Torquing the tibia

A small group of us are currently trying to tidy up certain aspects of the Conventional Gait Model (Newington, Helen Hayes, Davis, Kadaba, PiG, VCM) in such a way as to address a number of known problems but also preserves its strengths. We’ve had quite a bit of discussion about tibial torsion recently which flags a number of different issues. Looking back at my book, I see I spent quite a bit of space discussing how different degrees of femoral anteversion affect gait data but not quite so long on tibial torsion.

Tibial torsion is the twist along the shaft of the tibia. If you were to look along the long axis of the bone and imagine where the knee joint axis is in relation to the proximal tibia (in black in the picture below) and the ankle joint axis (trans-malleolar axis) in relation to the distal tibia (grey) then the angle between them is the tibial torsion (20° in this example).

Tibial torsion

At birth tibial torsion, thus defined, is small (<5° on average) and normally increases with age to about 15-20° at adolescence after which it remains constant. (Note that this is different to femoral anteversion which starts off large and reduces over time). There is, however, considerable variability between individuals.

The importance of tibial torsion clinically and to biomechanical modelling is fundamentally that it means that the distal tibia is pointing in a different direction to the proximal tibia. In other words your knee points in a different direction to your ankle. This is particularly important for understanding gait problems in the transverse plane – if for example you want to know why someone is walking with internal foot progression (pigeon-toed in old language).

PiG, the current Vicon implementation of the Conventional Gait Model, deals with the issue by defining both a proximal and distal tibial segment differing only in a rotation about the long axis of the tibia. The proximal tibial segment is used for most measures of knee kinematics and kinetics and the distal tibial segment is used for the ankle kinematics and kinetics (and knee kinematics when generating outputs from the static trial, for some reason!).

At one level this is quite logical but it has several disadvantages:

  • There are two segments but only one bone!
  • The way tibial torsion is incorporated in the model is quite different to the way femoral anteversion is incorporated and this leads to confusion about both. (This is a particular issue as one of the principal advantages of the CGM is that it is inherently quite easy to understand).
  • Tibial torsion becomes an implicit feature of the gait graphs rather than an explicit feature. Thus if you want to consider what factors are affecting transverse plane alignment  from the gait graphs you cannot do so without also knowing the value of tibial torsion that has been used. (This is particularly important if the value of tibial torsion has been calculated with the use of a medial malleolar marker placed for the static trial but not reported along with the gait data). There are a number of ways around this but it would be better if the information was explicit in the graphs.

I much prefer the way we do things when using Visual3D which is to use only the distal tibia. Knee rotation (transverse plane) is then defined pretty much as in the diagram above except the knee axis is taken as the transepicondylar axis in the femur. The measurement is thus the combination of tibial torsion and any true rotation in the knee joint (subject to soft-tissue artefact of course, but that’s another story). What makes things even nicer is that we plot joint angles from the static trial along with our consistency plots (see graph below).

Knee rotation.png

Thus the data in black is knee rotation from 5 walking trials showing a range  between about 12° and 22° internal rotation which indicates that the ankle joint axis (in the tibia) is internally rotated with respect to the the knee joint axis (in the femur). The solid red line is knee rotation measured during the static trial which is by definition equal to the torsion that the system measures in the tibia. It is clear that the (very) abnormal knee rotation is almost entirely explained by torsion within the tibia (and you can then sit around and debate whether the remaining signal is real or a consequence of soft tissue artefact – yes rigid clusters are vulnerable to STA as skin markers it is just that the artefact is different!).

Another nice feature about this approach is that if you have measured tibial torsion clinically then you can compare that measurement with that which the system has made (the static trial measurement) and very easily think through the clinical consequences of any difference, by thinking how your interpretation might change if the solid red line were higher or lower.

Final paragraph for the advanced reader!

Just when we thought we’d considered all the issues and agreed a sensible way forwards someone mentioned kinetics. PiG expresses joint moments in the coordinate system of the distal segment by default and it really doesn’t make any sense to report the knee joint moment about a coordinate system defined by the ankle joint axis! Perhaps this is the reason that the two tibial reference frames were defined in the first place. A much simpler solution, however, is to express the joint moments in the coordinate system of the proximal segment (until now I’ve generally considered this an arbitrary choice but, as result of reflecting on this issue, I’ve now convinced myself that if you are going to define rotation of segments about their long axis by distal landmarks or functional axes and want to use an orthogonal axis system then you have no real choice but to use the proximal segment). The other solution which I actually prefer is to express the joint moments as projections of the total moment vector onto the axes of the joint coordinate system (as recommended by Anthony Schache and myself in this paper). At first glance reporting “components” of a vector about a non-orthogonal axis system appears quite offensive to any self-respecting engineer but this is actually more appropriate if you want to interpret those moments clinically in the context of the requirement of different muscle groups to exert moments about the axes that we regard movement as occurring.



  1. Thank you Richard for this informative post.

    In our clinical meetings whenever there is an abnormal foot progression angle the surgeon will ask the question ‘where is the rotation coming from?’ and therefore I agree that the transverse plane knee rotation profile should include the combination of tibial torsion and any true rotation in the tibiofemoral joint. This way the gait analyst can infer the relative contributions from each joint (knee, ankle) and/or segment (pelvis, hindfoot, forefoot). In my opinion the only time we should inform the model with bony long axis torsion is for application with musculoskeletal modelling whereby such an adjustment becomes crucial for appropriate muscle-tendon pathways. I’m sure you will agree with my final point that it can be dangerous to inform any model with ‘clinical’ measures of bony torsion and that this should only be considered when such measures can be obtained from medical imaging.

  2. thanks for that. Have started graphing the static rotation onto gait consistency plots as suggested and for something so ridiculously simple (and using data we have anyway!) it has been very informative. I have also done the same for femoral rotation and that has also been very useful and helps try and work out what is a fixed bony excessive internal rotation at the hip versus those internally rotating dynamically for what ever reason e.g. to compensate for hip abductors

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s