Credit Ian Mackay VDU
7205
Over the weekend we saw a few hyperbolic headlines suggesting the H7N9 virus may be the `Deadliest’ bird flu ever, but the reality is, we haven’t enough data to begin to arrive at a reasonable estimate of its CFR (case fatality ratio).
As of today, out of 126 cases, the Flu Wiki lists 24 deaths, while 22 have been treated and discharged from the hospital.
Another 79 cases remain hospitalized, some reportedly still in serious condition after several weeks of treatment.
At this time we don’t know how many of those currently hospitalized will recover, and – just as importantly - how many additional `mild’ cases might be out there that have not been diagnosed and counted.
Dr. Ian MacKay’s Virology Down Under H7N9 website provides excellent commentary and graphs, including this one which deals with the `apparent CFR’ of the virus.
I use the word `apparent’ because we are dealing with incomplete data.
Right now, it sits at roughly 20%.
Dr. Mackay describes the `denominator problem’ (the total number of infected), this way:
Obviously, if the denominator is small (like it is now with so few cases and testing regents only just becoming widely available), then we see high proportions (percentages) of severe disease (x severe cases over a small number of total cases is a high proportion).
If there are mild cases of avian influenza A(H7N9) virus infection out there (as we have seen today, 15.04.13) and they are in large numbers, that denominator could be much larger and the severe disease cases will be "diluted" down to a smaller proportion (x severe cases over a large number of mild and severe total cases is a low proportion).
So, the big question is, are the 125 cases we’ve seen to date a fair representation of true spread of this virus, or are only `sickest of the sick’ showing up at the hospital?
If, as Hong Kong scientists predicted last week (see H7N9: Trying To Define The Size Of The Iceberg), there are actually twice as many cases out there than have been identified, then the `apparent’ CFR would be cut in half.
Of course, there could be uncounted deaths due to H7N9, attributed to other causes, as well. And they would serve to drive the CFR higher (if we knew about them, that is).
Figuring out the mortality rate from influenza remains a complex and controversial task. One that has sparked heated academic debate, which I’ve written about in the past:
- The H5N1 virus (see The Great CFR Divide)
- The 2009 H1N1 pandemic virus (see Lancet: Estimating Global 2009 Pandemic Mortality)
- Seasonal flu in the United States (see MMWR: Estimates Of Yearly Seasonal Influenza Deaths)
There are still debates – 95 years after the fact – over just how many people were killed during the 1918 Spanish flu. Estimates range from about 40 million, up to 100 million.
Similarly, the 1957 pandemic is said to have killed between 2 million and 4 million globally, and to have caused about 70,000 deaths in the United States.
But nobody really knows.
For now, it is safe to say that the H7N9 virus appears to produce severe, and all-too-often fatal illness in a significant portion of the people it infects.
But whether we are seeing 80%, 50%, or only 10% of the cases out there is unknown.
While we bandy numbers like 20% and 60% CFR (for H5N1) around, it is worth remembering that the worst pandemic on record – the 1918 Spanish Flu – had a highly variable fatality rate as it spread around the globe (see Study: Impact Of 1918 Pandemic In Mexico).
Its global CFR has been estimated at about 2.5%, but in 2006 in a Lancet journal (doi:10.1016/S0140- 6736(06) 69895-4) article cited as much as a 30-fold difference in mortality rates around the world:
Christopher JL Murray , Alan D Lopez , Brian Chin , Dennis Feehan , Kenneth H Hill
Excess mortality ranged from 0·2% in Denmark to 4·4% in India. Since there was some under-registration of mortality in India, total pandemic mortality could have been even higher.
Illustrating nicely just how variable the impact an emerging influenza virus can have, depending on things like geography, genetics, nutrition, healthcare standards, climate, locally circulating clades, and no doubt other factors that have yet to be identified.
While I expect we’ll get a better handle on the attack rate and CFR of this virus in the coming weeks, no one should get too comfortable with these numbers.
Influenza viruses have an amazing capacity to mutate and change (for better or worse), meaning the virus we see today may act little like the virus that we see next week, next month, or next year.
Stay tuned.
