Premature Physeal Closure Following Pediatric Ankle Physis Fractures

Abstract

Background:

Since there is no consensus on the risk factors of premature physeal closure (PPC) or on its incidence after pediatric ankle physis fractures, we studied these issues.

Methods:

We retrospectively analyzed a cohort of children younger than 16 years admitted to a single university hospital in Switzerland for an ankle physis fracture between 2004 and 2023. Demographic and radiological data were collected, fractures were classified using the Salter–Harris (SH) and Dias–Tachdjian classifications, and statistical analyses included the Fisher exact test for the statistical significance of correlations.

Results:

Twenty-eight (13%) of the 211 patients included (mean age = 12.7 years) experienced PPC. SH II fractures were the most common across the whole population and the subgroups with and without PPC. PPC required surgery in 12/28 cases because it was accompanied by a leg-length discrepancy or joint malalignment. The onset of PPC showed statistically significant correlations for younger age (p < 0.021), SH IV fractures (p < 0.047), and prereduction physeal gaps > 2 mm (p < 0.002).

Conclusion:

The risk of growth disorders after ankle physis fractures is low, but younger age, prereduction physeal gap, and SH IV fractures seem to constitute risk factors for PPC. Proving the statistical correlations between these potential risk factors and PPC will require larger populations.

Level of evidence:

III, retrospective cohort study. See Instructions for Authors for a complete description of levels of evidence.

Introduction

Ankle fractures account for approximately 5.5% of all childhood fractures, and 15% to 20% of these are physeal^1-4. The distal tibial physis is recognized as the third most injured growth plate (9.4% to 10.9% of physeal injuries), following injuries to hand phalanges (25.8% to 37.4%) and to the distal radius and ulna (17.9% to 28.3%)^1,2,5-7. Ankle physis fractures are a significant concern in pediatric orthopedics due to the risk of premature physeal closure (PPC), which can subsequently affect the injured limb’s growth.

The distal tibial physis supplies 30% to 40% of leg-length growth, accounting for 3 to 5 mm annually; the distal fibula physis contributes to a lesser extent^8,9. The consequences of an ankle physis fracture can thus be detrimental to the relevant lower limb’s overall growth, depending on the injured child’s residual growth potential^8,9. This can lead to growth disturbances, including leg-length discrepancies (LLDs), angular deformities, and/or articular incongruities^8-11.

The rates of PPC reported in distal tibial fractures vary widely, ranging from 1% to 66.7%^4,10,12-19, reflecting differences in study sample sizes and the distributions of fracture patterns across the Salter–Harris (SH) classification. Some of these studies may have suffered from selection bias and small-study effects as they focused on small groups with severe fractures requiring surgical treatment. In many studies, the incidence of PPC was primarily related to the overall study population, with no distinctions made between age groups and their related growth potential, which significantly affects PPC’s presentation and management.

Several factors are believed to contribute significantly to PPC, including the mechanism of injury, the energy delivered by the trauma itself, the fracture pattern, the treatment implemented, postreduction residual displacement, the number of repeated reduction maneuvers, and the spread of the periosteum or other soft tissue into the fracture^{10,11,13,14,20,21}. However, the full extent of these risk factors’ associations with PPC remained unknown or debated^{13,14,20,22-24}. We therefore aimed to study a large group of patients treated for ankle physis fractures at our hospital and evaluate the real incidence of PPC and the risk factors that might affect its occurrence.

Materials and Methods

Local ethics committee approved (CE 2023-01624) our retrospective review of the medical charts of all the pediatric patients (aged less than 16) admitted to our institution for an ankle physis fracture between January 1, 2004, and April 30, 2023. Patients with incomplete medical charts, neurological diseases, or a follow-up period of < 12 months were excluded.

We extracted relevant parameters from each patient’s computerized file, such as demographic data (age, sex) and injury information (side of the injured ankle, date, mechanism of injury). The Dias–Tachdjian (DT) classification was used to define mechanisms of trauma, and the SH classification was used to identify specific patterns of physeal fracture¹⁶. Prereduction and postreduction physeal gaps were measured from standard X-ray images¹³ or computed tomography (CT) when available. Pubertal status was noted, using thresholds of 11 and 13 years for girls and boys, respectively. Treatment methods (conservative, surgical) and the numbers of reduction maneuvers and secondary displacements were recorded, as were the treatments implemented after patients exhibited growth disturbances. Time elapsed between injury and the detection of a growth disturbance was recorded. Growth disturbances were defined as complete or partial and assessed as either clinically relevant and requiring surgical correction or not clinically relevant. The affected bone physis was noted, and the ankle’s varus, valgus, or sagittal plane deformities were measured from standard standing radiographs. Any increase or decrease in the patient’s mechanical distal lateral tibial angle of >5° or in the anterior tilt of the distal tibia of >3° was considered relevant. Any LLD >2 cm was also considered pathological^25-30.

Complete case analysis was performed using R software (version 4.2.2, R Development Core Team). The primary outcome was the onset of any growth disturbance, with a secondary outcome being a growth disturbance requiring surgical correction. Descriptive analyses included means, standard deviations (SD), and ranges (minimum; maximum) for continuous variables, and absolute counts and percentages for categorical variables. Univariate logistic regressions were used to calculate odds ratios (OR) with 95% confidence intervals (95% CI), along with their respective effect sizes (ES) and p-values for predicting primary and secondary outcomes. Statistical significance was set at p < 0.05.

Results

Data were collected from 211 patients having sustained ankle physis fractures (Table I). The mean follow-up was 64.0 weeks (SD 55.8), the mean patient age was 12.7 years (SD 2.3), and 127 patients (60.2%) were male. Most patients (150; 71.1%) were through puberty at the time of injury. The most common fracture pattern was SH II (109; 51.7%), followed by SH III (46; 21.8%) and SH IV (45; 21.3%) (Table I). The predominant mechanism of injury was supination-plantar flexion (72; 34.2%) (Table I). A prereduction residual physeal gap of > 2 mm was measured in 86 patients (40.8%) (Table I). Initial treatment was surgery for 171 patients (81.0%) (Table I).

Table I.

Study Cohort’s Characteristics (N = 211)

Characteristic	Overall Cohort (N = 211)
Age at trauma (years) mean (SD) [min; max]	12.7 (2.3) [3.0; 15.95]
Sex, N (%)
Female	84 (39.8%)
Male	127 (60.2%)
Pubertal status*, N (%)
Prepubertal	61 (28.9%)
Pubertal	150 (71.1%)
Salter–Harris classification, N (%)
I	11 (5.2%)
II	109 (51.7%)
III†	46 (21.8%)
IV‡	45 (21.3%)
Mechanism of trauma (Dias–Tachdjian), N (%)
Supination-Inversion	52 (24.6%)
Supination-Plantarflexion	72 (34.2%)
Supination-External Rotation	53 (25.1%)
Pronation-Eversion-External Rotation	33 (15.6%)
Not classifiable	1 (0.5%)
Physeal gap before reduction > 2 mm, N (%)	86 (40.8%)
Comminutive fracture, N (%)	22 (10.4%)
Follow-up (weeks), Mean (SD) [min; max]	64.0 (55.8) [12.0; 232.0]
Initial treatment, N (%)
Conservative	40 (19.0%)
Surgical	171 (81.0%)
Growth disturbance
Yes, n (%)	28 (13.3%)
No, n (%)	183 (86.7%)

Categorical variables are described using absolute counts and percentages.

SD = standard deviation, Min = minimum, and Max = maximum.

^*Pubertal status thresholds of ≥ 11 y.o. for girls and ≥ 13 y.o. for boys.

^†Including SH III McFarland.

^‡Including both biplanar and triplane SH IV fractures, as well as McFarland fractures

Twenty-eight patients (15 boys, 13 girls) exhibited PPC (28/211; 13.3%) (Table II). The mean time between injury and the detection of growth disturbances was 8.3 months. Patients with PPC were younger than those without a growth disturbance (mean age 11.8 vs. 12.9 years; p = 0.021) (Table II). PPC affected the distal tibia alone in 23/28 cases (82.1%), the distal fibula alone in just 1 case, and both the tibia and fibula in 4 cases (14.3%). Rates of PPC by SH classification were 9% for SH I fractures, 15.6% for SH II fractures, 4.3% for SH III fractures, and 17.7% for SH IV fractures. In patients experiencing PPC, SH II and SH IV fractures were the most frequent, representing 60.7% and 28.6% of cases, respectively (Table II). The mechanisms of injury and fracture patterns leading to PPC of both tibia and fibula were 2 SH II pronation-eversion-external rotation (PEER) injuries, 1 SH IV supination-inversion injury, and 1 supination-inversion with SH III of the distal tibia and an SH II of the distal fibula. PPC was not clinically significant in 10 cases but was accompanied by an LLD in 8 patients (28.6%), a joint malalignment (JMA) in 6 (21.4%), and a simultaneous JMA and an LLD in 4 (14.3%). Three transferred patients who required multiple fracture reductions and 1 with secondary displacement developed a physeal arrest.

Table II.

Predictors of Any Growth Disturbance Following Ankle Physeal Fracture

Predictor	Growth Disturbance 28/211 (13.3%)	No Growth Disturbance 183/211 (86.7%)	Odds Ratio (95% CI)	Effect size (ES)	p-Value
Age (years) Mean (SD)	11.8 (2.1)	12.9 (2.3)	0.79 (0.65-0.96)	0.51	0.021
Sex
Female (Ref)	13/28 (46.4%)	71/183 (38.8%)	1.00 (Reference)	—	—
Male	15/28 (53.6%)	112/183 (61.2%)	0.73 (0.33-1.62)	0.05	0.448
Pubertal status*
Prepubertal (Ref)	10/28 (35.7%)	51/183 (27.9%)	1.00 (Reference)	—	—
Pubertal	18/28 (64.3%)	132/183 (72.1%)	0.70 (0.30-1.61)	0.05	0.400
Salter–Harris Classification§
SH I	1/28 (3.6%)	10/183 (5.5%)	0.64 (0.08-5.20)	0.03	0.671
SH II	17/28 (60.7%)	92/183 (50.3%)	1.53 (0.70-3.36)	0.10	0.289
SH III (Ref)†	2/28 (7.1%)	44/183 (24.0%)	1.00 (Reference)	—	—
SH IV‡	8/28 (28.6%)	37/183 (20.2%)	4.76 (0.94-24.09)	0.19	0.047
Mechanism of trauma
SI (Ref)	6/28 (21.4%)	46/183 (25.1%)	1.00 (Reference)	—	—
SPF	9/28 (32.1%)	63/183 (34.4%)	1.10 (0.37-3.26)	0.02	0.866
SER	6/28 (21.4%)	47/183 (25.7%)	0.98 (0.30-3.21)	0.01	0.975
PEER	7/28 (25.0%)	26/183 (14.2%)	2.07 (0.62-6.87)	0.12	0.236
Physeal gap before reduction
≤2 mm (Ref)	9/28 (32.1%)	116/183 (63.4%)	1.00 (Reference)	—	—
>2 mm	19/28 (67.9%)	67/183 (36.6%)	3.65 (1.58-8.43)	0.22	0.002
Initial treatment
Conservative	6/28 (21.4%)	34/183 (18.6%)	1.19 (0.44-3.21)	0.03	0.733
Surgical (Ref)	22/28 (78.6%)	149/183 (81.4%)	1.00 (Reference)	—	—

Percentages refer to the populations specified in each column title.

CI = Confidence Interval, Ref = Reference category for odds ratio calculation, PEER = pronation-eversion-external rotation, SER = supination-external rotation, SH = Salter–Harris classification, SI = supination-inversion, and SPF = supination–plantarflexion.

^*Pubertal status, thresholds of ≥ 11 y.o. for girls and ≥ 13 y.o. for boys.

^†Including SH III McFarland fractures.

^‡Including both biplanar and triplane SH IV fractures as well as McFarland SH III and SH IV fractures.

^§The premature physeal closure rate considering SH classification in the overall population was 9% for SH I fractures, 15.6% for SH II fractures, 4.3% for SH III fractures, and 17.7% for SH IV fractures.

None of the following potential predictors exhibited a meaningful effect size, namely, injury mechanisms (p > 0.1), sex (p = 0.448), pubertal status (p = 0.400), and initial treatment type (p = 0.733) (Table II). SH IV fractures had a moderate effect size for the risk of a growth disturbance relative to SH III fractures (p = 0.047). Moreover, a prereduction physeal gap > 2 mm predicted a 67.9% (19/28) risk of a PPC vs. a 36.6% (67/183) risk for those without growth disturbance (p = 0.002). Finally, greater patient age had a protective role (ES = 0.51, p = 0.021, OR = 0.79), with a 21% decrease in risk per additional year of age (Table II).

Surgical correction of a clinically relevant growth disturbance was necessary for 12/28 patients (42.9%). Operated patients had a similar age than those nonoperated (mean 11.3 vs. 11.8, respectively), without a statistically significant difference (p = 0.282) (Table III). Although 6 patients with both an LLD and JMA met radiological thresholds for surgery, they did not have clinically significant growth disturbances.

Table III.

Predictors of Surgical Intervention Among Patients With Growth Disturbances (n = 28)

Predictor	Surgical Correction Growth Disturbance 12/28 (42.9%)	Nonsurgical Correction Growth Disturbance 16/28 (57.1%)	Odds Ratio (95% CI)	Effect size (ES)	p-Value
Age (years) Mean (SD)	11.3 (2.0)	11.8 (2.2)	0.83 (0.59-1.16)	0.44	0.282
Sex
Female (Ref)	7/12 (58.3%)	6/16 (37.5%)	1.00 (Reference)	—	—
Male	5/12 (41.7%)	10/16 (62.5%)	0.43 (0.10-1.88)	0.17	0.260
Pubertal status*
Prepubertal (Ref)	6/12 (50.0%)	4/16 (25.0%)	1.00 (Reference)	—	—
Pubertal	6/12 (50.0%)	12/16 (75.0%)	0.33 (0.07-1.60)	0.23	0.168
Salter–Harris Classification
SH I	0/12 (0%)	1/16 (6.3%)	0.00 (0.00–)	0.15	0.400
SH II	5/12 (41.7%)	12/16 (75.0%)	0.24 (0.05-1.17)	0.30	0.077
SH III	2/12 (16.7%)	0/16 (0%)	1.00 (Reference)	0.43	0.077
SH IV	5/12 (41.7%)	3/16 (18.8%)	3.06 (0.55-17.10)	0.24	0.200
Mechanism of trauma
SI (Ref)	2/12 (16.7%)	4/16 (25.0%)	1.00 (Reference)	—	—
SPF	3/12 (25.0%)	6/16 (37.5%)	1.00 (0.13-7.82)	0.00	1.000
SER	3/12 (25.0%)	3/16 (18.8%)	2.00 (0.24-16.66)	0.07	0.520
PEER	4/12 (33.3%)	3/16 (18.8%)	2.67 (0.37-19.24)	0.15	0.330
Physeal gap after reduction
≤2 mm (Ref)	3/12 (25.0%)	6/16 (57.5%)	1.00 (Reference)		—
>2 mm	9/12 (75.0%)	10/16 (62.5%)	1.80 (0.35-9.26)	0.12	0.480
Initial treatment
Conservative	4/12 (33.3%)	2/16 (12.5%)	3.50 (0.52-23.55)	0.24	0.200
Surgical (Ref)	8/12 (66.7%)	14/16 (87.5%)	1.00 (Reference)	—	—

*Pubertal status, thresholds of ≥ 11 y.o. for girls and ≥ 13 y.o. for boys. This analysis includes only the 28 patients who developed “any growth disturbance.” It identifies factors associated with the need for surgical correction within that subgroup.

CI = confidence interval, Ref = reference category for odds ratio calculation, PEER = pronation-eversion-external rotation, SER = supination-external rotation, SH = Salter–Harris classification, SI = supination-inversion, and SPF = supination-plantarflexion.

*Pubertal status, thresholds of ≥ 11 y.o. for girls and ≥ 13 y.o. for boys.

SH II and SH IV fractures had the highest rate of surgical intervention (5/12, 41.7% each), followed by SH III (2/12, 16.7%) (Table III). SH II fractures had nonsignificant lower odds of requiring surgery than other types (p = 0.077) (Table III). Moreover, no other predictors for surgical intervention in patients with PPC showed statistical significance (Table III). Surgical correction was used for JMA (3 cases), an LLD (8 cases), and combined JMA and an LLD (1 case).

Discussion

To the best of our knowledge, this study examined one of the largest case series of pediatric ankle physis fractures described to date and provided novel insights into robust epidemiological data on physeal growth disorders and PPC.

First, our findings revealed an incidence of PPC following pediatric ankle physis fractures of 13.3%, close to the lowest rates reported in previous series²⁰. An exhaustive review of the medical literature revealed that the incidence of PPC was extremely variable, ranging from 12% to 54%, depending on the sample sizes and fracture patterns analyzed^{10,11,13,14,20,21}. It thus appears essential to compare studies with broadly similar eligibility and inclusion criteria that do not select specific risk factors for PPC. In 2009, Leary et al. described a case series study with a similar design to ours²⁰ involving 124 pediatric patients presenting with any pattern of physeal fracture of the distal tibia and monitored for at least 1 year. They reported a 12.1% rate of PPC, very similar to our findings²⁰.

Second, our results showed no statistically significant associations between fracture types (characterized by SH classification) and PPC. In this regard, there is a critical lack of consensus on whether different SH fracture patterns are determinants of PPC^{11,13,14,23,31-33}. Indeed, many authors have considered SH I and SH II fractures to be at a low risk of growth arrest, with an incidence below 15%³⁴. Contrarily, other authors have reported surprisingly high rates of PPC with the same fracture types, ranging from 36% to 67% and suggesting that they may fall into a single, high or unpredictable risk category^13,20. In our series, SH IV fractures showed the greatest proportional risk of growth arrest (17.8%), consistent with previous studies on SH III and SH IV fractures that reported moderate rates of growth arrest, ranging from 0 to 11.8%^34-36. However, it is important to note that other studies have shown rates of up to 20%, suggesting that such fractures are the greatest risk factors for PPC^13,31,37.

Interestingly, the amount of residual physeal displacement after fracture treatment explained why SH II and SH IV fractures exhibited higher incidences of PPC. Indeed, various studies have suggested that the prognoses for SH III and SH IV fractures were related to the quality of their reductions, with a residual physeal gap > 2 mm predicting a greater risk of PPC^13,24,31,38. Two recent meta-analyses underscored the importance of a good reduction for preventing PPC^39,40. Unexpectedly, our study revealed evidence that an emphasis should also be put on the prereduction gap, since patients who exhibited growth disturbances were more likely to have a gap >2 mm. This observation suggests either that physeal damage was more extensive than observed or that soft tissues or periosteum were trapped in the fracture site—evidence supported by studies conducted on ovine and rodent models^41-43. It is thus possible that surgical procedures to remove soft tissue and periosteum from the fracture site explain the lower rate of PPC among operated patients, as animal models have suggested^41-43.

In this study, physeal gap measurements were performed from radiographs using a standardized approach described by Barmada et al¹³. Turgut et al. (2022) compared the inter-rater reliability of physeal gap measurements from radiographs and CT, finding that < 2 mm on radiographs corresponded to ≥ 2 mm on CT in 16.4% of cases⁴⁴. However, Banting et al. demonstrated a “fair” degree of inter-rater variability and an absence of intra-rater variability in radiographic measurements, indicating significant discrepancies⁴⁵.

Using the DT classification, our results could not confirm that any specific traumatic mechanisms constituted risk factors for PPC, which is inconsistent with other studies¹⁴. The lack of consensus regarding the suspected risk factors for growth arrest suggests, therefore, that the pathogenesis of PPC is probably multifactorial¹⁵.

The study’s results highlight that a good understanding of the clinical manifestations of PPC and their indications for treatment is relevant to clinical practice. Indeed, despite apparently similar lesions, growth disturbances can manifest themselves in different clinical forms between skeletally immature children and adolescents very close to the end of their growth, with the latter not requiring surgery. In our case series, fewer than half of the children exhibiting PPC required surgical treatment for a clinically significant PPC. It thus seems difficult to compare our findings among patients with growth disturbances to those in the existing literature, as other published studies involved small, essentially descriptive series focused on the specific treatments implemented^13,22,32,46.

Our study had some limitations. The small sample of patients exhibiting PPC likely affected our statistics. Properly exploring the multiple overlapping risk factors for PPC will require future studies with much larger sample sizes. Indeed, in inferential statistical calculations involving correlation analysis, a small sample size can influence coefficients and reduce data precision⁴⁷. We believe this factor was our study’s main limitation, as it was with the other series published on this subject and their derived meta-analyses. This makes it impossible to draw any robust conclusions about the most relevant risk factors for PPC in children with ankle physis fractures.

Conclusions

The incidence of PPC following pediatric physeal ankle fractures in this study was low (13.3%) and correlated with the lowest rates reported in previous series. Few clinical elements appeared to be related to potential risk factors for PPC, and prereduction physeal gap seems to represent the most significant risk factor for PPC. Prospective multicenter studies on this topic should be encouraged to provide larger sample sizes with equivalent numbers of patients for each fracture pattern of interest, thereby yielding more precise rates of PPC and more accurate statistical weights for its risk factors.