Bandshifts in ATR spectroscopy can be caused by anomalous dispersion. (Biochimica et Biophysica Acta 1422 (1999) 105-185). The detailed difference between the bands that you are looking at in transmission and ATR, however, mainly depend on the exact treatment of your sample in both cases. For instance, it matters whether you measure your sample in dilute, or concentrated solutions, as composites, in matrices, gels, etc. or as monolayers on substrates. Then differences in band positions may indeed tell you about decisive structural changes of your sample upon changes in the environment. From the information that you provide, one cannot tell more.

As above, these can be due to morphological effects in your sample, or to intrinsic chemical changes. 

"Reflection contributions to the dispersion artefact in FTIR spectra of single biological cells",

Paul Bassan, Hugh J. Byrne, Joe Lee, Frank Bonnier, Colin Clarke, Paul Dumas, Ehsan Gazi, Michael D Brown, Noel W Clarke, Peter Gardner,

Analyst,134, 1171 - 1175 (2009)

“Resonant Mie scattering in Infrared spectroscopy of biological materials - understanding the dispersion artefact”,

Paul Bassan, Hugh J. Byrne, Franck Bonnier, Joe Lee, Paul Dumas, Peter Gardner,

Analyst, 134, 1586-1593 (2009)

"Resonant Mie Scattering (RMieS) Correction of Infrared Spectra from Highly Scattering Biological Samples Analyst",

Paul Bassan, Achim Kohler, Harald Martens, Joe Lee, Hugh J. Byrne, Paul Dumas, Ehsan Gazi, Michael Brown, Noel Clarke and Peter Gardner,

Analyst, 135, 268 - 277 (2010)

Are your shifts reproducible when you take different samples of the same material or change tickness? Shifts can also be a result of mechanical force, which can bias an ATR spectrum. We've seen such effects in the spectra of polymer membranes (http://dx.doi.org/10.1016/j.memsci.2013.10.022).

unfortunately I didn't repeat the ATR, so I don't know whether the peaks are reproducible.However my composites are very thin polymeric films(50 um )