Mean Residual Life Causal Models for Survival Data with a Binary Instrumental Variable

Abstract

The causal effect of a treatment on a survival outcome is often of fundamental interest in scientific investigations. In the presence of unmeasured confounding, the instrumental variable (IV) is a valuable tool for estimating causal effects. In this article, we focus on modeling and inferring a family of generalized mean residual life (MRL) causal models, which are easily interpretable, for censored data. Utilizing a special characterization of the binary IV, we propose weighted estimating equations to estimate treatment effect and regression coefficients in MRL casual models. The weights are estimated using either logistic regression or kernel smoothing techniques. We establish the asymptotic properties of the proposed estimators and conduct extensive simulation studies to evaluate the finite sample performance. Finally, the proposed approach is applied to a dataset from the US Renal Data System to evaluate the causal effect of peritoneal dialysis in end-stage renal disease patients.

Keywords

Causal treatment effect / Censored data / Estimating equations / Instrumental variable / Mean residual life model / 62N01 / 62N02

References

[1] AbadieA. Semiparametric instrumental variable estimation of treatment response models. J. Econom., 2003, 113(2): 231-263

[2] AbadieA, AngristJ, ImbensG. Instrumental variables estimates of the effect of subsidized training on the quantiles of trainee earnings. Econometrica, 2002, 70(1): 91-117

[3] AngristJD, ImbensGW, RubinDB. Identification of causal effects using instrumental variables. J. Am. Stat. Assoc., 1996, 91(434): 444-455

[4] BaiocchiM, ChengJ, SmallDS. Instrumental variable methods for causal inference. Stat. Med., 2014, 33(13): 2297-2340

[5] BakerSG. Analysis of survival data from a randomized trial with all-or-none compliance: estimating the cost-effectiveness of a cancer screening program. J. Am. Stat. Assoc., 1998, 93(443): 929-934

[6] BiliasY, GuMG, YingZL. Towards a general asymptotic theory for cox model with staggered entry. Ann. Stat., 1997, 25(2): 662-682

[7] CaoY, YuJ. Adjusting for unmeasured confounding in survival causal effect using validation data. Comput. Stat. Data Anal., 2023, 180107660

[8] CorteseG, HolmboeSA, ScheikeTH. Regression models for the restricted residual mean life for right-censored and left-truncated data. Stat. Med., 2017, 36(11): 1803-1822

[9] CuzickJ, SasieniP, MylesJ, TyrerJ. Estimating the effect of treatment in a proportional hazards model in the presence of non-compliance and contamination. J. R Stat. Soc. Series B, 2007, 69(4): 565-588

[10] DharmarajanSH, LiY, LehmannD, SchaubelDE. Weighted estimators of the complier average causal effect on restricted mean survival time with observed instrument-outcome confounders. Biom. J., 2021, 63(4): 712-724

[11] HeafJG, LøkkegaardH, MadsenM. Initial survival advantage of peritoneal dialysis relative to haemodialysis. Nephrol. Dial. Transplant., 2002, 17(1): 112-117

[12] JinP, Zeleniuch-JacquotteA, LiuML. Generalized mean residual life models for case-cohort and nested case-control studies. Lifetime Data Anal., 2020, 26: 789-819

[13] KianianB, KimJI, FineJP, PengL. Causal proportional hazards estimation with a binary instrumental variable. Stat. Sin., 2021, 31(2): 673-699

[14] KimH, KimK-H, ParkK, KangS-W, YooT, AhnSV, AhnHS, HannHJ, LeeS, RyuJ-H, KimS-J, KangD-H, ChoiKB, RyuD-R. A population-based approach indicates an overall higher patient mortality with peritoneal dialysis compared to hemodialysis in korea. Kidney Int., 2014, 86(5): 991-1000

[15] KumarVA, SidellMA, JonesJP, VoneshEF. Survival of propensity matched incident peritoneal and hemodialysis patients in a united states health care system. Kidney Int., 2014, 86(5): 1016-1022

[16] LiJ, FineJ, BrookhartA. Instrumental variable additive hazards models. Biometrics, 2015, 71(1): 122-130

[17] LiS, GrayRJ. Estimating treatment effect in a proportional hazards model in randomized clinical trials with all-or-nothing compliance. Biometrics, 2016, 72(3): 742-750

[18] LiG, LuX. A bayesian approach for instrumental variable analysis with censored time-to-event outcome. Stat. Med., 2015, 34: 664-684

[19] LiW, MaH, FaraggiD, DinseGE. Generalized mean residual life models for survival data with missing censoring indicators. Stat. Med., 2023, 42(3): 264-280

[20] LinDY, WeiLJ, YangI, YingZ. Semiparametric regression for the mean and rate functions of recurrent events. J. R Stat. Soc. Series B, 2000, 62: 711-730

[21] LinDY, WeiLJ, YingZ. Checking the Cox model with cumulative sums of martingale-based residuals. Biometrika, 1993, 80(3): 557-572

[22] LoeysT, GoetghebeurE. A causal proportional hazards estimator for the effect of treatment actually received in a randomized trial with all-or-nothing compliance. Biometrics, 2003, 59(1): 100-105

[23] MansourvarZ, MartinussenT. Estimation of average causal effect using the restricted mean residual lifetime as effect measure. Lifetime Data Anal., 2017, 23: 426-438

[24] MansourvarZ, MartinussenT, ScheikeTH. Semiparametric regression for restricted mean residual life under right censoring. J. Appl. Stat., 2015, 42(12): 2597-2613

[25] MansourvarZ, MartinussenT, ScheikeTH. An additive-multiplicative restricted mean residual life model. Scand. J. Stat., 2016, 43(2): 487-504

[26] MartinussenT, VansteelandtS, TchetgenE, ZuckerD. Instrumental variables estimation of exposure effects on a time-to-event response using structural cumulative survival models. Biometrics, 2017, 73: 1140-1149

[27] MaHJ, ZhaoW, ZhouY. Semiparametric model of mean residual life with biased sampling data. Comput. Stat. Data Anal., 2020, 142106826

[28] NieH, ChengJ, SmallDS. Inference for the effect of treatment on survival probability in randomized trials with noncompliance and administrative censoring. Biometrics, 2011, 67(4): 1397-1405

[29] PollardDEmpirical Processes: Theory and Applications, 1990, Hayward. Institute of Mathematical Statistics.

[30] RudinWDPrinciples of Mathematical Analysis, 1976, New York. McGraw-Hill.

[31] SoveyAJ, GreenDP. Instrumental variables estimation in political science: a readers’ guide. American J. Political Sci., 2011, 55(1): 188-200

[32] SunLQ, SongXY, ZhangZG. Mean residual life models with time-dependent coefficients under right censoring. Biometrika, 2012, 99(1): 185-197

[33] SunLQ, ZhangZG. A class of transformed mean residual life models with censored survival data. J. Am. Stat. Assoc., 2009, 104(486): 803-815

[34] ThieryA, SéveracF, HannedoucheT, CouchoudC, DoVH, TipleA, BéchadeC, SauleauE-A, KrummelT. REIN registry: Survival advantage of planned haemodialysis over peritoneal dialysis: a cohort study. Nephrol. Dial. Transplant., 2018, 33(8): 1411-1419

[35] WeiB, PengL, ZhangM-J, FineJP. Estimation of causal quantile effects with a binary instrumental variable and censored data. J. R Stat. Soc. Series B, 2021, 83(3): 559-578

[36] YangF, KhinL, LauT, ChuaH, VathsalaA, LeeE, LuoN. Hemodialysis versus peritoneal dialysis: A comparison of survival outcomes in south-east asian patients with end-stage renal disease. PLoS ONE, 2015, 10100140195

[37] YuW, ChenK, SobelME, YingZ. Semiparametric transformation models for causal inference in time-to-event studies with all-or-nothing compliance. J. R Stat. Soc. Series B, 2015, 77(2): 397-415

Funding

National Natural Science Foundation of China(71931004)

RIGHTS & PERMISSIONS

School of Mathematical Sciences, University of Science and Technology of China and Springer-Verlag GmbH Germany, part of Springer Nature