A hybrid return to baseline imputation method to incorporate MAR and MNAR dropout missingness.
Contemp Clin Trials
; 120: 106859, 2022 09.
Article
in English
| MEDLINE | ID: covidwho-1959360
ABSTRACT
Missing data are inevitable in longitudinal clinical trials due to intercurrent events (ICEs) such as treatment interruption or premature discontinuation for different reasons. Missing at random (MAR) assumption is usually unverifiable and sensitivity analyses are often requested under missing not at random (MNAR) assumption. Return to baseline (RTB) imputation is a commonly used MNAR method. In practice, not all dropout missingness can be assumed MNAR. For example, missingness or dropouts due to COVID-19 can be reasonably assumed MAR. Therefore, traditional RTB is not applicable when there is both MAR and MNAR dropout missingness. Here we propose a hybrid strategy for RTB imputation which can handle missing data due to MAR and MNAR dropouts at the same time. Standard multiple imputation approach is proposed and an analytic likelihood based approach is derived to improve efficiency.
Keywords
Full text:
Available
Collection:
International databases
Database:
MEDLINE
Main subject:
COVID-19
Type of study:
Experimental Studies
/
Prognostic study
/
Randomized controlled trials
Limits:
Humans
Language:
English
Journal:
Contemp Clin Trials
Journal subject:
Medicine
/
Therapeutics
Year:
2022
Document Type:
Article
Similar
MEDLINE
...
LILACS
LIS