This claim has been widely repeated this week in many newspapers and news sites (e.g. arstechnica). Here is how the Daily Telegraph repeats it:

Three in four people in accident and emergency units at the weekend have been admitted because of alcohol, new figures have shown.

The new research, which was carried out at an A&E unit in the North East of England shows the huge burden that excess drinking places on the NHS.

Given the known pattern of attendance at emergency units (e.g. peak hours in the UK are 9-11am and weekends are quieter than weekdays) this seems somewhat suspicious.

Is it true?

Sub questions: what is the method used to derive the numbers? Are they consistent with what else we know about emergency attendance? Are other countries different?

  • Did you read the Telegraph article and follow its links to the scientific paper the report was based on? Dec 23, 2015 at 21:18
  • 3
    @DJClayworth Yes I did. But it is the newspaper claim that has attracted the attention and this site should be about analysing the reliability of the headline. Does it represent what the paper said, or not? Where is the discrepancy? Is the paper consistent with other research? Most people won't read the original research or understand it. Skeptics.SE can help. (feel free to answer before I do).
    – matt_black
    Dec 23, 2015 at 21:39
  • 1
    Don't forget that there are regional, cultural, geographical and social factors involved too. All hospitals are not alike. If the hospital in question is in the centre of a large city, near a lot of drinking venues, and that city has high rates of unemployment and social problems, (or maybe frequent visits by sailors on shore leave) then you'll get a lot of drunk admissions. If you took a hospital of the same size in a quiet, affluent rural area you'd get very few. Dec 29, 2015 at 23:10

1 Answer 1


No. Unless the weekend is redefined to cover a few quiet hours on friday and saturday night.

The sensational newspaper headlines are based on a relatively reliable piece of research which is consistent with previous research done both in the UK and in some other countries.

The language used in the research paper and the way data is presented somewhat obfuscate the results and lead to a temptation to quote sensational but insignificant results rather than the overall conclusion (of course, this might be a deliberate attempt by the paper's authors to get attention in newspapers).

The abstract reports the following:

The prevalence rates of alcohol related attendances were 12% and 15% for the retrospective and prospective cohorts, respectively. Prospectively, the rates ranged widely from 4% to 60% across week days, rising to over 70% at weekends.

This requires some explanation. The 12% is the ratio of people attending who have taken alcohol to total attenders over the observation periods. The higher 15% number is adjusted to account for future attendance outside the observation periods (and basically means that people who have attended once with alcohol-related issues are more likely to re-attend within a year). The vastly higher numbers are not the percentage of alcohol related attenders at the whole weekend, but the maximum ratio for any hour of the day. Alcohol-related attendance is focussed between the hours of midnight-4am which are also the quietest hours in typical emergency departments. So the percentage is a percentage of a low base and not at all representative of the whole day.

Here are two charts from the paper that illustrate this:

Hourly attendance and % alcohol related

To summarise the results: even at the weekend, alcohol-related admissions are low when general attendance is high, but high when attendance is low. So the proportion of alcohol-related attendance varies across the day with a peak in the early morning of about 70%. For the whole weekend, though, only about 15% of attendance is related to the demon booze.

Quoting the maximum hourly rate of alcohol-related attends greatly exaggerates the problem (and guarantees, perhaps, press attention). The paper actually makes it quite hard to see the complete picture for the whole weekend (the charts are the only place the data are presented in a clear and complete form).

The numbers may also be exaggerated by the choice to include other attendance by the same people outside the sample (the "prospective" cohort). It isn't clear from the method they use what the base is for the larger percentages in this cohort. It may exaggerate or it may not; it is hard to tell.

There is another problem with the paper's method. They use a positive breath test to categorise whether an attendance is alcohol-related. This doesn't clearly link the reason for attendance to to consumption of alcohol. This might not be a cause of exaggeration, especially in Newcastle which has a heavy weekend drinking culture, but it is impossible to be sure.

Summary The high numbers quoted in headlines are highly selective. It might be reasonable to say that 15% of attendances at weekends are from people who have recently consumed alcohol. In some or many of these cases the attendance will be indirectly caused by alcohol (e.g. falls). The 70% number comes from a highly selective view covering only a small percentage of the weekend where few people attend A&E anyway.

Update BBC Radio 4's weekly programme More or Less (which studies dubious claims in headlines) mostly agrees with the analysis above. The podcast is here (section starts at about 15m:30s). The programme clarifies what is meant (talking to the original researchers) and criticises them for fuzzy language that could be easily misinterpreted.

  • As you mentionned it the paragraph preceeding the summary, the proportion of alcohol-related attendance is also highly dependent on what is measured by studies. Some studies are based on systematic detection of alcohol rates, since other are based on clinical diagnosis. In the first case, 20 to 40% of emergency admissions are caused by alcohol. In the second case, the percentage falls to 4 to 10%. *source: Admission in emergency units for acute alcohol intoxication in France, in 2011. The OSCOUR® network contribution - page 196.
    – Graffito
    Dec 27, 2015 at 21:40
  • 1
    Did you mean "midnight to 4am" or "midnight to 4pm" ?
    – gnasher729
    Jan 27, 2016 at 16:26
  • @gnasher729 4am. Well spotted. Now corrected.
    – matt_black
    Jan 27, 2016 at 16:36
  • Great work here. I looked at the paper and couldn't make head or tail of it. Jan 27, 2016 at 17:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .