The draft ICH E9(R1) addendum stipulates that an estimator should align with its associated estimand and yield an estimate that facilitates reliable interpretations. The addendum further stipulates that assumptions should be justifiable and plausible, and that the extent of assumptions is an important consideration for whether an estimate will be robust because assumptions are often unverifiable. The draft addendum specifies 5 strategies for dealing with intercurrent events. The intent of this paper is to provide conceptual considerations and technical details for various estimators that align with these strategies. We include focus on how the nature and extent of assumptions influences the potential robustness of the various estimators. The content reflects the knowledge, experience, and opinions of the Drug Information Association’s Scientific Working Group on Missing Data. This group includes experienced statisticians from across industry and academia, primarily in the US and European Union.
RatitchBBellJMallinckrodtCH. Choosing estimands. Therapeutic Innovation & Regulatory Science. In press.
3.
MallinckrodtCHMolenberghsGRathmannS. Choosing estimands in clinical trials with missing data. Pharm Stat. 2017;16(1):29–36.
4.
MallinckrodtCHLipkovichI. A Practical Guide to Analyzing Longitudinal Clinical Trial Data. Boca Raton, FL: Chapman and Hall/CRC; 2017.
5.
RatitchBLipkovichIEricksonJSZhangLMallinckrodtCH. Points to consider for analyzing long-term efficacy outcomes in clinical trials. Pharm Stat. 2018;17(6):685–700.
6.
O’NeillRTTempleR. The prevention and treatment of missing data in clinical trials: an FDA perspective on the importance of dealing with it. Clin Pharmacol Ther. 2012;91(3):550–554.
7.
ZeidlerH. Paracetamol and the placebo effect in osteoarthritis trials: a missing link?Pain Res Treat. 2011;2011:696791.
8.
MallinckrodtCH. Preventing and Treating Missing Data in Longitudinal Clinical Trials: A Practical Guide. New York, NY: Cambridge University Press; 2013.
9.
MallinckrodtCHLinQLipkovichIMolenberghsG. A structured approach to choosing estimands and estimators in longitudinal clinical trials. Pharm Stat. 2012;11:456–461.
10.
MallinckrodtCHRogerJChuang-SteinC. Missing data: turning guidance into action. Stat Biopharm Res. 2013;5(4):369–382.
11.
CarpenterJRogerJKenwardM. Analysis of longitudinal trials with missing data: a framework for relevant, accessible assumptions, and inference via multiple imputation. J Biopharm Stat. 2013;23(6):1352–1371.
12.
AyelaBLipkovichIMolenberghsGMallinckrodtCH. A multiple-imputation-based approach to sensitivity analyses and effectiveness assessments in longitudinal clinical trials. J Biopharm Stat. 2014;24(2):211–228.
13.
O’KellyMRatitchB. Clinical Trials With Missing Data: A Guide for Practitioners. Chichester, UK: Wiley; 2014.
14.
ShihWJ. Problems in dealing with missing data and informative censoring in clinical trials. Curr Control Trials Cardiovasc Med. 2002;3(1):4.
15.
GouldAL. A new approach to the analysis of clinical drug trial with withdrawals. Biometrics. 1980;36:721–727.
16.
LachinJM. Worst-rank score analysis with informatively missing observations in clinical trials. Control Clin Trials. 1999;20:408–422.
17.
WangDPocockS. A win ratio approach to comparing continuous non-normal outcomes in clinical trials. Pharm Stat. 2016;15:238–245.
18.
PermuttTLiF. Trimmed means for symptom trials with dropouts. Pharm Stat. 2017;16:20–28.
19.
MehrotraDLiuFPermuttT. Missing data in clinical trials: control-based mean imputation and sensitivity analyses. Pharm Stat. 2017;16:378–392.
20.
WangMDLiuJMolenberghsGMallinckrodtCH. An evaluation of the trimmed mean approach in clinical trials with dropout. Pharm Stat. 2018;17(3):278–289.
21.
LittleRJARubinDB. Statistical Analysis With Missing Data. 2nd ed. New York, NY: John Wiley; 2002.
22.
MolenberghsGKenwardMG. Missing Data in Clinical Studies. Chichester, UK: Wiley; 2007.
23.
VerbekeGMolenberghsG. Linear Mixed Models for Longitudinal Data. New York, NY: Springer; 2000.
24.
HarvilleDA. Maximum likelihood approaches to variance component estimation and to related problems. J Am Stat Assoc. 1977;72:320–338.
RubinDB. Multiple Imputation for Nonresponse in Surveys. New York, NY: John Wiley; 1987.
27.
CarpenterJRKenwardMG. Multiple Imputation and Its Application. Chichester, UK: Wiley; 2013.
28.
Van BuurenS. Multiple imputation of discrete and continuous data by fully conditional specification. Stat Methods Med Res. 2007;16:219–242.
29.
Van BuurenS. Flexible Imputation of Missing Data. Boca Raton, FL: Chapman & Hall/CRC; 2012.
30.
MolenberghsGFitzmauriceGKenwardMTsiatisAVerbekeG. Handbook of Missing Data Methodology. Boca Raton, FL: CRC Press; 2015.
31.
MallinckrodtCHRogerJChuang-SteinC. Recent developments in the prevention and treatment of missing data. Therapeutic Innovation & Regulatory Science. 2014;48(1):68–80.
32.
RobinsJMRotnitzkyAZhaoLP. Analysis of semiparametric regression models for repeated outcomes in the presence of missing data. J Am Stat Assoc. 1995;90:106–121.
33.
LipkovichIRatitchBO’KellyM. Sensitivity to censored-at-random assumption in the analysis of time-to-event endpoints. Pharm Stat. 2016;15:216–229.
34.
HeitjanDFRubinDB. Ignorability and coarse data. Ann Stat. 1991;19:2244–2253.
35.
MallinckrodtCHLinQMolenberghsM. A structured framework for assessing sensitivity to missing data assumptions in longitudinal clinical trials. Pharm Stat. 2013;12:1–6.
36.
LittleRYauL. Intent-to-treat analysis for longitudinal studies with drop-outs. Biometrics. 1996;52(4):1324–1333.
LiuGFPangL. On analysis of longitudinal clinical trials with missing data using reference-based imputation. J Biopharm Stat. 2016;26(5):924–936.
39.
ImbensGAngristJ. Identification and estimation of average treatment effects. Econometrica. 1994;62:467–475.
40.
FrangakisCERubinDB. Principal stratification in causal inference. Biometrics. 2002;58(1):21–29.
41.
ZhangJLRubinDB. Estimation of causal effects via principal stratification when some outcomes are truncated by “death.”J Educ Behav Stat. 2003;28:353–368.
42.
BarnardJFrangakisCEHillJLRubinDB. Principal stratification approach to broken randomized experiments. J Am Stat Assoc. 2003;98:299–323.
43.
RoyJHoganJWMarcusBH. Principal stratification with predictors of compliance for randomized trials with 2 active treatments. Biostatistics. 2008;9(2):277–289.
44.
RubinD. More powerful randomization-based p-values in double-blind trials with non-compliance. Stat Med. 1998;17:371–385.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
