Between 2007 and 2023, 236 (302 physes) patients underwent guided growth surgery using a tension band plate for coronal plane angular deformity of the knee (distal femur and proximal tibia) at our centre. Data on 282 physes (193 patients) was included. To be included, all patients must have been followed for at least 2 years, to skeletal maturity (if this was less than 2 years) or until metalwork was removed; hence, 43 patients were excluded (Table 1). Patients included in our study had a diverse range of aetiologies but all had coronal plane angular deformity of the knee as measured by mechanical axis deviation on AP radiographs. The full range of diagnoses is shown in Table 2. Over 55% percent of patients were male and the mean age at surgery was 10.1 years (SD 3.5, range 1.7–17.3); the mean implant time was 16.2 months (SD 10.3, range 2–63); the mean follow-up time was 5.7 years (range 2.4–9.3).

Idiopathic deformity was the commonest patient cohort accounting for 114 knees (40.4%), (mean age of 11.9 years); Hurler’s syndrome (mucopolysaccharidosis type I) 38 knees (mean age 10.1 years) (13.5%); fracture related deformity in 16 knees (5.7%) (mean age 8.7 years). All other patient groups accounted for 12 or fewer knees.

The number of patients included and their diagnosis are represented in Table 2 and further subgrouped into 3 categories in Table 3.

In all, 227 knees (80.5%) were corrected to standard alignment with a mechanical axis between medial zone 1 to lateral zone 1. There was a total of 55 failures (19.5%) across all conditions where patients failed to achieve a mechanical axis to the physiological normal zone. The mean age at surgery of the patients who failed treatment was 9.8 years (SD 3.4, range 3.8–14.4). Table 4 shows the number of failures per condition.

For idiopathic angular deformity, which made up the most common aetiology in our cohort, there were 12 failures in 114 knees which equates to an 89.5% success rate. Hurler’s syndrome was the second largest patient cohort treated accounting for 6 failed treatments out of 38 (84.3% success). Fracture-related deformity had 3 failures in 16 patients (81.82% success) whilst in 12 patients with a diagnosis of spondyloepiphyseal dysplasia there were 3 failures (75% success). See Table 3 for complete results by condition.

For the purpose of logistic regression, all diagnoses were compiled into deformity attributed to three subgroups—‘idiopathic’, ‘congenital’ and ‘physeal disruption’. The congenital group was made up of patients with achondroplasia, pseudochondroplasia, cerebral palsy, arthrogryphosis, congenital fibular hemimelia, Ehlers-Danlos syndrome, Ellis-Van-Creveld syndrome, Greig syndrome, hemihypertrophy, Hurler’s syndrome, multiple hereditary exostosis, neurofibromatosis types 1 and 2, osteogenesis imperfecta, spina bifida, spondyloepiphyseal dysplasia, Stickler’s syndrome, Shwachman-Diamond syndrome, tuberous sclerosis and trisomy-21. The physeal disruption group was comprised of patients with coronal plane angular deformity following trauma or infection, after tumour, Juvenile Idiopathic arthritis, sickle-cell disease and also those with Rickets and Blount’s disease. Finally, the idiopathic group contained patients whose deformity could not be attributed to any underlying condition (one further patient with a diagnosis of transverse myelitis was added to the idiopathic group). Table 4 summarises treatment outcomes between the 3 deformity subgroups.

Logistic regression was performed to analyse the relationship between the three diagnosis subgroups and treatment success. It was found that the odds of treatment success decreased by 53% in congenital patients and by 76% in patients with physeal disruption when compared to patients in the idiopathic group. These results are illustrated in detail in Table 5.

Post-estimation analysis using marginal means revealed average predicted probability (of treatment success) scores of 89% (CI 83–94), 79% (CI 71–86) and 65% (CI 51–79) in idiopathic, congenital and physis disruption groups respectively. These results are summarised in Table 6 and a visual representation is illustrated in Fig. 2.

Radiographs of a patient pre-, during and post-coronal plane angular deformity correction

There were 32 complications documented at the final follow-up: hardware-related pain = 6; recurrences = 11; post-operative infection = 2; hardware failure = 8; overcorrection = 2; contracture deformity = 1. See below for detailed information regarding complications (Table 7).

Six patients had hardware-related pain following treatment: one patient with congenital fibular hemimelia, one with Hurler’s syndrome, one patient with trisomy 21, one patient with spondyloepiphyseal dysplasia and two patients with idiopathic angular deformity. One of the idiopathic knees required two steroid injections for their symptoms which resolved. One patient with spondyloepiphyseal dysplasia (failed treatment) required a release of tethered tissue which presented as pain at the plate site, whilst one patient with trisomy 21 reported pain whilst plates were in situ Fig. 3.

Average predicted probability of treatment success by diagnosis subgroup with 95% confidence intervals

Two-hundred eighty-two knees had an initial successful coronal plane deformity correction after their index procedure; however, 11 knees went on to develop a recurrence of the deformity. The highest number of recurrences occurred in knees with a fracture-related deformity, of which there were 3. There were two recurrences in knees with idiopathic deformity, congenital fibular hemimelia and Hurler’s syndrome respectively, whilst there was one recurrence in hereditary multiple exostosis and one in a patient with infection-related deformity. Of note, the recurrences in congenital fibular hemimelia were in the same patient who underwent 3 successful guided growth procedures of the same knee which recurred twice (Table 8).

There were 2 infections in the post-operative period (1 patient with a diagnosis of hereditary multiple exostosis and 1 with neurofibromatosis type 1 with anterolateral bowing both of which had successful guided growth surgery). Both had a superficial wound infection following routine removal of hardware and were treated successfully with a short course of antibiotics.

There were 8 documented hardware failures. Screw loosening occurred in 3 patients (2 with Hurler’s syndrome and 1 with neurofibromatosis type 1). Another Hurlers’ patient complained of hardware prominence. Among patients with idiopathic deformity, one patient suffered plate and screws backing out causing pain, whilst one needed a revision procedure for a loose screw. Another patient with neurofibromatosis type 1 suffered screw loosening whilst a further patient with spondyloepiphyseal dysplasia required a revision for screw loosening and another procedure to release tethered tissue.

One patient suffered a contracture deformity following failed treatment (Hurler’s syndrome), whilst two patients with idiopathic deformity were overcorrected. One patient with spondyloepiphyseal dysplasia developed knee instability following successful treatment requiring surgery.