Multiple imputation of missing repeated outcome measurements did not add to linear mixed-effects models

• Published in: Journal of clinical epidemiology Vol. 65; no. 6; pp. 686 - 695
• Main Authors: Peters, Sanne A.E, Bots, Michiel L, den Ruijter, Hester M, Palmer, Mike K, Grobbee, Diederick E, Crouse, John R, O’Leary, Daniel H, Evans, Gregory W, Raichlen, Joel S, Moons, Karel G.M, Koffijberg, Hendrik
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Inc 01.06.2012; Elsevier; Elsevier Limited
• Subjects: Algorithms; Atherosclerosis (general aspects, experimental research); Atherosclerosis - drug therapy; Atherosclerosis - pathology; Bias; Biological and medical sciences; Biometry; Blood and lymphatic vessels; Cardiology. Vascular system; Cardiovascular disease; Carotid Intima-Media Thickness; Cholesterol; Clinical trials; Clinical Trials as Topic; Confidence Intervals; Epidemiology; Female; Fluorobenzenes - therapeutic use; Humans; Hydroxymethylglutaryl-CoA Reductase Inhibitors - therapeutic use; Internal Medicine; Linear mixed-effects model; Longitudinal Studies; Male; Medical sciences; Methodology; Middle Aged; Missing repeated outcome measurements; Models, Statistical; Multiple imputation; Pyrimidines - therapeutic use; Research Design; Rosuvastatin Calcium; Studies; Sulfonamides - therapeutic use; Clinical trials; Methodology; Multiple imputation; Linear mixed-effects model; Missing repeated outcome measurements; Carotid intima-media thickness; Multiple; Prognosis; Cardiovascular disease; Rosuvastatin; Hypocholesterolemic agent; Linear model; Vascular disease; Thickness; Accuracy; Carotid; Atherosclerosis; Repeated measurement; Clinical trial; Vascular wall; Antilipemic agent
• ISSN: 0895-4356; 1878-5921
• DOI: 10.1016/j.jclinepi.2011.11.012

Abstract Objective To assess the added value of multiple imputation (MI) of missing repeated outcomes measures in longitudinal data sets analyzed with linear mixed-effects (LME) models. Study Design and Setting Data were used from a trial on the effects of Rosuvastatin on rate of change in carotid intima-media thickness (CIMT). The reference treatment effect was derived from a complete data set. Scenarios and proportions of missing values in CIMT measurements were applied and LME analyses were used before and after MI. The added value of MI, in terms of bias and precision, was assessed using the mean-squared error (MSE) of the treatment effects and coverage of the 95% confidence interval. Results The reference treatment effect was −0.0177 mm/y. The MSEs for LME analysis without and with MI were similar in scenarios with up to 40% missing values. Coverage was large in all scenarios and was similar for LME with and without MI. Conclusion Our study empirically shows that MI of missing end point data before LME analyses does not increase precision in the estimated rate of change in the end point. Hence, MI had no added value in this setting and standard LME modeling remains the method of choice.