Analysis of SARS-CoV-2 transmission in different settings, among cases and close contacts

Title Page 1 2 Manuscript Title 3 Analysis of SARS-CoV-2 transmission in different settings, among cases and close contacts 4 from the Tablighi cluster in Brunei Darussalam 5 6 Manuscript type: Original Research Article 7 8 Abstract word count: 150 words 9 Manuscript word count: 3510 words 10 11 Author Affiliations: 12 1. Liling CHAW, PhD 13 PAPRSB Institute of Health Sciences, Universiti Brunei Darussalam, Brunei 14 15 2. Wee Chian KOH, PhD 16 Centre for Strategic and Policy Studies, Brunei Darussalam 17 18 3. Sirajul Adli JAMALUDIN, MSc 19 Environmental Health Division, Ministry of Health, Brunei Darussalam 20 21 4. Lin NAING, MD 22 PAPRSB Institute of Health Sciences, Universiti Brunei Darussalam, Brunei 23 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10.1101/2020.05.04.20090043 doi: medRxiv preprint


67
We report the transmission dynamics of SARS-CoV-2 across different settings from the initial 68 COVID-19 cluster in Brunei, arisen from 19 attendees at the Malaysian Tablighi Jama'at 69 gathering and resulted in 52 locally transmitted cases. Highest non-primary attack rates(ARs) 70 were observed at a subsequent local religious gathering (14.8% [95%CI: 7.1,27.7]) and in the  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi. org/10.1101org/10. /2020  March 11. A rapid response by the global scientific community has described many aspects of 88 the SARS-CoV-2 virus. Its primary mode of transmission is through respiratory droplets, and 89 fomite and aerosol transmission may also play a role (1). Asymptomatic transmission has been 90 observed (2) and the peak infective period appears to be within the first few days of symptom 91 onset (3). Estimates suggest a basic reproduction number (R 0 ) of 2-3 in the early stages of an 92 outbreak (4). While the R 0 is valuable in assessing the spread of the outbreak, it can obscure 93 individual heterogeneity in the level of infectivity, among persons and in different settings (5,6). 94 Modelling of the 2003 SARS outbreak indicates that >70% of transmission occurred due to 95 super-spreading events (SSE) (7). Early reports suggest that similar dynamics may be in play in 96 the explosive propagation of SARS-CoV-2 (8).

98
As part of their mitigation strategies, many countries have adopted stringent yet blunt 99 'lockdowns' of whole cities and provinces (9) and are implementing exit strategies. In moving 100 towards a more targeted approach, countries should be able to answer two major questions Brunei (population 459,500) (10) detected its first COVID-19 case on March 9, arising from an 107 Islamic religious gathering (Tablighi Jama'at) in Kuala Lumpur, Malaysia, a known SSE lasting 108 four days and attended by >16,000 people, including international participants (11). Of the 135 109 confirmed cases in Brunei reported as of first week of April, 71 cases (52.6%) have an  Brunei's Ministry of Health has responsibility for communicable disease surveillance. Since 126 January 23, testing criteria have been implemented for suspected COVID-19 cases. Initially, 127 individuals with acute respiratory symptoms, and a travel history to a high-risk area were tested 128 for SARS-CoV-2. Over the next weeks, the program expanded to include: (i) contacts of a 129 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 10, 2020.    CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10. 1101 NP swabs were collected from all identified participants and tested with RT-PCR. Those who 152 tested positive were admitted to the National Isolation Centre (NIC), while those tested negative 153 were quarantined for 14 days from their return to the country at a designated community 154 quarantine facility where symptom and temperature screening was conducted daily. Those who 155 developed symptoms were re-tested. Activity mapping of confirmed cases was conducted, and 156 contact tracing initiated.  All confirmed cases were treated and isolated at NIC and followed-up until recovery. We 169 obtained clinical information on their history (including any prior presentation to health 170 services), examination, laboratory and radiological results from digital inpatient records on the 171 national health information system database. Additionally, oral history was taken to ascertain 172 whether they had symptoms up to 14 days prior to diagnosis. Cases were discharged following 173 two consecutive negative specimens collected at ≥ 24-hour intervals.

174
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. We categorised cases into two groups: primary (those presumably infected at the Tablighi event 178 in Malaysia), and non-primary (those who did not attend the Tablighi event but had an 179 epidemiological link to the primary cases).  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. For each setting, the attack rate (AR) was calculated by dividing the number of positive contacts 209 by the total number of close contacts (that is, the proportion of contacts that tested positive). To 210 identify risk factors of infection, log-binomial regression analysis was applied to estimate the 211 risk ratio for gender, age, and setting. Further stratification was done to assess differences in 212 infector symptom status across settings. The 95% confidence interval (95% CI) was estimated 213 using the normal-approximation method, or binomial method if the count was less than five.  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020.   n=46), and 5 cases (7.1%) had pre-existing chronic conditions. Compared to non-primary cases, 241 primary cases were significantly older and predominantly male. More than three-quarter of the 242 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. The incubation period was calculated from eight cases that had confirmed epidemiological links Of the 1755 close contacts in the Tablighi cluster, 51 local transmissions were detected, giving 263 an overall non-primary AR of 2.9% (95% CI: 2.2, 3.8). Case 121 (see Figure 2) was excluded 264 from this analysis because he was not detected during contact tracing. Highest AR was observed 265 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ARs also differed by symptom status of the infector (Table 3). ARs in households where the 278 infectors were symptomatic (14.4%) were higher than those who were asymptomatic (4.4%) or 279 presymptomatic (6.1%). AR for the local religious gathering could not be calculated as three 280 primary cases at the event had different symptom status; hence we could not ascertain how 281 transmission occurred. In the household setting, symptomatic cases have 2.66 times higher risk 282 of transmitting to their close contacts, when compared to asymptomatic and presymptomatic 283 (crude risk ratio: 2.66 [95% CI: 1.12, 6.34], Table S1).

285
Table 3 -Attack rates in different settings, stratified by symptom status of the infector 286 287 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10.1101/2020.05.04.20090043 doi: medRxiv preprint The mean observed R was highest in the local religious gathering (2.67), followed by the 288 household setting (0.67 [95% CI: 0.44, 0.96]) ( Table 4). The distribution of the observed R in the 289 household setting was skewed towards zero ( Figure S2). 71.4% of household infections (20 of 28 290 positive contacts) were from 16.7% of cases (7 of 42 possible links).  16,000 participants) may have been infected at that Malaysian event. Moreover, we find that the 310 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10.1101/2020.05.04.20090043 doi: medRxiv preprint highest overall non-primary AR and mean observed R was at the local religious gathering (AR= 311 14.8%, R= 2.67), which was higher than that observed in the household setting (AR= 10.6%, R= 312 0.67). These observations suggest a role for mass gatherings in facilitating SARS-CoV2 313 transmission.

315
The Tablighi is an apolitical Islamic movement with a presence in nearly 200 countries. Tablighi 316 adherents usually travel to gather at annual international events each lasting several days.

317
Communal prayers, meals and speeches form part of these events. In Malaysia, the participants 318 stayed and slept at the mosque, and several participants were deputised to cook meals and clean.  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi. org/10.1101org/10. /2020 To a lesser degree, our observations on the within-household transmission are similar to that 334 observed for the two religious gatherings. Out of 16 household contacts who subsequently 335 became first generation cases, 10 (62.5%) of these were from just three primary cases. As such, 336 even within similar settings, we can expect wide variability in transmission patterns. This 337 observation supports our finding of a moderately high household AR but an observed R of less 338 than one, suggesting that transmission is driven by a relatively small number of cases (5). High Our overall non-primary AR result of 10.6% in the household setting is comparable to other 346 studies that used contract-tracing datasets (16)(17)(18)(19). A study near Wuhan, China (20) reported a 347 higher AR of 16.3%; however, they detected 56.2% of their cases more than five days after 348 symptom onset. By contrast, 77.4% of the cases in our study were detected and isolated within 349 five days of symptom onset, suggesting that early case isolation can reduce AR. The aggressive 350 testing of contacts strategy employed may have contributed to this.

352
We note the low non-primary AR (<1%) and mean observed R (<0.3) for workplace and social 353 settings. While moderate physical distancing was implemented in Brunei following the 354 identification of this cluster, there was no community quarantine or lockdown, public services 355 and businesses remained open, and no internal movement restrictions were imposed.

356
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10. 1101 Combined with our observations on the role of SSE in driving SARS-CoV-2 transmission, we 358 suggest that in areas with limited community transmission full lockdown measures that adopt a 359 blunt approach by restricting all movement can be avoided, in favour of a more targeted 360 approach that includes a combination of case isolation, contact tracing, and moderate levels of 361 physical distancing that take into account the 'red flags' for mass gatherings identified earlier.   CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 While there are case reports of presumptive asymptomatic and presymptomatic transmission (24,379 25), observational studies quantifying such transmission are few. A study from Ningbo, China 380 analysed the overall ARs in symptomatic vs. asymptomatic cases and did not find significant 381 difference between the two groups (26). Another study, re-interpreted the same data and 382 theorized that under certain conditions, symptomatic cases could be more transmissible than 383 asymptomatic ones (27). In fact, our overall crude risk ratio for symptomatic cases showed no 384 significant difference when compared with asymptomatic and/or presymptomatic cases (Table 3,   385   Table S1). However, we suggest that this masks the true picture in transmissibility when 386 different settings are taken into account.

388
In our study, we do not find a significant difference in AR in non-household settings. These 389 settings usually practice some form of non-pharmaceutical interventions (NPI)-individuals with 390 moderate and severe symptoms may be on medical leave, and it is reasonable to expect some 391 physical distancing would be practiced by contacts of persons who display visible symptoms.

392
This is less feasible within the household setting; hence we suggest that transmission occurs 393 more frequently at the household level where control measures are less practical. We observed 394 that the household AR for symptomatic cases (14.4%) is higher than that of asymptomatic 395 (4.1%) or presymptomatic cases (6.1%), suggesting that the presence of symptoms is a host 396 factor in driving transmission.

398
The higher household AR observed among symptomatic cases suggest that testing for such cases 399 should be prioritised, especially in low resource areas with limited testing capacity. Nonetheless, 400 an AR of 4.4% and 6.1% in asymptomatic and presymptomatic cases, respectively, is not 401 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. precautions during an outbreak. Fourth, symptom status of the cases was reported during their 421 swab collection date. We assume this to be reflective of their actual condition when their close 422 contacts were exposed, however, this may not be necessarily true for all cases. Finally, the 423 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10.1101/2020.05.04.20090043 doi: medRxiv preprint generalizability of our results are limited due to no community transmission, a lack of cases in 424 settings such as residential care facilities and dormitories, and small number of cases.

426
The main strength of our study is the availability of a complete contact tracing dataset at the 427 national level. Since all contacts were tested, it is reasonable to assume that this study more 428 accurately detects SARS-CoV-2 transmission than those that only test symptomatic contacts. In 429 conclusion, our analysis highlights the variability of SARS-CoV-2 transmission across different 430 settings and in particular, the role of SSEs. We identify 'red flags' for the development of 431 potential SSEs, and describe environmental, behavioural, and host factors that drive transmission.

432
Overall, we provide evidence that a combination of case isolation, contact tracing, and moderate 433 physical distancing measures is an effective approach to containment.  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10.1101/2020.05.04.20090043 doi: medRxiv preprint   CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 10, 2020. ; https://doi.org/10. 1101