Comparison of intent-to-treat analysis strategies for pre-post studies with loss to follow-up

Wenna Xi, Michael L Pennell, Rebecca R Andridge, Electra D Paskett, Wenna Xi, Michael L Pennell, Rebecca R Andridge, Electra D Paskett

Abstract

In pre-post studies when all outcomes are completely observed, previous studies have shown that analysis of covariance (ANCOVA) is more powerful than a change-score analysis in testing the treatment effect. However, there have been few studies comparing power under missing post-test values. This paper was motivated by the Behavior and Exercise for Physical Health Intervention (BePHIT) Study, a pre-post study designed to compare two interventions on postmenopausal women's walk time. The goal of this study was to compare the power of two methods which adhere to the intent-to-treat (ITT) principle when post-test data are missing: ANCOVA after multiple imputation (MI) and the mixed model applied to all-available data (AA). We also compared the two ITT analysis strategies to two methods which do not adhere to ITT principles: complete-case (CC) ANCOVA and the CC mixed model. Comparisons were made through analyses of the BePHIT data and simulation studies conducted under various sample sizes, missingness rates, and missingness scenarios. In the analysis of the BePHIT data, ANCOVA after MI had the smallest p-value for the test of the treatment effect of the four methods. Simulation results demonstrated that the AA mixed model was usually more powerful than ANCOVA after MI. The power of ANCOVA after MI dropped the fastest as the missingness rate increased; in most simulated scenarios, ANCOVA after MI had the smallest power when 50% of the post-test outcomes were missing.

Keywords: ANCOVA; Missing data; Mixed model; Multiple imputation; Randomized trial.

Figures

Fig. 1
Fig. 1
Power comparisons for analysis models including WH and m=35 per group.
Fig. 2
Fig. 2
Type I Error Rate Comparisons for Analysis Models with WH and n = 35.

References

    1. David P., Buckworth J., Pennell M.L., Katz M.L., DeGraffinreid C.R., Paskett E.D. A walking intervention for postmenopausal women using mobile phones and interactive voice response. J. Telemed. Telecare. 2012;18(1):20–25.
    1. Allison P.D. Change scores as dependent variables in regression analysis. Socio. Meth. 1990;20:93–114.
    1. Cronbach L.J., Furby L. How we should measure “change”: or should we? Psychol. Bull. 1970;74(1):68.
    1. Glidden D.V., Shiboski S.C., McCulloch C.E. Springer; 2011. Regression Methods in Biostatistics: Linear, Logistic, Survival, and Repeated Measures Models.
    1. Oakes J.M., Feldman H.A. Statistical power for nonequivalent pretest-posttest designs the impact of change-score versus ancova models. Eval. Rev. 2001;25(1):3–28.
    1. Little R.J., Rubin D.B. Wiley; 2002. Statistical Analysis with Missing Data.
    1. Lachin J.M. Statistical considerations in the intent-to-treat principle. Contr. Clin. Trials. 2000;21(3):167–189.
    1. White I.R., Carlin J.B. Bias and efficiency of multiple imputation compared with complete-case analysis for missing covariate values. Stat. Med. 2010;29(28):2920–2931.
    1. Campbell D.T., Stanley J.C., Gage N.L. Houghton Mifflin Boston; 1963. Experimental and Quasi-experimental Designs for Research.
    1. Pape U.J., Millett C., Lee J.T., Car J., Majeed A. Disentangling secular trends and policy impacts in health studies: use of interrupted time series analysis. J. R. Soc. Med. 2013;106(4):124–129.
    1. Harris C.W. University of Wisconsin Press Madison; 1963. Problems in Measuring Change.
    1. Werts C.E., Linn R.L. A general linear model for studying growth. Psychol. Bull. 1970;73(1):17–22.
    1. Panel on Handling Missing Data in Clinical Trials; National Research Council . The National Academies Press; 2010. The Prevention and Treatment of Missing Data in Clinical Trials. ISBN 9780309158145.
    1. Ibrahim J.G., Chu H., Chen M.H. Missing data in clinical studies: issues and methods. J. Clin. Oncol. 2012;30(26):3297–3303.
    1. White I.R., Royston P., Wood A.M. Multiple imputation using chained equations: issues and guidance for practice. Stat. Med. 2011;30(4):377–399.
    1. Rubin D.B. Wiley; New York: 1987. Multiple Imputation for Nonresponse in Surveys.
    1. Yuan Y. Multiple imputation using SAS software. J. Stat. Software. 2011;45(6):1–25.
    1. Carpenter J., Kenward M. John Wiley & Sons; 2012. Multiple Imputation and its Application.
    1. Gomez E.V., Schaalje G.B., Fellingham G.W. Performance of the Kenward–Roger method when the covariance structure is selected using AIC and BIC. Commun. Stat. Simulat. Comput. 2005;34(2):377–392.
    1. Kenward M.G., Roger J.H. Small sample inference for fixed effects from restricted maximum likelihood. Biometrics. 1997;53(3):983–997.
    1. Littell R.C. SAS institute; 2006. SAS for Mixed Models.
    1. Salim A., Mackinnon A., Christensen H., Griffiths K. Comparison of data analysis strategies for intent-to-treat analysis in pre-test–post-test designs with substantial dropout rates. Psychiatr. Res. 2008;160(3):335–345.
    1. Liu G.F., Lu K., Mogg R., Mallick M., Mehrotra D.V. Should baseline be a covariate or dependent variable in analyses of change from baseline in clinical trials? Stat. Med. 2009;28:2509–2530.
    1. Hyer J.M., Waller J.L. SAS Global Forum Proceedings. 2014. Comparison of five analytic techniques for two-group pre-post repeated measures designs using sas. Paper 1798–2014.
    1. Mehrotra D.V., Li X., Liu J., Lu K. Analysis of longitudinal clinical trials with missing data using multiple imputation in conjunction with robust regression. Biometrics. 2012;68:1250–1259.
    1. Kiecolt-Glaser J.K., Fagundes C.P., Andridge R., Peng J., Malarkey W.B., Habash D. Depression, daily stressors and inflammatory responses to high-fat meals: when stress overrides healthier food choices. Mol. Psychiatr. 2017;22(3):476–482.
    1. Sullivan T.R., White I.R., Salter A.B., Ryan P., Lee K.J. Should multiple imputation be the method of choice for handling missing data in randomized trials? Stat. Meth. Med. Res. 2016 0962280216683570.

Source: PubMed

Sponsorer og samarbeidspartnere

Medisinsk tilstand

Legemiddelintervensjoner

3
Abonnere