Tracking COVID-19 in Andalucia and Spain

Created 20/3/2020. Updated 18/5/2020

©James Wimberley, March 2020. May be shared freely, though attribution would be courteous. Any modifications must be noted.

All data as of 17 May 2020. Sources at end.

To keep tabs on the progress of the epidemic I wanted two things:

- a historic table of total cases

- the growth rate.

How bad the epidemic is depends on how fast the infection is growing, and whether or not it slows down.

So I built a table and two charts. The table below just lists the cumulative cases in Spain and Andalucia. It’s obvious that infection is still spreading quickly both in our region and in Spain as a whole.

On the second aspect, I created two charts. These are on a logarithmic scale. The reason is that this shows a constant growth rate as a straight line (though parallel lines on log charts don’t mean that the rate is the same). When the line bends downwards, you can easily see that the growth rate is slowing. If you are an optimist, you can just see the beginning of this. The quarantine was only introduced at the weekend, and since there is a ≈ 5-day incubation, plus a few days for diagnosis, we should not expect to see a significant slowdown before early next week.

Finally I calculated moving 7-day growth rates, for the latest day and the five previous days. This will pick up short-term swings.

So far, the growth rate is 33% less in Andalucia than in Spain as a whole, though the gap is slowly narrowing. This is good news as at worst it will flatten the curve and reduce the burden on hospitals, and at best gives containment a better chance.

I plan to update this page regularly. If you find it useful, think of sending a message of encouragement to James dot Wimberley at gmail dot com. Also if you find a professional source of the same data, so I can give up.

I’m sorry it’s only in English. Adding Spanish would make the updating too much of a burden.

Update 18/5

Really good news all round: new cases down 28% to 285 – best since 9 March -, Andalucia back to 8 and Madrid only 5, new deaths down 32% to 59. Cross your fingers but this is what winning looks like.

Update 17/5

New deaths apparently below 100 per day. I say “apparently” because May 15 never appeared on the ISCIII website, though I checked several times. To rescue my charts, I invented numbers for this day by interpolation.

The mass antibody tests show that only 1 in 10 of infections reaches the stage of diagnosis and reporting. For most people, this is a mild flu. The problem is that it’s both very contagious (so numbers are high) and lethal to a significant minority (1% of a big total). They also indicate that 13% of those who test positive for the virus by PCR don’t have detectable antibodies, so they must have defeated the virus some other way. Half of those who do have antibodies (2.5% of all participants) are completely without symptoms.

Update 15/5

Deaths back on trend, cases stay a bit above it. This time Castile is to blame. New cases in Andalucia the same as yesterday – technically possible, but I suspect a reporting glitch.

Update 14/5

National upticks. Andalucia back up to 42 new cases. The Junta’s table reports one new case each in Vélez-Malaga and Torrox in the last 14 days, but they wren/t thee before.

I’m having second thoughts about the reason for the slow decline in deaths, much flatter than in cases. If it were down to testing and reporting, you would expect wide differences between countries, but it’s a universal pattern (see eg here). That suggests that it’s a real feature of the disease. Either the resistance of vulnerable patients erodes slowly, or they “recover” but are weakened and succumb to a secondary infection. A depressing prospect either way.

Update 13/5

Andalucia back up to 26 new cases. I can’t figure out where.

Update 12/5

Andalucia down to 8 new cases! Extremadura at 1 and Murcia back to zero. Sadly, the national rate is an uptick, driven by the usual suspects (Madrid, its commuter belt in Castile and La Mancha, and Catalonia). Deaths also up.

ISCIII have not corrected the total of new cases. They must be correcting back totals, which I can’t follow. I have stopped reproducing their headline number and am just reconstructing a total from the sum of the new cases by region. After all, that’s closer to the source data and therefore more reliable. What we are interested in is the recent trend.

Depressing thoughts on Twitter by my co-blogger Keith Humphreys, a psychology professor at Stanford, on American tolerance for the sort of intrusive TTI scheme public health experts assume is necessary to quit lockdown safely. If his fears are justified, the USA will end up with herd immunity (200m infected, 2m deaths, Trump flees to a warm country with no extradition treaty). And is Trump really having a test swab shoved painfully up his nostrils every day? My guess is that he is simply being checked for symptoms.

Update 11/5

Nothing much yesterday, but good news today: new cases down to 373, all regions below 100, deaths “only” 123. That is, if I’m right about that the reported total of cases (227,436) is a typo. It would give a ridiculous >3,000 new cases. I just added up the new cases by region and used that instead.

I’m still wondering what to believe about Vietnam (population 96m). Official total of cases 288, deaths zero (Johns Hopkins). As a ruthless Leninist oligarchy, the government would have no moral scruples about lying for reasons of state. But it’s hard to see why they should. Deaths on any scale are hard to conceal and the lie would blow up. On the other hand, it’s an efficient and competent Leninist oligarchy, overlaid on a Confucian culture of social discipline. Also, they’ve been primed by 2,000 years of Chinese imperialism to distrust what the Chinese government said early on about community transmission, and I believe closed their borders early. Like Korea, it’s not a touchy-feely culture in personal interactions.

Update 9/5

A day late for VE day, but good news resumes. Andalucia down to 11 new cases, I had to rescale the chart! Catalonia still leads the hall of shame, and theValencian Community is up – a warning against complacency. ISCII have revamped their Covid data web page. There is a world map, though it doesn’t do anything yet. A waste of resources, as anybody can go to Johns Hopkins.

Update 8/5

Not good. New cases back over 1,000 – and that’s by correcting yesterday’s total upwards, the raw figure was 1,410. As usual, Catalonia (543) and Madrid (148) are holding back the rest of Spain, Andalucia back down to 19. Deaths stubbornly refuse to fall further.

There is a disturbing pattern in the low-incidence regions, including ours. Murcia had some days at zero, then a handful popped up. Same for the islands. My inference is that there is no solid TTI in force – test, track isolate. Do this right, and you get to zero. There is a fantasy in Europe (see France and the UK) that technology will come to the rescue – see, we are trialling this awesome smartphone app! TTI worked in the past without phones – it was developed for animal diseases like foot-and-mouth before WWI. The key requirement is a small army of trained busybodies with narrow but real police powers. You try not answering the Covid tracker’s questions in Taiwan or Korea about where you’ve been, or refusing a test. Most of the resulting self-isolation of suspected contacts is voluntary, but everybody knows there is an “or else” in the background. The local authority will deliver food to your quarantined door, and find you a hotel room if you don’t have a fixed address. The is one alternative to this tested nanny-state scheme: wait for infections to rise again, and reimpose mass quarantine. Catalan nationalists will not like the Bourbon oppression of TTI one bit. Good.

Update 7/5

Same old, same old. I moved the updates for March and April to a separate page, here. Note that with the low numbers for Andalucia, the log chart exaggerates the day-to-day swings, which don’t mean much.

Update 6/5

New deaths up for the second day running. Concerning, but no need to panic yet. The new cases are doing better. Why are deaths falling so much slower? Are hospitals getting better at prolonging life but not preventing death? That does not seem likely given that the only treatment is mechanical ventilation, which just gives the body’s immune system a better chance. A better bet is steady improvement in post-mortem testing and reporting of cause of death. That means that the true peak was much higher than reported, which would give us the expected parallel decline to cases.

Update 5/5

Upticks in Spain, but new cases in Andalucia have fallen to 16 – the first time below 20. With low numbers generally, expect to see more noise in the day-to-day numbers. Paranoia special: Catalan nationalists have accused the government of sending 1,714,000 masks. Obviously a triumphalist Bourbon reminder of the disaster of 1714!

Update 4/5

It continues, with a 58% fall in new cases in one day to 356! Andalucia at 19, Madrid at 1, and even Catalonia down at 132. New deaths are exactly the same as yesterday – I suspect a reporting glitch. It looks (touch wood, then disinfect) as if we in Spain, citizens and foreigners together, have broken this thing. But it’s still out there and will try to come back. France and the UK have at least parts of a TTI plan (test-track-isolate), but what about Spain?

Update 3/5

Really good today on all fronts: new cases down below 100 at 838, Madrid stays low indicating yesterday’s crash was not a fluke, new deaths below 200 at 164, Andalucia stays low (though corrections have made the series a mess). Common sense would suggest keeping the lockdown in place in Catalonia, where new cases are high at 384, but I suppose politics will rule that out. I fear a resurgence there.

What I don’t see is a plan for aggressive case tracking, backed by mass free testing and targeted isolation (forcible if necessary). 3Ts or TTI. That’s how South Korea and Taiwan got early control of the epidemics, and kept it. Without these, an exit from lockdown is trusting to luck.

Update 2/5

Dramatic fall in new cases to a new record of 1,147. Driven by a welcome collapse in Madrid to only 25! Half of the total (543) is now in Catalonia. Perhaps they have a dumber type of secessionist than the Basques or Scots.

Update 1/5

Not much in the way of May Day cheer: upticks in new cases (Andalucia back over 100), deaths flat. But new cases sharply down in Madrid to 52 – wait for confirmation of the trend.

Older updates here


Spain To 16 March: WHO daily situation reports (archive)

From 17 March: Instituto de Salud Carlos III (no archive, can be checked against WHO)

Andalucia: Junta de Andalucia (archive)


* Before 12/3, Andalucian reports only cover new cases and are not consistent. Numbers for recoveries are only given on ISCIII website. The old Andalucian series seems to be misaligned by one day with the national series. I have aligned on the latter. This makes no difference to the growth rate.

Logarithms. Example: take a series growing 10 x at each step. It goes 1, 10, 100, 1000 .. The corresponding base-10 logarithms are 0, 1,2, 3 ... which is obviously linear. Change the base to e, and each term is multiplied by the constant (ln 10) = 2.3. Same thing.