Q: So how do we know that vaccines are safe?

A: Because they are tested!

What happens first is that relatively small trials are conducted. The trials will not usually have more than a few hundred “guinea-pig” subjects, if that. Such trials can to some extent show us how effective the vaccine is, and whether there are common ill effects, but in a small trial, you cannot expect to show up an ill effect that is as rare as, say, one in a thousand. To spot the less common side effects, you need a much larger trial.

Unfortunately, it just isn’t considered practicable to do large trials. For one thing, you are injecting large numbers of people with “untested” substances. To do this ethically you would need to explain to the subjects the risks that they were taking, and if you did that then you might struggle to find the large numbers of volunteers that you would need…

So large scale vaccination safety trials are just not done. Vaccine products are normally licenced for widespread – and in some countries mandatory – use on the basis of small trials only.

But we would know about it if a vaccine was licensed but turned out to have dangerous side effects. Wouldn’t we?

Once a vaccine is put into general use, we are then reliant on doctors reporting safety data from the vaccines in actual use. But those doctors are not instructed to look out for side effects – so, for all we know, many may go undetected. Indeed, those doctors may be told “in clinical trials, no side effects were detected” (or the like) and thereby actively deflected away from seeing side effects even when they happen. Moreover, vaccines are not generally administered and vaccinated children are not generally monitored in any controlled scientific way. Nor do suspected reactions appear to be investigated with any particular or routine thoroughness. As such, it is rarely going to be possible to analyse suspected reactions adequately so as to identify whether the vaccine really was the cause. This is not a scientific way to conduct safety trials.

A third methodology used to try and prove safety is by means of population studies. Take two large populations – one vaccinated and one not. Compare rates of certain conditions or diseases in the two populations. If, say, autism is higher in the vaccinated population then vaccination may cause autism. If not, then vaccination cannot be a cause of autism.

That all sounds plausible, but it is a massive over-simplification.

In a scientific trial you would normally expect to see a “test” group (receiving vaccinations) and a “control” group (not receiving vaccinations). The groups would be identical in all other material respects so that a comparison between them could genuinely be considered true and fair. Each group would have to be large enough that any trends identified would be statistically significant.

However, these factors are not normally present in population studies. Many, many factors contribute to trends in the health of a large population, which can vary from country to country and even within a country. Diet and exercise habits, pollution levels and ethnic differences are just a few factors that spring immediately to mind as potentially relevant. For these reasons, it must be near impossible to find two large groups of people, one group vaccinated and one not, with no other significant differences between them. The real world is complex with infinite variations, but scientific trials need simplicity with only one variation (i.e. the one being tested).

(Here’s a rather absurd illustration: the number of obese people in 1940s Britain was very low. Since then, vaccinations have been on the increase and so has obesity. Could it then be argued that obesity is caused by vaccinations? No. There are so many factors at work, that it would be impossible to draw any conclusions about whether obesity and vaccinations are linked in any way.)

Even if population studies were to some extent workable, they would still have serious data quality problems. There is a difficulty even with a straightforward questions, such as “How many children who have been vaccinated with MMR subsequently contract rubella, compared with those who have not been vaccinated?” What if the child has had mild rubella without being taken to the doctor and so without being included in the data? Even if the child was taken to the doctor, what if the doctor did not rcognise rubella, or overlooked the possibility precisely because he knew that the child had been vaccinated? How can the study control for these sorts of difficulty?

Where does that leave us?

Useful clinical safety trials are rarely if ever conducted.

The alternatives are reaction reporting and population studies. Reaction reporting by doctors is likely to be patchy and will almost certainly be scientifically inadequate. Population studies are very difficult to do in that appropriate “control” groups are very hard to come by. These options are also retrospective – they can only test the safety of a vaccine after it has been administered to thousands or even millions of children. This is hardly ideal.

So how do we know that vaccines are safe? Because they are tested?