Acting on historically offensive content in BMJ’s archive

How should we deal with published content that may be offensive or harmful? New or recently published articles attract the most attention from readers. They are also published in an era when peer reviewers and editors should be sensitised to potentially offensive content, whether, for example, on the basis of race, gender, sexual orientation, religion, geography, or culture. Mistakes and offences can still occur, but they are more likely to be quickly identified and corrected given the immediacy and visibility of the internet and social media.

But what of content that was published many years ago, possibly at a time when the language may not have been considered offensive? Some of the challenges of dealing with historical content mirror those of dealing with historical figures, artefacts, and records that are a product of their time and society.

Removing offensive content seems an attractive way of righting wrongs and preventing harm. However, for a journal this is not as simple as it first appears. Scientific publishers have a responsibility to preserve the scientific record. Best practice in scholarly publishing is that even retracted content remains retrievable. Even when publications decide to correct or remove content, the original version will remain in print editions and in other locations online through third parties such as indexers and libraries.

Defining offensive content is challenging, as is determining to what extent it is harmful, because perspectives vary on whether offensive language or derogatory terminology are inherently a cause of harm. Context is important when judging offence and harm, and guidance created by those who have experienced abuse is particularly useful.1 For example, censoring doctors who quoted racist language when describing their experiences of racial abuse feels unjust2 and may delay or prevent harm from being addressed.

Exploratory work at BMJ suggests that harmful content falls into four broad categories: offensive language (such as racial abuse), offensive views (language that may not be explicitly derogatory but the theme and tone of which would now be recognised as unacceptable); harmful science (research that harms certain groups); and misused content (language or an article that is not offensive but is used to support a harmful agenda) (fig 1). This categorisation is a first attempt and may help other editors and publishers to categorise content, and to decide whether action such as correction or retraction is needed.

Fig 1

Categories of offensive language

• Download figure
• Open in new tab
• Download powerpoint

It isn’t feasible for us to review everything ever published in The BMJ and other BMJ journals. But that doesn’t mean that no action should be taken. We will review any articles referred to us by readers or when we are concerned about a particular author or field of research. We will label those that we consider potentially offensive or harmful content with a disclaimer (box 1). Our aim is to strike a balance between acknowledging potential offence or harm and preserving the published scientific record. We reserve the right to correct or retract historical content as our understanding grows and the public debate evolves.

There is no perfect solution. BMJ recognises that offensive, uncensored content may hurt individuals or groups. Our approach was developed with advice from the BMJ ethics committee,3 individuals with lived experience, and external organisations representing marginalised groups. We continue to work closely with the Coalition for Diversity and Inclusion in Scholarly Communications (c4disc.org) on handling historically offensive content, and the Committee on Publication Ethics working group for best practice in scholarly publishing. We invite readers to share their views on our approach and to alert us to any potentially offensive or harmful content in any BMJ journal. We will keep this policy under review.