How much is safe to drink? Dame Sally boobs on booze

Dame Sally Davies tells us that ‘no level of alcohol consumption is safe’. Is she right? And why has the advice changed, and why is it different in different countries? Getting to the truth is tricky, and here is why. 

  

When she started the job as Chief Medical Officer Professor Dame Sally Davies – who was formerly one of the most specialised of super specialists – dealing with sickle cell disease- got on the wick of many doctors with her instructions. She said that we should lead by example in terms of lifestyle, and make every consultation count in terms of imposing an agenda of well meaning health advice on obesity etc. even when we are dealing with patients struggling with major medical problems.

Now she has really poured salt on the wound with her advice reducing the maximum safe limit for alcohol intake – especially with reference to cancer. For a while  GP’s will be getting an increase in patients presenting with minor symptoms who are terrified that they are caused by cancer which is related to their modest consumption of alcohol.

Meanwhile lots of health experts continue to insist that moderate drinking has overall health benefits, and guidance from around the world on safe levels of consumption varies widely even though the experts are looking at the same evidence. Surely, an issue which has been debated and studied by huge numbers of people for decades should have clearer answers.

Well, no actually. It is extremely difficult to be sure about issues such as this. The reason is that direct experimental evidence is very difficult to obtain – you could conceivably split imprisoned convicts into groups and administer alcohol in varying quantities – but it would not be allowed. Studying what happens to people in the population has a lot of potential errors and possible biases, and this is the reason why there is so much disagreement in interpreting the research that has been done. Similar issues apply to lots of other questions about the health effects of meat in the diet, sunlight, vitamin D, Hormone Replacement etc.

There are 4 sorts of evidence that are relevant to these sorts of issues and it is worth understanding each one as you usually need more than one type of evidence to really nail something as a probable causative factor in a certain disease. 

The easiest sort of evidence comes from large surveys of people with a particular disease asking them about their habits, occupations, drug use etc. and looking for statistical associations. This method is called a Retrospective Case Control study. It is quick and pretty good at excluding possible causes, but suffers from several possible sources of bias, which makes its positive conclusions about causation relatively unreliable. It is difficult to recall past habits correctly leading to ‘recall’ bias. (Although some studies get around this by searching health records for example for a history of medication use).  The control group who have responded to the request to take part in surveys may be unrepresentative of the general population in many ways. This is called ‘selection bias’. Lastly this sort of study can suffer severely from so called ‘confounding’ where some other causative factor that may be unknown and that is not being measured differs between the groups being compared. For example, lots of effort has been spent on comparing different sorts of occupations with the aim of working out to what extent ‘stress’ and poor job satisfaction causes disease. However it has been very difficult to rule out other factors such as smoking, poor diet, and lack of exercise as they tend to more common among lower status workers.

A more laborious, but more reliable method is to identify the suspected culprit at the start of the study and arrange to follow a cohort of subjects over a period of time and count the number of people falling ill in the separate groups. This is called a Prospective Cohort study and is much more reliable. However, it is still difficult to choose a control group that matches the study group who are being exposed to the risk or treatment, so that confounding and selection bias are still possible. ‘Recall’ bias is still possible if people lie about their habits – which is possible with questions about alcohol.  The other big problem is that it takes several years from when the study is started to the day the data is analysed.

One feature of both prospective and retrospective studies that strengthens their conclusions a lot is a so called ‘dose response’ effect. When the smoking habits of British doctors were studied back in the 1950s it was the fact that the risk of lung cancer strongly and progressively increased with increasing tobacco consumption that clinched the argument.

The most reliable evidence comes when the researcher is able to randomly select a group of subjects, and expose them to the relevant factor, while keeping the ‘control’ group unexposed. This is called a RCT or randomised controlled trial, and is accepted as the most reliable evidence. However, it is expensive, and usually has far fewer participants, and it can often be criticized as not being relevant to the ‘real world’.

The 4^th very important line of evidence is called ‘biological plausibility’. This is when there is evidence that the factor under suspicion has some known effect on the organ concerned. For example it is experimentally provable that a small concentration of CO2 in air will reduce the transmission of infra red radiation – so that the mechanism of man made climate change is clear.

If we have evidence of all 4 types pointing in the same direction then the case is very strong. Unfortunately that is not common and there are many subjects where we are not even close. We know that we have to be distrustful of conclusions based on one type of study alone as this has led to some famous errors in the past. The most famous example is that of Vitamin C. Retrospective studies showed that consuming fruit and veg was associated with much better health outcomes, and everyone thought that it must the anti-oxidant Vitamin C that was the important ingredient.  However, when Randomised Controlled studies were done with half the subjects taking a supplement, they showed no benefit whatever. We now think that people who eat fruit and veg are generally more careful with their health, and perhaps there are other chemicals rather that make the difference.

And so to Alcohol. We must of course first make clear that we are not talking about 5+ drinks a day or 35+units per week. It is abundantly clear that this does a lot of harm, raises blood pressure, leads to violence, social disorder, harms mental health etc etc. Everyone thinks of liver cirrhosis but this is actually fairly rare from alcohol and requires a much higher intake.

But is low to moderate consumption harmful - or possibly beneficial?

Retrospective studies have generally failed to show that mortality is increased, which is against a major harmful effect, but have suggested an increased cancer risk, but a reduction in heart disease.  One of the big problems with such studies is that smokers tend to drink more alcohol, and it is difficult to disentangle the effect of each. But in the last 10 years or so there have been some prospective studies which have probably produced more reliable evidence. There have only been a few of these and the results have varied which has led to fierce argument over their interpretation. The issues are whether Cancer risk is increased, and if so how much, and whether Heart Disease is prevented by low or moderate intake.

I myself am influenced by the 4^th sort of evidence, which is whether there is a likely mechanism that fits the facts observed. It does look as the effect of alcohol on cholesterol and also on blood clotting might well explain a protective effect, and it fits in well with the results of the trials. But the Sally Davies faction say that the observed beneficial effect is an artefact due to the tendency of some teetotallers to be reformed alcoholics so that their bad outcomes worsen the picture for the whole group of non drinkers. Perhaps I am biased but I do feel that they would not make similar arguments if the results came out in their favour, and the protective effect has been observed in countless studies. The size of the protection is not trivial – about a 10-20% reduction in heart attacks, which is as more than low dose aspirin. For comparison consistently taking a statin drug will reduce risk by around 30%.

As far as Cancer is concerned, the strongest effect is that on cancers of the mouth and throat and oesophagus. Here the smoking confounding is the big problem. A large prospective US study concluded that the effect of alcohol was confined to smokers.

Unfortunately, there is another result which is far less palatable (joke). That is that there is a pretty definite effect on Breast Cancer risk. Since this is a common and serious disease this has to be taken seriously. I had thought that alcohol intake might lead to obesity and that might explain the statistical correlation but they have looked at that and decided that drinking and obesity don’t actually go together that much in women. Overall, it is estimated that about 10% of all breast cancers are caused by alcohol.  So If all women stopped drinking altogether we would have 10% fewer cases. The mechanism is thought to be that alcohol affects the metabolism of female sex hormones in the liver.

Whether other solid cancers such as bowel and prostate cancer are affected by low to moderate alcohol intake is disputed. Some large studies have failed to show an effect. In a large recent prospective US study there was no statistically significant definite evidence of an increase in cancer in non smoking men.  

The new guidelines proclaimed by Sally Davies suggest the same limits for men and women. This disagrees with almost all other national guidelines on the subject and all the basic science about how men and women handle alcohol.

If there is a risk for women of breast cancer which seems likely, what can be done to reduce it?

 First of all be lucky.  Failing that you can stay slim, take exercise daily, and have babies and breastfeed them for a long time. In terms of preventative drugs there is a lot of interest in good old aspirin.

And if you are a man? -------------------- See you down the pub!