If the vacuum is attractive, one has to explain voids and the cosmic web ( https://en.wikipedia.org/wiki/Observable_universe#Walls,_fil... ), and it is not easy to do. This is exactly where most energetic-vacuum theories (of which entropic gravity is a flavour) fail to match observations. One can solve this by proposing extra fields, but then you have adapted your energetic-vacuum theory to concord with observations, and thus have produced an even more complicated theory than the standard "concordance" cosmology.

If anything, observations of distant galaxies suggest that one can treat deep intergalactic space as being very very slightly repulsive -- this is essentially what "dark energy" is.

The Casimir effect is only seen where there is manifestly not a vacuum: you need something in the vacuum to observe the effect: an evacuated box containing some conducting metal plates is not a box of classical vacuum any more than a box containing a human in a spacesuit is not a box containing nothing at all.

Because the plates are quantum objects with "free" electrons [1] sloshing around inside (otherwise they would not be conductors), the space around them has virtual photons sloshing around too. One can recalculate the vacuum energy (which is the basis for the pop science description you know) but one can alternatively consider relativistic Debye forces and London dispersion interactions (roughly, "the moving electrons result in inter-molecular attraction"; https://en.wikipedia.org/wiki/Casimir_effect#Relativistic_va... ).

Whichever way to understand the Casimir effect, it is not observed at anything but small distances. Put a couple metres between the plates and gravitational attraction between them will be measurably stronger than the Casimir effect. (Even in modern laboratory conditions it's hard to actually demonstrate concord with the theoretical prediction. [2]) There's no Casimir effect at work in the orbit of the Moon around Earth, but there's plenty of extremely high quality vacuum between the two bodies.

In short, the Casimir effect is the wrong tree to bark up to try to explain why the fast-moving outer stars in elliptical galaxies aren't flung out into intergalactic space.

- --

[1] https://en.wikipedia.org/wiki/Free_electron_model

[2] There are groups studying the Casimir effect, and there is a fairly clear relationship between the amount of jiggle ("temperature") and the strength of the effect, which leads one more to an electromagnetic interaction between the nearby objects. https://jila.colorado.edu/cornell-group/research/hybrid-trap is one such team and some abstracts and links to papers. https://web.archive.org/web/20070210090006/http://www.nist.g... discusses the most relevant one and is more accessible.