Skip to main content

Table 1 Options (not exhaustive) for reference-based and delta-based multiple imputation with a continuous outcome

From: A review of the use of controlled multiple imputation in randomised controlled trials with missing outcome data

Imputation option

Handling of missing outcome data

Last mean carried forward (LMCF)

Impute assuming all unobserved participants stayed at the mean value of their respective randomised group after their last observed time point.

Copy reference (CR)

Impute assuming all unobserved participants behaved similarly to the behaviour of the specified reference group for the entire duration of the study.

Copy increments in reference (CIR)

Impute assuming all unobserved participants followed the mean increments observed in specified reference group after their last observed time point.

Jump to reference (J2R)

Impute assuming all unobserved participants jumped to the behaviour of the specified reference group after their last observed time point.

Delta

Impute assuming all unobserved participants having a poorer or better response than those observed, by adding or subtracting a “delta” parameter’ to the expected value of the MAR imputed values. “Delta” can be implemented in all treatment groups, or in only one group, or may vary by treatment group or an alternative specified factor.