Or the UK reported infection rate is wrong. What you wrote there bothered me a lot, and I've several times tried to authenticate the figures, but without much success. Part of the problem is that the rise in reported infection rate coincided with differences in methodology. Early infections were confirmed by a local doctor doing blood tests. Now they have a "swine flu hotline" here, which you can phone if you think you have it. You get diagnosed over the phone and are then authorised to get Tamiflu.

The diagnosis is unreliable. Apparently the official figures are 5% of the hotline figures: ie 95% of the people given Tamiflu in England don't need it (and possibly do have some other illness which needs some other medication they aren't getting). Sometimes they send out swab kits which the patient is supposed to administer and send back, and those provide a statistical check - that's where the 5% comes from. Except it's highly variable - eg 10% some weeks, 5% others. There is probably bias in the samples. People who are feeling better are less likely to send back their swabs.

It still seems like there was a surge in July, and a tailing off in August. However, I'm now getting that from the death rate - presumably death statistics are a lot more reliable. For example Wolfram|Alpha shows a steep rise in the second half of July. That must have been preceded by a rise in infection rate, and probably it was that rise which triggered the Hotline launch, but I wouldn't like to put any figures on how big it actually was, and I don't trust the NHS figures.

(PS this is really just England and Wales - Scotland has a different system.)