Abstract–In Clinical Trials, not all randomized patients follow the course of treatment they are allocated to. The potential impact of such deviations is increasingly recognized, and it has been one of the reasons for a redefinition of the targets of estimation (“Estimands”) in the ICH E9 draft Addendum. Among others, the effect of treatment assignment, regardless of the adherence, appears an Estimand of practical interest, in line with the intention-to-treat principle. This study aims at evaluating the performance of different estimation techniques in trials with incomplete post-discontinuation follow-up when a “treatment-policy” strategy is implemented. To achieve that, we have (i) modeled and visualized as directed acyclic diagram a reasonable data-generating model; (ii) investigated which set of variables allows identification and estimation of such effect; (iii) simulated 10,000 trials in Major Depressive Disorder, with varying real treatment effects, proportions of patients discontinuing the treatment, and incomplete follow-up. Our results suggest that, at least in a “Missing at Random” setting, all studied estimation methods increase their performance when a variable representing compliance is used. This effect is more pronounced the higher the proportion of post-discontinuation follow-up is.

The Use of a Variable Representing Compliance Improves Accuracy of Estimation of the Effect of Treatment Allocation Regardless of Discontinuation in Trials with Incomplete Follow-up

Gallo C.
Supervision
2020

Abstract

Abstract–In Clinical Trials, not all randomized patients follow the course of treatment they are allocated to. The potential impact of such deviations is increasingly recognized, and it has been one of the reasons for a redefinition of the targets of estimation (“Estimands”) in the ICH E9 draft Addendum. Among others, the effect of treatment assignment, regardless of the adherence, appears an Estimand of practical interest, in line with the intention-to-treat principle. This study aims at evaluating the performance of different estimation techniques in trials with incomplete post-discontinuation follow-up when a “treatment-policy” strategy is implemented. To achieve that, we have (i) modeled and visualized as directed acyclic diagram a reasonable data-generating model; (ii) investigated which set of variables allows identification and estimation of such effect; (iii) simulated 10,000 trials in Major Depressive Disorder, with varying real treatment effects, proportions of patients discontinuing the treatment, and incomplete follow-up. Our results suggest that, at least in a “Missing at Random” setting, all studied estimation methods increase their performance when a variable representing compliance is used. This effect is more pronounced the higher the proportion of post-discontinuation follow-up is.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11591/437985
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 8
  • ???jsp.display-item.citation.isi??? 8
social impact