To our knowledge the current study is the first study to document how the PA and the hamstrings spasticity changes during childhood in a cohort of children with SBCP, GMFCS levels I, II, and III. The estimates from the statistical model showed that the PA was increasing during childhood, and that the GMFCS levels evolved with somewhat different patterns (Fig. 2). However, at 14 years of age no significant differences between the GMFCS levels were found. The hamstrings spasticity curves also changed during childhood, and as for the PA the patterns were different at the three GMFCS levels (Fig. 3). The spasticity curves increased throughout childhood, but were categorised as “a small increase in muscle tone” (mean < 1+) at all three GMFCS levels. At 10 years of age there was no significant difference in MAS scores between GMFCS levels II and III (Fig. 3).
For the PA curves there were 5° estimated differences between the GMFCS levels, and the significant differences seemed stable until the age of 8 years (Fig. 2 and Table 2). These findings are partly in line with previous published studies [1, 2]. At GMFCS level III the mean PA reached 41° (95% CI 39–43) at 8 years of age, which according to CPOP  indicates an increased PA, a need for more frequent follow-ups, and consideration of intervention initiatives. However, from the age of 12 years, the curve at GMFCS level III was levelling off. The curves at GMFCS levels I and II showed a slightly different slope, reaching 40° at 14 and 10 years respectively, and continuing to increase. The upper part of the CI band of GMFCS levels II and III were reaching 50° at about 14 years of age, which according to CPOP and others  indicates a need for treatment initiatives . At GMFCS level III, the CI band was wide in the highest age groups, overlapping the CI of both GMFCS levels I and II, indicating a statistical uncertainty, and no significant difference between the groups. This may be explained as a statistical artefact due to fewer children in the oldest age groups, especially at GMFCS level III. One explanation why the PA at GMFCS level III was levelling off may be seen in the hamstrings spasticity curves (Fig. 3), which were peaking at the age of 6 and then descending. This decreased spasticity after the age of 6 years may indicate a decreased risk of contracture . Seventy-two percent of the children at GMFCS level III had received BoNT-A, ITB, SDR, or surgery in the lower extremities during childhood (Table 3). In addition there were more interventions targeting the hamstrings at GMFCS level III compared to GMFCS levels I and II (Table 3), which may also have affected the shape of the curve. The separate analysis, excluding children who had undergone hamstrings tenotomy showed a decrease in both the MAS and PA curves at 6 and 9 years of age respectively, but only at GMFCS level III. No changes at GMFCS level I and II was probably due to few children who had undergone hamstrings tenotomy at these levels (Table 3). The hamstrings tenotomies were performed at a mean age of 8.6 years (±2.5). At the age of 9 years, the PA curve representing GMFCS level III, (Fig. 2) stopped to increase at about 43° and then levelled off, which may indicate that it was the children with the biggest PA who had received hamstrings surgery. Nordmark et al. , who studied all CP subgroups (spastic uni-, and bilateral CP, ataxia, and dyskinesia) of children with CP in Sweden also described a mean increase in the PA from 1 to14 years at all GMFCS levels. The curve estimates at GMFCS levels II and III were almost identical to our findings from about 5 to 10 years of age, however Nordmark did not describe a levelling or decrease of PA in the oldest age span at GMFCS level III as shown in the present study. They also performed additional analyses excluding the children (n = 6) who had undergone a hamstrings tenotomy, and the exclusion did not influence the PA curve. Reasons for the different findings in the two studies may be that they included all subgroups of CP and that fewer children had undergone a hamstrings tenotomy .
McDowell et al.  studied a sample of 178 children (4–17 years of age) with spastic CP and reported an increase in PA with increasing age and GMFCS level, with a higher PA in those having a bilateral involvement compared to those who were unilaterally affected. Compared to the findings in the present study, they found higher PAs at all GMFCS levels. One explanation might be that they excluded children who had undergone surgery the last year and those who had BoNT-A treatment the latest 6 months.
From 1 to 10 years of age the PA curves, GMFCS level I, showed CIs significantly narrower than GMFCS levels II and III (Fig. 2). One explanation might be that the level I group was the biggest group including 47% of the children, which might to a certain degree influence the distribution of the results. There was a continuous increase in the PA throughout childhood, reaching 41° at 14 years of age (Table 2, Fig. 2). This indicates that the PA in GMFCS level I increased despite of relatively good function, and a low spasticity level (Fig. 3). In comparison to studies reporting the PA in typically developing (TD) children, the children at GMFCS level I had a 10° to 15° higher PA. Mc Dowel et al.  reported a PA of 26° (±11) in TD children 4–10 years of age (n = 39) and 32° (±10) in TD children 11–17 years of age (n = 29). Moon et al.  reported in a group of TD adolescents 13–20 years old (n = 26) a PA of 34° (±10). Both studies showed increasing PAs with increasing ages, however, in the present study the change with age at GMFCS level I was more pronounced. As for the TD children, age seemed to be an important factor for the evolvement of the PA in walking children with CP. Rose et al.  followed 18 children with bilateral CP, mainly GMFCS levels I and II (mean age at inclusion 7.7 years) to evaluate the effect of time on their gait. The children performed 3D-gait analysis twice, the time intervals differed from 4.3–9.3 years. The results did not show any significant change in PA and flexed knee gait until the observation period reached at least 6 years. Rethlefsen et al.  studied 1005 gait records retrieved from ambulant children with CP. They reported that the odds for having excessive knee flexion in stands increased with increasing age at GMFCS level I, II and III, but only reaching significance at GMFCS level I.
In the present study the curve estimates for the hamstrings spasticity development the first 4 years were steep for all GMFCS levels (Fig. 3), which indicates a rapid change in spasticity during the first years of life, which is in line with previous published findings [6, 21]. However, the spasticity was significantly different in the three GMFCS levels. Compared to the PA curves (Fig. 2) the spasticity curves (Fig. 3) were steeper in the youngest age groups. The peak point of the spasticity curve at GMFCS level III was at the age of about 6 years, followed by a slightly decreasing curve. At the GMFCS levels I and II the curves continued to increase throughout the childhood, however the mean MAS was at the lower end of the spasticity scale (< 1+) at all GMFCS levels. Lindèn et al.  performed a register-based prospective cohort study including 4162 children with CP, between 0 and 15 years of age. The analyses included children treated with BoNT-A and oral baclofen and they also performed separate analyses for each GMFCS level. Additional analyses excluding ITB, SDR and Achilles tendon lengthening were performed and no change was found. They reported increased spasticity in the gastrocnemius-soleus muscle up to the age of 5 and thereafter a decreasing muscle tone up to 15 years of age in all CP subtypes. These findings correspond partly with our findings. Linden et al.  showed that the spasticity increased until the age of about 5 years for GMFSC level I, II and III and decreased until the age of 15 years. In the present study we found the same increasing tendency in hamstrings spasticity up to 5–6 years of age for GMFCS I, II and III. The spasticity pattern at GMFCS level III from the present study and Linden’s study showed almost the same pattern; the spasticity are decreasing from the age of about 5–6 years until 15 years of age. However, in the present study the spasticity at GMFCS levels I and II (Fig. 3), increased up to the age of 15 years, most pronounced at GMFCS level II. This indicates that the hamstrings spasticity at these two GMFCS levels seems to show a different longitudinal pattern compared to the pattern reported for the gastrocnemius-soleus muscle group .
Muscle contractures in CP has generally been associated with the presence of spasticity . Later research draws a more complex picture, also involving impairment of muscle growth and altered muscle adaptation . Hägglund and Wagner  found a relationship between spasticity in the gastrocnemius-soleus muscles and the development of contractures in the gastrocnemius-soleus muscle. In the present study, we also found that the PA curves (Fig. 2) and the hamstrings spasticity curves (Fig. 3) had quite identical shapes, especially at GMFCS levels II and III. At the GMFCS level I, the increasing spasticity after the age of 4 was modest (MAS < 1) (Fig. 3), however, the PA curve (Fig. 2) at the GMFCS level I had the highest increment of the three levels presented in this study (Fig. 3). This may indicate that there are additional factors than an increased stretch reflex registered as spasticity contributing to muscle contractures [14, 26, 40,41,42]. Reduced active terminal knee extension, either due to reduced selective motor control, muscle weakness or immobilization, may be contributing factors . In addition Gough and Shortland  suggested that in CP there might be multifactorial impairments of muscle growth which may lead to impaired muscle adaptation during growth . Recent published papers have also shown increased arrangement of collagen in the extra cellular matrix, and factors within the contractile elements in the muscles which may contribute to muscle contractures [14, 26, 40, 42].
In the present cohort a high rate of medical, surgical and physiotherapy interventions had been implemented (Table 3). The high frequency of procedures makes it difficult to assess the natural course of the PA. The reason for the changes is probably due to both natural growth and maturation and an effect of the interventions received during childhood, and in general, it is the most affected children who receive BoNT-A, ITB, SDR and orthopaedic surgery.
Previous studies evaluating reliability of goniometric joint measurements in children with CP [27, 28] have reported big measurement errors when measuring PA. To minimize the influence of confounding factors, and narrow the variability in the measurements, a written standardised protocol as well as trained assessors and, when possible, the same assessor over time has been underlined as important in CP . In the yearly routine for collecting data, written information was distributed from the CPOP , and the assessors were trained and experienced physiotherapists. However, to have the same assessor for each child over several years was not always possible. The big number of tests in the current study should limit the variability, but must be taken into account in the interpretation of the results.
The complexity of muscle spasticity makes it difficult to quantify. Several tests exist, however none of them seem to be superior, and validity and reliability are discussed . However, in clinical studies of neurological diseases and injuries the MAS  is the most frequently used instrument for assessing spasticity [21, 32]. The MAS added one score level (1+) at the lower end of the original Ashworth scale because the lower end of the spasticity score is more frequently seen in less involved children [18, 31]. This is in line with the results in the present study (Fig. 3).
There were some limitations to this study. The children were tested with different time intervals according to age and GMFCS levels. They were tested each year until the age of 6 years, and thereafter yearly at GMFCS levels II-III and every second year at GMFCS level I. This may indicate uncertainty in the oldest ages at GMFCS I. However, GMFCS level I is the biggest group, including 47% of the children in this cohort. There were fewer children in the oldest and youngest age groups, especially at GMFCS level III, making the confidence intervals wider and the estimates less reliable at this level (Figs. 2 and 3).