Distal radial fractures are the most common fracture that we see in clinics in this country. Across hospitals in England, there is a nine-fold variation in surgery or treatments offered to patients with distal radius fractures. This suggests that the fracture displacement threshold at which we offer treatment in the UK reflects clinical preferences rather than an absolute agreed threshold.

In this interview, Andrew Duckworth is joined by Mr Nick Johnson and Professor Joe Dias from the academic team of musculoskeletal surgery in Leicester to discuss their study, entitled ‘Prospective investigation of the relationship between dorsal tilt, carpal malalignment, and capitate shift in distal radial fractures’, which was published in the January 2020 edition of The Bone & Joint Journal. The aim of their study was to investigate the magnitude of dorsal tilt that leads to carpal malalignment, whether reduction of that dorsal tilt corrects the malalignment and which measure of malalignment is the most useful.

Can you give a brief introduction to the paper and the large international Delphi study you discussed that looked at the important radiological parameters for distal radius fractures?

In January of last year, we published the international Delphi consensus study, which captured 614 years of experience of 43 senior clinicians, both within the UK and internationally, trying to investigate whether there was a consensus displacement threshold at which we would offer intervention for distal radial fractures. This Delphi consensus established that 3 mm of shortening or a dorsal tilt of 10° is the point at which most offered treatment. But three out of four of our experts also pointed out that although we were focusing in the Delphi on the thresholds of displacement, carpal malalignment may be an important additional factor which they take into consideration.

What do we already know about carpal malalignment and outcome for our patients?

It was initially noted just over 100 years ago, in a paper from 1919 by Jean Mouchet; he suggested that it caused wrist pain and instability. Following that, there's little else in the literature about it apart from a few papers in the 1980s, including one from Leicester and Professor Dias here, that reported that carpal malalignment was related to dorsal tilt after a distal radius fracture. There was also a series from Tolesnik and Watson, who found that dorsal displacement of the longitudinal axis of the capitate in relation to the axis of the radius led to pain and instability. The interesting thing was that they did corrective osteotomies for this and found that the symptoms improved. It was only a small series of about 13 patients, but nevertheless, they had noted this. There was also a Margaret McQueen RCT looking at four different treatment methods for distal radius fracture; they found that carpal malalignment was a major predictor of function following the distal radius fracture.

What were the aim and hypothesis of your study and how did you hope to address these limitations of the current literature with your work?

The previous papers had observed or suggested an association between carpal malalignment and dorsal tilt and had suggested this as an explanation for persistent pain or weakness, and hence gave that as an indication for a corrective osteotomy. However, we did not know the strength of the association. We do not know how best to evaluate it and whether the alignment or the change in alignment was restored after the tilt was corrected when we reduce the fracture.

We also don't know what the normal ranges of capitate shift were either without a fracture or within an acceptably displaced fracture. That was what led us to raise these set of questions.

How was the study performed and put together?

It was a prospective study. It had 250 consecutive patients who had 252 fractures. They were aged 16+ years, and they presented with a distal radius fracture to our service over roughly a four-month period. Most were dorsally displaced although this wasn't exclusive; some of them were button fractures or Smith's fractures (we wanted a range of displacements on which to check carpal malalignment). We did not exclude patients as this was an observational study. It is based on radiographs taken for routine care so, therefore, there were no exclusion criteria. The whole set was consecutive.

How were the radiographic measurements collected?

Firstly, we got the 576 sets of posterior, anterior and lateral radiographs of the injured wrist, taken before and after treatment, and exported these into an open access software called OsiriX. For this particular study, we assessed four attributes. The first was fracture factors, for example, intraarticular extension comminution variance, not the primary focus of our assessment, but they could be core factors that changed our observations. The second thing we measured was the tilt of the distal radius. We did all the usual measurements for fracture assessment itself, the principle one being dorsal tilt. The third was to assess carpal malalignment, which we assessed on lateral radiographs by measuring the tilt of the lunate on the radius in degrees or the capitate on the lunate in degrees, which are well described in the literature. We then measured the capitate shift; this has not been clearly described before. We identified the centre of the head of the capitate by putting a circle on the capitate head and marking the centre. We defined the axis of the radius by using an axis tube within the cervix which is a reproducible tool to find out the axis of a bone, and then the perpendicular distance of the centre of the capitate from the axis gave us the shift of the capitate from the axis of the radius. This was negative if it was dorsal to the axis.

The fourth element that we assessed was signs of other explanations for carpal malalignment, principally ligament injury. We assessed this by measuring gaps between the scaphoid and lunate or between the lunate and triquetrum, noting breaks and Gilula lines.

The statistical analysis performed in the study is very nicely laid out, but could you give us a simple, concise overview of what was done?

Initially we did some scatter plots, which showed a clear linear relationship between the different measures of carpal alignment and dorsal tilt. We chose to use linear regression to assess the relationship, and for each measurement of carpal alignment we created individual linear regression models, to assess the relationship with dorsal tilt. We did that at three timepoints; the initial radiograph (first x-ray after injury), if they'd had a manipulation we did it again following reduction and also then the last available radiograph. Then we also did further regression models where we added in different variables, including age, gender and risk position to see how these affected the results.

Separately, as you've heard, we also wanted to find out what the actual normal value of capitate shift was because no one knows that. It would be unethical to take radiographs of 50 normal wrists but we had a series of patients who'd had a wrist x-ray for a possible injury, but were reported as completely normal. We used a random sample of 50 of these patients and measured the capitate shift on those to give us a normal sample.

We also wanted to know what capitate shift was likely to be within a population of patients who had a distal radius fracture that had been determined to be within an acceptable position. We defined that as a dorsal tilt of 0° to 11°. This was then used to determine the cut point.

Let’s move onto the results of the study.

There were 252 fractures, the mean age was 58 years and just over 70% were female, which is very in keeping with the normal epidemiological data for these injuries. Three-quarters were displaced fractures and just over a quarter were intra-articular. In terms of management, almost 80% were managed nonoperatively; in terms of the quality of the gradient, the vast majority were good, with about 92% with a small spattering of fair and only two poor cases. The right and left hands were equally represented as well.

What do you feel that the key findings of your study in terms of the results were in relation to the primary analysis performed to address your aims?

The first important thing was that we were unable to identify significant ligament injury which could explain carpal malalignment. Having excluded that as a cause of carpal malalignment our key observations were that the capitate shifts dorsally as the distal radius tilts after the fracture and this shift is reversed by reducing the fracture.

The mean capitate shift in our population was 3 mm dorsal to the axis of the distal radius. Although we measured radial lunated and capital lunated in addition to the capitate shift to assess carpal malalignment, we found a weaker association of the angle measurements with the tilt than we found with capitate shift. The mean tilt of the lunate in our population was 11° tilting back and the capitate seems to compensate for this by moving forward 7°.

This raised the question of what capitate shift we may accept, and we adjusted this, as pointed out, by defining a probable normal range of capitate shift in 50 wrists. We found that the normal capitate shift was 2 mm in front of the axis of the radius and the range was from 2 mm behind the axis to 7 mm in the front. The mean was 4 mm in front of the axis. The centre of the capitate seems to line up with the palmar cortex of the distal radius, which is an easy visual clue without doing measurements.

It is usual in clinical practice to be unconcerned if the dorsal tilt of the distal radius is at least 0 - we start getting more and more concerned when it tips back. We looked at the capitate shift in a separate population of distal radius fractures, which were 0 or palmar tilted and, therefore, the normal acceptable range. For these the mean location of the capitate centre was 2 mm in front of the axis of the radius and the dorsal limit; the 95% confidence limit was 6 mm behind the axis, and that is the one that we used as a cut point.

With regards to the regression analysis and the ROC analysis, what did you find for those?

The regression analysis showed that all measures of carpal alignment were associated with dorsal tilt up to each timepoint that we measured them. It was very clear that capitate shift had the strongest relationship with dorsal tilt. On the initial x-rays the R-squared value for capitate shift and regression calculation was 0.8 compared to 0.27 for the capitate lunate angle and 0.3 for the radial lunate angle, and there were similar values throughout. On the final x-ray the capitate shift was 0.65. Whereas capitate lunate was 0.09 and radial lunate 0.18; this can be clearly seen where the spread around the regression line is quite tight for capitate shift. In the additional regression where we added in the other variables, capitate shift was the only parameter not affected by age or risk position.

Our ROC analysis showed that the cut point of dorsal tilt to maintain the capitate shift within the 99th percentile following a distal radius fracture was 9° of dorsal tilt. This is interesting because it corresponds with what our Delphi palmar said; most people would accept 9° of dorsal tilt.

If we can move onto the implications of the work; what do you feel are the key findings of the work considering any potential limitations of the data?

We've confirmed that previous impression that dorsal tilt is related to carpal malalignment. People have suggested it before, but this clearly demonstrates that. We've also shown that if you correct the tilt by reducing the fracture well then you will improve the alignment.

In terms of the measures of carpal malalignment we've shown that capitate shift is the best measure. It's got the strongest association at all timepoints, and it's not related to the position of the wrist or the other variables such as gender or age. I think the paper is also useful because we've defined how to measure capitate shift. It's easy to do visually and it will aid people clinically.

What do you feel the implications are moving forward in terms of the measurements, but also for future studies in the distal radius?

I think about focus; the last hundred odd years of research have been on the alignment of the distal radius. This study will hopefully make doctors aware of how the risk changes after the fracture of the distal radius. We need to understand that when the radius tilts the carpal bones buckle, so a very subtle mechanism of stabilising the wrist changes by changing the tilt of the distal radius. A capitate shifts dorsally and low transmission across the wrist shifts backwards. We have shown how this can be easily assessed without needing to do any measurements. Doctors can just look at the image and say whether the centre of the capitate lines up in front of the axis of the radius or at least is aligned with the palmar cortex of the radius on a lateral view.

The aim of reducing the fracture should be to get the capitate aligned with the radius rather than to correct the tilt of the radius. You're now suddenly thinking differently about the distal radius. When you assess whether it is stable or not, you are actually saying you've achieved restoration or the best situation to transmit low across the broken wrist.

The study also establishes that the Delphi consensus of an acceptable tilt of 10° dorsally would keep the capitate shift within acceptable limits. I think all those are benefits and hopefully the outcome of this will be useful to clinicians treating these common injuries.

If you’d like to read the full paper you can do so here. You can listen to the podcast version of this interview here.