When we want to `cut to the chase’ when describing the impact of a particular infectious disease, we almost always go first to its CFR or Case Fatality Ratio; the percentage of people who contract the disease, and then die.
To calculate the CFR, you really only need to know two things; the total number of cases (the denominator), and the total number of deaths due to the illness (the numerator).
Unfortunately, in actual practice, both numbers can be maddeningly difficult to deduce. And without qualifiers, the CFR numbers that get bandied about are almost always misleading.
The example that anyone who follows avian flu is familiar with is the astronomical CFR for H5N1. With 630 cases reported globally since 2003, and 375 deaths, a quick calculation provides:
And so the CFR of 60% is widely used. But is it correct? Or even close to reality?
The problem is, with just about every illness or infection, only a fraction of the cases – usually the most severe – are identified. And it doesn’t matter whether we are talking about seasonal influenza, West Nile Virus, Salmonella, or avian flu.
As the pyramid chart above indicates, only a tiny fraction of infectious disease cases are actually reported to health authorities. - Credit CDC
With nearly every infectious disease, we see a wide spectrum of clinical illness, which can range from mild (or even asymptomatic) to severe. Even with SARS in 2003, retrospective testing found asymptomatic cases.
So how confident are we that the denominator in the above equation (630) accurately represents the total number of H5N1 cases since 2003?
And and how confident are we that numerator (375) accurately counts the number of deaths due to the virus?
Again, not very.
What we really know, is that among those cases that were sick enough to be hospitalized, tested, and diagnosed – 60% died.
The limited serological evidence we have suggests there may not be a lot of `missed’ cases in the general population (see The Great CFR Divide & Revisiting The H5N1 CFR Debate), but it would only take a few hundred unreported cases to cut the CFR in half.
Which brings us to an article appearing in The Lancet today, that attempts to quantify the relative severity of the H7N9 virus compared to H5N1 and the 2009 H1N1 pandemic virus.
Hongjie Yu MD, Benjamin J Cowling PhD, Luzhao Feng MD, Eric HY Lau PhD, Qiaohong Liao MD, Tim K Tsang MPhil, Zhibin Peng MD, Peng Wu PhD , Fengfeng Liu MD, Vicky J Fang MPhil, Honglong Zhang MD, Ming Li ME, Lingjia Zeng MSc, Zhen Xu MD, Zhongjie Li MD, Huiming Luo MD, Qun Li MD, Zijian Feng MD, Bin Cao PhD, Weizhong Yang MD, Dr Joseph T Wu PhD, Dr Yu Wang PhD, Prof Gabriel M Leung MD
Of 123 patients with laboratory-confirmed avian influenza A H7N9 virus infection who were admitted to hospital, 37 (30%) had died and 69 (56%) had recovered by May 28, 2013. After we accounted for incomplete data for 17 patients who were still in hospital, we estimated the fatality risk for all ages to be 36% (95% CI 26—45) on admission to hospital. Risks of mechanical ventilation or fatality (69%, 95% CI 60—77) and of admission to an intensive care unit, mechanical ventilation, or fatality (83%, 76—90) were high.
With assumptions about coverage of the sentinel surveillance network and health-care-seeking behaviour for patients with influenza-like illness associated with influenza A H7N9 virus infection, and pro-rata extrapolation, we estimated that the symptomatic case fatality risk could be between 160 (63—460) and 2800 (1000—9400) per 100 000 symptomatic cases.
Human infections with avian influenza A H7N9 virus seem to be less serious than has been previously reported. Many mild cases might already have occurred. Continued vigilance and sustained intensive control efforts are needed to minimise the risk of human infection.
The main finding here is that the mortality rate for those hospitalized with H7N9 was roughly 36%, and that risk increases with age. This rate is greater than that seen in hospitalized cases with the 2009 H1N1 virus, but lower than we’ve seen with H5N1.
The authors argue against trying to come up with a `one-size-fits-all’ CFR for the H7N9 virus, and instead devised a two-stage approach; estimation of fatality risk among hospitalized patients and then estimation of number of symptomatic infections.
Calculating the fatality risk among hospitalized patients was fairly straight forward, but the second part; estimating the likely number of symptomatic H7N9 infections across China as of May 28 was considerably less so.
For this reason the authors listed a number of limitations to their study, particularly when it came to attempting to extrapolate the total number of cases. The authors provide a wide range of possibilities, writing:
Our estimate that between 1500 and 27 000 symptomatic infections with avian influenza A H7N9 virus might have occurred as of May 28, 2013, is much larger than the number of laboratory-confirmed cases.
They further calculated that the `symptomatic CFR’ of the virus probably runs between .16% and 2.8%. The authors warn that this estimate relies on a number of `simplifying assumptions’, and should therefore be viewed cautiously.
As I’m not a statistical wizard, I’ll let others with a background in mathematics dissect and analyze their methods.
Lest anyone scoff at an estimated CFR of under 3%, I would remind them that the 1918 Spanish Flu – which killed somewhere between 50 and 100 million people – was estimated to have a CFR of roughly 2.5%.
Often, epidemiological data points like the CFR, CAR (Case Attack Rate), R0 (basic reproductive number) are only refined in retrospect, usually after years of analysis.
Complicating matters, these numbers are rarely static over time, or geography.
A second Lancet report (which shares a number of authors with the first study) - after comparing the epidemiological differences between H5N1 and H7N9 human infections - warns that public health officials should be preparing now for a possible resurgence of the H7N9 virus later in the year.
Comparative epidemiology of human infections with avian influenza A H7N9 and H5N1 viruses in China: a population-based study of laboratory-confirmed cases
Benjamin J Cowling PhD, Lianmei Jin MD, Eric HY Lau PhD, iaohong Liao MD, Peng Wu PhD, Hui Jiang MD, Tim K Tsang MPhil, Jiandong Zheng PhD, Vicky J Fang MPhil, Zhaorui Chang MD, Michael Y Ni MPH, Qian Zhang MD, Dennis KM Ip MPhil, Jianxing Yu MD, Yu Li MD, Liping Wang PhD, Wenxiao Tu MD, Ling Meng MD, Joseph T Wu PhD, Huiming Luo MD, Qun Li MD , Yuelong Shu PhD, Zhongjie Li MD, Zijian Feng MD, Weizhong Yang MD, Yu Wang PhD, Prof Gabriel M Leung MD, Dr Hongjie Yu MD
The differences in age distribution of patients with laboratory-confirmed infection with H7N9 and H5N1 are intriguing; presumably, immunity associated with different histories of influenza virus exposures has an important role in addition to differences in exposure patterns. Although we have reported the fatality risk for patients admitted to hospital, the symptomatic case-fatality risk remains to be established and a large portion of the “clinical iceberg” of infection might have remained undetected so far.
The warm season has now begun in China, and only one new laboratory-confirmed case of H7N9 in human beings has been identified since May 8, 2013. If H7N9 follows a similar pattern to H5N1 (figure 2B), the epidemic could reappear in the autumn. This potential lull should be an opportunity for discussion of definitive preventive public health measures, optimisation of clinical management, and capacity building in the region in view of the possibility that H7N9 could spread beyond China's borders.