How does this flu season compare with others?

10 08 2009

Just how much worse is this flu season here in Australia, compared to other seasons? I’ve looked at this question twice before, over the past few weeks (see here and here for good background). Well it’s time to take another look – because the numbers are in for July 2009.

As I’ve discussed before, the Department of Health and Ageing have a “notifiable diseases report generator” that will give you a table of disease notifications, which is then easy to graph. So, I’ve run this generator for influenza notifications and got some interesting stats.

The first graph (below) shows the number of influenza notifications in Australia for the years 2001 to 2008 (8 years all up), by month. At each month in this graph I show three points:

  • The bottom point is the minimum value for this month over the 8 years. In other words, I look for the year that had the lowest number of flu notifications for that month, and that’s the figure I plot. For example, if I choose the month of July, the year with the fewest notifications for flu was 2004, with only 96 notifications being made in the July of that year. So, I plot 96 as the minimum point for July.
  • The top point is the maximum value for the month. For example, for the month of July, the year with the most notifications was 2007 (of all the July’s from 2001 to 2008). In that year, July clocked up 2,470 notifications. So that’s the number I plot for maximum.
  • The middle point for each month is the average number of notifications for that month over all the years 2001 to 2008. So, for July we got an average of 770 notifications in that month for the years (2001 to 2008). So I plot 770 as my average.

Note that these are notifications for all types of flu (not just swine flu, which wasn’t around in its present form anyway):

year on year comparisons 2001 to 2008

You can double click on this graph to expand it.

The graph tells an interesting story (as I’ve remarked before). Flu notifications tend to peak in August. And there’s still a lot of flu notifications in Australia in September (on average, more than in July). In other words – we are not through the worst of our normal flu season yet.

The next graph then becomes very interesting. All I’ve done is included 2009 monthly notifications in the above graph, as a fourth (purple) line. I’ve changed the scale to fit 2009 in. Here it is (again, you can double click on it to expand it):

year on year comparisons 2001 to 2009I think this graph tells the story well enough as to just how bad this season is compared to a normal season. The only caveat I would add to this is that we have clearly undertaken a lot more testing for flu in Australia this season than in a normal season. So some of the marked increase in notifications has to be due to an increased testing effect. Just how much – I don’t know.


ECDC: Planning Assumptions for the First Wave of Pandemic A(H1N1) 2009 in Europe

30 07 2009

The ECDC have just released a “Public Health Development” entitled “Planning Assumptions for the First Wave of Pandemic A(H1N1) 2009 in Europe”. Read it here (with full references). This paper seems to have been prompted by Norway and the UK re-assessing the assumptions they are using in their pandemic planning. The paper starts this way:

As it is summer in Europe the 2009 pandemic has yet to really accelerate in EU countries but the experience in temperate Southern Hemisphere countries suggests it is inevitable that Europe will be affected by a major first A(H1N1) 2009 pandemic wave in the autumn and winter.  The 2009 pandemic is less severe than might have been expected and ECDC has been made aware by two European Union countries (Norway and the UK) of the updating they have made of their planning assumptions specifically for a first wave of an A(H1N1) 2009 pandemic.

The paper is short, but makes interesting reading and starts to answer a number of questions.

What do we think the “clinical attack rate” might be?

The “clinical attack rate” is the proportion of the population that will catch Swine Flu at some point in the pandemic’s “wave” and display symptoms. The paper says it is assuming a clinical attack rate of 30% (and that another 30% of the population will catch Swine Flu but display no symptoms). By the way, I’m guessing that “displaying symptoms” means getting sick enough to take at least half a day off work – my understanding is that that’s the requirement to be “symptomatic”. Here’s what the paper says about the clinical attack rate it’s assuming:

This is 30 % (The UK clinical attack rate is based on an assumption that half of the infected become symptomatic so this would imply a total infection attack rate of about 60 %). WHO assumptions are that two thirds become symptomatic. Whether the UK or WHO is correct will be determined later when the results from serology become available.   The UK assumptions imply a basic reproductive number Ro in the interval 1.4 – 1.5 which seems to be the case at present in the UK. A Ro of value 1.4 implies a total infection attack rate of about 50 % (which would imply a clinical attack rate of 25 % in the UK planning assumptions). A higher value of Ro of 2.0 implies a total infection attack rate of about 80 % (hence a clinical attack rate of 40% in the UK planning assumptions).

What do we think the “case fatality rate” might be?

This is the proportion of people who catch the virus (and are symptomatic?) who then go on to die. The paper states:

This is one of the most eagerly sought parameters but it is also amongst the hardest to determine with any accuracy. The earliest studies of this pandemic gave a high CFR of about 0.4 % compared to lower rates for the 1957 and 1968 pandemics but higher rates for 1918. The UK estimates are of a CFR of 0.1-0.2 though values of up 0.35% cannot be ruled out as impossible.

Later, the ECDC comments:

Case Fatality Rates (CFR) will also change as more data become available and more stable estimates will take some time to emerge. As some cases will be very mild and not reported the reported figures from official tables of cases and deaths will most often be an over-estimate of the true CFR. Equally though many deaths which result from influenza (seasonal or pandemic) are not attributed to the infection in official causes of deaths and so officially reported influenza deaths are always an underestimate, sometimes grossly so. In previous pandemics it has only been computed with any accuracy once the pandemics were over. It is also important to appreciate that CFR is especially subject to social effects. In poor social settings such as Africa even seasonal influenza can result in CFR’s that are higher than seen in pandemics.

What level of absenteeism might we see?

In the comments section the ECDC discusses the UK’s assumption of a peak in absenteeism at 12%:

The UK predicted peak absenteeism rate of 12 % of the workforce is interesting and fits with the mild illness seen for most people. It suggests that the social disruption effects of the pandemic will be less than feared for other pandemics and that severe social interventions will not be necessary given good business continuity planning.

Plan for a 30% “clinical attack rate” for the first wave: ECDC

21 07 2009

The European Centre for Disease Control and Prevention (“ECDC”) have just put out their “Interim Risk Assessment 20th July 2009” for Swine Flu (I was notified about this via CIDRAP‘s email alert service, which is a great daily update on all things Swine Fluish). You can read the ECDC’s full risk assessment here.

You can find the entire Executive Summary to the report reproduced below. Just a few terms to understand first:

  • “Clinical attack rate” means the total proportion of the population that catch swine flu over the entire wave of the pandemic. So, it’s a measure of the total number of people who get sick (most of them mildly, of course). It’s not a measure of the hospitalisation rate or the death rate.
  • “Hospitalisation rate” means the proportion of those who get sick who then need to go to hospital. Can one of my readers confirm this please? I understand that it’s the proportion of people who get infected who then have to go to hospital, NOT the proportion of the total population who will end up having to go to hospital?
  • “Case fatality rate” is the proportion of people, who have caught swine flu, who will then die.
  • “Oseltamivir” is Tamiflu.
  • “Zanamivir” is Relenza.

Here’s the executive summary:

 There are no reports as yet of unusual presentations or transmission routes for this influenza compared to normal seasonal influenza viruses. There is no indication of risk of infection through food or potable drinks.
 If the pandemic behaves like previous ones, cumulative clinical attack rates over the first major wave of infection in 2009–10 might be expected to be in the range of 20% to 30%, with a reasonable planning assumption of 30%.
 Based on experience in North America, clinical attack rates will be highest in children and younger adults.
 Adults over 60 years seem, at present, to be the least affected age group, though there are indications from the USA that those few that are affected experience the highest risk of severe disease of any age group.
 The groups experiencing most of the severe disease and death are those in the risk groups of people with chronic underlying medical conditions (this includes morbid obesity), pregnant women and young children (especially under two years of age).
 Most of those infected experience a mild self-limiting illness, even in people in risk groups. However, as for seasonal influenza there are some people who experience more severe disease and some of these die despite medical care. These include a few people without any known underlying condition and outside other risk groups.
 A reasonable planning estimate for hospitalisation rates in Europe using the overall clinical attack rate as a base is in the in the range 1% to 2%. However, in the winter this may rise because of the presence of other respiratory infections.
 Local experience from the USA (New York City) indicates that, without preparation, this pandemic can severely stress healthcare systems.
 The observed case fatality rate based on the largest population reported to date, from the USA, is 0.4%. While in Europe the observed rate in the earliest affected country (the United Kingdom) is 0.3%. However, this is likely to be higher than the true figure, which may at present be more than the range of 0.1% to 0.2% of all clinical cases.
 As in seasonal influenza, case fatality rates are high in the very young, low in children and young adults and then increase with age.
 At the individual level the highest risk of hospitalisation for an affected person is: a) in the risk groups; and b) for young children and those over 60.
 As yet almost all the viruses have been sensitive to the antivirals known as neuraminidase inhibitors (oseltamivir and zanamivir) but they are resistant to adamantenes (amantidine and rimantidine). There have been a few pandemic virus isolates that have showed resistance to oseltamivir (though sensitive to zanamivir).
 The current seasonal influenza vaccine that contains a component effective against another A(H1N1) virus is not effective against the new pandemic A(H1N1) 2009 virus.
 It is impossible to predict when European countries will be affected, but a proper first wave seems inevitable for the autumn. The experience in one country (the United Kingdom) suggests that countries could be affected considerably earlier in the autumn than happens with seasonal influenza.
 It is too early to predict what the mix of pandemic and seasonal influenza viruses will be this autumn, although there will also be B influenza viruses, as they do not compete with A viruses.
 Pandemic viruses are unpredictable, and can change their characteristics as they evolve. Even pandemics usually slow down in summer, only to pick up in autumn, and the virus may even then come back, perhaps in a more aggressive form, like it happened in 1918–19.
 ECDC will work with Member States, other European Agencies, the European Commission, WHO and its other international partners to gather more information to update this Risk Assessment at intervals. Special attention will be paid to how the pandemic is developing in the first affected European countries and the temperate Southern Hemisphere countries.

Exclude Victoria and the trend’s still ugly

10 06 2009

I’ve stripped out Victoria from the numbers of new cases and cumulative cases and then recut the graphs (see below). I’ve stripped out Victoria for two reasons:

  • Firstly, they’ve stopped reporting regularly, so I’m pissed.
  • Secondly, the large absolute number makes it very hard to see the trends in the other states of Australia.

So, the first graph (below) shows the cumulative number of cases of swine flu in Australia excluding Victoria.

The second graph – which is very interesting – shows the number of new cases reported in Australia (ex Vic) each day. And what it shows is that today we’ve seen the highest number of new cases in Australia yet (not counting Vic). In other words, the virus seems to be getting the foothold in the rest of the country that we all expected it would.

(PS – I’ve updated the Australian total graphs at the top of the blog as usual – but they’re getting less interesting).

Swine flu cases Australia ex Vic

 Swine flu NEW cases Australia ex Vic

Two things to worry about

2 06 2009

Laurie Garrett on Late Night Live again just now. She says we have two things to worry about with this virus.

Firstly, H1N1 and H5N1 get mixed up in a host (is that the right word?). That is, a person or a pig or a chicken catches both strains of flu at the same time. This is now a possibility given that H1N1 is hitting geographies where H5N1 is endemic. In that situation you risk mixing strains to get a new nasty strain. A new strain that has the virulence of H5N1 (which has killed 80%+ of the people who have contracted it in Indonesia) and the contagiousness of H1N1.

The second thing to worry about is if the current swine flu mixes with another H1N1 strain currently circulating in North America that is resistant to Tamiflu.

“Most rapid spread on the planet at the moment”

2 06 2009

Laurie Garret on Late Night Live on ABC radio talking about the spread of Swine Flu in Australia.

But, as she goes on to say, we might just be picking up the cases well because we have one of the best health systems in the Southern Hemisphere.

Australia has the most cases outside of North America: Bloomberg

2 06 2009

This report from Bloomberg describes the “tally” in Australia as “the highest outside North America”. Is this right?

It also says that the rate of spread in Australia may prompt WHO to raise the pandemic alert level to 6:

Victoria has 306 cases, up from 173 two days earlier, reflecting the spread of the virus within communities in the southeastern state. Evidence that swine flu has gained a foothold in Australia may compel WHO Director-General Margaret Chan to raise the pandemic alert to the highest of the agency’s six-step system, MacIntyre said in a telephone interview today.

“It must be fairly close to moving into phase 6,” said MacIntyre, who is also professor of infectious diseases epidemiology at the university in Sydney. “It’s clearly spreading in the community in Victoria.”

Disease trackers are looking for evidence of sustained, community transmission of the pig-derived virus outside North America to meet the WHO’s criteria for a pandemic. Such a global epidemic occurs when a new flu strain, to which most people have no immunity, appears and spreads worldwide.

I got alerted to this Bloomberg report via aCIDRAP feed. CIDRAP is worth keeping an eye on.