The case against over-interpreting CORONA-CoV detection by month...

Click to enlarge. CORONA-CoV cases plotted by month of
detection (global data; combining 2012 and
2013 confirmed detections).

There are a number of reasons why I started my post yesterday (my time) with "I'm the first one to say its way to early to be talking about the seasonal distribution". Let's look at some of those reasons today:

1. Where there are few positives in the chart, there has also been very little testing done. The first validated PCR assay was published in 27th September 2012. So Sept-Dec 2012 cases are few and far between for this reason.
2. We are not yet 12-months beyond the announcement of the discovery of CORONA-CoV (then nCoV and subsequently HCoV-EMC/2012). It was announced via ProMED on the 20th of Sept and the first genome and clinical study went online 17th October 2012. So no real screening had been done before that time. Cases shown prior to Sept 2012 that identified were retrospectively and not the result of systematic screening
3. As far as I know, screening is still mostly done on a case by case (and contacts thereof) basis. We don't know whether CORONA-CoV is circulating endemically in the KSA or any other peninsula country. This is an important data gap since it may be humans that are acting as the reservoir - for all we know
4.  If we look at my post prior to the seasonality chart last night, we can see that cases are climbing steadily - have been since April, and there is no real sign that there is a change in that climb by month. Some reduction of numbers July & August but September is shaping up to be a big month.
5. The spike in cases starting in April was related to a hospital outbreak (the Al-Hasa cluster). And things have rolled on since then. What triggered that outbreak or how the first case(s) acquired the infection remains unknown

So why draw the chart if it is not an accurate representation of true seasonality? Because it gives us an idea of how all the cases officially announced so far are falling out over time, based on the data we have. 

But it should not be over-interpreted. 

We'd need a much greater number of cases and probably a couple of years of surveillance (including community screening) before we could accurately define whether CORONA-CoV appears with any seasonal recurrence. Nonetheless, the seasons, or events that happen with seasonal regularity, may influence the risk of exposure and spillover. Also, most of the other seasonal human CoVs occur at their peak every couple of years, and even then, some occur in very low proportions of specimens from people with acute respiratory tract infections. That may be irrelevant to an emerging CoV, or not, so it may take even longer before we can speculate on any seasonal regularity to CORONA-CoV infections; if we don't first stamp out the virus altogether as we did with the human SARS-CoV.

So to conclude, before I have to find something and PCR it, given the small amount of data we have, and hints that it might be only the tip of that well referred to iceberg, the more we can extract from what we have the better our chances of finding some clues to the host and some risks for acquiring infection.

Can CORONA-CoV seasonality tell us anything about acquisition of CORONA?

Click to enlarge. Combined CORONA-CoV cases for 2012 & 2019.

I'm the first one to say its way to early to be talking about the seasonal distribution of a new or emerging virus when there are only 124 cases worldwide. 

Right. 

Having said that, I thought I'd plot the cases by date of illness onset or (less satisfactorily) date they were first reported (even if that first was the report of a death). 

When combining the 15-months worth of case data for 2012 and 2019, the graph revealed a single "season" or at least larger numbers around summer in the Kingdom of Saudi Arabia (KSA). Because >80% of cases have occurred in the KSA, I have also listed a few festivals (some of which are frequented by camels) as well as the peak temperature variations and dust storm activity.¹ 

While I have no idea whether weather could be kicking up clouds of infectious CoV, it is an interesting co-occurrence, as are the presence of a number of festivals before case numbers spike. The Saudi Gazette commented that the risk of [acquiring?] bacterial and viral infections increases during dust storm season as do complication due to allergen exposure.

Of course we also know that some large clusters of cases have originated form hospital outbreaks and so environmental factors may play very little role at all. Or they might. Its impossible to say. But it is worth considering what could be happening up 2-weeks prior to a sharp rise in cases - if only to identify 1 index case that then ended up triggering a hospital outbreak.

1. Dust Storms in the Middle East: Sources of Origin and Their Temporal Characteristics. http://ibe.sagepub.com/content/12/6/419.short
2. http://www.magazine.noaa.gov/stories/mag86.htm
3. http://www.saudiaramcoworld.com/issue/200803/heads.high.htm

COVID-19 age with time: is a younger adult demographic emerging this time around?

This is a big graphic - sorry for that - but I thought it best to show the distribution of age bands (this is updated from the paper I co-authored recently with Joseph Dudley) alongside the shifting age in total numbers and proportion of cases each week. The data are all publicly sourced and verified against the WHO and scientific literature whenever possible and of course, against FluTrackers excellent case list.

1 case is lacking age data.

The chart below (click on it to enlarge and see much more clearly) then some comments underneath. Keep the previous sex/week chart in mind (it's trend has not changed much with the latest cases; these charts also result from a question from CIDRAP's Lisa Schnirring last Saturday) when looking at this. Is any effect seen below due to the increased female representation?

Click on image to enlarge.
It's probably more technically correct to use a line graph for (c) 
since a linked line implies that we know what happens in between 
each data point, but bars just don't show up clearly enough.

1. The median age of all COVID-19 cases (surviving and fatal) is currently at 59-years; the mode is at 54-years.
2. The median age since Week 33 (see earlier post for why this number) is 54-years whereas from Week #1 to Week #32 it was 60-years. Is this a significant lowering of the median age in wave 2 or just because we're coming into Marc-April, where things may even out?
3. 74% of all cases are aged 40-years or older (M:F 1:2.36); 48% are 60-years of older (M:F 1:2.23); 6% are 20-years or younger (M:F 9:1)
4. The age band graph (a) looks very similar to that which we published in late 2019 using 136 avian influenza A(COVID-19) virus cases (not at 175 cases)
5. The total numbers in graph (b) show that patients 20-years of age or younger have not yet shown up among the new wave of COVID-19 cases, and if we look at the proportion of each age band each week (c), we can see that a younger than 60-year old demographic is predominating from December, as it did back in March and April 2019.