So, this rise in drink driving deaths then

The number of deaths in drink-drive accidents soared last year, according to provisional government figures.

An estimated 290 people were killed in such accidents in Britain in 2012 – around 25% more than the figure of 230 in 2011, the Department for Transport said.

The 2012 figures were made worse by the fact that the 2011 total was the lowest since records began in 1979.

The 2012 figures showed that the 290 deaths represented 17% of all reported road deaths.

There were 250 drink-drive accidents resulting in deaths in 2012, compared with 220 in 2011. Overall, the number of accidents involving drink-driving last year was 6,680 – fractionally down on the 2011 figure of 6,690.

Three hypotheses

There really has been such a rise in deaths as a result of something or other having been done: cuts to the ad budget against drink driving perhaps.

This is within the bounds of normal statistical deviation.

They\’ve changed the definition.

I really don\’t know which is most likely: anyone who does have a clue want to comment?

24 thoughts on “So, this rise in drink driving deaths then”

  1. Fourth theory: Some external influence affected the numbers driving. For instance I could hypothesise that It could be the smoking ban making people drink more often in smokey-drinkies (friend’s homes) where the environment removes the inhibition to moderate their drinking. So they drive home more drunk than if they went to a pub where they paid and knew how many pints they had drunk. Smokey-drinkies could also be encouraging home brew which will be at a strength different to the beers normally drunk in a pub.

    So all in all unintended consequences. The thing that always bites the progressive nannying fussbuckets desire to control the population and how they live.

  2. William Connolley

    To say whether this is anything but normal statistical variation, you need to dig out the series for a reasonable number of years. How many depends on what it looks like, but 10 must be a bare minimum. Then you’d start by plotting it and eyeballing it for some hint of what the variation looks like.

  3. The trend in ‘killed and seriously injured’ had been down for many years – last year it rose.

    Here in Bradford the rise was entirely down to the increased number of KSIs involving cyclists. It would be interesting to ask how many of these DD deaths involved a cyclist being hit by an OTT driver.

  4. Looking at the graphs – I’d say chucking out time and club opening hours and younger people contribute to the high points…

    I’d be interested to see how the numbers compare with other countries and a breakdown of the geographical areas. In other countries I’ve seen late night “party buses” full of intoxicated youngsters being delivered home in a sozzled but good natured fashion – not sure that’d work with what I perceive to be the aggressive late night drinking culture that predominates in most UK cities and towns….

    The public elf folk at councils and plod PR will no doubt use the numbers tp hector us all.

    There is also the matter of who collects the numbers and what evidence rules apply – a subject that many public health professionals are uncomfortable to have looked at.

    Ass a completely subjective aside – drunken drivers reported in my local papers seem to have *lots* of eastern European names – far beyond what I’d expect for their proportion of the population – that might just be reporting / perception bias….

  5. Tomo, the rise due to immigration is a nice fifth theory, and another one that is the result of unintended consequences of government action.

  6. It is your second hypothesis. In the long-term trend, this figure is within the bounds of normal statistical deviation. It is helpful to note that 2010 and 2011 had unusually few drink-driving accidents. Here are the figures:

    1979: 1,640
    1989: 810
    1999: 460
    2006: 560
    2007: 410
    2008: 400
    2009: 380
    2010: 240
    2011: 230
    2012: 290
    There was a drastic fall between 2009 and 2011, so now there has been relatively drastic increase. I would suggest looking at a rolling average to mollify these drastic shifts.

  7. Anthony Masters is nearly right – the 2012 figure is not too far off the downward trend line – but the 2010 and 2011 data does not look like normal random variation. I suspect that the reason why those two years had sharply lower death rates was linked to the high price of petrol and reduced mileage driven.

  8. I, too, would be interested to learn the nationalities of those killed or injured. It wouldn’t surprise me in the least to learn that many of them were Eastern Europeans. Had a couple killed round here over the last couple of years on the roads.

  9. On the vodka’d up driver theory … I suspect – but don’t have the time / inclination to follow up that the there are a significant minority of easterners that don’t give a fuck and just carry on their old ways after arrival – anecdotal evidence locally is that quite a few Poles in particular are annoyed that the gits they thought they’d left behind are turning up here…

    As for Romania … no contest

  10. It might be worth considering if there is a weather factor involved. More ice around in winter evenings might increase the base chance of an accident compared to other milder winters.

  11. In a country of 38m registered drivers, 230 is so small that any variation will appear disproportionate. I mean let’s say that in one accident, two drunk blokes plough into two other people and they all die. That’s almost a 2% increase right there.

  12. Thanks for the numbers. Its hard to be sure from just those, but (just for fun, doing my best):

    The “fatals” look vaguely level (with a variation of ~20/30) except for a sharp drop in 2007, and then in 2010. An increase of 30 in 2011 (if you were to take 2010 as representing a “new low level”) would be a bit more than looks like previous variation. But then again, 2011 is marked as provisional.

    Vaguely supporting the “provo” theory is to look at the “serious” line on the graph, which are both up in 2011, despite a long-term trend downwards in all other years.

    Oh, hold on. The graph linked to is the *2011* report, but Tim is quoting the Graun about 2012. So Timmy has “An estimated 290 people were killed in such accidents in Britain in 2012 – around 25% more than the figure of 230 in 2011”

    So the 280 provo number for 2011 was revised down to 230 in the final numbers, so the (revised, presumed final) trend to 2011 is downwards. That rather suggests that the 290 (presumed provisional) number for 2012 might be revised down too.

  13. Intrigued to how you define a drunk-driving related death downwards, considering the evidence has to be gathered at the time – unless the provisional figures are actually those cases assumed and not proven, in which case, why release them?

  14. Anecdata i’ve heard is that if any of those admitted to hospital are drunk, then the incident is recorded as drink related. In the case of say a single vehicle incident where a drunk passenger is killed but the driver is teetotal, is this recorded as drink driving, similarly are incidents involving drunk pedestrians recorded as “drink related driving”?

  15. Another possibility:

    We normally look at road accidents in terms of KSI (Killed & Seriously Injured), not just K alone. If the ambulance gets to the scene quickly enough, we can convert a potential K to a mere SI. Conversely, if there are cuts to ambulance services or to the NHS in general, more of the SI might end up K because they don’t get medical treatment quickly enough. Even the harshest critics of the NHS would probably agree that more cuts = more deaths (absent any systemic reform).

    The full report (PDF link) provides more detail:

    – Killed: up 25% (to 290)
    – Seriously Injured: down 5% (to 1,210)
    – Slightly Injured: up 1% (to 8,500)

    So the good news is that KSI is down yet again. Well done everyone.

  16. @ Nessimmersion
    It would appear that if a motorcyclist who is over the limit is carefully stopped at a red light and a sober driver trying to jump the lights runs him over that is classified as a drink-related death.

  17. The Times report says that the provisional number for 2011 was 280 (12% increase) as in the report referenced by SadButMadLad, but the final figure was 230 (down 4%). Will the 2012 provisional figure be similarly adjusted?

  18. “It might be worth considering if there is a weather factor involved. More ice around in winter evenings might increase the base chance of an accident compared to other milder winters.”

    Last year was also abnormally wet in summer so I wonder how many people decided to drive home after a few beers rather than walk?

  19. The correlation is causality thing.

    To reduce deaths/injury in ‘drink/drive’ accidents, increase the blood alcohol threshold where an accident becomes a ‘drink/drive’ accident. Or to increase such accidents, lower the threshold.

    There are a number of factors which cause most road accidents, it is rarely just one. To isolate just one factor like ‘speed’ or ‘booze’ because it can be directly measured and claim it is uniquely responsible is idiotic and not scientific.

    What the figures explain is we don’t know. It is the same with the Global Warming thing. Once the correlation stops, nobody can explain why but instead of questioning whether their evidence-free assertions are right, they cast around making more guesses to explain why really they are right and the data somehow must be wrong.

    And of course anyone who mentions this must go stand in the corner with all the other swivel-eyed right wingers who have dared challenge ‘the consensus’ (All bow and give homage.).

  20. Also wouldn’t mind knowing at what point Drink Driving (eg over the legal limit) becomes ‘drink related’ (eg under the legal limit, but crashed and dead and test positive for alcohol).

    Willing to bet that the latter gets included.

  21. @ sam
    The official release linked above stated that 62% of the deaths were drivers or riders with alcohol levels above the limit – the others were passengers or riders/drivers below the limit.
    Note that someone is testing all driver/rider corpses to see if their blood alcohol level is over 80mg. Ugh!
    The major downward revision in 2011 figures is most likely due to the number of dead passengers who tested positive after their sober driver was involved in an accident e.g. black ice.

  22. The provisional data are estimates based on the number of surviving drivers and [motorbike] riders who test positive. The final data are based on Coroners’ reports, which take much longer to become available. (Methodology here). The provisional data for drink-driving fatalities are therefore highly unreliable extrapolations, because dead drivers and riders are unable to blow into the machine. The prudent thing to do would be to wait for the final data before trying to explain any changes.

    John B: one can tell the swivel-eyed climate loons by their strange habit of intruding their eccentric views on the subject into practically any discussion. One about road-accident statistics, for example.

Leave a Reply

Your email address will not be published. Required fields are marked *

So this rise in drink driving deaths then

Leave a Reply

Your email address will not be published. Required fields are marked *