Reliability of patellar height indices in children with cerebral palsy and spina bifida.

BACKGROUND: The Koshino (KI) and Caton-Deschamps (CDI) indices are used to measure patellar height in children, with the CDI showing excellent reliability in typically developing (TD) children. Reliability of such measures in children with cerebral palsy (CP) and spina bifida (SB) is unknown. METHODS: Lateral knee radiographs were reviewed retrospectively for children with TD (n = 49), CP (n = 48) and SB (n = 42). Five raters took measurements from radiographs twice, at least two weeks apart. Measurements included the CDI, Insall-Salvati Index (ISI) and KI. Systematic variability (bias) and random variability were examined using repeated measures ANOVA, 95% limits of agreement (LOA) and coefficients of variation (CV). RESULTS: Mean values of all three indices differed among raters (p < 0.0001). A significant difference was seen between the first and second measurements for CDI and KI indicating a learning effect. LOA ranges were large for the CDI (intra-rater: 0.37-0.95, inter-rater: 0.60-1.04) and ISI (intra-rater: 0.25-0.49, inter-rater: 0.51-0.57) for all patient groups. The KI showed a clinically acceptable range for TD participants (intra-rater: 0.14-0.16, inter-rater: 0.11-0.14) with larger ranges for CP (intra-rater: 0.26-0.33, inter-rater 0.0.2-0.35) and SB patients (intra-rater: 0.23-0.27, inter-rater: 0.19-0.25). CVs were lowest (best) for KI (3.8% to 7.4%) and highest (worst) for CDI (14.7% to 23.1%) for all three groups. Results were similar for patients with both open and closed physes. CONCLUSIONS: The KI is the most reliable patellar height measure for paediatric patients with TD, CP and SB, with either open or closed physes. The KI is more complex and experience may be important for valid, reliable measurement.

Prediction of postoperative gait velocity in cerebral palsy.

Preoperative and postoperative gait analyses were reviewed for 47 patients with cerebral palsy in an attempt to ascertain predictors of surgical outcome as measured by gait velocity. Higher postoperative velocity correlated with higher preoperative velocity and younger age. Observed velocities in older children were smaller than predicted values. Older children showed either an average decline in velocity from normal for age or no change. Younger children showed an average increase in velocity toward normal for their age. Diagnosis, type of surgery, number of procedures performed and level of ambulation preoperatively were not predictive of postoperative velocity. These results suggests that gait velocity cannot be reliably increased in all children with cerebral palsy undergoing surgery, especially those older than age 12.

Variability in gait analysis interpretation.

The purpose of this study was to assess the reliability of interpretation of gait analysis data between physicians and institutions. Gait analysis data from seven patients were reviewed by 12 experienced gait laboratory physicians from six institutions. Reviewers identified problems and made treatment recommendations based on the data provided. Agreement among physicians for the most commonly diagnosed problems was slight to moderate (kappa range, 0.14-0.46). Physicians agreed on identification of soft tissue more than bony problems (intraclass correlation, 0.56 vs. 0.37). Variability regarding surgical recommendations for soft-tissue procedures (kappa range, 0.20-0.64) was similar to that for diagnosis of both soft-tissue and bone problems, although recommendation for hamstring lengthening showed substantial agreement (kappa = 0.64). There was less agreement in recommendation of osteotomies (kappa range, 0.13-0.22). Physicians agreed more on the number of soft-tissue procedures than bone procedures recommended (intraclass correlation, 0.65 vs. 0.19). There was an interinstitutional difference in the frequency of soft-tissue (p = 0.0152) and osseous problem identification (p = 0.0002), as well as in the frequency of recommendations for soft-tissue surgery (p = 0.0004) and osteotomies (p < 0.0001). Although gait analysis data are themselves objective, this study demonstrates some subjectivity in their interpretation. The interobserver variability reported here is similar to that reported for established classification systems of various orthopedic conditions.