BackgroundThe absolute change in the severity score between the baseline and pre-specified time frame (absolute criterion) was recommended as a criterion for myasthenia gravis (MG) treatment response. But heterogeneity of disease severity might dilute major changes in individual patients. The rationality of relative criterion (improvement percentage) had not been evaluated in treatment response in patients with MG. ObjectivesTo investigate the consistency between an absolute criterion and a relative criterion in the evaluation of treatment response in patients with MG. MethodsWe retrospectively analyzed the treatment response to a 3-month standardized treatment protocol with only glucocorticoid in 257 MG patients native to immunological treatments. With the commonly used absolute criterion, cut-offs of relative criteria were generated with the receiver operating characteristic (ROC) curve in the whole cohort and in patients with different degrees of baseline severity stratified by pre-treatment quantitative myasthenia gravis score (QMGS). The consistency between absolute and relative criteria was examined with Cohen's Kappa test and Venn diagrams. ResultsThe absolute and relative criteria had an overall substantial consistency (Kappa value, 0.639, p < 0.001) in the cohort. The Kappa values were substantial to almost perfect in mild and moderate groups and moderate in severe groups between the absolute and relative criteria (all p <= 0.001). More patients were classified as responsive with an absolute criterion while as unresponsive with a relative criterion in the moderate and severe groups. ConclusionsThe overall consistency between absolute and relative criteria was substantial in the whole cohort. The inconsistency between the two criteria was mainly from the moderate or severe patients at the baseline.