Boo to those saying we are opening up too early

A blog about COVID 19 in Malaysia.  Updated 10 Aug 2020 by integrating 2 other posts - Time to Panic? (4 May 2020) and Is there cause for concern? (3 May 2020)

On 2nd and 3rd May, there were spikes in the number of cases. This comes at the time when the country was about to relax the lockdown. There were thus 2 major related questions - is there any concern with the spike and are we relaxing the measures too early?  With hindsight, spikes are recurring phenomena so the analysis to look at whether there is cause for concern remains valid even today. Secondly, we are going through a phased relaxation and at every phase, there will be concern about whether we are moving too fast.

Designed by Freepik

Are there any concerns; is it time to panic?

The spike

• We had 105 cases on 2 May 2020 c/w an average of 54 daily new cases for the week before. Is there cause for concern?

• We had 122 cases on 3 May 202 and a plot of the virus trajectory is shown below. For those who have forgotten, when the MCO was implemented on 18 Mac, there were 117 cases reported for that day. Time to panic?

The Concerns

1)  While I have expected double digits daily new cases, the triple-digit figures two days in a row and the reversal in the trendline of the weekly highs make you wonder whether there are reasons to be concerned about a 3rd wave.  

2)  Life is full of ironies. When the MCO was first implemented we were experiencing exponential growth from the Tabligh cluster and there was a concern of the 3rd wave from those coming from their school holidays overseas tours. 

Fortunately, there was no school holiday spike. Today the numbers from the Tabligh cluster and sub-clusters are declining but there have been concerns about opening up too soon due to concerns from a new 3rd wave. 

The problem with those voicing their concerns about opening too early is that I have not seen any party providing numerical data to support their positions. Neither has the govt provided numerical analysis to refute the 3rd wave. 

The Analysis

There were several ways to analysis the situation

1. Comparing Optimistic views with Pessimistic views
2. Comparing differences between the 122 cases on 3 May and the 117 cases on 18 Mac.  
3. Looking at whether there is any change in the underlying conditions
4. Can charting tell us anything?
5. Try to trace the source of any third wave

1)  Comparing Optimistic views with Pessimistic views

We looked at the 105 cases of 2 May for this.

Optimistic view

The 105 cases can be broken down into

• 11 returnees
• 60 EMCO areas
• 34 balance

The MOH report is not very clear and I deduced from the tracking of clusters that we had:

• 41 from the Tabligh cluster. Some of those captured under the Tabligh cluster could be double counted as part of the 60 under EMCO.  There was also a report of 6 from a new Tahriz cluster which I suspect is part of the Tabligh cluster.

• 5 from the Church cluster.  I am sure this is part of the 34 balance.

Piecing together the Church and Tahfiz info, I am sure that the number of cases due to the local non-cluster group would be less than 23 cases.  

This is around what we have been experiencing from this group over the past week so I would not be worried about the 105 cases.  I would have worried if our PM had not announced the conditional MCO as these 105 cases would have weighed heavily on the lifting of the MCO. 

Pessimistic view

While the optimistic view is that the community spread from the 105 cases are manageable, the pessimistic view is that during the conditional MCO period, the surge from all of the groups may be more than what we can handle from the contact tracing and treatment perspectives. 

• While there has not been any update on the number of returnees since 28 April, I deduced that there are still people coming back and who have been tested positive. (Refer to Note 1).  

I would expect the number of cases from this group to continue for some time more.  The good news is that with mandatory quarantine, there is a low risk of this group infecting the general public

• There is nothing to suggest that there will not be any further cases from the foreign workers' community.  As many of them would be returning to work, the unknown is the asymptomatic cases.  The SOP that companies will follow will not be able to detect the asymptomatic cases until it is too late. 

• The Clusters are more perplexing.  The Tabligh and Church clusters are into their 5th generation of cases (where 1st generation is the one after the original index cases).  

It has been more than 2 months since the index cases first had their respective events and yet there were 41 new cases from the Tabligh cluster and 5 from the Church cluster yesterday.    

 As late as yesterday MOH reported a new sub-cluster ie the Lanchang, Pahang Tahfiz sub-cluster.  All the Tahfiz sub-clusters are subs of the Tabligh cluster.  

It makes you wonder how good has been our contact tracing and what would happen under the conditional MCO when more people can move about. 

• With a steady number of cases from the community spread during the MCO period, does it mean that the infections came from the asymptomatic cases?  Surely with the conditional MCO, the number of cases will go up.  Why isn't it mandatory for all to wear masks when appearing in public?

My view was that the conditional MCO was a sudden announcement as it was not part of some pre-opening plan. Even in the US, there was an unveiling of an opening plan before some of the states started to open.  The European Union and WHO also had their respective opening plans.  

I am guessing that since we have not announced such a plan before the conditional MCO, there is some "udang" under the conditional MCO "batu".  The announcements by Sarawak, Kedah and Selangor about not following the Federal govt conditional MCO only fuels this. 

2) Comparing differences between the 122 cases on 3 May and the 117 cases on 18 Mac

The 122 cases can be broken down into

• 52 returnees
• 58 – 65 from clusters/EMCO areas: note that different media had different figures
• 5 - 12 balance from the community or sporadic transmission

The profile from yesterday also showed a similar pattern of 11 from returnees, 71 from clusters/EMCO, and 23 from community spread. 

In contrast, the profile for 18 Mac was

• 0 returnees
• 100 clusters (Tabligh, Church, case 26)
• 17 balance from misc sources 

Furthermore in Mac, the number of daily new cases from both the clusters and community spread was growing whereas today, at least the number of cases from community spread appears to be steady if not declining. 

We can safely say that one of the drivers of growth in the number of new cases today is the returnees.  

However, due to our mandatory quarantine requirements, our Defence Minister is right to say that this group is unlikely to be the cause of exponential growth.

If there is any risk of exponential growth, it should come from the clusters and/or EMCO areas.  It is unfortunate but the current EMCO cases are mostly foreign workers related.

I am still grappling with why we see an increase in the number of new clusters and cases among the foreign workers so far into the MCO ie if the incubation period is 1 to 2 weeks shouldn’t all the clusters and/or infections among the foreign workers be already flushed out in the first 2 to 3 weeks of the MCO?  

This needs to be resolved in order to safely rule out the clusters and/or EMCO areas.  I have not got a good answer yet. 

3) Underlying conditions

The only thing that changed significantly during the past 2 weeks was the extension of the MCO till 12 May and the corresponding relaxation.  

I don’t think this change would impact on the number of cases from returnees, clusters or EMCO areas.  

If there was any impact, it would manifest in the community transmission part and I think it would come from contact with asymptomatic cases OR with the foreign workers.  So to have a satisfactory answer we need to look at these 2 sources. 

4) Charting

I had originally thought that as the number of cases was coming down, there is not much point in trying to project the number of cases.  The 105 jumps have given me second thoughts and I thus revised the model to account for the relaxation of the MCO.  

This new model projects that on 12 May, the number of daily cases would be 30 whereas, under the previous model, the projected cases would be 10. (refer to Note 2)

The charts below show the composite performance as compared to actual.  The first chart shows the actual number of cumulative cases c/w with the composite curve.  The second chart (which is derived from the cumulative chart) compares the actual daily number of cases with the best fit curve for the respective periods. 

I also reviewed my past 2 weeks postings to see whether there was anything that I missed out.

I had sort of anticipated the sideway move when I saw the higher “weekly high” in Wk 6 MCO and have been talking about focussing on the numbers from the community spread (ie non-returnees, non-EMCO, non-cluster).

However, there is a conceptual issue with the curve fitting model.  Although the current model is a composite one comprising several layers of curves, I have modelled the final MCO period as the tail end of an S curve. 

When you look at the trajectory of the actual number of cases ie starting from 19 April, you could argue that it could also be represented by either

• The start of a new S curve OR
• The middle portion ie flat top 

It goes without saying that the projected trajectory would be different depending on what is chosen.

The only way to answer this question is to wait for a few more days of data. If the number of daily cases continues to increase, then it has to be changed to reflect another wave of exponential growth ie 3rd wave.

But looking at the analysis so far, I think that at worst would be another flat top. Either a flat top or a tail end of an S curve would imply that there is no 3rd wave.

5) Try to trace the source of any third wave

At the peak of the MCO, we had about 200 cases per day. Over the weekend we had about 100 cases per day. Are those voicing concerns saying that the numbers will double ie 3rd wave? If so, where would this doubling come from?

• About 30 % of the 100 are returnees. Even if these were to double due to more returnees, it would not increase the risk to the general public.

• About 60 % are from clusters and EMCO areas. There is no increasing trend for this group. So why would it be doubled as the EMCO areas would still be under lockdown and those infected and PUI would continue to be hospitalized and/or quarantined 

• As for the balance from the community spread, the trend is downwards. For this to reverse direction and doubled, it must mean that there are many asymptomatic cases walking about and even if true, the various social distancing measures, makes, and the ban on gathering would reduce this transmission. 

Even if the cases doubled and remain steady at 200 cases per day, we still have the medical capacity to handle them (I have shown that it is about 370 cases per day).

Conclusion

The current cases from community spread are small.

  

Returnees, existing clusters, and the EMCO area are not likely to be sources of a 3rd wave.

There are probably asymptomatic cases and can be a source of further community spread,  But this is unlikely to be a driver of a 3rd wave 

Any source of a 3rd wave would be from new clusters and/or foreign workers   

I think that the likelihood of a 3rd wave is low cos

• We will continue to test the foreign workers' community and EMCO their residences

• While slow, MOH is tracing every member of the clusters

Transmission from asymptomatic cases will be minimized with the ban on gatherings, wearing of masks 

I must admit that more analysis is required for the clusters and foreign workers before we can confidently rule this out as the driver of any 3rd wave.

Are we opening up too early?

The MCO is supposed to end on 12 May. Why would waiting another week make a difference?

I would argue that waiting for another week for the number of cases to come down would not reduce the transmission risk further – definitely not from the returnee or clusters/EMCO groups. Any risk reduction would be from community spread and I think it makes no difference whether this is 20 or 10 cases per day as these tend to be distributed all over the nation.

Given that the virus incubation period is 1 to 2 weeks, waiting for another week is not sufficient time if the intention is to see the impact of any action before taking the next step.

If there was a shortcoming in the govt plan, it is the post MCO one. I suspect that since there has not been any indication of this, the likelihood comes 12 May is that the conditional MCO would be extended. Perhaps the barbers and saloons would be allowed to open but all else remain unchanged. 

It is not often that we come across socio-economic phenomena where science-based methods can provide an accurate near-term picture.  

Predictable

The COVID 19 pandemic is one such case where while we can debate about the causes, its immediate trajectory is very predictable. As such any efforts or claims that are not consistent with the data can be easily dismissed.

That is why I can understand the logic for Sarawak to be cautious about opening up the economy as the compounded daily growth rates (CDGR) of new cases for Kuching and Kota Samarahan were 1.6 % and 3.2 % respectively c/w the national average of 1.1 %

Where else are there significantly higher CDGR c/w with the national average?

• Lembah Pantai – 1.8 %
• Titiwangsa – 2.3 %
• Cheras – 1.7 %
• Putrajaya – 2.9 %
• Melaka Tengah – 4.0 %. But this is because the returnees are quarantined there

Given the above statistics and that many of the EMCO areas are in KL and Selangor, it is not surprising for Selangor to go slow although I would argue that as the cases are from known clusters and/or EMCO areas, the logic for the go-slow is not clear. 

Negri is probably worried about the Sendayan cluster. But I can’t understand Kedah, Sabah, or even Penang’s decision. 

And I am also waiting for data-driven explanations from the various states why they think that waiting for another week will help to prevent further spread of the virus. 

Just getting lower numbers alone is not going to help. Unless these states need another week to put in more measures to prevent community spread or cluster-based spread, it sounds like politics.

BTW, to those voicing concerns, you should base your challenge on where we have failed vis-a-vis the WHO 6 points for re-opening as this was cited by the PM as justification. 

Looking at these points as per below, you really need to be numbers-driven to make a credible case 

• transmission under control

• health system able to detect, test, isolate and treat every case and trace every contact

• hotspots risks are minimized eg nursing homes 

• schools, workplaces, and other essential places have established preventive measures 

• risk of importing new cases can be managed 

• communities are fully educated, engaged and empowered to live under a new normal

Areas of concern

But if I was MOH and I see the change in the direction of the trendline of the weekly highs, what would be the areas of concern?  I think they are: 

• Asymptomatic cases
• New clusters
• Foreign workers

The biggest concern from the data is the “missing cases” that arose from the backward-looking mortality analysis.  

While I have provided one tongue-in-cheek explanation that they were mild and/or asymptomatic cases and have been explained by the slow contact tracing, it is too hard to believe that ALL have been accounted for.

I came across a report that about 47 % of the cases in Sabah have been due to asymptomatic cases and numerous studies around the world indicate that there are significant asymptomatic cases. 

As I don’t think that there is a flaw in the backward-looking mortality method, my conclusion is that in the conditional MCO phase and thereafter, the increase in the number of cases attributable to community spread would most likely be from people coming into contact with the asymptomatic cases.

Since there is no indication how large this group is, it is thus best that we work as if such asymptomatic cases are present in the community and protect ourselves from them as we enter the conditional MCO stage eg

• We all should wear masks and avoid crowds
• MOH should have random tests of the general population
• Avoid crowds and stay at home whenever possible

New symptoms

I came across an article suggesting that there were many asymptomatic cases because we were only looking for common symptoms based on China’s experience.  

Now that we have more experience, the list of symptoms should be expanded to include the several new ones that were recently added to by CDC ie chills, muscle pain, headache, sore throat, repeated shaking with chills, and a loss of taste or smell.  I don’t think the current MITI SOP for businesses included these additional symptoms.

The positive news about the asymptomatic cases is that if this is correlated to the number of cases from community spread, then it may not be such a big number as the community spread is small and declining.  

The other positive news is that if you look at a map showing where the non-clusters, non-EMCO cases are located, you will find that they are quite spread throughout the country.  

No doubt it is concentrated in the towns but the distribution gives me confidence that one is equally likely to be infected by an asymptomatic case anywhere in the country. 

Technically the way to validate this is to look at the number of community spread cases per capita but unfortunately, MOH does not provide such a breakdown.

 

Controlling inter-state travel

This logic could help answer OP TCF posting of the 500,000 who have applied for inter-state travel.  

Unless the people traveling are healthcare workers or those dealing with clusters and EMCO areas, the risk of them being infected would be if they were in contact with asymptomatic cases prior to their travel.  I would think this risk is about the same wherever you are in the country. 

If I was the authorities I would 

• First, approve travel from areas without any virus history

• Next, allow those areas without any EMCO or cluster history

• The last would be those from the EMCO and location-based cluster areas.  Even then, I would suggest that the travelers are all tested first

Unfortunately, I have little hope that it would be done in this manner and so we have to figure out our risk mitigation plans if most of the 500,000 are coming back to the Klang Valley.  

The good thing is that we only need to worry about those coming back from a handful of areas eg

• Sungei Lui, Selangor
• Sendayan, Negri Sembilan
• Simpang Renggam

On the other hand, if any of the 500,000 are traveling away from the Klang Valley, there should be a plan to ensure that they have not been in contact with those in the many Klang Valley EMCO areas. 

Mystery 

A couple of days ago, I posted 3 mysteries and "solved" two of them. I think I now have some plausible answers for Q1 - How come the number of daily tests does not show an increasing trend even though MOH has repeatedly mentioned about ramping up its testing capacity 

I think it is a timing issue since it takes several days for the laboratory diagnosis. The statistics below seem to suggest this could be the reason. 

Period              Total number of tests for the period
Past 7 days                  64,342
7 days earlier              27,599
14 – 7 days earlier      26,401


Syabas to MOH

Mystery 4

Yesterday, I mentioned another mystery - why at this stage of the MCO are there new clusters being announced as well as many cases among the foreign workers. The logic was that if the incubation period is 1 to 2 weeks, shouldn’t all the clusters and/or infections among the foreign workers have already been flushed out? 

I think I have been muddled in my thinking. By right we should differentiate between types of transmissions eg clusters or community spread AND groups of people eg locals and foreigners. 

Looking at the issues from the above perspective would provide a clearer picture.

My hypotheses for Mystery 4 are:

• New clusters – today we have 34 clusters and sub-clusters. This is because MOH occasionally re-classify the cases through contact tracing. Many of the sub-clusters are actually linked to the Tabligh cluster. 

While not specifically identified as such by MOH, there were cluster transmissions among the foreign workers in the EMCO areas. So the new clusters are actually an administrative issue and it is wrong to see them as representing some trend.

• Foreign workers - My view is that MOH overlooked this group earlier on and by the time they tested them the virus had already spread to several generations. 

We are suggesting that many foreign workers may have fallen sick and already recovered. Those who were sick did not seek treatment either because they were asymptomatic or were afraid. The ones currently tested are the latest generation.

Talking about testing foreign workers, I find it amusing that our Defence Minister is asking for employers to be responsible for testing all the foreign workers in the country. 

Taking a conservative 3 million workers and assuming 10,000 tests per day (i.e. no else gets tested), it will take 10 months to test all the workers. It looks like someone forgot to inform the Minister of this in-practical ruling.

Note
1) As of 28 April, of the 22,671 returnees, only 14,279 samples have been collected with 154 positive, leaving a balance of 8,338 to be tested. Unfortunately, there has not been any further update on the status of the total under quarantine or tested. Considering that as of yesterday we had a total of 279 returnees tested positive and comparing this with the expected number if the 8,338 had all been tested, I deduced that there have been more returnees since 28 April.


2) For the statistically minded, taking into account the past week of a sideways move, I have revised my curve fitting model to be a composite one comprising 4 different curves – one for the 1st wave, the other for Pre-MCO, another for the Mid MCO and the final one to represent the final MCO stage. As the number of cases from the 1st wave was relatively small, we took the actual values in the model. The Pre-MCO curve was modeled as a Logistic curve while the other 2 were modeled as general bell curves. Curve fitting was done manually using the least square method based on the cumulative chart. The daily chart was derived from the cumulative chart.

- - - - - - - - - - - - - - - - - - - - - - - - - - -

Be fearless to get more to Stay Safe, Share this Blog

PS: This blog is for me to better understand COVID 19 as this will impact my investments. If you are also into equities, follow me at i4value.asia

Disclaimer:  I am not an epidemiologist, healthcare worker, pharmacist, or staff in the Ministry of Health, but rather is someone with a strong interest in numerical analysis.  The content is an attempt to understand what is happening in the battle against COVID 19 from a data-based perspective. The opinions expressed here are based on information extracted from readily available public sources but I do not warrant its completeness or accuracy and should not be relied on as such.