DO lockdowns work? The evidence to date has not been good, however vehemently lockdown zealots have asserted the contrary. The main study drawing international comparisons, published in the Lancet in August, found that neither the timing nor the severity of lockdown measures was associated with the number of Covid deaths. Within this finding is the fact that some of the longest, earliest and strictest lockdowns – Peru, Spain, Belgium – are associated with some of the highest death tolls.
Similarly, local lockdowns in the UK have been associated with stronger autumn resurgences, while analysis of American states by J P Morgan found no link between lockdowns and infection rates.
Last week a study by Dr You Li and colleagues appeared in the Lancet claiming to show that in fact lockdown measures (which they call non-pharmaceutical interventions or NPIs) lower transmission after all. Here’s the summary of their findings.
‘To the best of our knowledge, this study is the first to assess the temporal association between changing the status of a range of NPIs and the transmission of SARS-CoV-2, as measured by R, for all countries for which data were available. On the basis of the empirical data from 131 countries, we found that individual NPIs, including school closure, workplace closure, public events bans, requirements to stay at home, and internal movement limits, were associated with reductions in R of 3 to 24 per cent on day 28 after their introduction, compared with the day before their introduction. Reopening schools, lifting bans on public events, lifting bans on public gatherings of more than 10 people, lifting requirements to stay at home, and lifting internal movement limits were associated with increases in R of 11 to 25 per cent on day 28 after the relaxation. The effects of introducing and lifting NPIs were not immediate; it took around one week following the introduction of an NPI to observe 60 per cent of the maximum reduction in R and even longer (almost three weeks) following the relaxation of an NPI to observe 60 per cent of the maximum increase in R.
‘Our analysis suggests that, in the context of a resurgence of SARS-CoV-2, a control strategy of banning public events and public gatherings of more than ten people would be associated with a reduction in R of 6 per cent on day 7, 13 per cent on day 14, and 29 per cent on day 28; if this strategy also included closing workplaces, the overall reduction in R would be 16 per cent on day seven, 22 per cent on day 14, and 38 per cent on day 28. These findings provide additional evidence that can inform policy-makers’ decisions on the timing of introducing and lifting different NPIs.’
This might sound impressive, but in fact the study is riddled with problems that undermine its purported findings. The authors claim, for instance, that closing schools and banning public events have the largest impact on infection rates. However, they also acknowledge that these were usually the first interventions countries brought in and that the large impact may just reflect being early and first.
They support their conclusion about closing schools by claiming schools are major drivers of infection, citing one study about the high viral load in five-year-olds. But they ignore all the studies showing that closing schools made little or no difference to transmission. Indeed, their finding that closing schools is one of the highest impact interventions brings them into direct conflict with most of the studies on this question, raising wider concerns about the validity of their methods.
In terms of observed increases in infection rates after lifting restrictions, they acknowledge that they don’t allow for increases in testing, yet the early summer (when most of the restrictions were being lifted) was when testing was being ramped up worldwide, so much of the increase at that point must be attributed to that. By failing to filter out this effect they have seriously weakened the credibility of their findings in relation to opening up.
They admit they don’t take the seasonality of the virus into account, and appear to defend this by citing a model that claims to show that temperature and humidity don’t make any difference to transmission, despite it being clear that the virus faded in many places partly due to the onset of summer and is seeing a seasonal resurgence in the autumn.
They acknowledge that there were varying delays in the interventions having an impact, with a median of eight days to reach 60 per cent of the effect, which seems a very long delay for an impact that should really be immediate (their methodology takes into account the lag between infection and reporting tests results). They argue this is likely to be a result of behavioural inertia, which they say is backed up by Google mobility data, but don’t go into detail. It’s hard to see how behavioural inertia could explain a delay in the impact of the closure of schools, which is by its nature an immediate and universal behavioural change. Likewise, if public events are banned, they are banned. Why the delay and variation?
Crucially, there is no sign they have considered how much of the decline in R would have happened anyway, due to natural epidemic decline (herd immunity). As often happens with these studies, one gets a sense that they are assuming their conclusion (that NPIs work) and thus don’t give proper consideration to the possibility that the reduction in R is unrelated to the interventions.
In some ways, though, sceptics can welcome this study because it concedes that most interventions have no clear impact, and even for those that do the effect is very limited. If even the dubious studies of lockdown proponents, with their unrealistically favourable assumptions, show that lockdowns don’t really prevent transmission, and vaccines are known to be limited likewise, what argument is left against the strategy of protecting vulnerable people as best we can while otherwise getting back to normal? None that I can see.
