Wednesday 4 September 2019

Junk alcohol statistics laid bare

Geoffrey Rose, the doyen of modern public health, wrote in The Strategy of Preventive Medicine (1992) that:

..from the average alcohol intake of a population one can predict precisely the number of heavy drinkers.

He borrowed this idea from Sully Ledermann who had come up with the single-distribution theory in the 1960s. Rose's highly influential book essentially applied Ledermann's hunch to a broad range of other health issues, in each case concluding that the solution to extreme behaviour lay in changing average behaviour.

Evidence for this theory was always scant. At best, it came from the tautological observation that averages tend to be higher when there are many large numbers (eg. per capita alcohol consumption is higher when there are many heavy drinkers to push up the average).

Despite its logical and empirical shortcomings, the single-distribution theory became popular in 'public health' because it implied support for the 'whole population approach' in which blunt policies are applied to all consumers rather than targeted policies being applied to at-risk individuals.

As John Duffy and I showed in this IEA report five years ago, the theory hasn't stood up in the real world. There are many examples of per capita alcohol consumption going down while alcohol-related harm goes up and vice versa.

This seems to have been the case in the UK in the last fifteen years, as even the Sheffield University alcohol researchers had to acknowledge in this recent study.

This conclusion and the findings above therefore present major challenges to collectivity theory and the total consumption model. They suggest that a complex mixture of strong age and cohort effects, hard and soft collectivity, and some polarised trends underpin the findings of general collectivity observed when examining simple time trends in previous studies. These more complex findings do not imply a consistent relationship between population-level consumption and harm over time.

Today, the World Health Organisation published its 'status report' on alcohol in Europe. It was launched with a press release that lamented Europe having the highest rate of alcohol consumption in the world, and called for the usual grab bag of temperance policies. 

“Alcohol consumption has decreased in many European countries, but progress is grinding to a halt. Policy-makers need to implement the strategies we know are effective, such as increasing prices, limiting availability and banning advertising. With as many as 800 people dying every day in parts of the Region due to alcohol-attributable harm, we must do more to continue the fight,” said Dr Zsuzsanna Jakab, WHO Regional Director for Europe.

The report contains two points of interest. First, it provides more figures that do not fit the single-distribution theory. It finds that the European region saw a statistically insignificant decline in annual alcohol consumption from 11.5 litres per adult to 11.3 litres between 2010 and 2016. In other words, levels of consumption stayed essentially unchanged. If you believe in the whole population approach, you might expect the number of alcohol-related deaths to stay essentially unchanged. In fact there was a substantial decline...

• the age-standardized alcohol-attributable death rate decreased from 35.5/100 000 to 30.5/100 000, a proportional 14.1% reduction

• the absolute number of alcohol-attributable YLL [years of life lost] decreased by 11%, from 8.6 million to 7.6 million

• the age-standardized alcohol-attributable YLL rate decreased from 1234/100 000 to 1016/100 000, a proportional 17.6% reduction

Confused? You will be - because the WHO doesn't reckon the decline is real.

Socioeconomic inequalities are an important determinant of diminished life expectancy. At the same time, trends in alcohol-attributable mortality are greatly impacted by trends in overall mortality; if the death rates on which alcohol operates go down, alcohol-related death rates inevitably will also go down, even if there is no change in alcohol consumption.
This means that the observed changes seem to be driven mostly by overall improvement in health and health care for the EU+ population rather than by reduction in prevalence of heavy episodic drinking and alcohol consumption in current drinkers.

Take that in. It is the nearest you will get to the WHO admitting that their estimates of alcohol-related deaths, which are based on alcohol attributable fractions (AAFs), are worthless.

The AAF system simply hives off a certain proportion of deaths from different diseases and classifies them as alcohol-related. No clinician ever diagnoses them as 'alcohol-related'. It all comes from a dubious interpretation of epidemiological studies which is then applied to data from spreadsheets.

This is one of the reasons why there has been an implausibly large rise in the number of 'alcohol-related' hospital admissions in the UK in the last fifteen years despite alcohol consumption falling. The overall number of hospital admission has risen by 50 per cent in that time, largely thanks to the ageing population. There was bound to be a rise in the number of 'alcohol-related' admissions recorded, even if there was no actual rise.

In today's report, the system has backfired on the WHO and so they have had to admit the truth, or at least hint at it.

No comments: