The Observational Gait Scale Can Help Determine the GMFCS Level in Children With Cerebral Palsy : Pediatric Physical Therapy

Secondary Logo

Journal Logo


The Observational Gait Scale Can Help Determine the GMFCS Level in Children With Cerebral Palsy

Zapata, Karina A. DPT, PhD; Rushing, Charter L. PT, ScD; Delgado, Mauricio R. MD; Jo, Chanhee PhD

Author Information
Pediatric Physical Therapy 34(1):p 23-26, January 2022. | DOI: 10.1097/PEP.0000000000000851
  • Free


The Observational Gait Scale (OGS) is commonly used in clinic as an attempt to objectively rate gait in children with cerebral palsy (CP). The OGS evolved from the Physician's Rating Scale as a simple, low-technology tool to examine gait in 2 dimensions.1 Common reasons for performing the OGS include assessing for changes before and after botulinum neurotoxin type A injections, serial casting, orthotic use, surgery, and as part of yearly follow-up visits.

Several variations of the OGS exist. The OGS developed at our institution and published by Boyd has 8 subscores to evaluate gait parameters in the following order: knee position in midstance, initial foot contact, foot contact at midstance, timing of heel rise, hindfoot at midstance, base of support, assistive devices, and whether there has been a change.2 Lower OGS scores indicate more gait deviations. Boyd's OGS version was found to have acceptable interrater reliability (weighted kappa [wk] range 0.43-0.86) and intrarater reliability (wk range 0.53-0.91) and validity (wk range 0.38-0.94) for subscores 1 to 4 and poor reliability for subscores 5 to 6. Only subscores 1 to 4 were assessed for validity compared with 3-dimensional gait analysis. Interrater reliability was provided for subscores 1 to 7, but not for the total mean.3 We have used a modified version of the OGS consisting of 7 subscores and a total score of 19 points (Figure 1). Our subscores differ from Boyd's in that the version used in this study does not assess “timing of heel rise” or “change” and has a “speed of gait” subscore. In our clinical experience, the “change” subscore is open to varying interpretation, and should be captured by the other gait parameters. A third published version of the OGS consists of 24 subscores and a total score that was not reported.4,5

Fig. 1.:
Observational Gait Scale.

The Gross Motor Function Classification System (GMFCS) is a 5-level functional classification that differentiates children and youth with CP according to their current gross motor abilities and limitations, and need for assistive technology including wheeled mobility.6 Those with GMFCS level I can generally walk without restrictions but tend to have some limitations in more challenging situations. Those classified as level V are usually very limited in their ability to move themselves about, even with assistive technology.

In the past, the GMFCS has been found to be mildly related to gait speed, stride length, and cadence.7 However, we are unaware of any studies that have evaluated whether the OGS can assist in assigning the GMFCS level according to pairwise comparisons and post hoc analysis. We aimed to evaluate the association between the OGS and the GMFCS in walking children with CP according to Spearman correlations.



The study design is a single-center retrospective study of data collected and entered into a clinical database with a remote intent to use the data over time. Patients and caregivers provided a blanket assent and consent for use of their information in the database. A research coordinator entered clinical data obtained from a childhood movement disorder clinic at a tertiary care hospital into this institution's movement disorder database following each clinic visit. Entered data included participants' names, date of birth, date seen, gender, height, weight, diagnosis, type of CP, GMFCS level, and OGS scores. A research coordinator extracted previously entered data from a consecutive series of 533 walking participants with CP (GMFCS levels I-IV) presenting to the specialty clinic who had either GMFCS levels or OGS scores recorded between 1997 and 2016. After data were exported from the database, the research coordinator, first author, and statistician reviewed the data of each participant to ensure accuracy. Data from the exported files were confirmed for accuracy through comparison with the childhood motor disorders database and the child's medical record. Twenty-one children were excluded due to missing recorded OGS scores or GMFCS levels, resulting in a total of 512 participants for inclusion in the study. Ethical approval was obtained from the hospital and university institutional review boards. Participants' caregivers provided written informed consent and participants older than 10 years also provided assent for study participation.

Children's OGS score and GMFCS level were both recorded at the same initial clinic visit at baseline, prior to any intervention. Each child was instructed to walk barefoot with the knees in view back and forth along a 6-m walkway at least 6 times at a self-selected speed. A physical therapist (PT) viewed the child's gait 3 times in the frontal plane and 3 times in the lateral plane. Children with bilateral involvement will have 2 OGS scores reflecting their left and right sides. When there were 2 OGS total scores, the lower OGS total score was included for data analysis. The same PT was involved in rating both the OGS and GMFCS level. Each child's GMFCS level was determined by the treating physician who consulted with the multidisciplinary study team, including the PT, family, and medical team. All OGS scores and GMFCS levels used in this study were recorded from the same initial clinic visit at 1 visit at 1 time point.

Each child's OGS score was determined by any of 8 PTs trained in neuromuscular evaluation. Since we did not formally establish interrater reliability for the OGS, we reviewed the OGS total scores for each PT tester to evaluate whether scores were similar across testers after reviewing the exported data. After matching for the GMFCS level and type of CP, we used a subgroup of 9 participants to compare 8 testers. All OGS scores across testers were within a similar range with no significant differences between OGS scores across the testers (P = .69).

Statistical Analysis

Means and standard deviations were used to describe continuous variables and a Student's t test to compare those variables. Spearman correlations along with box plots were used to examine the relationship between the OGS scores (total and sub scores) and GMFCS levels. A 1-way analysis of variance was used to assess for differences in OGS total scores across GMFCS levels, followed by Tukey's multiple comparisons. We ran multivariate regression analyses where OGS was the dependent variable and the covariates were age, gender, body mass index, and GMFCS. Statistical significance was set at a P value less than .05. Statistical analyses were performed using SAS/STAT version 9.4 (SAS Institute, Cary, North Carolina) and IBM SPSS 26 Software (IBM Corp, Armonk, New York).


Participants averaged 4 years 7 months (1-18 years) (Table 1). The most common type of CP was bilateral (65%) (Table 1). Participant distribution of GMFCS levels was GMFCS I = 47%, GMFCS II = 28%, GMFCS III = 20%, and GMFCS IV = 5%.

TABLE 1 - Participant Characteristics (n = 512)
Parameters Mean (SD) Range
Age, y 4.6 (2.7) 1.3-18.0
Girls 278
Boys 234
Height, cm 105 (16) 77-171
Weight, kg 18.2 (8.0) 8.6-73.4
BMI, kg/m2 16.2 (2.4) 11.3-34.1
Type of cerebral palsy
Bilateral 331
Unilateral 181
Abbreviations: BMI, body mass index; SD, standard deviation.

The OGS scores decreased as GMFCS levels increased (Table 2 and Figure 2). A significant negative relationship was seen between the OGS total scores and GMFCS levels according to a Spearman correlation (r = −0.61; P < .001). In particular, each GMFCS level was different based on the OGS total score across all levels in a pairwise comparison (P < .001) (Table 2). Each GMFCS level was different based on the OGS total score across all levels according to post hoc analysis (P < .05) (Table 2). In addition, multivariate modeling analysis confirmed that the association between the OGS total score and the GMFCS was still valid after adjusting for age and gender (P < .001). In particular, the OGS total score did not differ according to gender, age, or body mass index. Furthermore, when the OGS total score was compared across GMFCS levels according to 2 age groups (≤5 years old vs >5 years old), no significant differences were found according to t tests. Finally, when the 7 subscores of the OGS were compared across GMFCS levels, a weak to moderate negative association was seen (r = −0.14 to −0.70; P < .01 for all subscores) according to Spearman correlations (Table 3).

Fig. 2.:
Box plot with actual values of OGS total scores across GMFCS levels. GMFCS indicates Gross Motor Function Classification System; OGS, Observational Gait Scale.
TABLE 2 - Observational Gait Scale Total Score Mean Across Gross Motor Function Classification System Levela
GMFCS Level OGS Total Score
Overall Bilateral Unilateral
n Mean (SD) Range n Mean (SD) Range n Mean (SD) Range
I 240 13.1 (2.8) 4-19 111 12.7 (2.9) 4-19 129 13.5 (2.6) 6-18
II 143 11.4 (2.7) 5-18 91 11.1 (2.7) 5-17 52 11.8 (2.7) 5-18
III 103 7.7 (2.7) 1-16 103 7.7 (2.7) 1-16
IV 26 6.1 (2.0) 2-9 26 6.1 (2.0) 2-9
P value <.001 <.001 <.001
Multiple comparisons
I vs II <.001 <.001 <.001
II vs III <.001 <.001
III vs IV .030 .031
Abbreviations: GMFCS, Gross Motor Function Classification System; OGS, Observational Gait Scale; SD, standard deviation.
aThe lower total OGS score was included.

TABLE 3 - Observational Gait Scale Subscore Associations Across Gross Motor Function Classification System Levels (n = 415)
OGS Subscore GMFCS Level
Spearman Correlation (r) P Value
Initial foot contacta −0.19 <.001
Foot at midstancea −0.26 <.001
Knee position at midstancea −0.21 <.001
Base of supporta −0.48 <.001
Hindfoot at midstancea −0.14 <.01
Speed of gaita −0.47 <.001
Assistive devicea −0.70 <.001
Abbreviations: GMFCS, Gross Motor Function Classification System; OGS, Observational Gait Scale.
aStatistical significance (P < .05).

A subgroup analysis of participants with bilateral CP (diparesis, quadriparesis) and unilateral CP (hemiparesis) also demonstrated that the OGS total score decreased as the GMFCS increased (Table 2). Participants with GMFCS I with bilateral CP demonstrated significantly lower OGS total scores compared with participants with GMFCS I with unilateral CP (P < .01) according to t tests, but these were not statistically significant for participants with GMFCS II with bilateral CP compared to participants with GMFCS II with unilateral CP (P = .15). Multivariate modeling analysis revealed that each GMFCS level was different based on the OGS total score across all levels (P < .05) (Table 2).


At our institution, we use the OGS as a tool to assess a child's gait and response to treatment at this institution since it is quick, convenient, and easy to learn. The OGS does not substantially lengthen a child's clinic visit or contribute to testing fatigue. The OGS can be performed in any space without requiring any special equipment. Although 3-dimensional gait analysis is the gold standard for gait, it is unavailable and impractical for most clinical settings.8 Components of the OGS have been found to be reliable and valid overall, but the reliability and validity of the total score have not been reported.3 Not only does this study include OGS total score findings, but we also add to the evidence supporting the validity of the OGS by finding that the OGS total score and GMFCS levels measure related constructs according to Spearman correlations.

Occasionally, discussions arise about a child's GMFCS level when it falls on the border of 2 GMFCS age bands and descriptors. For example, it can be difficult to define whether a child who just turned 4 years old (when the 4 to 6 age band starts) is GMFCS level I or II. The OGS may help establish the GMFCS level by determining the average OGS total score according to GMFCS level. Each GMFCS level showed significantly different OGS total score means according to post hoc analysis, with OGS scores decreasing more than 50% from GMFCS I to IV. Since we found that the OGS total score is moderately associated with the GMFCS level, the distribution of the OGS within the GMFCS level can help confirm a child's GMFCS level if there is uncertainty.

This study also provides expected OGS total scores based off a child's GMFCS level overall, by bilateral CP, and by unilateral CP. Participants with GMFCS level I with unilateral CP may have higher OGS scores compared with participants with bilateral CP since they can compensate with their unaffected leg. Clinicians can use these expected scores as a reference.

This study is limited in several ways. First, the OGS version used in this study is based on a modified version of Boyd's OGS. Therefore, the expected OGS total scores are specific to this version only. Another limitation is that numerous raters performed the OGS without establishing intrarater or interrater reliability since this is clinical retrospective data. However, this reflects the reality of clinical care. Raters were trained under direct supervision to perform the OGS to minimize interrater inconsistencies. Furthermore, standardization among raters also occurred during frequent discussions between the physical therapist rater and the interdisciplinary team in clinic regarding the OGS subscores regarding a child's possible or previous management. Future research should evaluate the reliability of this OGS version and whether the OGS total score for each participant changes over time. In conclusion, this study supports the OGS as another tool to help determine a child's GMFCS level, and provides expected total OGS scores based off a child's GMFCS level. The OGS is a clinically useful tool to quickly and objectively rate a child's gait.


The authors would like to acknowledge Deanna Carman and Blanch Carpenter for assisting in the development of the OGS, Debbie Baldwin and Jonathan VanPelt for assisting with the database, and Kim Kaipus, Stephanie Cone, Darla Kalb, Tony Hageman, Carol Chambers, and Hun Epps for performing the OGS.


1. Koman LA, Mooney JF III, Smith B, Goodman A, Mulvaney T. Management of cerebral palsy with botulinum-A toxin: preliminary investigation. J Ped Orthop. 1993;13(4):489–495.
2. Boyd RN, Graham HK. Objective measurement of clinical findings in the use of botulinum toxin type A for the management of children with cerebral palsy. Eur J Neurol. 1999;6(suppl 4):S23–S35.
3. Mackey AH, Lobb GL, Walt SE, Stott NS. Reliability and validity of the Observational Gait Scale in children with spastic diplegia. Dev Med Child Neurol. 2003;45(1):4–11.
4. Araújo PA, Kirkwood RN, Figueiredo EM. Validity and intra- and inter-rater reliability of the Observational Gait Scale for children with spastic cerebral palsy. Rev Bras Fisioter. 2009;13:267–273.
5. Bella GP, Rodrigues NB, Valenciano PJ, Silva LM, Souza RC. Correlation among the visual gait assessment scale, Edinburgh visual gait scale and Observational Gait Scale in children with spastic diplegic cerebral palsy. Rev Bras Fisioter. 2012;16(2):134–140.
6. Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol. 1997;39(4):214–223.
7. Ross SA, Engsberg JR. Relationships between spasticity, strength, gait, and the GMFM-66 in persons with spastic diplegia cerebral palsy. Arch Phys Med Rehabil. 2007;88(9):1114–1120.
8. Zanudin A, Mercer TH, Jagadamma KC, van der Linden ML. Psychometric properties of measures of gait quality and walking performance in young people with cerebral palsy: A systematic review. Gait Posture. 2017;58:30–40.

CP; Gross Motor Function Classification System; OGS

© 2021 Academy of Pediatric Physical Therapy of the American Physical Therapy Association