If you read anything about SARS-CoV-2, the novel coronavirus that causes COVID-19, transmission is described as person-to-person via direct contact and respiratory droplets released by coughing, sneezing, or even talking, with fomites as a less common possibility. Whether it is airborne is, well, up in the air.
But what everyone really wants to know is how you translate this information into practical recommendations. Is the advice for the widespread use of masks, frequent hand hygiene and disinfection of surfaces, and social distancing at six feet from others necessary? Is it enough? Many emergency physicians and other high-risk health care workers have gone further, devising elaborate decontamination rituals, and some even electing to live apart from their families. Their concern is not unwarranted—20 percent of health care workers in Italy were reported to have become infected. (Lancet. 2020;395:1225; https://bit.ly/2Yh1D2b.)
Health care workers don't want to get infected, but even more importantly, they don't want to bring the virus home to their families. Making matters worse, it has become increasingly apparent that COVID-19 can be transmitted by a person who is asymptomatic or presymptomatic. One study estimated that 44 percent of infections occur this way. Infectiousness is thought to start 2.3 days prior and peak 0.7 days prior to symptom onset. (Nat Med. 2020 Apr 15. https://go.nature.com/3f2vi4V.)
Yet, sparse information is available on the specific activities that increase the risk of transmission, and what health care workers can and should be doing to prevent infection for themselves and their families. To try to get this information, I reviewed case cluster reports in the literature and the media for specific details on exposure and transmission activities, articles reporting on secondary attack rates, and literature regarding virus surface stability, fecal shedding, and mask efficacy for aerosolized virus.
Lessons from Case Clusters
Table 1 shows the case clusters from the medical literature that I reviewed. As expected, transmission between close household contacts and family members is a frequent source of infection, presumably due to close direct contact, face-to-face interactions, and possibly contaminated surfaces and fomites. Table 2 summarizes reviewed media reports of case clusters from a variety of locations.
Close contact, especially in enclosed indoor spaces such as small conference rooms, is a risk factor that recurs in household and workplace clusters. The European Centre for Disease Prevention and Control defines high-risk exposure to an infected person as direct physical contact—being with the person in a closed environment (e.g., household, classroom, meeting room, hospital waiting room, etc.) for more than 15 minutes, having face-to-face contact within six feet for more than 15 minutes, and being on an aircraft within two seats in any direction. Cases where a shop owner became infected through interactions with customers reflect the potential for transmission from relatively brief and innocuous face-to-face encounters. Prior studies of adults show only slightly more respiratory droplets produced with coughing compared with talking. (J R Soc Interface. 2009;6 [Suppl 6]:S703; https://bit.ly/2YiqLpl; J Epidemiol. 2013;23:251; https://bit.ly/3aQl8ky.)
Prolonged close contact combined with large numbers of attendees, which increases the chance that at least one person present will be infectious, likely played a role in most of the media reports of outbreaks at religious gatherings, birthday parties, funerals, sporting events, spring break trips, cruises, and meat processing plants. These findings reinforce the need for social distancing and avoiding large gatherings and crowded conditions.
Beyond close contact, a few themes emerge, in particular, shared meals. Dining with someone who is infected but asymptomatic, presymptomatic, or minimally (not obviously) symptomatic may be especially risky due to respiratory droplets expelled while combining eating and talking. Contact tracing revealed several patients whose sole exposure was a shared meal or sitting near an infected person in a restaurant or at a banquet. Dining face-to-face from a close distance combined with conversation is likely even a higher risk. Previous research has shown an increase in respiratory droplets expelled during coughing and talking when food dye was used, possibly due to the food dye stimulating an increase in saliva. (J R Soc Interface. 2009;6 [Suppl 6]:S703; https://bit.ly/2YiqLpl.)
Another risk factor highlighted is singing or vocalizing in close proximity to others, as seen in the choir practice cluster (45 of 60 members at a 2.5-hour practice were infected), and the various church and other religious group clusters, which likely involved singing and other synchronous vocalizations. A prior case series documented a singer transmitting rubella to multiple contacts who visited the nightclub where he performed. (Am J Epidemiol. 1981;114:574.)
Religious gatherings, birthday parties, and funerals have figured prominently in reported clusters. Several factors come together at these events, including large numbers of attendees, crowded and enclosed spaces, singing and vocalizing, and the tendency to shake hands or embrace in greeting. Future gatherings will need to incorporate methods for social distancing and greeting without direct contact to lower risk.
Secondary Attack Rates
Table 3 summarizes secondary attack rates reported in the medical literature for family contacts, household contacts such as roommates and cabinmates, close contacts, and health care workers. Here the evidence is not very clear. Attack rates are likely to vary widely by pandemic timing (early on, with reduced awareness, protective measures are not taken, and secondary attack rates are higher), family customs such as meal and bed sharing, family living space crowdedness, health care worker use of personal protective equipment (PPE), and for small clusters, index patient characteristics such as the tendency to expel respiratory droplets with talking, hand hygiene frequency, and cough etiquette. All of the attack rates discussed below are ill-defined, and much more data are needed.
From the data we do have, family members unsurprisingly have the highest secondary attack rate at approximately 10-30 percent. Spouses have a higher rate, children are less affected than adult family members, and secondary attack is reduced if the index patient quarantines separately from the family immediately upon becoming symptomatic. ((Preprints with The Lancet. 2020 Mar 31. https://bit.ly/2y2FsCh; Chinese J Preventive Medicine. 2020:54 [Epub ahead of print]; (Clin Infect Dis. 2020 Apr 17; https://bit.ly/2yPV63L.)
The U.S. Centers for Disease Control and Prevention recently expanded sentinel symptoms beyond fever, cough, and shortness of breath to include chills and shaking, myalgias, headache, sore throat, and loss of sense of taste or smell.) (https://bit.ly/2VNPwIh.) The small amount of data currently available shows household member secondary attack rates of 8-16 percent, although this is much higher in closer quarters such as a cruise ship cabin or military unit quarantine tent.
Secondary attack rates in close contacts have been reported more often, and are generally less than five percent unless there is a high-risk situation such as a shared meal or enclosed environment such as a cruise ship. Likelihood of transmission is increased with longer duration, higher-risk types of contact, and whether the index patient was symptomatic.
The few reports of health care worker secondary attack rates estimated through contact tracing (when a patient was not initially diagnosed and PPE use was inappropriate) are mostly zero, which contradicts reports of 20 percent of Italian health care workers becoming infected (Lancet. 2020;395:1225; https://bit.ly/2Yh1D2b), 50 positive health care workers identified in King County, WA, between Feb. 28 and March 13, 2020 (JAMA. 2020 Apr 17. https://bit.ly/2SjDbt8), identification of 9,282 U.S. health care worker COVID-19 cases reported to the CDC from Feb. 12 to April 9, 2020 (MMWR. 2020;69:477; https://bit.ly/2WbNMYk), and 18 percent positive (282/1533) of U.K. symptomatic health care workers tested over a two-week period. (Euro Surveill. 2020;25; https://bit.ly/2Si4yDQ.)
Some health care workers may be exposed through community contacts, although the CDC study of U.S. health care workers found that 55 percent reported health care setting exposure only, 27 percent household exposure, 13 percent community exposure, and five percent multiple settings. (MMWR. 2020;69:477; https://bit.ly/2WbNMYk.)
Investigators in the Netherlands looked at the possibility of community transmission by testing a convenience sample of health care workers with no history of concerning travel or known COVID-19 patient contact; 4.1 percent (45/1097) tested positive. (Euro Surveill. 2020;25; https://bit.ly/2SeK8M7.) Interestingly, two of six hospitals accounted for 38 of the 45 infected health care workers, which could suggest that the workers were infecting each other (or that local communities surrounding those hospitals were more endemic.)
More data are needed on community v. workplace exposure for health care workers who become infected, efficacy of PPE, the risks of specific health care worker-patient interactions, and whether we are infecting each other while working together. A thorough review of methods health care workers can use to protect themselves during specific situations and procedures was published by Ferioli, et al. (Eur Respir Rev. 2020;29:; https://bit.ly/3aJFNqt.)
There's hope. The proportion of new COVID-19 positive cases in China among health care workers decreased from 5.72 percent from Jan. 11 to Jan. 20, 2020, to 2.68 percent from Feb. 1 to Feb. 11, 2020; this was attributed to better availability of PPE, enhanced vigilance, and improved knowledge and experience with protective measures. (Front Med. 2020 Mar 23. https://bit.ly/2VLIwvn.)
Fomite transmission of respiratory viruses is a known phenomenon, which is why we have recommendations for meticulous hand hygiene, not touching one's face, and frequent surface disinfection. There are few to no clear cases of COVID-19 fomite transmission found in the literature. A workplace investigation found the only contact between the transmitter and recipient of the virus to be when, while sitting back to back in a canteen, one turned around to the other and asked for the salt. (Preprints with The Lancet. 2020 Mar 31. https://bit.ly/2y2FsCh.)
An epidemiologic investigation into the case of a woman in Charlotte, NC, who had left her house only once in three weeks to go to the pharmacy concluded that she was infected by touching the pharmacy keypad. (WCNC, Charlotte, NC, April 21, 2020; https://bit.ly/3d0MEO4.) On the other hand, an article about a case cluster from a carnival party in Germany stated, “In Heinsberg, his team of coronavirus detectives could find scant evidence of the virus being transmitted via the surfaces of door handles, smart phones or other objects.” (The Guardian, 9 April. https://bit.ly/2Sk2LOI.) It may be difficult to answer this one. It is hard to say whether the virus was passed via a salt-shaker fomite or close-contact respiratory droplets despite sitting back to back except for a brief period.
Multiple studies have reported on swabs of patients' rooms and hospital areas for COVID-19 viral RNA, finding more samples positive in the ICU than the general ward, extensive contamination of COVID-19 patient rooms before cleaning but not after cleaning, and mixed findings on testing health care worker PPE. (Emerg Infect Dis. 2020;26; https://bit.ly/3aNDKC2.) Common contaminated sites include the floor, computer mice, doorknobs, telephones, hand sanitizer dispensers, trash cans, bedrails, bedside tables, remote controls, cell phones, window ledges, patient masks, self-service printers used by outpatients, and health care worker gloves. Viral RNA was identified in the cabins of the Diamond Princess cruise ship up to 17 days after they were vacated (but before cleaning). (MMWR. 2020 Mar 27;69(12):347; https://bit.ly/2xkADUh.) One important caveat to swab studies is that detection of viral RNA is not the same as detection of virus capable of transmitting infection.
Table 4 reports the results of two surface stability studies. The COVID-19 virus persists longer on hard surfaces such as plastic and steel compared with soft surfaces such as cardboard (although it persisted for only four hours on copper). (Lancet Microbe. 2020 Apr 2; https://bit.ly/2KJ82Lk; N Engl J Med. 2020;382:1564; https://bit.ly/2ShCZe4.) These durations represent experimental conditions and may not represent the actual ability to become infected from touching a contaminated surface. Although decontamination of objects brought in from outside the home is still recommended, the risk of fomite transmission from mail, packages, takeout delivery, and grocery bags is thought to be minimal.
The effects of heat are unclear. The virus is highly stable at a refrigerator temperature of 39°F; one study found that 30 minutes at 133°F or five minutes at 158°F inactivated the virus, but another was only able to inactivate the virus with 15 minutes at 198°F. (Lancet Microbe. 2020 Apr 2; https://bit.ly/2KJ82Lk.) Many experts recommend washing potentially contaminated clothing such as scrubs in hot water. Even if heat does not fully inactivate the virus, the laundry soap is likely to.
Unpublished data from the U.S. Department of Homeland Security Science and Technology Directorate uploaded to the internet and previewed in a recent White House briefing suggests that the coronavirus is inactivated on surfaces in full direct midday sunlight in three minutes and in aerosols in full direct sunlight in 10 minutes. (April 13, 2020. https://bit.ly/2Wad88L; The Washington Post. April 24, 2020; https://wapo.st/2YjnXZ4.) The half-life reported on surfaces in full-intensity sunlight (NYC/Washington, DC, during a clear day on summer solstice) is two minutes, in half-intensity sunlight is three minutes, and in quarter-intensity sunlight (NYC/DC on clear day at the end of February) is four minutes. The half-life for aerosols in full sunlight is 1.5 to three minutes. They concluded that outdoor daytime environments are at lower risk for transmission.
Fecal Shedding and Bioaerosols
A meta-analysis of 60 studies of 4243 COVID-19 patients found that 17.6 percent had gastrointestinal symptoms, and 48.1 percent of stool samples were positive for viral RNA. (Gastroenterology. 2020 Apr 3; https://bit.ly/2W9Fdxd.) Some patients, particularly children, have been shown to have viral RNA detectable in anal swabs longer than respiratory swabs, up to five weeks. (Lancet Gastroenterol Hepatol. 2020;5:434; https://bit.ly/3aQiTxV.)
These findings have raised concerns regarding the possibility of fecal-oral transmission, underscoring the importance of hand hygiene and surface disinfection. A case cluster where those affected were all women who may have used a common shopping mall restroom (Emerg Infect Dis. 2020 Mar 12;26; https://bit.ly/2YjbjJm) and another that occurred in a men's bathhouse (JAMA Netw Open. 2020 Mar 2;3:e204583; https://bit.ly/2KHLH0Q) with cases transmitted to men who visited one to six days after the index case hint at a fecal vector of transmission.
Besides fecal-oral transmission, fecal viral shedding increases risks due to toilet bioaerosols. Flushing contaminated toilets aerosolizes large droplets and droplet nuclei that can be carried on air currents, and this continues through multiple flushes. (Am J Infect Control. 2013;41:254; https://bit.ly/2y9wvqw.) Bioaerosols of particles predominantly less than 3 microns in size (the coronavirus is 0.06 to 0.14 microns [StatPearls. 2020 Apr 6; https://bit.ly/2yY9neY]) are aerosolized by toilet flushing and detectable about three feet away and beyond from the toilet 30 minutes after flushing, even when no waste was present during the flush (i.e., from previous waste). (Antimicrob Resist Infect Control. 2018 Jan 26;7:16; https://bit.ly/35eNynm.)
Previous studies have demonstrated enteric and respiratory viruses on 78 percent of hospital and office toilet surfaces (toilet seat, flush handle, door handle) and in 81 percent of toilet-adjacent aerosol samples. (Am J Infect Control. 2014;42:758; https://bit.ly/2W9gzNc.) Frequent cleaning of toilets and bathrooms, closing the lid when flushing (even if waste is not present), and assigning a toilet to high-risk family members who may give or get the virus are potential preventive measures.
Airborne transmission, in the strict definition used by infectious disease experts for varicella, measles, and tuberculosis, is not thought to occur with the coronavirus. A review of 10 studies of the horizontal distance travelled by respiratory droplets found that eight demonstrated a distance farther than six feet, and one showed that exhalations, sneezes, and coughs can create a multiphase turbulent gas cloud that can travel 23-27 feet. (J Infect Dis. 2020 Apr 16; https://bit.ly/3aOvvFJ; JAMA. 2020 Mar 26; https://bit.ly/2SkmHRI.)
Case clusters, including restaurant diners in the path of the air conditioning, choir singers, aircraft travelers, and cruise ship passengers infected after they were confined to their cabins, point toward the possibility of an “airborne-lite” transmission route. The COVID-19 virus was detectable in aerosol for three hours or longer, with a half-life of 1.09 hours. (N Engl J Med. 2020;382:1564; https://bit.ly/2ShCZe4.) Concern has also been raised for carriage of the virus via secondhand smoke generated by smoking and vaping. (Environ Sci Technol. 2020 Apr 22; https://bit.ly/2SlupL7.)
This realization has led to recommendations to wear masks in public, preferably non-medical cloth masks to conserve PPE for health care workers. Face masks help by reducing the amount of virus expelled by infected people, providing some filtering of inhaled air to prevent the virus from reaching the respiratory epithelium (homemade masks may not block particles measured in microns, but the majority of transmission is still thought to be via large droplets), and serving as a reminder not to touch your face. A case report recounted a newly symptomatic man who transmitted the virus to five of 39 bus passengers when he wore no mask, but none of 14 minibus passengers after he purchased a mask for the second leg of his journey. (Influenza Other Respir Viruses. 2020 Mar 29; https://bit.ly/2Wato9N.)
Two studies examined various materials that could be used to make homemade masks. (Table 5.) The first found that vacuum cleaner bags and tea towels filtered effectively, but they had too high of a pressure drop across them, making it difficult to breathe. Cotton T-shirts and pillowcases were the preferred material. The second found that cotton-chiffon and cotton-silk hybrids and high thread count cotton materials produced the best filtration, that gaps significantly reduced filtration, and that none of the materials tested had too high of a pressure drop for use. Mueller, et al., placed a particle counter inside various masks worn by a volunteer to sense 0.04 micron NaCl particles aerosolized in ambient air, and found that adding a nylon stocking overlayer to the mask improved virus blockade for all types, including surgical masks. (medRxvi.com. 2020 Apr 24; https://bit.ly/3cXL6Em.) This simple addition improved many of the homemade cloth masks to the baseline level of a surgical mask.
We know that the spouse/partner secondary attack rate is higher than that of other family members, but there aren't good data on whether sleeping in the same bed contributes. Human-to-human droplet nuclei transmission research shows increased transmission in face-to-face positioning, distances less than about 2.5 feet between people, and mouth (v. nose) exhalation by the infected individual. (Indoor Air. 2018;28:500.) Incidentally, these data also show the highest risk position for a doctor is standing over a COVID-19 patient who is lying on his side and facing the doctor, so examine suspected COVID-19 patients seated or supine, not face-to-face. Decisions regarding sleeping apart should take into account these factors and the risk levels for the partners to become infected from work exposures and the risk of experiencing a severe course if infected.
The Bottom Line
New data are being reported daily, and this literature synthesis will become rapidly outdated. Other caveats: Some of the studies reviewed have not yet been peer-reviewed, media reports may be inaccurate and biased, and decisions and recommendations must be made based on scarce, imperfect information at this time. With these caveats in mind, Tables 6 and 7 summarize possible preventive measures.