Suppose if there is a peak shift or all together different structure (millers index), how can be data be used to determine the stability of the thin film?
Well if you mean thermal stability for example against de-wettig (island formation) than in some way the answer is yes. In case you know your lattice parameter in the equilibrium, you will be able to determine elastic strain in your film. High strain will of course be a driving force for dewetting and island formation in order to permit for strain release.
XRD data can be used to learn the structure of a material and identify which crystalline material(s) is present. If the structure is very different on two substrates, I would suspect that the film has reacted with one of the substrates and is not the same material - or possibly that the material is the same basic material but has grown in different polymorphs due to templating by the substrates (although this does not often occur). I cannot see how it can predict future behavior of the film, however, such as solubility in water.