Hi Frida, I don't know which LCIA method and database you used but to me this sounds quite reasonable. After all, energy is required to produce electricity, or not? So if you change the latter, the former will be affected too.
If you look the LCIA method Cumulative Energy Demand LHV [v1.0.1, January 2015] for instance, you will see that the impact categories are quite likely to capture flows that are also connected to electricity generation processes.
If you use a database with unit processes (e.g. ecoinvent) you can also trace back where this affect happens (via the used in/generated by view).