A new surge in coronavirus cases across the world is threatening to stall the efforts made to revive the global economy. What is the likely path of the pandemic following this surge? Is the talk about a second wave warranted and what can be done about it? In this article, we assess the evolution of the situation in the United States (state by state). Here are the main takeaways of this analysis:
- The COVID-19 pandemic has moved from a concentrated outbreak in a few hotspots in March-April to a widespread nationwide circulation. The main reason for that is probably that contrary to China where the first infected areas – Wuhan and the Hubei province – have been locked-down and isolated from the rest of the country early on -, in the United States lock-downs were imposed with a delay and were removed before the virus was contained. In addition, with a few exceptions people could still freely move between States.
- The surge in new cases has been tamed in some of the states that experienced it (Florida, Texas, California) but the test-based incidence ratio remains elevated in those states. This means that the policy measures taken by some of these states are still insufficient to warrant a containment and significant reduction in the contagious force of the virus.
- In the states that managed to bring down significantly their incidence ratio over the March-May period, the surge in new cases over June-July coincided with a lower confirmed case fatality rate. Overall, this is a positive development although it cannot be taken for granted going that this disconnect between new cases and new deaths will continue going forward.
Let us start by assessing the evolution in the daily number of infections and deaths averaged over 7 rolling days.
Concentrating on the 14 most important states, which account for two thirds (67%) of US GDP and an almost equivalent (63%) share of US total population, we can see that the surge in COVID-19 infections that has been witnessed over the last two months in some of these states has reached a plateau in the second half of July in most of them (California, Florida, Georgia, Ohio, Texas, North Carolina). In other important states like Illinois, Massachusetts, Pennsylvania and Michigan the surge in virus cases does not show signs of abating but the case fatality ratio has been moving downward. It could indicate either that the medical response to the pandemic has improved or that the new cases are younger on average and therefore less exposed to severe forms of the disease. All things being equal, the change in the number of new deaths lags the change of new infections by ten to fifteen days. Therefore in states like Florida and Texas which managed to stabilise the number of new daily cases – something which by itself does not give much reason to rejoice as the objective is to reduce the number of new infections -, we could still witness over the coming days an increase in the number of deaths before an eventual peak and subsequent decline.
Predicting the short-term dynamic by looking at the test incidence ratio
To get a clearer view of the epidemic’s dynamic, we have to look at the incidence ratio which measures the number of new positive cases related to the total number of new tests. We can see that the incidence continues to move up in some states (Alaska, Alabama, Kansas, Oregon, Tennessee, Idaho) while it has receded to lows in other states which previously experienced stronger incidence ratios earlier (New York, Illinois, Pennsylvania, Virginia, Massachusetts). In between, there are states which managed to stabilise their incidence ratios albeit at high levels which tells us that it will take a while before the surge is curbed in those states.
Again we zoom on the 14 largest US states in terms of GDP. The tests-based incidence ratio remains very high in Florida and Georgia and Texas than in other economically and demographically important US states. California does better in level terms but not in terms of its ability to curb the surge as the incidence ratio remains almost stable since the start of July. Illinois, Michigan, Ohio, Pennsylvania showed upticks from lower levels of incidence. The state of New York stands out ut with a particularly low incidence, hovering around 1%.
This warrants a closer examination. Thanks to the logarithmic scale, which is better suited to observe relative growth between two variables, we can see on the chart below that the ramp-up in the incidence ratio (bars in colour) in Florida is related to a much higher increase in the number of new positive cases (dark line) than the increase in the number of tests even though the latter has been following an exponential growth in most US states. Basic statistical theory tells that the more individuals from a population we add to a sample the more accurate we can estimate a metric – or ratio of metrics – that relates to the whole population. Indeed, the law of large numbers says that if you take samples of larger and larger size from any population, then the mean of the sampling distribution tends to get closer and closer to the true population mean.
Let us not forget that the incidence ratio is not constant, as can be seen from the chart. Its evolution is related to changes in the effective reproduction ratio which measures the effective contagious force of the virus in a given setting. Even taking into account the stochastic nature of the incidence ratio, this does not change the validity of the (weak) law of large numbers. The more new tests are performed the better the test incidence ratio reflects the effective incidence of the disease – i.e. its contagious force – at a given point in time and space.
Assessing the death toll and the adequacy of the response to the pandemic state by state.
In the snapshot above we report and compare four indicators for every US state: the state’s share of US GDP, its share of US population, it share of COVID-19 cases (in % of confirmed cases in the US) and, finally, its share of COVID-19 deaths (in % of total deaths attributed to the pandemic nationwide). What we observe is that some States have been afflicted by a share of cases and deaths far above their economic and demographic weight, while others have managed to contain the spread of the virus and the related death toll. We can also see that depending on the state we consider, its share of total cases is higher or lower than its share of total deaths which means either that the demographic characteristics are different (e.g. the share of population aged 70 and over) and/or that the adequacy and efficiency of the local authorities response to the pandemic can differ substantially from one state to another. Interested readers can make their own analysis and draw their own conclusions based on this snapshot.
Perhaps the ultimate indicator of the pandemic’s impact is the number of deaths inflicted so far by the virus per 10 000 people. So far, New England and the Great Lakes regions, alongside some southern states like Louisiana, Mississippi and Arizona have taken more than their fair share of the virus-related death toll. But things could still change in the future according to the dynamic we described in this focus.