What do we know about physical activity? Knowledge from one week seminar

11 minute read


What do we know actually about physical activity? The past week, I attended a seminar in Kirstiansand that addressed exactly this question: What do we evidently know about physical activity? I guess everyone has a concept of it, but does this actually align with what the experts in the field are saying?

I had now for close to a year the pleasure to work on the intersection of physical activity research and Computer Science (Yes, this is actually a thing!), but with my background in Cognitive Science my knowledge about physical activity and its effects on health and the body itself were rather limited. I had some basic ideas or probably rather concepts in mind that are likely familiar to many people: Sport is good. Being active is good. But else? Due to my position as a PhD candidate at the intersection of these fields, I already knew that there are several guidelines for physical activity from different health authorities (among others including the WHO, the CDC and many others). I had further an idea of what they are: About 150min per week spent in moderate-to-vigorous activity (often simply called MVPA which is significantly more friendly for your tongue) in bouts of 10min. I even have one of these modern smart watches that give you estimates for this, but still I had no idea what I actually have to imagine by this. So what is this, what do we know and what does it mean? All these and many more questions got answered this week by top experts in the field.

Why are we actually interested in physical activity?

I think the first question that many people might ask is: Why actually care? Why does the tax payer actually fund people to work on this. Historically speaking, and this was mentioned throughout the course probably as often as psychologists show a picture of Wilhelm Wundt, the inventor of evidence-based or experimental psychology, in a bus in London. Morris et al. (1953) observed in London buses that the drivers had a significantly decreased life expectancy compared with conductors working in the same environment. As a fundamental difference between the two occupations was the amount of activity, an association between physical activity and mortality seems to exist.

Easy isn’t it? Sadly it is not also because if it was, the research of physical activity would probably have been settled already some decades ago. A huge problem, and this is seems to be a reoccurring theme in the field, is that a wide range of factors are all associated with each other. Later analysis of the Morris data, for instance, showed another equally important associations: Not only a difference in physical activity existed, also a difference in uniform size (as a proxy of weight and obesity) was found. In fact, many different closely related variables have been found which all to a certain extend seem to be associated with each other.

What causes what? The problem of causality

We humans like the principle of causality. Something alway has to be the reason for something else and if we cannot explain it is god, magic or the flying spaghetti monster that is causing it. Quite often, however, we humans mix associations or correlations with causation. In fact we cannot determine whether the observation Y is caused by the co-observation X. I can, just from observation, always be also the other way around (Y causing X) or funnily enough none of these (for some ridiculous not related correlations see here).

The British epidemiologist Sir Bradford Hill developed for this reason nine criteria that are required for causality. As especially, the criteria of temporarily is not given in cross-sectional observations (snapshots at one time step looking on what occurs together with what), Randomized controll trials are used to determine causal relationships. But before looking at what we know about these relationships, it is necessary to define what physical activity actually is.

Activity = Activity?

What is actually activity? The naive definition is that activity is everything which is not inactive (in PA research language called sedentary time; from latin sedere “to sit”). But is it so easy? Are a walk, a run and a sprint all have the same influence? Does it matter whether you spent hours on your bike or in the gym? What we know is that some intensity classes seem to exist. So activities can be grouped by their intensity into groups that share roughly the same amount of metabolic equivalents (METs). METs are a measure of the rate a person expends energy relative to the mass of the body and is used to objectively access the intensity of an activity.

Experiments suggested that three different classes of activity exist: light, moderate and vigorous activity. For each class thresholds exists for the different assessment methods of activity. The classes itself are backed by the physiological mechanisms underlying physical activity.

How are the different activity classes meassured?

So how do researchers access physical activity? Traditionally, physical activity is often accessed using questionnaires like the International Physical Activity Questionnaire (IPAQ). These method is nowadays considered as not really accurate which is easily relate-able when you think of how accurate you think your answer would be if someone asked how “How many minutes a day are you doing moderate to vigorous activities?”. This example might be a bit exaggerating, but many studies showed that this kind of assessment is evidently biased by giving answeres that tend to be what is socially expected.

The field is therefore moving slowly towards small devices called accelerometers which measure activity in terms of acceleration (no acceleration = no activity). However, these devices also come with some disadvantages as, for instance, a higher burden for the participants and a systematic underestimation of activity as participants likely sometimes do not wear them (and should not, for instance, if the sensor is not waterproofed, etc) under some activities. Despite these limitations, the use of accelerometers is usually concluded to be significantly more accurate than questionnaires.

I think it is important to highlight that biases are a problem all over the place in Science and knowing that these biases exist and aiming to figure out what these biases are is the best you can do as a scientist.

A small self-experiment

Interesting for me was also to experience the whole procedure of professional physical activity assessment myself. In the preface of the course we all got handed out accelerometers to measure our own activity in the days before and during the experiment. This was of particular interest, as I figured out that I had a significant misconception of what the respective activity intensities actually are. I always thought that MVPA is a workout style activity including sweat, resistance and endurance, but actually it isn’t.

Even though, I was also on two shorter runs in that week, the main contribution to my measured physical activity was given by a longer really relaxed hike on the weekend. Nothing exhaustive, just good old walking.

Guidelines, Interventions and eternal life

Getting accurate measures, understanding the physiological mechanisms, even screening and monitoring whole populations is one side, but at the end an important question lies also in the application of all of this research. As mentioned already in the beginning, a substantial amount of experts around the world are working in developing new or updating existing guidelines to increase global health and well-being. My impression was that the field here is currently in a paradigm shift caused by the use of accelerometers. For over a decade now the official guidelines of the world health organization recommend 150 minutes of 10 minute bouts of MVPA, recently updated to 150-300 minutes of MVPA. From what I could infer, it seems to exists some discrepancies as these values were derived based on questionnaires which as discussed earlier are subject to several biases overestimating the amount of physical activity that is needed for a healthy and long life.

A huge surprise to me, however, was the distribution of physical activity in various populations indicating a severe trend of a highly skewed distribution towards the physical inactivity side. In other words: We see a huge amount of people being really inactive and only a few people being really active. Given the bounded nature of the physical activity scale a skew is not unsurprising. We can observe this in Science in many occasions as for instance in reaction time experiments in humans and animals, but the amount of skewness still surprised me and my impression was that these distributional characteristics are understudied. Especially, as they appear to have a huge impact on how interventions and policies that should be implemented: Should interventions focus on increasing physical activity among all people or should we aim “unskewing” this distribution by intervening on those who are most inactive.

Ironically, these group of rather inactive people are on the same hand the least studied group because besides all the different variables that are associated with each other (physical inactivity, poor health, low socioeconomic status and many more), these are also associated with lower engagement and motivation to participate in this kind of studies, resulting in Science done by scientists researching on scientist producing results that give us insights about scientists as so often men did research on men reporting results about men. This is provocatively written and obviously exaggerating a lot, but a huge problem nevertheless. But and this I think is very important, I was highly impressed by how clear and open all these experts talked about this problem. How clearly they stated the problems and are aiming to overcome these and I think this is the best and probably only way to address problems like this as scientists.


So what did I infect learn? Definitively more than I can describe in this summary. The field of physical activity research is not easy. The scientists in this field have to deal with a whole bunch of highly interconnected and -correlated factors and have simultaneously to deal with many of the sampling and bias problems you have when you do research on and with humans. And still, they have come a long way. Even though, the magnitude of effect that physical activity has on different-cause mortality, quality of life or other co-factors is not exactly known, I feel it is save to say that being and staying physical active is really important and simultaneously the guidelines that are out there are not so difficult to fulfill. On the same side, I was really surprised about the magnitude of inactivity that seems to be apparent. I understand now why people sometimes speak about the physical inactivity pandemic and think more effort should be done addressing this problem.

So instead reading and procrastinating you should have spend this time, for instance, going for walk, do a small hike, or just for a quick bike trip to a cafe in the next city.