Circumferential Periosteal Release to Treat Paediatric Leg Length Discrepancy: Medium Term Outcomes
2Institute of Science and Technology in Medicine, The Robert Jones and Agnes Hunt Orthopaedic Hospital NHS Foundation Trust, Oswestry, Shropshire; Institute of Science and Technology in Medicine, Keele University, Keele, Staffordshire, United Kingdom
Corresponding Author: Benjamin Dougal Chatterton, Paediatric Orthopaedic Unit, The Robert Jones and Agnes Hunt Orthopaedic Hospital NHS Foundation Trust, Oswestry, Shropshire, United Kingdom, Phone: +447891852386, e-mail: firstname.lastname@example.org
How to cite this article: Chatterton BD, Kuiper JH, Williams DP. Circumferential Periosteal Release to Treat Paediatric Leg Length Discrepancy: Medium Term Outcomes. Strategies Trauma Limb Reconstr 2023;18(2):67–72.
Source of support: Nil
Conflict of interest: None
Received on: 21 July 2022; Accepted on: 01 May 2023; Published on: 21 October 2023
Aims: This study aims to report the medium term outcomes of circumferential periosteal release of the distal femur and distal tibia in treating paediatric leg length discrepancy (LLD).
Materials and methods: A retrospective case series was performed on all patients undergoing circumferential periosteal release of the distal femur and/or tibia between 2006 and 2019. Data collected included demographics, surgical indications, post-operative leg lengths, and complications. Leg length discrepancy was calculated as actual values and percentages of the longest limb length. Final actual and percentage discrepancies were compared to initial discrepancies using a paired t-test. Patterns of discrepancy over time were analysed using linear mixed models.
Results: Eighteen patients (11 males) were identified, who underwent 25 procedures. The mean age at first surgery was 5.8 (range, 2–13). The commonest indication was congenital limb deficiency (7 patients). Five patients underwent repeat periosteal release, and one patient had three releases. The mean follow-up was 63 months [standard deviation (SD), 33.9]. Fifteen patients had sufficient data for statistical analysis.
The mean actual discrepancy decreased from 2.07 cm (SD, 1.07) to 1.12 cm (SD, 1.62), and the mean relative discrepancy from 4.3% (SD, 2.8) to 1.5% (SD, 2.4). Significant mean reductions were seen in both actual discrepancies [0.61 cm (95% CI: 0.05–1.16; p = 0.034)], and percentage discrepancy [2.10% (95% CI: 1.0–3.1, p = <0.001]). In five patients, the operated limb overgrew the contralateral limb. Patients whose operated limb overgrew still had a reduction in LLD, with a mean residual discrepancy less than 1 cm (mean 0.7 cm, 95% CI: From −0.9 to 2.4).
Conclusion: Circumferential periosteal release produces a significant decrease in both actual and percentage LLD. We believe this procedure is best indicated in younger patients with congenital LLD in whom the discrepancy is predicted to increase as they age.
Clinical significance: Circumferential periosteal release produces a significant decrease in LLD. This procedure can be used to manage symptoms during growth, particularly at the point where orthotic usage may become problematic, and to potentially reduce the magnitude of surgery needed at an older age.
Keywords: Case series, Leg-length discrepancy, Lengthening, Paediatric, Periosteum.
Leg length discrepancy (LLD) in children may occur due to congenital or acquired reasons and remains a common presentation to the paediatric orthopaedic surgeon. The treatment of this problem is however complex, in part because it is unclear what degree of discrepancy may lead to long-term problems.1 Broadly, a short limb can be lengthened, or a long limb can be shortened. Treatment options for lengthening a short limb include distraction osteogenesis with an external fixator, or more recently, intramedullary lengthening devices.2 Whilst successful outcomes have been reported for these techniques, both are associated with unique and predictable complications.3,4 Although less preferable, the longer limb may be shortened acutely, or by manipulation of the physis with epiphysiodesis.
Circumferential periosteal release is a rarely reported technique of paediatric leg lengthening that avoids many of the complications associated with the use of frames or intramedullary devices and has the advantage of correcting the short leg rather than restricting the longer leg. The technique was first described in detail in 1987 by Wilde and Baker, with longitudinal stripping having previously been reported clinically in 1975 as a treatment for patients with limb length discrepancies following poliomyelitis.5,6 The procedure involves the circumferential excision of a strip of periosteum from the metaphysis of the distal femur, tibia and fibula. Wilde and Baker reported favourable outcomes in their original series of 38 patients, but since this report, there have been few studies reporting the results of this specific technique.
The primary aim of this study was to determine whether isolated periosteal release in the context of a LLD led to stimulation of growth and a reduction in the discrepancy. The secondary aims were to determine the pattern of discrepancy over time following the procedure and to find out if any reduction of discrepancy was associated with age or degree of initial discrepancy.
MATERIALS AND METHODS
A single-centre retrospective case series was performed for patients less than 18 years of age who underwent circumferential periosteal release of the lower limb between 2006 and 2019. Although patients had undergone this procedure prior to 2006, this was the earliest date that imaging was still available in either physical or digital format at our institution to allow formal assessment for this study.
Patients were identified via the search function in our electronic patient record system. Only patients who had a periosteal release in the lower limb were included. Patients who underwent only partial periosteal release for angular correction were excluded. Patients who underwent additional lower limb procedures at the same time as the periosteal release which could potentially affect bone growth were also excluded (for example, a contralateral epiphysiodesis). Data collected included demographics, indication for surgery, pre- and postoperative leg length measurements where available, and details of any complications.
This procedure was performed by three different surgeons at our institution, utilising the same surgical technique, broadly as described by Wilde and Baker in their original report.5 The operation is performed with a high thigh tourniquet and the patients’ lower limb is prepped and draped in the standard fashion. The physes of the distal femur and tibia are identified using the image intensifier and marked on the patient. For the distal femur, two incisions are made overlying the midline of the femur laterally and medially. A subvastus approach is taken to expose the metaphysis. In the tibia and fibula, distal release is preferred due to the improved access and elevation of surrounding soft tissue, with improved visualisation of the bones, minimising the risk of neurovascular injury. An incision is made distally just behind the posteromedial border of the tibia (Fig. 1). This allows extraperiosteal exposure of the tibia posteriorly and visualisation of the deep surface of the fibula. The anterior tibia is then exposed up to the anterior compartment muscles, after which the anterior compartment fascia is released to start blunt extraperiosteal exposure of the remaining tibia. A further incision over the fibula allows completion of the fibula exposure, and gentle elevation of the anterior compartment muscles to complete visualisation of the anterolateral tibia.
Following extraperiosteal exposure of the bone, a 0.5 cm strip of periosteum can then be excised from the metaphysis in a circumferential fashion (Fig. 2A). Care must be taken to ensure complete circumferential excision, or an angular deformity may occur. Confirmation of complete circumferential excision is indicated by the symmetrical retraction of the free edges of the cut periosteum (Fig. 2B), and excision of an intact periosteal strip (Fig. 2C). The wounds are washed and closed in layers, with absorbable sutures to the skin. The patient is allowed to weight bear as tolerated postoperatively.
Measurement of Leg Length
Variations in surgeon preferences meant that pre- and postoperative imaging protocols differed between patients. Patients’ leg lengths were assessed by either a computed tomography (CT) scanogram or long-leg plain radiographs where available. The total leg length was measured from the centre of the femoral head to the centre of the tibial plafond. The femoral length was measured from the centre of the femoral head to the lowest point of the lateral femoral condyle at the knee joint. The tibial length was measured from the lowest point of the lateral femoral condyle at the knee joint to the centre of the tibial plafond. In one patient who had bilateral avascular necrosis the femoral head could not be used as a fixed reference point, and therefore the greater trochanter was used. Each scan was measured by both clinical authors (BDC and DPW), and the means of the two measurements were used. Leg length discrepancy was then expressed in actual values and as relative (percentage) discrepancy, by dividing the total discrepancy, femoral discrepancy and tibial discrepancy by the longer leg length, as described by Wilde and Baker.5
The normality of the discrepancy measurements was assessed using quantile–quantile (QQ) plots. The reliability of the LLD measurement was assessed by its intraclass correlation coefficient (ICC). We determined ICC(1,A), assessing agreement assuming measurements from a single assessor was used, and ICC(k,A), assessing agreement assuming the mean value across the two assessors was used.7 The standard error of measurement (SEM) was calculated from the standard deviation (SD) of all discrepancy measurements and the ICC as . The minimum difference (MD) needed between two measurements to be considered “real” differences beyond random fluctuations was calculated as .8
The mean change in discrepancy (actual and percentage) was compared to zero (no change) using paired t-tests. The pattern of discrepancy over time was analysed using linear mixed models, with a patient-dependent (random) intercept. Multiple lengthening procedures were modelled as nested within the patient. Time was used as a fixed effect, and modelled as a fractional polynomial.9
The full set of fractional polynomials up to degree 2 [i.e., linear, squared, square root and log, inverse square root, inverse linear, and inverse square—each based on time (in months) + 1] was screened to find the one that best fitted the data. In these analyses, we classified the patient’s procedure depending on whether the discrepancy had increased, decreased or overcompensated, and added this classification as a fixed effect.
The best-fitting fractional polynomial was selected on the basis of Akaike’s Information Criterion (AIC), and the significance of fixed terms was assessed using likelihood ratio tests. The association of length gain with age and initial discrepancy was investigated by first determining for each child the 24-month discrepancy after their first procedure using linear interpolation, or if needed using the last observation carried forward. The length gain was defined as the difference between the initial discrepancy and the 24-month discrepancy, and a linear correlation analysis of gain vs age and gain vs initial discrepancy was performed. All statistical analyses were done using R vs 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria) using the packages irr and nlme. For all analyses, a two-sided p-value below 0.05 was assumed to denote statistical significance.
Thirty-nine patients were initially identified. Of these, 21 patients did not meet the inclusion criteria and were therefore excluded. This resulted in a study population of 18 patients, who underwent 25 procedures. Fifteen patients had adequate data for statistical analysis. In twenty-three cases both the distal femur and distal tibia periosteum were released. In only two was one segment of the lower limb operated on. Five patients had a repeat periosteal release (mean time to next procedure 5.3 years; range, 2.6–8.1), after the first procedure, and one patient had three periosteal releases, to stimulate further correction of residual length discrepancy. No major complications were noted in any of the patients. No angular deformities were seen after the procedure. One patient had a minor wound infection in their medial distal femoral scar.
The mean age at the first procedure was 5.8 years (range, 2–13). The mean follow-up was 63.1 months (SD, 33.9). The indications for surgery were congenital limb length discrepancy (7, 39%), fibular hemimelia (5, 28%), previous trauma or sepsis (4, 22%), skeletal dysplasia (1, 6%), and neuromuscular (1, 6%).
Normality and Reliability of Discrepancy Measures
The absolute (raw) values of the discrepancies were normally distributed, but their relative values were not. However, the differences between the initial and final assessments were normally distributed, and therefore the paired t-test could be used. The reliability of the discrepancy measures for a single assessor was 0.97 [95% confidence interval (CI): 0.94–0.98] and for the mean of multiple assessors was 0.98 (95% CI: 0.97–0.99), suggesting almost perfect agreement. If a single rater did the measurements, the SEM and MD were 2.6 and 7.2 mm, respectively. When using the mean of two raters, the SEM and MD were 1.8 and 5.1 mm.
Actual and Relative Leg Length Discrepancies
The mean actual discrepancy decreased from 2.07 to 1.12 cm, and the mean relative discrepancy from 4.3 to 1.5% (Table 1). However, in six cases the procedure ended with an over-correction (negative discrepancy; Fig. 3), and we therefore also calculated the means of the absolute final discrepancy values (i.e., their values disregarding the sign), which were 1.56 cm, or 2.2%. A significant mean reduction in actual and percentage LLD was seen (Table 1).
|Actual (SD; range or 95% CI)||Relative (SD; range or 95% CI)|
|Mean initial discrepancy||2.07 cm (1.07; from 0.5 to 4.2)||4.3% (2.8; from 1.1 to 11.4)|
|Mean final discrepancy||1.12 cm (1.62; from−1.2 to 3.9)||1.5% (2.4; from−1.8 to 6.8)|
|Mean final absolute discrepancy value||1.56 cm (1.16; from 0.3 to 3.9)||2.2% (1.8; from 0.4 to 6.8)|
|Mean reduction absolute discrepancy value||0.61 cm (95% CI, 0.05–1.16)||2.1% (95% CI, 1.0–3.1)|
|p-value mean reduction||0.034||<0.001|
Patterns of LLD over Time following the Procedure
The best-fitting fractional polynomial was the sum of the log and the square root function and included the interaction term of these two functions with the procedure classification (decreased/increased/overcompensated). Using this combination, we produced plots showing the pattern of actual LLD over time for individual patients (Fig. 3).
Besides the six patients whose operated leg overgrew the contralateral leg but still ended with a lower overall discrepancy, one patient (4) had a good initial response, and then the discrepancy began to increase, before a further positive response after repeat surgery. One patient (18) had a minimal response to the surgery.
Plots comparing the trend in mean actual LLD between those whose discrepancy decreased and those in whom the operated shorter leg eventually overgrew the other leg can be seen in Figure 4. In both groups, the change in discrepancy was initially rapid but steadied after 24 months. The mean gain in length (change in discrepancy) at 60 months was 0.91 cm (95% CI: 0.6–1.2, p < 0.001) for those with a decreased discrepancy and 1.7 cm (95% CI: 1.3–2.1, p < 0.001) for those who overgrew. Patients whose operated leg overgrew still had a reduction in LLD (mean 0.25 cm, 95% CI: From −3.0 to 3.5), and their mean residual limb length discrepancy was less than 1 cm (mean, 0.7 cm; 95% CI: From −0.9 to 2.4).
No evidence for an association between age and 24-month length gain was found, neither when analysed as actual length gained nor when analysed as relative length gained (Fig. 5). The same applied to an association between actual or relative gain with initial discrepancy (r = −0.28, 95% CI: From −0.70 to 0.34, p = 0.34 and r = −0.30, 95% CI: From −0.72 to 0.28, p = 0.30).
Management of congenital LLD in childhood can be challenging. Large discrepancies in early years, when alternative surgical interventions may not be suitable, may require significant orthotic modifications to compensate. Contralateral epiphysiodesis—while the child grows—may be an option in some cases to limit growth on the longer side. However, this has the disadvantage of intervening on the normal limb, reducing overall height, and in cases of significant shortening, may not achieve full correction of the discrepancy alone. Where large final discrepancies are anticipated, lengthening procedures may be required, but large corrections carry additional risks.
Surgical intervention in the early years that may help lengthen the shorter leg, potentially reducing the amount of contralateral shortening required or the extent of future lengthening, would therefore seem appealing. The periosteal release is a seldom reported technique for this, and our results show a significant mean reduction in LLD for this procedure.
The mechanism by which this technique induces growth in the limb is still unclear. The vascular model proposes that sectioning of the periosteum induces a local hyperaemic response which stimulates bone growth,10 the same effect that is seen in the overgrowth arising after paediatric femoral shaft fractures.11 A mechanical model has also been proposed, in which the periosteum acts as a mechanical restraint to the physis.12 Animal models have also shown that periosteal sectioning induces changes in chondrocyte activity within the physis itself.13
Our findings are largely in keeping with previous studies. D’Souza and Shah report a series of patients with residual LLD following poliomyelitis, in whom a gain in length was seen for both the femur and tibia when the circumferential release was performed, with a mean total gain of 2.2 cm (95% CI: 0.05–4.3).14 The authors recommend the technique in the treatment of modest LLDs. The aforementioned study by Jenkins et al. also reported successful increases in leg length following the procedure in 17/28 (61%) of femora and 17/26 (65%) of tibiae, again in a series of patients with poliomyelitis, although longitudinal stripping was used in this study rather than circumferential.6 In a further study examining the response to this technique in a series of 10 patients with a single diagnosis (Achondroplasia), little response was seen in either the femur or tibia.15
Wilde and Baker were the first to present the results of this technique in patients with a wide range of aetiologies, the most common of which was fibular hemimelia.5 They noted a fall in actual and percentage LLD of 0.8 cm or 1.8%, similar to our findings. The response was more pronounced in younger patients but was not influenced by the patient’s diagnosis or initial discrepancy. In our data, we did not find this association of length gain with age. This may be related to our smaller sample size, which reduces the power to find such a relation. However, replicating the correlation of r = −0.62 found by Wilde and Baker would need a sample size of only 18 (assuming 80% power at the two-sided p = 0.05 level), the number of children in our dataset. A more likely explanation is that the correlation coefficient may be smaller than that reported by Wilde and Baker If it equalled their lower 95% confidence limit (r = −0.33) a study would need 69 patients. In addition, Wilde and Baker correlated age with relative length gain, which will introduce a spurious correlation because limb length is highly correlated with age.
Even greater increases in leg length have been reported with a modification of this technique.16 Limpaphayom and Prasongchin report a series of 11 patients with a mean LLD of 6 cm. After surgery, leg lengths equalised in eight of the 11 patients at a mean of 25 months. Those who didn’t equalise started with a LLD of over 10 cm. However, the reported technique requires much larger incisions, involving full-length exposure of the tibia and femur, with periosteal striping at three different levels in each segment.
Although we noted low complication rates, one issue seen in our results was the operated leg “over-shooting” the previously longer limb. This phenomenon was also reported in the study by Limpaphayom and Prasongchin16 although to a greater extent than in our results, with one patient experiencing an overgrowth of greater than 2 cm. In our results, no patient overgrew by more than 1 cm, which would be unlikely to cause a clinical issue. However, this does highlight one of the pitfalls of this technique, namely that the response to surgery can be unpredictable, which has previously been highlighted by Jenkins et al.6 As such, we would suggest that surgery should be deferred until a discrepancy of 2–3 cm is established in order to minimise the risk of “overshoot.” This amount of discrepancy often correlates with the point that orthotic management becomes slightly more troublesome in the younger child.
In addition to the overgrowth phenomenon, one patient showed an increase in discrepancy after an initial period of good response. This patient was treated for fibular hemimelia, and the discrepancy began to worsen around the age of 10, a period of potential extra growth. It should be noted that this patient then underwent a second release, with an improvement in discrepancy. A further patient had a minimal response to the surgery, and in fact had a slight increase in discrepancy, although within the SEM for the leg lengths. This patient underwent the procedure at an older age (13 years), and may therefore have had a reduced response to the surgery, as was suggested for older children in the aforementioned study by Wilde and Baker.5
The main limitations to this study are the small patient population and the inconsistency in follow-up due to differences in practice. However, given that this is a rarely performed procedure we feel the study sample is appropriate and is in keeping with the limited previous studies that have been published. Future prospective studies could be performed after these cases have been operated on although these would obviously require significant follow-up times.
CONCLUSION AND CLINICAL SIGNIFICANCE
This study has shown that circumferential periosteal release can be a useful tool in treating LLDs in children. It is, however, unpredictable and some patients with mild a discrepancy may “over-shoot” to a small extent. The ideal age at which to intervene and the ideal limb length discrepancy are unclear from this study, although we noted no association between age and gain in length in our analysis.
We believe this procedure is indicated in patients with congenital LLDs in whom the discrepancy is predicted to proportionally increase as they age. In these patients, this procedure can be used to manage symptoms during growth, particularly at the point where orthotic usage may become problematic, which normally correlates to a LLD of 2–3 cm. Operating at this degree of discrepancy also reduces the risk of over-correction. Additional interventions to manage LLDs may still be required, but this technique may potentially reduce the magnitude of surgery needed at an older age.
The study was registered with, and approved by, the research and audit committee at the Robert Jones and Agnes Hunt Orthopaedic Hospital NHS Foundation Trust (1920-021).
4. Horn J, Steen H, Huhnstock S, et al. Limb lengthening and deformity correction of congenital and acquired deformities in children using the Taylor Spatial Frame. Acta Orthop 2017; 88(3):334–340. DOI: 10.1080/17453674.2017.1295706.
6. Jenkins DH, Cheng DH, Hodgson AR. Stimulation of bone growth by periosteal stripping. A clinical study. J Bone Joint Surg Br 1975; 57(4):482–484.PMID: 1194317.
10. Crilly RG. Longitudinal overgrowth of chicken radius. J Anat 1972;112(Pt 1):11–18. PMID: 5086208.
12. Warrell E, Taylor JF. The role of periosteal tension in the growth of long bones. J Anat 1979;128(Pt 1):179–184. PMID: 422478.
13. Taylor JF, Warrell E, Evans RA. The response of the rat tibial growth plates to distal periosteal division. J Anat 1987;151:221–131. PMID: 3654353.
14. D’Souza H, Shah NM. Circumferential periosteal sleeve resection: Results in limb-length discrepancy secondary to poliomyelitis. J Pediatr Orthop 1999;19(2):215–221. DOI: 10.1097/00004694-199903000-00016.
15. Edwards DJ, Bickerstaff DB, Bell MJ. Periosteal stripping in achondroplastic children. Little effect on limb length in 10 cases. Acta Orthop Scand 1994;65(3):333–334. DOI: 10.3109/17453679408995464.
16. Limpaphayom N, Prasongchin P. Surgical technique: Lower limb-length equalization by periosteal stripping and periosteal division. Clin Orthop Relat Res 2011;469(11):3181–3189. DOI: 10.1007/s11999-011-2013-9.
© Jaypee Brothers Medical Publishers. 2023 Open Access This article is distributed under the terms of the Creative Commons Attribution-Non Commercial-share alike license (https://creativecommons.org/licenses/by-nc-sa/4.0/) which permits unrestricted distribution, and non-commercial reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. If you remix, transform, or build upon the material, you must distribute your contributions under the same license as original. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.