Beyond randomised controlled trials (RCTs): understanding different types of evidence

When it comes to discussing health and lifestyle research, you’ll sometimes hear observational studies dismissed with a simple response: “But it wasn’t an RCT.” 

Randomised controlled trials (RCTs) are a useful way to answer certain research questions, particularly about drug treatments. By randomly assigning people to different groups, they help reduce other factors that could influence the results. When a trial is well designed, this can give us strong evidence about whether a particular intervention actually causes a change in health. 

But that doesn’t mean RCTs are the best tool for every question. 

There is a very important difference between establishing what occurs under controlled conditions and understanding what happens in the real world. 

RCTs aren’t real life 

Imagine you want to know whether changing someone’s diet improves their MS symptoms. 

In an RCT, researchers might carefully define the intervention, recruit participants who meet specific criteria, provide dietary advice, monitor adherence and follow participants for a set period. 

That’s extremely useful. But it also creates an environment that can be very different from everyday life. 

Outside a clinical trial, people don’t follow instructions perfectly. They eat differently at weekends, go on holiday, have stressful days, have meals with friends and change their behaviour over time. 

The same applies to exercise. It is relatively straightforward to randomise people to an exercise programme for a defined period. But that’s very different from observing what happens when thousands of people live their lives with different levels of physical activity over many years. 

This is where observational research can provide something an RCT often cannot: a picture of what actually happens among people living their ordinary lives. 

Observational doesn’t mean meaningless 

Observational studies have an obvious limitation – people aren’t randomly assigned to a particular behaviour or treatment. 

Someone who exercises regularly, for example, may also have a different diet, income, education, smoking history or access to healthcare. These factors can influence health outcomes and make it difficult to determine exactly what caused the difference. 

But the existence of these factors doesn’t mean observational research is worthless. Researchers can measure and adjust for many potential factors, use different study designs and analytical methods, and look for consistency across populations and studies. 

It is also important to recognise that observational research isn’t necessarily trying to answer exactly the same question as an RCT. Sometimes the question is not simply “Does this intervention cause this outcome under controlled conditions?” but “What happens to people who behave this way over time?” 

Those are different questions, and both are valid and useful. 

It is also important to distinguish association from causation, and prospective from retrospective research. In simple terms, prospective studies follow people forward over time, while retrospective studies look back at what has already happened. For example, a 7.5-year prospective study1 found that higher diet quality was associated with around a 50% lower risk of disability progression in people with MS. This does not prove that diet caused the difference, but it provides stronger evidence that diet quality came before the outcome. 

The “best available evidence” 

This is particularly important in lifestyle research. 

In areas such as diet, physical activity, sleep and other health behaviours, it is often difficult, or sometimes impossible, to conduct large, long-term RCTs that perfectly reproduce real life. 

In these cases, we need to look at the full range of available evidence. RCTs can tell us about the effects of specific interventions under controlled conditions. Cohort and other observational studies can help us understand what happens over longer periods in everyday life. Cross-sectional studies, which look at people at a particular point in time, can help us understand patterns and associations within a population. Systematic reviews can bring together findings from multiple studies to give us a broader picture. 

Each approach has strengths and limitations. The key is to understand what each type of study can, and can’t, tell us, and to consider the evidence as a whole. 

The question shouldn’t simply be, “Was it an RCT?” 

It should be: What question was the study designed to answer, how well was it done, and what does it add to the wider body of evidence? 

RCTs are an important part of evidence-based research but they are not the whole of it. 

You can read more about our approach to research and evidence here.