The success or not of a trial ultimately depends upon two key groups: The participants who take part and the clinic staff that take care of them, so training these staff well really matters.
Depending on your perspective, I was lucky or unlucky enough to first do ICH-GCP training quite a while back (1994). ICH-GCP requires that all staff involved in a trial at a site, say a hospital, should be appropriately trained. That’s sensible.
But, in my albeit limited experience, most of this training is dreadful. Whether online or live event, it’s terrible. Most importantly, trial staff at sites who regularly have to undergo such training regularly say that the training is dreadful, but they have to do it otherwise the sponsor company won’t let them be involved in the trial. An experienced research nurse told me that after having completed ICH-GCP training for one trial, she was told a few weeks later by another company that it didn’t matter about this previous training she still needed to do their own ICH-GCP training for their new trial. This is not the study-specific training that follows later, it’s the general introductory training on ICH-GCP and, yes, it’s a bit bonkers.
A quick analogy. I came back on an Emirates flight last week from Abu Dhabi. I got on, got comfortable and then stopped what was I doing to hear the safety announcement. Sensible. But, there are two versions, the English one and an Arabic one, same content, but different languages. As I’m barely lingual in my own mother tongue, I attentively listened to one and ignored the other. That makes sense, but imagine instead, that one of the stewards stood over me dangling a fragrant hot-towel in front of my nose and had said, “You can have this when you pay attention and watch our safety video again.” Implausible, but could you imagine somewhere an airline policy that says this, the company lawyer putting it in there just in case.
Back to firm ground and training clinic staff. The most fundamental problem with this training is it doesn’t teach people about why randomised trials matter and what are those few key things that really matter when doing a trial. Most importantly, it doesn’t focus on those specific things that the trial staff can do to help make the study a success. Why?
The reason for this is that they are not covered in the ICH-GCP guideline so don’t get covered in training, but the mentality of don’t ask questions, why you need to do this doesn’t matter, just shut up and follow the rules makes matters even worse. Also, much of this training is delivered by “professional trainers”, people who have never done a trial of any consequence so don’t have a clue what matters.
That’s my view, your experience maybe very different, but let’s now look at a real example of this.
Enter Peter Graham and Lois Browne. Peter is a Radiation Oncologist and Lois is a statistician out in Kogarah, Australia (I had no idea where that was, Kogarah, not Australia) They do cancer trials and are supporters of MoreTrials, in fact, they were drivers behind the campaign. Their cancer trials need to be done under ICH-GCP guidelines so staff get ICH-GCP training. All good so far. But a paper they wrote in 2014 highlighted that some really key principles of how to do trials well were not mentioned at all in ICH-GCP and so is never covered in the ICH-GCP training of site staff. Specifically, those key principles are intention-to-treat (ITT) analysis and minimising losses to follow-up in general. Technical, yes, but do they matter to the staff doing the trial in the clinic?
Short answer is yes, they do matter.
Long answer is that simply put, the whole point of randomisation, particularly with blinding is to avoid bias in treatment allocation and to take account of chance (randomness). The problem is that randomisation on its own doesn’t take care of these, but randomisation followed by ITT analysis and minimising losses to follow-up does.
ITT means that the analysis of the results is done on the basis of the randomised treatment allocation, not on the actual treatment taken. Let’s look at an example: Patient gets randomised to treatment A, but switches treatment to B. How should they be analysed? Analyse according to the randomisation (ITT analysis), that is put them in group-A you avoid bias, put them in group-B and you introduce all sorts of potential bias and destroy the whole point of going to all of the trouble of randomising in the first place. You might want to read that again as common sense might well be telling you that it would be better to analyse according to the actual treatment that the person takes. Bias, is very cunning.
Minimising losses to follow-up is easier to explain. Let’s say to answer a question reliably requires a 100 patients to be randomised, given that not everybody might not get to the end of the trial, it might be sensible to randomise more than this at the start and try to minimise any drop out during the trial. In a properly conducted trial, you would estimate the likely drop-out before starting and then monitor it during the trial, maybe adjusting recruitment if needed.
So, ITT analysis and minimising losses to follow-up really matter to the reliability of the result, but here’s the important part, they are something about which trial staff in clinics can do something about. This is not about heavy handed tactics to keep participants in a trial at all costs, but there is no harm (and much value) in a trial nurse including in the initial discussions about participating the importance of doing ones best to take part for the full duration of the trial. During the trial, there also seems no problem in the nurse pointing out to a participant who is thinking of withdrawing the importance of continuing to collect follow-up data. Or maybe, not to withdraw a patient who decides they want to stop taking the study treatment. The fact that these are not covered at all in ICH-GCP clearly demonstrates how bad it is, but it gets even worse.
When Lois and colleagues realised that ICH-GCP training for their site staff did not cover these, they added them to the training. Problem fixed?
Wrong………Some of the clinic staff made statements along the lines, “ITT can’t be correct, or it would have been included in ICH-GCP” or “ITT and avoiding missing outcomes is not consistent with statements in ICH-GCP about patient withdrawal.” Peter and Lois eventually fixed this – I’m going to add, think they’ve fixed this – by including new training on these essential aspects which are not even mentioned in ICH-GCP (side-note – there is still no mention of ITT analysis or the importance of minimising losses to follow-up in ICH’s recent update to ICH-GCP – that’s how awful it is).
I’ll let Salim Yusuf have the last word – taken from his 2010 Clinical Trial editorial – on ICH-GCP and training:
“They have also led to a large number of courses focused on the regulations and those who complete them have a mistaken belief that they are capable of running trials, but often have little knowledge of the scientific underpinnings of randomised trials. This is wasteful, and may even be harmful.”