4 ms·

This strikes me as a) not very well written, and b) not presenting all the relevant information. For example, if the study was designed to gather data for three

This strikes me as a) not very well written, and b) not presenting all the relevant information. For example, if the study was designed to gather data for three years after each procedure, then taking into account data beyond that period would possibly taint the analysis. (Of course, if people were dying in year four, there could/should be another study.) I had to read the article twice to understand the main point. Sentences like this make me scratch my head: "The data suggested more people fitted with stents were dying after three years."

satya71 7y ago
Exactly, the last two year survival of control group may be artificially enhanced because the ones that were going to die were already dead in the first three.
- James_Henry 7y ago
  Why would they compare the death rate in just the fourth year between the two groups instead of doing a normal survival analysis?
  - satya71 7y ago
    That's what the article seemed to imply. That mortality was higher after year 3 in the intervention group compared to the control group.
mohammad_ali85 7y ago
Agreed. It isn’t clear what the issue is here. If the research protocol states 3 years of data collection it’s technically unethical, without submitting an amendment, to collect data beyond 3 years as the patient hasn’t agreed to this. Sure, in hindsight the data should have been collected for longer but at the time of study design 3 years probably seemed appropriate. It’s really difficult to say how long to follow up - forever? Sure, but would be financially unfeasible. We do have a yellow carding system in the UK - the medicine and health regulatory authority (MHRA) run this and it’s for submission of adverse drug reactions as well device issues. But again, not sure how helpful it would have been for the instances of participant death after 3 years. Seems harsh on the doctors who carried out the research, unless I’m not reading this correctly.
- James_Henry 7y ago
  They were already collecting data past 3 years. The doctors who carried out this research are, it sounds, blatantly in the wrong, for they withheld the data that appears to make clear that stents have worse outcomes past year 3 until after the guidelines were written while they had already gathered this data before the guidelines were written. Sure they cannot openly share all their data because it is private, but it is very suspect that they do not share the external review that found them at fault As Prof John Ioannadis points out in the article, there are wide systemic issues in the trials and guidelines process.
  - u801e 7y ago
    > Sure they cannot openly share all their data because it is private Couldn't they just anonymize the data?
    - ska 7y ago
      This depends on jurisdiction of course, but in a lot of cases you can't share data that wasn't consented for that use originally - anonymized, de-identified, or otherwise. Chasing down previous patients for retrospective consent would be difficult to impossible.
James_Henry 7y ago
Here is a better written article with some of the controversy over the EXCEL trial. Medical decision making is something of a mess. https://www.dicardiology.com/content/excel-trial-authors-say-european-surgery-society-claims-bad-science-are-groundless https://www.dicardiology.com/content/excel-trial-authors-say...

satya71 7y ago

Exactly, the last two year survival of control group may be artificially enhanced because the ones that were going to die were already dead in the first three.

James_Henry 7y ago
Why would they compare the death rate in just the fourth year between the two groups instead of doing a normal survival analysis?
- satya71 7y ago
  That's what the article seemed to imply. That mortality was higher after year 3 in the intervention group compared to the control group.

mohammad_ali85 7y ago

Agreed. It isn’t clear what the issue is here. If the research protocol states 3 years of data collection it’s technically unethical, without submitting an amendment, to collect data beyond 3 years as the patient hasn’t agreed to this. Sure, in hindsight the data should have been collected for longer but at the time of study design 3 years probably seemed appropriate. It’s really difficult to say how long to follow up - forever? Sure, but would be financially unfeasible. We do have a yellow carding system in the UK - the medicine and health regulatory authority (MHRA) run this and it’s for submission of adverse drug reactions as well device issues. But again, not sure how helpful it would have been for the instances of participant death after 3 years. Seems harsh on the doctors who carried out the research, unless I’m not reading this correctly.

James_Henry 7y ago
They were already collecting data past 3 years. The doctors who carried out this research are, it sounds, blatantly in the wrong, for they withheld the data that appears to make clear that stents have worse outcomes past year 3 until after the guidelines were written while they had already gathered this data before the guidelines were written. Sure they cannot openly share all their data because it is private, but it is very suspect that they do not share the external review that found them at fault As Prof John Ioannadis points out in the article, there are wide systemic issues in the trials and guidelines process.
- u801e 7y ago
  > Sure they cannot openly share all their data because it is private Couldn't they just anonymize the data?
  - ska 7y ago
    This depends on jurisdiction of course, but in a lot of cases you can't share data that wasn't consented for that use originally - anonymized, de-identified, or otherwise. Chasing down previous patients for retrospective consent would be difficult to impossible.

James_Henry 7y ago

Here is a better written article with some of the controversy over the EXCEL trial. Medical decision making is something of a mess. https://www.dicardiology.com/content/excel-trial-authors-say-european-surgery-society-claims-bad-science-are-groundless https://www.dicardiology.com/content/excel-trial-authors-say...