Missing Data in Marginal Structural Models A Plasmode Simulation Study Comparing Multiple Imputation and Inverse Probability Weighting
Missing Data in Marginal Structural Models A Plasmode Simulation Study Comparing Multiple Imputation and Inverse Probability Weighting
复制标题
DOI：
10.1097/mlr.0000000000001063
复制
发表时间：
2019-03-01
期刊：
影响因子：
3
通讯作者：
Lapane, Kate L.
中科院分区：
文献类型：
--
作者：
Liu, Shao-Hsien;Chrysanthopoulou, Stavroula A.;Lapane, Kate L.
背景：流行病学研究中使用边缘结构模型（MSM）来调整时变混杂因素的情况有所增加。然而，在MSM的背景下，关于如何最好地处理缺失数据的建议是矛盾的。我们提出了一个plasmode模拟研究，比较的有效性和精度的MSM估计使用完整的情况下分析（CC），多重插补（MI），和逆概率加权（IPW）的缺失数据的存在下，时间无关的和随时间变化的混杂因素。材料与方法：模拟是基于一项队列子研究，使用来自骨关节炎倡议的数据，估计关节内注射使用对膝关节疼痛年变化的边际因果效应。我们模拟了81个场景，参数值因缺失机制（MCAR、MAR和MNAR）、缺失百分比（10%、20%和30%）、混杂因素类型（时间无关、时变、一种或两种）和分析方法（CC、IPW和MI）而异。CC，IPW和MI方法的性能进行了比较，使用相对偏差，均方误差的估计的利益，和经验的权力。结果如下：在由缺失数据机制、缺失数据程度和混杂因素类型定义的情景中，与IPW（相对偏倚：-5.3%至8.0%;精密度：0.19-0.53）相比，MI通常产生偏倚较小的估计值（范围：1.2%-6.7%）和更好的精密度（范围：0.17-0.18）。经验功效在使用MI的情景中是恒定的。结论：在简单而现实的情况下，MI似乎赋予IPW在MSM应用程序的优势。
Background: The use of marginal structural models (MSMs) to adjust for time-varying confounding has increased in epidemiologic studies. However, in the setting of MSMs, recommendations for how best to handle missing data are contradictory. We present a plasmode simulation study to compare the validity and precision of MSMs estimates using complete case analysis (CC), multiple imputation (MI), and inverse probability weighting (IPW) in the presence of missing data on time-independent and time-varying confounders. Materials and Methods: Simulations were based on a cohort substudy using data from the Osteoarthritis Initiative which estimated the marginal causal effect of intra-articular injection use on yearly changes in knee pain. We simulated 81 scenarios with parameter values varied on missing mechanisms (MCAR, MAR, and MNAR), percentages of missing (10%, 20%, and 30%), type of confounders (time-independent, time-varying, either or both), and analytical approaches (CC, IPW, and MI). The performance of CC, IPW, and MI methods was compared using relative bias, mean squared error of the estimates of interest, and empirical power. Results: Across scenarios defined by missing data mechanism, extent of missing data, and confounder type, MI generally produced less biased estimates (range: 1.2%-6.7%) with better precision (range: 0.17-0.18) compared with IPW (relative bias: -5.3% to 8.0%; precision: 0.19-0.53). Empirical power was constant across the scenarios using MI. Conclusions: Under simple yet realistically constructed scenarios, MI seems to confer an advantage over IPW in MSMs applications.