I've reviewed more than 50 papers, and my preparation to review them has varied widely. At least a few of them have drawn heavily on papers I've written. Generally, these are extending my results. Typically, I ask myself: What does this paper add? Is it something I would have thought of easily or not? If yes, why didn't I include it in my paper?
In contrast, I've also reviewed some papers that I was not well prepared for. They were in my general field, but in a subarea that I hadn't worked in. These papers took a lot more work. (Up to 10x as long for me to review as a paper in an area where I'm very familiar.) For papers in an area that's new to me, I often have to look up (and skim) at least a few of the references from the introduction to judge whether the paper is original. Originality has at least two very different levels. The easier one to achieve is: (1) this result is new, and doesn't follow simply from anything previous. The much harder one is, (2) this technique is fundamentally new and produces interesting results. If the results are interesting, most journals are happy to publish articles of type (1). Articles of type (2) are much rarer, and can potentially open entire new fields.
tl;dnr:
- Preparation: you need to either know the subarea well or be willing to read a lot to learn about it.
- Originality: What papers does it draw on? How likely is it that a reader of those would be able to write this paper? Answering this question may require you to read a lot of background.
