Tuesday, 13 December 2016

Public Health England's alcohol report: part 2

Last week, I wrote about the Public Health England report on alcohol which was released to the media with a lie that had to be retracted and which contained many more lies within. The report was shamelessly biased towards nanny state intervention and this is hardly surprising since the 'expert advisory group' and peer review team included the likes of Katherine Brown (from the organisation formerly known as the UK Temperance Alliance), anti-alcohol fanatic Ian Gilmore (Alcohol Health Alliance), Petra Meier (Sheffield Fantasy Modelling Club), Tim Stockwell (minimum pricing campaigner), Peter Anderson (temperance stalwart) and Gerard Hastings (socialist lunatic).

I covered some of the more egregious lies in the previous post. This post is about some of the other tactics used to mislead the public.

Doubt is their product

In common with the Chief Medical Officer's report in January, the PHE document embraces every epidemiological study that finds an association between alcohol and harm but treats the health benefits with extreme scepticism.

And so, for example, when it comes to the harms of drinking, findings from observational studies are considered to provide strong evidence.

There is strong evidence for an association between alcohol consumption and cancer including cancers of the oral cavity and pharynx, oesophagus, female breast, colorectum, larynx, liver, stomach, pancreas, lung and gallbladder. For certain cancers, including breast cancer, any level of drinking increases your risk so there is no ‘safe’ level of drinking.

But when it comes to the evidence that moderate drinking is good for health, epidemiological papers are no longer described as observational studies or cohort studies but become mere 'health surveys' and the authors emphasise all the problems that are inherent to this type of evidence. In particular, they focus on the zombie argument about 'sick quitters'.

Health surveys typically ask about current drinking levels and the classification of ‘non-drinkers’ can include former drinkers, occasional drinkers and people who have never consumed alcohol. This group of ‘non- drinkers’ is not a reliable comparison group as it may include individuals who never started drinking for a variety of reasons which may make them more susceptible to poor health (eg a lifelong disability), and former drinkers who may have stopped drinking due to poor health.

I've written enough about the tired old canard of 'sick quitters' so here's what the Department of Health working group said about it 21 years ago...

A key issue previously complicating the epidemiological data has been the so called “sick quitter” hypothesis outlined by Professor A G Shaper. This identified a number of relevant issues, but in particular it attributed higher mortality in people who did not drink to the effect of those who had given up drinking because of ill health. In addition, Professor Shaper has suggested that abstainers may carry a greater burden of ill health than drinkers and consequently never take up drinking, or that abstainers are constitutionally predisposed to high risk of disease, in contrast with moderate drinkers who are similarly predisposed to low risk of disease. However, a number of studies since 1987 have controlled for these factors so that we believe Professor Shaper’s reservations cannot be considered as a major explanation of the cardio-protective effect. Other confounding factors such as tobacco use, obesity, diet and age have now been controlled for in enough studies to allow us, on the basis of expert testimony, to be confident that the basic protective effect for CHD by alcohol is scientifically valid.

And when it comes to the benefits - but not the risks - of drinking, we are reminded that people underestimate the amount they drink.

Furthermore, health surveys generally underestimate alcohol consumption due to the exclusion or poor representation of people who are hard to access, less able to participate or do not live in private households in addition to inaccuracies in respondents recalling and reporting their drinking behaviour and problems with measurement error as people try to convert their consumption into units of alcohol or standard drinks.

People certainly under-report how much they drink. In many cases people may be drinking twice as much as they say they do. This should be of great interest to those who read the epidemiological literature and yet PHE are only interested in it in relation to the health benefits. They do not discuss to obvious implications for health harm, namely that people can drink considerably more alcohol before they assume the risks that are reported in observational studies.

Under-reporting makes the (already dubious) claim about there being 'no safe level' still more questionable. It suggests that the women who have a slightly elevated risk of breast cancer from light drinking are not drinking so lightly after all. And yet this point is never made in the report, nor do we hear about the significant confounding variables that affect studies of breast cancer. All we hear about is the ancient 'sick quitter' hypothesis that has been repeatedly shown to not be the explanation for lower mortality rates among moderate drinkers.

There are also big questions about the biological mechanisms that could explain how drinking causes cancer, especially in parts of the body that do not come in direct contact with alcohol, but the authors do not dwell on these questions except when talking about the benefits. And so we get weasel words like these...

There are no biological processes which have been robustly evidenced to explain the J-shaped curve effect.

That depends on what you call robust evidence. PHE do not elaborate on this point, but it is clear that they demand a far greater burden of proof when the benefits of drinking are involved. In reality, there are plenty of highly plausible biological processes that can explain lower rates of heart disease among drinkers.

When the government commissioned a group of non-partisan experts to review the evidence in 1995, they first of all concluded...

The epidemiological evidence alone linking alcohol and CHD [coronary heart disease] now makes the existence of a protective effect appear very likely. The written and oral evidence received by the Group confirmed that most of the technical epidemiological criteria for a causal association are now fulfilled.

And they had plenty to say about the biological process by which the J-Curve occurred:

5.6 Though a model of causation has not yet been established definitely, most of the evidence reflects increasing acceptance of biological explanations. It is established that the major physical cause of CHD is a life long deposition of fatty tissue in coronary arteries (atheromatous plaque). In addition, local coronary artery spasm and an increased tendency to form blood clots generate the symptom of chest pain (angina). These factors can lead to an acute narrowing or blockage of a major coronary artery and a clinical heart attack, leading to serious illness or death.

5.7 Atheromatous plaques consist largely of cholesterol. Cholesterol is transported in blood combined with specific proteins (lipoproteins) which affect its metabolism. CHD is mainly linked with low density lipoproteins (LDL) which carry 70% of plasma cholesterol. However, some forms of high density lipoproteins (HDO remove cholesterol from the surface of blood vessels. Put most simply it is acknowledged that a high ratio of LDL to HDL is associated with an increased risk of CHD mortality. Physical activity appears to raise HDL cholesterol but does not change LDL cholesterol levels. Alcohol, more than any other dietary factor, raises HDL levels in the blood. In addition, however, alcohol lowers LDL blood levels, and it has been speculated that it is through these lipoprotein cholesterol pathways that alcohol inhibits the formation of coronary artery atheroma.

5.8 Alcohol also directly affects advantageously a number of mechanisms associated with blood clotting and thrombosis. It has been found to reduce platelet stickiness and aggregation, to reduce fibrinogen and to increase fibrinolysis. Again, it has been speculated that through these mechanisms alcohol consumption directly reduces the likelihood of coronary heart disease.

Favouring low quality evidence and opinion over facts

The same exaggerated scepticism can be found when the authors look at other areas where the evidence goes against their tax-and-ban approach to drinking. Instinctively opposed to any solution that involves business, they go out of their way to dismiss the government's work with the drinks industry. They make brief mention of a Department of Health evaluation which found that the Responsibility Deal (which involved drinks companies watering down their products) helped reduce the nation's alcohol consumption by nearly two billion units, which is the twice the target of the 'billion unit pledge'.

In 2013, a Department of Health report evaluated the RD and concluded that between 2011 and 2013 the number of units of alcohol in the market reduced by 1.9 billion (286).

But the authors move quickly to cast doubt on this...

Low quality evidence suggests public-private partnerships involving voluntary pledges to reduce the number of units in the market are ineffective, given that most industry activity to reduce the number of units occurred regardless of the pledge. Furthermore, this activity related to the launch and promotion of new products, potentially increasing the size of the market.

There are two claims in the above paragraph that are little more than bald assertions. The idea that the size of the market is increased by the launch of more products is fanciful and the authors give no serious evidence for it other than the fact that it is 'possible'...

Reducing the number of units in an existing product is different to launching a new low-strength product. The latter potentially increases the total number of alcohol products on the market. It is possible people will then drink more as a result of increased product variety rather than switching from high-strength to low-strength products.

Moreover, the claim that the drinks industry would have launched lower-strength products in the absence of a voluntary agreement to do so is pure speculation. The authors admit that the evidence for this is of 'low quality' but that doesn't stop them treating it as fact when they get to their conclusion:

In conclusion, most alcohol pledge signatories appear to have committed to actions that they would have undertaken anyway, regardless of the RD [responsibility deal].

Oh, OK then. If you say so.

Incidentally, Public Health England is currently 'working with' the food industry to reduce sugar and fat content from food. When this happens, do you suppose PHE will claim that the food industry would have done it anyway?

While the authors treat their own hunches as fact, they cast doubt on the most patently obvious home truths when they do not suit their purposes. They want higher taxes on alcohol, for example, but higher taxes lead to substitution effects and illicit trade. How do they respond?

Cross-border trade, illicit trade and home production are other important phenomena that governments need to take into account when implementing taxation as well as pricing policies. However, there is a lack of data on the changes in alcohol price and tax avoidance and the illicit trade (201).

There will always be a 'lack of data' up to a point, but there is a wealth of evidence that taxation is a key driver of the illicit alcohol market and of the shadow economy in general. To not mention this is to lie by omission.

The report lies by omission a great deal. It cites a study that found a link between advertising expenditure and alcohol consumption but does not mention studies that have failed to replicate this. It mentions France's near-total ban on alcohol advertising but rather than look at whether the ban reduced underage drinking (which was supposedly the intention), the authors switch to various theoretical models which claim such bans work.

A reliance of models, particularly from Sheffield University, is a feature of the whole report. This is because the real world refuses to bend to dogmatic temperance fantasies such as the whole population approach which the authors mention as a theory at the start of the report...

Population consumption theory links population level consumption with alcohol-related harm and contends that overall consumption is directly, and dose-related to the level of alcohol-related harm in a population. As the consumption of a population increases, so does alcohol-related harm and vice versa. Contemporary alcohol policy rests on this fundamental assumption (1,37,38).

At least they admit that it is only a theory. However, by the end of the report it is being treated as gospel...

Given that alcohol-related harm is dose-related, population-level alcohol consumption can be viewed as an approximation of population-level harm (45).

When I wrote a report about the total consumption model with John Duffy in 2014, some alcohol researchers assured me that it was not central to alcohol policy and no one took in 'public health' too the theory literally. Just saying.

Given the lack of evidence for so many of their beliefs, the authors turn to the opinions of their mates. On alcohol advertising and self-regulation, they resort to anti-industry fanatic Anna Gilmore as their source for the following claim...

These tactics are similar to the strategies used by the tobacco industry, particularly the use of obfuscating tactics such as misrepresenting research evidence and using third parties and front groups to lobby government.

This is the kind of inflammatory claim you expect from a pressure group, not a government agency, but Public Health England has blurred the lines between the two. 

When the authors want to donwplay the effectiveness of current alcohol warnings (because they are the product of a voluntary agreement), they say this:

There is a general consensus among experts that alcohol labels have been poorly implemented and this may, in part, explain the finding that labels are ineffective in changing drinking behaviour (304,307).

If the citations are any guide, the 'general consensus' is the result of two anti-alcohol campaigners agreeing with one another. Reference 304 was written by Robin Room. Reference 307 was written by Tim Stockwell. Stockwell's article isn't even from a peer-reviewed journal.

With this 'consensus' in place, the authors turn to expert opinion on what alcohol packaging should look like:

With regards to health warnings, expert opinion informed by the experience of tobacco, suggests alcohol warning labels should be designed and implemented as follows (314):

- develop research to identify ‘direct’ and ‘evidence-based’ health warnings
- increase the visibility of the warnings
- incorporate pictorial health warnings
- consider plain packaging for alcohol products

'Expert opinion' here refers to the views of a single researcher, a doctoral student in occupational psychology called Mohammed Al-hamdani (reference 314).

I hope you spotted the reference to plain packaging, by the way. Quelle surprise.

No comments: