When this was not provided, SD change was estimated by using statistics provided (eg, t values, confidence intervals, SEs, and P values) or by using SDs for the baseline and follow-up score and the correlation between baseline and follow-up scores ( r). 16 The SD for the change score (SD change) (ie, mean difference between baseline and follow-up values) was used to calculate Hedges’ g*. 15 Effect sizes were calculated as Hedges’ g* to correct for bias associated with a small sample size. A random effects model was used because there was large variation between studies in terms of procedures, populations, and context. Weighted effect sizes were calculated by using the inverse-variance method. If data were reported on both limbs, data on the most affected limb or the right limb were extracted. If data were not reported for the most frequently reported muscle group, we extracted data for the next most frequently reported muscle group, and so on. For each study, we then extracted data for strength and spasticity, respectively, for the most frequently reported muscle group. Because authors typically reported muscle strength and spasticity for >1 muscle group, we ranked muscle groups in order of frequency of reporting across all included studies. We extracted data on passive knee extension and passive dorsiflexion only because these were the most commonly reported joints. 13 Finally, we extracted data on passive range of motion, muscle strength, and spasticity because these are commonly evaluated before and after MLS as part of a clinical examination and are used to inform clinical decision-making on surgery. For the purpose of this review, we extracted data on measures that mapped to the recently developed family of participation-related constructs. Since the publication of the ICF, there has been a lack of consensus regarding measurement tools to assess participation. In addition, we extracted data on participation, QoL, and satisfaction with surgery. We chose to extract data on gait summary scores rather than individual kinematic or kinetic variables because authors of many studies reported a large number of variables obtained from three-dimensional gait analysis. 10, – 12 When >1 summary score was reported in studies, we extracted data on the GPS. We extracted data on gait speed and summary statistics of gait (ie, the Gait Deviation Index, Gait Profile Score, or Gillette Gait Index). Although gross motor function was assessed in some studies by using the GMFCS, we did not consider the GMFCS to be a measure of gross motor function because it is primarily a classification system. We extracted data on gross motor function as measured by the Gross Motor Function Measure 66 (GMFM-66) or Gross Motor Function Measure 88, the Functional Mobility Scale (FMS), and the Gillette Functional Assessment Questionnaire (FAQ).
However, the review in 2010 revealed that authors of few studies evaluate the effect of MLS across multiple domains of the ICF. When developed, it was envisaged that 1 use of the ICF would be the evaluation of interventions. The ICF places greater emphasis on the role of the social and physical environment on functioning and conceptualizes that functioning results from a complex interaction between the person with the health condition and his or her context (consisting of personal and environmental factors).
7 Developed in 2001, the ICF is used to classify health-related domains by using the terms “body functions and structure,” “activity,” and “participation.” The ICF replaced the International Classification of Impairments, Disabilities, and Handicaps (ICIDH), which was used to conceptualize a health condition leading to impairment, disability, and handicap in a linear manner. The World Health Organization’s International Classification of Functioning, Disability, and Health (ICF) framework is useful for considering the impact of CP on the individual.