Doctors working on a clinical trial for treatment of heart disease held back key data, Newsnight has been told.
The Excel trial tested whether stents were as effective as open heart surgery at treating patients with a heart problem called left main disease.
The data suggested more people fitted with stents were dying after three years.
It was eventually published - but only after treatment guidelines that partly relied on the trial had been written.
These guidelines recommend both stents and heart surgery for certain patients with left main disease.
Trial authors defend standards
The authors of the trial said it was carried out rigorously and to accepted academic standards.
In the trial, sponsored by US stent manufacturer Abbott, half the patients were given stents, the other half had open heart surgery.
Not all the patients were recruited at the same time. Some were recruited in 2011, others over the years that followed.
So, when the first results were published in 2016, the doctors doing the trial knew there was data about what had happened to some of the patients five years after their stent or heart surgery procedure.
But they chose to look only at what happened up to three years after the patients' procedures and publish that data.
A spokesman for Abbott said: "The study's execution, data collection, analysis and interpretation were entirely performed by independent research organisations. The publication of three-year Excel data reflects the original follow-up period and endpoints the study was powered to assess."
Prof Nick Freemantle, a biostatistician at University College London, said: "If somebody had died three years and one day into the trial, that death wouldn't have been counted in the results.
"I'm absolutely appalled that they've done this," he said.
"I've taken a straw poll of my professional colleagues and it draws disbelief that people would do this," he said
The researchers said the outcomes of the study were analysed and reported according to the protocol.
Newsnight has seen information shared between people involved with the safety of the trial that suggested things were starting to look worse for people with stents after three years. More people were dying than those who had had surgery.
Emails from the the trial's safety committee warned that all the data about deaths should be viewed by the researchers and published.
"It might be very concerning if in the future, suspicions were raised that already available information on mortality was withheld from the cardiology and thoracic surgery community," Dr Lars Wallentin, the head of the safety committee, wrote to the researchers in 2017.
He was worried that major European clinical guidelines were being drawn up by heart doctors about how people with left main disease should be treated and the trial results would be used as part of their work.
But the doctors on the trial chose not to publish the data when the safety committee asked, despite the warning. They published further data after the guidelines were completed.
Even without this additional data, there was disagreement among those writing the guidelines about whether stents or surgery was the better treatment for patients.
Review 'not shared'
An external reviewer was brought in by the European Society of Cardiology to look at a number of trials and resolve the debate.
Newsnight has seen the review. It said that the evidence suggested stents were worse than surgery for those with left main disease.
"I think most patients would find these differences to be clinically meaningful, I do not believe that both these procedures should receive the same class of recommendation," it said.
But the review was not shared with everyone who believed they should have seen it. One of those people was Prof Freemantle, who was involved in the European guidelines.
He claims that this calls into question the neutrality of the whole process.
Newsnight has previously reported that the same trial failed to publish certain heart attack data that cast stents in a bad light.
The researchers said our leak data was fake and their methodology was the right one.
Following Newsnight's previous report, a number of major surgical organisations have called for a review of the trial.
The researchers carrying out the trial have agreed to an "independent" review of the raw data.
Various names have been put forward by the researchers and the European Society of Cardiology about who is doing the analysis. All have ties to the researchers, guidelines process or medical device industry.
When approached by the BBC they have all said they are not doing it.
Prof John Ioannadis, from Stanford University, an expert on medical research design, said the analysis must be completely independent.
"I think that if you have the same network, the same closed club passing the data from one member to another, that's not really very helpful," he said.
He believes the trial and guidelines process raise concerns which are indicative of a wider systemic problem with the way medical research is done.
All the main doctors working on the trial, and the lead doctor writing the guidelines for left main disease, have declared financial contributions to either themselves or their institutions from companies that manufacture stents.
"You have the same people who run the show at all levels. They design the trials. They set the agenda, they choose what to present.
"They are involved in disseminating the information and running the large conferences that are attended by tens of thousands of people, specialists in the field. And then they also populate the guideline panels that reach the recommendations," he said.
The organisations involved and researchers have declared the conflicts of interest, and say that they are effective in managing them. The conflict-of-interest declarations are intended to mitigate against conscious or unconscious bias - or the appearance of it.
You can watch Newsnight on BBC Two at 22:30 on weekdays. Catch up on iPlayer, subscribe to the programme on YouTube and follow it on Twitter.