Real-time monitoring the transmission potential of COVID-19 in Singapore, March 2020

Background As of March 31, 2020 the ongoing COVID-19 epidemic that started in China in December 2019 is now generating local transmission around the world. The geographic heterogeneity and associated intervention strategies highlight the need to monitor in real time the transmission potential of COVID-19. Singapore provides a unique case example for monitoring transmission, as there have been multiple disease clusters, yet transmission remains relatively continued. Methods Here we estimate the effective reproduction number, Rt, of COVID-19 in Singapore from the publicly available daily case series of imported and autochthonous cases by date of symptoms onset, after adjusting the local cases for reporting delays as of March 17, 2020. We also derive the reproduction number from the distribution of cluster sizes using a branching process analysis that accounts for truncation of case counts. Results The local incidence curve displays sub-exponential growth dynamics, with the reproduction number following a declining trend and reaching an estimate at 0.7 (95% CI: 0.3, 1.0) during the first transmission wave by February 14, 2020 while the overall R based on the cluster size distribution as of March 17, 2020 was estimated at 0.6 (95% CI: 0.4, 1.02). The overall mean reporting delay was estimated at 6.4 days (95% CI: 5.8, 6.9), but it was shorter among imported cases compared to local cases (mean 4.3 vs. 7.6 days, Wilcoxon test, p<0.001). Conclusion The trajectory of the reproduction number in Singapore underscores the significant effects of successful containment efforts in Singapore, but it also suggests the need to sustain social distancing and active case finding efforts to stomp out all active chains of transmission.

CC-BY-NC 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 (which was not certified by peer review) Singapore, Wizlearn Technologies and the SAFRA Jurong cluster (13,14). In China, substantial 94 hospital-based transmission of SARS-CoV-2 has been reported, with approximately 3400 cases 95 involving healthcare workers (15). This pattern aligns well with past outbreaks of SARS and 96 MERS (16), including substantial nosocomial transmission during the 2003 SARS outbreak in 97 Singapore (17). To minimize the risk of hospital-based transmission of SARS-CoV-2, the Ministry 98 of Health of Singapore has restricted the movement of patients and staff across hospitals (18). 99 Also, because multiple unlinked COVID-19 cases have been reported in the community (19) and 100 the recognition that a substantial proportion of asymptomatic cases may be spreading the virus 101 (20)(21)(22), strict social distancing measures have been put in place including advising the public 102 against large social gatherings in order to mitigate the risk of community transmission (23,24). 103 These social distancing measures reduce the risk of onward transmission not only within 104 Singapore, but also beyond the borders of this highly connected nation (25). A recent influx of 105 imported cases from Asia, Europe and North America into Singapore has triggered travel bans and 106 restrictions for travelers and citizens (9). 107 108 The reproduction number is a key threshold quantity to assess the transmission potential of an 109 emerging disease such as 27). It quantifies the average number of secondary cases 110 generated per case. If the reproduction number is below 1.0, infections occur in isolated clusters 111 as self-limited chains of transmission, and persistence of the disease would require continued 112 undetected importations. On the other hand, reproduction numbers above 1.0 indicate sustained 113 community transmission (16,27). Using epidemiological data and mathematical modeling tools, 114 we are monitoring the effective reproduction number, Rt, of SARS-CoV-2 transmission in 115 Singapore in real-time, and here we report the evolution of Rt by March 17, 2020. Specifically, we 116 characterize the growth profile and the effective reproduction number during the first transmission 117 wave from the daily case series of imported and autochthonous cases by date of symptoms onset 118 after adjusting for reporting delays, and we also derive an estimate of the overall reproduction 119 number based on the characteristics of the clusters of COVID-19 in Singapore. . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101 We obtained the daily series of 247 confirmed COVID-19 cases in Singapore between January 23-125 March 17, 2020 from public records of the Ministry of Health, Singapore as of March 17, 2020 126 (9). Individual-level case details including the dates of symptoms onset, the date of reporting, and 127 whether the case is autochthonous (local transmission) or imported are publicly available. Clusters 128 consisting of two or more cases according to the infection source were also assembled from case 129 descriptions obtained from field investigations conducted by the Ministry of Health, Singapore 130 (9). Single imported cases are analyzed as clusters of size 1 whereas unlinked cases were excluded 131 from the cluster analysis. 132 133

Transmission clusters 134
As of March 17, 2020, 18 different clusters of COVID-19 cases with 2-48 cases per cluster have 135 been reported in Singapore. A schematic diagram and characteristics of the COVID-19 clusters in 136 Singapore are given in Figure 1 and Table 1. The geographic location of the six clusters accounting 137 for 45.3% of the total cases is shown in Figure 2 whereas the corresponding distribution of cluster 138 sizes is shown in Figure 3. 139 140

Yong Thai Hang cluster 141
This cluster with 9 cases was the first to be reported in Singapore. It has nine traceable links, 142 including eight Chinese and one Indonesian national associated with the visit of Chinese tourists 143 to the Yong Thai Hang health products store, a shop that primarily serves the Chinese population, 144 on January 23, 2020. Four shop employees and the tour guide were first identified as a cluster on 145 February 4, 2020 (12, 28, 29 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 April 17, 2020. . https://doi.org/10.1101  . CC-BY-NC 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 April 17, 2020. 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 April 17, 2020. . https://doi.org/10.1101 The first case of this cluster was identified on March 11, 2020. This cluster is composed of 3 local 216 As an outbreak progresses in real time, epidemiological curves can be distorted by reporting delays 240 arising from several factors that include: (i) delays in case detection during field investigations, 241 (ii) delays in symptom onset after infection, (iii) delays in seeking medical care, (iv) delays in 242 diagnostics and (v) delays in processing data in surveillance systems (33). However, it is possible 243 to generate reporting-delay adjusted incidence curves using standard statistical methods (34). 244 Briefly, the reporting delay for a case is defined as the time lag in days between the date of onset 245 and date of reporting. Here we adjusted the COVID-19 epidemic curve of local cases by reporting 246 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101/2020.02.21.20026435 doi: medRxiv preprint delays using a non-parametric method that employs survival analysis known as the Actuaries 247 method for use with right truncated data, employing reverse time hazards to adjust for reporting 248 delays as described in previous publication (35-37). The 95% prediction limits are derived 249 according to Lawless et al. (38). For this analysis, we exclude 7 imported cases and 5 local cases 250 for which dates of symptoms onset are unavailable. 251 252

Effective reproduction number from case incidence 253
We assess the effective reproduction number over the course of the outbreak, Rt, which quantifies 254 the temporal variation in the average number of secondary cases generated per case during the 255 course of an outbreak after considering multiple factors including behavior changes, cultural 256 factors, and the implementation of public health measures (16,27,39) . Estimates of Rt>1 indicate 257 sustained transmission; whereas, Rt <1 implies that the outbreak is slowing down and the incidence 258 trend is declining. Hence, maintaining Rt <1 is required to bring an outbreak under control. Using 259 the reporting delay adjusted incidence curve, we estimate the most recent estimate of Rt for 260 COVID-19 in Singapore by characterizing the early transmission phase using a phenomenological 261 growth model as described in previous publications (40-42). Specifically, we first characterize 262 daily incidence of local cases for the first transmission wave (January 21-February 14, 2020) using 263 the generalized logistic growth model (GLM) after adjusting for imported cases. This model 264 characterizes the growth profile via three parameters: the growth rate (r), the scaling of the growth 265 parameter (p) and the final epidemic size (K). The GLM can reproduce a range of early growth 266 dynamics, including constant growth (p=0), sub-exponential or polynomial growth (0<p<1), and 267 exponential growth (p=1) (43). We denote the local incidence at calendar time " by " , the raw 268 incidence of imported cases at calendar time " by " , and the discretized probability distribution 269 of the generation interval by " . The generation interval is assumed to follow a gamma distribution

274
In this equation the numerator represents the new cases " , and the denominator represents the total 275 number of cases that contribute to the new cases " at time " . Parameter 0≤ ≤ 1 represents the 276 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101/2020.02.21.20026435 doi: medRxiv preprint relative contribution of imported cases to the secondary disease transmission. We perform a 277 sensitivity analyses by setting = 0.15 and = 1.0 (46). Next, in order to derive the uncertainty 278 bounds around the curve of ( directly from the uncertainty associated with the parameters 279 estimates (r, p, K), we estimate ( for 300 simulated curves assuming a Poisson error structure 280 (47).  (51)). This permits direct inference of the maximum likelihood 296 estimate and confidence interval for R and k. In this manuscript, we modify the calculation of the 297 likelihood of a cluster size to account for the possibility that truncation of case counts at a specific 298 time point (i.e. March 17, 2020) may result in some infections being unobserved. This is 299 accomplished by denoting x as the sum of the observed number of serial intervals in a cluster. Then 300 the likelihood that an observed cluster of size j containing m imported cases is generated by x 301 infectious intervals is given by: 302 . D→(.-;) > ?@@ , C (1) 304 Where the likelihood of a i infections causing j infections is given by: 305 306 . CC-BY-NC 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 April 17, 2020. (2) 307 where Γ is the gamma function. 308

309
To determine the number of observed serial intervals we observe in each cluster, we first estimate 310 the cumulative probability distribution of the serial interval. We assume the serial interval is a 311 gamma distribution, with a mean of 4.7 days and a standard deviation of 2.9 days (44). This 312 translates to a shape parameter of 2.63 and a scale parameter of 1.79. We then use the difference 313 between the onset data and the end of our study (March 17, 2020) to determine how much of the 314 infectious period was observed. For cases that only have a report date, but no onset date, we assume 315 an onset date that is six days earlier than the reporting date. This is based on the average duration 316 between onset date and report date that was observed in the data. When applied to the case series, 317 we are able to assign a total size, the number of imported cases and the observed number of 318 infectious periods for each cluster in the case series. When no imported cases are known to be in 319 a cluster, we assign the number of imported cases to be one as the cluster must have been initiated 320 by someone (e.g. the index case had contact with a foreign visitor). 321 322 When equation 1 is applied to the table of cluster size characteristics, the likelihood of the data can 323 be calculated as a function of R and k. Minimizing the likelihood produces the maximum 324 likelihood estimates of R and k. Applying the likelihood ratio test by profiling and R and k, 325 produces confidence intervals (52). Code was run in R version 3.6.1. 326 327

Incidence data and reporting delays 330
The COVID-19 epidemic curve by the date of reporting, stratified for local and imported incidence 331 case counts is shown in Figure 4. It shows that the majority of the imported cases are concentrated 332 at the beginning of the outbreak (January 23, 2020 -February 3, 2020) and after March 10, 2020 333 in Singapore, with an average of ~12 new cases reported per day between March 1, 2020 and 334 March 17, 2020 (Figure 4). Out of 88 imported cases, only 14 cases have been linked to secondary 335 . CC-BY-NC 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 April 17, 2020. The reporting-delay adjusted epidemic curve of local cases by date of symptoms onset roughly 339 displays two small waves of transmission reflecting the occurrence of asynchronous case clusters 340 ( Figure 5). Moreover, the gamma distribution provided a reasonable fit to the distribution of 341 reporting delays for all cases, with a mean reporting delay at 6.4 days (95% CI: 5.8, 6.9) (Figure  342 6).We also found that imported cases tend to have shorter reporting delays compared to local cases 343 (mean 4.3 vs. 7.6 days, Wilcoxon test, p<0.001), as imported cases tend to be identified more 344 quickly. The mean of reporting delays for the six large clusters ranged from 4.8-13.6 days (Figure 345 . CC-BY-NC 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 April 17, 2020.  (Table 2). However, some differences in the reproduction 389 numbers reported for the epidemic in China may result from different methods, differences in data 390 sources, and time periods used to estimate the reproduction number. Similarly, a recent study has 391 shown an average reporting delay of 6.1 days in China (67) which agrees with our mean estimate 392 for cases in Singapore (6.4 days). Moreover, the scaling parameter for growth rate (p) indicates a 393 sub-exponential growth pattern in Singapore, reflecting the effective isolation and control 394 strategies in the region. This is consistent with a sub-exponential growth pattern for Chinese 395 . CC-BY-NC 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 April 17, 2020. identifiable clusters in many parts of the world (2, 74-77). Moreover, Singapore has also produced 420 secondary chains of disease transmission beyond its borders (25). Although Singapore has been 421 detecting and isolating cases with diligence, our findings underscore the need for continued and 422 sustained containment efforts to prevent large-scale community transmission including 423 nosocomial outbreaks. Overall, the current situation in Singapore highlights the need to investigate 424 the imported, unlinked and asymptomatic cases that could be a potential source of secondary cases 425 and amplified transmission in confined settings. Although Singapore has a world-class health 426 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101 system including a highly efficient contact tracing mechanism in place that has prevented to 427 outbreak from getting out of control (25, 78), continued epidemiological investigations and active 428 case finding efforts are needed to contain the outbreak. 429

430
Our study is not exempt from limitations. First, the outbreak is still ongoing and we continue to 431 monitor the transmission potential of COVID-19 in Singapore. Second, onset dates are missing for 432 twelve cases, which were excluded from our analyses. Third, we cannot rule out that additional 433 cases will be added to existing clusters, which may lead to underestimating the reproduction 434 number based on the cluster size distribution. Fourth, some of the cases are associated with 435 generating secondary chains in more than one cluster, which were included in the most relevant CC-BY-NC 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 April 17, 2020. CC-BY-NC 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 April 17, 2020. 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 April 17, 2020. . https://doi.org/10.1101 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101  March 9, 2020 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101  CC-BY-NC 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 April 17, 2020.  CC-BY-NC 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 April 17, 2020. March 17, 2020 . The pink circles represent the cases linked to Wuhan, the green circles represent 722 the non-Wuhan related case importations and the blue circles represent cases with no travel history 723 to China. The larger dotted circles represent the COVID-19 disease clusters. Each blue arrow 724 . CC-BY-NC 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 April 17, 2020. CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101  CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101  CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101/2020.02.21.20026435 doi: medRxiv preprint 754 Figure 5: Reporting delay adjusted local incidence for the COVID-19 outbreak in Singapore as of 755 March 17, 2020. Blue bars represent the raw incidence, red solid line represents the adjusted 756 incidence, red dotted lines represent the 95% lower and upper bound of the adjusted incidence. 757 758 759 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101/2020.02.21.20026435 doi: medRxiv preprint CC-BY-NC 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 April 17, 2020. The effective reproduction number followed a declining trend with the latest estimate at 0.7 (95% 801 CI: 0.3,1.0) by February 14, 2020. 802 . CC-BY-NC 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 April 17, 2020. . https://doi.org/10.1101