The Data Hair Loss Studies Hide (massive problem)

Added: 2026-06-01

[00:00:00]I'm going to tell you about something that you almost never see in hair loss research papers that you see all the time in other fields of science that take truth-seeking seriously. Once you know about this, you can't unsee it. And in my eyes, this is really important because it has critical ramifications with respect to the efficacy and the safety of the interventions for hair growth that you might actually want to try one day. My name is Rob English. I'm a researcher who focuses on hair loss disorders. [music] I'm on the editorial board of a dermatology journal. I publish manuscripts related to hair loss. I referee papers on hair loss disorders. And I make content like this to help individuals who are fighting hair loss and looking for a straightforward path based on the evidence.

[00:00:41]Now, over the weekend, I was reading a paper in exercise science that sought to answer the question about training volume and its impact on athletic performance. These researchers randomized their participants into groups where they exercised only once per week all the way up to them exercising five times per week. And then they sought to determine how they performed on fitness tests at the end of those training windows. The researchers then represented the average outcomes across each of these subgroups. But they did something else. They showed you all the individual data points of each athlete within those subgroups. And this was critically important because then it allowed the researchers to do another thing. For individuals who were non-responders to that training stimulus, what happened when they upped their training volume? This is when those researchers took individuals at the bottom of those subgroups and moved them into a higher volume training load.

[00:01:32]And lo and behold, they found that non-responders to training just need more volume. That's what that study found. And I thought this was an absolutely fascinating study design. But it also made me reflect on something that this study did very well that is almost never seen in any hair loss research treatment-related studies. They showed us every single outcome across all participants for all training stimuli. In other words, we got n equals one data on every single person who was doing this study. We knew exactly where they fell for their training stimuli. We didn't just see the averages and the standard deviations, we saw every little data point.

[00:02:10]That is something that is almost never seen in hair loss research, full stop.

[00:02:15]Now, you might be listening to this and thinking to yourself, Rob, who cares about individual data points? We want average outcomes and standard deviations. That way we can capture roughly where we're going to net for any outcome related to the intervention we're trying. And I understand where you're coming from. But, let me give you three examples as to why this type of exercise is actually hugely impactful.

[00:02:36]Not only does it help to detect semi-fraudulent research design, it also helps to set expectations and potentially even enable new discoveries in hair loss research that I don't think are happening in part because of an absence of this level of evidence.

[00:02:53]First, when you show individual data points like this, alongside average expected outcomes, you can actually set realistic expectations for the variability in results for an intervention that you might want to try.

[00:03:06]So, let me give you an example of this.

[00:03:07]Back in 2019, we published a survey-based study on scalp massages, and we sought to determine the subjective impact on scalp massaging over a number of months to years for individuals who opted into this experimental intervention. And on average, we found that the scalp massages helped improve subjective scores for hair improvements. But, even with statistically significant results, the study had extreme limitations. We didn't have a placebo group, these were subjective metrics, not objective metrics from some sort of investigator who was observing these individuals, or objective hair counting data. And so, what we wanted to do was show as much data as possible in that paper. So, we showed the average outcomes and the trend lines, but we also showed every single individual data point. Now, this is one study on one experimental intervention, but there's a relationship that we can derive from what we're seeing here. And it's one that actually runs very consistent across almost all hair loss treatments. It's that when an intervention is more poorly supported, meaning that it doesn't have a ton of studies or replications behind it across different research groups. Those interventions, they tend to have hyper variability in their results. We see this with scalp massages, we see this with micro-needling, we see this with PRP, we see this with interventions like that. You will find hyper-responders to these things. In fact, we point one out in this study. It's one of the last data points here. You can see their before and after photos over around a 3-year period right here.

[00:04:34]And yet, is that an expected response?

[00:04:37]Does it score above that trend line? And these are the things that you can actually begin to glean when you have all the individual data points. And the relationship that we see here is that the lower the quality of evidence for an intervention, the more variability that you will get across responders. The wider those individual data points will stretch from the trend line. Meaning that you can get great results or you can get terrible results. That's just what we see consistently amongst interventions with one or maybe two studies supporting them. But, the higher the replication of research for a single intervention, the more groups of investigators that have studied that intervention, the better the hair metrics. We often see that band tighten.

[00:05:23]We see the spread of those individual data points cluster more closely around the trend line. So, that typically happens with things like minoxidil and finasteride and even oral dutasteride.

[00:05:33]The individual data points here, while there are certainly non-responders and there are certainly hyper-responders, the trend line is more tightly bound to the individual data points. And you can often infer that even without the individual data by looking at the standard deviations within these studies. They're just tighter than say the interventional research on PRP or micro-needling or scalp massaging. It's just the way that it is. So, having these individual data points can not only help you to determine your potential magnitude of growth in terms of what's expected, but it can also help you to determine if the variability in the results even make you want to try something that could be so experimental.

[00:06:11]Because you might not want to put time, money, and energy into something that has a positive trend, but a lot of variability.

[00:06:18]The next thing that happens when people include these individual data points in their study is that we can detect bias in hair counting methodologies or study design methodologies. One of the best examples of this is a CBD study that showed a 246% improvement to hair counts in the study itself. Now, as a headline, that sounds absolutely incredible, and we would all be excited in using CBD oil as a topical if those were the expected outcomes.

[00:06:49]But, this study, which I give it absolute credit for, also decided to include each individual before and after hair count in their analysis. And what we can see here is that these starting hair counts were from zones that had really, really low numbers of hair, like four or five hairs, six hairs, seven hairs. And if you start with four hairs, and I add four more hairs to a square centimeter area, that technically gives you eight hairs. I just increased your hair count by 100% relative to baseline. But, those four additional hairs add no cosmetic value.

[00:07:26]And because the researchers in the CBD study gave us this information at a participant level, we were able to determine that that 246% number was more so due to the law of small numbers, and less so due to real appreciable hair count changes had they used better haired areas like saying in the larger scale clinical trials on finasteride and minoxidil which had more hair in those areas and had better hair counting methodologies.

[00:07:54]And the third thing that I think that will happen when you have this level of granularity of n equals one data across all participants is that you will actually enable an ability to make bigger hair loss breakthroughs. First, I want you to watch this clip very briefly of Gary Nolan. He is an absolute juggernaut from Stanford in his field which has nothing to do with hair loss.

[00:08:14]>> It's not the data that falls in line that's that's so interesting. It's the spot off the graph that you want to understand.

[00:08:24]>> Yeah.

[00:08:24]>> When something is way off the graph, that's the interesting thing because that's usually where discovery is.

[00:08:31]And the number of times that I've stopped people in my lab and said, "Wait a second, go back a few slides. What was that?"

[00:08:39]And then it end up being something interesting that made their careers. I can, you know, count on a few hands.

[00:08:46]>> Now I want to take an example like that and walk you through how you can employ that same logic as a researcher to begin to explore why somebody becomes a hyper responder to a treatment or why somebody is a terrible responder to a treatment.

[00:09:01]So for this hypothetical example, let's take finasteride and let's say there's a range of responses and one person in this study is an absolute mega responder and another person in this study absolutely did terribly. They got even worse hair loss. Let's start with the mega responder. If we have this n equals one data on this mega responder, first we can sequence their DNA and begin to build a database repository of individuals who become hyper responders to finasteride and then ask the question, is there something in their DNA that explains this? Next, we can further analyze their hair situation.

[00:09:36]Did they actually fit the standard profile of somebody with androgenic alopecia, or did they have androgenic alopecia alongside another hair loss disorder like telogen effluvium, which self-resolved during the window of the study, and actually boosted hair counts artificially beyond what the treatment was supposed to do? Then we can look at their hair counting before and after photos, and determine to the best of our ability if they had a number of catagen hairs. These are hairs that are typically the ones that regrow from finasteride, minoxidil, dutasteride, etc. They're the ones that are stuck in between hair cycles. Individuals with a ton of catagen hairs and very minimal detachment from the arrector pili muscle are the mega responders that we see to hair loss interventions, both natural and conventional. If you want to learn more, watch our video is it too late to reverse hair loss. Watch our video about why I think full hair regrowth will one day soon be possible, and then watch our video about natural versus pharmaceutical hair loss interventions, teasing out the mega responders from the expected response. And these are just a few of the things that we can begin to do to understand why somebody might have hyper responded. And maybe we find something about that individual that we actually would be able to back engineer and apply to all other individuals in the study to get them closer to that mega response. I think that's the level of creativity and insightfulness that I'd like to see, and it's accessible when individuals start to include these n equals one data points, and then begin to explore them. Now, let's take the opposite example. The person who responded terribly to finasteride. Say that they got worse hair loss when they started this medication. With these individuals, we can ask, did their blood levels and scalp levels of DHT even decline? Because there's pharmacogenetic data that shows that there can be a 60-fold difference in 5-alpha reductase inhibition capacity from finasteride alone, meaning that your average person will get around a 70% decline in DHT levels with 1 mg of finasteride, but there can be a 60-fold difference at the edge cases, meaning that people can get far more DHT declines, and some people will actually see DHT increases. Is that a reason why this individual is not responding? We can go through the same genetic sequencing exercise, and then build that repository, that database. We can then look at this individual's hair loss and ask the same questions as the best responder. Do they actually have androgenic alopecia? Were they misdiagnosed? Is there a concomitant scarring alopecia? And in doing this, build a picture as to who a non- responder to finasteride might be. They can then bring them closer to the average response. And we can't do any of this because these companies publishing these studies almost never include the individual n equals one data points. And I wish this weren't the case. I wish it were the case that when research companies report their data on hair counting metrics, they show the entire data set, not just the averages and sometimes the standard deviations. And when these research companies come across a mega responder or a terrible responder, I wish they'd take the Gary Nolan approach and say, "Why is that?"

[00:12:35]And then maybe make a career out of that. Perhaps make a discovery that helps unlock new avenues of hair growth for individuals who are all fighting hair loss, not just that one individual who got a mega or terrible response. And I often wonder at that point, "Well, why isn't this happening in hair loss research, and why is it happening in a field like exercise science?"

[00:12:54]And at least so far, the only explanation that I can come up with is that the field of exercise science, at least the field that's publishing regularly, really values truth seeking.

[00:13:04]So, their research is often funded by grants and labs, and want to know how all athletes can become the best versions of themselves. Now, contrast this with the field of hair loss research, which focuses on treatments.

[00:13:16]This is done primarily from companies that compete with one another. And it's often self-funded or privately funded.

[00:13:23]This is a totally different dynamic.

[00:13:25]Now, we've moved from truth seeking to building a narrative that the product actually works. And that's less so about reporting all the individual data points because that might not tell as compelling a story as you'd like, and more so toward obfuscating the data to make it look as nice as possible. For more information on this, watch our video about topical clascoterone and how that research group obfuscated their clinical data to make it sound like their product was more impressive than it actually was. Watch our videos on PP405 where that research team did something similar. They excluded total changes to hair counts in their press releases and instead only reported the percent improvements for subgroups. And then in a later presentation, they used metrics like the reactivation of inactive hair follicles, and they confused upright regrowing hairs as vellus hairs, and then said that those transition to terminal hairs. And they invented all these other metrics that you never really see in other hair loss research papers, not because they aren't interesting, but because it looks like this is a giant exercise in statistics hunting, and it drives me crazy. I feel like the truth-seeking here doesn't exist in the same way that it exists in exercise science. So, my question to all of these research companies that publish this kind of content, that deliberately withhold key information, is are we truth-seeking or are we building a marketing narrative? That's what I want to know. Because when I see practices like this, I can't help but think this looks more like statistics hunting. This looks more like building a marketing narrative to get a major funding round from VCs like Google. And it bothers me because the data could very well be absolutely fascinating, and there could be n equals one insights that we want to derive from them, and they could be buried in there, and we might never actually see it. And that could bury with it a huge discovery for hair loss treatments. And it doesn't need to happen. You know, when we published our scalp massage survey study, some people ripped it to shreds. They said that this is a low-quality study. This is survey-based, so it's subjective. It doesn't have a placebo group, and it has confounders, such as the use of other interventions, which we teased out to the best of our ability statistically, but which still exists to some extent in the data. You can't fully rule out all confounders. And guess what? I was fine with that. That's why we published it.

[00:15:38]That's why we wanted the data out there.

[00:15:40]It's a free intervention, so you can give it a go if you'd like. The directions are there, and we showed all of the individual data points. Because that's what truth-seeking is supposed to look like, at least in my eyes. And I would ask other research groups to maybe do the same. There are some fascinating n = 1 case studies out there, which I think deserve further exploration. One of them was the story of a man who was totally bald, 78 years old, fell asleep in his rocking chair, slipped backwards, hit his head, burned his scalp on hot coals, and then accidentally, over the next 6 months, regrew all of his hair.

[00:16:13]Why are we not exploring that case study and asking more questions? Was this individual on any concomitant medications when the accident happened and during his recovery window? For example, was he taking a DHT reducer like finasteride or dutasteride to help lower his prostate size? Was he taking an immunosuppressant that might have been popular at the time, like cyclosporine A, which has been shown to help improve hair parameters, but only in subsets of individuals with androgenic alopecia, and which was prescribed to people like him? Because we do know that there's crosstalk between the immune system and DHT reduction, and I think that that pathway combined is potentially one which will unlock new levels of regrowth. And this case study could be the answer to all of that, and yet we don't have answers to any of those questions because these researchers, despite publishing an incredibly fascinating study, which I give them credit for, also didn't happen to include any of the information in that study to help us answer these questions. And that's where that curiosity starts to kick in, and that's where there could be a breakthrough.

[00:17:09]Keep in mind that the two FDA-approved medications for androgenic alopecia today, minoxidil and finasteride, were partially accidental discoveries.

[00:17:19]Minoxidil was originally a blood pressure-lowering medication. A consequence of taking it orally, people got more hair growth everywhere, including the scalp. Finasteride was originally used for benign prostatic hyperplasia. A consequence of using it, finasteride helped to regrow hair in the scalp, which paired really well with data on individuals who had a type 2 5-alpha reductase genetic deficiency and who never went bald later in life.

[00:17:43]That's the same enzyme that finasteride inhibits. It's my belief that this level of creativity could help to solve some of the bigger challenges with respect to hair growth today, like reattaching the arrector pili muscle to hairs that are fully vellus and basically matured in their balding process. That, once solved, will be an absolute major breakthrough. I wonder if wounding plus immunosuppression plus DHT suppression is the combination that gets us there. I don't know, but maybe one day we will, and it might start with people showing their n = 1 data across these types of studies.

[00:18:18]That's everything. Thank you so much for watching. And if you are in the US and you're fighting hair loss, we co-founded the telehealth brand ULO, that's ulo.co, to help facilitate your access to best-in-class hair loss treatments. We offer a level of personalization that is unrivaled by any other telehealth brand out there. We offer low-dose formulations of both topical finasteride and topical dutasteride. We offer high-strength minoxidil with add-ons like tretinoin or retinoic acid. We offer full-strength finasteride, dutasteride, minoxidil in oral and topical formulations and a whole host of other things designed specifically to get you the best possible outcomes and prioritize safety and your tolerability.

[00:18:57]Check out the brand if you can. In the meantime, thank you so much for watching. We'll make more content like this soon. Take care and have a good day.