Specimens for PCR testing should be obtained from the nasopharynx and oropharynx; if upper respiratory specimens are negative but clinical suspicion remains, obtain specimens from the lower respiratory tract as well.
  • After brief training, 530 symptomatic patients were asked to collect tongue, nasal, and mid-turbinate samples, in that order. Afterwards, trained staff collected swab samples from the nasopharynx and at least one additional location. Using the staff-collected sample as “the gold standard,” self-collected samples were 90% sensitive (one-sided 97.5% CI 78-100) for tongue samples, 94% sensitive (97.5% CI 84-100) for nasal samples, and 96% sensitive (97.5% CI 87-100) for mid-turbinate samples.117 This study suggests that patient-collected samples are reasonably accurate and have the potential to decrease the frequency of exposing staff to potentially infectious material.
  • A review of 7 published studies evaluated the accuracy of PCR by days from symptom onset. Virus is detectable beginning 3 to 4 days following exposure, with the highest sensitivity in the first 5 days following symptom onset.187
  • In a Chinese study of 51 patients eventually diagnosed with COVID-19, 15 (29%) had a negative PCR but positive CT at initial presentation; PCR became positive for these patients over the next 1 to 7 days. Only one patient had a positive PCR and negative CT which became positive for viral pneumonia 3 days later.62
  • Saliva testing
  • A systematic review of studies comparing saliva with NP sampling identified 16 studies with 5922 unique patients. Overall, the risk of bias in the included studies was high for patient selection but low otherwise. Overall, the pooled sensitivity and specificity for saliva-based tests were 83% (95% credible interval [CrI] 75%-91%) and 99% (95% CrI 98%-99.8%), respectively. This was comparable to those for nasopharyngeal-based tests: 85% (95% CrI 77%-92%) and 99% (95% CrI 97.4%-99.8%), respectively.224
  • After brief training, 530 patients were asked to collect tongue, nasal, and mid-turbinate samples, in that order. Afterwards, trained staff collected swab samples from the nasopharynx and at least one additional location. Compared with samples collected by trained staff, the patient-collected specimens had high sensitivity of 90% to 96%. 117
  • Point-of-care PCR and rapid antigen test
  • A Cochrane review of 11 studies of point of care PCR tests found 95.2% sensitivity and 98.9% specificity. However, this differed by manufacturer. The Abbott ID Now (5 studies) was only 76.8% sensitive but 99.6% specific (LR+ 192, LR- 0.23). The Cepheid Xpert XPress (6 studies) was 99.4% sensitive and 96.8% specific (LR+ 31, LR- 0.01).188
  • A Cochrane review of 8 studies concluded that rapid antigen testing has much lower sensitivity (56%) but good specificity (99.5%) and should only be used in symptomatic patients with higher viral loads.
  • Real world evaluation of the accuracy of rapid antigen tests also found significantly lower sensitivity than manufacturer reports. One study evaluated the Abbott BinaxNOW rapid antigen in the Pima County, Arizona Health Department. In 827 symptomatic patients, 80% of whom were within 7 days of the onset of symptoms, sensitivity was 64% and specificity was 100%. In 2592 asymptomatic persons, sensitivity was only 36% while specificity was 99.8%.212 A second study evaluated the Quidel Sofia SARS FIA test in patients at a Wisconsin urgent care center. In patients symptomatic ≤ 5 days, sensitivity was 82% and specificity 100%. However, if symptomatic >5 days, sensitivity dropped to 55% and specificity to 97.3%. 211
  • Antibody (IgG and IgM) tests
  • The FDA requires that antibody tests be at least 90% sensitive and 95% specific. These test characteristics lead to an unacceptably high false positive rate in low prevalence situations.
  • A meta-analysis identified 40 studies of IgG and/or IgM antibody tests for SARS-CoV-2. The overall risk of bias was high for almost all studies. Based on 15 studies of ELISA assays, they found the following test characteristics: sensitivity 84%, specificity 98%, LR+ 42, LR- 0.16. Based on 17 studies of lateral flow immunoassays, they found sensitivity 66%, specificity 97%, LR+ 22, LR- 0.35. Finally, based on 13 studies of chemiluminescent immunoassays they found sensitivity 98%, specificity 97% to 98%, LR+ 49, LR- 0.02. These values reflect testing 3 or more weeks following symptom onset.150
  • A rapid systematic review from the National Academies estimates that IgM can be detected a median of 5 days after symptom onset and IgG a median of 14 days (range 10-18 days) after symptom onset.
  • An Icelandic study tracked persons who had recovered from COVID-19, and found that antibody levels persisted for at least 4 months at high levels.177
  • In an outbreak on a fishing vessel prior to vaccine availability, of 122 persons on the boat only 3 had evidence of neutralizing antibodies prior to departure. While all tested negative prior to departure, at least one false negative occurred, and 103 of 122 were eventually infected on the voyage. However, all 3 of the persons with neutralizing antibodies due to previous infection remained uninfected. The likelihood of these 3 all being among the 19 uninfected was estimated to be 0.2%.193
  • Other blood tests
  • In early reports of hospitalized patients in Wuhan City, lymphocytopenia was common (83%). The median white cell count was 3700 in patients with severe disease and 4900 in those with non-severe disease. The median lymphocyte count was 800 in severe disease and 1000 in non-severe disease.15
  • C-reactive protein, d-dimer, interleukin-6, and LDH are commonly elevated. Lymphopenia and neutrophilia are commonly seen. These blood tests also serve as risk factors for poor outcomes.
  • A systematic review of 43 studies with 3600 Chinese patients found that the most common laboratory findings in patients sick enough to be admitted to hospital were elevated c-reactive protein (69%), elevated lactate dehydrogenase (52%) and elevated d-dimer (29%) (see Table 1).102
  • Imaging
  • A systematic review of 43 studies with 3600 Chinese patients found that he most common imaging findings in patients sick enough to be admitted to hospital were ground glass opacities on chest CT (80%) and bilateral pneumonia (73%). The absence of radiographic findings was reported in 18% with mild disease and 3% with severe disease in early reports.10215
  • Up to half of patients may have normal CT findings in the first 2 days after onset of flu-like symptoms. Early radiologic changes include peripheral focal or multi-focal ground glass opacities. Late radiologic findings may include crazy paving (ground-glass opacity with superimposed interlobular septal thickening and intralobular septal thickening) and consolidation peaking 9 to 13 days after onset of symptoms, followed by slow clearing over the next month.37
  • A retrospective review of CT scans in 21 Chinese patients without severe respiratory distress identified 4 stages: 1) early stage (0-4 days after onset) with ground glass opacities; 2) progressive stage (5-8 days) with rapid spread to involve multiple lobes with ground glass opacities, crazy-paving pattern, and consolidation; 3) peak stage (9-13 days) with increased findings and dense consolidation; and 4) absorption stage (14+ days) with consolidation gradually absorbed over the next month or more, with no further crazy-paving pattern.85
  • These radiographic findings may persist for months after recovery.17 In a study of 114 patients with severe COVID-19 pneumonia, 6-month follow-up CT showed lung fibrotic-like changes in 35%; changes were associated with older age, acute respiratory distress syndrome, longer hospital stays, tachycardia, non-invasive mechanical ventilation, and higher initial chest CT score.245

Approach to the Patient


Bottom Line

  • Systemic corticosteroids are highly effective at reducing mortality in patients with COVID-19 who are mechanically ventilated (NNT = 7) or who are on oxygen (NNT = 20) but are not recommended for hospitalized patients not requiring oxygen or for outpatients.161B
  • Two studies have found that inhaled budesonide in outpatients with early disease have a shorter duration of symptoms and may have a lower risk of hospitalization, death, and the need for urgent visits.B
  • A single open-label RCT concluded that remdesivir shortens the duration of hospitalization from 15 to 11 days; there was a non-significant reduction in mortality as well (6.7% vs. 11.9%, p = 0.059, at day 15 and 11.4% vs. 15.2% at day 29). Benefit is greatest in patients requiring only low-flow oxygen.108B
  • The monoclonal antibody bamlanivimab and the combination of casirivimab and imdevimab (Regeneron) have been given emergency use authorization for treatment of outpatients not on supplemental oxygen but at high risk for severe disease. They somewhat reduce the likelihood for an ED visit or hospitalization.218B
  • In newly hospitalized patients not requiring mechanical ventilation, the Janus kinase inhibitor tofacitinib 10 mg twice daily reduced the composite of death or respiratory failure (18.1% vs. 29.0%, p = 0.04, NNT = 9).265B
  • Multiple randomized controlled trials have confirmed that hydroxychloroquine (HCQ) is not effective for severe disease, mild disease, early disease, or as post-exposure prophylaxis.105A

Drug Therapy

  • Sites to link patients with trials have been developed and are active.
  • IDSA guidelines are frequently updated:
    • Strong recommendations against using anti-malarials, alone or in combination with macrolide antibiotics, and lopinavir/ritonavir.
    • Strong recommendation to use dexamethasone (or equivalent dose of other glucocorticoids if dexamethasone is not available) in critically ill hospitalized patients and a conditional recommendation for use in patients hospitalized with severe COVID-19 infections and against its use in hospitalized patients with non-severe illness and who do not need oxygen.
    • Recommends against using tocilizumab in hospitalized patients unless they are severely or critically ill (conditional recommendation).
    • Conditional recommendation in favor of using remdesivir in patients hospitalized with severe illness and those using supplemental oxygen but not on ventilators. The IDSA recommends against using remdesivir in hospitalized patients not needing oxygen.
    • Conditional recommendation when managing hospitalized patients with severe illness who have contraindications to corticosteroids, that patients receive baricitinib with remdesivir rather than remdesivir alone.
    • Conditional recommendation to use casirivimab/imdevimab as post-exposure prophylaxis in exposed persons who are at high risk of progression to severe COVID and to use bamlanivimab/etesevimab, casirivimab/imdevimab, or sotrovimab in ambulatory high-risk persons with mild to moderate COVID.
    • Recommends against using bamlanivimab in hospitalized patients and conditionally recommends against its routine use and against routinely using casirivimab/imdevimab in ambulatory patients (outside of high-risk patients who receive counseling for these agents).
    • Recommends against using the following specific agents except in the context of a clinical trial: convalescent serum, famotidine, baricitinib plus remdesivir plus corticosteroids, and ivermectin.
  • A regularly updated "living" guideline systematic review from the WHO makes a strong recommendation in favor of using Interleukin-6 receptor blockers (tocilizumab and sarilumab) for persons hospitalized with severe and critical covid-19, based on high-certainty evidence of benefit for mortality and mechanical ventilation. They make a strong recommendation for the use of corticosteroids only in patients hospitalized with severe or critical illness.They also make strong recommendations against using hydroxychloroquine (HCQ) and against using lopinavir-ritonavir in patients with COVID, regardless of disease severity. The WHO also recommends against using ivermectin for any patients with COVID except as part of a clinical trial. Additionally, due to low quality evidence, the WHO makes a weak recommendation against using remdesivir.165
  • The European Respiratory Society also has a living guideline for hospitalized patients. They recommend using corticosteroids for patients requiring oxygen, noninvasive ventilation, or mechanical ventilation, and against using them for anyone else. They recommend IL-6 inhibiting monoclonal antibodies for patients requiring oxygen or ventilatory support and not for others (conditional recommendation based on low quality evidence). They recommend anticoagulation for hospitalized patients, and high flow nasal cannula or noninvasive CPAP through a facemask or helmet for patients with hypoxemic respiratory failure not requiring mechanical ventilation. They make no recommendation regarding use of remdesivir for patients not requiring mechanical ventilation due to conflicting evidence from randomized trials, and recommend against offering it for patients requiring mechanical ventilation. Finally, they recommend against use of HCQ, azithromycin, azithromycin + HCQ, colchicine, lopinavir-ritonavir, and interferon-beta.264
  • Systemic corticosteroids
  • The UK RECOVERY trial randomized over 11,500 hospitalized patients with COVID-19 to one of 6 arms: azithromycin, lopinavir-ritonavir, tocilizumab, convalescent plasma, low dose dexamethasone (6 mg once daily for 10 days), or usual care. This study reports the results for the primary outcome of 28-day mortality between 2104 patients randomized to low dose dexamethasone and 4321 patients randomized to usual care. The primary outcome of 28-day mortality was significantly reduced overall (22.9% vs. 25.7%, rate ratio 0.83, 95% CI 0.75-0.93, NNT = 36). The degree of benefit was strongly associated with the severity of illness. For patients requiring mechanical ventilation mortality reduction was greatest (29.3% vs. 41.4%, RR 0.64, 95% CI 0.51-0.81, NNT = 8), while patients requiring oxygen but not mechanically ventilated benefitted somewhat less (23.3% vs. 26.2%, RR 0.82, 95% CI 0.72-0.94, NNT = 35). No benefit, and in fact a trend toward harm, was observed for hospitalized patients not requiring oxygen or mechanical ventilation (mortality 17.8% vs. 14.0%, RR 1.19, 95% CI 0.91-1.55). 161
  • The efficacy of corticosteroids for reducing mortality in patients requiring mechanical ventilation or oxygen was also supported by a meta-analysis of 7 trials with over 1700 patients.172
  • The REMAP-CAP trial randomized 384 patients to fixed-dose hydrocortisone, shock-dependent dosing of hydrocortisone, or no cortisone. The mean organ support-free days was 11.5 days for the fixed-dose hydrocortisone group, 9.5 days for the shock-dependent hydrocortisone group and 6 days for the no hydrocortisone group.171
  • Antiviral drugs
  • Remdesivir
  • The Adaptive COVID-19 Treatment Trial (ACTT) multicenter, multinational study recruited 1063 hospitalized patients with SARS-CoV-2 infection and evidence of lung involvement. They were randomized to remdesivir (200 mg IV on day 1 and then 100 mg IV daily for up to 10 days or hospital discharge) or matching placebo injection. The primary outcome was time to recovery, which was significantly faster in the remdesivir group (11 vs. 15 days, p <0.001). Clinical improvement was also more likely in the remdesivir group using a 7-category ordinal scale. There was also a trend toward lower mortality in the remdesivir group (6.7% vs. 11.9% at 15 days, p = 0.059, NNT = 21). Patients receiving high-flow oxygen, non-invasive mechanical ventilation, or mechanical ventilation did not appear to benefit, and benefit was greatest in those requiring low flow oxygen. Harms or other outcomes such as need for mechanical ventilation were not reported. 108
  • However, the WHO SOLIDARITY trial did not find a benefit to use of remdesivir. This trial took place in 405 hospitals in 30 countries, with 2750 randomized to remdesivir, 651 to interferon + lopinavir, 1412 to interferon only, 1411 to lopinavir + ritonavir, 954 to hydroxychloroquine, and 4088 to no study drug. They found no mortality benefit for any of the drugs. Looking specifically at remdesivir, there were trends toward benefit for non-ventilated patients (RR 0.80, 95% CI 0.63-1.01) and a trend toward harm (RR 1.16, 95% CI 0.85-1.60) for patients on mechanical ventilation.236
  • Another study randomized 397 patients to 5 days or 10 days of therapy and concluded that there was no difference in the likelihood of clinical improvement. The study did not mask or conceal allocation, and as a result a larger number of sicker patients ended up in the 10-day group. In the unadjusted analysis the 5-day group did better, but this difference was no longer significant after adjusting for baseline differences.118
  • Chinese researchers randomly assigned 237 patients hospitalized with COVID-19 pneumonia and hypoxia who were within 12 days of symptom onset to receive intravenous remdesivir (200 mg on day 1 followed by 100 mg on days 2-10 in single daily infusions) or the same volume of placebo infusions for 10 days in a 2:1 ratio. The treating teams were allowed to administer other therapies. In this study, there was no statistically significant improvement in time to recovery. While about ⅔ of patients in each group reported adverse events, more remdesivir-treated patients stopped treatment early due to adverse events than did placebo-treated patients (12% vs. 5%). They found no difference in mortality (12% vs. 13%). 109
  • A randomized trial in 199 hospitalized patients with severe COVID-19 infection found no benefit from lopinavir-ritonavir. There was no difference in time to clinical improvement (16 days median in both groups. There was a small but non-significant reduction in mortality in the treatment group (19.2% vs. 25.0%, p = NS).26
  • Lopinavir-ritonavir
  • The RECOVERY trial found that lopinavir-ritonavir was not effective. A total of 1616 patients received daily lopinavir-ritonavir (400 mg and 100 mg, respectively) and 3424 received usual care. The authors report that 23% of the patients allocated to lopinavir-ritonavir died by 28 days compared with 22% of control patients (p = ns).238
  • Janus kinase inhibitors
  • The Janus associated kinase inhibitor ruxolitinib 5 mg twice daily was studied in a small placebo-controlled randomized trial of 43 patients with severe COVID-19. The time to clinical improvement, assessed in a non-blinded manner, was 12 days in the treatment group and 15 days in the placebo group. There were 3 deaths in the treatment group and 0 in the placebo group. While perhaps promising, the small sample size makes it impossible to draw conclusions about efficacy.111
  • A trial randomized 1033 hospitalized patients with COVID-19 lower respiratory tract infection to remdesivir alone or remdesivir plus baricitinib, a Janus kinase 1 and 2 inhibitor. Patients who received remdesivir plus baricitinib recovered a median of 1 day faster than those who received remdesivir plus placebo (7 days vs. 8 days; p = .03). The outcome was most prominent in patients with a baseline ordinal score of 6 (those receiving noninvasive mechanical ventilation or high-flow oxygen) with a recovery time of 10 days compared with 18 days (RR 1.51; 1.10-2.08). The odds of improvement in clinical status at day 15 were also greater in the remdesivir plus baricitinib group (odds ratio [OR] 1.3; 1.0-1.6), again more likely in patients with a baseline ordinal score of 6 (OR 2.2; 1.4-3.6). Serious adverse events were less frequently noted in the remdesivir plus baricitinib group and there was no significant mortality difference between the 2 groups. 220
  • A study in Brazil randomized 289 adults with PCR confirmed COVID-19 and radiographic evidence of pneumonia to tofacitinib 10 mg or placebo twice daily for 14 days or until hospital discharge. Those receiving mechanical ventilation or ECMO were excluded. Patients were also receiving glucocorticoids (79%) and all were anticoagulated. Monoclonal antibodies, IL-6 or IL-1 inhibitors, interleukins, and other Janus kinase inhibitors were not allowed. A total of 289 patients were randomized to tofacitinib 10 mg or placebo twice daily for 14 days or until hospital discharge. The primary outcome was the composite of death or respiratory failure at 28 days which was significantly less likely in the tofacitinib group (18.1% vs. 29.0%, p = 0.04, NNT = 9). Deaths were numerically less likely but this was not a statistically significant difference (2.8% vs 5.5%, HR 0.49, 95% CI 0.15-1.63). There was no significant difference between groups in serious adverse events or serious secondary infections.265
  • Monoclonal antibodies targeting IL-6
  • Monoclonal antibodies targeting IL-6 include tocilizumab, sarilumab, and siltuximab. A systematic review identified 27 good quality RCTs with 10,930 patients (19 with 8048 for tocilizumab, 9 with 2826 of sarilumab, and 1 with 149 of siltuximab). All-cause mortality at 28 days was lower in those receiving IL-6 antagonists (21.8% vs. 25.8%; NNT = 25, 95% CI 18-42). When you look at the specific agents,only tocilizumab significantly decreased 28-day mortality (22.3% vs. 27.3%; NNT = 21, 95% CI 15-33). There was no overall mortality benefit to IL-6 antagonists without corticosteroids. The co-intervention with steroids is probably why the authors report an “association” between IL-6 antagonist use and lower mortality.275
  • A trial randomized 377 hospitalized patients on supplemental oxygen (room air O2 saturation <94%) in a 2:1 ratio to 1 or 2 doses of tocilizumab or placebo. More patients in the placebo group received dexamethasone (67.2% vs. 55.4%). For the primary efficacy outcome, significantly fewer patients in the tocilizumab group progressed to mechanical ventilation or death by day 28 (12% vs. 19%; hazard ratio 0.56; 95% CI 0.33-0.97: P = .04) although this was driven entirely by less need for mechanical ventilation. 219
  • Data from the RECOVERY trial support the use of tocilizumab for more severely ill patients. Patients had clinically suspected or laboratory confirmed SARS-CoV-2 infection and evidence of clinical deterioration: O2 <92% on room air or receiving oxygen therapy, and CRP levels ≥75 mg/L). They were randomized to a single weight-based parenteral dose of tocilizumab (n = 2022; dosing: 800 mg if weight >90 kg; 600 mg if weight >65 and ≤90 kg; 400 mg if weight >40 and ≤65 kg; and 8 mg/kg if weight ≤40 kg) or to usual care (n = 2094). Over 80% of these patients were also receiving systemic corticosteroids. The main outcome, 28-day mortality, was lower in tocilizumab-treated patients (30.7% vs. 34.8%; NNT = 25, 95% CI 15-82). Fewer tocilizumab-treated patients required mechanical ventilation (15.1% vs. 19.1%; NNT = 26, 95% CI 16-68).262
  • Monoclonal antibodies specific to coronaviruses
  • Bamlanivimab is a monoclonal antibody that targets a region of the coronavirus spike protein. Researchers recruited 577 ambulatory patients with confirmed mild to moderate COVID-19 infections. Over the first two months of the study, the researchers randomized patients to a single infusion of placebo or one of three doses of bamlanivimab ranging from 700 to 7000 mg (n = 309) to another wing where patients received a combination of bamlanivimab plus etesevimab (2800 mg each, n = 112) or placebo (n = 156). The secondary outcome of hospitalizations or ED visits occurred in 5.8% of placebo-treated patients and in 1.4% of patients receiving bamlanivimab either as monotherapy or as combination therapy (NNT = 23, 95% CI 11-85).218 The WHO living systematic review found a similar NNT of 21 to prevent one hospitalization in high risk outpatients.280 However, it was not effective in hospitalized patients.232
  • A combination of two monoclonal antibodies, casirivimab and imdevimab (trade name Regeneron) was given emergency use authorization on November 21, 2020 for outpatients 12 years and older with mild to moderate COVID-19 at risk of progression to severe disease. It is given as a single IV infusion. A systematic review found an NNT of 24 to prevent one hospitalization.280
  • Inhaled corticosteroids
  • In the STOIC trial, researchers identified adults with less than 7 days of cough and either fever or anosmia or both (n = 146), of whom 94% had a positive swab for SARS-CoV-2. They randomized them to budesonide 400 mcg actuations, with two actuations twice daily, or to usual care. In the entire population, the primary outcome of urgent emergency department visit or hospitalization occurred less often in the budesonide group (3% vs. 15%, p = 0.009, NNT = 8). Time to recovery was also about 1 day faster, and symptom resolution was also more rapid. Adverse events (4 with sore throat, 1 dizziness) were minor and self-limited.256
  • The larger PRINCIPLE trial in the UK randomized (open label) patients to usual care (n = 1988), usual care + inhaled budesonide 800 mcg twice daily for 14 days (n = 1073), or usual care plus other treatments (n = 1639). Time to self-reported recovery was significantly faster in those given budesonide (2.94 days, 95% Bayesian credible interval 1.19-5.12 days). For the outcome of hospitalization or death, the budesonide group showed a trend toward benefit (6.8% vs. 8.8%, absolute difference 2.0%, 95% BCI -0.2%-4.5%). 268
  • Antibiotics
  • There is no evidence to recommend azithromycin for treatment of COVID-19. The RECOVERY trial randomized 2582 patients to receive azithromycin (500 mg once daily for 10 days or until discharge) and 5181 patients to usual care alone. The 28-day mortality was the same (22%) for each treatment group. They also found no significant difference in length of stay or survival to discharge. Additionally, they found no improved outcomes in patients on mechanical ventilation.213
  • For patients who are mechanically ventilated and in respiratory failure, Surviving Sepsis Campaign guidelines recommend that empiric antimicrobial therapy be considered based on expert opinion.
  • Anticoagulants and antiplatelet agents
  • In a retrospective cohort study, 4297 patients were hospitalized with COVID-19 between 3/1/20 and 7/31/20 in the US Veterans Affairs health system. The researchers compared patients who had an initial dose of prophylactic anticoagulation in the first 24 hours of hospitalization with those who did not, with a primary outcome of 30 day mortality. The analysis used a propensity score matched approach to adjust for comorbidities, tobacco use, medications, and laboratory results. Overall, 84.4% of patients received anticoagulation within 24 hours of admission; of those receiving it, most received subcutaneous heparin (30%) or enoxaparin (69%). In the propensity score matched analysis, the 30-day mortality was significantly lower in the group receiving early thromboprophylaxis (14.3% vs. 18.7%, NNT = 23).214
  • A study in 12 US centers randomized patients to standard therapy (prophylactic or intermediate dose LMWH or unfractionated heparin) or therapeutic dose enoxaparin (1 mg/kg twice daily). Patients were at high risk for VTE based on elevated d-dimer or coagulopathy score. The primary outcome of symptomatic venous thromboembolism or arterial thromboembolism or death from any cause within 30 days occurred in less often with therapeutic dose enoxaparin (29% vs. 42%, RR 0.68; 95% CI 0.49-0.96). The incidence of major bleeding was higher with therapeutic dose enoxaparin (4.7% vs. 1.6%). The primary outcome was reduced in non-ICU patients (17% vs. 36.1%, RR 0.46; 95% CI 0.27-0.81) but not in ICU patients (51.1% vs. 55.3%, RR 0.92; 95% CI 0.62-1.39).279
  • Fluvoxamine
  • An RCT in 11 Brazilian cities randomized high risk outpatient adults with symptomatic COVID to receive 10 days of the SSRI fluvoxamine (100 mg twice daily; n = 741) or placebo (n = 756). The main outcome of hospitalization that could include admissions for observation occurred in 11% of the fluvoxamine-treated participants compared with 16% of the controls (NNT = 20, 95% CI 12-61). There were no significant differences in adverse event rates between the two groups.281
  • Colchicine
  • In a small Greek trial, 105 hospitalized COVID-19 patients were randomized to receive colchicine or standard medical therapy for up to 3 weeks. The primary clinical outcome was time from baseline to clinical deterioration, defined as a 2-grade deterioration on a 7-point ordinal scale. Mean event-free survival time was 18.6 days in the control group vs. 20.7 in the colchicine group (log rank P = .03). Fewer patients in the colchicine group had clinical deterioration (1.8% vs. 14%, p = 0.02). A larger trial is needed to see if colchicine leads to meaningful clinical benefit for COVID-19 patients.137
  • The COLCORONA study recruited high-risk patients with COVID who were not yet hospitalized and considered not likely to need hospitalization. They randomized 4488 patients to colchicine (0.5 mg twice daily for the first 3 days followed by once daily) or placebo.286 Using intention-to-treat, they found the rate of a composite 30-day outcome (death or COVID hospitalization) was similar in both groups of patients (4.7% vs. 5.8%, respectively; p = 0.08). There was also no difference in the rate of each of these events individually. When they restricted their analysis to the 4159 patients with a positive PCR result, the differences became statistically significant for the composite outcome (4.6% vs. 6.0%, respectively; NNT = 71, 95% CI 36-1974), driven mainly by a barely statistically significant reduction in hospitalizations (4.5% vs. 5.9%). In the real world, clinicians seeing a high-risk patient with symptoms and a COVID exposure are more likely to apply data from the main analysis and not this last one. More colchicine-treated patients developed pulmonary emboli (0.5% vs. 0.1%; NNTH = 244, 95% CI 124-1163).
  • Chloroquine and hydroxychloroquine
  • There was initially interest and widespread use of hydroxychloroquine (HCQ) for COVID-19. However, randomized trials to date have all been negative, with no effect on mortality, viral shedding, or symptom duration.95 and well-designed observational studies have also failed to find any benefit. However, they do find harm, primarily prolonged QT intervals and cardiac arrhythmia and in some cases increased mortality.
  • The UK RECOVERY Trial randomized hospitalized patients to HCQ (n = 1561) or usual care (n = 3155). Patients in the HCQ group got two 800 mg loading doses 6 hours apart, followed by 400 mg twice daily for up to 9 days. There was no difference between groups with regards to 28-day mortality (26.8% HCQ vs. 25.0% usual care, rate ratio 1.09, 95% CI 0.96-1.23). Patients in the HCQ group had a longer time to hospital discharge (16 vs. 13 days) and were less likely to be discharged alive within 28 days (60.3% vs. 62.8%, RR 0.92, 95% CI 0.85-0.99, NNT = 40). Patients in the HCQ group were also more likely to experience the combined outcome of mechanical ventilation or death (29.8% vs. 26.5%, RR 1.12, 95% CI 1.01-1.25, NNT = 30).246
  • A living systematic review in BMJ confirmed an increased risk of adverse events with HCQ.165
  • A Brazilian study randomized 665 hospitalized patients to HCQ (400 mg twice daily), HCQ plus azithromycin 500 mg daily, or usual care. Of 665 patients, 504 had confirmed COVID-19 and made up the primary study population. They found no difference between groups for any outcomes including length of stay or mortality (although there were only 2 deaths).149
  • A Spanish study randomized 293 outpatients with onset of symptomatic COVID-19 less than 5 days previously to HCQ 800 mg on day 1 and 400 mg once daily for 6 more days. There was no difference in viral loads at 3 and 7 days, or in symptoms or days to resolution. Adverse events were more common in the HCQ group.157
  • A US study randomized symptomatic outpatients with confirmed or suspected COVID-19 (mostly epidemiologically linked cases) to HCQ 800 mg loading dose followed by 600 mg per day for 4 days or placebo. There was no difference between groups in symptoms duration or severity, hospitalizations, or deaths.158
  • A Brazilian study randomized 81 patients to high-dose chloroquine (600 mg twice daily for 10 days) or low-dose chloroquine (450 mg twice daily on day 1 and once daily on days 2-5). The high-dose group had a high rate of prolonged QT (25%) and a trend toward higher mortality than the low-dose group and overall mortality similar to historical control rates of death in untreated patients.141
  • Ivermectin
  • A systematic review identified 24 trials of ivermectin with 3328 participants. The included trials assessed a range of outcomes, such as viral clearance and inflammatory markers. Additionally, the trials used a variety of comparators, including usual care, placebo, hydroxychloroquine, and azithromycin. Overall, the studies were at low to moderate risk of bias. Six trials reported time to clinical recovery, variably defined within the studies. Three of the six trials reported that ivermectin-treated persons recovered more quickly (range = 1-7 days). Seven trials reported hospital length of stay: although the data were heterogeneous, the ivermectin-treated patients had a shorter length of stay by 1.6 days. Eleven trials (2127 participants) reported mortality in persons with moderate to severe infections. The authors report that mortality was lower in the ivermectin-treated patients than in the control patients (3% vs. 8.7%). However, since publication, the authors discovered that 1 of the included studies, with 400 persons (representing 19% of the patients in the studies that reported mortality), has since been withdrawn due to fraudulent data. The authors report they will re-analyze all their data and issue an updated report, so stay tuned. Re-analysis notwithstanding, the messiness of the remaining studies and concerns over fraud indicate that better studies are needed to determine the true role of ivermectin in managing patients hospitalized with COVID-19.287
  • ACE inhibitor and angiotensin receptor blockers
  • Because the SARS-CoV2 virus binds human angiotensin converting enzyme (ACE)-2 receptors, it has been hypothesized that the increased risk in patients with hypertension may be due to use of ACE inhibitors increasing the number of these receptors.33 Others hypothesize that ACE inhibitors and angiotensin receptor blockers (ARBs) might be useful as therapy.4632Current guidance recommends continuing ACE inhibitors and ARBs in patients with COVID-19 infection as the observational studies summarized below found no evidence of harm.
  • Among 12,594 patients in New York who were tested for COVID-19, 5894 (46.8%) were positive and 1002 (17.0%) had severe illness. Antihypertensive medication (ACEI, ARB, beta-blockers, calcium-channel blockers, or thiazide diuretics) use was higher in the COVID-infected patients than in those who tested negative. However, using a propensity analysis to account for known confounders, the authors found no association between any antihypertensive class and contracting COVID-19 or in the risk of having severe disease.80
  • A Danish cohort study found no difference in COVID-19 mortality among 4480 patients when comparing those taking vs. not taking an ACEI or ARB after adjusting for age and comorbidities. In fact, the non-significant trend was for lower mortality in ACEI/ARB users (hazard ratio 0.83, 95% CI 0.67-1.03). The same researchers also found no difference in the likelihood of becoming infected in a nested case-control study with 571 COVID-19 patients compared to 5710 age and sex matched controls.131

Other Treatment

  • Guidelines are available from the World Health Organization, Infectious Disease Society of America, and Massachusetts General Hospital.
  • Guidance for using telehealth consultations was published in BMJ and includes a suggested algorithm.59 CDC also provides guidance for telehealth consultations.
  • Oxygen supplementation
  • Adults: Give supplemental oxygen to patients with severe acute respiratory infection (SARI) and respiratory distress, hypoxemia or shock, with a target O2 of >94% (WHO interim guidance).
  • Children: Give supplemental oxygen with target O2 of >90%. If emergency signs such as obstructed breathing or apnea, severe respiratory distress, central cyanosis, shock, coma or convulsion provide airway management and oxygen to target O2 >94%. (WHO interim guidance).
  • The Surviving Sepsis Campaign Guidelines recommend supplemental oxygen if SpO2 <90% and to maintain it at no higher than 96%. If needed, high flow nasal cannula is recommended over noninvasive positive pressure ventilation depending on availability.
  • Use of prone positioning appears to improve oxygenation in patients with COVID-19 who are in respiratory distress, preventing the need for intubation and mechanical ventilation in many patients.142 Randomized trials are lacking.77 Use of home pulse oximetry should be considered, if available, for patients recovering from COVID-19 at home as hypoxic patients do not always feel dyspneic and may wait too long at home before coming in for help.
  • Clinical trials to evaluate nitric oxide as a treatment for patients with COVID-19 and ARDS are underway, as well as trials to evaluate the efficacy of nitric oxide inhalation for healthcare personnel to prevent infection.
  • Convalescent plasma
  • The UK RECOVERY trial indentified patients with clinically suspected or laboratory confirmed SARS-CoV-2 infection and evidence of clinical deterioration: oxygen saturation <92% on room air or receiving oxygen therapy, and CRP levels ≥75 mg/L. They were randomized to convalescent plasma (n = 5795) or usual care (n = 5763). The protocol called for two units of high-titer convalescent plasma, the first administered as soon as possible after randomization and the second unit the following day. Over 90% of patients in each group also received corticosteroids. The main outcome, 28-day mortality, was the same (24%) in both groups, as was the proportion of those achieving the composite outcome of progression to mechanical ventilation or death (29%).266
  • An RCT concluded that convalescent plasma can prevent progression of COVID-19 disease in the population of older adults with mild symptoms when administered within 72 hours. They randomized 160 adults aged 65 to 74 with comorbidities or age 75 and older regardless of comorbidities, all of whom had mild symptoms, to convalescent plasma or placebo infusion within 72 hours of onset. The plasma had a high titer of antibodies. The primary endpoint was severe respiratory disease, defined as a respiratory rate of 30 breaths per minute or more, an O2 saturation of less than 93% on room air, or both. Fewer patients in the treatment group developed severe respiratory distress (16% vs. 31%, p = 0.03, NNT = 6).228
  • In a randomized trial of 33 hospitalized patients with severe COVID-19 pneumonia (PlasmAr), there was no benefit of convalescent plasma in this group that was also being treated with corticosteroids.239
  • In an open label trial in 464 inpatients at 39 Indian hospitals, patients randomized to convalescent plasma were no more or less likely to progress to severe disease or die than those receiving standard care (19% plasma vs. 18% control).195
  • Stem cell therapy
  • Several companies are investigating stem cell therapy with trials getting underway.
  • Sepsis and shock
  • See detailed recommendations in the WHO interim guidance document and the Surviving Sepsis Campaign Guidelines.
  • Monitor patients for signs of clinical deterioration and sepsis. Monitor hematologic parameters and biochemistries as needed to monitor for complications such as acute hepatic or renal injury, shock, or acute cardiac complications. (WHO interim guidance)
  • Avoid fluid overload by using conservative fluid management in patients with SARI but no evidence of shock. (WHO interim guidance)
  • For acute management of shock, assess fluid responsiveness by measuring dynamic parameters, and use a conservative fluid strategy with balanced crystalloids over colloids or unbalanced crystalloids. Norepinephrine is recommended as the first-line vasopressor, with vasopressin or epinephrine as second line if norepinephrine is not available (SSCG,
  • For patients with severe pneumonia complicated by sepsis, even if suspected of having COVID-19, appropriate empiric antimicrobials should be given promptly.
  • See WHO interim guidance for details regarding management of ARDS in the ventilated patient, including prone ventilation if severe ARDS and using lower tidal volumes (4 to 8 ml/kg predicted body weight) and lower inspiratory pressures (plateau pressure less than 30 cm H20).
  • The Surviving Sepsis Guidelines generally agree with the WHO interim guidance regarding management of ARDS in ventilated patients, noting that a higher PEEP is recommended over lower PEEP strategy. If severe ARDS and continued hypoxemia, a trial of an inhaled pulmonary vasodilator can be considered, as well as referral to an ECMO center if available.
  • Test of cure
  • For outpatients: When ample testing supplies are available, patients may be considered cured using a test-based strategy: clinically recovered (no fever without use of antipyretics and no cough or dyspnea) plus two negative PCR tests 24 hours apart. A symptom-based strategy omits PCR testing but recommends that isolation be maintained for at least 10 days after illness onset and at least 3 days (72 hours) after recovery. The CDC defines recovery as resolution of fever without the use of fever-reducing medications with progressive improvement or resolution of other symptoms. (CDC interim guidance)
  • For hospitalized patients: A test-based strategy as described above is recommended, with recovery from fever without antipyretics, recovery from respiratory symptoms, and two negative PCR tests at least 24 hours apart. (CDC interim guidance)
  • A briefing paper from the National Academies of Medicine and Sciences suggested two sequential negative PCR tests be used to indicate that virus is no longer being shed.


Bottom Line

  • Overall prognosis
  • Approximately 85% to 90% experience a mild illness, while 10% to 15% experience severe disease requiring hospitalization, including 5% who require ICU admission.54 22
  • In a report of the initial 41 confirmed cases of COVID-19 in Wuhan, China, days from illness onset to dyspnea in 21 patients was 8 days (5.0-13.0) and to ARDS, occurring in 11 patients, was 9 days (8.0-14.0).41
  • Viral shedding persisted a median of 20 days in hospitalized patients who recovered, and as long as 37 days.18
  • Mortality
  • Mortality was very high early in the pandemic. In a retrospective cohort study of 191 adults hospitalized in Wuhan, China with confirmed COVID-19 infections who had been discharged or had died by January 31, 2020, 72% were discharged and 28% died as inpatients.18
  • The mortality estimates have changed as the pandemic evolved. This is due in part to incomplete case ascertainment, underreporting of deaths, the evolution of newer strains, and of course changes in therapy. For example, a study comparing monthly rates among hospitalized patients at a large New York medical center found that the case fatality ratio decreased from 25.6% in March to 7.6% in August, after adjusting for patient factors and severity of illness on admission.196
  • Based on modeling by the COVID-19 research unit at Imperial College in London, the infected fatality ratio is estimated to be 0.9% overall (95% credible interval 0.4% to 1.4%). It is lowest in children (0.002%), and is higher with increasing age (0.08% for 30-39, 0.60% for 50-59, 2.2% for 60-69, 5.1% for 70-79 and 9.3% for 80 and older). Mortality is also higher in those with comorbidities such as hypertension, diabetes and chronic cardiopulmonary conditions. It is important to note that any estimates of fatality ratios during a pandemic will be in accurate and should be considered preliminary.
  • The same group published a more detailed and updated report on March 30, 2020. They provide updated estimates of the infection fatality ratio (IFR, deaths/all infections including mild and asymptomatic) and the case fatality ratio (CFR, deaths/symptomatic or confirmed infections). They estimate a mean duration from symptom onset to death of 18 days, and for survivors the time from symptom onset to hospital discharge of 25 days. They do a good job of trying to adjust for biases in the data due to oversampling of severe cases early in a pandemic, failure to adjust for age, and the lag between case identification and death. The overall IFR is estimated to be 0.66%, and the overall CFR 1.38%. The CFR increases from 0.06% for those in their 20’s to 0.15% in their 30’s, 0.30% in their 40’s, 1.3% in their 50’s, 4.0% in their 60’s, 8.6% in their 70’s, and 13.4% for those 80 and older. The proportion hospitalized increased from 1% for patients in their 20’s to 4% in their 40’s to 12% in their 60’s.65
  • A registry study of 8 million admissions to 1030 US hospitals of non-pregnant adults age 18 to 34 years found 3222 admissions, of whom 21% required intensive care, 10% mechanical ventilation, and 2.7% died.185
  • A Canadian retrospective cohort study found that compared with ancestral strains, N501Y-positive variants (Alpha/B.1.17, Beta/B.1.351 and Gamma/P.1) were associated with a 52% (95% CI 42%-63%) increase in the likelihood of hospitalization, an 89% (95% CI 67%-117%) increase in the risk of ICU admission, and a 51% (95% CI 30%-78%) in mortality. The Delta variant was associated with a 108% (95% CI 78%-140%) increase in the risk of hospitalization, 235% (95% CI 160%-331%) increase in the risk of ICU admission, and 133% (95% CI 54%-231%) increase in the risk of death271.
  • It is known that some variants appear to spread more easily, such as the B.1.1.7 ("UK variant"). It appears to be transmitted about 40% faster than the original strain in the US, and may become the dominant strain in the US.216 In one analysis, 384 of 2,583 deaths were due to the B.1.1.7 strain, and the relative hazard of death within 28 days was 1.35 (95% CI 1.08-1.68). Another case-control analysis of community testing data linked to mortality data found a mortality hazard ratio of 1.91 (95% CI 1.35-2.71). In a matched cohort analysis of 14,939 people infected with the B. strain compared to 15,555 infected with the original strain, there were 104 deaths (0.2%) compared to 65 deaths (0.1%), a mortality ratio of 1.65 (95% CI 1.21-2.25). However, in a hospital based study, the risk of death from the B.1.1.7 strain was not statistically increased (odds ratio 0.63 (95% CI 0.20-1.69), though this small study could not exclude an excess mortality of up to 69%.215
  • Morbidity
  • A VA study compared outcomes for patients hospitalized with COVID-19 with those for patients hospitalized with seasonal influenza in the 3 previous years. After adjusting for potential known confounders, those with COVID-19 were more likely to experience complications such as acute kidney injury (37.2% vs. 29.0%; OR 1.5, 95% CI 1.4-1.7), severe septic shock (8.8% vs. 2.4%; OR 4.0, 95% CI 3.4-4.8), pulmonary embolism (3.2% vs. 2.1%; OR 1.5, 95% CI 1.2-1.9), and arrhythmias and sudden cardiac death (3.8% vs. 2.2%; OR 1.8, 95% CI 1.4-2.2). More COVID-19 died (18.6% vs. 5.3%; HR 5.0, 95% CI 4.4-5.6), required mechanical ventilation (15.0% vs. 4.2%; HR 4.0, 95% CI 3.5-4.5), and were in intensive care units (36.8% vs. 18.6%; HR 2.4, 95% CI 2.3-2.6). 235
  • Prognosis in subgroups
  • In the UK, standardized mortality rates were less than 1.0 in whites and were higher in persons of Black African (SMR 3.2), Black Caribbean (2.2), Pakistani (3.3), Indian (1.7), and Bangladeshi (2.4) origin.103
  • A study comparing mortality rates for Black and white patients in Louisiana found that after adjusting for severity at presentation, age, comorbidities and socioeconomic factors there was no difference in case fatality ratios.107 A second larger study in a US 92 hospital system similarly found no difference in mortality after adjusting for age, sex, insurance status, comorbidity, neighborhood deprivation, and site of care (HR for Black vs. white 0.93, 95% CI 0.83-1.09).164
  • A study compared mortality in Black and white patients hospitalized in different hospitalis for COVID-19. After fully adjusting the analysis, they found no difference in outcomes between Black and white patients in the same hospital or health system. However, they also concluded that Black patients are more likely to be hospitalized in hospitals with higher overall mortality rates.267
  • A study