A good write-up should probably look at several hypotheses to explain the data, followed by a discussion on which might be the most plausible.
Eg It is more likely that severe symptoms prevent an individual from being able to work,
than it is likely that the work itself causes the individual to have fewer symptoms.
It is interesting that this paper does explore a variety of options for the data they collated.
And as rightly pointed out elsewhere the BPS folk tend to just suggest one explanation, and without another for contrast, theirs might seem a fairly plausible suggestion to a casual reader.
Muddling the discussion in complicated language is another BPS technique used to distract the reader from considering alternative explanations for the same data.