Viral fireworks

Around New Year, the Swedish newspaper DN had an article about the spread of contagious diseases, like influenza and covid-19 (Letmark 2024). According to the article, the viruses are spreading rapidly because we have lots of contacts around Christmas and New Year. Sweden has indeed had to influenza seasons in a row with reporting cases peaking in last week of the year. At the same time, one must remember that this is due to gradual build-up of infections the weeks leading up to Christmas, when people have lots of contacts at schools and workplaces. If cases peak around New Year, this implies that \(\r\) is on its way down. As the article also states, the spread often increases after start of spring semester, but if a lot of immunity has been built up, we may avoid later peaks.

In my Swedish 25 November post, I wrote about infectious disease trends in Sweden during fall, with covid-19 on a plateau, Mycoplasma pnemumoniae infections at high levels, and flu still at low levels. Updated reports around the holidays are sparse, but data up until week 52 are available for Stockholm (Karolinska Universitetslaboratoriet 2025). Not much has happened with covid-19, and M. pnemumoniae infections are now on their way down. Influenza A has increased during December, but is still at lower levels than during the December months 2021–23, and the increase has stalled during week 52. As noted, we have yet to see whether there will be large increases after the holidays.

An important reason for the early influenza peeks during 2021–23 is most certainly that population immunity has been lower than usual, because influenza almost stopped circulating during the pandemic. Now, when we have had both A(H1N1)pdm09, A(H3N2) and B/Victoria circulating after 2021, we may see a return to a pattern with influenza peaking after New Year. Such a pattern for influenza peaks has been most common in Sweden for several decades, or more. In my Swedish 22 June 2020 post, using data back to 1981, I noted that peaks in total mortality already in December has coincided with severe waves of seasonal influenza, as in 1988 and 1993.

On one of my old blogs, I wrote a Swedish post 20 January 2011 about cause-specific mortality in Sweden per month, for the years 1969–86, when ICD-8 was used and monthly statistics on underlying causes of death in Sweden were included in published reports, later made available via SCB (2019). For the ICD-10 years, 1997 and onwards, such statistics has recently been made available via Health and Welfare (2024). For the ICD-8 period, the monthly statistics is available for all ages, and ages above 75, but for the ICD-10 period, only statistics for all ages is currently published.

In the monthly ICD-8 statistics, important categories for respiratory infections include acute respiratory infections and influenza (ICD-8 460–478) and pneumonia (ICD-8 480–486), while the newer statistics uses the block influenza and pneumonia (ICD-10 J09–J18), where the use of the and conjunction often causes confusion, both in Swedish¹ and English as it might be interpreted as referring to having both influenza and pneumonia, not (at least) one of the diseases. Fig. 1 and fig. 2 show number of deaths among females and males in Sweden, all ages, per month 1969–86, for the mentioned ICD-8 categories, and fig. 3 shows these data for influenza and pneumonia 1997–2024, with preliminary data available up until September 2024.²

The graphs clearly show that mortality peaks already in December or January have been relatively uncommon during both periods. In 1969, the A(H3N2) pandemic peaked around New Year, and caused significant peaks in both of the respiratory categories. After a calm 1970/71 season, there was a new peak in late 1971, which perhaps may be seen as a second pandemic wave in Sweden. Otherwise, there are no clear December peaks during the 1969–86 period. The severe A(H3N2) seasons 1988/89 and 1993/94 occurred both during the ICD-9 period, for which no monthly cause-specific statistics has been published yet, even though the excess total mortality is clear, as noted above. For the ICD-10 period, December and January peaks in the respiratory categories can be seen for 1999/2000, 2003/04, 2016/17, and the latest years, from 2021/22. Note that the latest seasons are not large in size, compared to e.g. 2018. Even though immunity debt after the decreased circulation in 2020–21 seems to have influenced the timing of the influenza seasons, they have not necessarily been much larger than usual in terms of e.g. mortality. In this respect, antigenic drift, relative to the strains people have been exposed to earlier in life, may be more important than waning immunity for influenza (Munro and House 2024).

One must remember that these respiratory captures by no means capture all excess mortality caused by influenza. The smaller ICD-8 category, acute respiratory infections and influenza is probably rather specific for influenza-related mortality, but has low sensitivity. The categories including pneumonia are probably neither very sensitive or specific. As I noted in my 2011 post, much influenza-related mortality is ascribed to circulatory causes, and in recent years probably often dementia. For example, fig. 4 shows deaths among females and males in Sweden 1969–86 with other ischemic heart disease (ICD-8 411–414) as cause of death.

There are clear peaks in this type of mortality coinciding with, for example, the flu peaks during March 1976 and March 1983. One may note that the category is rather diffuse, and that many of these deaths in recent years probably would have been ascribed to other causes, such as dementia, in accordance with the trends I discussed in my 9 December 2024 post.

For several seasons during the late 1970s and early 1980s, there is very little excess mortality from influenza. This coincides with the reintroduction of A(H1N1) 1977, where older generations were protected by immunity they had acquired when this subtype circulated before 1957. For similar reasons, they 2009 A(H1N1)pdm09 pandemic did not cause much excess mortality either. There are also seasons dominated by influenza B, with low average age among cases, and similarly low excess mortality. One may take the 1983/84 as an example. During the winter, Moscow reportedly had high influenza activity, mostly influenza B, and most cases occurring among children and young adults (CDC 1984). Eventually, the epidemic reached Sweden, and newspaper articles, searchable via Kungliga biblioteket (2024), indicate a peak in Stockholm in early April, while mortality was at low levels.³ In contrast, the relatively high mortality during the possibly last B/Yamagata season, 2017/18, when cases were much older on average, as I wrote in my Swedish 8 February 2020 post, shows that this depends on population immunity, and not on influenza B being more benign by itself.

References

---

1. I wrote about a Swedish example in my 14 March 2020 post.↩︎

2. The figures may be reproduced in R by cloning the blog repository and running 2025-01-03-fire.r in the subdirectory 2025-01-03-fire. Note that the figures for 1969–86 are based on data in the PDF reports and may contain transcription errors.↩︎

3. Some may have caught it and recovered, only to succumb to more sinister fates in early summer.↩︎

Webbplatsen drivs av Karl Pettersson och har genererats med Hakyll. Källkod. Kakor används vid kommentering med Isso, för att möjliggöra editering av kommentarer inom 15 minuter efter att de postats, vilket inte kommer att fungera korrekt om kakor stängs av i webbläsaren. Kommentarer som skickas sparas för att publiceras med namn och webbsida, om dessa angetts: en fungerande e-postadress kan, om så önskas, anges för att få möjlighet att kommunicera kring kommentaren i efterhand men publiceras inte. Om JavaScript inaktiveras, kommer varken kommentarer eller visning av formler med MathJax att fungera. Om fel uppstår vid besök, kan uppgifter om besökares IP-adress och tid för felet lagras i interna loggar och användas vid felsökning.