Treatment principles, prognostic factors and controversies in radial neck fractures in children: A systematic review

Abstract

Introduction

Radial Neck fractures are rare injuries in children. There is controversy surrounding their prognosis and management.

Aims

This review aims to produce an up-to-date summary to clarify prognostic factors and management principles, in the light of recent, better-quality evidence.

Methods

A systematic review was undertaken in accordance to PRISMA guidelines, applying pre-defined selection criteria. 6 papers were found suitable after quality assessment. All were observational cohort studies, one prospective and the rest retrospective. A semi-qualitative review was undertaken as heterogeneity, especially in the fracture classification and outcome assessment tools used, prevented quantitative synthesis.

Results

Majority of these fractures occur at the metaphysis. Consistently good results are seen with simple immobilization in fractures angulated<30° with translation<50%, with nearly all achieving a good outcome. In more displaced fractures, results are poorer and only about 70% patients achieve a good outcome. Both higher fracture displacement and more invasive treatment are associated with worse outcomes, but also with each other. Associated injuries are common, with Proximal Ulna fractures being commonest (71%), but their effect on outcomes is unclear. Age more than 10 years is associated with worse displacement, more invasive treatment and worse results. There is much confounding among all these factors which remains to be convincingly addressed. An algorithmic approach is advisable for these fractures, with stepwise application of more invasive treatment only if less invasive methods fail. Percutaneous fixation with either K-wires or retrograde intramedullary elastic nails is acceptable. Fractures reduced closed in theatre fare better if fixed percutaneously to prevent re-displacement. Open treatment should be considered only if the fracture can’t be reduced to within the displacement limits of angulation<30° and translation<50%. Incidence of serious complications is generally low (3–5%), but their effect on outcomes is unclear.

Conclusions

The treatment of higher-grade radial neck fractures is still controversial, needing further research, possibly through multi-center prospective data collection in pediatric fracture registries using validated outcome measures.

1. Introduction

Radial Neck fractures in children are rare, comprising about a tenth of all paediatric elbow injuries.1, 2, 3 As such, there is lack of standardised treatment guidelines for these fractures. Many different classification systems, treatment selection criteria, methods of treatment and outcome assessment tools are used, contributing to controversies and a reliance on expert opinion.⁴ The last evidence summary was published 20 years ago and was based largely on small cohorts and case-series.⁵ This review aims to incorporate better quality evidence that has since emerged, to produce an up-to-date summary regarding these fractures, clarify broad principles of treatment and identify areas for further research.

2. Methods

The study was conducted in accordance to PRISMA guidelines.⁶ A study protocol was made but not registered on a database.

Research question- Factors affecting outcomes after radial neck fractures (treated by operative or non-operative means) in children younger than 16 years.

Time frame- We aimed to include studies not considered by Evans & Graham in their review published in 1999.⁵ We began our search from 01/01/1997 with an intention to capture any articles published while it was being written and peer reviewed.

Selection criteria- These were determined before undertaking the literature search.

Inclusion criteria: All studies published in English, which include patients-

• •with radial neck fractures;
• •aged <16 years;
• •treated by both non-operative and operative methods; and
• •more than 20 in number

Exclusion criteria: Studies which include patients-

• •with intra-articular radial head fractures only;
• •treated by a single treatment modality (operative or non-operative); or
• •with previous elbow injuries.

The threshold of 20 patients was applied as it was felt that meaningful conclusions could not be drawn from smaller series. SK conducted a systematic search of three online databases (MEDLINE, EMBASE, CINAHL) on 05/08/2018. Different combinations of MeSH terms were used to broaden the scope of the search (see Table 1). SK and SO then independently screened the titles and abstracts against the selection criteria. Differences were resolved in consultation between all authors.

Table 1.

Medline search strategy (05/08/2018).

#	Database	Search term	Results
1	Medline	exp "RADIUS FRACTURES"/	8819
2	Medline	(RADIUS FRACTUR∗).ti,ab	8134
3	Medline	(RADIAL FRACTUR∗).ti,ab	5706
4	Medline	("RADIAL FRACTUR∗").ti,ab	937
5	Medline	("RADIUS FRACTUR∗").ti,ab	3087
6	Medline	(head).ti,ab	283068
7	Medline	(neck).ti,ab	186240
8	Medline	(6 OR 7)	384766
9	Medline	(1 OR 2 OR 3 OR 4 OR 5)	15113
10	Medline	(9 AND 8)	2565
11	Medline	("radial head fractur∗").ti,ab	573
12	Medline	("radius head fractur∗").ti,ab	16
13	Medline	("radius neck fractur∗").ti,ab	5
14	Medline	("radial neck fractur∗").ti	116
15	Medline	(11 OR 12 OR 13 OR 14)	702
16	Medline	(10 OR 15)	2565
17	Medline	16 [DT 1997–2018] [Human age groups Infant OR Child, preschool OR Child OR Adolescent] [Languages English]	402

The full text of retrieved articles was reviewed, including the references to ensure a comprehensive coverage of literature. We used the National Heart, Lung and Blood Institute (NIH) quality assessment tool for observational cohort studies,⁷ and the studies that scored Fair or Good were used for synthesis (PRISMA flow chart-see Fig. 1). As the papers used different classification systems and outcome measures, a meta-analysis was not possible. Hence a qualitative synthesis was undertaken. Results have been presented according to PRISMA reporting standards.

Fig. 1.

PRISMA flow chart.

3. Results

A total of 841 articles were obtained after the primary search of which 310 were duplicates. After applying our selection criteria, a total of 7 studies were found suitable for inclusion. All the studies scored fair on quality assessment (see Table 3) except one which scored poor⁸ and was excluded. For a summary of the results see Table 2.

Table 3.

Study quality using NIH Quality Analysis Tool for Observational Cohort Studies.

Question	Schmittenbecher et al.	Tan and Mahadev	Zimmerman et al.	Basmajian et al.	De Mattos et al.	Badoi et al.	Guitterez- de la Iglesia et al.
Was the research question or objective in this paper clearly stated?	+	–	+	+	+	+	+
Was the study population clearly specified and defined?	+	+	+	+	+	+	+
Was the participation rate of eligible persons at least 50%?	na	na	na	na	na	na	na
Were all the subjects selected or recruited from the same or similar populations (including the same time period)? Were inclusion and exclusion criteria for being in the study prespecified and applied uniformly to all participants?	++	++	++	++	++	−/+	++
Was a sample size justification, power description, or variance and effect estimates provided?	–	–	–	–	–	–	–
For the analyses in this paper, were the exposure(s) of interest measured prior to the outcome(s) being measured?	+	+	++	+	+	+	+
Was the timeframe sufficient so that one could reasonably expect to see an association between exposure and outcome if it existed?	+	+	+	+	+	nr	+
For exposures that can vary in amount or level, did the study examine different levels of the exposure as related to the outcome (e.g., categories of exposure, or exposure measured as continuous variable)?	+	+	++	+	+	+	+
Were the exposure measures (independent variables) clearly defined, valid, reliable, and implemented consistently across all study participants?	+	+	+	+	+	+	+
Was the exposure(s) assessed more than once over time?	na	na	na	na	na	na	na
Were the outcome measures (dependent variables) clearly defined, valid, reliable, and implemented consistently across all study participants?	+	+	+	+	+	+	+
Were the outcome assessors blinded to the exposure status of participants?	nr	nr	nr	nr	nr	nr	nr
Was loss to follow-up after baseline 20% or less?	+	nr	+	nr	–	–	+
Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)?	–	+	++	–	+	–	+
Comments	Prospective, No statistical analysis	Retrospective, limited attempt to address confounders	Retrospective, Statistical method addressed most confounders	Retrospective	Retrospective, Large cohort, Some confounders addressed	Retrospective, Historial control, Incoherent statistical analyses, follow-up not mentioned	Retrospective, Judet Grade 3/4 only, Some confounders addressed
	Fair	Fair	Good	Fair	Good	Poor	Good

Table 2.

Summary of reviewed studies.

Study	Schmittenbecher et al.	Tan and Mahadev	Zimmerman et al.	Basmajian et al.	De Mattos et al.	Guitterez- de la Iglesia et al.
Journal/Year published	JPO 2005	JPO 2011	JBJS(Am) 2013	JPO B 2014	JPO 2016	JPO 2017
Design	Prospective cohort	Retrospective cohort	Retrospective cohort	Retrospective cohort	Retrospective cohort	Retrospective cohort
Patients analysed (considered)	66 (94)	108	140 (151)	78	193; 307 (756)a	51
Key selection criteria	Inclusion: open physis of radial head	NR	Exclusion: Treated by manipulation outside Operation Theatre; Cognitive impairment; Radial head dysplasias	Exclusion: Age <3 or >15; Not followed till clinical/radiographic union	Inclusion: any open physes around elbow Exclusion: inadequate radiographic/follow-up data	Inclusion: >30° angulation/50% displacement (Judet 3+) Exclusion: Follow-up <6 m; Associated injuries
Mean age (years)/Gender mix (% female)	8.6 (1.8–14.3) NR	8.7 (2–14) 46%	8.4 ( ±2.9) 60%	7.8 (3–15) 55%	8.4 (1.3–16.1)a NR	8 (3–15)
Mean follow-up (months)	<5 (Gr 1/2); >6 (Gr 3/4)	32.4 (12–60)	13.3 ( ±2)	4.4 (0.75–28)	Op:2.5 (0.3–39.8) Non-op:6.75 (0.8–39)	14 (8–39)
Fracture Severity grades	Modified Judet	Modified Judet	O’Brien; Exact an-gulation/translation	Judet	O’Brien	Modified Judet
Outcome scales/measures	Metaizeau	Tibone	Defined Successful/Unsuccessful	Tibone	Leung/Peterson	Metaizeau; Mayo Elbow Performance Score
Proportion with lower fracture grades/non-op Rx	55%/35%	78%/87%	23% of those Rx in OT	35%/44%	NR/87%	Excluded
Outcomesb in lower fracture gradesc	100% (36/36)	100% (85/85)	Not reported results by grade	81% (22/27)	Not reported results by grade	Excluded
Outcomes in higher fracture gradesd	77% (23/30)	83% (19/23) (Gr 3–88%; Gr 4–70%)	Not reported results by grade	56% (26/45)	Not reported results by grade	88% (45/51) with Metaizeau (Gr 3–95%, 4a- 88%, 4b- 77%); 100% using MEPS
Outcomes in non-operative treatment	100% (40/40)	100% (94/94)	NR	83% (29/35)	89% (124/139)	NA
Outcomes in patients with casting alone	100% (36/36)	100% (86/86)	NR	89% (17/19)	NR	NA
Outcomes in Closed Reduction (CR) +cast	100% (4/4)	100% (8/8)	NR	75% (12/16)	NR	85% (6/7)
Proportion of CR + cast (of those going to theatre)	36% (8/22)	NR	42% (63/151)	NR	NR	NR
Proportion needing surgery	39% (26/66)	13% (14/108)	58% (88/151) of those going to theatre	55% (43/78)	13% (100/756)	Gr3- 71% 4a- 94% 4b- 100%
Proportion needing open surgery (of all patients needing operative treatment)	17% (5/30)	50% (7/14)	53% (46/88)	40% (17/43)	40% (27/68)	39% (17/44) Gr3-7% Gr 4a-38% Gr 4b-85%
Outcomes in those needing surgery	73% (19/26)	71% (10/14)	NR	58% (25/43)	73.5% (39/53)	88% (39/44)
Outcomes in percutaneous reduction + fixation	81% (17/21)	86% (6/7)	NR	73% (19/26)	NR	93% (25/27)
Outcomes in those needing OR	40% (2/5)	57% (4/7)	NR	35% (6/17)	NR	82% (14/17)
Incidence of associated injuries	39% (26/66) (Gr 1/2–33%; Gr 3/4–47%)	19% (21/108)	46% (70/151)	37% (29/78)	Operatively treated cohort- 34% (23/68)	Excluded
Incidence of Complications/Common complications reported	NR	21% (25/118) Cubitus Valgus- 12%	27% (35/131) HO- 20%	36% (28/78) Stiffness- 31%	4.6% (14/307) Op-12% Non-op-3%	NR Radial head overgrowth-31% Malunion- 20%
Heterotrophic Ossification	NR	4.6% (5/108)	20% (26/131)	1.3% (1/78)	1.6% (5/307) Op-2.9% Non-op-1.3%	NR
Avascular Necrosis	NR	0	3% (4/131)	2.6% (2/78)	1.3% (4/307) Op-4.4% Non-op-0.4%	NR
Radio-Ulnar Synostosis	NR	1.8% (2/108)	NR	2.6% (2/78)	0.3% (1/307) Op-1.5%	2% (1/51)
Nerve Injury	1.3% (1/76)	3.7% (4/108)	2% (3/131)	0	0.3% (1/307) Non-Op- 0.5%	2% (1/51)
Physeal closure	6.5% (5/76) Gr3/4–16%	NR	3% (4/131)	0	0	NR

NR= Not Reported/NA= Not applicable/OT= Operation Theatre/Op = Operatively treated cohort/Non-op = Non operatively treated cohort.

a) Demographic data is from 307 pts; 193 were included for ROM outcome analysis.

b) Proportion of Good/Excellent outcomes according to respective grading system used.

c) Grades 1/2 in Judet classification and 1 in O’Brien classification- Angulation <30°/Translation<50%.

d) Grades 3/4 in Judet classification and 2/3 in O’Brien classification- Angulation >30°/Translation>50%.

3.1. Description of studies included

All were observational cohort studies, and all were retrospective except one.⁸ The average duration of follow up was more than 6 months in all except two studies.^10,11 The average age of the children was 8.4 years (7.8–8.7).

3.2. Exposure measures

All studies used a classification system to grade fractures based on displacement-one study¹⁰ used Judet’s classification system,¹⁵ three^9,13,14 used Metaizeau’s modification,³ and two^11,12 followed O’Brien’s classification¹⁶ (see Table 4). One study factored in the exact displacement.¹² Two studies^11,13 noted the mechanism of injury. Associated injuries were noted by all except one study which excluded them.¹⁴ Only one study recorded the Body Mass Index (BMI).¹² We have classified any treatment method that used invasive means to achieve or hold the reduction as ‘Operative’.

Table 4.

Classification systems used in the papers reviewed.

Judet’s classification with Metaizeau Modificationa
Grade 1	Undisplaced
Grade 2	Angulation <30° and Translation <50%
Grade 3	Angulation 30-60° or Translation 50–100%
Grade 4a	Angulation 60-80°	Or Translation >100%
Grade 4b	Angulation >80°
O’Brien’s classification
Grade 1	Angulation <30°
Grade 2	Angulation 30-60°
Grade 3	Angulation >60°

^aMetaizeau subclassified Grade 4 into 4a/4b

3.3. Outcome measures

Four different outcome scales were used in the studies (Metaizeau,³ Tibone,¹⁷ Leung-Peterson¹⁸ and Mayo Elbow Performance Score - MEPS¹⁹) which have similarly named grades (Excellent to Poor) but define them very differently. Zimmerman et al. used their own definition to categorise ‘Successful’ or ‘Unsuccessful’ outcomes.¹²

3.4. Fracture severity

Three studies report outcomes in lower-grade injuries (Grade 1/2 injuries according to Judet’s classification corresponding to Angulation <30° with Translation<50%). Schmittenbecher et al. (36/36) and Tan & Mahadev (85/85) report excellent or good outcomes in all their patients, while Basmajian et al. report it in 81% (22/27).

In higher grade fractures (Grades 3/4 according to Judet and Grades 2/3 according to O’Brien- Angulation >30° or Translation>50%), most studies report worse outcomes. Schmittenbecher et al. and Guitterez-de la Iglesia et al. report excellent or good outcomes in 84% (pooled- 68/81) using the Metaizeau scale, while Tan & Mahadev and Basmajian et al. report it in 83% (19/23) and 56% (26/45), using the Tibone outcome scale.

It must be noted here that Basmajian et al. report worse outcomes than Tan & Mahadev using the same fracture classification (Modified Judet) and outcome measure (Tibone). This may be explained by a much shorter duration of follow up and suggests that results improve with time.

Increased fracture displacement was significantly associated with poorer outcomes in all the studies except De Mattos et al. Zimmerman et al. have reported a significant linear trend of worsening outcomes with more translation, but not with more angulation or higher O’Brien grade, in their regression analysis.

3.5. Operative versus non-operative treatment

Two studies have shown significantly worse results with operative treatment compared to non-operative treatment.^10,11 This observation is confounded by the higher likelihood of operative intervention in higher injury grades, and this is acknowledged in one of the papers.¹⁰

3.6. More versus less invasive treatment

All studies except one have reported significantly worse outcomes with the use of more invasive treatment. Guitterez-de la Iglesia et al. have found no correlation within their cohort of higher-grade fractures. Increasing fracture severity confounds the picture as it leads to a higher incidence of invasive treatment. This has been shown to be significant in three papers.^10,12,13 One of them¹⁰ also reported that Grade 4 injuries (Judet’s classification) had a significantly higher rate of open reduction. This confounding effect has not been addressed convincingly in any paper. It is not clear if Zimmerman et al. have factored the intervention used in their regression analysis.

3.7. Open versus percutaneous techniques

The rate of open reduction (OR) among patients treated operatively is 41% (126/287) calculated across studies. Guitterez-de la Iglesia et al. have noted that this increases from 7% in Judet Grade 3 (angulation 30-60°) to 85% in Grade 4b (>80°). Poorer outcomes with open reduction compared to percutaneous techniques have been reported significant by two studies.^10,11 Confounding by fracture grade has not been addressed convincingly. One paper¹³ has reported an analysis by grade of fracture and shown it to be significant in Grade 3 fractures (Judet’s classification), but not in grade 4 injuries.

3.8. Associated injuries

The incidence of associated injuries is 36% (pooled, 169/471) in 5 studies.9, 10, 11, 12, 13 Most studies report an incidence of around 40% (34–46%), but one of them¹³ reports a much lower rate at 19%. The commonest associated injury are fractures of the proximal Ulna including the Olecranon with a pooled frequency of 71% (105/147).

Schmittenbecher et al. note a higher incidence in higher-grade fractures (47% compared with 33%) but they have not analysed this for significance. One of the studies¹³ found a significant effect of associated injuries on the outcomes, while three studies10, 11, 12 did not, though confounders such as severity of injury or intervention used were not addressed by any.

3.9. Age

Four authors10, 11, 12, 13 found age more than 10 to be significantly associated with worse outcomes. Among confounding factors, older children have also been shown to sustain more severe fractures¹³ and undergo more invasive treatment.^11,12 Two studies have shown worse outcomes with increasing age while adjusting for severity of displacement.^12,13

3.10. Other factors

Zimmermann et al. did not find BMI to be a determinant of outcome. De Mattos et al. did not find an association between energy of injury and outcome.

3.11. Complications

We noticed variation in the reporting of complications among studies. The commonest complication reported in any of the series is stiffness (defined as >20° loss of range of motion (ROM) in any plane) in 31% patients by Basmajian et al. However, this is directly accounted for in their outcome scoring as a fair/poor result. Tan & Mahadev mentioned a relatively minor Cubitus Valgus (3–10°) in 12%, which counted towards a ‘Good’ result in their outcome assessment. Other studies report a total of 2 patients with this complication. Gutierrez-de la Iglesia et al. reported radiographic angulation of 10-30° as mal-union and reported a 19.6% incidence. However, they have accepted angulation up to 30° which probably explains this. They also report Radial head overgrowth in 31%, the functional effect of which is not clear.

Among more serious complications, Heterotopic Ossification (HO) was noted in 11 patients in 3 papers^10,11,13 (pooled incidence 2.2%), with 2 leading to complete arthrodesis.¹¹ In contrast Zimmerman et al. report an incidence of HO in 20% of their patients, which appears to be quite high, even accounting for the larger proportion of this cohort managed operatively. Avascular Necrosis (AVN) and Radio-Ulnar Synostosis had a low incidence of 1.3% (10/751) and 0.7% (6/751) respectively, across series. Transient neurological deficit following injury/intervention has been recorded to affect the Ulnar and Radial Nerves in 1.3% (10/751), with complete recovery. Premature physeal closure has been identified by two papers,^9,12 with a pooled incidence of 1.5% (9/592).

A significant association between complications and outcomes was not found after analysis in one of the papers.¹¹ Two studies^11,12 have found a significant association of complications with operative treatment. Tan & Mahadev have found ‘strong’ (but not significant) association between HO and associated injuries.

4. Discussion

Radial Neck Fractures have been reported in the literature to comprise 8–10% of all elbow fractures in children.1, 2, 3 In this review, De Mattos et al. reported it in 9.8% of all Elbow fractures from their data in 28,091 upper limb fractures in children.

These fractures occur around the growth plate, with half or more previously reported to be metaphyseal.^1,20 Among papers in our review, Schmittenbecher et al. have described 41 (62%) metaphyseal fractures and 24 (36%) Salter-Harris Type 1/2 injuries.

Jeffery²¹ postulated two mechanisms of isolated radial neck fractures-valgus injuries to the elbow leading to failure in compression and fractures sustained during injury or reduction of Elbow dislocations. Propagation of the valgus force can result in associated injuries such as oblique shear fracture of the Olecranon, Lateral Condyle fracture (compression), Medial Collateral ligament injury or avulsion of the Medial Epicondyle (tension). Associated Olecranon fractures can occur with elbow dislocations. Radial neck fractures can also result from extension forces in association with fractures of the proximal ulna (Monteggia pattern- Bado type 1²²) or distal Humerus. All of these associated injuries and co-incident upper limb fractures have been reported in the papers reviewed, with the commonest being fractures of the proximal Ulna (70%), with most involving the Olecranon. The pooled incidence of associated injuries at 36% was lower than the figure of 50% or more previously reported in the literature^5,17,20

Among outcome measures used in the included studies, the Mayo Elbow Performance Scale (MEPS) is the only one to factor in the ability to perform everyday tasks.¹⁹ Guitterez-de la Iglesia et al. have used both the purely ROM based Metaizeau scale and MEPS in their study. All five patients with an outcome graded ‘fair’, and one with outcome graded ‘good’ on the Metaizeau scale are graded ‘good’ and ‘excellent’ respectively on the MEPS, leading to all of their patients having a good or excellent result when MEPS was used. Thus, outcomes appear to be better with MEPS. We wonder whether this suggests that the ROM outcome doesn’t accurately represent functional outcome, or merely a lack of standardization. Either way, there is a definite need for a Patient/Carer Reported Outcome Measure (PROM) for elbow function validated in children, with the DASH/Quick-DASH being good candidates,²³ though needing development for this population.

The lower-grade (Judet 1/2) fractures have consistently good results with non-operative treatment.

Two papers,^9,13 which report excellent or good outcomes in all their patients with Judet Grade 1 and 2 fractures, have stated categorically that they have been treated by immobilization alone without manipulation. It must be clarified here that though Tan & Mahadev have made a recommendation in this regard to accept up to 45°, in line with some other authors,^24,25 they have reported only one patient with angulation between 30° and 45° treated non-operatively, making it difficult to accept their recommendation. While De Mattos et al. and Zimmerman et al. haven’t stated their view in this regard, we note some O’Brien Grade 1 fractures (<30° angulation) among their cohort treated by manipulation; this may have been due to a large translation, which is not considered in this classification. Schmittenbecher et al. have further specified that manipulation wasn’t attempted even when associated injuries needed surgery. Basmajian et al. differ in this regard but they also report smaller numbers and worse outcomes. Finally, there are no reports of further displacement in any lower-grade fractures managed without manipulation, requiring further intervention. Based on these observations, we suggest that Judet Grade 1/2 fractures (those with <50% translation and <30° angulation) can be safely managed with cast immobilization alone. It is probably best to avoid the temptation to improve the position or fix these fractures even when surgery is performed for associated injuries.

In higher-grade injuries, results are uniformly worse, though most papers^9,13,14 report excellent or good results in more than 75% patients.

Most published work to date, including papers we reviewed, consistently report poorer outcomes with open reduction (OR).⁵ There is unanimous agreement in the literature that OR should be reserved as a last resort after all closed methods, including percutaneous manipulation, have been exhausted.⁵ However, it is also established that higher grade fractures are more likely to require OR, and this has been confirmed statistically in some of the papers we considered.^10,14 We contend therefore that it is not clear if OR itself is the sole cause for poorer outcomes, or fracture severity contributes as well. A more accurate statement is that a ‘need for open reduction’ is associated with poorer results, and confounders need to be clarified by further research.

In terms of acceptability criteria for closed or percutaneous techniques, Guitterez-de la Iglesia and Basmajian et al. have explicitly stated it as <30° angulation, though they vary regarding translation (<30%/<50% respectively). We can only presume that Tan & Mahadev and Schmittenbecher et al. have also used similar criteria, as they have not manipulated fractures within these limits. It is postulated that angulation leads to a Varus deformity while translation leads to a cam-shaped radial head which impacts on rotational movements of the forearm.^5,26 These displacements have been separately considered by Zimmerman et al., who concluded that Translation, but not angulation, is independently predictive of poor outcomes, with >65% translation leading to unsuccessful outcomes in half the patients. They haven’t provided similar figures for >30% or >50% translation and have not made any suggestions about acceptability criteria. Thus, there is lack of clarity, but we suggest the criteria of residual angulation>30° or translation>50% or to consider open methods.

Several authors have recommended a graded application of successively more invasive interventions, and some have suggested the use of clinical algorithms.^5,12, 13, 14 While there is no direct evidence this pragmatic position finds broad acceptance. There is a controversy around repeating closed and percutaneous reduction attempts before moving on to open reduction, as it is thought that multiple attempts may damage the vascularity of the radial head and affect outcomes. Zimmerman et al. have not found a significant association between multiple attempts and outcomes, and this the only study in the literature to have looked at this using statistical methods.

Many percutaneous fixation methods are described, with the Metaizeau method (using retrograde intramedullary elastic nails)^3,5 and percutaneous trans-physeal K-wires^5,27 being the most widely used. As both these methods either traverse or impinge upon the growth plate, and patients are typically immobilized after application, there may not be much to choose between them. However, Metaizeau’s technique allows reduction in addition to fixation. Evans & Graham found better results in fractures treated by this method, but this was based on many small case-series of patients treated exclusively by this method. This arouses a suspicion of bias arising out of enthusiasm for a new technique at the time.⁵ Only one study¹² has statistically compared these techniques and found no difference regardless of the reduction method used. Accordingly, we suggest the use of either of these methods the surgeon is familiar with.

Zimmerman et al. have reported quite a high rate of loss of reduction (43%, 27/63) in fractures treated by closed reduction under anaesthesia without any fixation. All of these required re-operation with percutaneous fixation. No other authors reviewed have reported this, though they haven’t mentioned such high rates of closed reduction and immobilization, without fixation, either (42%, 63/147). While we are unsure whether this was because of more stringent reduction criteria, or the lack of fixation, it seems prudent to always employ a percutaneous fixation method when closed reduction is successful.

Higher age (defined as >10 years) is predictive of poorer outcomes, and Zimmerman et al. have shown it to be an independent predictor. It is not clear if an age >10 years warrants separate criteria for acceptability of reduction, however. On one hand, remodelling potential decreases as skeletal maturity approaches, with a higher risk of a residual deformity affecting forearm rotation. On the other hand, the contribution of the radial neck physis to longitudinal growth is minor. Importantly, more stringent criteria may lead to higher rates of open reduction and poorer results, proving to be counterproductive.

While a higher incidence of complications was associated with more invasive interventions,^11,12 we found a low rate of serious complications such as avascular necrosis and radio-ulnar synostosis. Other complications such as premature physeal closure have not been consistently reported, but the rate is similarly low. A few nerve injuries, both primary and iatrogenic have been reported but all were transient. Most series have reported low rates of heterotrophic ossification which is encouraging as this is a feared complication in a child’s elbow. The high incidence of this complication in Zimmerman’s work is unexplained, but it may be due to a broader definition based more on radiographic appearances. It is also not clear if it had any impact on outcomes in their series.

This review brings the literature up to date with regards to this rare injury using standardised methods and data from large cohort studies, all of which report on a broad spectrum of injury severity and management. In terms of sheer numbers, this review includes more than 700 patients, with no study reporting less than 50 cases. Finally, we have focussed on associations shown to be significant by statistical methods, and highlighted attempts to address confounders, or lack thereof.

There are also many limitations, in line with this being a rare injury. There are no randomized controlled trials, comparative cohort studies or case-control studies. All except one of the included studies are retrospective. Each paper used different grading systems and outcome measures making it difficult to pool data or compare findings. While all the authors report their observations using statistical methods, they have not addressed confounding factors convincingly. Only one paper has used a statistical model (multivariate regression analysis) that can address multiple confounders.¹²

More information is certainly required, focusing on areas identified in the text. Trials of any sort are likely to be very difficult given the rarity of the injury, the population in question and the association of other injuries. Probably, the way forward is a large multi-centre pediatric elbow trauma registry, recording demographic data, injury parameters and treatment methods in a pre-defined manner and registering outcomes at multiple time points using a standardized PROM validated for this purpose. Any analysis of data collected must use robust statistical methods to provide clear associations while addressing confounders.

5. Conclusion

Lower grade, stable fractures with <30° angulation and <50% translation, which constitute majority of pediatric radial neck fractures, are best treated with immobilization alone. Higher grade injuries should be approached pragmatically with the aim of achieving and holding reduction with the least invasive technique possible. Open reduction should be used as a last resort only if the fracture can’t be reduced to the limits above with closed or percutaneous methods. Further research, possibly based on pediatric fracture registries with standardized data collection, is needed to provide clear answers.

Authors contribution

Mr Sachin Kumar: Conceptualisation, Methodology, Investigation, Data Curation, Formal Analysis, Writing-original draft, review and editing. Mr Arya Mishra: Data Curation, Formal Analysis, Writing-original draft, review and editing. Mr Saurabh Odak: Conceptualisation, Investigation, Data Curation, Writing-review. Mr Jonathan Dwyer: Conceptualisation, Supervision, Investigation, Writing-review.

Declaration of competing interest

None of the authors have any conflicts to declare.