Reposted from Dr. Judith Curry’s Local weather And so on.
By Nic Lewis
I believed it was time for an replace of my authentic evaluation of 28 June 2020. As I wrote then, the course of the COVID-19 pandemic in Sweden is of nice curiosity, as it’s one in every of only a few superior nations the place no lockdown order that closely restricted individuals’s actions and different primary freedoms was imposed.
Sadly, a number of the touch upon how the COVID-19 epidemic has developed in Sweden has been ill-informed. Certainly, a shadowy group of teachers, opinion leaders, researchers and others who’re upset about Sweden’s technique and are actively in search of to affect it has been unmasked. They’ve been coordinating efforts to criticize media protection of Sweden’s technique and to break each the picture of Sweden overseas and the popularity of people who work on this discipline.
I current right here up to date plots of weekly new circumstances and deaths, with accompanying feedback.
- Regardless of Swedish Covid circumstances falling to low ranges in the summertime, they resurged within the autumn
- This second wave, which was very probably a seasonal impact, now seems previous its peak
- Extra deaths in East Sweden had been excessive within the first wave and low within the second; for South Sweden the other is true. This means that inhabitants immunity and/or the remaining variety of frail previous individuals are key elements within the severity of the second wave.
- Extra deaths in Sweden to finish 2020 had been modest, significantly for 2019 (when deaths had been abnormally low) and 2020 mixed. They look like a lot decrease relative to the inhabitants than in England, regardless of far harsher restrictions being imposed there.
- Solely 3% of recorded 2020 COVID-19 deaths in Sweden had been of individuals aged below 60
Total improvement of the epidemic
Determine 1 exhibits the general image for confirmed weekly complete new COVID-19 circumstances, intensive care admissions and deaths in Sweden, as much as knowledge launched on 9 February 2021. The standards for testing had been widened throughout the early months, so case numbers as much as June 2020 aren’t comparable with subsequent ones. Weekly new circumstances have been divided by 50 so as to make their scale akin to that for ICU admissions and deaths. Dying numbers for the 2 remaining weeks can be noticeably understated as a consequence of delays in dying registrations.
Fig. 1 Complete weekly COVID-19 confirmed circumstances, intensive care admissions and deaths in Sweden
In late summer season 2020 it appeared as if the epidemic had burnt itself out, nevertheless a powerful second wave developed throughout September to December. Though begin of faculty and college time period, together with extra relaxed behaviour, might have began the second wave off, over the interval as a complete the first driver was nearly actually a seasonal enhance within the virus’s transmission and therefore copy quantity. Research that indicated an absence of considerable seasonality in transmission  have been confirmed incorrect.
Evaluation by age group
The altering age composition of recent circumstances over time is proven in Determine 2. Case numbers earlier than and after June 2020 aren’t comparable due to the foremost widening of testing throughout June 2020. Nevertheless, it’s clear that the second wave has been dominated by infections of individuals aged 10 to 59 years.
Fig. 2 Weekly COVID-19 confirmed circumstances by age group in Sweden
After falling to very low ranges in late July 2020, weekly COVID-19 recorded deaths rose strongly from late October on, throughout all age teams (Determine 3). The information present the variety of individuals with confirmed COVID-19 who died, no matter the reason for dying. In complete, about 50% of deaths occurred as much as and after 30 September 2020, that’s within the first wave and within the second wave (which is, nevertheless, not over but). Through the second wave, a barely larger proportion of deaths have been of individuals aged 80+ years than within the first wave.
Fig. 3 Weekly COVID-19 recorded deaths by age group in Sweden
I flip now to regional evaluation. Determine Four exhibits weekly confirmed new circumstances for every of the 21 areas in Sweden. Though widening of testing (primarily within the second quarter of 2020) different between areas, it’s evident that Stockholm and Västra Götaland, which dominated circumstances throughout the first wave, had been additionally two of the three areas dominating the second wave, with Stockholm area main each waves. Nevertheless, whereas Skåne had comparatively few first wave circumstances, it broadly matched Stockholm within the second wave, albeit with a delay. Instances have fallen fairly sharply in nearly all areas because the flip of the 12 months.
Fig. 4 Weekly COVID-19 confirmed circumstances by area in Sweden
Areas have various populations, so confirmed circumstances per 100,000 head of inhabitants give a greater image of relative illness incidence (Determine 5). There may be negligible correlation between the areas that had the very best incidence of COVID-19 circumstances throughout the first wave (together with or excluding June to August) and the post-September 2020 interval. Within the absence of rising inhabitants immunity having an impact, one would possibly anticipate that in these areas wherein the virus unfold most simply previous to September 2020 (by which period it was nicely ensconced in all areas) – as an illustration, as a consequence of larger urbanisation – it will even have unfold most simply within the second wave, within the absence of adjustments in different elements. A scarcity of correlation between circumstances within the first and second waves is per larger inhabitants immunity in these areas that had been tougher hit within the first wave counteracting, throughout the second wave, the larger ease with which infections unfold there when inhabitants immunity was low.
Fig. 5 Weekly COVID-19 confirmed circumstances per 100,000 head of inhabitants by area in Sweden
As for circumstances, it’s troublesome to discern an apparent relationship throughout areas between COVID-19 deaths per 100,000 individuals within the first and the second waves, and the correlation between them is negligible. The non-identity between recorded COVID-19 deaths and people really brought on by the illness could also be one motive for this. A considerably clearer image comes from analyzing weekly extra deaths in geographical super-regions, as proven in a current Swedish report.
Fig. 6 Weekly deaths (purple line) as much as week Three of 2021 in Sweden in contrast with the anticipated regular dying toll (strong inexperienced line) and its 95% confidence interval (dashed inexperienced line)
Determine 6 exhibits the place for Sweden as a complete. Information go as much as week 3 2021; knowledge for newer weeks are incomplete. Peak extra deaths had been larger within the first wave than within the second wave, the other relationship to that for recorded COVID-19 deaths. Whereas this probably partly displays a component of undercounting of deaths brought on by COVID-19 on the peak of the primary wave, it seems to be primarily due extra to a significantly bigger over counting of COVID-19 deaths all through the second wave. Whereas the second wave shouldn’t be over but, it does seem that extra deaths have peaked.
Determine 7 exhibits deaths for East Sweden, the inhabitants of which is dominated by Stockholm area. Extra deaths within the first COVID-19 wave had been additional above regular than for Sweden as a complete, however extra deaths throughout the second wave peaked at a degree not a lot above that within the 2017/18 flu and pneumonia season, and fell again inside the 95% confidence interval by the top of 2020 (and to shut to regular for Stockholm area alone).
Fig. 7 As Determine 6 however for East Sweden (Stockholm, Uppsala, Södermanland, Östergötland, Örebro, Västmanland)
Nevertheless, in Southern Sweden, the image is kind of completely different (Determine 8), with the second wave of COVID-19 extra deaths being significantly bigger than the primary, which was smaller than within the 2017/18 flu season.
Fig. 8 As Determine 6 however for South Sweden (Jönköping, Kronoberg, Kalmar, Gotland, Blekinge, Skåne, Halland, Västra Götaland)
The inhabitants of South Sweden is dominated by that of Västra Götaland within the north west and Skåne within the south, which include respectively Sweden’s second and third largest cities (Gothenburg and Malmö). In southernmost Sweden, the primary wave barely breached the higher sure of the 95% confidence interval for regular deaths, whereas peak extra deaths within the second wave had been 3 times that degree (Determine 9). Within the the rest of South Sweden, extra deaths within the second peaked at a broadly comparable degree to within the first wave. The identical is true for North Sweden, which is comparatively sparsely populated and has few sizeable cities.
Fig. 9 As Determine Eight however for southernmost Sweden (Blekinge, Skåne) alone
For my part, the sample of extra deaths in waves one and two in Stockholm-dominated East Sweden, in comparison with that in different elements of Sweden, means that a lot of the pool of individuals in East Sweden susceptible to dying from COVID-19 had already succumbed by the top of wave one. Then again, though the extent of earlier infections and therefore inhabitants immunity in Stockholm area on the finish of wave one was greater than ample to inhibit massive scale unfold of COVID-19 throughout the summer season, on the degree of inhabitants mixing occurring then, with hindsight it was at that stage clearly inadequate to offer herd immunity within the winter, when transmission is larger, inflicting each the virus’s copy quantity (R0) and the herd immunity threshold to rise.
Though it’s too early to make certain, at current it seems that inhabitants immunity in each Stockholm area and Sweden as a complete is now ample to forestall large-scale COVID-19 epidemic development even in winter, no less than on the present degree of inhabitants mixing. Nevertheless, there’s a caveat in that the B.1.1.77 (UK-discovered) variant, which is estimated to be about one-third extra transmissible – and therefore sooner rising – solely turned obvious in Sweden throughout December. Whereas current in 35% of all Swedish sequenced genomes throughout the first three weeks of 2021, it’s not but dominant, so transmission could be anticipated to rise because it achieves dominance over the following two or three months.
Complete Swedish deaths as a consequence of COVID-19
Sweden had 10,082 deaths with confirmed COVID-19 an infection for the 53 reporting weeks of 2020, ending Three January 2021, together with these reported subsequently. On one other measure, there have been 9,432 deaths. Solely 0.9% of deaths had been of individuals below 50 years previous, and solely 3% had been of below 60 12 months olds. Folks over 70 years previous accounted for over 91% of COVID-19 deaths.
The definition of COVID-19 deaths imposed by the WHO is more likely to over rely deaths brought on by COVID-19, since the place there are a number of causes contributing to a dying clinically-compatible with COVID-19 (usually respiratory failure or acute respiratory misery syndrome) will probably be recorded as a dying as a consequence of COVID-19 the place SARS-CoV-2 an infection is confirmed or suspected, even when COVID-19 shouldn’t be thought of to be the principle explanation for dying. Furthermore, some international locations have adopted much more over-pessimistic definitions of COVID-19 deaths. Others have probably undercounted COVID-19 deaths. And in lots of international locations some deaths brought on by COVID-19 in the beginning of the epidemic had been most likely not recognised as being such. Due to this fact, extra mortality over a standard degree is often considered the very best measure of deaths as a consequence of COVID-19.
Extra mortality is primarily affected by the severity of respiratory illness (primarily influenza) winter seasons. A extreme flu season, which can be brought on by a brand new influenza virus pressure, leads to many extra very frail unhealthy previous individuals dying than a gentle flu season. Extreme flu seasons might happen in pairs in adjoining years, as an illustration as a consequence of widespread vulnerability to a brand new pressure.
It follows that, different issues being equal, fewer deaths will are likely to happen in a flu season that follows a extreme flu season, much more so the place that’s the second of a pair of extreme flu seasons, as there can be fewer than regular very previous and frail individuals alive. Correspondingly, extra deaths than common will are likely to happen in a season following a number of gentle flu seasons. This is called the “dry tinder” impact. It has been proven, for instance, that throughout 32 European international locations there’s a vital destructive correlation (–0.63) between flu depth in winter 2018/19 and 2019/20 mixed and the COVID-19 mortality fee within the first wave (Determine 10).
Sweden had unusually low mortality in 2019, which is basically a mirrored image of gentle late 2018/19 and early 2019/20 flu seasons (the early and late a part of every flu season falling in several calendar years). It thus had larger than common “dry tinder” when the COVID-19 epidemic began.
An in depth evaluation by a Danish researcher of the affect of “dry tinder” in Sweden, revealed by a US financial analysis institute, concluded that it accounted for a lot of COVID-19 deaths.
Equally, an evaluation by an economics researcher at a US college, which checked out 15 elements other than severity of presidency interventions that may clarify the upper COVID-19 deaths in Sweden than in different Nordic international locations, concluded that the “dry tinder” issue was essentially the most vital one. That paper additionally thought of it believable that Sweden’s lighter authorities interventions accounted for less than a small a part of Sweden’s larger Covid dying fee than in different Nordic international locations.
Fig.10 Dying fee from COVID-19 as much as 10 June 2020 by complete 2-year flu depth for 32 international locations. The R2 of 0.396 (r=−0.63) is critical on the 1% degree. A copy of Determine 1 in reference 10.
A good estimate of extra deaths in Sweden brought on by COVID-19 in 2020 ought to mirror the unusually massive variety of very previous and frail individuals who survived 2019. That may be carried out by evaluating precise and predicted deaths for 2019 and 2020 mixed.
I calculated extra mortality in Sweden for annually from 2000 on, by 5-year age group and intercourse, because the distinction between precise mortality and regular mortality predicted by a regression match to precise mortality charges over both 2000–2018 or 2009–2018, after which used inhabitants knowledge to derive the anticipated variety of deaths in a standard 12 months for 2019 and 2020. Mortality charges have been declining since 2000 in all age teams, though extra slowly during the last decade. Nevertheless, the impact on total mortality of declining mortality charges at every age is partially counteracted by the rising common age of the inhabitants.
When estimating regular mortality from traits in mortality over ,alternatively, 2000–2018 or 2009–2018, the surplus mixed 2019 and 2020 deaths had been respectively 4,500 or 2,100, representing 0.043% or 0.020% of the mid-2020 Swedish inhabitants. For 2020 by itself, calculated extra deaths are 6,900 or 5,600 for the 2 regression bases (0.066% or 0.054% of the mid-2020 Swedish inhabitants).
Extra deaths for 2019 and 2020 mixed had been largely of males aged 65–79 and (to a lesser extent) aged 80–89 and 90+. Extra deaths of ladies had been below 30% of these of males, primarily based on mortality predicted by regressing over 2000-2018, and had been really destructive primarily based on regressing mortality over 2009-2018. On each regression bases and for each sexes, 2019 plus 2020 deaths of below 65 12 months olds had been decrease than predicted. And common total mortality for 2019 and 2020 mixed was decrease than for any earlier 12 months this century (and really most likely ever).
A extra detailed evaluation of Swedish mortality in 2020, however which used incomplete deaths knowledge, was revealed a month in the past by the blogger swdevperestroika; it’s nicely value studying. That article made comparable factors, and reached comparable conclusions, to my very own evaluation.
Comparability of Swedish and English extra deaths
I utilized the same evaluation methodology to derive extra deaths in England for 2019 and 2020. The information revealed in England are reasonably much less passable than in Sweden, so the derived estimates must be considered approximate. I used knowledge from Desk 1 of the UK Workplace of Nationwide Statistics (ONS) month-to-month mortality evaluation for December 2020, which spans 2001 to 2020. Doing so offers greatest estimates for mixed 2019 and 2020 extra deaths of 113,000 (0.20% of the inhabitants) when predicting regular deaths by regressing age-standardised annual mortality charges over 2001–2018, or 44,000 (0.08% of the inhabitants) when regressing over 2009–2018. The estimated extra deaths for 2020 alone had been respectively 95,000 and 58,000. Different knowledge revealed by the ONS suggests 2020 extra deaths in England had been modestly beneath the common of those two estimates, and represented some 0.13% of the inhabitants.
Whether or not the longer or shorter regression intervals present higher estimates of regular mortality in 2019 and 2020, it appears clear that extra deaths, as a proportion of the inhabitants, had been a lot larger in England than in Sweden. Extra deaths in England per 100,000 inhabitants had been about 4 occasions these in Sweden for 2019 and 2020 mixed, and about double these in Sweden for 2020 alone, .
Nicholas Lewis 18 February 2021
Initially posted right here, the place a pdf copy can also be out there