A comparison of covariate adjustment approaches under model misspecification in individually randomized trials

Trials

Methodology

Clemence Leyrat¹  Elizabeth Williamson¹  Karla Diaz-Ordaz²  Mia S. Tackney³  Ian White⁴  Tim Morris⁴ 

[1] Department of Medical Statistics, London School of Hygiene and Tropical Medicine, London, UK;Department of Medical Statistics, London School of Hygiene and Tropical Medicine, London, UK;Department of Statistical Science, UCL, London, United Kingdom;Department of Medical Statistics, London School of Hygiene and Tropical Medicine, London, UK;MRC Biostatistics Unit, University of Cambridge, Cambridge, United Kingdom;MRC Clinical Trials Unit at UCL, London, UK;

关键词: Covariate adjustment;    Randomized controlled trials;    Misspecification;    ANCOVA;    G-computation;    IPTW;    AIPTW;    TMLE;   

DOI  :  10.1186/s13063-022-06967-6

 received in 2021-11-05, accepted in 2022-11-28,  发布年份 2022

来源: Springer

【 摘 要 】

Adjustment for baseline covariates in randomized trials has been shown to lead to gains in power and can protect against chance imbalances in covariates. For continuous covariates, there is a risk that the the form of the relationship between the covariate and outcome is misspecified when taking an adjusted approach. Using a simulation study focusing on individually randomized trials with small sample sizes, we explore whether a range of adjustment methods are robust to misspecification, either in the covariate–outcome relationship or through an omitted covariate–treatment interaction. Specifically, we aim to identify potential settings where G-computation, inverse probability of treatment weighting (IPTW), augmented inverse probability of treatment weighting (AIPTW) and targeted maximum likelihood estimation (TMLE) offer improvement over the commonly used analysis of covariance (ANCOVA). Our simulations show that all adjustment methods are generally robust to model misspecification if adjusting for a few covariates, sample size is 100 or larger, and there are no covariate–treatment interactions. When there is a non-linear interaction of treatment with a skewed covariate and sample size is small, all adjustment methods can suffer from bias; however, methods that allow for interactions (such as G-computation with interaction and IPTW) show improved results compared to ANCOVA. When there are a high number of covariates to adjust for, ANCOVA retains good properties while other methods suffer from under- or over-coverage. An outstanding issue for G-computation, IPTW and AIPTW in small samples is that standard errors are underestimated; they should be used with caution without the availability of small-sample corrections, development of which is needed. These findings are relevant for covariate adjustment in interim analyses of larger trials.

【 授权许可】

CC BY   
© The Author(s) 2022

【 图 表 】

41116_2022_35_Article_IEq51.gif

40249_2022_1049_Article_IEq2.gif

41116_2022_35_Article_IEq50.gif

Fig. 4

Fig. 1

Fig. 3

40249_2022_1049_Article_IEq44.gif

Fig. 2

40249_2022_1049_Article_IEq1118.gif

40249_2022_1049_Article_IEq1117.gif

Fig. 1

Fig. 12

40249_2022_1049_Article_IEq39.gif

40249_2022_1049_Article_IEq38.gif

Fig. 11

Fig. 5

Fig. 10

40249_2022_1049_Article_IEq35.gif

40249_2022_1049_Article_IEq34.gif

Fig. 9

Fig. 8

40249_2022_1049_Article_IEq33.gif

40249_2022_1049_Article_IEq32.gif

Fig. 6

40708_2022_182_Article_IEq12.gif

Fig. 8

Fig. 5

Fig. 4

Fig. 5

Fig. 7

Fig. 3

Fig. 5

40854_2022_439_Article_IEq324.gif

41116_2022_35_Article_IEq47.gif

41116_2022_35_Article_IEq46.gif

41116_2022_35_Article_IEq45.gif

Fig. 6

41116_2022_35_Article_IEq43.gif

Fig. 1

Fig. 1

41116_2022_35_Article_IEq40.gif

41116_2022_35_Article_IEq39.gif

41116_2022_35_Article_IEq38.gif

41116_2022_35_Article_IEq37.gif

40249_2022_1049_Article_IEq26.gif

Fig. 5

Fig. 3

41116_2022_35_Article_IEq36.gif

41116_2022_35_Article_IEq35.gif

Fig. 4

41116_2022_35_Article_IEq32.gif

41116_2022_35_Article_IEq30.gif

41116_2022_35_Article_IEq29.gif

41116_2022_35_Article_IEq28.gif

Fig. 1

Fig. 1

40249_2022_1049_Article_IEq21.gif

Fig. 2

40249_2022_1049_Article_IEq57.gif

12936_2022_4438_Article_IEq29.gif

Fig. 1

Fig. 3

Fig. 3

40249_2022_1049_Article_IEq56.gif

Fig. 3

40249_2022_1049_Article_IEq20.gif

Fig. 2

Fig. 1

Fig. 11

Fig. 7

Fig. 1

42004_2022_803_Article_IEq55.gif

12936_2022_4438_Article_IEq9.gif

41116_2022_35_Article_IEq27.gif

41116_2022_35_Article_IEq25.gif

41116_2022_35_Article_IEq23.gif

41116_2022_35_Article_IEq21.gif

Fig. 2

41116_2022_35_Article_IEq17.gif

41116_2022_35_Article_IEq15.gif

41116_2022_35_Article_IEq13.gif

41116_2022_35_Article_IEq10.gif

41116_2022_35_Article_IEq7.gif

41116_2022_35_Article_IEq5.gif