Charlotte,

Generally, the peer review process precludes the need to do this. With that said whenever claims of misconduct, often plagiarism or falsification of data are purported, editors can address these concerns with authors. We have had to ask authors to clarify or withdraw their submissions in the past. I have not, at the journals I have edited, had to pull an article after publication as of yet. There have been several studies on medical research that has led to such retractions.

http://www.nejm.org/medical-articles/retraction

http://www.nature.com/nm/journal/v9/n9/full/nm0903-1093.html

http://retractionwatch.com/

http://www.the-scientist.com/?articles.view/articleNo/38743/title/Top-10-Retractions-of-2013/

Hi Charlotte,

At PLOS ONE (the journal I used to run), if a complaint came in about a paper, for example, questioning the integrity of the data, we would first of all write to the author asking for them to comment on the allegation. If the author admitted that the data had been manipulated (usually by blaming an errant postdoc or lab technician!), we might move to retract quite quickly. More often, however, the author would not admit that anything untoward had occurred, and in those cases we usually referred the case to the author's institution. It is very difficult for journals to know who is telling the truth in most cases, and so we rely on institutions to perform investigations, for example by interviewing those involved and looking at the lab notebooks. These investigations usually take a few months, hence the long delays in the process. 

Unfortunately not all journals take complaints seriously enough (they often feel it shows a failing in their peer review and so they try to brush away the allegations), so it's not always clear whether the delays are a result of ongoing investigations or deliberate stalling by the journal. A good journal should keep the complainant updated of the progress, but it is sometimes worth chasing to make sure things are moving!

Hope that helps.