Nigel Hawkes: How serious is the threat? They're probably just guessing...

A month ago the Government warned us that by the end of August 100,000 people would be getting swine flu every day. Yesterday it produced figures showing that the disease has apparently peaked, with only 30,000 new cases last week. How did they get it so wrong?

The trouble with warning people about the dangers of swine flu is that you're damned if you do, and damned if you don't. Striking the right balance is impossibly difficult. There are always plenty of commentators ready to scent a scare, but following their insouciant advice is not an option. "Calm down, everybody, it's only a couple of days feeling a bit peaky" was never the first sentence likely to spring to Sir Liam Donaldson's lips.

That said, it's worth asking if the message has been right, and if it has been based on sound evidence. It's also worth asking why the Scottish Government has done so much better in providing timely and complete data than the Health Protection Agency in England.

Take, for example, the estimates of the numbers of new cases – 110,000 in England in the week up to 27 July. How exactly were these figures reached? After prompting from the Royal Statistical Society, the HPA provided a limited explanation of how it analysed swabs from patients who have consulted their GP about flu-like illnesses.

But it gave no indication of how many swabs had been taken, which makes it impossible to gauge how accurate its predictions are. To provide an answer to an accuracy of plus or minus 5 per cent would have needed 700 swabs across England. The suspicion is that the number was much lower than this. If not, why did the HPA not disclose it?

Without proper data collection, we might as well guess – which is more or less what statisticians suspect has happened. Last week Professor David Hand, President of the RSS, wrote to Sir Liam demanding better monitoring.

Sir Liam agreed this was a good idea, but added that the NHS already faced "a heavy burden" in gathering information. He might have added, but didn't, that the great bulk of this is for accounting, not clinical purposes – which is why, despite mountains of data, we still have a very poor handle on how well the NHS performs in treating disease.

The latest HPA figures suggest that swine flu has peaked, with just 30,000 new cases last week. But once again, the HPA has not explained how this estimate was arrived at. And while it is good news that hospital deaths have peaked, we still haven't been told the dates on which the 36 victims of H1N1 actually died.

Did the Government exaggerate the threat? Certainly its early guess of 100,000 cases a day by the end of August looks wildly out. That, too, was never explained properly: nor, it now appears, was it right. Greater frankness over what it knew, and what it was merely guessing, might have helped people to make their own minds up how great a threat they faced.

Nigel Hawkes is Director of Straight Statistics (