I had read there were various ways to handle missing data and the modern imputation methods might be the best solution to handle missing data. However, following planned statistical analysis plan in our RCT protocol, Last observation carried forward (LOCF) method was chosen.

I had done complete case analysis/ per protocol analysis using repeated measure ANOVA. As for the power of the study, we had achieved the minimum sample size (since we had included attrition rate during sample size calculation).

As for ITT analysis, I had a problem with drop-out with missing value at each timepoints and also missing all measurement value including baseline measurement (no data at all). So, is it okay if use mean imputation for to impute the data from those who had no data at all combine with LOCF method for drop-out who have baseline value.

Please advise, I will really appreciate it since statistics always making me confuse. Thank you.