Skip to content
#UnblindingResearch: Blinding episode artwork

Dec 21, 2018 15 min

#UnblindingResearch: Blinding

from Take Aurally Hosted by Take Aurally

In this live recording of the ninth #UnblindingResearch session on 'Blinding' from DREEAM through the medium of mince pies. It's a seasonal episode that hopefully shows how blinding is a piece of cake! The blog entry containing the slides, #TakeVisually and more info can be found at https://www.takeaurally.com/new-blog-1/2018/11/26/session-ten-blinding You can subscribe to Take Aurally on SoundCloud and iTunes Remember you can find Take Aurally and NUH DREEAM on both Facebook and Twitter The amazing DREEAM Research Team can be found on both Facebook and Twitter Our next #UnblindingResearch session will be on Medication Trials on 16th January! We're off for a Christmas break so we'd like to thank you all for listening in 2018, have a very Merry Christmas and a Happy New Year! We'll be back in 2019!

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

#UnblindingResearch: Blinding

0:00 15:26
of matches

Transcript Auto-generated, may contain errors

Hello everyone, can you guess, I've gone through quite a seasonal theme for today when we're talking about blinding, also blinding also has a pastry connotation for the bakers in the room because you can blind, blind bake your pastry, no that's a shortcrust, no shortcrust pastry for blinding, it's for quiches isn't it, you put it and you put the beads, I remember that from Foodtech. Is that to stop, you get the soggy bottom? Yeah you do the crust first don't you and you put the beads in the bottom and then you make your mix and you put that in and you put it under the grill for a bit. That's how you make the quiches, that's how you make the quiches. I'm a master baker. We did about randomisation last month and you guys ate a lot of sweets and I tossed a coin a lot and things like that. Now we're talking about blinding. LinkedIn along with randomisation, why do we blind? Why is blinding the importance of trials that we do? And what does Matt say?

Randomization is all about removing selection bias and making sure that we're not sticking our best patients into the treatments that we want to be blind. Blindings are a little bit different. We do this for sight reasons along those lines. So you were saying about we don't treat them differently. So that's one aspect about it.

What about? You do things like preventing the patient describing their symptoms differently. So if they know they're getting a placebo, they may be a bit more, oh, this is really painful because it's not working. Whereas if they don't know what they're getting, they get a bit more organised result. Yeah, brilliant. So if a patient knows that they're getting a brand new wonder drug, they are more likely to go, this is the greatest thing I've ever had. Brilliant. I feel so much better every day. And we write that and then we record that. If they know that they've got the sugary water or the sugar pill, they're more likely to...

a be negative in their reporting but also seek other options so we talk about back pain they may think well i'm only on placebo so i'm going to go get some physio as well i'm going to get one of those thames machines as well and things like that oh i've got some benefits and that completely messes up our research as well. So blinding is a very, very important part of research. Is it about hiding different bits of information from different stakeholders? Absolutely. So it's all about a concealment. I like that. So essentially within the relationship, you've got your patient, you've got the patient and their research staff, that could be me as a research doctor or our research nurses, so our research team. And then you've also got the investigators, essentially the three stakeholders. And depending on how many of them are blinded, that depends on how the research is conducted.

which we'll go through in a bit. Cool. Once again, it's a nice friendly environment where we all hold hands and look into the sunset. So as I've already said, say I was a master baker and I opened up my own bakery. That's really real. Making croissants here, as you can see. But say I wanted to make a brand new type of mint pie and I wanted to be the best mint pie in the world and I've got you as my taste, as my guinea pigs, to taste my brand new mint pie.

What trouble have I got straight away knowing human beings the way they are? I would never agree that the person that you've baked would be nasty. Well that's one thing, okay. The fact that you don't trust me to make you a nice mint pie, that's one thing. We might want to tell you it's nice because we don't want to hurt your feelings. There is that, yeah. We might tell you it's not nice because we want to hurt you. There is that, it's probably more likely as well. Good, okay.

So this is something we call performance bias in research and this is what we already talked about with our patient subscribe you know going oh I've got I've been given the wonder drug I suddenly feel so much better or I've only been given the placebo every day my life is getting worse and worse etc etc. So, as an example of that, we all have our favourite mince pies. So as you know, I went around and I asked everybody who their favourite mince pie was. And Aldi and Stainsbury's was there with one each. Local bakery in which we both got two. Three people in Dream don't like mince pies, one of which Amy is one, and Louise is the other one. So we have two of them in our room. And Ben's just out there because he doesn't like mince pies. He'd be in the same room as him at the time. And then, we're a very industrious group. The majority prefer to just make their own mince pies. If you pre-get your chocolate pastry and you pre-get your mincemeat, you've just got to put them together and that's not really making your own.

No, that's like buying a pre-made curry mix. It's like a Lego mint pie. I like that. Anyway, so already I know that people have got their favourites, and if I was to give them, say you as a group preferred the Waitrose one, and I'd go, so this is a Waitrose mint pie, this is my mint pie, and you know straight away, oh, Waitrose, I love this, this is a staple of my Christmas, warm and lovely, yours is rubbish, Jamie, go away. And there's a risk of that. So this is what we call performance bias. This is one of the aspects that we need to avoid, which is why we're blind. The other one is ascertainment bias, and that is a problem on our side as a research group. If you know that a patient has been given a placebo, we are less likely to follow them up properly, give them as much attention as the patient we know has been given the drug.

So if you've got two patients and you know one has been given tranexamic acid, one has just been given some sugary water, the risk is that we follow that one up with tranexamic acid much, much better, follow all procedures properly, as opposed to this one who we know, oh, they've just got placebo, off they go to one side. Are you not more likely to look for problems occurring? There is the other option as well, there is the flip side of going, well, you've been given your tranexamic acid, so I'm going to report everything. Patient reported a headache a day after they had their car crash, must be the tranexamic acid, you know, for example. Patient who was in a car crash and didn't get it, didn't get a headache, oh, that's the tranexamic acid. So there is that flip side as well. So it could be a positive, it could be an negative.

Cool. So as we said, there are three stakeholders, as Matt brilliantly said. You guys are our patients. What we're going to do now is demonstrate single blinding. I'm going to ask Amy to step up. Amy, you are our investigator for this. I am the research team. You are our patients. Hey guys, so you've had three. It's high there. So, over to Amy again. An A, one A, yeah. An A, one A, yeah. An A, one A, yeah. An A, one A, yeah.

I think I'm going for A as well. 2 for B. Anybody for a C? I'm going to go for a C. I'm going to go for a C. I'm going to go for a C. There we go. Okay, that's interesting. So that's an example of double blinding. Does that make sense? So you didn't know, I didn't know, the only person who knew was the investigator. So now we're going to go for triple blinding. Who's going to give out these? I'm blinded. Yeah. Oh, Louise is blinding. So this is where I don't know, you don't know, and neither does Amy. I'm going to go to all blind. Sorry? I've got this on the plates.

So Gus is the one who did this blinding, and he's written under the plates, so we'll find out afterwards. So please do. So the one that says A, please, Louise. Guys, get ready. You can now take off your blindfolds. It's now time to once again go over to Amy, our investigator. Anybody for an A? Oh, I can't remember. Two. Anybody for a B? Three. Anybody for a C? No. Okay, right. So, this is interesting. It is very interesting. So... We're not doing that again.

So this is interesting. So, for your single blind, the one that I did the blinding for, A was a Sainsbury's basic mince pine. B was Waitrose's finest. And C was Louise's home mother. No, Megan, sorry. Megan's homemade mince pie. So for the double, I don't know which was which. Yeah, for the double, was number one, Megan's. Number two was Waitrose.

And number three was Sames Brees. That was a really good buy! Do you vote for your own? And for triple blind, tell me. So, your taste buds have changed you guys. So, for Single, where nobody likes Sames Brees, three of you like Waitrose, two of you like Megans, where the double one, two of you like Megans, two of you like Waitrose, suddenly somebody decided they liked Sames Brees.

and now for triple we are now going to find out we're now unblind in the situation now okay so see what best part number a is homemade oh number two is waitrose and then number three is sainsbury's own brand basic i think it's basic brand as well i mean it's it's yeah So we've got two wins for Waitrose and a draw between Leg and Waitrose. So we've done a bit of blinding, it's a bit of fun, and we've demonstrated a single double, triple blinding. Can you think of any circumstances, just as we're wrapping up, any circumstances in the trial where we can't blind, or blinding is near impossible? Such as? What is too difficult when you see a triple blinding?

Yes, absolutely. If, you know, it's a trial of blood versus something, if you want to make it completely the same color, the same viscosity and all of that, very, very difficult, very, very expensive, but actually you can no longer find your trial good. What sort of interventions definitely do you think? So it's very, very difficult. But it's very, very difficult, especially if you've got a long-term follow-up. If I'm testing this procedure versus this procedure, and I want to follow you up a year later, I'm not going to keep you blinded for a year as to whether or not I may do a new hip. Yes or no. So there's something there, they're called expertise-based

Research, have you heard of that expression? Expertise-based trials. So this is one specifically used in surgery, where you'll find surgeon A likes to do a procedure laparoscopically, surgeon B does it open, surgeon C does it this way, and you get randomised into one of those. So one surgeon only does it this way, another surgeon only does it this way, and another surgeon only does it that way. And then you compare it that way, and that's what we call expertise-based trial in EPT. And that's a way around it, because as I said, it's very difficult to do a placebo of a surgery. You could do it in the first few hours, but a year later you're going to be difficult.

Of course, this only really works really well in RCTs. So in a cohort study, people are going to know if they're a smoker. They're going to know if they have high blood pressure. They'll have been told these things. So it's very difficult to be blinded in cohort. RCTs are really where our timing happens. Always beware a paper that hasn't done any blinding and doesn't admit that it hasn't blinded.

There is no statistical way of getting around the facts out of the data of somebody that's published without doing any blinding. There's no statistical tool to get you around it. So if you find that there's a trial that they say this and they've not done any blinding, always read it with a great deal of trepidation. There's no way around it to manipulate the data to counter for the fact that they haven't done any blinding. There are some things that you can do.

in order to get around if you feel you can't do any blinding in your study. Standardizing your practice as much as possible, so that even if you know that your patient's not had a drug and they're on the placebo, you've got a very clear checklist you have to follow for every single patient in your study as a way of getting around that. Thinking of outcomes as specific as possible. There's another way of getting around that. Duplicate your assessment, so it's not just one research nurse that's recording data off your patient where there isn't any blinding. It's more than one person, so if there's any confidence you can get around that. And first and most important, always acknowledge and be very honest in all of your research. If you haven't done any blinding, please point it out and be open and honest about it as a major flaw that's in your data. Because if you don't, somebody else will. Any questions about all of that? Pull up a minute to panic? And on that note, thank you. Merry Christmas, everybody. Merry Christmas. May you have an evidence-based new year. Blinded. Blinded.

Frequently asked questions

How long is this episode of Take Aurally?

This episode is 15 minutes long.

When was this Take Aurally episode published?

This episode was published on December 21, 2018.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this Take Aurally episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.