Rapid Expert Consultation Update on SARS-CoV-2 Surface Stability and Incubation for the COVID-19 Pandemic (March 27, 2020)
March 27, 2020
Kelvin Droegemeier, Ph.D.
Office of Science and Technology Policy
Executive Office of the President
Eisenhower Executive Office Building
1650 Pennsylvania Avenue, NW
Washington, DC 20504
Dear Dr. Droegemeier:
You requested an update and elaboration on our previous rapid expert consultation dated March 15, concerning issues of virus survival on surfaces and in the air, and virus/disease incubation period. Here, we provide an update and elaboration on these issues, as well as some caveats about the work performed so far and as yet unmet needs. As with other questions and issues related to SARS-CoV-2 and COVID-19, work on these two topics is proceeding at a rapid pace at many locations across the globe. Consequently, aspects of this update may rapidly be superseded by new data.
This rapid expert consultation is organized by question and summarizes published and unpublished studies that were deemed most useful, as well as personal communications with experts (cited below). We have selected studies that are most relevant and critical, rather than attempting to be comprehensive. For each of the questions, data are presented for experimental studies and natural history studies, followed by comments on caveats and unmet needs.
This document was prepared by me with support from staff of the National Academies of Sciences, Engineering, and Medicine. Harvey Fineberg approved this document as chair of the Standing Committee on Emerging Infectious Diseases and 21st Century Health Threats. The following individuals served as reviewers: Kathryn Edwards, Vanderbilt University Medical Center; James LeDuc, University of Texas Medical Branch; and Linsey Marr, Virginia Tech. Ellen Wright Clayton, Vanderbilt University, and Susan Curry, University of Iowa, served as arbiters of this review on behalf of the National Academies’ Report Review Committee and their Health and Medicine Division.
QUESTION 1: ENVIRONMENTAL SURVIVAL
In general, there are two basic approaches to study this issue: (A) experimental studies, typically involving the deliberate dissemination of a laboratory-propagated virus under controlled environmental conditions and subsequent sampling; and (B) natural history studies, typically involving the characterization of environments naturally contaminated by a virus, such as hospital rooms recently occupied by patients. Each approach has strengths and weaknesses: with experimental studies there is control over important parameters, but almost always the conditions fail to adequately mimic those of the natural setting; with natural history studies, the conditions are relevant and reflect the real world, but there is typically little control of environmental conditions and potentially confounding factors. Since March 15, there have been advances with studies of each type.
A. Experimental Studies
In a recent study from Hong Kong, Chin et al. examined the stability (using viral culture) of SARS-CoV-2 as a function of temperature, type of surface, and following the use of disinfectants.1 With respect to temperature, using a starting suspension of 6.7 log TCID50/ml in virus transport medium,2 at 4oC there was only a 0.6-log unit reduction at the end of 14 days of incubation in this medium; at 22oC, a 3-log unit reduction after 7 days, and no detection at 14 days; and at 37oC, a 3-log unit reduction after 1 day and no virus detected afterward. No virus was detected after 30 minutes at 56oC or after 5 minutes at 70oC. With respect to survival on surfaces using a 5 µL droplet of virus culture at 7.8 log TCID50/ml, no infectious virus was recovered from printing and tissue paper after 3 hours; no infectious virus was detected on cloth after 2 days or on stainless steel after 7 days. However, on the outside of a surgical mask, 0.1% of the original inoculum was detected on day 7. The persistence of infectious virus on personal protective equipment (PPE) is concerning and warrants additional study to inform guidance for health care workers. Such studies should also examine the effects of various treatments that might be used to disinfect PPE when they cannot be discarded after single use.
Chad Roy from the Tulane University National Primate Research Center shared via telephone some preliminary results of dynamic aerosol stability experiments with SARS-CoV-2 conducted over the past several weeks at the Infectious Disease Aerobiology Core program at Tulane.3 His group generated an aerosol with a fairly uniform distribution of 2 micron particles, using virus grown in DMEM tissue culture (TC) medium and suspended in a rotating drum at an ambient temperature of ~23oC and ~50% humidity. The aerosol was sampled longitudinally for up to 16 hours, and the virus was assessed for viability by growth (enumeration of plaque forming units [PFUs]) and morphology (electron microscopy). He reports surprisingly that SARS-CoV-2 has a longer half-life under these conditions than influenza virus, SARS-CoV-1, monkeypox virus, and Mycobacterium tuberculosis. He is still waiting for some growth results, but expects to post a manuscript describing these findings to bioRxiv on March 27. This result is also concerning, but is quite preliminary; importantly, the details have not yet been shared.
1 Chin et al. 2020. Stability of SARS-CoV-2 in different environmental conditions. https://www.medrxiv.org/content/10.1101/2020.03.15.20036673v1.full.pdf (accessed March 24, 2020).
2 TCID50 is the Median Tissue Culture Infectious Dose.
3 Personal communication, Chad Roy, Tulane University National Primate Research Center, March 24, 2020.
George Korch and Mike Hevey from the National Biodefense Analysis and Countermeasures Center (NBACC), which was created by the U.S. Department of Homeland Security, shared their plans for an extensive series of experiments on SARS-CoV-2 environmental survival.4 Because they have shared these plans with the White House Coronavirus Task Force, only a few observations are provided here. The NBACC is well suited for the kinds of studies it has planned, and the scope and relevance are noteworthy. In particular, it plans to create simulated infected body fluids, including saliva and lower respiratory secretions. It plans to test simulated solar radiation on virus survival, which is important. It also has already examined a wider range of relative humidity and temperature than some other groups, which is again, important. And they will compare RNA semi-quantitative measurements with viral growth (PFUs) on samples from all conditions, which is critical.
At Rocky Mountain Laboratories (RML), part of the National Institutes of Health, current studies include the effect of temperature and humidity on virus stability; virus stability in human body fluids, including urine and feces; and the effectiveness of decontamination procedures for PPE, including N95 respirators.5
As follow-up, the study by van Doremalen et al. mentioned in our rapid expert consultation on March 15, which was at that time an unpublished preprint, has since been published by the New England Journal of Medicine.6
B. Natural History Studies
In a recent published study from Singapore, Ong et al. sampled environmental surfaces at 26 sites in each of 3 SARS-CoV-2 patient isolation rooms, as well as PPE worn by physicians exiting patient rooms and air in the patient rooms and anterooms.7 All samples were tested using reverse transcriptase-polymerase chain reaction (RT-PCR). There were no efforts to assess virus viability. Patient A’s room was sampled on days 4 and 10 of illness while the patient was still symptomatic after routine cleaning. All samples were negative. Patient B was symptomatic on day 8 and asymptomatic on day 11 of illness; samples taken on these 2 days after routine cleaning were negative. Samples collected from Patient C’s room before routine cleaning had positive results at 13 (87%) of 15 room sites (including air outlet fans) and 3 (60%) of 5 toilet sites (toilet bowl, sink, and door handle). Anteroom and corridor samples were negative. Patient C had upper respiratory tract involvement with no pneumonia and had 2 positive stool samples for SARS-CoV-2 on RT-PCR, despite not having diarrhea. Only 1 PPE swab, from the surface of a shoe front, was positive. All other PPE swabs were negative. All air samples were negative. However, the lack of detection of the virus in air samples does not necessarily contradict the finding of the virus on the air outlet fan in Patient C’s room, which presumably deposited from
4 Personal communication, George Korch and Mike Hevey, National Biodefense Analysis and Countermeasures Center, March 24, 2020.
5 Personal communication, Vincent Munster, Rocky Mountain Laboratories, March 24, 2020.
6 van Doremalen et al. 2020. Aerosol and surface stability of SARS-CoV-2 as compared with SARS-CoV-1. New England Journal of Medicine. DOI: 10.1056/NEJMc2004973.
7 Ong et al. 2020. Air, surface environmental, and personal protective equipment contaminated by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) from a symptomatic patient. JAMA. https://jamanetwork.com/journals/jama/fullarticle/2762692 (accessed March 24, 2020).
air onto the surface of the fan. There are at least three explanations for the negative findings in air: (1) a high ventilation rate of the room would dilute concentrations to a level that would be difficult to detect except with a large volume of air; (2) the sample volume was only a fraction of the total room volume; and (3) the air outlets were located above the head of the bed, and it is likely that any virus released into air would be transported directly upward to the outlet, so an air sampler would need to intersect this pathway to optimize chances of detection. Again, it is important to underscore that samples from the two surface-negative rooms were collected after the rooms had been cleaned.
In a recent unpublished study from Changchun, China, Jiang et al. collected 158 environmental surface and air samples from inside and near isolation wards where persons under investigation (PUIs) and known infected patients were housed.8 Samples were collected just before daily cleaning procedures. Only 2 of the 158 samples were RT-PCR-positive: one from surfaces at a nursing station, and the other from an air sample from the room of an intensive care patient.
The Centers for Disease Control and Prevention (CDC) Cruise Ship Environmental Investigation Team mentioned in the CDC’s Morbidity and Mortality Weekly Report (MMWR) on March 23, 2020, the results of environmental sample analysis from the Diamond Princess cruise ship.9 In total, 601 samples were collected and tested, of which 58 were positive (9.7%) by RT-PCR. According to the Discussion, “SARS-CoV-2 RNA was identified on a variety of surfaces in cabins of both symptomatic and asymptomatic infected passengers up to 17 days after cabins were vacated on the Diamond Princess but before disinfection procedures had been conducted (Takuya Yamagishi, National Institute of Infectious Diseases, personal communication, 2020). Although these data cannot be used to determine whether transmission occurred from contaminated surfaces, further study of fomite transmission of SARS-CoV-2 aboard cruise ships is warranted.”
Santarpia et al. recently completed a study (as yet unpublished and not yet posted on a preprint server) of air and surface samples from 11 isolation rooms at the University of Nebraska Medical Center that were used to care for SARS-CoV-2 patients.10 Samples were collected from common room surfaces, personal items, and toilets, as well as high volume air samples and low volume personal air samples. Many commonly used items, toilet facilities, and air samples had evidence of viral contamination: 76.5% of all personal items and 80.4% of all room surfaces were positive for SARS-CoV-2 by RT-PCR (0.22-0.82 gene copies/microliter of swab resuspension); 63% of room air samples were positive (mean 2.86 copies/L of air); 81% of toilet samples were positive. The percentage of positive samples from each room ranged from 50% to 100%. There was no clear correlation between severity of illness, cough or fever, and the prevalence of viral RNA. Of note, air collectors positioned more than 6 feet from each of two patients yielded positive samples, as did air samplers placed outside patient rooms in the hallways. Although the results are preliminary, it appears that some samples are positive for infectious virus, including an air
8 Jiang et al. 2020. Clinical data on hospital environmental hygiene monitoring and medical staffs protection during the coronavirus disease 2019 outbreak. https://www.medrxiv.org/content/10.1101/2020.02.25.20028043v2.full.pdf (accessed March 25, 2020).
9 Moriarty et al. 2020. Public health responses to COVID-19 outbreaks on cruise ships—worldwide, February–March 2020. Morbidity and Mortality Weekly Report 69(12):347-352. http://dx.doi.org/10.15585/mmwr.mm6912e3.
10 Santarpia et al. In preparation. Transmission potential of SARS-CoV-2 in viral shedding observed at the University of Nebraska Medical Center. Soon at medRxiv.
sample collected well more than 6 feet from a patient.11 These results require urgent confirmation under a variety of conditions as they have significant implications for current public health messaging regarding necessary distancing between nearby individuals to prevent virus transmission. In addition, and in this case anecdotal, the highest airborne RNA concentrations were recorded by personal samplers while a patient was receiving oxygen through a nasal cannula (19.17 and 48.21 copies/L). The possibility of aerosol generation by oxygen delivery via nasal cannula and other mechanisms is currently being explored. Overall, these data support the possibilities of both direct (droplet and person-to-person) and indirect (contaminated objects, airborne) forms of transmission.
A recent study by Liu et al. provides additional information regarding aerodynamics, concentrations, and distribution of aerosols containing SARS-CoV-2.12 A total of 35 aerosol samples (30 samples with total suspended particles, 3 samples with size-segregated particles, and 2 aerosol deposition samples) were collected in two hospitals and public areas in Wuhan, including patient areas, ICUs, medical staff areas, and toilet areas. In regard to patient areas, the highest concentrations of airborne SARS-CoV-2 were observed inside the patient mobile toilet room (19 copies m–3), suggesting the importance of frequent disinfection of patient toilets. In regard to medical staff areas, the protective apparel removal rooms had the highest airborne virus concentrations (18 to 42 copies m–3). In regard to public areas, airborne concentrations were generally below 3 copies m–3, except for a crowded site near the entrance to a department store and a busy site next to a hospital. The peak concentrations of SARS-CoV-2 aerosols appear to exist in two distinct size ranges: 0.25 to 1.0 µm and those larger than 2.5 µm. Aerosols smaller than 2.5 µm can remain suspended in the air for many hours. The study observed that the negative pressure ventilation and high air exchange rate inside some locations were effective in minimizing airborne SARS-CoV-2. Additional findings suggest that virus-laden aerosol deposition may play a role in surface contamination and thus subsequent human infection. The authors believe that a direct source of SARS-CoV-2 may be due to a resuspension of virus-laden aerosol from the surface of medical staff protective apparel during removal, which may come from direct deposition of respiratory droplets while medical staff are working. Floor dust aerosol containing the virus is also subject to resuspension—meaning that virus-laden aerosols could first deposit on the surface of protective gear and then fall to the floor to be resuspended by medical staff movement. Outside of the hospital, only 2 crowd gathering sites (of 11 sites sampled) had detectable concentrations of SARS-CoV-2 aerosol, which may contribute to sources of virus-laden aerosol during sampling. It is important to note that the sample size for the aerosol samples, and notably the size-segregated samples (3) and aerosol deposition samples (2), were small—a limitation of this study. Furthermore, TRIzol LS Reagent (Invitrogen) was added to inactivate SARS-CoV-2 to extract the RNA, which should be noted as a limitation to the study because the authors measured viral RNA, not infectious virus.
There are a number of published studies that examine the relationship between the geographic incidence of COVID-19 cases and ambient temperature and humidity. Some suggest possible but modest correlations between geographies with higher temperature or humidity, and lower rates
11 Personal communication, Josh Santarpia, University of Nebraska Medical Center, March 25, 2020.
12 Liu et al. 2020. Aerodynamic characteristics and RNA concentration of SARS-CoV-2 aerosol in Wuhan hospitals during COVID-19 outbreak. https://www.biorxiv.org/content/10.1101/2020.03.08.982637v1 (accessed March 26, 2020).
of disease; however, there are a number of confounding factors, including disease reporting practices and quality of and access to health care. We did not scrutinize these studies carefully nor perform an extensive search for related studies.
C. Caveats, Needs
A notable limitation of most of the natural history studies described above is a reliance on RT-PCR to assess the presence of SARS-CoV-2 on surfaces and air. Although viral RNA was detected in many environmental samples across the various studies, infectivity is not known. It is important to note that there are no available data to our knowledge that speak to the possible linkage between the presence of environmental viral RNA or even infectious virus and the risk of transmission from these environmental sites to humans. This is a key issue, and relates in part to another major issue and unanswered question: What is the infectious dose of SARS-CoV-2 for humans? Studies to address this question are planned, and in fact may be under way with nonhuman primates at several laboratories, but these studies will be limited by the relevance of nonhuman primate susceptibility to human susceptibility. The use of other laboratory animals will provide even less relevant information on incubation time.
Questions have been (appropriately) raised about whether there are relatively easy-to-perform, quick, and safe measurements one might undertake on environmental samples for predicting the presence of viable virus, rather than reliance on cultivation (PFU) assays. One idea recently discussed by Wölfel et al.13 is to look for subgenomic mRNAs made by the virus during its life cycle in a human cell but not packaged into mature virions. These subgenomic mRNAs, if detected directly in a clinical sample, signify that the virus has been actively replicating in host cells in the sample at the time the sample was expelled from the body. This approach was used by Wölfel et al. to argue for active SARS-CoV-2 replication in the throat of COVID-19 patients during the first 5 days after symptoms onset. This approach could conceivably be used to assess the possibility of recent active viral replication in environmental swab samples.
An important caveat regarding the results from experimental studies relates to their relevance to real-world conditions. For example, many of the experimental environmental survival studies have used virus grown in TC media. It is quite possible that virus from naturally infected humans when directly disseminated to the nearby environment has different survival properties than virus grown in TC media, even when the latter is purified and spiked into a relevant human body fluid such as saliva. However, environmental dissemination of clinically relevant human fluids spiked with TC-grown virus will be more predictive of real-world environmental survival than environmental dissemination of TC-grown virus in TC media. Important human clinical matrices into which virus should be spiked include saliva, respiratory (including nasal) mucus and lower respiratory tract airway secretions, urine, blood, and stool. In addition, nebulized saline should be spiked and studied. Another issue related to experimental conditions is the effect of humidity on viral stability. Aerosol studies to date have tended to use humidity levels for culture media that are more favorable for viral decay (e.g., 50-65% relative humidity). Real respiratory fluid is likely to be more protective of infectivity, and indoor relative humidity in wintertime in temperate regions is usually 20-40%, a range that is more favorable for virus survival.
13 Wölfel et al. 2020. Virological assessment of hospitalized cases of coronavirus disease 2019. https://www.medrxiv.org/content/10.1101/2020.03.05.20030502v1.full.pdf (accessed March 25, 2020).
Consequently, the half-lives reported to date may represent the lower end of the range. Differences in experimental conditions across studies (e.g., viral growth media, viral titer determination methods, infectivity of the inoculum) would be expected to contribute to variation in study results.
Before too many public health decisions are made on the basis of experimental or natural history studies using just one virus strain, some attention should be paid to the possibility of variation among different SARS-CoV-2 strains in their environmental survival properties. Different isolates from early and late in the pandemic, and from different geographic regions, should be studied and compared.
Registries of patient data and patient samples (e.g., nasopharyngeal, sera, urine, stool) are being created and can be used in future studies examining environmental persistence of the virus. For example, such samples could be used as clinical matrices to look at SARS-CoV-2 persistence on surfaces.
QUESTION 2: INCUBATION PERIOD
We approach this question in a similar way, examining first experimental studies and then natural history studies: (A) experimental studies, typically involving the inoculation of animals in the laboratory using a laboratory-propagated virus under controlled conditions and subsequent monitoring for onset of viral shedding, signs of disease, or other physiological responses; and (B) natural history studies, typically involving longitudinal or cross-sectional studies of naturally exposed humans and the collection of data on time of exposure and time of onset of signs, symptoms, and virological and molecular features of infection and disease. Each approach has strengths and weaknesses: with experimental studies there is control over time of exposure and various features of the inoculum, but non-human animals to varying degrees fail to reflect the natural history of infection in humans; with natural history studies, the host is relevant, but the time and nature of the exposure is less well understood and sample availability is uncertain.
A. Experimental Studies
As mentioned above, experimental infections in non-human primates are planned or are under way at several sites in the United States, including the Tulane University National Primate Research Center and RML,14 and presumably in other countries. While animal models are very important for understanding pathogenesis and responses to therapeutic and vaccine candidates, they are not as helpful with the incubation period studies given physiological differences across species.
B. Natural History Studies
In a recent preprint from Shaanxi, China, and New York, Men et al. examined confirmed cases of COVID-19 from 10 regions in China, other than Hubei province, for whom there were data on
14 Personal communication, Chad Roy, Tulane University National Primate Research Center, March 24, 2020.
time of exposure and time of disease onset.15 A Monte Carlo simulation was employed to estimate incubation period, along with additional statistical analysis to assess relationships between different age and gender groups. In this study, the mean and median incubation periods were estimated to be 5.84 and 5.0 days, respectively. Patients 40 years or older had a longer incubation period and larger variance than did patients younger than 40 years. There was no statistically significant difference in incubation period based on gender. These findings suggest that different periods of quarantine may be advisable based on age. However, these results need to be confirmed through additional studies and with further stratification of incubation period results by age group.
In a recent preprint from the National Institute of Allergy and Infectious Diseases, Peking University, and the Chinese Center for Disease Control and Prevention, Qin et al. identified asymptomatic individuals at their time of departure from Wuhan and followed them until symptoms arose.16 This method was reported to offer enhanced accuracy by reducing recall bias and by utilizing forward time data. More than 1,000 cases were collected from publicly available data. They found that the estimated median incubation period was 8.13 days, the mean was 8.62 days, the 90th percentile was 14.65 days, and the 99th percentile 20.59 days. Compared to other studies, this incubation period is longer. They conclude that ~10% of patients with COVID-19 do not develop symptoms until 14 days after infection.
In a recent preprint from Guangzhou and Hong Kong, He et al. reported on temporal patterns of viral shedding in 94 laboratory-confirmed COVID-19 patients and modeled COVID-19 infectiousness from a separate sample of 77 infector-infectee transmission pairs.17 They observed the highest viral load in throat swabs at the time of symptom onset, and inferred that infectiousness peaked on or before symptom onset. They estimated that 44% of transmissions may occur before the first symptoms of the index case.
C. Caveats, Needs
Robust estimates of the distribution of the incubation period and the period of infectiousness for SARS-CoV-2 are critically important to inform public health messaging. Differences in incubation period findings among existing studies may relate to methodological differences, limited sample sizes, recall bias, or inadequate follow-up (potentially missing people who have longer incubation periods). Given the small number of human studies evaluating these disease characteristics for COVID-19, additional studies to confirm incubation period estimates and infectiousness prior to symptom onset are urgently needed. For public health management, it makes a great deal of difference whether 1% of patients will develop the disease after 14 days (if the mean incubation is approximately 5 days) or whether the fraction is 10% of patients (if the mean incubation period is approximately 8 days). Additional studies should examine variables that may have an impact on incubation period, which, besides age (see Men et al. above), may
15 Men et al. 2020. Estimate the incubation period of coronavirus 2019 (COVID-19). https://www.medrxiv.org/content/10.1101/2020.02.24.20027474v1.full.pdf (accessed March 25, 2020).
16 Qin et al. 2020. Estimation of incubation period distribution of COVID-19 using disease onset forward time: A novel cross-sectional and forward follow-up study. https://www.medrxiv.org/content/10.1101/2020.03.06.20032417v1.full.pdf (accessed March 25, 2020).
17 He et al. 2020. Temporal dynamics in viral shedding and transmissibility of COVID-19. https://www.medrxiv.org/content/10.1101/2020.03.15.20036707v2.full.pdf (accessed March 25, 2020).
include inoculum size, immune competency of host, co-infecting agents, and underlying morbid conditions. Prospective longitudinal studies are most effective for addressing this issue. An obvious challenge is precise identification and timing of natural exposures. Additionally, as mentioned above, it is conceivable that the evolution of new SARS-CoV-2 strain variants will be accompanied by different properties, including incubation period. Prior to changing current public health guidance, it may be prudent to compare observed incubation periods among different SARS-CoV-2 strains. Future studies related to incubation period and viral loads in asymptomatic patients may help to inform pressing questions related to, for example, the role of super spreaders and children in transmission.
David A. Relman, M.D.
Standing Committee on Emerging Infectious Diseases and 21st Century Health Threats