Nutritional science is difficult and demanding. Difficult because of the complexity of human biology and behaviour, and demanding because people want answers right now.
The thing is that the real answers take time and money.
Generally, nutritional research progresses through three stages that I will refer to as S1-3. The research starts out in the laboratory (S1) and progresses to small-scale human studies (S2). These stages are to see if there is sufficient evidence to warrant a large-scale clinical trial (S3). The answer to the health question doesn’t usually come until S3 is complete. In the following, I will use salt (from my previous post) as an example, so the question might be: Does eating salt increase the risk of heart disease?
The results from S1 and S2 studies investigating aspects of this question get published along the way. That is the scientific system. These results are important but do not answer the question. Still, they are the first publications, they are novel and therefore of interest. They usually get the attention of the media, lay nutritionists and health advisory bodies that then mount a campaign (e.g. ‘eat less salt’). But this is premature – the science hasn’t had the chance to answer the question yet.
S1: In the laboratory
There are two types of laboratory studies. First, in vitro studies can be thought of as those taking place in a petri dish or a test tube (in vitro is Latin for ‘in glass’). They mostly use cultured cells or experimental preparations. Their strength is that these systems can be studied in great detail, they tell us about mechanisms. Their weakness is that they are not performed in living things.
The second type of laboratory study is in vivo (‘in life’) with animals such as rats or mice. Their strength is that there is still control over parameters (e.g. rats can be fed salt and compared to another group that is not, keeping everything else equal). Their weakness is that rats aren’t humans (although some might argue that the converse occasionally applies).
The purpose and importance of the S1 studies is to accumulate enough evidence to indicate that human studies are warranted.
S2: The first human studies
These start out small and manageable. They often use a biomarker of disease to see if a nutrient affects the biomarker. So, for salt, the biomarker is blood pressure, while for fat it might be cholesterol (if heart disease is of interest). An alternative approach is to use cross-sectional studies (sometimes called retrospective studies) which take people who have heart disease and determine if they ate a lot of salt.
The biomarker studies are indirect – they are not looking at actual health outcomes (heart disease) but rather something normally associated with that outcome (high blood pressure). The cross sectional studies have a range of limitations and biases (relying on self-reporting salt intake for example).
Still, these S2 studies are important because, despite their limitations, they are part of the process of accumulating enough evidence to embark on the costly and time-consuming S3 studies.
S3: Answering the question
To answer the salt question, salt intake and the incidence of heart disease need to be measured over time. The gold standard method is a clinical trial. This requires recruiting large groups of people with different levels of salt intake and comparing the incidence of heart disease as time progresses. It takes many years (decades even) until sufficient health outcome (heart disease) data is accumulated. Furthermore, because there are so many variables to account for (lifestyle, genes, other dietary factors, exercise, compliance etc), large numbers of participants are required, usually around 100,000 or more. Typically, these are multi-centre and even multi-country studies.
Which explains why the science does not start with an S3 experiment. There has to be seriously good data to support a hypothesis before the S3 studies are undertaken. Research funds are limited, and funding bodies will not direct their precious resources to an S3 study without compelling evidence from S1/S2 results. As well, scientists are not going to dedicate themselves to a tedious decade-long study without a good reason.
The next time a headline makes a firm or startling claim, check whether the article is referring to a laboratory study (S1), an indirect biomarker or cross sectional study in a small group of people (S2), or a definitive health-outcome study in a few hundred thousand individuals (S3). They are not equal, but they are a part of the systematic and pragmatic scientific process.
The problem is that nutritional advice gets formed around S1/S2 data, before the definitive S3 data is available. For example, S3 studies of salt only began appearing ~2004, even though the ‘eat less salt’ campaign has been going in earnest for over 40 years (without adequate scientific evidence). I am not aware of any nutritional campaign that started out with adequate scientific evidence.
Because they appear late in a nutritional campaign, the significance of S3 studies can be easily overlooked. But their results are the most important and can be dramatically different to the established nutritional wisdom – low salt consumption might increase the risk of heart disease, for example.
Still, scientists are not happy with just one S3 study, they know that even the most well-designed study will have its limitations. They continue with S3 studies until there is sufficient replication that a consensus can form and further S3 studies are no longer justified.
Only then do we have an answer to the question, or, as with salt, a new question to answer.
Print This Post