The ‘Fat but fit’ debate
* A study was published on January 22nd, 2021 in the European Journal of Preventive Cardiology.
* It set out to answer the ‘fat but fit’ debate – if someone has higher BMI but is active, does the activity make up for the obesity? Can the obese but active person be as healthy as the normal weight inactive person?
* This study examined over half a million working Spanish people with an average age of 42, one third of whom were women.
* Normal BMI ranges were used (normal weight, overweight and obese) and World Health Organization guidelines were used to define active, insufficiently active, and inactive. Three factors were used for cardiovascular (CVD) risk factors: diabetes, hypercholesterolemia, and hypertension.
* The researchers produced an excellent diagram to show the three way relationship between BMI, activity level and CVD risk factors. They also produced an excellent table to show the risk ratios for these three variables and with male/female results on top.
* Hypercholesterolemia wasn’t so different between the groups and there were some inconsistent findings (maybe it’s a poor risk factor?) But diabetes and hypertension were strikingly associated with obesity.
* At all three weight levels (normal BMI, overweight and obese), it was better to be active than insufficiently active and it was better to be insufficiently active than inactive, but being active did not counter the impact of overweight and particularly obesity. Some of the risk ratios were large – more than seven fold the difference, for example – comparing an obese inactive person with a normal weight active person.
* The study had some limitations, which I’ve discussed, but these are unlikely to diminish, let alone override, these findings. This large study seems to have answered the ‘fat but fit’ debate quite definitively.
There has long been a debate as to whether someone can be ‘fat but fit’ (Ref 1). The phrase is blunt, rather than sensitive, but it’s a well known one and it’s a very interesting one to explore. A study was published on January 22nd, 2021, in the European Journal of Preventive Cardiology. It was called “Joint association of physical activity and body mass index with cardiovascular risk: a nationwide population-based cross-sectional study” (Ref 2). It turned out to be just three pages long – a research letter more than a study – but it packed a lot into the few pages with excellent use of one graphic and one table. An accompanying press release was also informative (Ref 3).
The study was Spanish – led by many researchers in Madrid. It used data from 527,662 working adults insured by a large occupational risk prevention company in Spain. The average age of participants was 42 years and 32% were women.
The study aim was simple – to clarify the existence of the ‘fat but fit’ (i.e. higher body mass index (BMI) but active) paradox, in an observational study. The researchers assessed the association (important word) between different BMI categories and Physical Activity levels. The following BMI categories were used – normal (BMI 20-24.9); overweight (BMI 25-29.9) and obese (BMI 30 or above). The World Health Organization recommendations were used for Physical Activity (PA) categories (Ref 4). Participants were categorized as ‘inactive’ (performing neither moderate nor vigorous PA), ‘insufficiently active’ (not meeting World Health Organization (WHO) minimum PA recommendations for adults, i.e. <150min/week and < 75min/week in moderate and vigorous PA, respectively), or ‘regularly active’ (meeting WHO guidelines of ≥ 150min/week of moderate PA or ≥ 75min/week of vigorous PA, or a combination thereof.) The researchers retrieved information from medical examinations on the prevalence of diabetes (defined as medicated or glycaemia >125mg/dL), hypercholesterolemia (defined as medicated or total blood cholesterol ≥ 240 mg/dL), and hypertension (defined as medicated or systolic/diastolic blood pressure ≥ 140/90mmHg). These three factors were described as cardiovascular risk factors. One of these I would dispute (cholesterol levels), but let’s go with them.
Having gather information on BMI, physical activity level and three medical conditions, the researchers performed a statistical technique (logistic regression) to see if the three factors were associated with each other. Would the cardiovascular risk factors be more prevalent in more obese people and/or in less active people? Would higher activity levels in overweight and obese people mitigate the prevalence of risk factors?
Approximately 42% of participants were normal weight, 41% were overweight, and 18% were obese (rounding errors). The majority were inactive (64%), while 12% were insufficiently active, and 24% were regularly active. The paper reported that 30% of people examined had high cholesterol, 15% had high blood pressure, and 3% had diabetes.
The results were presented in one of the best graphics I have seen in an academic paper. I have replicated it below.
This is a three-by-three diagram. The chart under the green person presents the data for normal weight participants. The chart under the amber person presents the data for overweight participants. The chart under the red person presents the data for obese participants.
Within each of these charts, the left-hand section is for active participants, the middle section for insufficiently active participants and the right-hand section for inactive participants.
The third three-way is that the orange dot represents hypercholesterolemia, the green dot represents diabetes and the purple dot represents hypertension.
The final thing to note is that the active and normal weight person is used as the comparator for all other groups.
If we start by looking at the chart under the green person, we can see the three dots for the normal weight active person are all on the line 1.0. Everything else is going to be compared to this, so that we can establish risk ratios relative to this. Continuing to look at the chart under the green person, we can see that all three risk factors rise slightly in the insufficiently active section and then they rise a bit more in the inactive section. The increased risk ratios – even for the highest dot (hypertension in inactive people) – are small. (I’ll come on to the numbers for risk ratios). The message from the chart under the green person is that normal weight people – even if inactive – don’t have large risk ratios for diabetes, cholesterol, and blood pressure – relative to normal weight/active people.
Moving to the chart under the amber person, you now know how to read the chart, so you can see that cholesterol and diabetes are higher for every level of activity in the overweight person relative to the normal weight person. Only hypertension has a blip in the pattern – active overweight people have slightly lower hypertension (on average) than inactive normal weight people.
Moving to the chart under the red person, everything rises again and by multiples in many cases. Cholesterol is not dramatically different between the groups (because it’s not a risk factor?) but hypertension and then especially diabetes are. Diabetes and hypertension are strikingly associated with obesity.
It is the case that in all three chart areas (green, amber, red), it is better to be active than insufficiently active and it’s better to be insufficiently active than inactive, but being active does not counter the impact of overweight and particularly obesity.
The one table in the paper was also very informative. The data are split into another level – men and women – and this enabled us to see the three-by-three of the diagram (weight, activity level and risk factors) and with the sex data presented separately.
Any result in brackets above that includes 1.0 – the line of no effect – is not (statistically) significant and could have happened by chance. For example, hypercholesterolemia in men – comparing the insufficiently active with the active gives a result of 0.97 (0.93-1.02). The brackets include 1.0 – the line of no effect – and thus this result is not significant. A number of the brackets above start with the number 1.0. These are also not significant.
Points to note from the above table:
– Hypercholesterolemia and diabetes were more common in inactive normal weight men than inactive normal weight women.
– The cholesterol marker generally gave some spurious results – e.g. lower risk ratios at higher levels of inactivity – the opposite of a risk factor. For this reason, I ignored the hypercholesterolemia middle part of the above table beyond the first finding in the point above.
– Hypertension risk ratios were higher in women than men across the board. Whether overweight or obese, women had higher risk ratios for hypertension than men at every activity level.
– The same applied with diabetes with just one negligible exception. The differences were large too. Inactive obese women had a 6.26 risk ratio relative to normal weight active women. Inactive obese men had a 5.42 risk ratio relative to normal weight active men.
– The risk ratios were large generally. We talk about risk ratios of double being needed for the Bradford Hill criteria to be worthwhile exploring for causation and not just association. Taking normal weight and active as 1.0, a risk factor of double this would be 2.0. The risk factors in the obesity section of the above table are almost all way bigger than 2.0. The largest is 7.28 – (obese inactive women for hypertension) – that’s a seven-fold greater risk. That is highly likely to be causal.
– Apart from the hypercholesterolemia anomalies, risk ratios increase with each level of lower activity within and across overweight and obese categories. It is almost universally the case that being obese and inactive is worse and then obese and insufficiently active and then obese and active and then overweight and inactive and then overweight and insufficiently active and then overweight and active.
The short article noted two possible limitations of the paper. First analysis was not controlled for diet. The researchers examined BMI and while this is likely to be associated with diet, the study cannot rule out that a dietary factor impacted diabetes, cholesterol, and/or hypertension independent of weight. For example, intake of plant sterols will likely impact cholesterol levels independent of weight. Intake of alcohol might impact cardiovascular risk factors independent of weight.
Second, leisure time physical activity levels were self-reported and people may report things differently and there may be differences in activity that don’t get reported. The adults were all working adults insured by a large occupational risk prevention company in Spain. However, the type of work they did could have been very different. One might have been a desk worker and the other might have been building site surveyor. The occupational activity levels might have been very different, but the two people might only have reported the half hour walk they both did daily.
There are other possible limitations of the study:
– BMI is not the best marker – especially in men. BMI works less well in taller people and not well at all in athletes. We must assume that professional athletes were not included in this survey, as professional rugby players, for example, would likely have BMIs indicating obesity and very low incidence of diabetes or hypertension.
– The World Health Organization physical activity guidelines are not necessarily the best markers. The use of discrete categories ignores what could be vast differences in activity. For example, I go for at least a 30-minute walk daily. I thus meet the guidelines for “regularly active” according to the WHO. However, I am in the same category as someone training for ultra-marathon competitions. Are we comparable?
Cut offs are also blunt instruments in population studies. The person doing 70 minutes vigorous activity a week is “insufficiently active”; the person doing 75 minutes vigorous activity a week is “regularly active.” Is that fair?
– Few people had diabetes (3%) and this could therefore drive the small comparator group issue that we have talked about previously. When one group is very small, small changes in numbers in that group can make a big difference.
– There could be a reverse causation issue. Someone may have diabetes and find it difficult to control their blood glucose levels while exercising. They may choose to sacrifice exercise (especially vigorous) to better manage their diabetes. Someone may have hypertension and worry that their blood pressure goes up with vigorous exercise and avoid this as a result.
However, I think that the limitations are unlikely to diminish, let alone override, these findings. This is an important and valuable, succinct study. It has shown that it is better to be active than insufficiently active and it’s better to be insufficiently active than inactive, but being active does not counter the impact of overweight and particularly obesity. It looks like this has answered the ‘fat but fit’ debate quite definitively.
Until the next time
All the best – Zoë
Ref 1: Ortega FB, Ruiz JR, Labayen I, Lavie CJ, Blair SN. The fat but fit paradox: what we know and don’t know about it. Br J Sports Med 2018.
Ref 2: Valenzuela et al. Joint association of physical activity and body mass index with cardiovascular risk: a nationwide population-based cross-sectional study. European Journal of Preventive Cardiology. 22 January 2021. https://academic.oup.com/eurjpc/advance-article/doi/10.1093/eurjpc/zwaa151/6105192
Ref 3: https://www.escardio.org/The-ESC/Press-Office/Press-releases/Being-fat-linked-with-worse-heart-health-even-in-people-who-exercise#
Ref 4: The WHO recommends that all adults do at least 150 minutes per week (and up to 300 minutes per week) of moderate to vigorous aerobic physical activity such as walking or brisk walking, or at least 75 minutes per week (and up to 150 minutes per week) of more vigorous activities such as jogging, or a combination thereof.