Is Iceland clear of Sars-CoV-2 ?


In this study, we propose a model of the Covid-19 epidemics in Iceland calibrated on official “Confirmed Cases” and official death record curves (as of May 3, 2020 there were 1799 cases and 10 deaths). The aim is here to make the prior assumption that these data are correct and observe some of the outcomes. An immediate consequence of such hypothesis is the model obviously adopts an IFR of 1 in 180 or 0.56%, since the current outbreak of the epidemic is now almost over. An excellent match of recorded Confirmed Cases history was obtained with a R schedule starting from R0=1.78 in February, declining to 1.2 in March. The time of introduction was found to be February 17. The model predicts a last occurrence of the disease this week (on May 10, 2020). Based on the very low initial value for R0 obtained from history-match we conclude that early containment has certainly played a crucial role in controlling the disease, but country-specific if not unknown factors must have helped in keeping the epidemic in almost total control.

Model match curves

Figure 1: new Cases and Death incidence. Data in red are official Confirmed Cases. Model is solid line. Death incidence is statistically too low to be accurately reproduced, but we give a fractional representation matching cumulative deaths (10 fatalities in total).
Figure 2: cumulative case match: model and Confirmed Cases data are in perfect agreement
Figure 3: death curve match. Note that 99.4% of the Iceland population is still susceptible to the virus as of May,4 (in this scenario).


Seroprevalence is expected very low, below 1%. If the true IFR is indeed around 0.56%, we expect that a serology survey would not return more than 1% with antibodies – below detectable rate in fact. Serology testing are highly recommended though. A higher number of sero-positives would infirm the central assumption made in this study, and would imply a higher group immunity for the country. We have not heard of any serology survey being reported or planned yet.

Figure 4 : simulated seroprevalence in Iceland, assuming an IFR of 0.56%.

Model parameters – How match was obtained

Match was obtained from adjusting time of introduction and R as detailed in Tables below.

The R schedule can be related to the epidemics history in Iceland on

Time of introdutionto17-02-20 
Basic reproduction numberRo1.78 
Infectious period1/sigma4.5days
Transmission coefbeta0.3961/days
Time infection – deathPsi17days
Prob of dying with severe diseasethet0.14 
Proportion at risk of severe diseaserho                       0.040 
Basic parameters. R0 and time of introduction were obtained from curve match
8209-05-20Forecast: 1.5
8916-05-20Forecast: 1.5
R schedule with time, obtained from history-match of Confirmed Case curve. Forecast values were to demonstrate resurgence of epidemics in case distancing policies were released a FEW DAYS too early


An unusually low initial R0

An excellent match of recorded Confirmed Cases history was obtained with a R schedule starting from R0=1.78 in February, declining to 1.2 in March. Because the time of introduction was found to be February 17, and it takes 17 days to materialise in fatalities, we conclude that R0 reflects the reproduction number before or just when any policy action or voluntary distancing measures were taken. Cultural or country-specific if not unknown factors certainly have played a key role in keeping the epidemic in almost total control, or at least making isolation policies feasible. For example, Iceland is a low population density country, and many were indeed not susceptible to be in contact with the disease. First case was discovered on 28 February, and strict case identification and isolation policies implemented immediately.

Comparison of IFR with previous country results

An IFR of 0.56% or 1 fatality for 180 infected was found for Iceland. This ratio is only based on the assumption that the country managed to account, exactly, for all its infected cases. This is unlikely, since asymptomatic Covid-19 cases are often reported and Iceland, despite its remarkable performance on this matter, had “only” tested 5% of its population by end of March ( It results that an IFR of 0.56% is a highest possible value for the Covid-19 epidemics, in Iceland at least.

This only apparently contradicts an IFR of 0.9% found for New York City in a study of serology results we published on April, 23. First, fatality rate in a dense urban city with hospital system under strain is expected higher. Second Second, low death number in Iceland may not be statistically representative. And, most importantly, third: one should not pre-assume that the Covid-19 epidemics must result in same epidemic dynamic parameters in all countries (a simple situation where this may happen is if different virus strains result in different clinical effects).

Other previous serology survey interpretations by CovModel have returned an IFR of 0.21% for Denmark (and an extremely low number of 0.03% for Scotland and Finland with low level of trust).

As of time of writing we retain an IFR range of 0.21% to 0.9% for the Covid-19 epidemics.

Has Iceland Suppressed Sars-Cov-2 ?

Both latest statistics and our numerical modelling converge to agree that the “spring” outbreak is almost over in Iceland, the country being one of the only in the world having succeeded to suppress the virus by mid May. Our model predicts the last case on May, 10. Though, we would not advise the country announces this to the world yet.

Assuming an R factor back to R=1.5 on 9 May, epidemic relapse is immediate. A high IFR and extremely low mortality, reflecting a total control of the epidemics, means a paradox that very few in the population have acquired immunity and the country is now very vulnerable to a resurgence if or when international travel resumes.

Figure : resurgence of epidemics assuming R=1.5 from 9 May 2020 (date when only one infected patient theoretically remains in model).

Photo by Rudolf Kirchner from Pexels