TWC2 volunteers monitor the media and government statements to stay abreast of developments regarding Covid-19 in SIngapore, particularly the situation among migrant workers residing in dormitories. Unless the Covid situation among migrant workers deteriorates again, this is likely to be the last of the monthly summaries we started in August.

If readers spot any errors or can help fill any gaps in the picture, please drop a note to <> with the header “Covid-19 summary, attention Kellock”.

Quick Summary

  • Barring some major unexpected setback, the end of the COVID crisis in the dorms seems now to be in its final phase, with new cases now averaging less than one per day.
  • Since the outbreak of COVID in the dorms, about 17% of all dorm residents have been infected.
  • Analyzing patterns of infection since the start of the crisis, small dorms have generally fared better than larger dorms. The smallest dorms, those with capacity less than under 25, have as a group experienced a dramatically lower infection rate, less than a tenth of the average across all sizes. The very largest dorms also seem to have fared somewhat better than the average, while midsize dorms (500 to 5000 capacity) have suffered the highest infection rates.

Key numbers

Data here is derived from Ministry of Health (MOH) figures, primarily their Daily Reports:

Cumulative cases (since start of the pandemic)

  • As of 31 October, a total of 54,497 dorm residents have been declared infected, a 0.1% increase in total cases during the month of October.
  • The 54,497 cases represent about 17% of the total dorm population of 323,000. Despite some predictions early in the epidemic that the dorm outbreak might only halt when infection levels reached “herd immunity”, it has been halted (at least for now) well short of the 60% to 70% required for typical herd immunity.

New cases

  • In the week to 31 October, 4 new dorm cases were declared, a 95% drop compared to the 83 dorm cases in the final week of September and averaging less than one per day.
  • For October as a whole, a total of 48 new cases were detected in dorms, an average of about 1.5 new cases per day. The following graph shows total new cases in dorms per month from July to October.

Deaths and critical illness

On 12 October Singapore announced the first COVID fatality in almost 3 months, a 64-year-old permanent resident (not a migrant worker) who had been in intensive care. There have been no other COVID fatalities or ICU cases in the month of October.

Hospitalised and in care facilities

As of 31 October there were 74 people in hospitals and Community Care Facilities (CCFs), about 70% down compared to the figure of 250 at the end of September and about 99% down compared to the figure of 5,687 at the end of July. (These are total numbers in Hospitals plus CCFs and they are across the total population, not just migrant workers.)

Dorm infection patterns

With new COVID infections in the dorms now reduced to less than one per day, we have been looking back through the data we collected over the last few months, analysing the pattern of infections across dorm types, – Purpose-built (PBDs) and Factory-converted (FCDs) – and sizes. The following chart shows cumulative infections (total since the start of the outbreak) as a percentage of the total approved capacity of dorms in the size category (capacity including infected and non-infected dorms in the category).

Two things can be seen from this chart:

  • The smallest dorms – those with under 25 men – have fared best by far, with an infection rate of about 1.3%, less than a tenth of the average across all sizes [footnote 1]. This is primarily because the majority of the smallest dorms escaped infection completely, an outcome which makes intuitive sense: the probability that at least one worker will introduce an infection to his dorm is proportional (albeit not linearly) to the number of workers in the dorm. And when one case arrives in a dorm, it is likely to spread rapidly to others in the dorm. Most of the smallest dorms completely avoided COVID infection.
  • There is a “bump in the middle” in the capacity range of about 500 to 5000: the worst infected categories were a) the largest FCDs which house 500 to 999 men, and b) the smallest of the PBDs which house 1000 to 4999 men. The very largest dorms, especially the handful which house more than 10,000 men, fared better than those in the middle. Why this should be the case is unclear, but one possible reason is that the largest dorms have more effective partitioning of workers by block, section or floor, thus in a certain sense being more akin to multiple smaller dorms.

Dorm plans for the future should consider the above carefully: in an epidemic there’s an inherent advantage to having many small dorms instead of a few large ones. However if large dorms are deemed necessary it’s encouraging to hear that plans are afoot to partition workers better within them: for example recent news reports say that new Quick-Build Dorms (QBDs) will house just 5 workers per room with its own ensuite bathroom instead of today’s huge shared bathrooms, and will have kitchen/dining areas for 30 to 40 men instead of large communal dining halls [footnote 2].  Not only do workers prefer these more “home-like” arrangements, but this data for COVID infections strongly suggests they will slow the spread of any future viral infection in the dorms.

An article in our recent newsletter presents additional statistics about the infections by dorm type and size [footnote 3].


Since early August, all dorm residents have undergone Rostered Routine Testing (RRT), an initiative in which they (and some other workers) are being tested every two weeks [footnote 4].  In our August & September reports we mentioned that we didn’t understand how the daily testing numbers required by the RRT regime (about 28k per day in order to test all 387k workers for whom RRT is required) tallied with cumulative testing numbers released by MOH approximately weekly [footnote 5].  A reader wrote in to us pointing out that this was probably due to a  misinterpretation of the limited information MOH provides in the weekly releases. Using the data for “Total Swabs Tested” instead of “Total Unique Persons Swabbed”, the figures match up well. For example for the 7-day period from 20 to 26 October, the weekly MOH release implies that about 27.7k tests were carried out per day, very close to the 28k per day required by RRT. Assuming this new interpretation is correct, it also implies that RRT comprises the vast majority of swab tests – which of course makes sense given that about 94% of all COVID cases in Singapore have been among the migrant workers.

Another deduction from the weekly MOH figures is that the average number of tests per tested person has risen steadily since early May. Figures released back in early May give a ratio of about 1.3 tests per person from the start of testing up to that time. The equivalent figure for 26 October is about 3.4, more than 2.5 times greater. This fits with the move to RRT: most tests now are of people who have been tested previously.

Additional information

Wikipedia has a detailed timeline of the COVID pandemic in Singapore, with many dorm-specific details at

1. The overall infection rate across all FCDs and PBDs as a function of approved capacity is about 14.7%. This is slightly lower than the 17% infection rate as a function of dorm population, primarily because some dorms are not at full capacity, ie the total population is somewhat lower than the total approved capacity. There are some other factors which may contribute to the difference between these figures (for example construction site dorms are not included in our capacity figures), but these differences are insignificant in arriving at the conclusions above. For example, whether one takes 17% or 14.7% as the average, the figure of 1.3% for the smallest (under 25) category is less than a tenth of the average.


3. Research Spotlight section of TWC2 newsletter, Sep-Oct 2020.