Raw, centile and individualised minimal clinically important differences for the MMN-Rasch-built Overall Disability Scale

backgroundOptimal clinical application of the minimal clinically important difference (MCID) for the MMN-Rasch-built Overall Disability Scale

methodsWe retrospectively studied subjects with MMN from 2 UK neuropathy centres. Raw and centile MMN-RODS scores were collected, at initial, two intermediate, and latest assessment, and distribution-based MCIDs determined at studied time-points. The sensitivities of raw, centile and individualised MCIDs, were compared.

resultsWe included 32 consecutive subjects with MMN on individualised immunoglobulin dosing regimens. First intermediate, second intermediate, and latest assessments were performed at a median of 17.0, 54.1 and 74.6 months from onset, respectively. Progressive amelioration of mean raw MMN-RODS score occurred between initial and latest assessment. The distribution-based raw MMN-RODS MCID was of 3, 4 and 5, and the distribution-based centile MMN-RODS MCID was of 6, 7 and 7, at first intermediate, second intermediate and latest assessment, respectively. At first intermediate assessment, raw and individualised MCIDs were equally sensitive, and both more sensitive than centile MCID (McNemar's Test: p < 0.001, p < 0.001). Raw MCID, centile MCID and individualised MCID, all had equivalent sensitivity at the second intermediate and latest assessment. Neither initial MMN-RODS score, nor amplitude of treatment-induced changes in early disease stages, independently predicted total improvement amplitude or final outcome.

conclusionsMMN-RODS MCID cut-offs may require adaption to assessment timing. In earlier stages, raw/individualised MCID may be more sensitive than centile MCID, whereas all three methods may subsequently be equivalent. Initial disability and early treatment response do not predict achievable total improvement amplitude nor final outcome.