Reliability and clinical utility of paediatric ankle fracture classifications

Abstract

Aims

Despite the availability of various classification systems for paediatric ankle fractures, there is a lack of consensus regarding the reliability and utility of each classification, and the corresponding impact on clinical practice. This systematic review aimed to investigate current paediatric ankle fracture classification systems, by assessing their inter- and intraobserver reliability, and their clinical utility in guiding clinical decisions and outcomes.

Methods

A systematic search was performed on PubMed, Embase, and Cochrane Library from inception until 31 October 2025. The inclusion criteria were studies which evaluated the reliability and clinical utility of different paediatric ankle fracture classification systems. Exclusion criteria were adult fractures, paediatric non-ankle fractures, and lack of classification usage.

Results

There were 42 studies covering 11 classification systems, including Salter-Harris, AO, Dias-Tachdjian, Lauge-Hansen, Gerner-Smidt, Ashhurst-Bromer, Carothers-Crenshaw, transitional fractures, atypical triplane fractures, Li-La, and Nenopoulos, in this systematic review. Salter-Harris, AO, Dias-Tachdjian, and transitional fracture classifications are among the more widely used systems in the current field of paediatric orthopaedic surgery. In terms of interobserver reliability, both the AO classification and the Dias-Tachdjian demonstrated substantial to almost perfect agreement for paediatric ankle fractures, followed by the Salter-Harris classification, which showed substantial agreement. Anatomical classification such as the Salter-Harris classification is better at predicting the rate of premature physeal closure. Mechanistic classification such as the Dias-Tachdjian classification is preferred for predicting the rate of angular deformity.

Conclusion

Dias-Tachdjian classification demonstrated the greatest interobserver and intraobserver reliability among paediatric orthopaedic surgeons, while the Salter-Harris classification was the most commonly used and provided the highest predictive value for premature physeal closure. Combining anatomical (e.g. Salter-Harris) and mechanistic (e.g. Dias-Tachdjian or Carothers-Crenshaw) systems may improve prognostic accuracy and treatment planning.

Cite this article: Bone Jt Open 2026;7(6):777–791.

Introduction

Paediatric ankle fractures are fractures that involve the hinge joint formed by the distal tibia, distal fibula, and dome of talus. They represent approximately 5% of paediatric fractures and are the most common type of paediatric lower-limb fractures.^1,2 Common causes of injuries include sports activities, falls, and road traffic accidents.³

Special consideration must be given towards complications that are specific to paediatric ankle fractures. Since the tibia accounts for 15% to 25% of lower-limb length growth in children, premature physeal closure is a detrimental issue which can lead to interrupted height growth and limb length discrepancy.⁴ Moreover, the faster healing rate of fracture in children increases the likelihood of malunion when prompt management is delayed.⁵ Other important complications include angular deformity and neovascular injury.⁶

As a result, accurate clinical diagnosis and appropriate management are crucial in lowering the chance of complications in paediatric ankle fractures. In order to guide clinicians in their decisions on the corresponding treatment plans for different fracture cases, various classifications have been proposed and adopted in the field of paediatric orthopaedic practice. The earliest classification system used was proposed by Ashhurst and Bromer⁷ in 1922, which was originally designed for adult patients. For paediatric patients specifically, the classification system proposed by Bishop⁸ in 1932 was the earliest adopted. Currently, the Salter-Harris classification and the Dias-Tachdjian classification,^9,10 which are based on the anatomy and trauma mechanism of the ankle fracture, respectively, are the more commonly used classification systems. Nevertheless, despite the availability of various classification systems, there is a lack of consensus regarding the reliability and utility of each classification, and the corresponding impact on clinical practice.

Hence, we conducted this systematic review to investigate the paediatric ankle fracture classification systems, by assessing their inter- and intraobserver reliability, and their clinical utility in guiding clinical decisions and outcomes.

Methods

Literature search

This systematic review was performed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. This study was registered at PROSPERO (ID: CRD42023484277). PubMed, Embase, and Cochrane Library were searched from the date of inception using the following search strategy: ‘fracture’ AND (ankle OR tibia) AND (immature OR physis OR physeal OR adolescent OR adolescents OR child OR children OR paediatric OR paediatric). The latest search was conducted on 31 October 2025. Bibliographies of included studies were further screened for any additional eligible literature. Covidence (Australia) was used to manage the available search results and remove duplicates.

Inclusion and exclusion criteria

Two reviewers (LST, LX) independently screened the studies according to a set of inclusion and exclusion criteria. Conflicts were resolved through discussion with a third reviewer (SHST) until a consensus was reached. The inclusion criteria were studies which assessed the reliability and clinical utility of different paediatric ankle fracture classification systems. The exclusion criteria were 1) individual case report or case study, conference report, literature review, or systematic review; 2) animal studies; 3) studies on adult patients; 4) studies on paediatric patients with non-ankle fractures; 5) studies with no established classification system used to assess ankle fractures; and 6) studies with no mention of the impact of the classification system on clinical decision and patient outcomes.

Data extraction

Two reviewers (LST, LX) independently collected data from the included studies. Discrepancies were resolved through discussion with a third reviewer (SHST). The main variables that were collected included: study design, study aim, publication year, country, sample size, classification system used, methods of assessing reliability of the classification system, and important findings including intra- and interobserver reliability, complication rates, and predictability of post-fracture growth pattern.

Risk of bias assessment

The Newcastle-Ottawa Scale for Cohort Studies was used to assess the bias of the included studies.¹¹ All of the studies are cohort studies, cross-sectional studies, case-control studies, and case series and only studies with a score of > 6 were included. Bias was assessed by the same two independent reviewers and any discrepancy was resolved with by the third reviewer (Supplementary Material).

Quality assessment of included studies

The quality of evidence was assessed using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach (Table I).¹²

Table I.

Grading of Recommendations, Assessment, Development and Evaluation (GRADE) table for formal assessment of evidence.

Outcome	No. of studies (design)	Overall certainty (GRADE)
Interobserver reliability of Salter-Harris classification	1 (cross-sectional study)	Very low
Intra-/interobserver reliability of Dias-Tachdjian classification	2 (cross-sectional studies	Low
Interobserver reliability of AO classification	4 (cross-sectional studies)	Low
Inter-/intraobserver reliability reliability with radiograph vs CT using Rapariz classification	1 (cross-sectional)	Very low
Interobserver reliability for Li-La classification	1 (cross-sectional)	Low
Predictive ability for premature physeal closure for Salter-Harris	2 (retrospective cohort studies)	Very low
Prediction of angular deformity for Salter-Harris	3 (retrospective cohort studies)	Very low
Predictive ability for PPC and growth arrest for Dias-Tachdjian classification	1 (retrospective cohort study) 1 (retrospective case series) 1 (cross-sectional) 1 (case-control)	Very low

PPC, premature physeal closure.

Intraobserver and interobserver reliability

According to the guidelines proposed by Landis and Koch,¹³ the kappa values were interpreted as: < 0: poor agreement, 0.0 to 0.20: slight agreement, 0.21 to 0.40: fair agreement, 0.41 to 0.60: moderate agreement, 0.61 to 0.80: substantial agreement, and 0.81 to 1.0: almost perfect agreement.

Statistical analysis

Descriptive analysis of the data was performed, with comparison to the existing evidence in the relevant field of research.

Results

A total of 19,377 studies were identified by the search strategy, and five studies were retrieved from citation searching. Overall, 4,050 duplicates were automatically detected and removed by Covidence, and 15,332 studies were screened for title and abstract. In total, 80 studies were screened for full texts and 42 studies were included in the final review.^10,14-54 The PRISMA flowchart of the screening progress is shown in Figure 1. Out of the 42 included studies, there were seven case-control studies,^{20,28,36,38,40,51,53} ten retrospective cohort studies,^{14-18,21,27,30,37,44,45} 11 retrospective case series,^{10,18,19,25,29,34,35,39,49,52,54} two prospective case series,^26,31 two prospective cohort studies,^33,42 and ten cross-sectional studies.^{22-24,30,32,43,46-48,50}

Fig. 1.

PRISMA flow diagram illustrating the study selection and screening process for the systematic review.

Study characteristics

A total of 42 studies, covering 11 classification systems were included in this systematic review. These included Salter-Harris, AO, Dias-Tachdjian, Lauge-Hansen, Gerner-Smidt, Ashhurst-Bromer, Carothers-Crenshaw, transitional fractures, atypical triplane fractures, Li-La, and Nenopoulos classifications. A total of 4,033 specified paediatric ankle fractures and 1,120 general paediatric fractures were investigated. Eight studies assessed the intra- and/or interobserver reliability of the classification systems,^{22-24,43,46-48,50} and 34 studies compared the classification systems using parameters including complication rate and recovery time.^{10,14-21,25-42,44,45,49,51-54} General characteristics of the included studies are summarized in Table II.

Table II.

General characteristics of the included studies.

No.	First author, year	Study design	Sample size	Classification studied	Study aim
1	Zomorrodi 201153	Case-control	247	SH	To determine the diagnostic and management differences between emergency physicians and orthopaedic physicians for patients with distal fibular physis pain without radiological evidence of fracture
2	Thawrani 201150	Cross-sectional	50	SH	To determine whether or not the addition of CT would improve the reliability of fracture classification and treatment decision
3	Oktay 202240	Case-control	75	SH, DT	To determine the factors that affect prognosis of distal tibial physeal fractures and analyze whether SH or DT is more predictive for outcomes
4	Kärrholm 1982a32	Case-control	26	SH, GS	To evaluate the possibility to predict growth disturbance as determined by the Roentgen stereophotogrammetric method after supination-eversion ankle fractures using GS and SH
5	Kärrholm 1982b28	Cross-sectional	176	SH, GS, LH	To evaluate the use of a combination of traumatological and anatomical classifications in determining the treatment of supination-eversion ankle fractures in children
6	Kärrholm 1983a30	Cross-sectional	147	SH, GS	To evaluate the use of a combination of traumatological and anatomical classifications in determining the prognosis and treatment of supination-adduction ankle fractures in children
7	Kärrholm 1983b33	PCS	55	SH	To evaluate the feasibility of predicting the post-traumatic growth pattern after ankle fractures in children using factors including sex, skeletal maturity, classifications, displacement, treatment
8	Rohmiller 200642	PCS	137	SH, LH	To define the importance of injury mechanism on outcome in SH types I and II fractures of the distal tibia, specifically the complication of PPC
9	Binkley 201916	RCS	141	SH, DT	To determine how fracture pattern in SH II ankle fracture is associated with PPC and the development of angular deformity
10	Kärrholm 1982c31	Prospective case series	10	SH, GS	To determine the growth pattern after supination-adduction ankle fracture in children with GS, SH classifications
11	Aslantas 202014	RCS	27	SH, DT	To evaluate the role of trauma mechanism, fracture pattern, and fixation technique on clinical outcomes and epiphyseal growth arrest in the surgically treated distal tibial epiphyseal fractures
12	Kärrholm 1983c29	Retrospective case series	83	SH, GS, LH	To evaluate the prognosis of pronation type ankle fractures in children with GS, LH, and SH classifications
13	Vahvanen 198051	Case control	310	SH, LH, AB	To compare between AB, LH, SH classification systems in terms of the success in classifying ankle fractures in children
14	de Sanctis 200021	RCS	158	SH, CC	To review paediatric patients with ankle fractures based on CC and SH classifications, and to verify the prognostic ability of these criteria
15	Kärrholm 1982d27	RCS	366	SH, GS	To evaluate the incidence of different fracture types from a defined population, also to compare GS traumatological classification with the SH classification
16	Park 201941	Retrospective case series	50	SH, transitional	To investigate the incidence and location of periosteal entrapment in SH types II, III, or IV fractures and the angle of the fracture plane of metaphyseal fragments on axial plane
17	D’Angelo 201720	Case-control	46	SH, DT	To evaluate risk factors that can influence the outcome of distal tibia physeal fractures, like fracture pattern, fracture displacement, mechanism of injury, and treatment method
18	Landin 198336	Case-control	373	SH, DT	To describe the pattern of ankle fractures in childhood and adolescence by applying an extended classification based on that proposed by DT; to calculate the incidence of ankle fracture in children aged under 16 years; and to reveal possible secular changes
19	Leary 200937	RCS	124	SH, transitional	To determine the incidence PPC after physeal fractures of the distal end of the tibia in children, grouped by SH classification, and also to identify clinical predictors for PPC
20	Caterini 199117	RCS	68	SH	To have a long-term follow-up of paediatric patients with ankle fracture grouped under the SH classification and assess their outcomes
21	Mac Nealy 198238	Case-control	194	SH, transitional, CC	To explore the various types of distal tibial epiphysis fractures according to different classifications: SH, transitional, CC; and to show the prevalence of each sub-type
22	Seel 201145	RCS	225	SH, transitional	To review distal tibial physeal fractures treated at the institution and to characterize the original injury, the treatment given, and any subsequent complications
23	Barmada 200315	RCS	92	SH, transitional	To investigate the incidence and predictors of PPC after paediatric distal tibial fractures, classified according to SH
24	Cottalorda 200819	Retrospective case series	48	SH	To establish the frequency of partial growth arrest of the distal end of the tibia in patients with SH type III and IV medial malleolar fractures
25	Schurz 201044	RCS	376	SH, CC	To evaluate the treatment of distal physeal injuries retrospectively and explain the relationship between the trauma mechanism (CC), the radiological injury pattern (SH), the subsequent therapy, and the functional outcome, as well as to further deduce and verify prognostic criteria
26	Demirel 202422	Cross-sectional	56	DT	To determine intra- and interobserver reliability of DT and investigate the effect of clinician’s experience and expertise on the reliability of the classification
27	Eduardo 201623	Cross-sectional	53	DT, AO	To compare two DT and AO for classifying paediatric ankle fractures
28	Dias 197810	Retrospective case series	71	DT	To develop a new ankle classification validated for paediatric ankle fractures
29	Eismann 201524	Cross-sectional	25	Transitional	To compare the reliability of triplane fracture classification, displacement measurement, and treatment planning with the use of radiographs with and without CT
30	Yung 201952	Retrospective case series	13	Atypical triplane fractures	To evaluate atypical triplane fractures occurrence, mechanism of injury, and fracture patterns in the paediatric population
31	Choudhry 201418	RCS	78	Transitional	To analyze the function outcome of triplane and juvenile Tillaux fractures after closed reduction and percutaneous fixation
32	Tan 201349	Retrospective case series	28	Transitional	To review 28 cases of paediatric triplane fractures of the distal tibia over a 7-year period in a tertiary paediatric hospital
33	Kärrholm 199726	Prospective case series	21	Transitional	To follow-up and review the clinical outcomes of triplane fracture patients
34	Kim 201035	Retrospective case series	14	Transitional	To identify the fracture configurations of triplane and Tillaux fractures using CT and assess the treatment outcomes
35	Kaya 200734	Retrospective case series	10	Transitional	To report the outcomes of open reduction and internal fixation in displaced juvenile Tillaux fractures
36	Slongo 2007a46	Cross-sectional	70	AO	To assess the reliability and accuracy of using standard radiological images to classify paediatric long-bone fractures and their variability among surgeons of various experience levels
37	Slongo 200648	Cross-sectional	*267 paediatric fractures	AO	To introduce AO classification and test its interobserver reliability
38	Slongo 2007b47	Cross-sectional	*275 paediatric fractures	AO	Another phase of study after the introduction of AO classification, to validate the interobserver reliability specifically regarding the severity component of the classification code
39	Joeris 201725	Retrospective case series	*548 lower limb paediatric fractures	AO	To present the most relevant fracture patterns in the lower limbs of a representative population of children classified according to the AO classification
40	Schneidmuller 201143	Cross-sectional	*30 paediatric fractures	Li-La classification	To develop a child-specific classification system for long bone fractures and to examine its reliability and validity on the basis of a prospective multicentre study
41	Nenopoulos 200539	Retrospective case series	83	Nenopoulos classification	To introduce a classification that helps in the method of treatment of prognosis of paediatric ankle fractures involving the distal tibial epiphysis
42	Faggion 202554	Retrospective case series	7	Shin classification	To describe a multicentre cohort of atypical triplane ankle fractures with intramalleolar involvement of the epiphysis, providing insights into the affected population, treatment outcomes, and complications

AB, Ashurst-Bromer classification; CC, Carothers-Crenshaw classification; DT, Dias-Tachdjian classification; GS, Gerner-Smidt classification; LH, Lauge-Hansen classification; PCS, prospective cohort study; PPC, premature physeal closure; RCS, retrospective cohort study; SH, Salter-Harris classification; Transitional, transitional fractures.

Description of included classifications

A total of 11 paediatric ankle fracture classifications were identified across the included studies, comprising of anatomical, mechanistic, and mixed approaches (Figure 2, Supplementary Material). The Salter-Harris classification describes physeal injuries according to the pattern and growth plate involvement. The AO paediatric classification provides a comprehensive coding system involving fracture location and morphology which is applicable to various long bone fractures. The Dias-Tachdijan classification categorizes fractures based on injury mechanism and physeal involvement, derived from the Lauge-Hansen classification, which was designed for adults. The Li-La and Nenopoulous classifications introduce coded and anatomical schemes for paediatric distal tibial injuries. Transitional fracture classifications, such as the Rapariz system for triplane fractures, use descriptors based on the type of fracture fragment, whereas the Shin classification for atypical triplane fractures depends on their relationship to the tibial plafond and whether they are intra- or extra-articular. Three historical classifications were identified, of which Gerner-Smidt and Carothers-Crenshaw follow mechanistic principles, whereas Ashurst-Bromer incorporates both force-based and anatomical descriptors. These classification systems collectively represent the historical and contemporary approaches used to categorize paediatric ankle fractures.

Fig. 2.

Summary of paediatric ankle classification.

Salter-Harris classification

The Salter-Harris classification has been one of the most commonly used systems for describing physeal fracture since it was proposed in 1963.⁹ Although not specific to ankle fracture, its detailed description of the different degrees of physis involvement has been widely used by clinicians to determine the prognosis and treatment of paediatric ankle fractures. There are five conventional fracture types under this system (Supplementary Material).

In terms of the Salter-Harris classification’s reliability, Thawrani et al⁵⁰ evaluated the intra- and interobserver reliability of this classification system assessed by paediatric orthopaedic surgeons. Thawrani et al⁵⁰ concluded that the intraobserver reliability for Salter-Harris classification showed substantial agreement (κ = 0.76 with radiograph and κ = 0.80 with CT) and the interobserver reliability showed moderate to substantial agreement (κ = 0.67 with radiograph and κ = 0.57 with CT) (Table III). Nevertheless, Zomorrodi et al⁵³ pointed out the lack of consistency between Emergency Department and orthopaedic surgeons in the diagnosis of Salter-Harris I ankle fractures specifically. The frequency of Salter-Harris type I ankle fracture diagnoses differed significantly between the two groups (p = 0.01). This could be attributed to the lack of radiological signs on radiograph, as Salter-Harris type I fracture line entirely crosses the growth plate.

Table III.

Intra- and interobserver reliability of different classification systems.

Classification	First author, year	Intra-/interobserver reliability	Qualification of assessors	Results	Remarks
Salter-Harris	Thawrani, 201150	Interobserver	5 paediatric orthopaedic surgeons 2 paediatric orthopaedic fellows	Moderate agreement κ = 0.64	N/A
		Intraobserver		Substantial agreement κ = 0.76
AO classification	Slongo, 2007a46	Interobserver	77 orthopaedic surgeons	Almost perfect κ = 0.95	Specific to ankle fractures
	Slongo, 200648	Interobserver	5 paediatric orthopaedic surgeons	Substantial Metaphyseal - κ = 0.79	Not specific to ankle fractures
	Slongo, 2007b47	Interobserver	70 paediatric orthopaedic surgeons	Slight - fair agreement Epiphyseal - κ = 0.13 Metaphyseal - κ = 0.34	Not specific to ankle fractures
	Eduardo, 201623	Interobserver	2 orthopaedic surgeons 3 orthopaedic fellows	Substantial agreement κ = 0.76	Specific to ankle fractures
Dias-Tachdjian	Demirel, 202422	Intraobserver	Paediatric orthopaedic surgeons General orthopaedic surgeons Orthopaedic residents radiologists	Substantial to almost perfect agreement All surgeons - κ = 0.77 to 0.95 Paediatric orthopaedic surgeons - κ = 0.9, 0.82	N/A
	Eduardo, 201623	Interobserver	2 paediatric orthopaedic surgeons 3 paediatric orthopaedic fellows	Almost perfect κ = 0.90 to 1.0	N/A
Rapariz for triplane fracture	Eismann, 201524	Interobserver (with radiograph)	1 paediatric radiologist 1 musculoskeletal radiologist 3 paediatric orthopaedic surgeons	Slight agreement κ = 0.17	N/A
		Intraobserver (with radiograph)		Fair agreement κ = 0.31
		Interobserver (with CT)		Moderate agreement κ = 0.41
		Intraobserver (with CT)		Moderate agreement κ = 0.54
Li-La classification	Schneidmuller, 201143	Interobserver	Li-La paediatric expert groups	Fair agreement κ = 0.71	Not specific to ankle fractures

N/A, not applicable.

As one of the most commonly used classification systems in assessing paediatric fractures, many studies have evaluated the impact of SH classification on the treatment and clinical outcome of cases. Ten studies acknowledged the utility of SH classification in predicting the likelihood of developing complications including premature physeal closure, disturbed growth pattern, and angular deformity.^{15-17,19,21,37,41,42,44,45} One of the commonly discussed complications, premature physeal closure, was found to be most prevalent in Salter-Harris type II injuries.^37,45 This is because type II injuries create a large metaphyseal fragment, which increases the rate of bony bridge formation across the physis, while for type V injuries, despite being the most severe form, the low prevalence and under-reporting make them not as statistically common as type II injuries when it comes to premature physeal closure (Table IV). In terms of the utility of this classification system, seven studies by Kärrholm et al^27-33 and one study by Vahvanen et al⁵¹ concluded that, being an anatomical classification system, the Salter-Harris classification would have shown better prognostic ability if used together with a mechanistic classification based on the mechanism of fracture, such as the Dias-Tachdjian and the Gerner-Smidt classification systems.

Table IV.

Complications described in different classification systems.

Classification	First author, year	Complications	Findings
Salter-Harris	Leary, 200937 Seel, 201145	PPC	SH type II showed the highest rate due to metaphyseal fragment promoting physeal bar formation SH type V, though severe, has a low prevalence and is under-reported, hence not statistically common as SH type II to cause PPC
	Cottalorda, 200819	Growth arrest	Open reduction for SH type III and IV medial malleolar fracture had high growth arrest rate
Dias-Tachdjian	Dias, 197810 Aslantaş, 202014 Eismann, 201524 Oktay, 202240	PPC	Salter-Harris classification is more predictive of PPC than Dias-Tachdjian due to its greater emphasis on anatomy and fracture pattern
	D’Angelo, 201720	Angular deformity	PEER sub-type was associated with angular deformity and severe displacement
Lauge-Hansen	Rohmiller, 200642	PPC	Degree of post-reduction fracture displacement was the strongest predictor of PPC PAB injuries was associated with greater displacement and higher closure risk but was not statistically significant
Carothers-Crenshaw	De Sanctis, 200021	Growth arrest	Adduction-supination injuries (SH type III, IV, V) had greater risk of permanent physeal damage Abduction, external rotation, plantarflexion injuries (SH type I, II) had lower rates of growth arrest
Rapariz for triplane fracture	Eismann, 201524	PPC	Rates of PPC in triplane fractures are up to 21% Long-term follow-up of ≥ one year recommended
Transitional fractures	Choudhry, 201418	PPC	Occur in adolescents nearing physeal closure Lower risk of long-term complications compared with younger patients A ≥ one-year follow-up for closure monitoring
Atypical triplane fractures	Yung, 201952 Kim, 201035	Joint incongruity	AMES fragment most common (50%) Emphasis on anatomical reduction quality to prevent joint incongruity and arthritis

AMES, anteromedial epiphyseal sleeve; PEER, pronation-eversion-external rotation; PPC, premature physeal closure; SH, Salter-Harris classification.

AO classification

In view of the need to achieve an ubiquitous and reliable way to classify paediatric fractures, the AO Paediatric Classification Group proposed this classification system based on the existing Müller AO classification for adults, with the consideration of child-specific fracture characteristics.⁵⁵ The system describes the fracture location and pattern by assigning a specific code, and the details of the code are summarized in the Supplementary Material.

Though not specific to ankle fracture, the efforts to achieve comprehensiveness and detailed description of the fracture feature demonstrated the original intention of proposing this system: to create a common language in which different surgeons can view and describe fractures in similar ways.

To validate this, four studies have been done to assess the intra- and interobserver reliability of the classification system.^23,46-48 Overall, the AO classification was found to be considerably accurate in assessing paediatric fractures. Slongo et al⁴⁶ reported an almost perfect interobserver reliability of κ = 0.95 in classifying paediatric ankle fracture, and Eduardo et al²³ reported a substantial interobserver reliability of κ = 0.76 (Table III).

In terms of the potential implication of the AO classification for the clinical outcomes, one study pointed out its usefulness in analyzing the epidemiological data in paediatric fractures.²⁵ No studies were performed to validate its value in determining prognosis and choice of management in a clinical setting. While the AO classification provided a broader range of fracture sub-types, Eduardo et al²³ also concluded that its complexity required extensive training for comfortable use by orthopaedic surgeons.

Dias-Tachdjian classification

The Dias-Tachdjian classification system was first described in 1978 by two paediatric orthopaedic surgeons.¹⁰ It is a widely used method for classifying paediatric ankle fractures, adapted from the adult Lauge-Hansen classification. This system categorizes ankle fractures according to physeal involvement and the injury mechanism in skeletally immature patients (Supplementary Material). The classification also characterizes the deforming force, allowing surgeons to reverse the force to facilitate closed reduction and detect associated fractures.²²

The applicability of the Dias-Tachdjian classification for paediatric ankle fractures has been evaluated in terms of intra- and interobserver reliability. Both Demirel et al²² and Eduardo et al²³ reported substantial to almost perfect intraobserver reliability for the classification, with kappa values ranging from 0.77 to 0.95 (p < 0.01) and 0.90 to 1.0, respectively. Paediatric orthopaedic surgeons showed the highest interobserver kappa values (κ = 0.90, 0.82, p < 0.01), suggesting that increased expertise in paediatric orthopaedics leads to more consistent classification of ankle fractures (Table III). Eduardo et al²³ also compared the Dias-Tachdjian classification with the AO classification, and both systems showed consistent reproducibility.

However, in terms of clinical utility, Eduardo et al²³ noted that most fractures were classified as Dias-Tachdjian stage 1, despite differences in surgical management and implants. As a result, they concluded that both the Dias-Tachdjian and AO classifications had limited utility for surgical planning.

Despite this, it was noted that the pronation-eversion-external rotation (PEER) mechanism sub-type exhibited the greatest prereduction displacement compared with other Dias-Tachdjian sub-types.²⁰ Additionally, it was associated with a higher rate of angular deformity,¹⁶ making it a more severe injury pattern in paediatric ankle fractures. Four studies have examined the impact of the Dias-Tachdjian and Salter-Harris classifications on premature physeal closure and epiphyseal growth arrest.^10,14,16,40 Neither classification showed significant difference in the rates of epiphyseal growth arrest among the sub-types of each classification.¹⁴ However, the Salter-Harris classification was more effective in predicting premature physeal closure compared with the Dias-Tachdjian classification.⁴⁰ This is likely because the Salter-Harris classification places greater emphasis on the anatomy and fracture pattern, which are key factors in determining the prognosis for distal tibial physeal fractures (Table IV).¹⁴

Lauge-Hansen classification

While the Lauge-Hansen classification is primarily used to classify adult ankle fractures according to the mechanism of injury, its applicability in the paediatric population remains unvalidated.⁵⁶ This can be attributed to the complexity of identifying trauma mechanisms in a paediatric population. Vahvanen and Aalto⁵¹ noted that this classification cannot effectively classify biplane and triplane fractures and only occasionally identifies distal tibial epiphyseal separations and intra-articular fractures which the Salter-Harris classification can do. Thus, the Lauge-Hansen classification should not be used in isolation, as it focuses solely on the mechanism of injury and neglects fractures involving the growth plate, which may lead to premature physeal closure if undetected.

Pronation-abduction (PAB) injuries were found to have significantly greater prereduction displacement. Although there was a higher rate of premature physeal closure in PAB injuries compared with supination-external rotation (SER) injuries, this difference did not reach statistical significance.⁴² Nevertheless, the authors advised surgeons to inform patients with PAB injuries about the potential risk of premature physeal closure. Ultimately, the degree of post-reduction fracture displacement was the strongest predictor of premature physeal closure.

Gerner-Smidt classification

The Gerner-Smidt classification was introduced in 1963.⁵⁷ It is a mechanistic classification of ankle fractures in children based on the terminology used by the Lauge-Hansen classification.⁵⁸ It described five types of fractures: supination-adduction, supination-eversion, supination-inversion, pronation-abduction, and pronation-eversion.²⁸ Both the Gerner-Smidt and the Dias-Tachdjian classifications were developed based on the Lauge-Hansen classification, while the Gerner-Smidt classification has been adopted more by European surgeons, the Dias-Tachdjian has been quoted more often in North America. The original article is in Danish and is unfortunately unretrievable.⁵⁷ Several studies done by Kärrholm et al^27-33 have discussed the utility of this classification system and come to the conclusion that Gerner-Smidt classification would be of greater clinical value if applied together with an anatomical classification like the Salter-Harris Classification system.

Ashhurst-Bromer classification

The Ashhurst-Bromer classification was developed for adult ankle fractures in 1922 and is the first modern classification for ankle fractures.⁷ It is primarily focused on the trauma mechanism of the ankle fracture, as the authors believed that mechanistic classification is better understood and more easily remembered. Its sub-types include external rotation, abduction, adduction, compression, and direct violence. Interestingly, Ashhurst and Bromer⁷ also included an additional anatomical classification for the purpose of completeness, which resembled the Danis-Weber classification proposed decades later.^56,59 Details of the classification are summarized in the Supplementary Material.

One study investigated the usage of the anatomical segment of the Ashhurst-Bromer classification,⁵¹ which was found to be unable to adequately classify two-plane and triplane fractures, as well as distal tibial epiphyseal separations and intra-articular fractures. It is less effective than the Salter-Harris classification in categorizing common fractures in the paediatric population.⁵¹ This is likely due to its failure to account for epiphyseal lesions, which are prevalent in paediatric ankle fractures.

Carothers-Crenshaw classification

The Carothers-Crenshaw classification is based on the mechanism of trauma for paediatric ankle fractures and is modified from the Ashhurst-Bromer classification system.⁶⁰ It includes six sub-types: abduction, external rotation, plantar flexion, axial compression, direct violence, and adduction. This classification emphasized the importance of accurate reduction in adduction injuries.⁶¹

The application of trauma mechanism classifications, such as the Carothers-Crenshaw classification, stressed that identifying the involvement of physeal plates is vital to ascertain the extent of damage and prognosis for distal tibial fractures. Fractures causing permanent damage to physis are due the mechanism of adduction supination, which results in a Salter-Harris type III, IV, and V fractures of the distal tibia, whereas, injuries involving abduction, lateral rotation, and plantarflexion resulting in Salter-Harris type I and II fractures have a lower complication rate (Table IV).²¹

Li-La classification

Apart from the aforementioned classifications, several new classifications that may apply to paediatric ankle fracture have been proposed, of which Li-La classification is one.⁴³ Similar to AO classification, Li-La classification consists of a code made of five or six digits. The notable differences include that Li-La classification has a different, simpler code regarding the pattern of fracture morphology; it also does not include a code describing the severity of the fracture.

Nenopoulos classification

In addition to the Li-La classification, Nenopoulos et al³⁹ proposed a new anatomical classification system which specifically focused on the fractures involving the distal tibial epiphysis. This classification splits the types of ankle fractures into three groups. Group 1 encompasses fractures of the medial malleolus region, and it is divided further into Group 1a – separation of the medial part of the distal tibial epiphysis and Group 1b – fracture of the medial malleolus; Group 2 describes fractures of the lateral portal of the distal tibial epiphysis; and Group 3 includes fractures of the distal tibial metaphysis-epiphysis in three dimensions.

Transitional fractures

As the distal tibial physis matures, it closes asymmetrically: starting centrally, then medially, and finally anteriorly and laterally, over 18 months.⁵⁴ This asymmetry in adolescents increases the risk of transitional fractures, including juvenile Tillaux and triplane fractures, as the ankle transitions from skeletally immature to mature. Therefore, injuries during this period may result in different fracture patterns compared with younger children with open physes.⁵⁴

Triplane fracture involves fracture lines through the physis, epiphysis, and metaphysis, in each of the axial, sagittal, and coronal planes. On an anterior-posterior radiograph, triplane fracture appears as a Salter-Harris type III fracture, while on a lateral radiograph, it appears as a type II fracture, reflecting its complex nature that requires specific classification. The Rapariz classification is specific to triplane fractures and consists of seven configurations. Using only radiographs to classify paediatric triplane fractures with the Rapariz system showed slight interobserver (κ = 0.17) and fair agreement for intrarater reliability (κ = 0.31).²⁴ However, adding CT improved interobserver (κ = 0.41) and intrarater reliability (κ = 0.54) to moderate levels. Therefore, CT is recommended as a valuable tool for assessing triplane fractures, enhancing surgeons’ visualization of fracture lines and aiding preoperative planning.²⁴ Juvenile Tillaux fracture is a less complex transitional fracture, involving fracture line through the anterolateral physis and epiphysis, with extension into two planes.

Triplane and juvenile Tillaux injury typically occur in patients aged between 12 and 15 years who have minimal growth remaining, resulting in a lower incidence of growth disturbances.⁴⁹ Nevertheless, premature physeal closure is still a known complication of triplane fractures due to articular involvement, with rates reported at 21% and 7%.^15,37 Three studies recommend long-term follow-up of at least one year to monitor for premature physeal closure and growth complications (Table IV).^15,18,37

Atypical triplane fractures

Atypical triplane fractures are extra-articular triplane fractures and have been described in case reports and series. Yung et al⁵² reported the largest series of atypical triplane fractures, identifying the anteromedial epiphyseal sleeve (AMES) fragment as the most common pattern, accounting for 50% of cases. The AMES fragment, a previously undescribed injury, is now classified as a type 4 triplane fracture. It likely results from external rotation or eversion of a plantarflexed foot, causing posterolateral widening of the metaphyseal fragment with avulsion of the anteromedial cortex.

Shin et al⁶² evaluated triplane fractures that did not involve the tibial plafond and developed a classification system, ranging from type 1 to 3. A treatment algorithm based on Shin’s classification suggests anatomical reduction for type 1 and 2 fractures to restore joint congruency and prevent osteoarthritis (Table IV).⁶² Displaced fractures require closed or open reduction and internal fixation, while non-displaced fractures can be treated with casting. For type 3 and 4, the decision for surgery depends on displacement; closed reduction and cast immobilization for displaced fractures, and cast immobilization for non-displaced fractures are recommended. Faggion et al⁵⁴ further applied Shin’s classification and found that most atypical triplane fractures were type 3. Using the same treatment algorithm, they reported no subsequent complications and demonstrated favourable functional outcomes, reflected by high AOFAS scores.

Discussion

Various classification systems have been used for paediatric ankle fractures. Our systematic review demonstrated that Salter-Harris, AO, Dias-Tachdjian, and transitional fracture classifications are among the more widely used systems in the current field of paediatric orthopaedic surgery. They demonstrated good reliability, with proven values in predicting clinical outcomes. Among the classifications evaluated, the Dias-Tachdjian classification demonstrated the best reliability and reproducibility among paediatric orthopaedic surgeons, while Salter-Harris is the most useful in terms of prediction of growth arrest. In addition, systems like Rapariz classification for transitional fractures and AO classification require advanced imaging or specific training, which may limit their practical application in general paediatric settings but could be highly beneficial in specialized cases.

Prior to our systematic review, there were a few literature reviews which mentioned the existing knowledge of the common classification systems of paediatric ankle fracture. Wuerz and Gurd⁶³ in 2013 and Venkatadass et al² in 2021 discussed Salter-Harris and transitional fracture classifications and the specific management for each type of fracture under the respective classification. Cancino et al⁶⁴ in 2021 described four classification systems including Johnson-Fahl, Salter-Harris, Dias-Tachdjian, and transitional fractures in a chronological manner, demonstrating the evolution of understanding of ankle fracture in the field of paediatric orthopaedic surgery. To our knowledge, no prior systematic review has been conducted; our systematic review gathered evidence on most of the previously proposed classification systems and evaluated the reliability and clinical utility of several major classification systems.

The Salter-Harris classification is the most commonly used classification in paediatric fracture. Given its moderate to substantial interobserver reliability when it comes to ankle fracture,⁵⁰ it is feasible and beneficial to adopt it widely among both emergency physicians and orthopaedic surgeons. In contrast, despite demonstrating higher reliability than the Salter-Harris classification,^23,46 the newly proposed AO classification uses a relatively complex coding system for fractures, thereby creating a steeper learning curve when it comes to clinical application. The Dias-Tachdjian classification, which primarily accounts for injury mechanism and integrates the Salter-Harris classification under its sub-categorization, also shows high intraobserver and interobserver reliability among experienced paediatric orthopaedic surgeons. Nevertheless, its clinical applicability has limits, particularly for surgical planning.²³ This classification system demonstrates that experience level plays a significant role in classification accuracy, which aligns with findings in other systems that also emphasize the importance of specialized knowledge for accurate fracture categorization. The Rapariz classification for transitional fractures was found to have poor agreement in terms of interobserver reliability when using radiograph and fair agreement with the addition of CT scan.²⁴ This is not unexpected, as the dynamic bone changes during the physeal closure process make accurate classification challenging. Hence, transitional fractures should be carefully assessed on an individual-case basis.

In terms of clinical utility, the Salter-Harris classification offers more value in terms of predicting the rate of premature physeal closure, due to its focus on the pattern of physeal involvement of the fracture. The Dias-Tachdjian classification adds prognostic value by accounting for the mechanism of injury, which some studies indicate may predict certain complications more accurately than anatomical classifications alone. Specifically, fractures with PEER mechanisms had higher rates of angular deformity and may require greater care to ensure anatomical reduction.²⁰ The AO classification is still relatively new, and hence there are insufficient data to justify its clinical utility at the current stage. Other, less common classifications like the Gerner-Smidt and Carothers-Crenshaw classifications were modified from even earlier pre-existing classifications. Together with their predecessors, they have been mostly superseded by more recent classification systems (Table IV).

Looking into the evolution of classification systems for paediatric ankle fracture, we observed a convergent trend between anatomical classification and mechanistic classification. The Ashhurst-Bromer classification was the first ankle fracture classification, proposed in the year of 1922,⁷ primarily based on the mechanism of trauma. The Carothers-Crenshaw classification in 1955 was modified from it.⁶⁰ The Lauge-Hansen classification, a mechanistic classification used in adult ankle fracture,⁵⁸ was proposed in 1950, and two other trauma-focused classifications in paediatric patients – Gerner-Smidt (1963) and Dias-Tachdjian (1978) – were developed from it.^10,57 The popular Danis-Weber classification in adults is another mechanistic classification, proposed in 1966.^56,59 On the other hand, Bishop (1932),⁸ Aitken (1936),⁶⁵ and Johnson-Fahl (1957)⁶⁶ classifications were the earliest anatomical classifications before the Salter-Harris system came in 1963.⁹ The aforementioned Dias-Tachdjian (1978) system,¹⁰ while being primarily mechanistic, sub-categorizes the fractures based on the Salter-Harris classification system. The 2007 AO classification is a mixed system,⁵⁵ which has incorporated both anatomical and mechanistic characteristics of the fracture, with further consideration of the fracture’s severity. This trend aligns with the recommendations made by several of our included studies, which state that a combination of anatomical and mechanistic classifications will offer greater clinical value.^27-33,51

This review has highlighted various evidence gaps within the current literature on paediatric ankle classifications. First, only a small number of studies have formally assessed interobserver or intraobserver reliability, especially for newer fracture classifications such as the Shin classification for atypical triplane fractures. AO classification and Li-La classification only analyzed interobserver reliability, without intraobserver reliability. Additionally, few studies have been conducted to compare between fracture classifications directly. Eduardo et al²³ compared the Dias-Tachdjian and AO classifications, and several studies by Kärrholm et al^27-33 examined the combined use of Gerner-Smidt with Salter-Harris classification, but comprehensive comparative analyses across multiple classification systems remain absent. Most available studies are retrospective and single-centre with small sample sizes, further limiting the certainty of conclusions. Moreover, a considerable portion of the evidence base comprises older studies that predate the use of advanced imaging systems in paediatric trauma assessment. Future research priorities should focus on validating the clinical utility of existing classifications and designing multicentre prospective studies with larger sample sizes, to determine which system is reliable and clinically useful in guiding treatment and predicting outcomes in paediatric ankle fractures.

Our study has several strengths: this is the first comprehensive review of paediatric ankle fracture classification systems that evaluates their utility, as well as interobserver and intraobserver reliability, across different groups of clinicians, including paediatric orthopaedic surgeons, general orthopaedic surgeons, and emergency physicians. The findings provide guidance on the most appropriate classification systems to use based on the clinical context and user profile. In addition, strict inclusion and exclusion criteria were used. Studies discussing paediatric ankle fractures without using a clear classification system were not included, because they would have shifted the focus of this review towards other aspects of ankle fracture such as aetiology and management options. Furthermore, a broad and comprehensive search strategy was adopted in order to capture a large number of relevant studies, given the relatively specific nature of our discussion.

Nevertheless, our study has its limitations. Only a few studies have quantitively evaluated the reliability of classification systems in terms of intra- and interobserver agreement, which made combined statistical analysis challenging. The variation in the sample size of the fracture cases, and the difference in number and background of the observers, would also have introduced heterogeneity if a statistical analysis were performed. Other than that, there is a substantially larger number of studies on the more commonly used classification systems like Salter-Harris and Dias-Tachdjian, as compared with the rest. Therefore, findings that favoured the more widely adopted systems may have inherent publication bias. In addition, the strength of the available evidence is limited by the predominance of retrospective study designs, which are subject to incomplete documentation, selection biases, and incomplete follow-ups. Consequently, the certainty of evidence is low to very low, underscoring the need for more rigorous prospective studies.

In summary, our findings suggest that no single classification system is universally optimal for paediatric ankle fractures. The Dias-Tachdjian classification demonstrated the greatest interobserver and intraobserver reliability among paediatric orthopaedic surgeons, while the Salter-Harris classification was the most commonly used and provided the highest predictive value for premature physeal closure. Combining anatomical (e.g. Salter-Harris) and mechanistic (e.g. Dias-Tachdjian or Carothers-Crenshaw) systems may improve prognostic accuracy and treatment planning. Given these findings, future research should aim to develop integrative frameworks that further blend anatomical and mechanistic insights to improve their prognostic reliability, clinical utility, and treatment outcomes in paediatric ankle fractures.

Take home message

- There are currently 11 classification systems for paediatric ankle fractures. The Dias-Tachdjian classification demonstrated the greatest interobserver and intraobserver reliability among paediatric orthopaedic surgeons, while the Salter-Harris classification was the most commonly used and provided the highest predictive value for premature physeal closure.

- Combining anatomical (e.g., Salter-Harris) and mechanistic (e.g., Dias-Tachdjian or Carothers-Crenshaw) systems may improve prognostic accuracy and treatment planning.

Data Availability

All data generated or analyzed during this study are included in the published article and/or in the supplementary material.