A comparison of methods to adjust for continuous covariates in the analysis of randomised trials

BackgroundAlthough covariate adjustment in the analysis of randomised trials can be beneficial, adjustment for continuous covariates is complicated by the fact that the association between covariate and outcome must be specified. Misspecification of this association can lead to reduced power, and potentially incorrect conclusions regarding treatment efficacy.MethodsWe compared several methods of adjustment to determine which is best when the association between covariate and outcome is unknown. We assessed (a) dichotomisation or categorisation; (b) assuming a linear association with outcome; (c) using fractional polynomials with one (FP1) or two (FP2) polynomial terms; and (d) using restricted cubic splines with 3 or 5 knots. We evaluated each method using simulation and through a re-analysis of trial datasets.ResultsMethods which kept covariates as continuous typically had higher power than methods which used categorisation. Dichotomisation, categorisation, and assuming a linear association all led to large reductions in power when the true association was non-linear. FP2 models and restricted cubic splines with 3 or 5 knots performed best overall.ConclusionsFor the analysis of randomised trials we recommend (1) adjusting for continuous covariates even if their association with outcome is unknown; (2) keeping covariates as continuous; and (3) using fractional polynomials with two polynomial terms or restricted cubic splines with 3 to 5 knots when a linear association is in doubt.

[1]  H Brenner,et al.  Controlling for Continuous Confounders in Epidemiologic Research , 1997, Epidemiology.

[2]  B. Kahan,et al.  Adjusting for multiple prognostic factors in the analysis of randomised trials , 2013, BMC Medical Research Methodology.

[3]  S. Pocock,et al.  Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practiceand problems , 2002, Statistics in medicine.

[4]  Tim P Morris,et al.  The risks and rewards of covariate adjustment in randomized trials: an assessment of 12 outcomes from 8 studies , 2014, Trials.

[5]  P. Royston,et al.  Selection of important variables and determination of functional form for continuous predictors in multivariable model building , 2007, Statistics in medicine.

[6]  D. Stoyan.  Stereology and stochastic geometry , 1990 .

[7]  V. Berger.  Valid Adjustment of Randomized Comparisons for Binary Covariates , 2004 .

[8]  Sunil J Rao,et al.  Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis , 2003 .

[9]  Douglas G Altman,et al.  Dichotomizing continuous predictors in multiple regression: a bad idea , 2006, Statistics in medicine.

[10]  R. Simon,et al.  Flexible regression models with cubic splines. , 1989, Statistics in medicine.

[11]  学 岩崎,et al.  Committee for Proprietary Medicinal Products (CPMP): points to consider on adjustment for baseline covariates. , 2006, Statistics in medicine.

[12]  Ewout W Steyerberg,et al.  Covariate adjustment increased power in randomized controlled trials: an example in traumatic brain injury. , 2012, Journal of clinical epidemiology.

[13]  C. Doré,et al.  Update on the transfusion in gastrointestinal bleeding (TRIGGER) trial: statistical analysis plan for a cluster-randomised feasibility trial , 2013, Trials.

[14]  N. Jewell,et al.  Some surprising results about covariate adjustment in logistic regression models , 1991 .

[15]  B. Kahan,et al.  Improper analysis of trials randomised using stratified blocks or minimisation , 2012, Statistics in medicine.

[16]  Ewout W Steyerberg,et al.  Covariate adjustment had similar benefits in small and large randomized controlled trials. , 2015, Journal of clinical epidemiology.

[17]  J. Ganju.  Diagnostics for assumptions in moderate to large simple clinical trials: do they really help? by Jonathan J. Shuster, Statistics in Medicine 2005; 24:2431–2438 , 2006, Statistics in medicine.

[18]  P. Royston,et al.  Multivariable model-building , 2008 .

[19]  H A Wendel,et al.  Randomization in clinical trials. , 1978, Science.

[20]  V. Durkalski,et al.  The impact of covariate adjustment at randomization and analysis for binary outcomes: understanding differences between superiority and noninferiority trials , 2015, Statistics in medicine.

[21]  David Moher,et al.  SPIRIT 2013 explanation and elaboration: guidance for protocols of clinical trials , 2013, BMJ.

[22]  Frank E. Harrell,et al.  Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis , 2001 .

[23]  A. Tsiatis,et al.  Efficiency Study of Estimators for a Treatment Effect in a Pretest–Posttest Trial , 2001 .

[24]  W W Hauck,et al.  Should we adjust for covariates in nonlinear regression analyses of randomized trials? , 1998, Controlled clinical trials.

[25]  Ewout W Steyerberg,et al.  Randomized controlled trials with time-to-event outcomes: how much does prespecified covariate adjustment increase power? , 2006, Annals of epidemiology.

[26]  Brennan C Kahan,et al.  Bias in randomised factorial trials , 2013, Statistics in medicine.

[27]  Ewout W Steyerberg,et al.  Covariate adjustment in randomized controlled trials with dichotomous outcomes increases statistical power and reduces sample size requirements. , 2004, Journal of clinical epidemiology.

[28]  M. Gail,et al.  Biased estimates of treatment effect in randomized experiments with nonlinear regressions and omitted covariates , 1984 .

[29]  Patrick Royston,et al.  The cost of dichotomising continuous variables , 2006, BMJ : British Medical Journal.

[30]  T. Pruett.  Intrapleural Use of Tissue Plasminogen Activator and DNase in Pleural Infection , 2012 .

[31]  B. Kahan,et al.  Reporting and analysis of trials using stratified randomisation in leading medical journals: review and reanalysis , 2012, BMJ : British Medical Journal.

[32]  B. Kahan,et al.  Assessing potential sources of clustering in individually randomised trials , 2013, BMC Medical Research Methodology.

[33]  P. Rosenbaum.  The Consequences of Adjustment for a Concomitant Variable that Has Been Affected by the Treatment , 1984 .

[34]  Patrick Royston,et al.  Combining fractional polynomial model building with multiple imputation , 2015, Statistics in medicine.

[35]  T. Cleophas.  The performance of the two-stage analysis of two-treatment, two-period crossover trials. , 1991, Statistics in medicine.

[36]  G M Raab,et al.  How to select covariates to include in the analysis of a clinical trial. , 2000, Controlled clinical trials.

[37]  M Schumacher,et al.  Effects of covariate omission and categorization when analysing randomized trials with the Cox model. , 1997, Statistics in medicine.

[38]  S. Lipsitz,et al.  Does Clustering Affect the Usual Test Statistics of no Treatment Effect in a Randomized Clinical Trial , 1998 .

[39]  Kristopher J Preacher,et al.  On the practice of dichotomization of quantitative variables. , 2002, Psychological methods.