The prime minister admitted last week that supplies of seasonal three-component influenza vaccine in some English general practices had run out. The health minister, Andrew Lansley, had to appear on Newsnight to defend using old stocks of the single swine flu vaccine to meet demand. Leaving the ordering of vaccine to individual GP practices instead of maintaining a central stock was clearly a flawed policy.

The swine flu virus – Influenza A (H1N1) 2009 – is behaving as expected: it’s back as the dominant seasonal flu. Maybe a little early, but so is the winter. It’s also behaving like all previous influenza-A strains in that some infections have been fatal; usually, but not exclusively, in people with pre-existing health problems. We’re much better at handling flu than we used to be. Severe infections can be treated in intensive care units; the last pandemic before swine flu was in 1968-69 when ICUs hardly existed, and the development of extracorporeal membrane oxygenation (ECMO) machines was a long way off. Essentially, ECMO does the work of patients’ lungs for them; most of the 14 machines in England are currently being used to treat flu cases. We have effective anti-virals. And vaccine development and delivery is now very quick: six million doses were given in response to swine flu without significant safety issues. But vaccine uptake in those who need it most has been disappointing.

Deirdre Hine’s ‘independent review of the UK response to the 2009 influenza pandemic’ was published yesterday. Her team was based in the Cabinet Office, which played a central role in the implementation of the pandemic plan. That’s one reason the word independent had to be in the title of her review. But there’s no doubting Hine’s independence. As the chief medical officer for Wales, she got into trouble during the BSE crisis for criticising the work of officials above her pay grade at the Department of Health in London. She was right then, and she has got it just about right this time too.

In recent times the usual response to scientific uncertainties about risk has been to apply the precautionary principle. Action is taken to prevent potentially dangerous events when there is no robust evidence about their likely magnitude, or sometimes even about the likelihood of their occurrence. With both Eyjafjallajökull's eruption and swine flu, pessimism and a heavy reliance on a very small number of historical events drove the policy response. It is on record that major damage has occurred on the few occasions when planes have flown through thick volcanic eruptions. In the century before swine flu there were only three flu pandemics, in 1918, 1957 and 1968. So when the precautionary principle led to the roll out of planned policies – zero tolerance for ash and very vigorous controls for the newish flu virus – the science was very imperfect.

At the Royal Mail you are sometimes made to come into work even when you are sick or injured, on threat of dismissal. It’s called an ‘Attendance Procedure’. They monitor your attendance. If you are off work for sickness or injury too many times, or for too long, you are given a Stage 1 warning. If you go over the limit a second time while still on the Stage 1 warning, you are given a Stage 2 warning. If you exceed the limit for a third time you are given a Stage 3 warning and threatened with dismissal. The limits are: either three absences in the space of a year, or one absence of three weeks or more. This is whether or not you are actually ill. All illnesses are assumed to be genuine, but all illnesses, no matter how desperate, also count towards your warnings. They don’t take any mitigating circumstances into account.

Swine flu has been spreading in Britain for three months. The virus has got about quite well, although the great majority of infections have been mild. Until two weeks ago reassurance about our preparedness for a pandemic was the order of the day. But the media tone changed with the reporting of the deaths of six-year old Chloe Buckley and Dr Michael Day. Chloe was said to have been infected with the virus but didn’t have the ‘underlying health conditions’ usually present in fatal cases, and Day was the first healthcare worker to have a lethal infection. Coincidentally, the tenor of official public pronouncements altered too. The chief medical officer for England mentioned the possibility of 65,000 deaths. On television he was quick to qualify: that figure was a worst-case scenario, necessary for planning, not a prediction. But the number, not the caveat, got the publicity. There was also a change in the way that case statistics were announced, with a shift from laboratory confirmation to estimates based on GP consultation rates and clinical diagnoses. The overnight five-fold increase in ‘cases’ was inevitable. Lab tests tend to underestimate, and consultation rates increase because of the media coverage.

The first death caused by swine flu virus outside the Americas occurred in Scotland on Sunday. The announcement generated more media interest than the declaration three days before by the World Health Organisation that the spread of the virus had moved into pandemic mode. But the declaration was expected and generated less fear than anticipated. The public can see that in Britain the virus is doing well – which is all that was needed to meet the pandemic criterion of sustained community spread in a region outside the Americas – and the message that the virus is mild is also well established, tempering the notion that the word 'pandemic' carries lethal overtones. But this means that a death requires explanation. There is no such thing as a naturally avirulent influenza virus. Even the mildest ones that infect humans can kill. They do it routinely every winter.

The spread of the novel influenza A(H1N1) virus through North America is nearly complete. Only three continental US jurisdictions (Wyoming, West Virginia and Alaska) and three Canadian provinces or territories (Newfoundland, Nunavut and the Northwest Territories) haven't reported cases. Its progress elsewhere is still slow, however. Japan (163 cases), Spain (103), the UK (102) and Panama (54) lead; vigorous containment is still the order of the day in the UK. But unless the North American epidemic slows soon, the continued export of the virus – in the coughs and sneezes of infected travellers returning home (particularly to the southern hemisphere, which is just entering its flu season) – has a good chance of defeating all best-laid plans. And it is doing well in Japan.

Influenza virus has only eight genes. The molecular structure of the most important proteins they code for is known in intimate detail. The coming and going of its epidemics have been studied by statisticians continually since the 1840s. But predicting pandemics remains a fools’ game. It falls into the category of Alvin Weinberg's 'trans-science' – a question of fact that can be stated in the language of science but is unanswerable by it. Weinberg’s examples focused on the impossibility of predicting the probability of extremely improbable events. There have only been three influenza pandemics in the last century: in 1918, 1957 and 1964. The uncertainty is massively amplified by evolution – the random and frequent genetic mutations and the swapping of genes between bird, pig and human viruses.