Skip to content
Skip to main navigation
Skip to first column
Skip to second column
Does EMA or FDA have a preferred method for dealing with missing data for when companies submit new drugs for approval?
I am planning a placebo-controlled trial in a chronic pain condition and know that I will get a high percentage of patients dropping out, but don't yet know if it will be at random or related to the test article.
If I have fully observed baseline covariates but missing response and the reponse is only measured at endpoint, is it worthwhile to do a multiple imputation?
What should I write in my analysis plan about the main analysis I'm going to use when I only know there're missing data but nothing else, such as % of missing or the missing mechanism, etc?
When would you advise using LOCF (last observation carried forward) in the analysis of a clinical trial?
What is the difference between last observation carried forward, reporting of as-observed data, and how inferior are these analyses to intent-to-treat analysis?
Submit a question to the Missing Data Google Group
Introduction to missing data
REALCOM Impute and Stata
Substantive model compatible FCS
DIA working group
Courses / Workshops
Book: MI & its application
Google Discussion Group
Registered Users Area
Monograph: missing data in clinical trials
MLwiN macros for multiple imputation
SAS code for sensitivity analysis using multiple imputation
LEMMA Missing Data Module
Login / Logout
Copyright © London School of Hygiene and Tropical Medicine