Journal of Pediatric Orthopaedics
23:292–295 © 2003 Lippincott Williams & Wilkins, Inc., Philadelphia
Gait Analysis Alters Decision-Making in Cerebral Palsy
Robert E. Cook, F.R.C.S.Orth., Ingo Schneider, M.D., M. Elizabeth Hazlewood, M.C.S.P.,
Susan J. Hillman, M.Sc., and James E. Robb, F.R.C.S.Ed.
Study conducted at the Anderson Gait Analysis Laboratory, Princess Margaret Rose Orthopaedic Hospital, Edinburgh, Scotland
Summary: This study was designed to assess the impact of
gait analysis on the treatment of patients with cerebral palsy.
One hundred two ambulant patients with cerebral palsy were
assessed clinically and with gait analysis. Separate treatment
proposals for each patient were recorded after clinical exami-
nation and after gait analysis. The results of the two methods of
assessment were compared. After clinical assessment, 71 of the
102 patients evaluated were recommended for a surgical pro-
cedure and 31 for nonoperative treatment. After gait analysis,
the indications for treatment were confirmed in 91 cases (89%).
Clinical assessment by the same orthopedic surgeon was in
There has been debate concerning the role of gait
analysis in the overall assessment of patients with cere-
bral palsy (2,6). Proponents believe it is a valuable tool
in the assessment of the dynamic and static functions of
the musculoskeletal system. Others argue that it is little
more than an expensive research tool of dubious clinical
benefit.
At least two previous studies have considered the in-
fluence of gait analysis on surgical decision-making
(1,4). The first, by DeLuca et al. (1), compared the sur-
gical recommendations for patients with cerebral palsy
assessed clinically and with video recordings before and
after gait analysis. Surgical recommendations were
changed in 52% of cases after gait analysis. The study
involved numerous assessors, and this may have been a
source of inconsistency. The second study, by Kay et al.
(4), assessed the impact of preoperative gait analysis on
the orthopaedic care of 97 pediatric patients with a va-
riety of diagnoses. Eleven different physicians referred
patients for gait analysis. The pre-gait analysis plan was
available in only 70 of these patients. The surgical treat-
ment plan was altered after gait analysis in 62 of these 70
patients (89%), and 39% of operations from the pre-gait
analysis plan were not performed.
Address correspondence and reprint requests to Mr. J. E. Robb,
F.R.C.S.Ed., Orthopaedic Department, Royal Hospital for Sick
Children, Edinburgh EH9 1LF, Scotland (e-mail: J.E.Robb@
btinternet.com).
From the Anderson Gait Analysis Laboratory, Princess Margaret
Rose Orthopaedic Hospital, Edinburgh, EH10 7ED, Scotland.
None of the authors received financial support for this study.
292
close agreement with gait analysis in identifying an indication
for surgery. There was less agreement in the type or level of
operation recommended. Gait analysis altered the decision in
106 of 267 operations (40%). There was good agreement for
bone surgery, suggesting that clinical evaluation of torsional
problems was fairly reliable. The poorer agreement seen for
soft tissue operations probably reflects the difficulties in as-
sessing tone-related problems in these patients clinically. This
study confirms the value of gait analysis for decision-making in
cerebral palsy. Key Words: Cerebral palsy—Decision mak-
ing—Gait analysis.
A further study by Skaggs et al. (5) demonstrated that
despite the objective nature of information from gait
analysis, there is some subjectivity in its interpretation.
The interobserver variability reported was similar to that
found for established classification systems of various
orthopaedic conditions. They also found differences be-
tween institutions in the interpretation of the data. To
obviate these interobserver differences, we compared de-
cision-making before and after gait analysis using the
same orthopaedic surgeon and the same staff in the gait
laboratory throughout.
MATERIALS AND METHODS
All new patients with cerebral palsy who were ambu-
lant and had disorders affecting their lower limbs were
seen and examined by the same orthopedic consultant. A
treatment plan was formulated on the basis of this con-
sultation. A decision was made as to the need for surgical
intervention or nonsurgical treatment with physiotherapy
or orthotics. If surgery was indicated, the level at which
this surgery should be performed was recorded. The out-
come and decisions made at this consultation will be
referred to as the “clinical assessment.”
After the clinical assessment, all patients were studied
using three-dimensional gait analysis. Gait analysis was
performed using a Vicon VX system (Oxford Metrics,
Oxford, UK), five infrared cameras, a Kistler force plate
(Kistler Instruments, AG Winthur, Switzerland), and
real-time coronal and sagittal plane video recording. The
results were analyzed by a team consisting of the same
 GAIT ANALYSIS IN CEREBRAL PALSY
TABLE 1. Operation levels of groups
Group Operation level
1 Hamstrings
2 Gastrocnemius
3 Femoral derotation osteotomy (FDRO)
4 Tibial derotation osteotomy (TDRO)
5 Psoas
6 Foot and ankle (see legend)
7 Hip adductors
8 Rectus femoris
Foot and ankle operations included tibialis anterior tendon transfer,
lengthening of tibialis posterior, flexor digitorum superficialis or flexor
hallucis longus, subtalar fusion, and calcaneal osteotomy
orthopedic surgeon, a research fellow in gait analysis, a
senior physiotherapist, and a bioengineer. All had exten-
sive experience in gait analysis. Management recommen-
dations were recorded for each patient after gait analysis,
including the need for operative or nonoperative treat-
ment and the level and type of operation that should be
performed.
The recommendations from gait analysis were com-
pared with those from the clinical assessment by an in-
dependent observer, and the results were analyzed. The
frequency of operative or nonoperative recommenda-
tions from each assessment method was calculated first.
Subsequently, the operative procedures recommended
for each patient were compared. Statistical comparisons
between groups were made using the chi-square test for
nonparametric data.
To simplify the data, the operations were grouped into
eight categories (Table 1). These groups consisted of
muscles with the same function or osteotomies of the
femur or tibia. No differentiation was made as to the
exact operation to be performed on any particular
muscle; for example, no distinction was made between
an intratendinous and intramuscular lengthening.
RESULTS
The study population consisted of 105 patients. Three
were subsequently excluded because a complete set of
293
clinical assessment results and gait recommendations
could not be found. All 102 patients had cerebral palsy
and were ambulant. After clinical assessment, 71 were
thought to require some form of surgical intervention,
whereas 31 were deemed appropriate for nonoperative
treatment. After gait analysis there was agreement in 91
cases (89%), 61 of the 71 in whom surgery was origi-
nally proposed and 30 of the 31 not requiring surgery.
Gait analysis, therefore, did not agree that surgical inter-
vention was required in 10 patients identified on clinical
assessment; in one patient in whom no surgery was ini-
tially proposed, gait analysis suggested that surgical in-
tervention would be of some benefit.
There was far less agreement between the two meth-
ods of assessment in the actual level at which the surgical
procedure should be performed. Because of the multi-
level abnormalities in most of the patients studied, many
patients had more than one operation proposed. A total of
267 operations were proposed after the clinical assess-
ment and the gait analysis. After clinical assessment, 215
procedures were proposed in 71 patients, approximately
three operations per patient. After gait analysis, 213 pro-
cedures in 62 patients were recommended. There was
agreement between gait analysis and the clinical assess-
ment for 161 of these operations. A further 54 operations
were proposed by clinical assessment but not supported
by gait analysis, whereas gait analysis proposed a further
52 operations not identified by clinical assessment.
Overall, there was agreement between the clinical as-
sessments and gait analysis recommendations in 60% of
procedures. The results for each group or level of surgery
are presented in Figure 1. The level of agreement be-
tween the two assessment techniques is shown in the left
column for each group; the other two columns show the
number of times clinical assessment or gait analysis rec-
ommended operations in isolation. There was greater
agreement between clinical assessment and gait analysis
for operations involving bone surgery (femoral or tibial
derotation osteotomy; agreement in 47 of 61 procedures
[77%]) compared with soft tissue surgery (muscle or
tendon lengthening or transfer; agreement in 114 of 206
FIG. 1. Results for each of the
eight surgical sites. The level of
agreement between clinical as-
sessment and gait analysis is
shown in the left column for each
group. The other two columns
show the number of times clinical
assessment or gait analysis rec-
ommended operations in isolation.
The vertical axis depicts the num-
ber of procedures. FDRO, femoral
derotation osteotomy; TDRO, tibial
derotation osteotomy.
J Pediatr Orthop, Vol. 23, No. 3, 2003
294 R. E. COOK ET AL.
FIG. 2. Results for osteotomy versus soft tissue surgery.
procedures [55%]) (Fig. 2). The difference between these
groups was statistically highly significant using the chi-
square test (P < 0.01). Agreement was also high for the
hamstrings (69%) and gastrocnemius (64%). It was not
as good for the hip adductors (48%), psoas (44%), foot
and ankle (36%), and rectus femoris (33%).
DISCUSSION
We have tried to recreate the typical clinical scenario
in which patients are examined by an orthopedic surgeon
and then treated using the additional information from
gait analysis. DeLuca et al. (1) used clinical examination
and videotape recordings of gait as part of their clinical
assessment. We believe that video recording of gait is an
integral part of gait analysis, but it is not usually avail-
able to clinicians in the outpatient setting. They excluded
analysis of specific foot and ankle deformities, including
equinovarus and pes planovalgus. We believe that these
patients are among those most likely to benefit from gait
analysis, because the dynamics of the foot are difficult to
assess clinically. For this reason they were included in
our study.
By using the same set of personnel in the assessment
of gait analysis, we removed the chance of interobserver
variation, which is clearly possible if numerous observ-
ers are involved, however experienced they may be.
Agreement between the team members in the gait labo-
ratory was gained after careful discussion of each case.
After gait analysis, there was agreement for surgery in
61 of the 71 (86%) patients identified from clinical as-
sessment, whereas there was agreement in 30 of 31
(97%) identified by clinical assessment for nonoperative
treatment. Overall, the results of gait analysis led to a
reduction in the number of patients recommended for
surgery. In addition, surgical decision-making was sub-
stantially altered by the use of gait analysis: the recom-
mendations for the level or type of surgery to be per-
formed were altered in 40% of procedures.
There was good agreement between the two methods
of assessment for bone surgery, suggesting that clinical
evaluation of torsional problems was fairly reliable in
J Pediatr Orthop, Vol. 23, No. 3, 2003
this group of patients. The poorer agreement seen for soft
tissue operations probably reflects the added difficulties
of assessing tone-related problems clinically in these pa-
tients. Agreement for the hamstrings and gastrocnemius
was greater than for other muscle groups, perhaps be-
cause their length facilitates clinical assessment better
than shorter muscle groups. Interestingly, the study by
Skaggs et al. (4) showed greater agreement between
physicians in the identification of soft tissue rather than
bony problems; this may reflect the interinstitutional
differences in interpretation of the data seen in their
article.
When considering the requirement for derotational os-
teotomy of the intertrochanteric region of the femur
(FDRO), the senior author (J.E.R.) performing the clini-
cal assessments did not routinely specify additional sur-
gery on the psoas muscle, because derotation could give
a gain in functional psoas length. The psoas muscle was
often identified as requiring further operations in addi-
tion to an FDRO on the basis of gait analysis when this
had not been identified in the clinical assessment. Six of
the 11 patients identified as requiring psoas lengthening,
in addition to an FDRO, from gait analysis, had been
listed for only an FDRO on clinical assessment. This
anomaly in the data collection would improve the con-
cordance rate for psoas to 20 of 27 (74%).
Table 2 summarizes the main differences between the
results of our article and those from DeLuca et al. (1) and
Kay et al. (4). DeLuca et al. (1) suggested that gait analy-
sis led to an increased rate of surgery on the gastrocne-
mius muscle and discussed their aggressive policy with
respect to this muscle based on previous kinematic and
kinetic data. We have shown the opposite. This may
reflect the different treatment protocols of different in-
stitutions. They also reported that gait analysis led to a
reduction in the rate of psoas surgery recommended. We
have shown an increase, probably caused by the anomaly
discussed.
Assessing rectus femoris activity is difficult on clini-
cal examination alone, and electromyographic data of
muscle contraction are often required for a more com-
plete assessment before considering a rectus transfer (3).
The operations on the foot and ankle are relatively dis-
parate. We believe that this is a demanding area for di-
agnosis and operative planning that would be greatly
assisted by the freeze-frame video recording performed
as part of gait analysis.
TABLE 2. Results in comparison with other articles
De Luca Kay Present
et al. et al. series
Hamstrings Less Less Less
Gastrocnemius More More Less
Femoral derotation osteotomy
(FDRO) Less Less Less
Tibial derotation osteotomy
(TDRO) Less Less Less
Psoas Less More More
Hip adductors Less Less Less
Rectus femoris More Less More
 GAIT ANALYSIS IN CEREBRAL PALSY
CONCLUSIONS
The results of this study show that gait analysis would
substantially alter the surgical management of many of
our patients and confirm the value of gait analysis for
decision-making in cerebral palsy. The two assessment
methods agreed in identifying which patients required
surgery. This study has not shown that the outcome after
treatment was necessarily changed by gait analysis, but
we believe that the assimilation of extra information has
been of benefit in fine-tuning the treatment of our pa-
tients to ensure they receive the correct operation at the
appropriate level. Randomized controlled studies are re-
quired to evaluate further the benefits that gait analysis
may have in the management of these patients.
295
REFERENCES
1. DeLuca PA, Davis RB, Õunpuu S, et al. Alterations in surgical
decision making in patients with cerebral palsy based on three-
dimensional gait analysis. J Pediatr Orthop 1997;17:608–614.
2. Gage JR. The role of gait analysis in the treatment of cerebral
palsy. J Pediatr Orthop 1994;14:701–702.
3. Gage JR, Perry J, Hicks RR, et al. Rectus femoris transfer to
improve knee function of children with cerebral palsy. Dev Med
Child Neurol 1987;29:159–166.
4. Kay RM, Dennis S, Rethlefsen S, et al. The effect of preoperative
gait analysis on orthopaedic decision making. Clin Orthop 2000;
372:217–222.
5. Skaggs DL, Rethlefsen SA, Kay RM, et al. Variability in gait
analysis interpretation. J Pediatr Orthop 2000;20:759–764.
6. Watts HG. Gait laboratory analysis for preoperative decision mak-
ing in spastic cerebral palsy: is it all it’s cracked up to be? J Pediatr
Orthop 1994;14:703–704.
J Pediatr Orthop, Vol. 23, No. 3, 2003