Three years ago the British Medical Journal investigated hospital whistleblowing policies. The investigation involved asking hospitals for a copies of whistleblowing policies and using the Freedom of Information act, then entering the information into a spreadsheet and analysing the data.
Deborah Cohen, investigations editor of the BMJ, said after reporting on a whistleblowing-related story in 2009 it decided to look at the issue more broadly by accessing whistleblowing policies from NHS trusts.
"You want to foster open dialogue in hospitals where there is patients' welfare at stake," Cohen said. "What we thought we'd do is look at how punitive they were towards whistleblowing and how much they fostered an open environment through their own policies."
She explained that they did not simply look at the number of gagging orders in place, as there can be a number of reasons for one being issued, including business confidentiality when a member of staff leaves.
Faced with more than 600 trusts, Cohen said they needed "to keep the numbers manageable".
"You need to set parameters, you don't want to cherry pick", she said, explaining that they could have asked a random sample of trusts for policies.
"What we decided to do was focus the investigation on the most business-oriented of hospitals, the foundation trusts." This gave a sample size of 122, the number of foundation trusts in November 2009 when the investigation started.
The BMJ journalists then needed to get hold of the policies. They did this by phoning hospital switchboards and getting put through to the HR department or press office.
There were a few that required Freedom of Information requests, Cohen said. "On the whole hospitals were happy to give us the policies."
Four refused and Cohen decided they had enough data so not to appeal, but did mention those which had refused within the article.
Analysing the data
The team then entered the information into a spreadsheet.
"We looked at several aspects through the policy. Do they specifically say gagging clauses are not in place? Do they infer that they are in place? Do they say things like 'if you have concerns you should not go outside the hospital'? Is there threatening language? Do they provide whistleblowers with adequate structure and support and alternatives? It might be if later you have to go to your manager first and the problem might be with your manager, so mandating that in a whistleblowing policy is not a good thing. So did they provide employees with a list of alternatives if they had concerns?"
Once all the data was in the spreadsheet the BMJ team was able to question foundation trusts on particularly good or questionable parts of the policies, asking why they had opted for a particular approach to whistleblowing.
But Cohen said it is vital to take care when dealing with data. "When it comes to data be careful, interpretation is everything. And there is a qualitative element to it in that it is our judgment."
She warned: "You do have to be very careful in how you interpret what you find. I know there's a massive trend for data journalism at the moment. I've seen it done very badly and I've seen it done well, but there are pitfalls with numbers and how you interpret what you find."
An example of over extrapolation would be finding that "more people with blond hair die of cancer and therefore blond hair causes cancer", she said.
"There might be an association that is worth further investigation, but you can't go beyond that statement 'more people with blond hair die of cancer'".
The whistleblowing investigation is just one of many investigations undertaken by the BMJ, including an in-depth study into Tamiflu. Some investigations are carried out independently, while for others they collaborate, teaming up with investigative teams from Panorama or Channel 4 News, for example.
Cohen has led the investigative unit at the BMJ since it was created three and a half years ago. She is medically trained, studying journalism after training to be a doctor.
Free daily newsletter
- Bureau Local announces fund to support local reporting from journalists and members of the public
- Tip: Recommendations for working with inconsistent data
- Tool for journalists: Flourish, for creating data visualisations without coding
- Tip: How to tell stories using small data sets
- One day, 5 cities, 160 people: Inside the Bureau Local's investigation into local budgets