Short Answer: Provide your formal recommendation about acceptance to the editor and not in your comments to the author.
In my field, most publication systems include a multiple choice question where reviewers indicate their recommendation to the editor (e.g., reject, major revisions, minor revisions, accept, etc.).
I generally find that most reviewers do not include a recommendation in their review that goes out to the authors.
Instead, the explicit recommendation is usually made using the recommendation check box in the editorial system and may be combined with additional information supplied in the "comments to editor" box.
There a few reasons why a reviewer may not want to place their recommendation in the "comments to author" box.
- It is the reviewers role to evaluate the quality and importance of the manuscript. The editor is responsible for the decision about what is suitable for the journal, all things considered.
- By not putting a recommendation in the author comments box, the editor has a little more flexibility in how they frame the response to the authors. The editor can choose to share the reviewer recommendations about acceptance with the author or not. For example, the reviewers might both recommend major revisions, but the editor decides to reject. The author sees all the problems identified by the reviewers and may more readily accept the editor's decision.
Of course, the review that goes to the authors may be suggest what the reviewers recommend. Most reviews involve an initial paragraph that (a) summarises the reviewer's understanding of the manuscript, and (b) provides an overall synopsis of whether the topic is important, interesting, and suitable for the journal, and whether the paper is of acceptable quality. More detailed comments will also imply the degree to which the paper could readily be refined (i.e., suggestive of revisions) or has fundamental shortcomings (i.e., suggestive of rejection).
As an additional note, I generally try to keep any note to the editor quite short. I like the transparency of as much as possible of my review going to the authors.