Knee arthritis is known to be aggravated by excess body weight. Many studies have shown a connection between obesity and osteoarthritis.
But is the reverse also true? That is, if you reduce body weight, will the knee osteoarthritis problem reduce? And if it is does, is there a percentage or amount of weight loss that helps improve knee arthritis?
Results of the earlier trials
A paper published in 2000 tested older, obese, osteoarthritis patients who lost weight over 6 months due to diet and/or exercise. In these patients, there was an improvement in pain, disability, and performance.
A paper published in 2005 tested older, obese, osteoarthritis patients over 18 months. Some patients were told to go for weight loss with diet and exercise; some others were given only a healthy lifestyle treatment (standard treatment, that does not ask for weight loss). The patients who underwent weight loss showed better improvement in parameters such as stair-climb, 6-minute walk, and pain.
A paper published in 2013 tested older, obese, osteoarthritis patients over 18 months, with diet, exercise, or both. They found reduction in Interleukin-6 as well as improvement in health-related quality of life. Interleukin-6 shows the level of inflammation, or swelling, in the joint — a kind of marker for the extent of osteoarthritis problem.
Thus, the earlier trials have shown that weight loss helps improve the osteoarthritis condition. So weight loss is associated with osteoarthritis condition.
However, it was not known if more is the weight loss, more will be the improvement. In other words, does osteoarthritis improvement correlate with weight loss?
Technical, but not Medical, Information
The words such as association, correlation, and causation cause a lot of confusion. In normal practice, they are used loosely and interchangeably. However, all three mean different things.
Let us take an example of two parameters, obesity and fitness. Now, (we think) we know that obesity and (lack of) fitness are associated as well as correlated. Perhaps, (lack of) fitness even causes obesity, or obesity leads to lack of fitness. While we think we know all this, do we really know, for sure?
Association
Association between two variables, A and B, means one variable (A or B) provides information about the other (B or A).
The association is, obviously, both ways —if A is associated with B, B is automatically associated with A.
Correlation
Correlation is more specific than association. It means the two variables follow either an increasing or decreasing trend. If A increases, B will increase or decrease (but never both, for the correlation to hold).
Correlation also works both ways —if A is correlated to B, B is automatically correlated to A.
Causation
Causation between two parameters means if A is changed, B changes. This is different from saying ‘if A changes, B changes’. Note the exact words: ‘if A changes’ versus ‘if A is changed’. The first phrase is observational, the latter is action-oriented.
Causation can not be established by just observing two variables. By simple observation, you can have only a hypothesis, but not a proof.
Now, you might say that you may observe situations such as someone jumping off a rooftop and committing suicide (please don’t silently observe; intervene, in such cases). So you can claim to have proof that action (jumping off) causes result (death).
However, from a purely statistical angle, it is not a proof of causation; it is a hypothesis. Unless you can make the person jump off the rooftop (please don’t), you can not verify any causation between the two. Of course, with millions of such observations over millennia, the mankind has developed ‘proof’ of this causation.
In some situations such as cosmic events, you cannot control any of the parameters. There, you cannot have causation at all; you simply develop more and more confidence in your hypothesis about causation, as you observe more events.
Unlike association and correlation, causation is not both ways: if A causes B (suicide?), it does not mean B causes A nor does it mean that B is caused only by A. B could also be caused by multiple other things.
Coming back to our parameters, fitness and obesity, we might notice that increased obesity reduces fitness. However, there may not be a correlation. For example, a thinner person may be more fit. But, an even thinner person may be underweight. It is possible that an underweight person may be so because of some underlying medical problem or smoking. In either of these cases, his fitness will be lesser than a similar person who is not as thin.
So we may notice that obesity and fitness have an association —a reverse U-shaped association.
Gaunt person (negative values of obesity) — low fitness;
Normal weight (zero obesity) — high fitness;
Overweight (positive values of obesity) — once again, low fitness.
On the other hand, obesity has no correlation with fitness (actually, in the strictest statistical sense, you should say that they have low correlation, instead of saying no correlation), because there is no perfectly increasing or decreasing trend between the two. This is, of course, if you define obesity in the widest terms, as above.
If you define obesity as being zero, at or below (gaunt person) the normal weight, then obesity will indeed be correlated to fitness since the higher the obesity, the lower the fitness.
Causation between obesity and fitness can only be established by actually changing one and seeing if the other one changes.
And then it gets trickier. For example, you may make an obese person lose weight by making him smoke (as some fashion models do to lose weight). In that case, you may not get improvement in fitness levels. This is called Confounding; but let us leave it for another time.
In summary
1. Association does not imply correlation nor causation.
2. Correlation does not imply association nor causation.
3. Causation does not imply correlation but it implies association.
All these can be grasped better, especially the summary point 2 above, if one can see graphical representations of such relations. A superb article on this in published in the journal Nature: Association, correlation and causation.
And if you want to fry your brains with overload of knowledge on this subject, try this compilation in the same journal: Statistics for Biologists.
So is an improvement in osteoarthritis correlated with weight loss?
Results of the new trial
A paper published in Nov 2018 in the journal Arthritis Care and Research studied obese patients with osteoarthritis for 18 months. They were evaluated based on pain, functioning, 6–min walking distance, physical and mental health-related quality of life (HRQoL), knee joint compressive forces, and interleukin–6. Based on these parameters, the trial tried to find out if the extent of weight loss correlated with the extent of improvement in osteoarthritis.
The trial grouped patients into 4 groups:
Group 1: those who lost less than 5% weight
Group 2: those who lost between 5% to 10% weight
Group 3: those who lost between 10% to 20% weight
Group 4: those who lost more than 20% weight
It found out:
The greater the weight loss, the higher the clinical and mechanical benefits.
Group 4 was better off on all parameters compared to groups 1 and 2.
Group 4 had 25% less pain and better functioning, as well as better physical health-related quality of life.
Comments on the results
The earlier test results had shown that:
Osteoarthritis improvement is associated with weight loss (association)
Osteoarthritis improvement is caused by weight loss (causation)
This test’s results show that:
Osteoarthritis improvement is correlated with weight loss (correlation)
Of course, we do not know if the trend in osteoarthritis improvement will be unidirectional (always improving, with more and more weight loss). However, within the range that we are considering (obese to normal body weight), the relationship is unidirectional (and so the two are correlated).
In conclusion
If one has osteoarthritis and is obese, it is a good idea to lose weight.
It is better to lose more weight, for improvement in arthritis condition.
A weight loss of 20% or more should become the advice, instead of 10% weight loss, which is the conventional advice.
First published on: 11th November, 2018