I am not a statistician but FWIW I think “negative selection” is a false friend here.
“Negative selection” would be the elective removal/exclusion of patients who might detrimentally affect the results of a study/clinical trial or not be suitable for treatment as in the example in #1 (i.e. not selected) and their exclusion would give rise to negative selection bias.
As MusicToMyEars pointed out in #2, this is not how Negativauswahl is being used in the context of the two examples given in the OP. These patients had negative characteristics that might have introduced selection bias but were still included.
In OP link 1, these patients were included but adversely affected the results due to their pre-existing conditions – in other words there was an error in choosing the individuals to be part of the study.
It would be better to re-write, e.g.:
=> “In addition, there was a clear selection bias in the population due to the high proportion of previously operated patients with inflammatory complications.”
“In addition, the population included a high proportion of patients who met negative criteria of previous surgery and inflammatory complications.”
OP link 2:
The second example is not of “negative selection” – these patients were not excluded from the study. This was the deliberate inclusion of patients who had received at least three months’ unsuccessful treatment:
=> “These were patients who also met negative criteria: they had been treated unsuccessfully for at least three months with other drugs, including corticosteroids and immunosuppressants".
[Edit: multitasking so hadn't seen Mattes' response #4]