I found an error in an earlier study. Should I point it out?

The first question: "While reviewing the literature for the discussion of my clinical study, I found an important earlier study in which the authors seem to have made an error in handling the data, so the direction of the effect and the conclusion are wrong. Can I write a letter to the editor? In my own paper, can I explain their error in the discussion, and show that my study handled the numbers correctly?"

The second question: "In my meta-analysis, I found an earlier meta-analysis by another group. Going back to the original studies, I found they copied values incorrectly and made many calculation errors. In my updated meta-analysis, do I need to describe their errors in the discussion and say that we corrected them?"

These two questions are similar: while working carefully through the literature, you find errors in earlier papers, and you wonder how to respond, or whether to write to the journal.

Different researchers handle this differently. What follows is my advice for beginners. If it differs from what others tell you, weigh it for yourself and take my view as one reference.

Some senior researchers are more combative. They believe that when you find an error in someone's paper, you should seize the chance to write a letter to the editor. It increases your visibility in PubMed and shows people that you are skilled and careful in this field.

I think that early in your career, trying this once or twice may be fine, but I don't recommend making it your path. The main reason is that once you have made original contributions and become known in your field, people will look up how you got started. We don't want other scholars, or even our own grandchildren, tracing our footprints in PubMed and concluding that we were someone who liked to find fault with others everywhere, and was proud of it.

So if I really do write to correct something, I usually choose a gentle tone: I ask a question, or present my own calculations and ideas, and ask the authors for their view. That is calmer. I also choose topics carefully; I don't write every time I see an error.

In letters to the editor, I more often write the kind that says, "This study is really interesting, and I would like to add my own experience," rather than the kind that says, "You did not do this well enough; let me show you how, and please respect and thank me."

In the discussion of our own papers, when we discuss the literature, we explain what earlier studies achieved, what different things we add, and how we move science a step forward. Discussing the literature is not about dismissing others and then saying how good we are. These two ways of writing are completely different and leave very different impressions.

It is like online: some people must first put others down to look superior, while others always politely thank people for sharing information and then share their own views.

The reason not to dismiss others is that you will certainly cite the paper you criticize. When the original authors later track who has cited their work and see the criticism, they may easily take it personally and see it as an attack on them. Later, they might check your papers one by one in detail, and write to the journal about each error they find.

Since we intend to work in this field for the long term, I suggest making fewer enemies.

As for meta-analysis, if I find that earlier authors extracted data incorrectly, I also don't write in my paper where they went wrong. I simply extract the correct values, add the update, and write a complete paper. The errors in earlier work are, for us, just "someone else's" errors. I only need to make sure my own analysis this time is correct.

In the discussion, I emphasize what new meaning my update brings: for example, new subgroup analyses, a new meta-regression, twice as many included studies, or a particularly interesting new perspective, rather than saying that others copied data wrongly or miscalculated and that this time everything is correct.

In research we handle so many numbers that no one can guarantee they will never make a mistake. Being a little more forgiving of others' honest errors is also a kindness to your future self.