Hawamdeh et al. [26] reported a study of the effects of three successive injections (intervals of 3–4 months) of botulinum toxin A (either Botox or Dysport) on calf muscle tone (modified Ashworth score (MAS)), passive ankle dorsiflexion range and gross motor function (GMF), at 3 months and 18 months (after last injection) in children with spastic diplegia, aged 3–15 years. Patients were excluded from the trial if they had fixed contractures or deformities, ‘previous surgery to the lower limb at least 1 year before the start of the study’, previous botulinum toxin treatment, treatment with phenol, intrathecal baclofen or other antispasticity agent. There were 60 patients: 40 in the treatment group, and 20 in the control group. At 18 months, the study reported the treatment group to have (i) better muscle tone (modified Ashworth score), involving a manoeuvre of questionable validity to convert categorical to numerical data, (ii) a statistically significant (but clinically insubstantial) improvement in passive ankle dorsiflexion range (12 vs 15 degrees), and (iii) an increase in median gross motor function from grade 4 “reasonably useful non-ambulant locomotion and/or walking when assisted” to 5 “walking with a walking aid” (both control and treatment groups improved from a baseline of grade 3 – it is not clear how many more patients improved by how much more to result in the observed difference in median GMF score).

The authors alleged the study to be randomized and single-blind. This study was not placebo-controlled, so patients and parents were aware of the treatment received; the two groups had their ancillary physical treatments under different physiotherapists on different days. It is questionable therefore whether physiotherapists would be ‘blind’ to the nature of treatment of their group (as alleged). Furthermore “All outcome measurements were performed by an experienced senior paediatric physiotherapist who knew whether a child belonged to the treatment or the control group”. It would appear therefore that the study was not blinded as to either participants or assessor, and should not be classed as ‘single blind’ by the conventional meaning for this term. Additionally, there is concern about the mode of local randomization ‘by flipping a coin’, apparently dividing subjects into two homogeneous groups: treatment group (40) and control group (20). This study is subject to major bias, sufficient to downgrade the level of evidence accorded, and the results should be treated/interpreted with major reservation.

Moore et al. [28] reported a single-center randomized, placebo-controlled, parallel group, double-blind trial of the effect of botulinum toxin A (Dysport) injection (repeated 3 monthly, if clinically indicated; i.e. 8 injection cycles) on motor function (Gross Motor Function Measure (GMFM) and Paediatric Evaluation of Disability Index (PEDI)) at 2 years (a ‘trough’ rather than ‘peak’ effect), on children with cerebral palsy aged 2–6 years (or 2–8 years, if targeting the hamstring muscles), with clinically significant spasticity of one or both lower limbs, and excessive involuntary muscle activity. Patients were excluded from the trial if they had ‘a disease, handicap or situation likely to make treatment too difficult or dangerous or to invalidate or prevent assessments’, or had been treated with botulinum toxin previously. There were 64 patients, with 32 in each group (placebo and botulinum toxin). There was no difference in adverse events between the two groups. The study found no evidence of cumulative or persisting benefit (mean change scores for both GMFM and PEDI, for either total or subscale values) from treatment at either 1 or 2 years. They commented that (i) the ceiling effects of the scales used may have limited responsiveness, and (ii) that ‘the small numbers may have caused a type 2 error although, if anything, the trends suggest that placebo is better’.

Tedroff et al. [31] reported a single center study on the effect of botulinum toxin A (Botox – two injections, 6 months apart, in gastrocnemius) on muscle tone (modified Ashworth score), development of contractures and gait (gait analysis, GMFM-66 and PEDI) in 15 young children (six in the botox treatment group, and nine in the control group; mean age 16 months) with spastic cerebral palsy (either unilateral or bilateral), after 1 and 3.5 years. The treatment phase was initiated when children were ‘pulling to stand’. Both groups received a daily stretching programme. Unfortunately, the groups allegedly contained heterogeneous GMFCS proportionality at the start of the study (the botox group: GMFCS I (4), GMFCS II (2); the control group: GMFCS I (4), GMFCS II (4), GMFCS III (1)), though GMFCS is hard to apportion at <2 years age. Furthermore, although the baseline characteristics of the two groups were held as showing no statistical significant difference except for plantarflexor tone, there were differences in mean values (eg ankle dorsal extension 10 degrees in the botulinum group and 17.8 degrees in the control group) with large standard deviations (unsurprisingly given the small numbers in each group). The authors hold that plantarflexor tone was significantly reduced after 3.5 years in the treatment group. Furthermore, they found the ‘change-score in knee flexion muscle tone’ to be significantly different between the two groups after 3.5 years. There was no significant difference between the groups in (i) changes in ankle dorsiflexion range at any time, (ii) gait analysis, (iii) GMFM-66, or (iv) PEDI. The limitations of this study are profound, including small sample size, initial assessment of subjects, heterogeneity of patients within groups, differences in baseline characteristics between groups, lack of blinding of either treatment or assessment (except 3D-gait analysis). These concerns, together with the difficulty in deriving meaningful values from the presented data (much is presented as ‘change from baseline’ or ‘effect sizes’, themselves “calculated by dividing the absolute mean difference between the groups by the SD for the control group at the baseline”), largely negate any value of the study.

Graham et al. [27] reported a multicenter (5 centers) randomized controlled trial of the effect of 6 monthly intramuscular botulinum toxin A injection (adductors and hamstrings) and SWASH hip abduction orthosis (6 h per day) on hip displacement (Reimers’ migration percentage (RMP) on radiography, 6 monthly) over 3 years, on children with bilateral spastic cerebral palsy (including hip adductor reactivity), aged 1–5 years, and with hips at risk (RMP 10–40 %). Patients were excluded from the trial if they had pseudobulbar palsy, previous hip surgery, hip flexion contracture > 30 degrees, or scoliosis with Cobb angle > 20 degrees. There were 91 patients, at a 98 % recruitment rate; 44 and 47 patients were allocated to control and treatment groups, respectively. Of note, there were 12 and 33 adverse events (major and minor, respectively) after 204 injection episodes (incidences of 6 and 16 %). Two modes of analysis were used to compare the treatment and control groups – one showing no statistically significant effect, the other suggesting a small decrease in the rate of displacement for the treatment group (3.1 % unweighted; 1.4 % weighted), indicative of a possible delay in the time to surgery, but not a difference on the number requiring surgery ultimately. The authors summarized: “There was no evidence of a beneficial effect of treatment that was clinically important.”

Willoughby et al. [30] reported the long-term outcome data (hip morphology and surgery requirements) for 46 patients (23 in each of the two groups – control and treatment (3 year programme botulinum toxin and hip abduction orthosis); 31 male, 15 female; 36 quadriplegic, 10 diplegic; mean follow-up after recruitment was 10 years and 10 months) from the major single centre in the trial outlined above [27]. There was no significant difference between the treatment and control groups in ultimate mean migration percentage, hip morphology (Melbourne Cerebral Palsy Hip Classification System (MCPHCS)), need for preventive surgery (21 ex 23, and 19 ex 23 in treatment and control groups respectively) and reconstructive surgery (10 ex 23, 8 ex 23). Furthermore, for children requiring surgery, ‘there was no statistically significant difference between the groups in mean age at either preventive or reconstructive surgery.’ i.e. there was no long-term benefit from the treatment, either in terms of reducing/delaying the need for surgery or improving later hip morphology.