My understanding is that it doesn't make sense. You have the correct reasoning: LEDs emit broadband light, with emission at practically any frequency. The emission is dominated by two rather wide peaks, one at blue, another at yellow. Theoretically I guess a filter might somewhat diminish those peaks, but that filter must be explicitly designed to work with LED lights.

For emission nebulae, there are filters that pass only the emission wavelengths of interest, and those make a lot of sense even with LED lights, as the emission wavelengths are very narrow.

However, something might help. If you use a telescope, you want to use a hood (usually sold as dew shield) to prevent light from diffusely reflecting from the internal surfaces of the telescope. Even if the telescope is black on the inside, it may have some screws for holding a mirror (Newtonian telescopes) that are not painted black. Also black still reflects a minor amount of light. If you use a telephoto lens, use a lens hood.

I have experience with telescope astrophotography in light polluted area. I find that without telescope hood, if I photograph in any direction where streetlight gets to the telescope, I get a varying light background, strongest usually on one corner, that's difficult to remove as it's not constant background but varying. Siril's background removal does an acceptable job, but still I'd prefer avoiding that varying background entirely.

I also find that with snow on the ground, the Bortle number of the location bumps up by two numbers, measured using a light pollution meter. This is especially true when snow is on those roads where the streetlights are. If snow on the plowed roads are melted but there's snow elsewhere, the situation is not as bad as with snow on the roads. Moon also hurts, don't even plan on photographing galaxies that are too near a full moon.

Of course, the best images are always made from dark locations. But without snow on the ground and avoiding full moon, you can make an acceptable result from light-polluted areas too. You just need a telescope hood, background removal, good flat frames (for getting rid of the light pollution in the most accurate possible manner) and some image color curve processing skills.

Also the amount of light collected helps. For example, something that would take 15 minutes from a dark area, could maybe be reproduced in 4 hours of exposures from a light polluted area. If there's light pollution, you just need more signal which means longer total exposure time.