The WHO’s Erroneous Risk Assessment

by Mikko Paunio, MD (University of Helsinki), MHS (Johns Hopkins Bloomberg School of Public Health), adjunct professor (general epidemiology) at the University of Helsinki

At the heart of the WHO’s risk assessment, at the start of the pandemic, was the assumption that only 1% of those infected would show no symptoms.¹²

The claim that few of the infections would be symptomless – and thus that everyone would become ill and that many people would die – paved the way to weeks of horror stories on the BBC, CNN, and in the New York Times, the Washington Post and the Guardian. And even more sober outlets like the Financial Times and the Economist followed suit, with little by way of analysis of what was actually known. In fact, the WHO’s claim was quickly rebutted by a member of its own Infectious Disease Catastrophe Committee,¹ but too late to prevent panic spreading. The result was a lockdown across much of the world, the collateral damage from which will do far more harm than the virus.

A major serological survey from Spain³ now shows how wrong the initial WHO risk assessment was.

Spanish serosurvey

The first wave of the pandemic caused around 27,000 deaths in Spain, a figure so small in comparison with the size of the population that it will probably have little effect on the overall 2020 mortality rate. Those who succumbed were mostly frail old people in nursing homes and had only short life expectancy.

The WHO claimed at the end of February that the virus might kill up to 4% of those infected.¹ I have now tabulated data for the case fatality rate (CFR), infection fatality rate (IFR) and the mortality in Spain (Table 1). The rightmost column shows the age-specific COVID-19 mortality. Even amongst the most vulnerable – people over the age of 90 – the survival rate is above 99%. Below 60 years of age, the death rate so far – with the first wave of the pandemic about to come to an end – is 3.6 per 100,000 population.

Symptomless infections

The Spanish serological survey was based on a random sample of 102,803 individuals, although only 60,983 individuals were traced and agreed to be tested. A quick serological test determined the presence of antibodies in the cohort, and the study authors also asked them about respiratory symptoms during the epidemic period. It turned out that around 5% had SARS-CoV-2 antibodies.

Among those surveyed, 257 had previously tested positive via a PCR test, and 87% of these had antibodies. These few people were the only ones who knew for sure they were already infected, though 14% thought they had COVID-19 like symptoms. Based on this PCR data it can be deduced that the quick serological test had 87% sensitivity.

Those without any respiratory symptoms (40,202) and those with mild flu-like symptoms had an antibody prevalence rate of 3.1%. This means there were 1,629 individuals who had become infected without knowing it. When these 1,629 people are divided by the number of former PCR-positive individuals, we get a ratio of 6.6:1. After correcting for the sensitivity of the test, we get a ratio of 7.6:1. This tells us that for each infection established by PCR there are 7.6 people who are infected but do not know it.

Someone who tests positive for the virus under PCR may never develop symptoms. Thus the ratio of symptomless to symptomatic infections may be even higher than 7.6. When you look at Table 1 you can see that there are almost nine times more serologically established infections than there are confirmed COVID-19 cases. The 60,000 people tested in Spain will be re-tested twice in the coming month, so we will soon have a better idea on this question. It is also worth noting that not all who lack antibodies lack immunity.⁴

It is also worth noting a study of an outbreak of the virus on a cruise ship on a voyage to Antarctica. All passengers and crew were tested, and found that 81% of those infected were non-symptomatic.⁵ And even then, the non-symptomatic/symptomatic ratio of 4.4 reported will be an underestimate, since the PCR test used would not pick up all those who were infected.

Mortality

Next I compared the mortality rate you would expect based on the WHO assumption of few asymptomatic infections to reality: the number of infections in Spain based on the serological data, both on relative and absolute scales (Table 2).

Back in March, and for several weeks, the mainstream media were reporting COVID-19 case fatality rates of around 10% in Spain and Italy. But the reality of large numbers of asymptomatic infections means that the infection mortality rate is only one tenth of this figure. When the case fatality rates publicised by the WHO are compared to the true rates, one finds that the WHO overstated elevenfold the risk to under-60s, although that risk was very small. For older age groups, where the true risk was much higher, the overstatement was less – but especially significant in absolute scale. For example, for those aged 80–89 years, the IFR was 9%, or roughly a third of the level claimed by the WHO.

An infection fatality rate (IFR) of 1% is 10 times higher than that of seasonal flu. This figure, derived from the Spanish data, has been widely interpreted in the media to be generalisable to other countries, but this is not the case. In Southern Europe there is much more contact between young and old people than in Northern Europe, increasing the number exposed there. Also the population of Spain is relatively elderly, and wide-spectrum antibiotic use is rampant, potentially hampering treatment of secondary bacterial pneumonias and other conditions. These factors may well explain the high IFR in Spain.

Iceland serosurvey

The suggestion that the IFRs of each country or even subpopulations within societies must be considered one by one becomes clear and confirmed by recent data from Iceland (Table 3). Currently, it has 1,804 confirmed COVID-19 cases and just ten deaths. There is nobody in hospital and, to all intents and purposes, the epidemic is over. Interestingly, among those who contracted COVID-19, all eight of those aged over 90 survived.⁶ The very low level of deaths means that it is likely that the IFR will be very low too. The ratio of asymptomatic infections to confirmed cases can be as low as five and would still imply an IFR similar to that of seasonal flu.

The CFR was 18 times higher in Spain than Iceland, with the notable differences seen across the age groups, (80–89 years is an exception), which tells us that the older age distribution of COVID-19 cases in Spain does not explain all the differences. It is possible that aggressive PCR-screening of the population in Iceland might have skewed confirmed COVID-19 cases towards those with a milder natural history than in Spain; i.e. Iceland might have diagnosed more mild cases than Spain. If Iceland had the same age distribution of COVID-19 cases as Spain, there would have been 86 deaths, not ten. If Iceland had both Spain’s COVID-19 age distribution and its CFR, the death toll would have risen to 178. These remarkable differences show that it is not possible to generalise the widely disseminated one percent SARS-CoV-2 infection fatality rate of Spain to other countries. Iceland seems to have able to protect its elderly much better than Spain.

Other serosurveys

An even less worrying picture emerges from the serosurvey of LA County, California published in JAMA⁷ Here, on around 10^th April 2020, there were 44 times more serologically established infections than confirmed COVID-19 cases – a remarkable contrast to the WHO claim that only 1% of infections were asymptomatic. The researchers involved have said that there was a selection bias in their study, and it was likely that symptomless infections were under-represented. If one calculates IFR of SARS-CoV from this material by assuming that there were around 1,220 cumulative deaths in early May in La County, one gets an IFR of 0.33%.

Conclusions

The outbreaks of COVID-19 we have seen across the world have remained exponential for only a very short period.⁸ We do not fully understand why, but it is possible that non-SARS coronaviruses, which account for around 20% of common colds, might have given cell-mediated immunity to some sections of the population.⁹ Either way, the global impact of the virus remains small, despite what the media would have you think.

Table 1. Key metrics calculated from Spanish data.
Age	Deaths	Confirmed cases	Antibody prevalence (%)	Real number of SARS-CoV-2 infections in population	Population (m)	Case fatality rate (%)	Infection fatality rate (%)	Corrected infection fatality rate (%)	COVID-19 death rate in Spain
<60	1,269	131,688	4.2	1,470,000	35.0	1.0	0.09	0.08	3.6/100,000
60-69	2,376	39,806	6.0	312,000	5.2	6.0	0.76	0.66	4.6/10,000
70-79	6,696	38,443	6.5	253,500	3.9	17.4	2.64	2.30	1.7/1000
80-89	11,151	42,805	5.4	124,200	2.3	26.0	8.98	7.81	4.8/1000
90+	5,481	19,903	5.8	34,800	0.6	27.5	15.75	13.70	9.1/1000
Total	26,973	272,645	5.0	2,194,000*	47.0	9.9	1.22	1.06	5.7/10,000

* 5% × 47,000,000 = 2,350,000; figure differs as age-specific serology prevalence means are not weighted

Table 2. WHO overstatement of risk
	Risk assessment (%)		Exagerration
Age	Per WHO	Per serological data	Relative (×)	Absolute (%)
<60	1.0	0.08	11.1	0.80
60–69	6.0	0.66	7.9	5.34
70–79	17.4	2.30	6.6	15.1
80–89	26.0	7.81	2.9	18.9
90+	27.5	13.70	1.7	13.8
Total	9.9	1.06	8.2	8.84

Table 3. COVID-19 case fatality rate (CFR) in Spain and Iceland and CFR ratio Spain vs. Iceland and expected number of deaths in Iceland assuming age distribution of COVID-19 of Spain and CFR of Iceland and expected number of deaths in Iceland assuming age distribution and CFR of Spain.
	CFR					COVID-19 age distribution		Iceland Covid-19 cases assuming same age distribution of cases as Spain	Expected deaths in Iceland assuming:
Age group	Spain		Iceland		Ratio	Spain	Iceland		age distribution of COVID-19 of Spain and CFR of Iceland	age distribution and CFR of Spain
<60	1269/131688	0.96	1*/1501	0.067	14.3	48.3	83.2	871	0.6	8.4
60-69	2376/39806	5.97	2/215	0.93	6.4	14.6	11.9	263	2.4	15.7
70-79	6696/38443	17.42	3/62	4.83	3.6	14.1	3.4	254	12.3	44.2
80-89	11151/42805	26.05	4/16	25.00	1.0	15.7	0.9	283	70.8	73.7
90+	5481/19903	27.53	0/8	0.00	N/A	7.3	0.4	132	0	36.3
Total	26973/272645	9.89	10/1804	0.55	18.0	100	99.8	1803	86.1	178.3