Inverse probability weighting to estimate causal effect of a singular phase in a multiphase randomized clinical trial for multiple myeloma

Annalisa Pezzi, Michele Cavo, Annibale Biggeri, Elena Zamagni, Oriana Nanni, Annalisa Pezzi, Michele Cavo, Annibale Biggeri, Elena Zamagni, Oriana Nanni

Abstract

Background: Randomization procedure in randomized controlled trials (RCTs) permits an unbiased estimation of causal effects. However, in clinical practice, differential compliance between arms may cause a strong violation of randomization balance and biased treatment effect among those who comply. We evaluated the effect of the consolidation phase on disease-free survival of patients with multiple myeloma in an RCT designed for another purpose, adjusting for potential selection bias due to different compliance to previous treatment phases.

Methods: We computed two propensity scores (PS) to model two different selection processes: the first to undergo autologous stem cell transplantation, the second to begin consolidation therapy. Combined stabilized inverse probability treatment weights were then introduced in the Cox model to estimate the causal effect of consolidation therapy miming an ad hoc RCT protocol.

Results: We found that the effect of consolidation therapy was restricted to the first 18 months of the phase (HR: 0.40, robust 95 % CI: 0.17-0.96), after which it disappeared.

Conclusions: PS-based methods could be a complementary approach within an RCT context to evaluate the effect of the last phase of a complex therapeutic strategy, adjusting for potential selection bias caused by different compliance to the previous phases of the therapeutic scheme, in order to simulate an ad hoc randomization procedure.

Trial registration: ClinicalTrials.gov: NCT01134484 May 28, 2010 (retrospectively registered) EudraCT: 2005-003723-39 December 17, 2008 (retrospectively registered).

Keywords: Causal effect; Compliance; Propensity score; RCT; Selection bias; Weighting sample.

Figures

Fig. 1
Fig. 1
a Study design of a phase III open-label RCT carried out in 73 Italian hospitals. Eligible untreated symptomatic multiple myeloma patients aged 18–65 years were randomized (1:1 ratio) to receive experimental (Arm A) versus standard (Arm B) treatment as induction therapy before a maximum of two planned autologous stem cell transplantations (ASCT) followed by a consolidation phase consisting on the same arm of therapy as induction phase. b Miming an Ad hoc RCT to evaluate the role of consolidation therapy. Eligible untreated symptomatic multiple myeloma patients aged 18–65 years who had received at least 1 ASCT after having been prepared with induction therapy (Experimental or Standard) were randomized to receive the same arm of therapy (Experimental or Standard) as a consolidation phase or to not receive any therapy
Fig. 2
Fig. 2
Weighted Kaplan-Meier survival estimates for PFS from last ASCT evaluation date, by consolidation treatment
Fig. 3
Fig. 3
Plot of the cumulative regression coefficient (95 % CI) for the consolidation phase as a function of follow-up time. Aalen’s additive hazard model of progression-free survival from last autologous stem cell transplantation

References

    1. Armitage P. The role of randomization in clinical trials. Stat Med. 1982;4:345–52. doi: 10.1002/sim.4780010412.
    1. Hernan MA, Hernandez-Diaz S, Robins JM. A structural approach to selection bias. Epidemiology. 2004;15(5):615–625. doi: 10.1097/01.ede.0000135174.63482.43.
    1. Frangakis CE, Rubin DB. Addressing complications of intention-to treat analysis in the combined presence of all-or-none treatment-noncompliance and subsequent missing outcomes. Biometrika. 1999;86:365–379. doi: 10.1093/biomet/86.2.365.
    1. Mealli F, Rubin D. Assumptions when Analyzing Randomized Experiments with Noncompliance and missing Outcomes. Health Serv Outcome Res Methodol. 2002;3:225–232. doi: 10.1023/A:1025802028890.
    1. Mealli F, Imbens GW, Ferro S, Biggeri A. Analyzing a randomized trial on breast self-examination with noncompliance and missing outcomes. Biostatistics. 2004;5:2,207–22. doi: 10.1093/biostatistics/5.2.207.
    1. Rosenbaum P, Rubin D. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70:41–55. doi: 10.1093/biomet/70.1.41.
    1. Cavo M, Brioli A, Tacchetti P, Zannetti BA, Mancuso K, Zamagni E. Role of consolidation therapy in transplant eligible multiple myeloma patients. Semin Oncol. 2013;40(5):610–7. doi: 10.1053/j.seminoncol.2013.07.001.
    1. D’Agostino RB. Tutorial in biostatistics. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Statist Med. 1998;17:2265–2281. doi: 10.1002/(SICI)1097-0258(19981015)17:19<2265::AID-SIM918>;2-B.
    1. Becker SO, Ichino A. Estimation of average treatment effects based on propensity scores. Stata J. 2002;2(4):358–377.
    1. Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Sturmer T. Variable selection for propensity score models. Am J Epidemiol. 2006;163(12):1149–56. doi: 10.1093/aje/kwj149.
    1. Pearl J. Letter to the editor. Remarks on the method of propensity score. Statist Med. 2009;28:1415–1424. doi: 10.1002/sim.3521.
    1. Guo SY, Fraser MW. Propensity Score Analysis: Statistical Methods and Applications, Second Edition. Sage; 2015. .
    1. Austin P. The performance of different propensity score methods for estimating marginal hazard ratios. Statist Med. 2013;32(16):2837–2849. doi: 10.1002/sim.5705.
    1. Austin P. The use of propensity score methods with survival or time-to-event outcomes: reporting measures of effect similar to those used in randomized experiments. Statist Med. 2014;33(7):1242–1258. doi: 10.1002/sim.5984.
    1. Xu S, Ross C, Raebel MA, Shetterly S, Blanchette C, Smith D. Use of stabilized inverse propensity score as weight to directly estimate relative risk and its confidence intervals. Value Health. 2010;2(13):273–7. doi: 10.1111/j.1524-4733.2009.00671.x.
    1. Hernan MA, Robins J. Estimating causal effects from epidemiological data. J Epidemiol Community Health. 2006;60:578–586. doi: 10.1136/jech.2004.029496.
    1. Cole SR, Hernan MA. Constructing Inverse Probability Weights for Marginal Structural models. Am J Epidemiol. 2008;168(6):656–664. doi: 10.1093/aje/kwn164.
    1. Westreich D, Cole SR, Schisterman EF, Platt RW. A simulation study of finite-sample properties of marginal structural Cox proportional hazards models. Statist Med. 2012;31(19):2098–2109. doi: 10.1002/sim.5317.
    1. Cole SR, Hernan MA. Adjusted survival curves with inverse probability weights. Comput Methods Prog Biomed. 2004;75:45–9. doi: 10.1016/j.cmpb.2003.10.004.
    1. Xie J, Liu C. Adjusted Kaplan-Meier Estimator and Log-rank Test with Inverse Probability of Treatment Weighting for Survival Data. Statist. Med. 2005;24(20):3089–110. doi: 10.1002/sim.2174.
    1. Xu S, Shetterly S, Powers D, Raebel MA, Tsai T, Ho PM, Magic D. Extension of Kaplan-Meier methods in observational studies with time-varying treatment. Value Health. 2012;15(1):167–74.
    1. Lin DY, Wei LJ. The robust inference for the proportional hazards model. J Am Stat Assoc. 1989;84:1074–8. doi: 10.1080/01621459.1989.10478874.
    1. Hosmer DW, Royston P. Using Aalen’s linear hazards model to investigate time-varying effects in the proportional hazards regression model. Stata J. 2002;2(4):331–350.
    1. Huffer FW, McKeague IW. Weighted least squares estimation of Aalen’s additive risk model. J Am Stat Assoc. 1991;86:114–129. doi: 10.1080/01621459.1991.10475010.
    1. Cavo M, Tacchetti P, Patriarca F, Petrucci MT, Pantani L, Galli M, et al. Bortezomib with thalidomide plus dexamethasone compared with thalidomide plus dexamethasone as induction therapy before, and consolidation therapy after, double autologous stem-cell transplantation in newly diagnosed multiple myeloma: a randomised phase 3 study. Lancet. 2010;376(9758):2075–85. doi: 10.1016/S0140-6736(10)61424-9.

Source: PubMed

Sponsorit ja yhteistyökumppanit

Sairaudet

Huumeiden interventiot

3
Tilaa