Read on Twitter
1/ (thread) Many in the lockdown debate seem to forget that evidence of problem ≠ evidence the intervention works ≠ if it works the intervention is cost-effective or there are no better alternatives. Let me do a mini literature review.
2/ This paper compares 50 countries with most COVID cases to explore what factors affected cases and deaths: https://thelancet.com/journals/eclinm/article/PIIS2589-5370(20)30208-X/fulltext. It found that "Rapid border closures, full lockdowns, and wide-spread testing were not associated with COVID-19 mortality per million people."
3/ This is a similar study but covered 175 countries. It looked at various factors (e.g. age) as well as an index for non-pharmaceutical intervention measures (NPIM) such as lockdown policies, and found NPIM index was not statistically significant https://www.sciencedirect.com/science/article/pii/S1876034120305724
4/ This paper conducted a cost-benefit analysis of the lockdown policy by assuming lockdown works using Professor Ferguson's (aka Prof Lockdown) model from Imperial. In no scenario (20-400k lives saved) it suggests the lockdown is cost-effective. https://onlinelibrary.wiley.com/doi/10.1111/ijcp.13674
5/ What about alternative solutions? A re-analysis of Ferguson's model suggests "a broad lockdown, as opposed to a focus on shielding the most vulnerable members of society, would reduce immediate demand for ICU beds at the cost of more deaths long term." https://www.bmj.com/content/371/bmj.m3588
6/ In addition to monetary cost, lockdown also caused huge deaths due to disruptions to diagnosis and treatment of other illnesses. For this, there is a good collection papers on this issue: https://collateralglobal.org/health
7/ UK govt's own estimate (by Department of Health, @ONS etc) is around 200,000 extra deaths could be caused by lockdown due to delays to treatment https://metro.co.uk/2020/07/20/coronavirus-lockdown-cause-200000-extra-deaths-13014848/
8.1/ Economic recession will also cost lives. This paper calculated that impoverishment will lead to a loss of average 830,000 lives ( http://www.colbas.org/ntp/opnAxs/N07TH20A.pdf). The net effect (minus lives saved) is comparable to World War II in the UK.
8.2/ Note the paper assumes all restrictions will be lifted in March 2023. Even let's say the 830k figure can be halved (given the recent news on vaccine
https://twitter.com/KateAndrs/status/1327909894385770502) this is still a staggering figure.
https://twitter.com/KateAndrs/status/1327909894385770502) this is still a staggering figure.
8.3/ For a more plain-English explanation of this author's paper see https://www.spectator.co.uk/article/is-the-cost-of-another-lockdown-too-high-
9/ This is a simple correlation analysis (so not as robust as others) comparing 188 countries in the first 8 months of 2020. It found "Stringency of the measures settled to fight pandemia, including lockdown, did not appear to be linked with death rate." https://www.frontiersin.org/articles/10.3389/fpubh.2020.604339
10/ This is another paper on school closure. It was estimated that by end of May, an American child has on average 0.26 years of life lost (YLL) due to decreased educational attainment. Overall 4.06 million (5.53-1.47) YLL was caused by school closure. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2772834
11.1/ New paper looking at the effect of mandatory stay-at-home and business closure on case growth. Using Sweden & South Korea as counterfactuals vs 8 countries, they found "no clear, significant beneficial effect" of these more restrictive interventions. https://onlinelibrary.wiley.com/doi/10.1111/eci.13484
11.2/ In fact, the point estimates of 12 out of the 16 comparisons were positive, meaning mandatory stay-at-home and business closure would INCREASE case growth.
11.3/ It is worth noting that, by using fixed-effects panel data model, the country-level confounding effects are controlled for. This means the results are unbiased whatever (different) properties/characteristics of the countries in the analysis. See https://drwangstatsconsulting.wordpress.com/2019/05/21/myths-on-randomised-controlled-trials-and-fixed-effect-panel-data-model/.
12/ There are papers claiming lockdown works of course. Here I have looked at some of them. https://twitter.com/excel_wang/status/1351175571523448835. It is disingenuous to pretend a statistically insignificant result is effective. No one would do similar for vaccine efficiency, but lockdown is somehow ok
.
. https://twitter.com/excel_wang/status/1351175571523448835. It is disingenuous to pretend a statistically insignificant result is effective. No one would do similar for vaccine efficiency, but lockdown is somehow ok.
