Rehashed fake statistics of the week

From The Torygraph* (my numbering):

(1) One in 20 hospital admissions in England is due to smoking and treating illnesses related to the habit costs the NHS £5.2 billion a year, official statistics suggest.

(2) There were around 1.4 million hospital admissions in 2007/8 for diseases caused by smoking in people aged over 35, a report from the NHS Information Centre said.

(3) For people aged over 35, smoking accounts for five per cent of all hospital admissions and one in five deaths is attributed to smoking, the report said.

(4) An estimated 83,900 people died as a result of smoking in 2007/8 – 18 per cent of all deaths for adults aged 35 or over.

(5) The proportion of adults smoking has dropped slightly from 22 per cent in 2006 to 21 per cent in 2007 but many diseases take years to develop so ill health related to the high smoking rates of the 1970s and 1980s are only now manifesting.

*Ahem*

(1) It is miraculous that the 'cost' has doubled in the space of a year - from £2.7 billion on 6 October 2008 to £5.2 billion now. Either way, this isn't costing the NHS a single penny, the cost is borne by taxpayers. Seeing as VAT and duties on tobacco raise about £10 billion per annum and smokers' shorter life expectancy cuts about £12 billion off the annual state pensions bill, non-smokers are making a handsome net profit on the whole thing. And is "official statistics suggest" a polite way of saying they're lying through their back teeth?

(2) OK, 1.4 million is a Big Scary Number, but it's the one-in-twenty that's relevant, which doesn't seem too terrible. There are (say) ten million smokers, so that would mean one-in-seven have to go to hospital once a year (possibly plausible, I suppose). Alternatively, we could assume that each of their alleged 83,900 deaths is preceded by seventeen hospital visits (which seems a tad on the high side). But the 83,900 figure is itself plucked out of thin air, so of the ones they actually know about (having treated them) the number of visits must be much, much higher (and hence even less plausible).

(3) Or we could assume it's true that "smoking accounts for five per cent of all hospital admissions and one in five deaths" (query WTF does "attributed" mean?) and compare 20% (deaths) with 5% (admissions) and conclude that three-quarters of smoking related deaths don't "cost" the NHS a penny.

(4) & (5) These factoids suggest either; that nearly all smokers die of smoking-related diseases (i.e. 21% smoke and 18% die of it, allegedly); or that only half of them do (because "ill health related to the high smoking rates of the 1970s and 1980s are only now manifesting"); or that in fact only a third of smokers die of smoking-related diseases (83,900 out of 250,000 smokers who die each year of whatever cause) NB, 250,000 = ten millions smokers divided by average life expectancy of a 35-year old smoker of forty years, i.e. ten years less than normal life expectancy of a 35-year old, which is fifty years.

* Via Tim W.