期刊论文详细信息
Trials
A comparison of covariate adjustment approaches under model misspecification in individually randomized 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
PDF
【 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

【 预 览 】
附件列表
Files Size Format View
RO202305113887615ZK.pdf 2556KB PDF download
41116_2022_35_Article_IEq5.gif 1KB Image download
41116_2022_35_Article_IEq7.gif 1KB Image download
41116_2022_35_Article_IEq10.gif 1KB Image download
41116_2022_35_Article_IEq13.gif 1KB Image download
41116_2022_35_Article_IEq15.gif 1KB Image download
41116_2022_35_Article_IEq17.gif 1KB Image download
Fig. 2 220KB Image download
41116_2022_35_Article_IEq21.gif 1KB Image download
41116_2022_35_Article_IEq23.gif 1KB Image download
41116_2022_35_Article_IEq25.gif 1KB Image download
41116_2022_35_Article_IEq27.gif 1KB Image download
MediaObjects/12974_2023_2697_MOESM1_ESM.tif 955KB Other download
12936_2022_4438_Article_IEq9.gif 1KB Image download
42004_2022_803_Article_IEq55.gif 1KB Image download
Fig. 1 499KB Image download
Fig. 7 239KB Image download
Fig. 11 95KB Image download
Fig. 1 332KB Image download
MediaObjects/12888_2022_4501_MOESM1_ESM.docx 13KB Other download
Fig. 2 73KB Image download
40249_2022_1049_Article_IEq20.gif 1KB Image download
Fig. 3 119KB Image download
40249_2022_1049_Article_IEq56.gif 1KB Image download
Fig. 3 87KB Image download
MediaObjects/12951_2022_1516_MOESM1_ESM.docx 1524KB Other download
Fig. 3 1840KB Image download
Fig. 1 471KB Image download
12936_2022_4438_Article_IEq29.gif 1KB Image download
MediaObjects/13068_2022_2254_MOESM3_ESM.tif 5470KB Other download
40249_2022_1049_Article_IEq57.gif 1KB Image download
Fig. 2 1426KB Image download
40249_2022_1049_Article_IEq21.gif 1KB Image download
Fig. 1 1888KB Image download
Fig. 1 162KB Image download
41116_2022_35_Article_IEq28.gif 1KB Image download
41116_2022_35_Article_IEq29.gif 1KB Image download
41116_2022_35_Article_IEq30.gif 1KB Image download
MediaObjects/12888_2023_4528_MOESM1_ESM.docx 13KB Other download
41116_2022_35_Article_IEq32.gif 1KB Image download
Fig. 4 582KB Image download
41116_2022_35_Article_IEq35.gif 1KB Image download
41116_2022_35_Article_IEq36.gif 1KB Image download
Fig. 3 2334KB Image download
Fig. 5 1046KB Image download
40249_2022_1049_Article_IEq26.gif 1KB Image download
41116_2022_35_Article_IEq37.gif 1KB Image download
41116_2022_35_Article_IEq38.gif 1KB Image download
41116_2022_35_Article_IEq39.gif 1KB Image download
41116_2022_35_Article_IEq40.gif 1KB Image download
Fig. 1 1380KB Image download
Fig. 1 82KB Image download
41116_2022_35_Article_IEq43.gif 1KB Image download
Fig. 6 900KB Image download
41116_2022_35_Article_IEq45.gif 1KB Image download
41116_2022_35_Article_IEq46.gif 1KB Image download
41116_2022_35_Article_IEq47.gif 1KB Image download
40854_2022_439_Article_IEq324.gif 1KB Image download
MediaObjects/12936_2022_4438_MOESM3_ESM.xlsx 56KB Other download
Fig. 5 1054KB Image download
Fig. 3 511KB Image download
Fig. 7 544KB Image download
Fig. 5 2122KB Image download
Fig. 4 172KB Image download
Fig. 5 236KB Image download
Fig. 8 288KB Image download
40708_2022_182_Article_IEq12.gif 1KB Image download
Fig. 6 171KB Image download
MediaObjects/41408_2023_784_MOESM1_ESM.pdf 1800KB PDF download
40249_2022_1049_Article_IEq32.gif 1KB Image download
40249_2022_1049_Article_IEq33.gif 1KB Image download
Fig. 8 176KB Image download
1546KB Image download
Fig. 9 216KB Image download
40249_2022_1049_Article_IEq34.gif 1KB Image download
40249_2022_1049_Article_IEq35.gif 1KB Image download
Fig. 10 171KB Image download
Fig. 5 1065KB Image download
Fig. 11 206KB Image download
40249_2022_1049_Article_IEq38.gif 1KB Image download
40249_2022_1049_Article_IEq39.gif 1KB Image download
Fig. 12 1003KB Image download
Fig. 1 344KB Image download
40249_2022_1049_Article_IEq1117.gif 1KB Image download
40249_2022_1049_Article_IEq1118.gif 1KB Image download
Fig. 2 35KB Image download
40249_2022_1049_Article_IEq44.gif 1KB Image download
Fig. 3 159KB Image download
Fig. 1 264KB Image download
Fig. 4 215KB Image download
41116_2022_35_Article_IEq50.gif 1KB Image download
40249_2022_1049_Article_IEq2.gif 1KB Image download
41116_2022_35_Article_IEq51.gif 1KB Image download
【 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

【 参考文献 】
  • [1]
  • [2]
  • [3]
  • [4]
  • [5]
  • [6]
  • [7]
  • [8]
  • [9]
  • [10]
  • [11]
  • [12]
  • [13]
  • [14]
  • [15]
  • [16]
  • [17]
  • [18]
  • [19]
  • [20]
  • [21]
  • [22]
  • [23]
  • [24]
  • [25]
  • [26]
  • [27]
  • [28]
  • [29]
  • [30]
  • [31]
  • [32]
  • [33]
  • [34]
  • [35]
  • [36]
  • [37]
  • [38]
  • [39]
  • [40]
  • [41]
  • [42]
  文献评价指标  
  下载次数:2次 浏览次数:1次