highly significantP < .001
Two-proportion Z tests revealed that the differences in proportions of populations in the EHR-vs registry-derived datasets were highly significant for all cohorts ( P < .001) ( Table 1 C). 3.2 Cohort modelling The EHR- and registry-derived data were used to calculate the predicted proportion, and statistically representative range, for each demographic cohort for hypothetical US SLE trials of 100 subjects and 300 subjects.