“Flu cases jump by 256% in just five years” screams a newspaper headline from the Daily Mercury, a paywalled News Corp publication.
It sure seems like scary stuff doesn’t it?
The headline implies a great leap in the number of Australian cases of influenza (flu). And the jump has occurred over a short time.
It suggests something must have changed.
And all of that makes us feel like we should be worried. And of course, we all really need more things to feel worried about don’t we?
But is this headline’s implication correct? Is it even accurate? Is it a good way to look at or talk about the data? Let’s take a closer look into what’s been happening.
Flu Numbers in Australia
Australia’s national flu numbers can be found at the National Notifiable Diseases Surveillance System (NNDSS) site hosted by the Department of Health.
Graph 1 is what the past 25 years of reported data look like when we plot each total number of laboratory-confirmed flu cases by month.
From this graph, you could take away that the biggest flu year in two and half decades was…last year.
You’d be partly right – but it’s a bit more complicated.
Two non-viral things known to drive flu numbers upwards
Firstly, you probably already expect flu numbers to rise because influenza is caused by incredibly easy-to-transmit viruses.
As our population size grows, so to our densely packed cities get bigger. This creates more potential hosts. So sure, there are more flu cases as time goes by. No need for a changed doomsday virus here.
Secondly, laboratory testing has changed to cope with increasing populations and increasing demand for testing and for rapid provision of test results. Some of those really low years in the NNDSS data are because flu virus testing was not requested as often, took ages, wasn’t done as much, and may not have been centrally reported. Testing has become more sensitive, more specific and a lot faster – which also means Doctors will be more likely to request something they can get a result for in a patient-relevant timeframe (and because flu is a notifiable disease and there are antivirals). In Australia, 2009 was an important point in time for testing because of the pandemic of “swine flu” that swept the world.[4,5,6] From then onwards PCR-based techniques became the favoured method to detect flu viruses. Firstly as in-house lab methods, then as kits and in the latest iteration in self-contained platforms that can be reliably used outside of the lab, as “point-of-care-tests” (POCTs).
The older methods, based on relatively insensitive and much slower cell culture, are now the domain of the specialist laboratory. Culture is still used for typing patient-derived flu strains and for determining vaccine effectiveness but it has not been the best choice for screening patients for years.
In Australia, POCT technology for flu has been intensively rolled out over the past few years. With wider access to leading-edge testing technology, more cases of flu are being reported year-on-year. Laboratory-based PCR and POCT PCR have undoubtedly pushed notification numbers up over the past few decade years – but just how much of a push still remains unclear.
Why cherry-picking numbers can be biased and is often misleading
So back to the article I described above, It picked from laboratory-confirmed flu case (notifications) totals to June for 2013 and 2018. I have slightly updated numbers for the NNDSS so now, there has been a 308% increase in notifications in 2018 compared to 2013. Ahhhhhh…………………………!!!!!
……….okay, I’m back.
The first thing I do when I see two numbers picked out of a time series like this is to ask: “what’s happening before and between these points (also after, but that doesn’t apply here)?
Flu notifications rise because of the reasons above but can also rise because of a novel or changed flu virus emerging onto the scene. This will be a virus we don’t have immunity against. And that’s the implied threat in these sorts of quickly thrown together “news” articles.
Anyway, let’s look at this totals. Graph 2 shows a trend of increasing notification numbers, especially as better testing has been implemented.
In other words, if you look at any year from about 2010 onwards, and compare that to a previous year, you’ll often see a “jump”. The exceptional years are the pandemic of 2009, and 2017 which seems to have been driven by a divergent H3N2 virus combined with a vaccine virus production-induced change that worsened the mismatch. Yes, there are special cases. But trends matter.
If we compare January to June 2013 to 2017 we see a 442% jump – the total of 2017 was 4.4 times higher than that of 2013. When you put it like that though (instead of in hundreds of something), it doesn’t really sound quite as bad? Maybe that’s a less inflammatory way to talk about this?
The 2014 January to June period had 1.8 times more cases confirmed than 2013, but what about going backwards in time? If we look at 2012, we find 1.8 times more notifications than in 2013, so it’s not always about more cases as we go forwards in time. 2010 only saw a quarter (0.26) of the notifications of 2013 while the 2009 pandemic year had 3.3 times the notifications of 2013.
If we look at specific months instead of year-to-date, I could tell you that Feb-2018 notifications were 2764% higher than those of Mar-2009, and 2009 was a pandemic year!! How could this be? Pandemics are bad mmkay? Of course, they are, but numbers alone don’t convey the story.
For instance, May-2018 was only 66% (two-thirds) the total of May-2009 and May-2015 but 9853% that of May-2001 when there were few notifications collected because of the type of testing and perhaps the type of reporting systems and data sharing among the Australian States and Federal health departments.
Getting the gist? Where you pick the numbers from makes a big – and biased (if you’re after a particular impact) – difference to the story. If you also don’t have an understanding of that story, you fail to write something that is accurate, useful or long-lasting.
Picking out two single numbers from a spectrum and deliberately ignoring trends doesn’t necessarily inform an audience of anything, but it drives clicks because of the associated alarmist banner headlines. I firmly believe that good story-telling could also drive clicks so write constructively not destructively; if your organisation allows that of course.
Is 2018 the year of flumageddon?
Has a new virus emerged in 2018? In brief – no – because there is no sign – so far – that this year is being hit by an unusual strain of flu virus. The virology doesn’t say that. The epidemiology doesn’t say that. The bed use doesn’t say that. The infected age distribution doesn’t say that. Deaths don’t say that. At least, not at the half-way point for 2018! We’re not even halfway through the Australian winter; our traditional flu season has yet to engage.
There certainly remains a chance that we’ll have a big year. But watch and see.
2018 started with the biggest first quarter of any year on record as we’ve looked at earlier. That may have been due to increased access to fast, accurate testing or to something virus-related. But after that early spike, Flu cases dived down to levels among the smallest monthly totals seen over the past 5-years.
Flu. Whatcha gonna do?
Find reliable sources and dump alarmist biased and destructive ones
As the flu season looks to be finally revving up down under, we get to see what sort of year we’ll have after the biggest on record. Until we in eth midst of it though, don’t believe media outlet claims that seek to instil fear about flu.
There have been some particularly poor quality, confusing and misleading claims made by some articles on Australia’s upcoming flu season in the “popular press” this year; except for what’s on this blog of course. 😃 Ignore the headlines that seek to create fear and seek out more reliable sources. Or ask me and I’ll try and get to the bottom of your concern or point you to those who already have.
- Influenza diseases by month and year – select this from the page below.
- Is increased testing at the root of Australia’s biggest ever first quarter of Flu cases…?
- Epidemiology of Australian Influenza-Related Paediatric Intensive Care Unit Admissions, 1997-2013
- Diagnostic testing in influenza and pertussis-related paediatric intensive care unit admissions, Queensland, Australia, 1997-2013.
- Pandemic (H1N1) Influenza 2009
- AUSTRALIAN INFLUENZA SURVEILLANCE REPORT No. 01, 2018
21 May to 3 June 2018