Covid-19 and Singapore in April

(Originally on my Medium account, shifting here for better filing with some edits)

As I’m writing this, we’re in the middle of a mostly-but-not-quite-lockdown here in Singapore. We’re not afforded the liberty of eating at a diner, even alone but we can exercise – alone – in uncrowded areas.

I’ve been talking to friends the last days – since the lockdown Circuit Breaker started- and I’m convinced that our statistics are not necessarily dire. As a country, we do have a lot of transparency with data which means lots of numbers to look at.

So, I’d like to explore the information that’s publicly available. I’ll try to make sense of the situation now and conjecture how we ended up here.

My sources are as follows:

3 bits of information

  • Symptom onset date is different from press release date. I would speculate that the higher sharper peaks on the PR numbers are due to things like labs/reporters being slowed down for non-work days. You can also see a rough trough separation roughly every 5–6 days which matches a work week. It’s also likely that test results are released in batches for whatever reason so you could have your positive results come in the same day as someone else who started their testing a few days after.
Image from local situation report 12 Apr 2020, MOH
  • There is a buffer between actual infection and a positive result that leads to isolation (at least for symptomatic people). This includes some or all of time taken to see a doctor, getting your medical leave days, revisiting for a test and actually testing positive. Before 26 March, the 14-day moving average was at about 8–9 days. There’s a sharp drop thereafter to roughly 5–6 days. My takeaway is that the bar chart doesn’t start bulking up til roughly the 14-day period ending with the 26th so a 5–6 day onset-to-positive result time frame might be most reliable for estimation and might be unlikely to go down significantly further.
Image from local situation report 12 Apr 2020, MOH
  • Also, incubation period observed in travelers from Wuhan = 6.4 days

Ok, with that, on to my thoughts

We are at Wave 3

While the media and official statements talk about us being in Wave 2 of the pandemic, I believe we are at Wave 3.

Wave 1 was at the very start when we started getting travelers from China and regional affected territories. We got this under control before the chaos that subsequently consumed Europe/USA.

Wave 2 started when we started receiving our overseas-based locals.

Wave 3 just started when our foreign worker population — most of whom were in mass housing arrangements — started coming down with the bug.

The data I have starts from roughly before the onslaught of Wave 2 when Wave 1 has more or less tapered off.

Here’s What Wave 2 Looks Like

Let’s look at a portion of Wave 2. From 17 Mar — 4 Apr, imported cases formed > 50% of our total cases. This wave lasted roughly 20 days. My logic here is that I’m going to use our domestic cases as a baseline here to judge for ‘normalcy’. When a cause of a wave starts consistently going <50%, it stops being the major cause of that wave’s existence, so to better allocate resources, priority should be shifted to the new major cause.

In our case, since we’re dealing with a small population, rather than radically-increasing numbers, the decreasing percentage is due to our infection ratio decreasing so I’ll be assuming that the shift from a significant to non-significant cause is the best case scenario of when we catch on to a trend, stop it before it spreads and aggressively control it.

15 Mar: We recalled students / 29 Mar: All travellers, even locals had to get permission before returning
  • Assuming the bulk of returnees will scramble for a flight in the first third of this time period, we’d have most of these travelers back by 20 March.
  • Anyone who is symptomatic will be tested on return. Results of that test is likely to take 3–5 days and give or take a few days (say 3) for PR to kick in (total 6–8 days) so there’s expected to be a steady increase of positives from 18–28 March. Looks more or less like the data above.
  • Now the more interesting bit here. For any returnees without symptoms, if they return before 21 March, there is NO Stay-Home-Notice (That’s the bulk of our returnees!). From 25 March, Stay-Home-Notice is served in hotels for ALL returnees. I’d explore the implications of this later on.
  • 6.4 + 6 (onset to PR) ~ 12/13 days for most of the returnees who were infected while abroad to have developed symptoms, get tested and to make it to press release. This would mark the end of the wave which would be on 10/11 April. We’re seeing 0 counts of this as of 12 April so this estimation holds, even if it’s on a very safe side.

And Now, Wave 3

Here is where we are now at. The current situation is where we’re starting to wake up to the horrors that our foreign worker population — those guys who are largely blue-collared workers — are now falling sick.

The implications of this is many-fold, the most important of which is how they’re the main means for our waste disposal and hygiene upkeep. This is troublesome because if they’re barred from working, our disease-spreading rubbish and litter will never be ridden of.

Whether by coincidence or by design (I would like to believe the latter), the fact that we’re all in a lockdown means there are less public areas that needs cleansing so our reliance on aforementioned labor is vastly reduced. Because of our shutdowns as well, these same people are allowed a legitimate reason to stay quarantined regardless of what their slave-driver bosses think.

Anyway, here’s where we are now. This graph considers any infected patients and cumulative total as of 30 Mar, where we started seeing a sizable uptick (and data separation) of foreign workers.:

  • From about 8 April, infected Foreign Workers form the bulk of our new cumulative total. They probably got infected from around 26 March before spreading it among themselves (12 days buffer, roughly). Sounds kind of like the time frame where our returnees are making their exodus back yea?
  • We started our movement to separate and quarantine around 5 Apr. As of last weekend (roughly 11 Apr), there have been press releases announcing healthy workers have been moved to makeshift locations. I would assume that this is the case for a vast majority of our workers. Given the aforementioned buffer times, it would be expected that anyone who’s sick and put in quarantine during this period of time would add to the count from 11–19 Apr. We will continue to see a rise til then.
  • For those who are in ‘safe’ locations, there is a likelihood that some will be false negatives. These cases will also start to surface around that time but are likely not as large in numbers as those identified in the point above. I believe these numbers will be more controllable.
  • Given our response to travelers and if I had to estimate a peak from there, it’s likely that we’ll see a real decrease in proportion of sick foreign workers from around 28 April (ie. ~ 20 days) with the peak being roughly around 18–19 Apr. It seems to add up so far so I’m quite convinced the estimation wouldn’t be too far off.

How Did We Get to Wave 3

So here’s the interesting question for me. I’d like to look into a few graphs to find some support for the data.

This is a chart we’ve been seeing quite a bit of which shows how we’re trending upwards. Before 30 Mar, the statistics broadcast had no differentiation between foreign workers and Singapore Citizens and we’re all broadly classified as Domestic cases. Not because of xenophobia but because, as explained above, Wave 2 was all about imported cases vs locally-transmitted cases and that’s where data made most sense.

Anyway, I feel this chart doesn’t give an accurate depiction of what is really going on with Wave 3. What we need to be seeing is this:

Daily new cases. After 30 Mar, data excludes tally of Foreign Workers

Ok here’s the fun part. There’s a peak just before 30 March. We have 2 suspect problems here:

  1. Returnees from abroad from 15–21 Mar who do not have Stay Home Notices. They’re allowed to roam and if they develop symptoms, it would likely fall in the 16–27 Mar. HOWEVER, if they do fall sick, they would be classified as an imported case which won’t fall in these statistics. What WILL be captured here are people who’d get infected by these imported cases. Assuming we’re isolating these sick people within 5 days (as mentioned above, Figure 1.4), these secondary infections will take place in roughly 12 days (5 + 6.4 for incubation) which will mean anything from 23Mar–8 April, give or take. We will see them in the PR from 26 Mar–11 Apr.
  2. On 30 Mar when we started separating our foreign workers from Domestic count, we were already on a rising trend. If you look deeper into the graphs on MOH’s site, there had started to be a noticeable growth in numbers from as early as the 25 Mar. That uptrend in my graph above isn’t about more locals getting infected but simply that our workers have shown their vulnerabilities.

Since the rough start of the lockdown for our foreign workers only began on 5 Apr, anyone who they could have infected would start factoring in our count from 6.4+3+5 = 14.4 days. Anyone infected will register sick from 6–20 Apr with the last of the sick coming in roughly at the tail end of 20 Apr.


Like I’m said at the start, I’m writing this and exploring more in depth as I go along so it’s not like I had an agenda planned before. I guess the TL;DR version is really these —

  • I think we screwed up by not implementing any sort of forced isolation measures on people returning from abroad in March.
  • I think we also screwed up in not being better for the people amongst us who cannot afford the luxury of social spacing or washing our hands regularly. I read somewhere an argument that being able to practice social distancing comes from a place of privilege. I agree.
  • If we aren’t getting ourselves infected through whatever methods (ie. licking toilet seats, hugging strangers and their cats, blowing bubbles into our neighbors’ homes), we’ll probably be seeing numbers for local transmission taper towards the end of this week.
  • I think we should expect really high numbers these few days since our sick foreign workers (and their bosses) are finally getting enough scrutiny to have to be tested.
  • If foreign workers continue to form the bulk of our new cases, we can expect this to last til roughly the end of April.
  • On the bright side, since we’re all stuck at home, this hugely lessens the reliance on cleaners, sweepers and other blue-collar workers. It serves to both reduce their risk of exposure to our dirty ways and also reduce the manpower strain.
  • I’d like to think we’ll not see a lockdown extension.

(Update: We did. We eventually went to a more relaxed state on 1 Jun)

  • Lastly, I’d really like to find some places with good data to run some hypothesis testing on whether asymptomatic people are a serious threat and also the true value of masking up.

Please feel free to discuss the above and to correct any mistakes on my part! 🙂