COVID-19. Comparison of results in Sweden and the USA. OECD.stat data

OECD.Stat provides weekly mortality numbers and rates for the pandemic period. This data can be used to compare the outcome of various anti-pandemic strategies. In this post, we compare the USA and Sweden, as the two countries have almost opposite policies against COVID-19. The COVID-19-related estimates are presented in two different ways: the absolute number of the deviation from the long-term weekly averages and the rates (% change from average). There are two parameters: total values and those related to COVID-19 (definitions of COVID-19-related cases are likely to vary between the OECD countries). Figure 1 shows the long-term average weekly numbers, the total weekly numbers, and the COVID-19-related weekly numbers. OECD.Stat gives the deviation from the average and we just add the basic weekly values and the deviations to obtain the actual death numbers. As a check of the basic average (seasonally adjusted) values we have calculated the total weakly death numbers for each week between 2020 and the last estimate in 2022 as shown in Figure 1 by a blue curve. Then we shifted this curve 53 months (2020 is the leap year) ahead (the red line) and compared the average values for the same week with a one-year shift. One can see that the average values are all pre-pandemic and this makes it possible to estimate the COVID-19 effect on excess mortality.

In the USA, the COVID-19-related curve is almost always below the total curve, but there are some segments where these two curves almost coincide. The highest mortality was observed at the beginning of 2021 and this peak is fully related to COVID-19. Instructively, the first peak in 2020 was the only one not related to the (fall-winter) seasonal increase. The absolute values are important for a given country but cannon be used from the cross-country comparison.  Figure 2 presents the % change (rate) as obtained from Figure 1. One can see that all peaks in Figure 2 have the same amplitude because the basic average values vary with time periodically. We have calculated the rates using the same average estimates for the total and COVID-19 cases as presented in Figure 1. OECD.stat proposes a different approach and uses the actual death numbers (black curve in Figure 1) instead of the average numbers. As a result, the OECD estimates are not consistent and underestimate the peak amplitudes as the red dotted line shows: the higher the difference between the actual and average numbers the lower the OECD curve. The reason for this approach is not clear, but it definitely leads to the comparison of apples (total rate) and oranges (COVID-19 rate). The COVID-19 numbers are not separated from the total numbers and the same average basis is straightforward to separate the COVID-19 and all other reasons.  The lay public will be rather confused without digging into the real data.

Figure 2. The % change in the total and COVID-19-related mortality in the USA.   

The same procedure was applied to Sweden. Figures 3 and 4 repeat Figures 1 and 2, correspondingly. The total deviation from the average values in Sweden was negative during relatively long periods of time while COVID-19-related deviation numbers are almost always positive. The average curve and the shifted average curve almost coincide with the accuracy of calculations using the OECD data with just one decimal digit. The % change from the average in Figure 4 draws an unexpected picture: the total mortality rate drops below the average in the pandemic conditions! The COVID-19-related rates are positive in both representations – used in this study and proposed by the OECD. This rate has to be non-negative because there can be no negative deviations in the number of COVID-19-related cases. The total numbers include all reasons and thus the deviation in non-COVID-19 cases can be a large negative number since these reasons are behind the average curve itself. 

Overall, the OECD data for Sweden shows that the COVID-19 pandemic was harsh for the country and the excess death rate is high mainly because of the virus itself. At the same time, the pandemic suppressed the severity of all other illnesses taken together and the total death rates fell below the long-term averages despite the pandemic. The driving forces behind such an outstanding finding are not clear and need deeper investigation of various death causes and corresponding definitions.

Figure 3. OECD statistics for Sweden between 2020 and 2022. The total numbers are shown.

Figure 4. The % change in the total and COVID-19-related mortality in Sweden.   

The dynamic view in Figures 1 through 4 does not present the long-term trends in the efficiency of the specific anti-COVID-19 policies during the pandemic. Figures 5 and 6 present the integral behavior in the total excess deaths and those related to COVID-19. The total curves are the ratio of the running sum of all excess deaths and the corresponding sum of all average numbers of deaths from week 14 of 2020. The COVID-19 curves are the ratio of the running sum of the COVID-19 numbers and the corresponding sum of all average numbers of deaths from week 14 of 2020. Figure 5 shows that Sweden has chosen a better long-term approach that guarantees lower total life losses over the pandemic period. In August 2022, the integral excess death rate during the pandemic was 23% in the USA and less than 4% in Sweden. At the same time, the COVID-19 results are closer for both countries as Figure 6 demonstrates. The advantage of Sweden is not in the treatment of COVID-19 as such (here Germany and Australia are much better) but in the reduction of other health risks.  

Figure 5. Integral total excess death rate since week 14, 2020.

Figure 6. Integral COVID-19-related excess death rate since 2020.