I am a scientist myself (obviously in one narrow field) and my approach to trusting the science is that unless I can myself evaluate the evidence, data, analyses, I'll trust the scientific consensus. I will not necessarily trust single papers, as people may make mistakes, be dishonest etc. Obviously the scientific consensus may also turn out to be wrong (it happened before), but realistically it's unlikely that I'll have unique knowledge to find this out myself not being expert, so 'trusting' the consensus is my approach.

Also in rare cases where I can actually read and understand the paper outside of my field that may or may not increase or decrease the belief in its correctness. Realistically outside my field of expertise I'll be only able to see things like dodgy statistical methods, small number statistics, crappy plots which are clues of untrustworthy results, but not proofs.