WORLDVIEW: Why Covid-19 fatality rates are all wrong

The Covid-19 crisis is a fast-moving target, which makes it hard to study. One thing seems clear, however: our understanding of how lethal the virus is has a long way to go..Initial fatality rates from the World Health Organisation (WHO) estimated that around 3.4% of people infected by the novel coronavirus who develop Covid-19 would die. Since then, we have seen a very wide range of national fatality rates..The table below shows the current best data for the world overall and the 15 countries with the highest number of reported cases (please note that all the numbers presented here are current at the time of writing, but obviously will change daily)..___STEADY_PAYWALL___.As you can see, fatality rates vary dramatically, and there is no clear pattern – populous countries may have higher or lower fatality rates than less populous ones, and there's no clear relationship between testing and fatality rates either..CountryTotal CasesTotal DeathsFatality RateDeaths/1M popTests/1m PopWorld3,077,166211,9646.927USA1,010,50756,8035.617217,211Spain229,42223,52110.350328,779Italy199,41426,97713.544629,600France165,84223,29314.03577,103Germany158,7586,1263.97324,738UK157,14921,09213.431110,605Turkey112,261 2,9002.63410,895Russia93,558 8670.9621,511Iran91,4725,8066.3695,147China82,836 4,6335.63Brazil67,4464,6036.8221,597Canada48,5002,7075.67219,440Belgium47,3347,33115.563318,468Netherlands38,2454,51811.826411,319India29,4519393.21519.What this data highlights is how uncertain any guesses about the real fatality rate may be. There are two reasons for this. First, we don't know how many cases there are. Some scientists think that there are many hundreds of times more cases than we have recorded, which would mean a much lower Covid-19 fatality rate..For example, a much-publicised study led by Stanford scholars suggested that the real fatality rate of Covid-19 in Santa Clara is as low as 0.12%. But, as scholars have pointed out, if the fatality rate of Covid-19 is 0.12%, that would imply that in New York City, where 15,400 people have died, there would have had to be 12.5 million people infected. The entire population of NYC is only 8.3 million..Indeed, if we assume that 100% of New Yorkers had Covid-19, we'd still have a fatality rate greater than 0.12%. And it's clearly not true that everyone in the city is coronavirus-positive. Testing data shows that..This simple example highlights how limited our current understanding of case numbers is. We are certainly undercounting cases, but at this point, we just don't know how much we're undercounting. And any guesses must be reasonable – they have to explain deaths in many places and cannot ignore data from places that have been hit hard by the virus. In addition, we have to remember that fatality rates will probably be different in different places due to population differences, lifestyle differences, and differences in the quality of healthcare..The second reason for our inability to assess the fatality rate is that we don't know how many people have died from Covid-19. Even in a developed country like the UK, we know that deaths are being undercounted because the numbers only include people in England and Northern Ireland who die in hospital, and Scottish hospital and home deaths. With many deaths at home and in old age homes, the UK numbers are inevitably an undercount..The extent of the problem is brilliantly illustrated in this piece from The Economist (free to read), which looks at excess deaths – that is, recorded deaths that are above the number of deaths we'd expect to see in a normal year..The figures are stark. In a normal late April, we'd expect about 50,000 deaths a week in the countries the data covers. We are experiencing around 73,000 – meaning around 23,000 extra deaths a week. Recorded Covid-19 deaths explain only around 60% of those extra deaths, meaning that there are thousands of deaths that we have not accounted for. According to The Economist, those are likely to be mostly Covid-19 deaths that were not counted, probably due to a lack of testing or death recording quirks. This is especially true since auto deaths and workplace accidents are down (due to lockdown) and recession-associated deaths, such as suicides, usually only peak many months after the economy sours..In short, our estimates about how lethal Covid-19 is are wrong. We are missing lots of cases – although it's not yet clear how many – and we are missing a chunk of deaths. If you see anyone saying "It's just a bad flu" be very suspicious of their motives. No one has any idea how deadly this disease is yet. We need a comprehensive death count and much better prevalence data before we can make a guess. And keep in mind that places like Italy, New York City, Wuhan, and Spain highlight just how deadly Covid-19 can be when there is an uncontrolled outbreak in a densely populated area. South Africa's townships – overcrowded and home to many vulnerable people – could become death camps if coronavirus is allowed to rampage through them.