Aspects of Placebo Response

Transcript

Aspects Of Placebo Response

[00:00:00] Alexander: Welcome to another episode of the effective statistician. Today we want to talk about placebo. There are a lot [00:00:10] of misperceptions actually about placebo. The first misperception is that it is no treatment. [00:00:20] Well, that is in most cases not true because placebo means that you [00:00:30] just have a placebo treatment on top of standard of care.

Subscribe to our Newsletter!

Do you want to boost your career as a statistician in the health sector? Our podcast helps you to achieve this by teaching you relevant knowledge about all the different aspects of becoming a more effective statistician.

SUBSCRIBE

[00:00:36] Alexander: And so, That usually doesn’t mean no [00:00:40] treatment whatsoever. So that is the first thing that we need to understand. It doesn’t mean no treatment. It just means that [00:00:50] beyond the all the usual things that the patient gives and that the protocol allows, of course, There is also this placebo injection [00:01:00] pill whatsoever.

[00:01:01] Alexander: Now even if there’s very, very little standard of care, there is still usually a [00:01:10] lot of placebo response. Well, of course, it depends on the different indications, but it’s surprisingly big and [00:01:20] widespread. Now, there’s a couple of things that can help. drive placebo response. And one of the important things is the [00:01:30] perception of the patients.

[00:01:32] Alexander: Now, what is driven? What actually drives the perception of the patients? So first, there is [00:01:40] the ratio of the randomization. If you have a one to one randomization, The placebo response [00:01:50] is sometimes actually smaller than if you, let’s say, have a randomization of three to one in [00:02:00] favor of the active group.

[00:02:02] Alexander: That has been studied in depression studies. And so that ratio, so the, the [00:02:10] expectance of whether, how likely it is that patients get placebo actually drives the placebo response. Another thing is of course [00:02:20] the strength of the anticipated treatment. So these can go in various directions. If the strength of the [00:02:30] treatment response is very big, Patients will notify and notice that they are not on treatment.

[00:02:37] Alexander: Yeah, because they [00:02:40] know that, well, if I take that medication, then my symptoms surely will go away. And only if I don’t take it, then they will stay. So the [00:02:50] perception about the strength of the medication will definitely drive it. Actually, also the on the other side. So there’s also the so called [00:03:00] nocebo effect.

[00:03:01] Alexander: So patients will experience side effects. Under placebo if they think these side effects [00:03:10] should occur under the actual treatment. Also, the pre-treatment might drive placebo response. So if they [00:03:20] already had exposure to similar treatments, things like that, that can have an in impact on the strengths of the placebo response.

[00:03:29] Alexander: [00:03:30] So naive patients usually behave. differently to placebo than pre created patients. The other thing is then [00:03:40] of course also the perception from the physicians because there is very often a very close relationship between the physicians and the patients and [00:03:50] so the perception about of, you know, the physician likewise As the patients will drive these kind of different things [00:04:00] and don’t think that patient reported outcomes are an exception from that.

[00:04:06] Alexander: There is a very, very strong bond [00:04:10] between physicians and patients in many cases. And so any expectations on the physician side, but also transition over to the patient [00:04:20] side. Now, these are. Very often, you know, the most commonly mentioned placebo responses. And [00:04:30] when you talk to non statisticians, they say, Oh, yeah, well, placebo response, but, you know, you can also see it in horses or, you know, all kinds of other [00:04:40] animals.

[00:04:40] Alexander: And they don’t have these perceptions and things like that. Well, Yeah, there’s still the observer bias. However, [00:04:50] there’s also the regression to the mean and just the natural variability of the condition. Usually, we [00:05:00] study patients only when their symptoms are very, very strong. And when there is a natural [00:05:10] variation over time, then just by chance, and just by, you know, how patients would usually [00:05:20] evolve.

[00:05:20] Alexander: That is especially true, for example, in depression, anxiety, pain, and other many other areas where [00:05:30] symptoms fluctuate over time. And when we only include patients with a certain level of symptoms, yeah, [00:05:40] certain severity of such symptoms, then naturally they will always kind of improve afterwards. Okay.

[00:05:48] Alexander: There’s another aspect and [00:05:50] that is the intensity of how we kind of study the patients in a sense. [00:06:00] So if we have a lot of visits, if we’ll have a lot of instruments, if we have a [00:06:10] lot of different measurements, all kind of different things, This will drive placebo response, because this kind [00:06:20] of intensity of care of the interactions will drive placebo response.

[00:06:27] Alexander: This is especially true [00:06:30] in mood disorders, for example. If you have more interactions with the physicians, just that by itself. can drive improvement [00:06:40] in terms of the symptoms. The administration wrote might also play a big role. So different [00:06:50] administration routes can lead to different placebo response, just because the different perceptions that it comes with.

[00:06:59] Alexander: [00:07:00] Patients will think about a pill very different to, for example, an injection or multiple injections, many injections, yeah? This [00:07:10] might drive a bigger placebo response. And finally, just different other design aspects within the study [00:07:20] may have an influence on the placebo response. The duration of the treatment or the placebo treatment can have [00:07:30] an influence and also the ability whether patients can switch to another treatment thereafter [00:07:40] once they haven’t responded.

[00:07:42] Alexander: can have an impact on that. So be careful for all these different factors. [00:07:50] There are lots of different design aspects that can help you to, for example, blind these things. Different depression, [00:08:00] anxiety, and other studies have, for example blinded lead in periods. So the patients actually don’t know when they will [00:08:10] switch from placebo to active treatment.

[00:08:13] Alexander: or whether they, when they switch from active to placebo treatment, or they [00:08:20] might, you know, have always a certain time period and only thereafter they can start with the new treatment. [00:08:30] There are many, many different ways you can reduce or manage the placebo response. You can even have a run in period [00:08:40] on placebo and then only randomize.

[00:08:43] Alexander: those that haven’t responded to placebo. Now these are of course all kind of different [00:08:50] things that also potentially have an influence on the, on the estimate and they definitely should be discussed with the different regulators. [00:09:00] Helping your non statistical colleagues to better understand placebo response and all the different [00:09:10] factors that Influences will also be really, really important so that they understand all these [00:09:20] different origins of placebo response and don’t assume that they know everything about placebo response.[00:09:30] 

[00:09:30] Alexander: So this was a shorter episode about a little bit more technical problem. Depending on your therapeutic area you’re working on in it, this [00:09:40] will be a bigger or smaller problem. I know that in certain indications, it’s no problem whatsoever, at least on the efficacy side, [00:09:50] because the treatment effects are that big.

[00:09:53] Alexander: And then of course the study more or less unblinds itself. or side effects are that [00:10:00] big that it also unplugs itself. Well, nevertheless, Think about what happens in your [00:10:10] specific area, and sometimes it even makes sense to invest more into understanding placebo [00:10:20] response in your area. For these cases, it is really helpful to work across different companies across the [00:10:30] industry, so said Esa Hall.

[00:10:33] Alexander: You can understand better placebo response and what drives it. For that, there’s [00:10:40] lots of different data sharing opportunities. And there’s, for example, the clinical study data request consortium that you [00:10:50] can have a look into. And there’s some other consortium. It could also be said, potentially you can work with regulators on that.

[00:10:58] Alexander: Maybe they can [00:11:00] be a help there. Thanks a lot. I don’t know that specifically. So if you found this episode helpful, [00:11:10] then please share it with your colleagues. This is an episode that is beyond the 350 mark [00:11:20] episode of this podcast. And there’s. Lots to learn within the statistics field, not just about technical things like this, but also many, many [00:11:30] other aspects. So please recommend this podcast to your colleagues.