Inspired by an online conversation and the hard work of others, I set off to examine age groupings among Australians confirmed as COVID-19 cases, based on testing using a quality-laboratory method, from 2020 to April 2026. The findings revealed interesting patterns, especially in the youngest age groups contributing to annual infection totals, as well as some more questions. Wanna have a look with me?

Preface

First up, a little background. You may have seen me asking questions about observations Ryan Hisner (teacher, citizen scientist, mutation and amino acid whisperer extraordinaire) noted and about which molecular virologist Professor Marc Johnson recently wrote (see his blog piece). In short, based on submitted SARS-CoV-2 sequences with attached age data, Ryan found that a greater proportion of SARS-CoV-2 BA.3.2 variant sequences were from children under 10, or 18 (why two thresholds?), years of age (yoa) than any non-BA.3.2 variants in circulation at the site at that time (December 2025 to present). When looking at 5 countires with the highest number of submitted BA.3.2 sequences, a similar pattern was identified. They didn’t see this pattern when they looked at XBB or BA.2.86/JN.1 variants – although they didn’t seem to compare each to the non-XBB or non-BA.2.86 sequences at the same time as they did for BA.3.2. Before Ryan and Marc, Professor Stefan Pohlmann, Josette Schoenmakers, JPWeiland, Euan Arnott, d_pruss6 and gwladwr (and possibly others – apologies if so) had been noting an association between SARS-CoV-2 BA.3.2 sequences or positivity and a trend towards young children. BA.3.2 has been around since emerging in November 2024, but hasn’t taken off…it also hasn’t disappeared.

Ryan has since offered a hypothesis, all over on X, if you want to read more. BA.3.2 has a number of mutations, but Ryan was particularly interested in the complete deletion of some ‘genes’ (for the sake of simplicity) called Open Reading Frame (ORF) 7 and 8 (see the image below), which are usually present in SARS-CoV-2 variants.

Anyhoo! BA.3.2 variants seem to be turning up among sequences from children more so than from adults. My questions were mostly about sampling for sequencing, which is a small subset of total SARS-CoV-2 detections, even at a single site, let alone across larger jurisdictions. There isn’t always a lot of transparency about how a sample is selected for sequencing. It’s possible that sample selection can be subject to biases, such as requiring a higher viral load (lower C_T), relying on which lab or hospital sends what and when, how accurately sampling reflects the general COVID-19 population’s disease burden, as well as delays to sequence uploads and probably a myriad of other decisions behind the scenes. But Ryan, Marc and others looked over as much of that as was possible, and the trend for BA.3.2 being found in sequences from children moreso than adults seems to hold from sources all over the world.

I wanted to have a look at what was going on in Australia, but not in terms of sequences or sequencing being done – some of the questions I had could not be answered by those data. Instead, I thought I’d look at Australian laboratory-confirmed SARS-CoV-2 detections and break them down by age and by year from 2020 (the first confirmed case in Australia) to 2026. A look backwards. Also, I wanted to see how that compared to the same breakdown for influenza detections in Australia.

Reading the graphs you’re about to see

The zoomed-in snippet of a graph I’m about to show is presented below. I get that it’s a busy graph, but hopefully this walk-through will make it easier to see the trends. Fire me questions in the comments section at the end.

Key features of the age band graph, which apply to the full versions later:

• Each bar represents one year: 2020, 2021, 2022, 2023, 2024, 2025, 2026. We are living through the rare and ongoing opportunity to watch COVID-19 “settle in” to the human population since it emerged in late 2019.
  Below each cluster of years is the “age band” (=age grouping). This one shows the data are from those who tested positive between 20 (years of age) yoa and 24 yoa.
• The height of each bar represents the percentage of positives in that year attributed to that age band. In 2020 (first bar), 10.3% of all confirmed cases were in this age band. That peaked in 2021 at 14.0%, then the contribution dropped from 2022 onwards to a fairly steady 3.5%.
• I’ve also used this approach to graph the influenza age and year data later (orange bars).

• These are the same data, just organised differently to make a different graph.
• In those grpahs below, each bar represents one of the 18 different 5-year age band groupings: 0-4, 5-9, 10-14, 15-19, 20-24, 25-29, 30-34, 35-39, 40-44, 45-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79. 80-85 and 85+ yoa.
• The height of each bar represents the percentage of positives in that year attributed to that age band. In 2020, 2.6% of cases were contributed by the 0 to 4 yoa band, while 5.1% were 85 yoa or older. Those aged 25 to 29 yoa were the largest contributors to cases, at 12.1% that year.
• I’ve also used this approach to graph the influenza age and year data later (purple bars).

A big caveat to all this is that the number and nature of people tested changed dramatically from late 2022/early 2023 onwards (see below). That is seen in a drop in total positive numbers (graphed below with either a log₁₀ y-axis (left graph) or a linear y-axis (right graph), but not necessarily a change in the proportion of each age band that tested positive. The proportions of different age bands contributing to the full year’s caseload may be affected by immunity, whether there’s an epidemic peak, and perhaps the variant itself, in ways not linked to antigenicity. Perhaps. But in the age-band snippet above, the proportion of positives in the 20-24 yoa band used above did drop from 2022 onwards.

Another caveat is that I don’t have a denominator that includes all the samples tested – just the number who tested positive and their age breakdown. Did we get more samples from the younger age bands and thus more positives? Did we get fewer adult samples and thus fewer positives? That’s unclear because those data weren’t in the public-facing National Notifiable Disease Surveillance System [NNDSS] database.

Testing times

In 2022, we tested more people for a respiratory virus than at any time in Australia’s history. You can see that really clearly in the linear y-axis graph above. That included sick people and well people, such as incoming and outgoing travellers, people needing clearance to enter or leave a hospital or workplace, people in quarantine or isolation, contacts and those who just rocked up to a lab for a test to see whether their illness was due to SARS-CoV-2.

From 2023, the focus was on detecting the cause of illness rather than supporting quarantine and preventing the entry and spread of a novel pathogen.

Those getting tested shifted mainly to those ill enough to be in the hospital or to those visiting a primary care doctor who recommended a lab test. Even then, SARS-CoV-2 testing regimens may vary according to local guidelines and standards and are now often part of a ‘triplex’ test that includes influenza viruses and respiratory syncytial virus (RSV; all of which are modifiable by vaccination). Rapid antigen tests (RATs) are not represented in these data. The number of samples for which age data were not provided, recorded, or legible ranges from 0.03% (2026) to 0.16% (2022) of the annual totals.

Changes in detection proportions among different age bands over time

That’s a bit of a mouthful. Basically, the full graph takes the first example graph above and adds back the rest of the age bands. Click the image to view full screen – it will look better on a computer than on a mobile.

This shows the change in the proportions (%) of each age band over time.

There are many unknowns about the causes of those changes, so I’ll try to focus on some key observations and the few guesses my tiny brain has come up with. Feel free to add your expert thoughts in the comment section below.

The youngest kids make up most of those who test positive for SARS-CoV-2 in 2026, but this has been a growing trend since 2024

Ryan’s observations for 2026 hold up in Australia, kind of. Children make up the highest proportion of those testing positive for SARS-CoV-2, but mostly in the 0 to 4 yoa band, at 17.2% of all detections for 2026. The next-biggest contributions come from those aged over 85 yoa, then those aged 5-9 yoa; then it gets messy, with those aged 10-14, 30-44, and 70-84 yoa being similar. Of course, we are only in April and have not had an epidemic peak since July (winter) 2025 (see below). Epidemics may see different spread with foci at places of gathering, perhaps schools initially, then households and offices later, for example.

If we look at the age extremes – those under 4 yoa and those 65 yoa or older over the 6 years and 4 months we’ve lived with COVID-19 in Australia, a couple of interesting things jump out at me in the next age-band graph.

Firstly, that proportion of positives among the youngest is very clear and has been rising since 2023 (the yellow arrow highlights this trend).

Secondly, there is an increasing proportion of cases contributed by the age groups we tried to protect the most early on during the pandemic – those aged 65 yoa and older. Specifically, a greater share of the nation’s cases came from those aged 75 to 85+ yoa from 2023 onwards, compared with the first three years of COVID-19 data. As we’ll see soon, the proportions of positive people in the 65 to 85+ yoa bands are lower than those in the 20 to 55 yoa bands, during the early years of the pandemic.

However, annual proportions aside, when accounting for age-band population sizes, the eldest Australians actually contributed to some of the highest rates per 100,000 during the first year of the pandemic, especially among people in the 90+ age band.

It’s all a bit subjective, and the percentage changes aren’t always very large, but I’ve used the arrows to give a quick idea of how the proportion of positives within an age band has changed in the most recent 4 years compared to the first 3 years of the pandemic.

Looking at the age bands I excluded from the first graph.

While the next two age bands for children (5 to 9 and 10 to 14 yoa) have become larger contributors to the caseload in 2026, they haven’t really surpassed the proportions of positives they represented in the early years of the pandemic. It’s unclear whether this is a cycle that will repeat or a major, permanent shift.

It would be good if Ryan and Marc looked back in time to check the contribution of children to earlier sequencing data sets as well. It will be interesting to follow these groups this year as the balance of variants changes. And if we get a winter epidemic.

The younger adult age bands that were hit hard when the pandemic was unleashed from New South Wales (20 to 24, 25 to 29 and 30 to 34 yoa) showed the most marked reduction (green dowards arrow) in contribution to case detections in recent years, with other older adults showing a generally steady (orange horizontal arrow) or slightly decreased (blue downards arrow) proportion year-on-year from 2023 onwards.

Patterns in the percentages

Below is the table of proportions I used to make my graphs. If you stand back, lift up one leg and squint a bit, you can see – well, I can see – two things.

Firstly, a trend of drifting toward older ages among the highest percentage age bands, moving from the older teens to young adults, then to older adults, between 2020 and 2023.

Secondly, quite a change after 2023. Among people aged 75 yoa and older, the proportion of positives in later pandemic years compared to earlier as much as doubled; the trend starts in 2023 for those between 65 and 75 yoa. Remember that this is commonly the age group in which the most severe COVID-19 outcomes occur. And again, you can see the proportion of positives from people in the youngest age band more than doubles after 2023. Those between 5 and 14 yoa rise in case share in 2026, but we’re only a third of the way through, so that may change.

Same data, plotted differently; aged bands grouped by year

This next graph changes it up, in case this interests you, as it did me. I changed the plot so the x-axis shows the year (instead of age), so you can see a bit more clearly how each age band contributed to the positives for a single year at a time.

This allows you to pick an age band and compare its height in that position across years. For me, it highlights that sudden rise in the youngest (lightest bars) and oldest (darkest bars) age bands. Nothing new to add – just a different way of looking at the same data.

How do influenza cases compare when plotted this way?

Same data source (NNDSS), same years examined, and the same two graphing formats, but some very different patterns.

There is no dramatic and ongoing increase over time in the youngest age band’s contribution, although the proportion does rise and fall within the 6.5-year window – imagine if we looked at this graph in 2023 – we’d have been asking what’s going on among people of 5 to 9 yoa!

There’s generally a bigger proportion of detections deriving from the first 3 or 4 age bands, not just the first one. Younger adults contribute relatively steady proportions of positives over the 6-year period, with a decline over time among those aged 30-44 yoa. Those aged 45-59 held steady after the pandemic years, when total influenza case numbers were historically small, but those aged 65 to 85+ yoa contribute a steadily rising proportion of total positives year-on-year, with a pattern that also differs from that of people in the same age bands who were positive for SARS-CoV-2. Is something happening among our elderly population? Are we failing to protect them?

But influenza is complicated, and an older companion

This is all complicated by the fact that there are 3 influenza virus subtypes, each affecting different age groups, evolving at different rates, influenced massively by international travel and constantly changing subunit vaccines, as well as existing for many, many years before the COVID-19 pandemic, therefore creating in each of us a complex mosaic of exposure and immunity. Nonetheless, for this comparison, the COVID-19 and influenza patterns are largely distinct (except for a shared lack of protection for our elderly populations against endemic, recurrent, virus-driven inflammatory infections).

Final words on age and COVID-19

Ryan and Co.’s sequence analysis has definitely picked up on a pattern at its most active in 2026.

But when you look back in time, at least using Australian national COVID-19 confirmed case data, you can see this isn’t just due to BA.3.2 now; it’s been a growing trend, driving a larger, although still minor, share of COVID-19 cases.

Those aged 20 and older currently comprise two-thirds (67.5%) of confirmed Australian COVID-19 cases in 2026 so far. To put that in context, they accounted for 76% to 86% of cases in each previous year, although that percentage has been declining since 2023 (86.0%➡️84.6%➡️80.0%➡️67.5%). So something appears to have changed. Or might a fast-moving seasonal epidemic smooth out this disparity for 2026? Winter’s coming.

If we choose 10% as an arbitrary threshold above which an age band is considered a major contributor to COVID-19 cases, then:

• Those in the 0 to 4 yoa band are the only ones to have crossed that threshold in 2026 (currently comprising 17% of those who are NNDSS-confirmed cases in 2026).
• In 2025, those in the 0 to 4 yoa and 85+ age bands crossed that threshold.
• In 2024, it was only those in the 85+ yoa band among the over 10%’ers.
• 2023 and 2022 had no line crossers.
• In 2020 and 2021, the young adults – mobile, parents of children who are “getting infected with the virus much less frequently than adults“ (🤨), and working on the frontline, composed those age bands crossing the 10% line.

Currently, Australia has 3 distinct co-circulating variants (also XFG and NB.1.8.1), of which BA.3.2’s latest derivatives are just one, but nowhere does anyone characterise variants from all cases. The proportion of those 0 to 4 yoa, but not really other children, has been rising for several years. Overall, adults, not children, make up the greatest proportion of those testing positive for SARS-CoV-2, but the balance has been shifting since 2023.

The analysis above, extending the timeline backwards by 5 years, doesn’t exclude the possibility that there’s something different about the BA.3.2 variant that has led it to target very young children moreso than adults. It just highlights that something has been growing over time. If viral change, these changes seem to have been present in variants since around 2023. And there’s a comment in Marc’s blog that may hold a hint:

“the displacement of XBB lineages by BA.2.86/JN.1 in late 2023 to early 2024“

He’s noted a variant changeover that aligns with some of the age patterns above. The specific comparison differed from what Marc used previously – there was no XBB versus non-XBB or BA.2.86/JN.1 versus non-BA. 2.86/JN.1 comparison which may have shown nothing extra, but would have been nice for completeness.

I’m still of the opinion that this could simply be related to the accrual of immunity over time, with a hole in that immunity in the young, allowing more effective infection and transmission within these age bands than to and among adults.

I’d like to see more detail on how sequencing samples are chosen at each site for tighter uniformity in what’s being compared, just to firm up that analysis. That said, the trend toward a larger role for sick children in transmission appears solid, regardless of the underlying mechanisms and variants involved. And, in Australia, and this may differ elsewhere, it’s been going on for a few years.

We’re dealing with a newly emerged virus and watching how it interacts with its human host in real time, and I acknowledge that we’re still learning (I know I am!) and need to stay open-minded. This isn’t a process we’ve lived through before, nor did the past have the analytical tools we have today to dig into what we observe, so the playbook is still being written.

By now, with vaccination (even if only using one protein), boosters and infection, hybrid immunity among adults should be pretty complex, so initially healthy adults should be less likely, relative to 2020-2022 anyway, to become severely symptomatic after infection. That might explain the mild or absent COVID-19 epidemics we’ve had recently. As you may remember, severe illness was never the majority outcome in COVID-19. And of course, in older adults, immunity wanes, so they may need more frequent vaccinations (as is already recommended, though I’m unsure of uptake) or perhaps a higher dose vaccine formulation to prevent what we’re now seeing – a rise in their share of infections each year. Could vaccine uptake among the elderly be a problem we need to address?

Young children in Australia are unlikely to be vaccinated and therefore rely on infection to build immunity because healthy children up to 18 yoa are “not eligible” to receive a primary course of the COVID-19 vaccine, and it’s not recommended that they get boosters. Perhaps it makes sense that they now account for a larger, seemingly growing share of detectable symptomatic infections than adults do. Just noting that there isn’t an obvious spike in COVID-19 hospitalisations among those 0 to 4 yoa in 2026.

If it is truly a variant-driven phenomenon resulting from mutational inactivation of important viral genes, that would be really intriguing. It would also be interesting to consider what evolutionary pressures may have selected a novel virus’s novel variant to drop key genes and seemingly favour spread in a specific younger age band, driven by mutational rather than immune pressures. Has this been a process common to other respiratory viruses as they first emerged into humans, then, after some years, settled into children (200+ respiratory viruses dominate in relatively immune-naive young children)? Or are we getting ahead of our data?

Beyond the questions above, I’ll highlight these final two based on what might be changes in COVID-19 epidemiology: are we now failing to shield our eldest from the harms of SARS-CoV-2 after its initial emergence and the flurry of protective measures we took (applied to flu as well), and should we reconsider the importance of vaccinating infants?

Questions, questions, questions!

References

1. Is BA.3.2 Disproportionately Infecting Children? A Look at the Age Distribution Data.
   https://lung.fish/blog/2026-04-06-ba32-age-distribution/
2. National Notifiable Disease Surveillance System dashboard
   https://nindss.health.gov.au/pbi-dashboard/
3. The first year of COVID-19 in Australia: direct and indirect health effects
   https://www.aihw.gov.au/reports/burden-of-disease/the-first-year-of-covid-19-in-australia/summary
4. COVID-19 Australia: Epidemiology Report 32: reporting period ending 3 January 2021. Communicable Diseases Intelligence Volume 45.
   https://www1.health.gov.au/internet/main/publishing.nsf/content/C50CAE02452A48A7CA2587320081F7BF/$File/covid_19_australia_epidemiology_report_32_four_week_reporting_period_ending_3_january_2021.pdf
5. COVID-19 vaccine advice and recommendations
   https://www.health.gov.au/our-work/covid-19-vaccines/getting-your-vaccination?language=en
6. The critical importance of COVID-19 vaccination in older Australians: an evidence-based perspective
   https://www.australianageingagenda.com.au/sponsored_content/the-critical-importance-of-covid-19-vaccination-in-older-australians-an-evidence-based-perspective/
7. NSW Respiratory Surveillance Report – week ending 4 April 2026 [Issued 9 April 2026], Epidemiological summary, Week ending 4 April 2026.
   https://www.health.nsw.gov.au/Infectious/covid-19/Documents/respiratory-surveillance-20260404.pdf
8. Queensland COVID genomics epidemiology summary 26 March 2026
   https://www.health.qld.gov.au/__data/assets/pdf_file/0024/1361922/sars-cov-2-genomic-analysis-report.pdf?version=23
9. Kids may be more likely to get the new ‘Cicada’ variant of Covid-19, scientists say. Here’s what to know about BA.3.2
   https://edition.cnn.com/2026/04/02/health/new-covid-variant-cicada
10. Evolution and viral properties of the SARS-CoV-2 BA.3.2 subvariant
    https://academic.oup.com/ve/article/12/1/veag011/8490867

Updates

15APR2026

• Fixed the aspect of the post’s thumbnail graphic
• Updated the intro blurb and Preface. Included more of those who noticed the age shift in 2026
• Update the COVID-19 2026 age spread (and corrected the title of that section) with some more details about all the age bands.

17APR2026

• Added a host of other people’s handles who had already been discussing this before I stuck my oar in
• Added a graph explainer for the ‘age-band grouped by year’ graph
• Further clarified and tidied the age rate section
• Updated the influenza text
• Added to and tidied up the Final words section.

20APR2026

• Added a second caveat about the data lacking total sample numbers – just those testing positive – not how many were tested to get those numbers.