Covid 19 update: Hong Kong is the Snuff-out champion, but will it have to suffocate itself to hold on to the title?
This is the second in our series of articles on Hong Kong. Here we aim to show you how our modelling of the COVID 19 outbreak provokes critical questions and informs answers to how Hong Kong can navigate an exit from a “walled city trap”.
Hong Kong has set an example for other cities and regions for containing its outbreak but at the cost of aggressively locking down and leaving almost everyone in the territory susceptible to infection by COVID-19. Without extraordinary vigilance a new outbreak could start and grow very rapidly.
After a recap of the history there is a review of the insights from setting up the model and some scenarios that show the consequences of certain actions the people of Hong Kong could decide to implement. In later articles we will expand on this.
Hong Kong is not alone in what must be a frustrating situation. Many other cities around the world have adopted Hong Kong’s approach. We hope that, after reading this, the reader will share ideas for how Hong Kong could escape. The obvious one is, of course, a vaccine … but we can’t help with that and it could take some time. We have been working to identify other ideas to help Hong Kong as soon as possible and we hope this will help other locations facing a similar dilemma, possibly including: Greece, Ghana, Singapore, Los Angeles, Australia and New Zealand, and every city or region that has succeeded in containing COVID-19 at the risk of stifling their livelihood.
History of the Outbreak and implications for setting up the simulation
The first reported case in Hong Kong was on January 23, subsequent to the outbreak in Wuhan in November 2019. Despite that the HKSAR government declared a “serious response level” to the outbreak on January 8, when Hong Kong announced 8 suspected cases, it did not impose the most stringent precautionary measures in January — based on the advice from the Wuhan authority that there was “no obvious evidence” of human-to-human transmission of the then unidentified virus.
Nevertheless, the government implemented measures requiring face masks at hospitals and tightened medical screening at airports and train stations with connections to Wuhan, but these eventually were inadequate to prevent the spread of the virus to Hong Kong.
Ever since the first case, many of the cases were “imported” either by infected people crossing the border from China (mostly with developed symptoms) and/or from passengers and crew on flights to Hong Kong from overseas. This resulted in a steady stream of cases coming in over a relatively extended period without very many new domestic transmissions within Hong Kong. For example, on January 23, when the first cases were recorded: two men travelling from Mainland China were confirmed by tests as positive for the virus. It was another two weeks before a significant jump occurred on 9th February and then another four weeks with again a steady stream of about 3 new cases per day. (Note : The simulation parameters reflect the Worldometer data and Wikipedia data).
Snuffing out COVID 19 — Hong Kong Style
Hong Kong is one of the most densely populated cities in the world (6,659 people per sq km vs an average of 5,590 people in London), and Hong Kong has a relatively older population with an average age of 43.8 years. These factors make Hong Kong people more likely to transmit the virus and more likely to die from the infection.
These factors impacted our thinking when we set up the simulation. Once the new cases started to grow the pace accelerated very rapidly: faster than in any other case. To illustrate, the implied normal contacts per day from the initial growth rate of infections is 12 versus 5 in Lombardy and 5 in Jakarta.
At the same time, Hong Kong is in close proximity to China and there were the painful memories of the SARS outbreak in 2003. By the end of January Hong Kong citizens began implementing Hygiene Measures by wearing face masks in public and practising regular sanitisation of hands and surfaces (Exhibit 1). In parallel a voluntary effort was increasingly underway to avoid contacts using social distancing especially during the Chinese New Year which started on January 25.
We set up the simulation to match the data on new cases. This revealed an insight: hygiene measures implied a reduced probability of virus transmission through contacts by two-thirds (note1, Exhibit 1); and the average contacts/person per day dropped from 12 to less than 7 by Day 50 (23 March) (note 2, Exhibit 2).
Simulation and Actual are Synchronised
However, in early March, new cases per day started to grow much faster and the government decided on 28th March to implement much more stringent guidelines (See label 3 on Exhibit 2). To replicate the actual number of cases in this phase, the probability of infectious contacts of 4% (as a consequence of hygiene) was used in the simulation whereas the potentially infectious encounters every day (as a consequence of reduced interactions) was reduced to the absolute minimum after the imposition of the latest social distancing regulations. As such, the number of new confirmed cases in past few days has approached zero.
The model was formally benchmarked against the Imperial College Report — the UK and US Governments responded to this report by implementing lockdowns. We used this report and reference sources in the report and other sources published since to adapt to the situation in Hong Kong. This setup has been arrived at because the steady stream of early infections resulted in deaths in Hong Kong from something like a two-phase outbreak. The deaths are linked to the steady stream of early infections arriving from elsewhere, and no other cause of deaths have been identified to date. In that way, the first deaths are disconnected from the exponential ramp up that was choked off by voluntary and enforced social distancing. We have not altered the expected death rates in Hong Kong from the Imperial College references which in turn come from data sourced in Wuhan China.
Two other parameter choices are needed for the simulation to synchronise with the data. The fraction of infections that are reported as a case is 56%. This is similar to the levels implied in Athens, 65%, where aggressive measures were taken to track and trace infections. By contrast in London where testing is not widely available the implied reported cases are just 5% of infections. Finally, in Hong Kong, the recovery rates have been slowed to match the actual by reducing the percentage of people who recover each day from recuperating in hospital to 8% of patients in the wards and ICU. That translates to an average stay in hospital of 12.5 days after admission.
Together these parameter changes result in a rather different picture than the last version of this case study:
A few observations from the dashboard (Exhibit 3 above):
- The simulation closely follows the available historical data for new cases, cumulative and active.
- As stated above, the deaths relate to the first of two phases. So far no deaths have occurred from the second exponential growth and choking off of new cases has been made possible from the extreme measures on social distancing. However, recoveries and resistant do not match the data, because social distancing does not kill the virus and thus up to 85% of infected never know they have had the virus. The model is simulating and adding the asymptomatic infections to the recorded recoveries
- The actions taken to limit social distancing have at first gradually impacted infections and the result is a sudden significant reduction in new cases after April 9 (with new confirmed cases of 13 versus 25 on the previous day).
- By contrast to our first publication of this case where we were projecting a rebound in new cases, the numbers have continued to drop. Now it appears the social distancing measures have almost entirely choked off any new infections.
- Hong Kong is the most extreme of all the cases implementing social distancing. But as in the first publication there is consequence: only a tiny fraction of the population have acquired resistance. Close to 99% of the population remain Susceptible.
This result shaken us — it is much more extreme that we were able to discern just two weeks ago and it is an insight deriving this parameterisation from the data.
However, it does not result in a blinding insight about what to do next in Hong Kong. The territory is now effectively a walled city living under strict lockdown and the authors are aware of the concerns about a second wave of infections from repatriated citizens from returning students and stranded citizen from overseas. One of our authors, Hayden Cheng is part of this group who have returned since mid-March (we are happy to report that he is healthy and free of infection).
What Can (or Should) Hong Kong Citizens Expect To Happen ?
The status quo may not be unsustainable as the citizens of Hong Kong are experiencing increasingly psychological and economic pressure, whereas others (though a minority) are becoming defiant towards the more stringent social distancing measures. At the same time, we are expecting several testing moments that will cause challenges to the government and overall population of Hong Kong, such as :
- the Hong Kong Diploma of Secondary Education Examination (HKDSE) currently scheduled for April 24 -25, 2020, whereby approximately 53,00 students and teachers will be participating in across Hong Kong;
- the 5-Day May holiday, whereby approximately 840,000 mainlanders crossed border to Hong Kong in the first 3 days last year (2019) and generated sales of over USD 5.1 Bn in May last year;
- looking further ahead, there will be the celebration (and associated civil activities) on and about the same time as the anniversary of the Hong Kong Handover on the July 1;
In order to enable activities to take place and achieve the maximum benefits, Hong Kong will have to significantly lift the social distancing limitations. Meanwhile, by testing the impacts by assuming different levels of social distancing, it is found that, in order that the current infectious rate be capped and Hong Kong continues to be a relative “safe haven” against the virus, Hong Kong will have to continue with its stringent “social distancing” and “travel ban” limitations.
Meanwhile, to gain further insights as whether the limitations can be relaxed based on certain criteria, we have created the following scenario by assuming that contact rate will be increased on 1th May (day 100) as illustrated in Exhibit 4 below:
(a). [orange line] Strict Social Distancing measures (which were implemented on 28th March) will be lifted on May 1. Therefore, contact rate will be relaxed to the level pre-enforcement whilst assuming that the hygienic discipline (wearing face masks, regular hand wash etc) continue amongst the population of Hong Kong. In this case, the simulation projects that the number of active cases will exponentially increase from late June to a maximum of 350,000 in mid-September;
(b). [grey line] all the Social Distancing measures will be totally-relaxed on May 1 (that is, back to pre-Covid-19 normality) from May 1, provided that demand on hospital isolation beds to cater for the consequential Covid-19 infections does not exceed supply. Once the demand cannot meet supply of isolation beds, the Strict Social Distancing measures will be immediately re-imposed. In this case, it expected that (according to the simulation) the Strict Social Distancing measures will be necessary only 1.5 months after the total relaxation. Therefore, taking into account a margin for error, the period of returning to “normality” will be very short to be logical for both the government and/or the citizens) to take the risk of suffering from the potential negative effects. Nevertheless, the total number of deaths from this scenario will be lowest of the three, but only 10% less than the others.
(c). [ red line] in this case, the Strict Social Distancing measures will be gradually relaxed from May 1 until 100 days later (mid-September). It can be seen from the charts that the impact of this scenario is a mere delay of Scenario (a) [orange line] with the same projected number of deaths and confirmed cases.
Let’s Take Stock of What To Do :
So, based on the above analysis, what is next for Hong Kong? The territory has bought some time, and it has been able to claim fame for being the city that has effectively snuffed the spread of the Covid-19 virus. But then, the psychological and sociological pressure, not to mention that economic scars inflicted on the population of Hong Kong, could caused the wall of defence to crack (and thereby leaks). On top of that, there are external forces which can exaggerate the deteriorations.
The results show a policy to ‘hold fast’ until medical breakthroughs on treatment or a vaccine are discovered seem to be the best option for now. This is consistent with the government of Hong Kong announcement of an extension of the Strict Social Distancing measures until May 7th. However, signs we see in the streets of Hong Kong today that vigilance is still needed. It is not our place to judge or prescribe. But the following remedies could be the means to enable the eventual “release” of the citizens of Hong Kong and return to normality:
This concludes the second update of the Hong Kong case at this time. We will be updating it and refining it over the coming weeks as events evolve.
[i] Alvin Cheng has been the CEO & Executive Director of several listed companies listed on the SGX. He was also previously regional head of GE Container Leasing and an Investment Banker with Chase (now JPMorgan) in the late 90s. He is currently in lockdown in Hong Kong and maintains constant contact with family members in Shanghai who have experienced the very stringent government actions to contain COVID 19. Alvin is the co-author of a related case on policy analysis and recommendations for managing COVID 19 in Indonesia.
[ii] Hayden Cheng is a nephew of Alvin Cheng. He is now in his final year of secondary school, studying in the UK, with an academic interest in forensic psychology. He has had first hand experience of fleeing from the outbreak of the Covid-19 threat in the UK, and was subject to two rounds of 14-days quarantine (in Hong Kong and in Shanghai) during his return trip to Shanghai in mid March. Hayden is instrumental in establishing the data base for conducting the simulation of the Hong Kong cases.
[iii] Maurice Glucksman is an Investor and former Equity Analyst now focused on Disruption. He is a published author on business and policy analysis using Simulation methodologies and was an Associate Principal with McKinsey & Co. He is now also an advisor to businesses in Aquaculture Health, Artificial Intelligence, Marine Shipping, Pharmaceuticals and others. He is one of the architects of this project, a developer of the simulation model and co-author of a number of the localisation cases.