Minimum Clinically Important Difference of Gross Motor Function and Gait Endurance in Children with Motor Impairment: A Comparison of Distribution-Based Approaches

Abstract EN

Objective.

The minimum clinically important difference (MCID) is a standard way of measuring clinical relevance.

The objective of this work was to establish the MCID for the 6-minute walking test (6minWT) and the Gross Motor Function Measure (GMFM-88) in pediatric gait disorders.

Methods.

A cohort, pretest-posttest study was conducted in a hospitalized care setting.

A total of 182 patients with acquired brain injury (ABI) or cerebral palsy (CP) performed 20 robot-assisted gait training sessions complemented with 20 sessions of physical therapy over 4 weeks.

Separate MCIDs were calculated using 5 distribution-based approaches, complemented with an anonymized survey completed by clinical professionals.

Results.

The MCID range for the 6minWT was 20-38 m in the ABI cohort, with subgroup ranges of 20-36 m for GMFCS I-II, 23-46 m for GMFCS III, and 24-46 m for GMFCS IV.

MCIDs for the CP population were 6-23 m, with subgroup ranges of 4-28 m for GMFCS I-II, 9-19 m for GMFCS III, and 10-27 m for GMFCS IV.

For GMFM-88 total score, MCID values were 1.1%-5.3% for the ABI cohort and 0.1%-3.0% for the CP population.

For dimension “D” of the GMFM, MCID ranges were 2.3%-6.5% and 0.8%-5.2% for ABI and CP populations, respectively.

For dimension “E,” MCID ranges were 2.8%-6.5% and 0.3%-4.9% for ABI and CP cohorts, respectively.

The survey showed a large interquartile range, but the results well mimicked the distribution-based methods.

Conclusions.

This study identified for the first time MCID ranges for 6minWT and GMFM-88 in pediatric patients with neurological impairments, offering useful insights for clinicians to evaluate the impact of treatments.

Distribution-based methods should be used with caution: methods based on pre-post correlation may underestimate MCID when applied to patients with small improvements over the treatment period.

Our results should be complemented with estimates obtained using consensus- and anchor-based approaches.