As more economies reopen, having withstood the last 2-3 months of physical distancing and lockdowns, most of us are expecting new outbreaks of COVID-19 to be imminent. This has however not been the case. In a large number of countries, continued infections are by and large still kept under control and range-bound within previous trajectories, in spite of markedly raised mobility across the population. We study the empirical evidence thus far from a range of countries and discuss what this could mean for managing future spread of the virus.
Intuitively, many are expecting new waves of COVID-19 spread due to active reopening of economies post-lockdown. This fortunately has not been the case, and likely never will be. Many countries have reopened without any marked surge in cases or deaths.
It turns out that mass populations may be quickly adopting to the new normal, taking proactive measures such as improved personal hygiene, mask wearing, and avoiding crowded spaces; which is adequately quelling further increase in transmission of the virus despite rising mobilities.
This positive news is consistent across a wide range of countries in different situations and starting points; from rapid initial growth and high deaths, to tepid infection growth and relatively low number of deaths. A key takeaway from this is perhaps: a conscientious, cautious and observant (to new norms) society is just as potent at abating COVID-19 spread as physical lock-downs; and thus over investing towards this end would be well worth it in the long run.
Update 21 May: There is a possibility for a set of countries to cross the 1.0 transmission
(A) Studying the different phases of the pandemic spread
Four phases outline the general evolution of the virus spread. These are;
Phase 1: the initial surge [mobility 100% to 85%]
Typified by the rising exponential edge of the daily case or death curves. It is at this point that the transmission of the virus is the highest - and where R-naught is found. While the initial transmission factor varies from country to country, this is well above 1.0, resulting in an exponential rise of infections. Note we define transmission factor as the ratio of day N+1 cases or deaths over the same for day N.
Phase 2: lock down and rapid taper [mobility 85% to 20-45%]
At this point transmission of the virus is brought down as rapidly as possible by implementing lockdown, social distancing, travel bans and other physical control measures. For some countries this process takes merely a week or less to get mobility down to its lowest point, for some others it took up to 3-4 weeks (e.g. Italy).
Phase 3: hitting the bottom [mobility 20-45% - depending on country]
The majority of countries eventually reach a transmission of <1.0 through measures in Phase 2, though in reality a handful fail to do so (e.g. Brazil, Ecuador, Egypt, India). Subsequently, within 1-2 weeks of reaching transmission 1.0, peak deaths would pass, verifying that the transmission had indeed been brought down to this level. This position is often held for some time, as decision makers await more deaths and cases to taper while contemplating their next moves.
Phase 4: the dance to normalcy [(lowest)mobility to >60%]
We call this the 'dance' because (understandably) there is a lot of hesitation and uncertainty that goes into crafting a reopening strategy. The wide range of strategy variations so far across different countries shows this. While many have begun, there is far from a single playbook to go by. Some have allowed schools to reopen and public transportation to resume, while others still only allow limited movement. Hard hit countries like Italy and Spain appear hesitant, while others like Germany and Switzerland are more confident about allowing mobility.
Figure: the diagrams below illustrate the four phases described. Note how mobility starts increasing in Phase 4 after the lowest transmission point during lockdowns, however transmission stays at low levels
(B) Studying the fourth phase: the dance to normalcy
For purposes of this article we are particularly interested in the fourth phase.
Firstly, during the lockdown period both mobility and transmission are reduced linearly with one another. See the top right hand chart in the figure above for Germany; this occurred during the period of 10 March through to 19 March (bottom left chart). A sizeable 0.2 reduction in transmission arose from some 50% reduction in mobility, a correlative factor of 0.4. Crucially bringing transmission from 1.25 to close to 1.0.
Secondly, towards the end of the sharp decline in mobility to its lowest point, there is a departure from the linearity of its relationship to transmission. While mobility is shaped as a sharper reverse logistic curve, the transmission curve has a far softer corner towards its low point. Even though mobility is halted, spread at the community level, for example within families and closed groups likely is able to continue.
Thirdly, the rise in mobility post-lockdowns has been gradual but distinct and often starts shortly after the lowest mobility point. For Germany, even at the beginning of April, mobility had started to rise across the board - particularly in transit and retail, and it has yet to slow down since. If the current trajectory continues the country would be back to 'normal' by mid June.
This brings us to the start of the dance. For Germany, transmission has remained low despite the increase in mobility - in other words, more movement does not translate to greater virus spread.
In Germany transmission has been held below 1.0 throughout this dance, despite mobility rising from 40-50% levels up to some 75% of initial baseline - see bottom right chart. In fact at the same mobility level of ~75% just 2 months prior during lockdown transmission was at 1.3 versus only 0.95 or so today.
What we believe is happening is that populations in general are quickly learning how better to manage living with the virus. A more cautious and conscientious population, borne from many weeks of stay at home orders, social distancing, and being pummelled with daily news of the virus' deathly impact, has been conditioned to adjust their behaviours accordingly.
This represents a more cautious approach in everyday practices and interactions. People wash their hands more, they are careful about mingling with others, masks are generally warn and highly populated public spaces avoided. You could say that while physical distancing has been reduced on the back of increased mobility, social distancing i.e. person to person interactions still remain sequestered.
We note the same observation across a range of countries - see figure section later below for reference. Among the largest increases in mobility with little to no effect to transmission are noted in Switzerland, Germany, Austria, and Sweden.
(C) Modelling the fourth phase: hysteresis within transmission change due to mobility
Intuitively, mobility and transmission are linearly correlated. More mobility leads to more person to person interactions and therefore further transmission of the virus. This is a corollary to lockdowns, where rapidly reduced mobility was used to abate the spread; an increase in mobility should therefore have the reverse effect. However, as we have established this is not necessarily the case.
To model this effect we have opted to introduce a hysteresis in the correlation between mobility changes and transmission. While reduced mobility has an accordant reduction effect to transmission, similar to Phase 2 lockdowns, increased mobility has the same effect in the other direction but to a lesser extent. The amount of hysteresis is set to vary from country to country, and further data will be used to improve our assumptions.
An example of the effect can be seen in the top right hand chart in the figure below. Notice how after transmission <1.0, repeated increases and decreases in mobility result in a net transmission decline, albeit only very gradually. For the Netherlands correlation between transmission and mobility is set to 0.2, with 'hysteresis' of 80%, i.e. an 80% overlap between the correlation of down- and up-cycles of mobility. We note our model with the yellow line in the top right chart labelled MOB MODEL CONCAT in the legend.
(D) What this means for future actions to combat COVID-19.
It is likely that too much nervousness about reopening of economies may be unfounded or at least not fully justified. Existing data gathered points to a credible ability for populations to adapt to practices necessary to abate the virus spread. Not a single country has yet to give rise to a subsequent major wave after the initial major wave they encountered.
If second waves do come, they are not likely to be as devastating as the first. There is embedded human conditioning and habit forming evident, and any future transmission from this point onwards would likely bring to bear those learnings from the past. We, as populations, are distinctly changed from the past when the first waves hit.
Efforts to educate and guide populations should really be emphasised. There is in theory tremendous value in this, and it could help avert shutting down the economy and letting livelihoods suffer. While all countries had invested heavily in stay home campaigns, the same if not more should be afforded to inculcate good practices among the population, around permissible interactions, hot spots and particular places to avoid, good hygiene practices, mask wearing practices and so on. The impact is very visible - comparing pre- and post-lockdown periods at equivalent mobilities, achieving non epidemic transmission is a tremendous feat.
We hope this article serves as a helpful guide to foster discussion that could help provide decision makers the confidence to design workable strategies as to how best to reopen their economies. The economic impact of just two short months of lockdown has been devastating. It is unimaginable to think that countries would have to resort to the same strategy again as the primary tool to manage COVID-19 spread in the future. We have been able to show and continue to be hopeful that we populations are indeed resilient and adaptable enough to be guided on how to behave differently in this new normal. Lockdowns can thus hoprfully be left as an emergency cord that can be pulled if and only if all else fails.
Figures section (for reference)
(A) Continued transmission plateau / decline in spite of rising mobilities
Notice on bottom right hand chart how mobility rises distinctly without the same effect to transmission, The top right hand chart also confirms this. Example countries are; Germany, Netherlands, Switzerland, Sweden, US, Austria.
(B) Less conclusive transmission plateau / decline with more gradual rise in mobilities
For these countries mobility has risen since lockdown, however not as dramatically. More time may be required to conclusively determine whether increased mobility may result in future outbreaks where transmission jumps to epidemic levels (significantly above 1.0).
(C) Higher possibility of an additional wave within coming month given aggressive increase in mobility
Mobility is rising significantly, risking additional wave in spite of hysteresis in mobility-transmission relationship, e.g. US (estimates updated 21 May)