Does sex influence outcome in ambulant children with bilateral spastic cerebral palsy?


  • See end of paper for list of abbreviations.

* Correspondence to first author at One Small Step Gait Laboratory, Guy’s and St Thomas’ NHS Foundation Trust, St Thomas Street, London SE1 9RT, UK.


To investigate the effect of sex on the phenotype of bilateral spastic cerebral palsy (CP) we reviewed the gait analysis data of 116 children (78 males, mean age 8y 1mo [SD 3y 1mo] and 38 females, mean age 8y 9mo [3y 1mo]) with bilateral spastic CP (Gross Motor Function Classification System [GMFCS] Levels I [four males, six females]; II [41 males, 19 females]; III [26 males, 12 females]; and IV [7 males, 1 female]) who had been referred for gait analysis to inform treatment. Although there were no differences between males and females in terms of gestational age, chronological age, or GMFCS level, males were more likely to have had nonoperative intervention before the referral (p=0.024), had a greater degree of knee flexion in stance phase when walking (p=0.003), and had a higher Gillette Gait Index (p<0.001) when compared with females. Males were also more likely to have surgery recommended on the basis of gait analysis (p<0.001). Sex seems to influence the development of the musculoskeletal system and mobility in ambulant children with bilateral spastic CP, and this may need to be considered when planning intervention or when assessing the outcome of intervention.

Although males have a higher incidence of cerebral palsy (CP) than females,1 sex is not generally thought of as a factor that may influence outcome. Sex does seem, however, to affect brain development in children born preterm,2 and males born extremely preterm seem to have a greater degree of cognitive impairment at the age of 6 years than females do3. Sex also seems to influence the development of CP in that male infants with birthweights at the upper and lower ends of the normal range seem to be more likely to develop severe CP.4 Females, in contrast, are reported to have a better outcome after preterm birth, which may be related to sex differences in metabolic rates, in the timing of lung maturation, and in the incidence of neonatal complications.5 Johnston and Hagberg, in a recent review of this topic,6 suggest that factors such as neurobiological development and the response to hypoxic–ischaemic injury may be influenced by sex and that sex thus contributes to the pathogenesis of CP.

Could sex, by means of its potential effects on the central nervous system in children with CP, influence musculoskeletal growth? If so, it may, as Johnson and Hagberg suggest, be a confounding variable in published and planned outcome studies of intervention in children with CP. It is particularly likely to be important in ambulant children with bilateral spastic CP, in whom a gradual deterioration in walking ability with growth has been reported.7,8

We hypothesize that sex does have an impact on outcome and mobility in children with bilateral spastic CP, with males being affected to a greater degree than females. This cross-sectional retrospective study looks at 116 ambulant children with bilateral spastic CP (78 males, 38 females) referred to our unit for three-dimensional gait analysis, and looks at the effect of sex on walking ability, previous nonoperative intervention, and subsequent treatment recommendations.


The One Small Step Gait Laboratory (Guy’s and St Thomas’ NHS Foundation Trust) is a specialist centre where ambulant children with CP are referred for clinical gait analysis to inform their management. We reviewed all of the children with a diagnosis of bilateral spastic CP who had been referred for gait analysis between 2003 and 2006 and who had not previously had surgical intervention to their lower limbs. The diagnosis was made in each case by the paediatrician or paediatric neurologist involved in the care of the child. Children who had previously had surgery were excluded because this could mask the natural history. The children referred were distributed equally throughout the region, and all were regularly reviewed by community paediatric physiotherapists and community paediatricians.

Each gait analysis was performed by the same team according to a standardized protocol, which included a clinical history and examination followed by the acquisition of video data, three-dimensional motion data, electromyographic data, and kinetic data (for independent ambulators) using a Vicon optical motion capture system (Vicon Motion Systems, Oxford, UK; Helen Hayes marker set). A minimum of four walking trials were recorded and assessed for each child. Children were encouraged to walk barefoot at a self-selected speed and used assistive devices for ambulation as needed. All of the treatment recommendations were made by the same orthopaedic surgeon (MG). The data recorded for each child included their chronological and gestational age, their level of home and community mobility (from which their Gross Motor Function Classification System9 [GMFCS] level was obtained), details of previous nonoperative management (if any), and the treatment recommendations made after gait analysis. Their popliteal angles (PAs) were reviewed to provide an indication of the degree of fixed deformity in their lower limbs. Because of interdependence between the joints of the lower limbs, individual joint kinematics were not selected for analysis, with the exception of minimum knee flexion in stance (MKFS). Instead, the Gillette Gait Index (GGI) was calculated for each child. The GGI uses multivariate analysis to derive a single dimensionless figure from 16 variables including motion data and normalized velocity, with a greater GGI value being interpreted as a greater variation from normal gait.10 The values for the able-bodied and reference groups in our calculations of the GGI were similar to published values. The GGI was used because it has been shown to correlate closely with other functional measures,11 seems to be independent of site,12 and has been recommended for use as a global outcome measure of gait analysis in children with CP.13 The motion data were reviewed and the GGI values were calculated independently by the second author (RS), who was not involved in the gait analysis or in the clinical management of the children in the study but was aware of the hypothesis of the study beforehand. We reviewed the MKFS because this is not specifically included in the GGI but seems to influence the support moment of the lower limbs significantly.14

Statistical analysis was performed with GraphPad Prism version 4.00 for Macintosh (GraphPad Software, San Diego California USA; An arbitrary level of 5% statistical significance (two-tailed) was assumed. The Shapiro–Wilkes test was used to assess whether individual data sets were taken from a normal distribution: groups were then compared by using an unpaired t test or Mann–Whitney U test as appropriate. Spearman’s test was used to investigate correlation, and relationships between data sets were assessed by using the odds ratio (OR) or logistic regression. The study was approved by the hospital research ethics committee and used anonymized data that had been collected during routine clinical management.


The study group consisted of 78 males and 38 females, giving a male:female ratio of 2.05:1. The groups are compared in Table I. The males were of similar chronological and gestational ages and had similar overall GMFCS levels (p=0.169, Mann–Whitney U test; Table II) when compared with the females. PAs were also similar in both groups. All of the children in the study received regular physiotherapy. Males were more likely to have had nonoperative intervention in addition to physiotherapy before being referred for gait analysis; the most frequently used nonoperative interventions in both groups are shown in Table III. The use of botulinum toxin A (BoNT-A) injections in isolation was similar in both groups, but males were much more likely to have had BoNT-A injections together with serial casting or the use of ankle–foot orthoses. The gait analysis data showed that males had a greater degree of MKFS (p=0.003, unpaired t test) and had greater GGI levels (p=0.002, Mann–Whitney U test). There was no significant correlation between gestational age and GGI, or between age and GGI, in either sex (Spearman’s test). Logistic regression analysis showed that MKFS (p=0.004) and GGI (p<0.001) were closely related to sex.

Table I.   Comparison of groups
  1. aInterquartile ranges shown; b95% confidence intervals shown; cMann–Whitney U test; dunpaired t test. PA, popliteal angle; MKFS, minimum knee flexion in stance; GGI, Gillette Gait Index.

Mean (SD) age, y:m8:1 (3:1)8:9 (3:1)0.342c
Median gestational age, weeksa31.5 (28–42)31.5 (28–42)0.59c
Median PA, degreesa68 (60–95)65 (57–80)0.066c
Mean MKFS, degreesb21.7 (18–25.4)15 (10.1–19.8)0.003d
Median GGIa982 (572–14 116)583 (261–853)<0.001c
Table II.   Gross Motor Function Classification System (GMFCS) levels in both groups
GMFCS levelMalesn=78 (%)Femalesn=38 (%)
I4 (5)6 (15)
II41 (53)19 (50)
III26 (33)12 (32)
IV7 (9)1 (3)
Table III.   Most common nonoperative interventions before gait analysis
GroupMalesn=78 (%)Femalesn=38 (%)Odds ratio(95% CI)
  1. BoNT-A, botulinum toxin A; AFOs, ankle–foot orthoses; CI, confidence interval.

No intervention14 (18)14 (37)2.67 (1.1–2.7)
BoNT-A only17 (22)8 (21)1.04 (0.4–2.7)
BoNT-A and casting/AFOs28 (36)4 (11)4.76 (1.5–14.8)

The recommendations made after gait analysis are shown in Table IV. Males were more likely to have surgery recommended after gait analysis, with an OR of 6.24 (95% confidence interval [CI] 2.1–18.2). The decision to recommend surgery was, however, significantly influenced by the PA (p<0.001, logistic regression), by MKFS (p<0.001, logistic regression), and by the GGI (p<0.001, logistic regression). The recommendations made were also influenced by the history of intervention before gait analysis, but the OR that males who had previous nonoperative intervention would have surgery recommended (12.4 [95% CI 2.0–76.9]) was greater than that for females (3.0 [0.74–12.1]). In those children for whom surgery was thought necessary, there was no significant difference between males and females in the number of children who were thought to be ready for surgery and those in whom lower limb muscle weakness was a concern (Table IV).

Table IV.   Treatment recommendations after gait analysis
GroupMalesn=78 (%)Femalesn=38 (%)Odds ratio(95% CI)
  1. CI, confidence interval.

Surgery not needed6 (8)13 (34)
Surgery recommended72 (92)25 (66)6.24 (2.1–18.2)
Ready for surgery58 (81)18 (72)
Not ready (weakness)14 (19)7 (28)1.61 (0.5–4.6)


Ambulant males with bilateral spastic CP are more likely to have had nonoperative intervention before gait analysis, more likely to have a greater MKFS and GGI, and more likely to have surgery recommended after gait analysis than females of similar gestational and chronological ages and GMFCS levels. This suggests that sex influences musculoskeletal growth and mobility in ambulant children with bilateral spastic CP.

There are limitations to the study. Children are likely to be referred for gait analysis because of a concern about their gait or level of mobility and because intervention is being considered. To assess the degree to which our results can be generalized, we need to know how representative the study group is of the general population of ambulant children with bilateral spastic CP in our area. Scrutton and colleagues reviewed all of the children with bilateral CP born in our area between 1989 and 1992 and have published their data as part of a study on the factors influencing the development of hip dysplasia.15 Their male:female ratio for children with bilateral spastic CP who were ambulant at the age of 5 years is 2.02:1. This is very close to our male:female ratio (2.05:1) and suggests that although this study may be biased towards children with a greater degree of involvement for whom gait analysis was thought to be needed, the ratio of males and females referred for analysis closely matches the population ratio, suggesting that referrals were not influenced by sex.

It is possible that females might have had surgery at an earlier age, and that although the numbers involved were small, the exclusion of children who have had previous lower limb surgery from the study may miss this group and thus bias the results. We have recently reviewed all of the ambulant children with bilateral spastic CP aged 8 years or less seen in our unit for whom surgical intervention to the lower limbs was recommended and for whom follow-up gait analysis was performed.16 Of the 24 children reviewed, 22 were males and two were females, suggesting that males are more likely to have surgery recommended at an early age because of the development of lower limb deformity that is unresponsive to nonoperative intervention. This is supported by the finding in the present study that males were more likely than females to have casting or orthoses used in addition to BoNT-A injections, again suggesting that their deformities were less responsive to nonoperative management.

Our treatment recommendations could potentially be biased by sex or by the history of a limited response to nonoperative intervention. The OR that surgery would be recommended for males who had previous nonoperative intervention was higher than for females, which suggests that the treatment recommendations are not explained fully by the clinical history. Logistic regression analysis showed that the recommendations after gait analysis were significantly influenced by the clinical examination and by the gait analysis findings. As the MKFS and GGI were more abnormal in males, they were therefore more likely to have surgery recommended.

Does this mean that young ambulant males with bilateral spastic CP have a poorer prognosis than females? This may be so, but there is an alternative possibility. The male:female ratio for all children with bilateral CP in the paper by Scrutton and colleagues15 was 1.44:1, but their male:female ratio for ambulant children at the age of 5 years was 2.02:1. It is possible that the Y chromosome in some way helps to maintain mobility in males with a greater level of impairment, resulting in an overrepresentation of males at higher levels of the GGI, with a resultant disproportionate degree of nonoperative and operative intervention when compared with females. It is of interest that Rose and colleagues17 did not note any sex difference in fractional anisotropy levels when reporting their findings on the association between alterations in the internal capsule of newborn infants and later gait abnormalities, but the numbers in their study with alterations in the microstructure of the internal capsule were limited, and this may have masked any differences related to sex.


This study suggests that nonoperative intervention may be less effective in ambulant males with bilateral spastic CP, possibly because of a greater tendency towards the development of deformity in males than in females. Sex may also affect the outcome of surgical intervention, but the variability of the GGI values in our study suggests that investigation of this will require a large and possibly multi-centre prospective study. It seems, however, that sex should be considered when assessing the natural history of deformity and mobility in children with bilateral spastic CP and their response to intervention.

List of abbreviations

Gillette Gait Index


Minimum knee flexion in stance


Popliteal angle