This paper provides an overview of multiple imputation and current perspectives on its use in medical research. We begin with a brief review of the problem of handling missing data in general and place multiple imputation in this context, emphasizing its relevance for longitudinal clinical trials and observational studies with missing covariates. We outline how multiple imputation proceeds in practice and then sketch its rationale. We explore the problem of obtaining proper imputations in some detail and distinguish two main classes of approach, methods based on fully multivariate models, and those that iterate conditional univariate models. We show how the use of so-called uncongenial imputation models are particularly valuable for sensitivity analyses and also for certain analyses in clinical trial settings. We also touch upon other forms of sensitivity analysis that use multiple imputation. Finally, we give some open questions that the increasing use of multiple imputation has thrown up, which we believe are useful directions for future research.
Get full access to this article
View all access options for this article.
References
1.
Rubin DBMultiple imputations in sample surveys — a phenomenological Bayesian approach to nonresponse . Proceedings of the Survey Research Methods Section of the American Statistical Association, 1978, pp. 20—34.
2.
Rubin DBMultiple imputation for nonresponse in surveys. Wiley , 1987.
3.
Rubin DB , Schenker N.Multiple imputation in health-care databases: an overview and some applications . Statistics in Medicine1991; 10: 585—98 .
4.
Rubin DBMultiple imputation after 18+ years . Journal of the American Statistical Association1996; 91: 473—90 .
5.
Schafer JLAnalysis of incomplete multivariate data. Chapman and Hall , 1997.
6.
Schafer JLMultiple imputation: a primer. StatisticalMethods in Medical Research1999; 8: 3—15 .
7.
Horton NJ , Lipsitz SRMultiple imputation in practice: comparison of software packages for regression models with missing variables . American Statistician2001; 55: 244—54 .
8.
Little Rja , Rubin DBStatistical analysis with missing data, second edition. Wiley , 2002. 9 Allison PDMissing data. Sage Publications , 1994.
9.
Little Rja.Pattern-mixture models for multivariate incomplete data . Journal of the American Statistical Association1993; 88: 125—34 .
10.
Rubin DBInference and missing data . Biometrika1976 ; 63: 581—92 .
11.
Robins JM , Rotnitzky A. , Zhao LPAnalysis of semiparametric regression models for repeated outcomes in the presence of missing data . Journal of the American Statistical Association1995; 90: 106—121 .
12.
Molenberghs G. , Thijs H. , Jansen I. , Beunkens C. , Kenward MG , Mallinkrodt G. , Carroll RCAnalyzing incomplete longitudinal clinical trial data . Biostatistics2004; 5: 445—64 .
13.
Dempster AP , Laird NM , Rubin DBMaximum likelihood from incomplete data via the EM algorithm (with discussion) . Journal of the Royal Statistical Society, Series B1977; 39: 1—38 .
14.
McCulloch CEMaximum likelihood algorithms for generalized linear mixed models . Journal of the American Statistical Association1997; 92: 162—70 .
15.
Reilly M. , Pepe M.A.mean score method for missing and auxilliary covariate data in regression models . Biometrika1995; 82: 299—314 .
16.
Best NG , Spiegelhalter DJ , Thomas A. , Brayne Ceg.Bayesian analysis of realistically complex models . Journal of the Royal Statistical Society, Series A 1996; 159: 232—342 .
17.
Horvitz DG , Thompson DJA generalization of sampling without replacement from a finite universe . Journal of the American Statistical Association1952; 47: 663—85 .
18.
van der Laan MJ , Robins JMUnified methods for censored longitudinal data and causality. Springer , 2002.
19.
Tsiatis AASemiparametric theory and missing data. Springer , 2006.
20.
Carpenter JR , Kenward MG , Vansteelandt S.A comparison of multiple imputation and inverse probability weighting for analyses with missing data . Journal of the Royal Statistical Society, Series A2006; 169: 571—584 .
21.
Li KH , Raghunathan TE , Rubin DBLarge-sample significance levels from multiply imputed data using moment-based statistics and an F reference distribution . Journal of the American Statistical Association1991; 86: 1065—73 .
22.
Li KH , Meng XL , Raghunathan TE , Rubin DBSignificance levels from repeated p-values with multiply-imputed data . Statistica Sinica1991; 1: 65—92 .
23.
Meng XL , Rubin DBPerforming likelihood ratio tests with multiply-imputed data sets . Biometrika1992; 89: 267—78 .
24.
Carpenter J. , Kenward MGMissing data in clinical trials. UK National Health Service, National Centre for Research on Methodology , 2006, in press.
25.
Fay REWhen are inferences from multiple imputation valid ? Proceedings of the Survey Research Methods Section of the American Statistical Association, 1992, pp. 227—32.
26.
Schafer JL , Schenker N.Inference with imputed conditional means . Journal of the American Statistical Association2000; 95: 144—54 .
27.
Gill R. , van der Laan M. , Robins J.Coarsening at random: characterizations, conjectures and counterexamples . Proceedings of the First Seattle Symposium on Survival Analysis , 1996, pp. 255—94.
28.
Robins JM , Gill RDNon-response models for the analysis of non-monotone ignorable missing data . Statistics in Medicine1997; 16: 39—56 .
29.
Herzog TN , Rubin DBUsing multiple imputations to handle nonresponse in sample surveys. Incomplete data in sample surveys, Volume 2: Theory and Bibliographies, 1983, Academic Press , pp. 115—42.
30.
Heitjan DF , Little Rja.Multiple imputation for the fatal accident reporting system . Applied Statistics1991; 40: 13—29 .
31.
Little Rja , Rubin DBStatistical analysis with missing data. Wiley and Sons , 1986.
32.
SAS Institute Inc., SAS Campus Drive, Cary, North Carolina 27513, USA.
33.
Splus 6.1 for Windows: Insightful Corporation, 1700 Westlake Avenue N, Suite 500, Seattle, Washington 98109, USA.
34.
Raghunathan TE , Lepkowski JM , Van Hoewyk J , Solenberger P.A multivariate technique for multiply imputing missing values using a sequence of regression models . Survey Methodology2001; 27: 85—95 .
35.
Taylor Jmg , Cooper KL , Wei JT , Sarma AV , Raghunathan TE , Heeringa SGUse of multiple imputation to correct for nonresponse bias in a survey of urologic symptoms among African—American men . American Journal of Epidemiology1987; 82: 528—50 .
36.
van Buuren S. , Boshuizen HC , Knook DLMultiple imputation of missing blood pressure covariates in survival analysis . Statistics in Medicine1999; 18: 681—694 .
37.
Stata Corporation, 702 University Drive East, College Station, Texas 77840, USA.
38.
Royston P.Multiple imputation of missing values . The Stata Journal2004; 3: 227—41 .
39.
Gelman A. , Raghunathan TEUsing conditional distributions for missing-data imputation . Statistical Science2001; 15: 268—69 .
40.
Schafer J.Imputation of missing covariates under a general linear mixed model. Technical report, Department of Statistics , Penn State University , 1997.
41.
MLwiN: Centre for Multilevel Modelling, Institute of Education, 20 Bedford Way, London WC1H 0AL, UK.
42.
Carpenter JR , Goldstein H.Multiple imputation in MLwiN . Multilevel Modelling Newsletter2004; 16: 9—18 .
43.
Liu M. , Taylor Jmg , Belin TRMultiple imputation and posterior simulation for multivariate missing data in longitudinal studies . Biometrics2000; 56: 1157—63 .
44.
Kenward MGSelection models for repeated measurements with nonrandom dropout: an illustration of sensitivity . Statistics in Medicine1998 ; 17: 2723—32 .
45.
Robins JM , Rotnitzky A. , Scharfstein DOSemiparametric regression for repeated outcomes with non-ignorable non-response . Journal of the American Statistical Association1998; 93: 1321—39 .
46.
Carpenter JR , Pocock S. , Lamm CJCoping with missing values in clinical trials: a model based approach applied to asthma trials . Statistics in Medicine2002; 21: 1043—66 .
47.
Meng X-L.Multiple-imputation inferences with uncongenial sources of input . Statistical Science1995; 10: 538—58 .
48.
Longford NTAn experiment in primary health care evaluation . Journal of the Royal Statistical Society, Series A1999; 162: 291—302 .
49.
Longford NTHandling missing data in diaries of alchohol consumption . Journal of the Royal Statistical Society, Series A 2000; 163: 381—402 .
50.
Little Rja.A class of pattern-mixture models for multivariate incomplete data . Biometrika1994; 81: 471—83 .
51.
Kenward MG , Molenberghs G. , Thijs H.Pattern-mixture models with proper time dependence . Biometrika2003; 90: 53—71 .
52.
Little Rja , Yau L.Intent-to-treat analysis for longitudinal studies with drop-outs . Biometrics1996; 52: 1324—33 .
53.
WinBUGS, MRC Biostatistics Unit, Institute of Public Health, University Forvie Site, Robinson Way, Cambridge CB2 2SR, UK.
54.
Wang N. , Robins JMLarge-sample thoery for parametric multiple imputation procedures . Biometrika1998; 85: 935—48 .
55.
Robins JM , Wang N.Inference for imputation estimators . Biometrika2000; 85: 113—24 .
