Associations between wearing masks, washing hands, and social distancing practices, and risk of COVID-19 infection in public: a cohort-based case-control study in Thailand =========================================================================================================================================================================== * Pawinee Doung-ngern * Repeepong Suphanchaimat * Apinya Panjangampatthana * Chawisar Janekrongtham * Duangrat Ruampoom * Nawaporn Daochaeng * Napatchakorn Eungkanit * Nichakul Pisitpayat * Nuengruethai Srisong * Oiythip Yasopa * Patchanee Plernprom * Pitiphon Promduangsi * Panita Kumphon * Paphanij Suangtho * Peeriya Watakulsin * Sarinya Chaiya * Somkid Kripattanapong * Thanawadee Chantian * Chawetsan Namwat * Direk Limmathurotsakul ## Summary **Background** Effectiveness of personal protective measure against COVID-19 infection is largely unknown. **Methods** We conducted a retrospective case-control study, using a cohort of contact tracing records in Thailand. A total of 1,050 asymptomatic contacts of COVID-19 patients between 1 and 31 March 2020 were retrospectively interviewed by phone about their protective measures against COVID-19 infection. Cases were defined as asymptomatic contacts who were diagnosed with COVID-19 by 21 April 2020. Multilevel mixed-effect logistic regression models were used **Findings** Overall, 211 (20%) were diagnosed with COVID-19 by 21 Apr 2020 (case group) while 839 (80%) were not (control group). Fourteen percent of cases (29/210) and 24% of controls (198/823) reported wearing either non-medical or medical masks all the time during the contact period. Wearing masks all the time (adjusted odds ratio [aOR] 0.23; 95%CI 0.09-0.60) was independently associated with lower risk of COVID-19 infection compared to not wearing masks, while wearing masks sometimes (aOR 0.87; 95%CI 0.41-1.84) was not. Shortest distance of contact >1 meter (aOR 0.15; 95%CI 0.04-0.63), duration of close contact ≤15 minutes (aOR 0.24; 95%CI 0.07-0.90) and washing hands often (aOR 0.33; 95%CI 0.13-0.87) were significantly associated with lower risk of infection. Sharing a cigarette (aOR 3.47; 95%CI 1.09-11.02) was associated with higher risk of infection. Those who wore masks all the time were more likely to wash hands and practice social distancing. We estimated that if everyone wore a mask all the time, washed hands often, did not share a dish, cup or cigarette, maintained distances >1 meter and spent ≤15 minutes with close contacts, cases would have been reduced by 84%. **Interpretation** Our findings support consistently wearing masks, washing hands, and social distancing in public to protect against COVID-19 infections. Combining measures could substantially reduce infections in Thailand. **Funding** The study was supported by the DDC, MoPH, Thailand. DL is supported by the Wellcome Trust (106698/Z/14/Z). Keywords * COVID-19 * SARS-CoV-2 * mask * medical mask * non-medical mask * hand washing * social distancing * contact tracing ## Introduction There is an urgent need to evaluate the effectiveness of wearing masks by healthy persons in the general public against COVID-19 infections. During the early stages of the outbreak of COVID-19, the World Health Organization (WHO) announced on 27 February 2020 that, “For asymptomatic individuals, wearing a mask of any type is not recommended”.1 The rationale, at that time, was to avoid unnecessary cost, procurement burden, and a false sense of security.1 A number of systematic reviews also found no conclusive evidence to support the widespread use of masks in public against respiratory infectious diseases such as influenza, SARS and COVID-19.2-4 However, China and many countries in Asia including South Korea, Japan and Thailand have recommended the use of face mask among the general public since early in the outbreak.5 There is also increasing evidence that COVID-19 patients can have a “pre-symptomatic” period, during which infected persons can be contagious and, therefore, transmit the virus to others before symptoms develop.6 This led to the change of the recommendation of the US Centers for Disease Control and Prevention, updated on 4 April 2020, from warning the public against wearing face masks to advising everyone to wear a cloth face covering when in public.7 On 6 April and 5 June 2020, WHO updated their advice on the use of masks for the general public, and encouraged countries to conduct research on this topic.6 Thailand has been implementing multiple measures to mitigate transmission of COVID-19 since the beginning of the outbreak. The country has established thermal screening at airports since 3 January 2020, and detected the first case of COVID-19 outside China, a traveler from Wuhan arriving at Bangkok Suvarnabhumi airport, on 8 January 2020.8 The country utilized the Surveillance and Rapid Response Team (SRRT), together with Village Health Volunteers, to perform contact tracing, educate the public about the disease and monitor the close contacts of COVID-19 patients in quarantine. The SRRT is an epidemiologic investigation team trained to conduct surveillance, investigations and initial controls of communicable diseases; including H5N1, SARS and MERS.9 Currently, there are more than 1,000 SRRTs established at district, provincial and regional levels in the country,9 working on contact tracing for COVID-19. In February 2020, public pressure to wear masks was high, medical masks were difficult to procure by the public, and the government categorized medical masks as price-controlled goods and announced COVID-19 as a dangerous communicable disease according to the Communicable Disease Act 2015 in order to empower officials to quarantine contacts and close venues. On 3 March, the Ministry of Public Health (MoPH) announced the recommendation of cloth mask for the public. On 18 March, schools, universities, bars, nightclubs and entertainment venues were closed. On 26 March, while the country was reporting approximately 100-150 new COVID-19 patients per day, the government declared a national state of emergency, prohibited public gatherings, and enforced everyone to wear a face mask on public transport.10 On 21 April, 19 new PCR-confirmed COVID-19 patients were announced by the MoPH, Thailand, bringing the total number of patients to 2,811 patients.10 Given the lack of currently available evidence, we evaluated the effectiveness of mask wearing, hand washing, social distancing and other protective measures against COVID-19 infection in public in Thailand ## Methods ### Study design and participants We conducted a retrospective case-control study in which both cases and controls were drawn from a cohort of contact tracing records of the central SRRT team, Department of Disease Control (DDC), MoPH, Thailand (Figure 1). Contacts were defined by the DDC MoPH as individuals who had activities together with or were in the same location(s) as a COVID-19 patient.11,12 The main aim of contact tracing was to identify and evaluate contacts, perform PCR diagnostic tests, and quarantine high-risk contacts. Contacts were tested with PCR assays per national contact tracing guidelines (Supplementary Text).11,12 All PCR tests were performed at laboratories certified for COVID-19 testing by the National Institute of Health of Thailand. Data on risk factors associated with COVID-19 infection, such as type of contact and use of mask, were recorded during the contact investigation, but not complete. ![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.11.20128900/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/F1) Figure 1. Study flow diagram **Footnote of Figure 1**. SRRT= Surveillance and Rapid Response Team (SRRT), Ministry of Public Health (MoPH), Thailand The central SRRT team was tasked to perform contact investigations for any cluster with at least five PCR-confirmed COVID-19 patients from the same location(s) within a one-week period.11 We primarily used these data to identify asymptomatic contacts of COVID-19 patients between 1 and 31 March 2020. To reduce the bias of the selection of asymptomatic contacts, all contact tracing records of the central SRRT team were used in the study. ### Procedures We then conducted telephone calls and asked details about their contacts with COVID-19 patients (e.g. date, location, duration and distance of contacts), whether they wore masks, washed their hands and performed social distancing during the contact period, and whether the COVID-19 patient, if known, wore a mask. We also asked, and checked using records of the DDC, whether and when they were sick and diagnosed with COVID-19. To include only asymptomatic contacts in the study, we excluded people from the analysis who already had any symptoms of COVID-19 on the first day of contact. We also excluded contacts whose contact locations were healthcare facilities because this study aimed to focus on infection in the public. Asymptomatic contacts, cases, controls, index patients, primary index patients and COVID-19 patients were defined as described in Table 1. The reporting of this study follows the STROBE guidelines. View this table: [Table 1.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/T1) Table 1. Definitions used in the study ### Selection of cases and controls We defined asymptomatic contacts who were later diagnosed as COVID-19 patients using PCR assays by 21 Apr 2020 as cases (Table 1). All asymptomatic contacts who were not diagnosed as COVID-19 patients using PCR assays by 21 Apr 2020 were controls. We arbitrarily used 21 days after 31 March as the cutoff based on the evidence that most COVID-19 patients would likely develop symptoms within 14 days13 and it should take less than another 7 days for symptomatic patients to present at healthcare facilities and be tested for COVID-19 with PCR assays. ### Statistical analysis Odds ratios (OR) and 95% confidence intervals were estimated for associations between development of COVID-19 and baseline covariates, such as wearing masks, washing hands and social distancing using logistic regression with a random effect for location and a random effect for index patient nested within the same location. The interviewer identified the index patient, the symptomatic COVID-19 patient who had the closet contact, if an asymptomatic contact had contacted with more than one symptomatic COVID-19 patient. The percentage of missing values in the variable whether the COVID-19 patients wore a mask was 27%, and the variable was not included in the analyses. We assumed that missing values were missing at random and used imputation by chained equations. We created 10 imputed datasets and the imputation model included all listed confounders and the case-control indicator. We developed the final multilevel mixed-effect logistic regression models on the basis of previous knowledge and a purposeful selection method.14 Compliance of wearing mask was strongly associated with lower risk of COVID-19 infection in multivariable models, while type of mask was not. Due to the collinearity between the two variables, type of mask was not included in the final model. We also estimated OR and 95% confidence intervals for associations between compliance of mask wearing and other practices; including washing hands and social distancing using multinomial logistic regression models and the imputed data set. Logistic regression was also used to estimate p-values for pairwise comparisons. Bonferroni correction was not performed. We estimated secondary attack rate using definitions as described in Table 1, to allow for comparison with other studies. We conducted a sensitivity analysis by including type of mask in the multilevel mixed-effects logistic regression model for COVID-19 infection. We also tested a pre-defined interaction between type of mask and compliance of wearing masks. To respond to national policy, we also estimated population attributable fraction (PAF) using the imputed dataset and a direct method based on logistic regression as described previously (Supplementary Text).15,16 The final multivariable model was modified by considering each risk factor dichotomously, and PAF was calculated by subtraction of the total number of predicted cases from total number of observed cases, divided by the total number of observed cases. STATA version 14.2 and R version 4.0.0 were used for all analyses. ### Role of the funding source The study sponsor did not participate in study design, data collection, analysis, interpretation of data, or writing of the article. PD, RS and DL had access to all the data in the study. ## RESULTS Contact tracing by the central SRRT team consisted of 1,716 individuals who had contact with or were in the same location as a COVID-19 patient, and who were associated with three large clusters in nightclubs, boxing stadiums and a state enterprise office in Thailand (Figure 1). Overall, we considered 18 individuals as primary index patients because they were the first to have symptoms at those places, had symptoms since the first day of visiting those places, or were considered to be the origin of infection of cases based on the contact investigations; 11 from the nightclub cluster, 5 from the boxing stadium cluster and 2 from the state enterprise office cluster. Timelines of primary index patients from nightclub, boxing stadium and state enterprise clusters are described in details in Supplementary Text and Supplementary Figure 1-3. All 18 primary index patients were excluded from the analysis of the case-control study. ![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.11.20128900/F2.medium.gif) [Figure 2.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/F2) Figure 2. Development and transmission of COVID-19 among asymptomatic contacts included in the study **Footnote of Figure 2**. A, B and C represent the nightclub cluster, boxing stadium cluster and state enterprise office cluster, respectively. Black nodes represent primary index patients, red dots represent cases, and green dots represent controls. Orange dots represent index patients (confirmed COVID-19 patients) who could not be contacted by the study team. Black lines represent household contacts, purple lines represent contacts at workplaces and gray lines represent contacts at other locations. Definition of index patients, cases and controls are listed in Table 1. After retrospectively interviewing each contact by phone and applying the exclusion criteria (Figure 1), we included 1,050 asymptomatic contacts who had contact with or were in the same location as a symptomatic COVID-19 patient between 1 and 31 March 2020 in the analysis. The median age of individuals was 38 years (IQR 28-51) and 55% were male (Table 1). Most asymptomatic contacts included in the study were associated with the boxing stadium cluster (61%, n=645), with 36% (n=374) with the nightclub cluster, and 3% (n=31) with the state enterprise office cluster. Overall, 211 (20%) asymptomatic contacts were later diagnosed with COVID-19 by 21 Apr 2020 (case group) and 839 (80%) were not (control group). Of the 211 cases, 150 (71%) had symptoms prior to the diagnosis of COVID-19 using PCR assays. The last date that a COVID-19 case was diagnosed was 9 April 2020. Of 839 controls, 719 (86%) were tested with PCR assays at least once. Figure 2 illustrates contacts (and possible transmission of COVID-19 infections) between index patients to asymptomatic contacts included in the study. A total of 228, 144 and 20 asymptomatic contacts had contact with index patients at nightclubs, boxing stadiums and the state enterprise office, respectively. The others were contacts of cases related to nightclubs, boxing stadiums and the state enterprise office at workplaces (n=277), households (n=230) and other places (n=151). ### Primary analysis Table 2 shows there was a negative association between risk of COVID-19 infection and shortest distance of contact >1 meter (adjusted odds ratio [aOR] 0.15, 95% confidence interval [CI] 0.04-0.63), duration of contact ≤15 minutes (aOR 0.24, 95%CI 0.07-0.90), washing hands often (aOR 0.33, 95%CI 0.13-0.87) and wearing masks all the time (aOR 0.23, 95%CI 0.09-0.60). Wearing masks sometimes was not significantly associated with lower risk of infection (aOR 0.87, 95%CI 0.41-1.84), whereas sharing cigarettes was associated with higher risk of COVID-19 infection (aOR 3.47, 1.09-11.02). Type of mask was associated with infection at the univariable model. Compliance of wearing masks was strongly associated with lower risk of COVID-19 infection in multivariable models, while type of mask was not. Due to the collinearity between the two variables, type of mask was not included in the final model. View this table: [Table 2.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/T2) Table 2. Factors associated with COVID-19 infections ### Association between compliance of mask wearing and other social distancing practices Since wearing masks all the time was found to be negatively associated with COVID-19 infection, we wanted to explore characteristics of those patients because of a potential false sense of security caused by wearing masks. We found that those who wore masks all the time were more likely to have shortest distance of contact >1 meter (25% vs. 18%, pairwise p=0.03), have duration of contact ≤15 minutes (26% vs 13%, pairwise p<0.001) and wash their hands often (79% vs. 26%, pairwise p<0.001) compared with those who did not wear masks (Table 3). We found that those who wore masks sometimes were more likely to wash their hands often (43% vs. 26%, pairwise p<0.001) compared with those who did not wear masks. However, they were more likely to have physical contact (50% vs. 42%, pairwise p=0.03) and duration of contact >60 minutes (75% vs. 67%, pairwise p=0.04) compared with those who did not wear masks. View this table: [Table 3.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/T3) Table 3. Factors associated with compliance of mask wearing ### Secondary attack rate Overall, 982 (94%) were contacts with high-risk exposure. All 68 asymptomatic contacts without high-risk exposure were controls. Among asymptomatic contacts with high-risk exposure included in the study, the nightclub secondary attack rate was 16% (35/213), the boxing stadium secondary attack rate was 87% (125/144), the workplace secondary attack rate was 4% (11/250), the household secondary attack rate was 17% (38/230), and the secondary attack rate at other places was 1% (2/145). ### Sensitivity analyses To determine if type of mask was independently associated with infection, we included type of mask in the multivariable model (Supplementary Table 1). Type of mask was not independently associated with infection (p=0.55). Interaction between type of mask and compliance of mask wearing was not observed. ### Population attributable fraction (PAF) Using the direct method to calculate PAF, we estimated that the proportional reduction in cases that would occur if everyone wore a mask all the time during contact with index patients was 0.28 (Table 4). Among modifiable risk factors evaluated, PAF of shortest distance of contact <1 meter was highest at 0.40. If everyone wore a mask all the time, washed hands often, did not share a dish, cup or cigarette, had shortest distance of contact >1 meter and had duration of close contact ≤15 min, cases would have been reduced by 84%. The reduction could be as high as 97% to 99% in most places under evaluation, except boxing stadiums with an estimated reduction of 75%. View this table: [Table 4.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/T4) Table 4. Population attributable fraction (PAF) of risk factors for COVID-19 infection ## DISCUSSIONS This cohort-based case-control study provides a supporting evidence that wearing masks, washing hands and social distancing are independently associated with lower risk of COVID-19 infection in the general public. We observed that wearing masks all the time when exposed to someone with COVID-19 was associated with lower risk of infection, while wearing masks sometimes was not. This supports the recommendation that people should be wearing their masks correctly at all times in public and at home when there is an increased risk.5-7 We quantified the effectiveness of different measures that could be implemented to protect against COVID-19 infections in nightclubs, stadiums, workplaces and other public gathering places. We estimated that adopting all recommendations (wearing masks all the time, washing hands often, not sharing dishes, cups or cigarettes, maintaining a distance of <1 meter and spending ≤15 minutes with a close contact) could result in controlling 84% of the burden of COVID-19 infections in our setting during the study period. We recommend that all public gathering places consider all measures to protect against COVID-19 and new pandemic diseases in the future. The effectiveness of wearing masks observed in this study is consistent with previous studies; including a randomized-controlled trial (RCT) showing that adherent use of a face mask reduce the risk of influenza-like illness17 and case-control studies which found that wearing masks is associated with lower risk of SARS infection.18-20 While previous studies found use of surgical masks or 12–16-layer cotton masks demonstrated protection against coronavirus infection in the community,18-20 we did not observe a difference between wearing non-medical and medical masks in the general population. Therefore, we strongly support wearing non-medical masks in public to protect against COVID-19 infections. Even though the risk perception of COVID-19 threat can increase the likelihood of wearing medical masks in other settings,21 we maintain that medical masks should be reserved for healthcare workers. This study found a negative association between risk of COVID-19 infection and social distancing (i.e. distance and duration of contact), which is consistent with previous studies which found that at least 1-meter physical distancing was strongly associated with a large protective effect, and distances of 2 meters could be more effective.20 Effectiveness of hand hygiene is consistent with the previous studies.22 The household secondary attack rate in our study (17%) is comparable with those reported ranging from 11% to 19%,23,24 and relatively high compared to workplaces and other places. While challenging and sometimes impractical, household members should immediately separate a person who develops any possible symptoms of COVID-19 from other household members (i.e. a sick person should stay in a specific room, use a separate bathroom, if possible, and do not share dishes, cups and other utensils in the households).25 All household members should be encouraged to wear masks, keep washing hands and perform social distancing to the extent possible.26 The high number of COVID-19 patients associated with nightclubs in Bangkok is comparable to COVID-19 outbreak associated with Itaewon nightclub cluster in Seoul, Korea, in May 2020.27 Similarly, we also found individuals who visited several nightclubs in the same area during the short period of time. The high number of COVID-19 patient cluster associated with boxing stadiums in Bangkok is similar to a cluster likely to have been associated with a football match in Italy in February 2020.28 The secondary attack rate of COVID-19 at a choir practice in the U.S. was reported to be as high as 53%,29 and the secondary attack rates in public gathering places with high density of people shouting and cheering, such as football and boxing stadiums, are still largely unknown. It is likely that clear and consistent public messaging from policy makers prevents a false sense of security and promotes compliance with social distancing in Thailand. We found that those who wore masks all the time were also more likely to wash hands and perform social distancing. It is recommended that both mainstream and social media support public health responses by teaming with government in providing consistent, simple and clear messages.30 Both positive and negative messages can influence the public.30 In Thailand, daily briefings of Thailand’s Centre for COVID-19 Situation Administration (CCSA) gave clear and consistent messages on social distancing every day, as well as how to put on a mask and wash hands. The regular situation reports and advices by CCSA may have improved the confidence in the public and compliance with the recommendations. Nonetheless, public messaging on how to wear masks correctly needs to be consistently delivered, particularly among those who wear masks sometimes or incorrectly (e.g. not covering both nose and mouth). This is because, based on our findings, those who wear masks intermittently could be a group that did not practice social distancing adequately. There are several limitations of the study. First, our finding might not be generalizable to all settings, since findings were based on contacts associated with three major COVID-19 clusters in Thailand during March 2020. Second, the estimated odds ratios were based on a condition that the contact with index patients occurred. Our study did not evaluate or take into account the probability of contacting index patients in public. Our study also lacked power to evaluate whether wearing masks by index patients can prevent infecting others in public. Third, our findings were based on PCR testing per national contact tracing guideline,11,12 and as such the estimated odds ratios might not take account of all asymptomatic infections. Fourth, it is impossible to identify every potential contact an individual has and some individuals may have been contacts to more than one COVID-19 patient. Hence, our estimated secondary attack rates among contacts with high-risk exposure could be over or under-estimated. Fifth, findings were subject to common biases of retrospective case-control studies; including memory bias, observer bias and information bias. Nonetheless, we used structured interviews, whereby each participant was asked the same set of defined questions, to reduce potential biases. As measures against COVID-19 are being implemented or relaxed in many countries worldwide, it is important that we continue to expand our understanding about the effectiveness of each personal protective measure. Wearing masks, washing hands and social distancing are strongly associated with lower risk of COVID-19 infections. We strongly support wearing non-medical masks in public to protect against COVID-19 infections. We also suggest that medical masks should be reserved for healthcare workers. Everyone should also wash their hands frequently and comply with recommendations of social distancing. ### Research in context #### Evidence before this study We searched PubMed and medRxiv for peer-reviewed articles, preprints and research reports on effectiveness of wearing masks, washing hands, and social distancing against COVID-19 and related diseases; including severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS). We identified two case-control studies from China and one cohort study from Vietnam evaluating effectiveness of wearing masks against SARS in public. A recent meta-analysis estimated an overall relative risk of wearing masks in public (0.56; 95%CI 0.40-0.79) based on the three studies. A meta-analysis including six case-control studies on SARS estimated odds ratio of hand washing more than 10 times daily in preventing disease transmission (0.45; 95%CI 0.36-0.57). The recent meta-analysis found a negative association between risk of infection and physical distancing; six of 38 studies included were on COVID-19. No comparative studies of wearing masks or washing hands in public against COVID-19 infections were identified. #### Added value of this study We did a cohort-based case-control study of 1,050 asymptomatic contacts of COVID-19 patients in the general public in Thailand. Our findings are, to the best of our knowledge, the first comparative study to quantitatively assess the protective effects of wearing masks and washing hands against COVID-19 infections in the general population. We observed that wearing masks all the time when exposed to someone with COVID-19 was associated with lower risk of infection, while wearing masks sometimes was not. We also found that those who wore masks all the time were more likely to perform social distancing, while those who wore masks intermittently were less likely to. We estimated that adopting all recommendations (wearing masks all the time, washing hands often, not sharing dishes, cups or cigarettes, maintaining a distance of <1 meter and spending <15 minutes with a close contact) could result in controlling 84% of the burden of COVID-19 infections in our setting during the study period. The reduction could be as high as 97% to 99% in most places, except boxing stadiums with an estimated reduction of 75%. #### Implication of all the available evidence As lockdowns are being relaxed in many countries, our findings provide some evidence supporting consistently wearing non-medical mask, washing hands, and performing social distancing to protect against COVID-19 in public gathering places. The effectiveness of complying with all measures could be high. However, in places with a high density of people, additional measures may be required. Policy makers should strive to deliver public messaging on recommendations against COVID-19 infection clearly and consistently, particularly among those who wear masks intermittently or incorrectly. Our data also showed that no protective measure was associated with complete protection from COVID-19 infection. All measures, including wearing non-medical masks, washing hands, and social distancing are required to protect against COVID-19 infections in public. ## Data Availability All data in aggregate are reported in the manuscript. ## Contributors PD, RS, CN and DL contributed to design of the study. PD, RS and DL contributed to setting up the database and quality control. AP, CJ, DR, ND, NE, NP, NS, OY, PaP, PiP, PK, PS, PW, SC, SK and TC contributed to data collection. DL carried out the main statistical analysis. PD and RS coordinated the study and contributed to the statistical analyses. PD, RS and DL contributed to interpretation of the results and drafted the manuscript. All authors commented on drafts and read and approved the final manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. PD is the guarantor. ## Declaration of interests We declare no competing interests. ## Data sharing All data in aggregate are reported in the manuscript. ## Funding The study was supported by the DDC, MoPH, Thailand. DL is supported by the Wellcome Trust (106698/Z/14/Z). The study sponsors did not participate in study design, data collection, analysis, interpretation of data, or writing of the article. ## Ethical approval As this study was part of the routine situation analysis and outbreak investigation of the DDC MoPH Thailand, it was not required to obtain ethics approval and no written informed consent was collected. However, the study team strictly followed ethical standards in research; that is, all individual information was strictly kept confidential and not reported in the paper. The DDC MoPH Thailand approved the analysis and reporting of data in aggregate. ## Supplementary Text ### Supplementary Methods #### Study design and participants Contacts were classified by MoPH as high-risk contacts if they were family members or lived in the same household as a COVID-19 patient, if they were within 1-meter distance longer than 5 minutes of a COVID-19 patient, if they were exposed to cough, sneeze or secretions of a COVID-19 patient and were not wearing a protective gear, such as mask, or if they were in the same closed environment (e.g. room, nightclub, stadium, vehicle) within 1-meter distance longer than 15 minutes of a COVID-19 patient and they were not wearing a protective gear, such as mask.12,13 Contacts were classified as low-risk contacts if they had activities together with or were in the same locations as a COVID-19 patient, but did not fulfil the criteria of a high-risk contact.12,13 All high-risk contacts with any symptoms were tested with a PCR assay and quarantined in a hospital or a quarantine site.12,13 All high-risk contacts without any symptoms were self-quarantined at home.12,13 Before 23 March 2020, all high-risk contacts without any symptoms were tested using PCR assays on day 5 after the last date of exposure to a case.12 As of 23 March 2020, all household contacts were tested using PCR assays regardless of their symptoms. Other high-risk contacts were tested only if they developed any COVID-19 symptoms.13 All low-risk contacts were recommended to perform self-monitoring for 14 days, and visit healthcare facilities immediately for PCR-assays if they develop any symptoms of COVID-19.12,13 ### Statistical analysis To respond to the national policy, we estimated direct population attributable fraction (PAF) using the imputed dataset and the direct method as previously described.27,28 Direct PAF can be obtained by calculating PAFs directly from individuals’ data using logistic regression.27,28 First, we had to modify our final logistic regression model by considering each risk factor dichotomously. Then, irrespective of exposure to each risk factor for each individual, that factor was removed from the population by calculating probability based on all observations as unexposed. The predicted probability of developing COVID-19 infection for each asymptomatic contact, with the assumption that there was no exposure to a certain risk factor, is: ![Formula][1] Pki is representative of predicted probability of COVID-19 infection in individual asymptomatic contact k, assuming no exposure to a specific risk factor (xi); βj indicates the regression coefficient of risk factor (xj), except risk factor number i (xi). Subsequently, the sum of all predicted probabilities for all individuals in the study would be equal to adjusted estimate of total cases, which is anticipated in the absence of that specific risk factor (xi). Then, PAF was estimated by subtraction of the total number of predicted cases from total number of observed cases, divided by the total number of observed cases: ![Formula][2] ### Supplementary Results For the pub cluster, we identified 11 primary index patients who started having symptoms from 4 to 8 March and were diagnosed (and isolated) from 3 to 10 March (Supplementary Figure 1). Those primary index patients visited multiple nightclubs included in the analysis during the study period, and 35 of 228 (15%) asymptomatic contacts at nightclubs had PCR-confirmed COVID-19 infections after the contact (Figure 2, Cluster A). For the boxing stadium cluster, we identified 5 primary index patients who started having symptoms from 6 to 12 March and were diagnosed (and isolated) from 11 to 21 March (Supplementary Figure 2). Those primary index patients visited multiple boxing stadiums included in the analysis during the study period, and 125 of 144 (87%) asymptomatic contacts at the boxing stadiums had PCR-confirmed COVID-19 infections after the contact (Figure 2, Cluster B). Of the two primary index patients for the office cluster; one had had symptoms since 15 March 2020 (Primary index patient C1 in Supplementary Figure 3) and was considered as the source of infection to one new case in the office during the study period. The other primary index patient (Primary index patient C2 in Supplementary Figure 3) was a household member of a staff at the office, and was considered as the source of infection to that staff via household contact. View this table: [Supplementary Table 1.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/T5) Supplementary Table 1. Factors associated with COVID-19 infections in a multivariable model including type of mask ![Supplementary Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.11.20128900/F3.medium.gif) [Supplementary Figure 1.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/F3) Supplementary Figure 1. Timeline and possible transmission of primary index patients of the pub cluster ![Supplementary Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.11.20128900/F4.medium.gif) [Supplementary Figure 2.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/F4) Supplementary Figure 2. Timeline and possible transmission of primary index patients of the boxing stadium cluster ![Supplementary Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.11.20128900/F5.medium.gif) [Supplementary Figure 3.](http://medrxiv.org/content/early/2020/06/19/2020.06.11.20128900/F5) Supplementary Figure 3. Timeline and possible transmission of primary index patients of the state enterprise office cluster ## Acknowledgement We thank all participants and all COVID-19 patients involved in providing information. We thank all SRRT members at the central, regional, provincial and district levels, as well as all Village Health Volunteers in Thailand. We thank Pattraporn Klanjatturat and Inthira Yamabhai for the technical assistance. We thank Dr. Suwannachai Wattanayingcharoenchai, Dr. Sombat Thanprasertkul, Dr. Panithee Thammawijaya and Dr. Walairat Chaifoo of the DDC, MoPH, and Dr. Virsasakdi Chongsuvivatwong from Prince of Songkhla University for their advices and direction. * Received June 11, 2020. * Revision received June 18, 2020. * Accepted June 19, 2020. * © 2020, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution 4.0 International), CC BY 4.0, as described at [http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/) ## References 1. 1.WHO. Rational use of personal protective equipment for coronavirus disease (COVID-19): interim guidance, 27 February 2020. [https://apps.who.int/iris/handle/10665/331215](https://apps.who.int/iris/handle/10665/331215) (accessed 16 June 2020). 2. 2.Long Y, Hu T, Liu L, et al. Effectiveness of N95 respirators versus surgical masks against influenza: A systematic review and meta-analysis. J Evid Based Med 2020. 3. 3.Cowling BJ, Zhou Y, Ip DK, Leung GM, Aiello AE. Face masks to prevent transmission of influenza virus: a systematic review. Epidemiol Infect 2010; 138(4): 449–56. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1017/S0950268809991658&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20092668&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) 4. 4.Jefferson T, Del Mar CB, Dooley L, et al. Physical interventions to interrupt or reduce the spread of respiratory viruses. Cochrane Database Syst Rev 2011; (7): CD006207. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/14651858.CD006207.pub4&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21735402&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) 5. 5.Feng S, Shen C, Xia N, Song W, Fan M, Cowling BJ. Rational use of face masks in the COVID-19 pandemic. Lancet Respir Med 2020. 6. 6.WHO. Advice on the use of masks in the context of COVID-19: interim guidance, 5 June 2020. [https://apps.who.int/iris/handle/10665/332293](https://apps.who.int/iris/handle/10665/332293) (accessed 16 June 2020). 7. 7.Centers for Disease Control and Prevention. Coronavirus disease 2019 (COVID-19) How to Protect Yourself and Others 2020. [https://www.cdc.gov/coronavirus/2019-ncov/prevent-getting-sick/prevention-H.pdf](https://www.cdc.gov/coronavirus/2019-ncov/prevent-getting-sick/prevention-H.pdf) (accessed 16 June 2020). 8. 8.Okada P, Buathong R, Phuygun S, et al. Early transmission patterns of coronavirus disease 2019 (COVID-19) in travellers from Wuhan to Thailand, January 2020. Euro Surveill 2020; 25(8). 9. 9.Chunsuttiwat S. Response to avian influenza and preparedness for pandemic influenza: Thailand’s experience. Respirology 2008; 13 Suppl 1: S36–40. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1440-1843.2008.01256.x&link_type=DOI) 10. 10.WHO Thailand. Coronavirus disease 2019 (COVID-19) WHO Thailand Situation Report 21 April 2020 2020 [https://www.who.int/docs/default-source/searo/thailand/2020-04-21-tha-sitrep-59-covid19-new-template-final.pdf?sfvrsn=31bc500a_0](https://www.who.int/docs/default-source/searo/thailand/2020-04-21-tha-sitrep-59-covid19-new-template-final.pdf?sfvrsn=31bc500a_0) (accessed 16 June 2020). 11. 11.Department of Disease Control, Ministry of Public Health, Thailand. Coronavirus Disease 2019: COVID-19, upated on 3 March 2020 2020. [https://ddc.moph.go.th/viralpneumonia/file/guidelines/G\_Invest\_03\_2.pdf](https://ddc.moph.go.th/viralpneumonia/file/guidelines/G_Invest_03_2.pdf) (accessed 16 June 2020). 12. 12.Department of Disease Control, Ministry of Public Health, Thailand. Coronavirus Disase 2019: COVID-19, updated on 23 March 2020 2020. [https://ddc.moph.go.th/viralpneumonia/file/g\_srrt/g\_srrt\_250363.pdf](https://ddc.moph.go.th/viralpneumonia/file/g_srrt/g_srrt_250363.pdf) (accessed 16 June 2020). 13. 13.Lauer SA, Grantz KH, Bi Q, et al. The Incubation Period of Coronavirus Disease 2019 (COVID-19) From Publicly Reported Confirmed Cases: Estimation and Application. Ann Intern Med 2020. 14. 14.Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source Code Biol Med 2008; 3: 17. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/1751-0473-3-17&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19087314&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) 15. 15.Azimi SS, Khalili D, Hadaegh F, Yavari P, Mehrabi Y, Azizi F. Calculating population attributable fraction for cardiovascular risk factors using different methods in a population based cohort study. J Res Health Sci 2015; 15(1): 22–7. 16. 16.Ruckinger S, von Kries R, Toschke AM. An illustration of and programs estimating attributable fractions in large scale surveys considering multiple risk factors. BMC Med Res Methodol 2009; 9: 7. 17. 17.MacIntyre CR, Cauchemez S, Dwyer DE, et al. Face mask use and control of respiratory virus transmission in households. Emerg Infect Dis 2009; 15(2): 233–41. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3201/eid1502.081167&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19193267&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000263230700012&link_type=ISI) 18. 18.Wu J, Xu F, Zhou W, et al. Risk factors for SARS among persons without known contact with SARS patients, Beijing, China. Emerg Infect Dis 2004; 10(2): 210–6. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3201/eid1002.030730&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15030685&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000188867700010&link_type=ISI) 19. 19.Lau JT, Tsui H, Lau M, Yang X. SARS transmission, risk factors, and prevention in Hong Kong. Emerg Infect Dis 2004; 10(4): 587–92. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3201/eid1004.030628&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15200846&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000220578600005&link_type=ISI) 20. 20.Chu DK, Akl EA, Duda S, et al. Physical distancing, face masks, and eye protection to prevent person-to-person transmission of SARS-CoV-2 and COVID-19: a systematic review and meta-analysis. Lancet 2020. 21. 21.Huynh TD. The more I fear about COVID-19, the more I wear medical masks: A survey on risk perception and medical masks uses. medRxiv 2020: 2020.03.26.20044388. 22. 22.Jefferson T, Del Mar C, Dooley L, et al. Physical interventions to interrupt or reduce the spread of respiratory viruses: systematic review. BMJ 2009; 339: b3675. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjE3OiIzMzkvc2VwMjFfMS9iMzY3NSI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDIwLzA2LzE5LzIwMjAuMDYuMTEuMjAxMjg5MDAuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 23. 23.Bi Q, Wu Y, Mei S, et al. Epidemiology and transmission of COVID-19 in 391 cases and 1286 of their close contacts in Shenzhen, China: a retrospective cohort study. Lancet Infect Dis 2020. 24. 24.Jing Q-L, Liu M-J, Yuan J, et al. Household Secondary Attack Rate of COVID-19 and Associated Determinants. medRxiv 2020: 2020.04.11.20056010. 25. 25.Centers for Disease Control and Prevention. Coronavirus Disease 2019 What To Do If You Are Sick 2020. [https://www.cdc.gov/coronavirus/2019-ncov/if-you-are-sick/steps-when-sick.html](https://www.cdc.gov/coronavirus/2019-ncov/if-you-are-sick/steps-when-sick.html) (accessed 16 June 2020). 26. 26.Centers for Disease Control and Prevention. Households Living in Close Quarters: How to Protect Those That Are Most Vulnerable 2020. [https://www.cdc.gov/coronavirus/2019-ncov/daily-life-coping/living-in-close-quarters.html](https://www.cdc.gov/coronavirus/2019-ncov/daily-life-coping/living-in-close-quarters.html) (accessed 16 June 2020). 27. 27.PROMED. PRO/AH/EDR> COVID-19-update (176): China, S. Korea, countries w/ high local transmission 2020. [https://promedmail.org/promed-post/?id=7325551](https://promedmail.org/promed-post/?id=7325551) (accessed 16 June 2020). 28. 28.Boccia S, Ricciardi W, Ioannidis JPA. What Other Countries Can Learn From Italy During the COVID-19 Pandemic. JAMA Intern Med 2020. 29. 29.Hamner L, Dubbel P, Capron I, et al. High SARS-CoV-2 Attack Rate Following Exposure at a Choir Practice - Skagit County, Washington, March 2020. MMWR Morb Mortal Wkly Rep 2020; 69(19): 606–10. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.11.20128900.atom) 30. 30.Hopman J, Allegranzi B, Mehtar S. Managing COVID-19 in Low- and Middle-Income Countries. JAMA 2020. [1]: /embed/graphic-8.gif [2]: /embed/graphic-9.gif