The Science of Masking to Control COVID-19

cdc.gov/coronavirus
Valid as of November 16, 2020
Most SARS-CoV-2 Infections Are Spread by People without Symptoms

Infection is spread primarily through

exposure to respiratory droplets exhaled
by infected people when they breathe,
talk, cough, sneeze, or sing
▪ Most of these droplets are <10 μm,
often referred to as aerosols
▪ The amount of these fine droplets and
particles increases with volume of
speech (e.g., loud talking, shouting)
and respiratory exertion (e.g., exercise)

Adapted from Aslved et al. 2020, Aerosol Sci Technol; https://doi.org/10.1080/02786826.2020.1812502.

Valid as of November 16, 2020
Most SARS-CoV-2 Infections Are Spread by People without Symptoms

▪ 40-45% of infected people are

estimated to never develop symptoms
Before symptoms After symptoms ▪ Among people who do develop
symptomatic illness
– Transmission risk peaks in the
days just before symptom onset
(presymptomatic infection) and
for a few days thereafter
– Accordingly, the number of
infections transmitted peaks
when virus levels peak

He et al. 2002, Nat Med; 26(5):672-675 and 26(9):1491-1494. Valid as of November 16, 2020
Most SARS-CoV-2 Infections Are Spread by People without Symptoms
Never symptomatic: 24% ▪ CDC and others estimate that more
Pre-symptomatic: 35%
than 50% of all infections are
Symptomatic: 41%
transmitted from people who are not
exhibiting symptoms
▪ This means, at least half of new
infections come from people likely
unaware they are infectious to others
(red and orange in the figure, left)*

* Figure assumes peak infectiousness occurs 5 days after infection

and that 24% of infections are asymptomatic. With these
assumptions, 59% of infections would be transmitted when no
symptoms are present but could range 51%-70% if the fraction of
asymptomatic infections were 24%-30% and peak infectiousness
ranged 4-6 days.
Moghadas et al. 2020, Proc Natl Acad Sci USA;117(30):17513-17515. Johansson et al. 2020, CDC unpublished data; submitted. Valid as of November 16, 2020
Three Levels of Scientific Evidence Demonstrate the
Benefit of Community Masking to Control SARS-CoV-2
1. Controlled laboratory-based experimental studies of cloth masks’ capacity to
– Block exhaled emission of virus-laden respiratory particles (source control)
– Reduce inhalation of these droplets by the wearer (personal protection)

2. Epidemiological investigations
– Outbreaks
– Cohort and case-control studies

3. Population-level community studies

– Across multiple levels (e.g., hospital system, city, state, country, multi-country)

Valid as of November 16, 2020

Community Masking to Control SARS-CoV-2
Experimental Studies

Valid as of November 16, 2020

Laboratory Assessment of Cloth Masks Effectiveness:
Source Control (exhalational)

Cloth masks provide source control

▪ Cloth masks block most large (>20-30 μm)
exhaled respiratory droplets
▪ Multi-layer cloth masks substantially
block respiratory droplets <1-10 μm
– Comprise the greatest fraction of
exhaled respiratory droplets
– Reductions as high as 50-70%
▪ Some on par with surgical masks

Figure from Lindsley et al. 2020; medRxiv: doi 10.1101/2020.10.05.20207241. . See “Appendix” at end of slide set for full set of references.
Valid as of November 16, 2020
Laboratory Assessment of Cloth Masks Effectiveness:
Filtering Protection (inhalational)

Cloth masks also filter inhaled droplets

▪ Their performance filtering inhaled small
droplets is not as good as their
performance blocking exhaled small
droplets
▪ Improvements possible with more layers,
multiple materials
– Static charge, hydrophobic
▪ Opportunities for innovation

Image from Konda et al. 2020, ACS Nano; 14(5):6339-6347. See “Appendix” at end of slide set for full set of references.
Valid as of November 16, 2020
Laboratory Assessment of Cloth Masks Effectiveness:
Two-Headed Experimental Masking Evaluation using SARS-CoV-2

Ueki et al. 2020, mSphere; doi.org/10.1128/mSphere.00637-20.

Valid as of November 16, 2020
Laboratory Assessment of Cloth Masks Effectiveness:
Two-Headed Experimental Masking Evaluation using SARS-CoV-2

Ueki et al. 2020, mSphere; doi.org/10.1128/mSphere.00637-20.

Valid as of November 16, 2020
Laboratory Assessment of Cloth Masks Effectiveness: Summary

▪ Focus on the relative effects, not the absolute values from these laboratory studies
– All experiments are proxies for human experience and biological processes
▪ Source control is substantial, but there is also measurable and meaningful personal
protection with the use of cloth masks
– Masking reduces the wearers’ viral exposure
▪ Cloth masks are comparable to surgical masks when used together for community
control (i.e., when combined for both source control and personal protection)

Valid as of November 16, 2020

Community Masking to Control SARS-CoV-2
Epidemiological Investigations

Valid as of November 16, 2020

Multiple Epidemiologic Investigations of Cloth Mask Effectiveness
▪ High-risk exposure events
– May 2020: 2 symptomatically ill hair stylists
• Interacted closely, for 15 minutes on average, with 139 clients over an 8-day period
• The stylists and all clients wore masks per local ordinance and company policy
• 0 of 67 clients subsequently reached for interview and tested developed infection
– March and April 2020: Outbreak aboard the USS Theodore Roosevelt
• Environment notable for congregate living quarters and close working environments
• Use of face coverings on-board was associated with a 70% reduced risk
▪ Retrospective case-control study of exposed contacts (Thailand)
– March 2020: People who reported having always worn a mask during high-risk exposures
• Experienced a greater than 70% reduced risk of acquiring infection compared with
people who did not wear masks under these circumstances

Hendrix et al 2020, MMWR; doi.org/10.1101/2020.05.22.20109231. Payne et al. 2020, MMWR; 69(23):714-721.

Doung-Ngern et al. 2020, Emerg Infect Dis;26(11).
Valid as of November 16, 2020
Multiple Epidemiologic Investigations of Cloth Mask Effectiveness
▪ Household surveys
– February and March 2020: Within 124 Beijing households with > 1 laboratory-confirmed case
of SARS-CoV-2 infection
• Mask use by the index case and family contacts before the index case developed
symptoms reduced secondary transmission within the households by 79%
▪ Air travel
– January 2020: symptomatically ill person was the sole air passenger wearing a surgical mask
• 15-hour flight (Wuhan to Toronto)
• 0 of 25 close contacts were infected in subsequent 14 days
– June and July 2020: At least 6 known infected passengers on 5 flights
• 11-hour flights (Dubai to Hong Kong)
• 100% enforced mask mandate on-board
• 0 new infections among other passengers in the subsequent 14 days

Schwartz et al. 2020, CMAJ; 192(15):E410. Freedman et al. 2020, J Travel Med; doi: 10.1093/jtm/taaa178.
Wang et al. 2020, BMJ Glob Health; 5:e002794. doi:10.1136/bmjgh-2020-002794.
Valid as of November 16, 2020
Frequently Cited Study that Cloth Masks Are Not Protective
▪ MacIntyre et al. 2015: 1,607 healthcare workers in 15 Vietnamese hospitals
– Compared: Regular use of surgical masks (3-ply), regular use of cloth masks (2-ply),
control (standard masking practice)
– Endpoint: Respiratory illness identified through self-monitoring or lab-confirmed
infection with flu, rhinovirus, or human metapneumovirus
– Outcome: Despite equal compliance wearing surgical and cloth masks, cloth masks
were statistically no better than the control situation and inferior to surgical masks
against
• Clinical upper respiratory illness
• Lab-confirmed viral infection

MacIntyre et al. 2015, BMJ Open; 5:e006577, doi:10.1136/bmjopen-2014-006577.

Valid as of November 16, 2020
Frequently Cited Study that Cloth Masks Are Not Protective
▪ Generalization of these findings to community masking is limited
– Study did not include SARS-CoV-2 infection
– Study did not include a true “no mask” group
– Study took place in a healthcare setting and not a general community setting
– Hospitalized patients and other staff were not masked (limited source control)
– Assignment to study arms was unblinded
• Possible mask-type preferences could influence self-reporting of illness
– Cloth masks were washed by users and re-used (risk of self-inoculation handling mask)
• Re-analysis of the data in 2020 found increased risk of infection from self-washing masks
– HR of infection for self-washing was 2.04 (95% CI 1.03-4.00); p=0.04
• “Healthcare workers whose cloth masks were laundered in the hospital laundry were
protected as well as those who wore disposable medical masks.” MacIntyre et al., 2020

MacIntyre et al. 2020, BMJ Open ;10:e042045, doi:10.1136/bmjopen-2020-042045.

MacIntyre et al. 2015, BMJ Open; 5:e006577, doi:10.1136/bmjopen-2014-006577.
Valid as of November 16, 2020
Community Masking to Control SARS-CoV-2
Community Studies

Valid as of November 16, 2020

Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking

▪ Seven published reports examined changes in new diagnoses or deaths with mask
mandates
– Massachusetts General Brigham (MGB) Integrated Health Care System
– Jena city, Germany
– Arizona state, United States
– 15 states and District of Columbia, United States (two analyses)
– Canada, national
– United States, national
▪ All observed reductions in new COVID-19 diagnoses (n=6) or deaths (n=3) following
recommendations for universal masking
Wang et al. 2020, JAMA; 323(14):1341-1342. Gallaway et al 2020, MMWR; 69(40):1460-1463. Lyu and Wehby 2020, Health Affairs (Millwood); 39(8):1419-1425.
Mitze et al. 2020, Institute of Labor Economics Report; DP No. 13319, http://ftp.iza.org/dp13319.pdf.
Karaivanov et al. 2020, National Bureau Of Economic Research; Working Paper 27891, http://www.nber.org/papers/w27891.
Hatzius et al. 2020, Goldman Sachs Research report https://www.goldmansachs.com/insights/pages/face-masks-and-gdp.html.
Chernozhukov et al. 2020, medRxiv: https://doi.org/10.1101/2020.05.27.20115139. Valid as of November 16, 2020
Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking

▪ MGB required masking for all health

care workers (HCW) followed two
weeks later by required masking for all
patients and visitors
▪ Despite interventions locally and
within the MGB system (see bars
below figure)
– New diagnoses among HCWs first started
to decline within ~1 week* after
implementation of full masking mandate

* Median incubation period is 4-6 days

Wang et al. 2020, JAMA; 324(7):703-704. doi:10.1001/jama.2020.12897
Valid as of November 16, 2020
Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking

▪ Political leaders mandated universal

community masking in the city of Jena
(Germany) on April 6, 2020
▪ New diagnoses leveled off starting
about 10 days later*
▪ Cumulative decline in number of new
diagnoses of about 25% within 20 days
– >50% for persons aged > 60 years
▪ Other interventions had already been
introduced (e.g., social distancing,
hand hygiene)
* Median incubation period is 4-6 days
Adapted from Mitze et al. 2020, Institute of Labor Economics Report; DP No. 13319, http://ftp.iza.org/dp13319.pdf.
Valid as of November 16, 2020
Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking

▪ Arizona mandated masking on June 17

▪ Decline in number of new cases began
about 12 days later*
▪ Further interventions applied June 29
– These interventions were coincident
with the start of the decline
– Their effect could not have been
instantaneous
– This observation suggests start of decline
was due to earlier masking mandate

Gallaway et al. 2020, MMWR; 69(40):1460-1463.10.15585/mmwr.mm6940e3. * Median incubation period is 4-6 days
Valid as of November 16, 2020
Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking

▪ Arizona mandated masking on June 17

▪ Decline in number of new cases began
about 12 days later*
▪ Further interventions applied June 29
– These interventions were coincident
with the start of the decline
– Their effect could not have been
instantaneous
– This observation suggests start of decline
was due to earlier masking mandate

Gallaway et al. 2020, MMWR; 69(40):1460-1463.10.15585/mmwr.mm6940e3. * Median incubation period is 4-6 days
Valid as of November 16, 2020
Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking
▪ Masking mandates in 15 states led to 2% decline in rate of new diagnoses by 21 days*
▪ Rate of decline steadily increased with time after mandate, doubling by 21 days

Before mandates After mandates

Daily decline in rate

of new cases
grew larger the
longer the mandates
were in place

Lyu and Wehby 2020, Health Affairs (Millwood); 39(8):1419-1425. *included D.C. and controlled for other major COVID-19 mitigation policies as time-varying.
Valid as of November 16, 2020
Jurisdictional Declines in New Diagnoses Associated With
Organizational/Political Leadership Directives for Universal Masking
▪ Mandatory masking prevented both infections and deaths; could avert more lockdowns
▪ With 15% increase in masking, estimated potential GDP savings of $1 trillion (5% GDP)
Daily Average Case Rate Daily Average Fatality Rate
* *
* *

** pp = percentage points
** pp = percentage points

Hatzius et al. 2020, Goldman Sachs Research report, https://www.goldmansachs.com/insights/pages/face-masks-and-gdp.html.

Valid as of November 16, 2020
Country-Level Declines in Deaths Associated With Timing of
Universal Masking Adoption or Mandates

▪ Evaluated mortality rates stratified by time

– From date of first diagnosis to date masking
was mandated or otherwise universally
adopted in 200 countries (including U.S.)
through May 9, 2020
– Used 3 strata based on time since infection
first identified in country
▪ During each week without masks,
mortality increased 59%

Leffler et al 2020, medRxiv; doi.org/10.1101/2020.05.22.20109231.

Valid as of November 16, 2020
The Science of Masking to Control COVID-19: Summary
▪ Cloth masks reduce community exposure to SARS-CoV-2
▪ Cloth masks offer both source control and personal protection
– The relationship is likely complementary and possibly synergistic
– Community benefit derives from the combination of these effects
– Individual benefit increases with increasing community mask use
▪ Wearing masks by both the infected and uninfected person gives the uninfected
person the most protection
– “Masking can protect you and works best for you when everyone does it”
– “When you wear a mask, you protect others as well as yourself”
▪ Universal masking policies can help avert the need for shutdowns
– Especially if combined with other non-pharmaceutical interventions such as social distancing,
hand hygiene, and adequate ventilation

Valid as of November 16, 2020

Appendix: Additional References
Slide 7: Laboratory Assessment of Cloth Masks Effectiveness: Source Control
Bandiera et al. 2000, medRxiv; https://doi.org/10.1101/2020.08.11.20145086. Davies et al. 2013, Disaster Med Public
Health Prep; 7(4):413-418. Leung et al. 2020, Nature Medicine; 26(5):676-680. Fischer et al. 2020, Sci Adv;
6(36):eabd3083. Lindsley et al. 2020, medRxiv: doi 10.1101/2020.10.05.20207241. Verma et al. 2020, Phys Fluids (1994);
32(6):061708. Alsved et al. 2020, Aerosol Science and Technology; doi 10.1080/02786826.2020.1812502. Asadi et al.
2019, Sci Rep; 9(1):2348. Morawska et al. 2009, J Aerosol Science; 40(3):256-269.Abkarian 2020; Proc Natl Acad Sci;
117(41):25237-25245.

Slide 8: Laboratory Assessment of Cloth Masks Effectiveness: Filtering Protection

Rengasamy et al. 2010, Ann Occup Hyg; 54(7):789-798. Konda et al. 2020, ACS Nano; 14(5):6339-6347. Long et al. 2020,
PLoS One; 15(10):e0240499. O'Kelly et al. 2020, BMJ Open; 10(9):e039424. Aydin et al. 2020, Extreme Mech Lett;
40:100924. Bhattacharjee et al. 2020, BMJ Open Respir Res; doi 10.1136/bmjresp-2020-000698. Maurer et al. 2020, J
Aerosol Med Pulm Drug Deliv; doi 10.1089/jamp.2020.1635. Hill et al. 2020, Nano Lett; 20(10):7642-7647. Whiley et al.
2020, Pathogens; doi:10.3390/pathogens9090762. Hao et al. 2020, Int J Hyg Environ Health; 229:113582. van der Sande
et al. 2008, PLoS One; 3(7):e2618. Chu et al. 2020, Lancet; doi 10.1016/S0140-6736(20)31183-1. Zhao et al. 2020, Nano
Lett; 20:5544−5552. Parlin et al. 2020, PLoS One; 15(9):e0239531. Kahler et al. 2020, J Aerosol Sci; 148:105617. Ueki et
al. 2020, mSphere; doi.org/10.1128/mSphere.00637-20.

Valid as of November 16, 2020

For more information, contact CDC
1-800-CDC-INFO (232-4636)
TTY: 1-888-232-6348 www.cdc.gov

The findings and conclusions in this report are those of the authors and do not necessarily represent the
official position of the Centers for Disease Control and Prevention.

Valid as of November 16, 2020