Data-adaptive methods have been proposed to estimate nuisance parameters when using doubly robust semiparametric methods for estimating marginal causal effects. However, in the presence of near practical positivity violations, these methods can produce a separation of the two exposure groups in terms of propensity score densities which can lead to biased estimates of the treatment effect. To motivate the problem, we evaluated the Targeted Minimum Loss-based Estimation procedure using a simulation scenario to estimate the average treatment effect. We highlight the divergence in estimates obtained when using parametric and data-adaptive methods to estimate the propensity score. We then adapted an existing diagnostic tool based on a bootstrap resampling of the subjects and simulation of the outcome data in order to show that the estimation using data-adaptive methods for the propensity score in this study may lead to large bias and poor coverage. The adapted bootstrap procedure is able to identify this instability and can be used as a diagnostic tool.
HernanMARobinsJM. Causal inference, Boca Raton, FL: Chapman and Hall-CRC, 2016.
2.
RosenbaumPRRubinDB. The central role of the propensity score in observational studies for causal effects. Biometrika1983; 70: 41–55.
3.
RobinsPRHernanMABrumbackB. Marginal structural models and causal inference in epidemiology. Am J Epidemiol2000; 11: 550–560.
4.
van der Laan MJ and Rose S. Targeted learning: causal inference for observational and experimental data. New York, NY: Springer Series in Statistics, 2011.
5.
ScharfsteinDORotnitzkyARobinsJM. Adjusting for nonignorable dropout using semiparametric nonresponse models, (with discussion and rejoinder). J Am Stat Assoc1999; 94: 1096–1120. 1121–1146.
6.
BangHRobinsJM. Doubly robust estimation in missing data and causal inference models. Biometrics2005; 61: 962972–962972.
7.
van der Laan MJ, Polley EC and Hubbard AE. Super Learner. U.C. Berkeley Division of Biostatistics Working Paper Series 2007; 222.
8.
PirracchioRPetersenMLvan der LaanMJ. Improving propensity score estimators robustness to model misspecification using super learner. Am J Epidemiol2015; 181: 108119–108119.
9.
SchnitzerMELokJJGruberS. Variable selection for confounder control, flexible modeling and collaborative targeted minimum loss-based estimation in causal inference. Int J Biostat2016; 12: 97–115.
10.
ColeMJHernanMA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol2008; 168: 656–664.
11.
PetersenMLPorterKEGruberS, et al.Diagnosing and responding to violations in the positivity assumption. Stat Meth Med Res2012; 21: 3154–3154.
12.
GruberSvan der LaanMJ. An application of collaborative targeted maximum likelihood estimation in causal inference and genomics. Int J Biostat2010; 6: 18–18.
13.
HorvitzDGThompsonDJ. A generalization of sampling without replacement from a finite universe. J Am Stat Assoc1952; 47: 663–685.
14.
RubinDB. Estimating causal effects of treatments in randomized and nonrandomized studies. J Educ Psychol1974; 66: 688–701.
15.
ColeSRFrangakisCE. The consistency statement in causal inference: a definition or an assumption?Epidemiology2009; 20: 3–5.
BembomOvan der LaanMJ. Data-adaptive selection of the truncation level for Inverse-probability-of-treatment-weighted estimators. U.C. Berkeley Division of Biostatistics Working Paper Series2008, pp. 230–230.
18.
van der LaanMJRubinD. Targeted maximum likelihood learning. Int J Biostat2006, pp. 2–2.
19.
GruberSvan der LaanMJ. A targeted maximum likelihood estimator of a causal effect on a bounded continuous outcome. Int J Biostat2010, pp. 6–6.
20.
Gruber S and van der Laan MJ. An R package for targeted maximum likelihood estimation. U.C. Berkeley Division of Biostatistics Working Paper Series, 2011. pp. 275.
PangMSchusterTFilionKB, et al.Targeted maximum likelihood estimation for pharmacoepidemiologic research. Epidemiology2016; 27: 570–577.
23.
GruberSvan der LaanMJ. Targeted maximum likelihood estimation: a gentle introduction. U.C. Berkeley Division of Biostatistics Working Paper Series2009, pp. 252–252.
van der LaanMJGruberS. Collaborative double robust targeted maximum likelihood estimation. Int J Biostat2010, pp. 6–6.
26.
MyersJARassenJAGagneJJ, et al.Effects of adjusting for instrumental variables on bias and precision of effect estimates. Am J Epidemiol2011; 174: 1213–1222.
27.
EfronB. Bootstrap methods: another look at the jackknife. Ann Stat1979; 7: 1–26.
28.
LendleSDFiremanBvan der LaanMJ. Targeted maximum likelihood estimation in safety analysis. J Clin Epidemiol2016; 6: 91–98.
29.
FranklinJMSchneeweissSPolinskiJM, et al.Plasmode simulation for the evaluation of pharmacoepidemiologic methods in complex healthcare databasess. Comput Stat Data Analysis2014; 72: 219–226.
30.
CossetteBForgetABeauchesneMF, et al.Impact of maternal use of asthma-controller therapy on perinatal outcomes. Thorax2013; 68: 724–730.
31.
FirooziFLemireCBeauchesneMF, et al.Development and validation of database indexes of asthma severity and control. Thorax2007; 62: 581587–581587.
32.
BreimanL. Random forests. Thorax Online2007; 45: 532–532.
33.
BlaisLBeauchesneMFReyE, et al.Use of inhaled corticosteroids duing the first trimester of pregnancy and the risk of congenital malformations among women with asthma. Thora Onlinex2007; 62: 320–328.
34.
Zheng W and van der Laan MJ. Cross-validated targeted minimum-loss-based estimation. In: Targeted learning: causal inference for observational and experimental data. New York, NY: Springer Series in Statistics, 2011, pp.459–474.
35.
CaroneMDiazIvan der LaanMJ. Higher-order targeted minimum loss-based estimation. U.C. Berkeley Division of Biostatistics Working Paper Series2014, pp. 331–331.
36.
BenkeserDCaroneMvan der LaanMJ, et al.Doubly robust nonparametric inference on the average treatment effect estimation. U.C. Berkeley Division of Biostatistics Working Paper Series2016, pp. 356–356.
