So this post is inspired by a blog I recently read on the Run3D website (see here) which was entitled “You aren’t still using 2D gait analysis… are you?” I’d recommend giving it a full read before continuing here so that you can see the context of my response.
The opening sentence reads like a tabloid headline: “2D gait analysis just isn’t good enough anymore” This is the theme of their blog entry and certainly worthy of further discussion. From their perspective 2D gait analysis is apparently “ancient technology”. Is this factually accurate, or possibly just marketing hyperbole from a company with a financial interest in all clinicians moving toward using (and paying for) 3D analysis?
The argument made throughout the piece is that 3D analysis is superior to 2D; it is more accurate, it picks up all the finite rotation and twisting movements happening as you move, and it can compare you to a normative database.
Apparently “research has shown that ONLY three dimensional gait analysis assessments are accurate enough to detect the small movement patterns that might cause an injury”. Sadly, they do not link to any of the research they refer to, chiefly as I suspect it does not exist (happy to be proven wrong on this). Here’s the uncomfortable truth – regardless of the accuracy of 3D analysis, kinematic patterns are poor predictors of injury. End of story. What benefit is ‘accuracy’ when this is the case?
The suggestion is then made that their 3D system will allow you to identify the CAUSE of what is taking place (with the inference that 2D will not allow this). A powerful statement indeed. Given the lack of robust predictive models for almost every lower extremity pathology I’m just not sure how anyone could be so confident about this. There is no denying that more data is generated with a 3D analysis – the question here is what that information truly informs? Does it really “find the source of the problem faster than 2D”? Given most of the runners attending for analyses are already injured, are we not concerned that any data we are seeing may be as a result of pain/injury, rather than the pain/injury being a result of the data we are seeing? The old correlation-causation issue rearing its ugly head as always.
How about the benefits of being compared to their normative uninjured database? This sounds like something good to do, right? Well it is key to understand that the use of the word normal here is in the statistical/distribution sense, and NOT a narrative on ideal/optimal. It is an average (mean) for each measure taken from a group of uninjured runners (I believe around 800 individuals from what I’ve heard). Remember that an average is made from just as many values above it as it is below it. Ironically there may be a ‘normal’ value for a kinematic measure provided within their database that not a single individual in that database actually exhibited themselves. Given what we know about human variation (which is high and this IS normal), the poor correlation between kinematic measures and injury, and that kinematic measures alone do not give any information about internal loads/stresses within human tissue it should become clear this is far more complex than a simple ‘compare the runners’ scenario. I’d wager there are highly functioning and pain free individuals that when compared to the normative database are significantly different to it. I’d also bet there are plenty of people who have identical measures to the normative database and yet cannot complete a 5km run without pain. Thus, employing the kinematic averages of less than a thousand pain free people and then applying this to an injured runner sitting in front of you is of no use here. No matter how cool and marketable it sounds.
I’m sure I’ve said this before, but a gait analysis does not start and end with the pressing of a button on a camera and treadmill. We cannot bring something as complex as the human body experiencing increased sensitivity in the lower limb down to a page of graphs and angles. We need to be mindful of the limitations of a gait analysis (patients usually already injured, treadmill Vs overground differences, the Hawthorne effects, the observer-expectancy effect, the subject-expectancy effect) and understand that they apply EQUALLY to both 2D and 3D approaches to assessment. We need to practice within a biopsychosocial model, be aware that pain is multidimensional and most importantly be aware that the language (and technology) we use has the potential to lead people toward pain rather than away from it. Interestingly, there could be an argument put forward that a 3D approach carries a greater risk of this.
To be clear, I’m not anti-3D analysis (and certainly not anti-Run3D in particular). I’m anti-BS. I just don’t feel that the claims and promises made in the Run3D blog linked above are currently scientifically supported. It reads rather more like an advert to me, which I guess it is. Particularly when their bold headline and opening sentence is in contradiction of published research which is found with the easiest of online literature searches.
I’m sure this post may gather some responses from people using 3D set ups. Let me try and nip this in the bud: there is no need to be defensive or confrontational about this post. I don’t have a problem with people using them. I’d happily use one myself if someone paid for it for me. And I am not saying it has no value at all. I am simply asking everyone (those who use it, those who don’t, and the runners googling about it) to be aware of and to consider all of my above comments rather than the media sizzle and tabloid headline style marketing in isolation.
To conclude:
Outside of a research setting and in the clinical world, do we need 3D analysis to help us assess our runners and formulate management plans for them? No.
Is the way 2D analysis has been framed as unequivocally ancient/old/inferior/inaccurate appropriate? No.
