It’s almost impossible to investigate suspected fraud unless you have access to the raw data. That may seem pretty obvious, but it raises the more interesting question of who should be responsible for looking after these data and making sure they are available, if needed.

Cases that frustrated journal editors brought to COPE (the Committee on Publication Ethics) in which authors claimed that their data had, rather conveniently, been destroyed by lab fires, floods, or catastrophic computer crashes, or more bizarrely, eaten by termites, have made me realise that this task cannot be left to researchers. And, while I’m sceptical about the disasters that sometimes appear to strike as soon as an author is confronted with suspicions, let’s not forget that disasters can occur, even to honest researchers, which makes it sensible to store data safely, even if there’s no whiff of misconduct.

A recent investigation into suspected misconduct by Imperial College, London (which, incidentally, found no evidence of fraud) highlights this issue. The investigation report notes that it required “protracted negotiation” between the college and the drug company that had sponsored the research to obtain the lab notebooks. While that’s unfortunate, and apparently delayed the investigation, I wonder, though, what would have happened if the research had not had an industrial sponsor since, in this case, the whereabouts of the accused researcher “remain unknown” and Imperial College failed to contact him and apparently his supervisor (who took part in the investigation) didn’t have the data. The report sensibly recommends that the college should make sure that, in future, it “retains access to all data and laboratory books”. But in COPE’s experience, this vital raw data is all too often solely in the hands of individual researchers and disappears with them if they vanish.

Rather cynically, I wonder if the fact that the lab notebooks were locked away by Novartis might not actually have been a blessing in disguise. Although it led to delays, and an unfortunate tussle between the institution and commercial sponsor, it did ensure that the records were safe. While Imperial College has recognised that it needs to scrutinise sponsorship agreements carefully to ensure similar problems don’t occur, it and other institutions should consider whether they would always be able to access data generated by their researchers should questions arise in the future. Until now, institutions have often seemed reluctant to take responsibility for data curation – this case should remind them of this important duty.

Liz Wager PhD is a freelance medical writer, editor, and trainer. She was the chair of the Committee on Publication Ethics (COPE) (2009-12).