Comparison of statistical approaches dealing with time-dependent confounding in drug effectiveness studies

The BeAMS study group

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

In longitudinal studies, if the time-dependent covariates are affected by the past treatment, time-dependent confounding may be present. For a time-to-event response, marginal structural Cox models are frequently used to deal with such confounding. To avoid some of the problems of fitting marginal structural Cox model, the sequential Cox approach has been suggested as an alternative. Although the estimation mechanisms are different, both approaches claim to estimate the causal effect of treatment by appropriately adjusting for time-dependent confounding. We carry out simulation studies to assess the suitability of the sequential Cox approach for analyzing time-to-event data in the presence of a time-dependent covariate that may or may not be a time-dependent confounder. Results from these simulations revealed that the sequential Cox approach is not as effective as marginal structural Cox model in addressing the time-dependent confounding. The sequential Cox approach was also found to be inadequate in the presence of a time-dependent covariate. We propose a modified version of the sequential Cox approach that correctly estimates the treatment effect in both of the above scenarios. All approaches are applied to investigate the impact of beta-interferon treatment in delaying disability progression in the British Columbia Multiple Sclerosis cohort (1995–2008).

Original languageEnglish (US)
Pages (from-to)1709-1722
Number of pages14
JournalStatistical Methods in Medical Research
Volume27
Issue number6
DOIs
StatePublished - Jun 1 2018

Fingerprint

Confounding
Drugs
Time-dependent Covariates
Cox Model
Structural Model
Pharmaceutical Preparations
Structural Models
Proportional Hazards Models
Multiple Sclerosis
Causal Effect
Longitudinal Study
Disability
Treatment Effects
Progression
Estimate
Simulation Study
British Columbia
Interferon-beta
Scenarios
Alternatives

Keywords

  • Bias (epidemiology)
  • causality
  • confounding factors (epidemiology)
  • epidemiologic methods
  • inverse probability weighting
  • longitudinal studies
  • models
  • survival analysis

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability
  • Health Information Management

Cite this

Comparison of statistical approaches dealing with time-dependent confounding in drug effectiveness studies. / The BeAMS study group.

In: Statistical Methods in Medical Research, Vol. 27, No. 6, 01.06.2018, p. 1709-1722.

Research output: Contribution to journalArticle

@article{4348764d9e0a43f586aa4f1ae077af53,
title = "Comparison of statistical approaches dealing with time-dependent confounding in drug effectiveness studies",
abstract = "In longitudinal studies, if the time-dependent covariates are affected by the past treatment, time-dependent confounding may be present. For a time-to-event response, marginal structural Cox models are frequently used to deal with such confounding. To avoid some of the problems of fitting marginal structural Cox model, the sequential Cox approach has been suggested as an alternative. Although the estimation mechanisms are different, both approaches claim to estimate the causal effect of treatment by appropriately adjusting for time-dependent confounding. We carry out simulation studies to assess the suitability of the sequential Cox approach for analyzing time-to-event data in the presence of a time-dependent covariate that may or may not be a time-dependent confounder. Results from these simulations revealed that the sequential Cox approach is not as effective as marginal structural Cox model in addressing the time-dependent confounding. The sequential Cox approach was also found to be inadequate in the presence of a time-dependent covariate. We propose a modified version of the sequential Cox approach that correctly estimates the treatment effect in both of the above scenarios. All approaches are applied to investigate the impact of beta-interferon treatment in delaying disability progression in the British Columbia Multiple Sclerosis cohort (1995–2008).",
keywords = "Bias (epidemiology), causality, confounding factors (epidemiology), epidemiologic methods, inverse probability weighting, longitudinal studies, models, survival analysis",
author = "{The BeAMS study group} and Karim, {Mohammad Ehsanul} and John Petkau and Paul Gustafson and Platt, {Robert W.} and Helen Tremlett and A. Shirani and Y. Zhao and C. Evans and E. Kingwell and {van der Kop}, {M. L.} and J. Oger",
year = "2018",
month = "6",
day = "1",
doi = "10.1177/0962280216668554",
language = "English (US)",
volume = "27",
pages = "1709--1722",
journal = "Statistical Methods in Medical Research",
issn = "0962-2802",
publisher = "SAGE Publications Ltd",
number = "6",

}

TY - JOUR

T1 - Comparison of statistical approaches dealing with time-dependent confounding in drug effectiveness studies

AU - The BeAMS study group

AU - Karim, Mohammad Ehsanul

AU - Petkau, John

AU - Gustafson, Paul

AU - Platt, Robert W.

AU - Tremlett, Helen

AU - Shirani, A.

AU - Zhao, Y.

AU - Evans, C.

AU - Kingwell, E.

AU - van der Kop, M. L.

AU - Oger, J.

PY - 2018/6/1

Y1 - 2018/6/1

N2 - In longitudinal studies, if the time-dependent covariates are affected by the past treatment, time-dependent confounding may be present. For a time-to-event response, marginal structural Cox models are frequently used to deal with such confounding. To avoid some of the problems of fitting marginal structural Cox model, the sequential Cox approach has been suggested as an alternative. Although the estimation mechanisms are different, both approaches claim to estimate the causal effect of treatment by appropriately adjusting for time-dependent confounding. We carry out simulation studies to assess the suitability of the sequential Cox approach for analyzing time-to-event data in the presence of a time-dependent covariate that may or may not be a time-dependent confounder. Results from these simulations revealed that the sequential Cox approach is not as effective as marginal structural Cox model in addressing the time-dependent confounding. The sequential Cox approach was also found to be inadequate in the presence of a time-dependent covariate. We propose a modified version of the sequential Cox approach that correctly estimates the treatment effect in both of the above scenarios. All approaches are applied to investigate the impact of beta-interferon treatment in delaying disability progression in the British Columbia Multiple Sclerosis cohort (1995–2008).

AB - In longitudinal studies, if the time-dependent covariates are affected by the past treatment, time-dependent confounding may be present. For a time-to-event response, marginal structural Cox models are frequently used to deal with such confounding. To avoid some of the problems of fitting marginal structural Cox model, the sequential Cox approach has been suggested as an alternative. Although the estimation mechanisms are different, both approaches claim to estimate the causal effect of treatment by appropriately adjusting for time-dependent confounding. We carry out simulation studies to assess the suitability of the sequential Cox approach for analyzing time-to-event data in the presence of a time-dependent covariate that may or may not be a time-dependent confounder. Results from these simulations revealed that the sequential Cox approach is not as effective as marginal structural Cox model in addressing the time-dependent confounding. The sequential Cox approach was also found to be inadequate in the presence of a time-dependent covariate. We propose a modified version of the sequential Cox approach that correctly estimates the treatment effect in both of the above scenarios. All approaches are applied to investigate the impact of beta-interferon treatment in delaying disability progression in the British Columbia Multiple Sclerosis cohort (1995–2008).

KW - Bias (epidemiology)

KW - causality

KW - confounding factors (epidemiology)

KW - epidemiologic methods

KW - inverse probability weighting

KW - longitudinal studies

KW - models

KW - survival analysis

UR - http://www.scopus.com/inward/record.url?scp=85016026303&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85016026303&partnerID=8YFLogxK

U2 - 10.1177/0962280216668554

DO - 10.1177/0962280216668554

M3 - Article

C2 - 27659168

AN - SCOPUS:85016026303

VL - 27

SP - 1709

EP - 1722

JO - Statistical Methods in Medical Research

JF - Statistical Methods in Medical Research

SN - 0962-2802

IS - 6

ER -