"There Is No Glory In Prevention."

   moon of  alabama      May 06, 2020

When the threat of an Covid-19 epidemic emerged the infectious disease epidemiologists began to build their mathematical models to predict how it would develop. They had to work with low quality data mostly from China and later from Italy. The main parameters where the replication rate R of the disease and the percentage of severe cases. Using the available numbers they predicted a high peak of serious cases that would overwhelm the health care system.

Their next step was to look at non-pharmaceutical measures that they hoped would lower the peak of cases. Some of these were less controversial than others. Closing cinemas and bars is a bit inconvenient but can be done without much protest. Closing down public traffic or schools is more controversial as the effects on the public and personal lives are way more serious.

We have little experience in taking such measures. The model builders do not know how much each of those restrictions will contribute to the lowering of the peak. They have to estimate those parameters. Until this month it was not even clear if children could get infected or were infectious. Arguing for closing schools without knowing that is quite difficult.

Clinical epidemiologists, who mostly work on randomized trials which produce hard data, are often critical of the model builders.

 They dislike the many assumptions that go into modeling and demand more hard data. Stanford's professor John Ioannidis, who ran the Santa Clara antibody study, is one of them. He is somewhat right. All models are wrong, but some are useful. A recent Boston Review piece looks at the differences between the two tribes of epidemiologists. It finds that we need both.

When the politicians take measures they are only in part based on the predictions the modelers made. They also have to look at economic outcomes, at other security issues and they have to take public opinion into account. Quite strict measures were taken in many western countries. They worked well in some of them. 

Germany has hardly any 'excess death' from Covid-19. Other countries, like Britain, acted too late or not to a sufficient degree and had to pay the price for that.

As the epidemic now starts to recede a bit there is quite a lot of criticism of the lockdown in Germany.  'The models were wrong,' some people claim. 'The lockdown measures were unnecessary.' Then follow demands for the immediate lifting of most restrictions.

"There is no glory in prevention" is the frustrating aspects in the life of an epidemiologist. If they do their job too well everyone will bash them.

A month ago Max Abrams saw this development coming and commented:

1. Models make assumption of how much people will social distance.
2. Based on this assumption model predicts virus cases.
3. More social distancing is practiced than assumed.
4. Model over-predicts virus cases.
5. Idiots say models are wrong so we don't need social distance.

Others point to Sweden and claim that its decision to let the epidemic burn without much intervention was a much better way than to go for lockdowns. But the evidence for that isn't there. The numbers show a different picture:

Barry Ritholtz @ritholtz - 18:03 UTC · May 3, 2020Sweden’s Coronavirus death rate > its neighbors https://www.worldometers.info/coronavirus/#countries
Sweden
Total Cases: 22,317
Per /1m Pop: 2,210
Deaths: 2,679
Recovered: 1,005

Denmark
Cases: 9,523
/1m: 1,644
Deaths: 484
Recovered: 6,987

Norway
Cases: 7,809
/1m: 1,440
Deaths: 211
Recovered: 32

Sweden in fact had the very same problems with its medical systems that some other countries also had. It had to ration ICU beds by denying them to people above a certain age. Its economy was hit as bad as other ones:

The effect of virus-fighting efforts on the Swedish economy has been devastating. A very large number of small businesses have collapsed. All but essential industries closed down almost immediately and many face bankruptcy. People have been told to refrain from all non-essential travel. Virtually all air travel has been suspended. Unemployment figures are soaring. The opposition parties deem government counter-measures to be too little too late.
...
Contrary to impressions created in American media, Sweden’s approach to handling the pandemic has not been “relaxed,” but essentially the same as in other Western countries. This country of 10 million has been at least as preoccupied with the pandemic as other countries. Whether its approach has been as efficient remains to be seen. What may stand out as exceptional in the end is Sweden’s glaring lack of preparedness for a pandemic, especially for protecting its elderly, and that the dead are disproportionately recent immigrants.

While Sweden may not have ordered everyone into a total lockdown the people have largely done that by themselves simply out fo fear.

As a comment by one Richard England here (May 6, 2020 at 3:40am) describes that effect:

There are two kinds of lock-down, lock-down by fiat and lock-down by fear (or for that matter, self-preservation). The importance of lock-down by fear explains why Sweden has not done as badly as would be expected. Both forms of lock-down are economically destructive. Lock-down by fiat is usually either too slow or too incomplete to be much different from lock-down by fear, and both are more than enough to knock over a weak economy. Fear dissipates, and the economic life resumes more quickly where the disease has been essentially eliminated.

The effect is also captured in this graph by the German equivalent to the CDC, the Robert Koch Institute. It shows the replication factor R of the epidemic in Germany and three points in time where official lockdown measures were taken.

bigger

The replication factor of the disease in Germany was already decreasing in mid March before the more severe measures were ordered. R was below 1 even before March 23 when the government ordered the lockdown.

The simple reason for that is the people heard the news and watched TV. The pictures and death numbers from Italy in late February were quite brutal. When herd animals sense that an epidemic is taken place within their herd they distance themselves from each other. 

Humans behave similarly. As in Sweden many people in Germany went into some kind of lockdown and practiced social distancing even before it was ordered.

Some now claim that the RKI graph shows that the measures were not necessary. They are wrong. The data was not known when the measures were taken. The first of the simulations shown in the graph was done on April 1. In late March the R seemed to go again above 1 which meant that the epidemic was again expanding. Only the lockdown measures taken on March 23 pressed R below 1 and led to a slow decrease of new daily cases.

Germany is now slowly coming out of its lockdown. The U.S. is doing this too but at a point of the epidemic where it is way too early. There are economic reasons to do so but the early lifting of lockdown measures will likely cost the U.S. many human lives.

Fear will help to overrule that overhasty political decision. The news will continue to report new mass outbreaks in this or that part of the country. The fear will therefore also continue and the people will keep distancing themselves from each other.  How much that will help to slow down the epidemic is difficult to estimate. 

There is now some evidence that the summer will bring some relief from the onslaught of bad news. A study with data from 166 countries and published in Science of The Total Environment finds:

A 1 °C increase in temperature was associated with a 3.08% (95% CI: 1.53%, 4.63%) reduction in daily new cases and a 1.19% (95% CI: 0.44%, 1.95%) reduction in daily new deaths, whereas a 1% increase in relative humidity was associated with a 0.85% (95% CI: 0.51%, 1.19%) reduction in daily new cases and a 0.51% (95% CI: 0.34%, 0.67%) reduction in daily new deaths. The results remained robust when different lag structures and the sensitivity analysis were used. These findings provide preliminary evidence that the COVID-19 pandemic may be partially suppressed with temperature and humidity increases. However, active measures must be taken to control the source of infection, block transmission and prevent further spread of COVID-19.

A hot and wet summer is likely to lower the number of new Covid-19 cases. But after the summer come fall and winter during which we are likely to see a new peak. The fear will be back, social distances will again be practiced and the economic damage will further increase.

We had the chance to do otherwise. China gave us time to take the right measures. It has, like Hong Kong, Vietnam, South Korea and New Zealand, practically eradicated the disease within its boarders. It now has an advantage that will be difficult to beat.

Posted by b on May 6, 2020 at 18:57 UTC | Permalink

https://www.moonofalabama.org/2020/05/there-is-no-glory-in-prevention.html#more