Intended for healthcare professionals

Editor's Choice

Predicting and preparing for pandemic flu

BMJ 2009; 339 doi: (Published 23 July 2009) Cite this as: BMJ 2009;339:b2988
  1. Fiona Godlee, editor, BMJ
  1. fgodlee{at}

    “Never make predictions,” said the American baseball player Casey Stengel, “especially about the future.” But we have to forecast if we are to plan. For more accurate forecasting in this flu pandemic, it’s essential for us to refine our methods and to understand their limitations. This is what Tini Garske and colleagues seek to do in their fast tracked article in this week’s journal (doi:10.1136/bmj.b2840).

    The world needs to know how many people are likely to die from the new flu virus. Data emerging from different countries offer widely different pictures. The initial reports from Mexico suggested high fatality ratios among those infected, sparking a sense of panic. Mexico’s fatality ratios are still among the highest in the world, but Gaske et al explain that the cause is unlikely to be an especially virulent version of the virus—instead, under-reporting of mild cases could explain the trend. If we take these cases into account, the true case fatality ratios could be much lower and more in line with those for seasonal flu. But because there’s no herd immunity to this new virus, many more people will catch it and the absolute numbers of cases and deaths will almost certainly be much greater than for seasonal flu, they say.

    The authors identify a second pitfall in forecasting for this pandemic: the delay between onset of symptoms and death. This “censoring bias” means that at any one time there will be people who will die from the infection but who are currently still alive, causing an underestimate of the case fatality ratio, especially in the early days when incidence is growing exponentially. Then as the pandemic unfolds, case fatality ratios will grow, raising concerns that the virus is becoming more virulent.

    The only answers, say the authors, are good data and rigorous methodology. Their wish list is long: close monitoring of the first few hundred cases, good ongoing surveillance for symptomatic cases, large scale testing in well defined populations, prospective household surveys to estimate attack rates for mild disease, and serological testing to assess rates of asymptomatic infection.

    It’s good to see that the UK seems to be doing well in its ascertainment of cases. But how well are we managing the flu pandemic? Also pretty well, says Roy Anderson in his editorial (doi:10.1136/bmj.b2897), with good pre-pandemic planning, good stocks of antivirals and pre-ordering of vaccine, and a sensible and timely shift from prevention to treatment of those most at risk.

    This rosy picture may not fit with people’s experiences either as patients (doi:10.1136/bmj.b2969) or as health professionals overwhelmed with conflicting information and advice. Anderson acknowledges that the detailed logistics could have been better and need improving. The new flu service to be launched this week in England should take the pressure off primary care (doi:10.1136/bmj.b2932). Questions will remain, including how to decide which doctor should be the “flu lead” in a practice. Daniel Sokol outlines the options and the ethical dilemmas (doi:10.1136/bmj.2567). As our contribution to bringing clarity, we are launching a pandemic flu website ( with daily updates and links to the most reliable sources of information. We would also like to hear from you on our flu forum on doc2doc (


    Cite this as: BMJ 2009;339:b2988

    View Abstract