Time scale and adjusted survival curves for marginal structural Cox models

Daniel Westreich, Stephen R. Cole, Phyllis C. Tien, Joan S. Chmiel, Lawrence Kingsley, Michele Jonsson Funk, Kathryn Anastos, Lisa P. Jacobson

Research output: Contribution to journalArticle

39 Citations (Scopus)

Abstract

Typical applications of marginal structural time-to-event (e.g., Cox) models have used time on study as the time scale. Here, the authors illustrate use of time on treatment as an alternative time scale. In addition, a method is provided for estimating Kaplan-Meier-type survival curves for marginal structural models. For illustration, the authors estimate the total effect of highly active antiretroviral therapy on time to acquired immunodeficiency syndrome (AIDS) or death in 1,498 US men and women infected with human immunodeficiency virus and followed for 6,556 person-years between 1995 and 2002; 323 incident cases of clinical AIDS and 59 deaths occurred. Of the remaining 1,116 participants, 77% were still under observation at the end of follow-up. By using time on study, the hazard ratio for AIDS or death comparing always with never using highly active antiretroviral therapy from the marginal structural model was 0.52 (95% confidence interval: 0.35, 0.76). By using time on treatment, the analogous hazard ratio was 0.44 (95% confidence interval: 0.32, 0.60). In time-to-event analyses, the choice of time scale may have a meaningful impact on estimates of association and precision. In the present example, use of time on treatment yielded a hazard ratio further from the null and more precise than use of time on study as the time scale.

Original languageEnglish (US)
Pages (from-to)691-700
Number of pages10
JournalAmerican Journal of Epidemiology
Volume171
Issue number6
DOIs
StatePublished - Mar 2010

Fingerprint

Structural Models
Proportional Hazards Models
Survival
Acquired Immunodeficiency Syndrome
Highly Active Antiretroviral Therapy
Confidence Intervals
Kaplan-Meier Estimate
Therapeutics

Keywords

  • Acquired immunodeficiency syndrome
  • Antiretroviral therapy, highly active
  • Bias (epidemiology)
  • Causal inference
  • Confounding factors (epidemiology)
  • Proportional hazards model
  • Survival curve
  • Survival time

ASJC Scopus subject areas

  • Epidemiology

Cite this

Westreich, D., Cole, S. R., Tien, P. C., Chmiel, J. S., Kingsley, L., Funk, M. J., ... Jacobson, L. P. (2010). Time scale and adjusted survival curves for marginal structural Cox models. American Journal of Epidemiology, 171(6), 691-700. https://doi.org/10.1093/aje/kwp418

Time scale and adjusted survival curves for marginal structural Cox models. / Westreich, Daniel; Cole, Stephen R.; Tien, Phyllis C.; Chmiel, Joan S.; Kingsley, Lawrence; Funk, Michele Jonsson; Anastos, Kathryn; Jacobson, Lisa P.

In: American Journal of Epidemiology, Vol. 171, No. 6, 03.2010, p. 691-700.

Research output: Contribution to journalArticle

Westreich, D, Cole, SR, Tien, PC, Chmiel, JS, Kingsley, L, Funk, MJ, Anastos, K & Jacobson, LP 2010, 'Time scale and adjusted survival curves for marginal structural Cox models', American Journal of Epidemiology, vol. 171, no. 6, pp. 691-700. https://doi.org/10.1093/aje/kwp418
Westreich D, Cole SR, Tien PC, Chmiel JS, Kingsley L, Funk MJ et al. Time scale and adjusted survival curves for marginal structural Cox models. American Journal of Epidemiology. 2010 Mar;171(6):691-700. https://doi.org/10.1093/aje/kwp418
Westreich, Daniel ; Cole, Stephen R. ; Tien, Phyllis C. ; Chmiel, Joan S. ; Kingsley, Lawrence ; Funk, Michele Jonsson ; Anastos, Kathryn ; Jacobson, Lisa P. / Time scale and adjusted survival curves for marginal structural Cox models. In: American Journal of Epidemiology. 2010 ; Vol. 171, No. 6. pp. 691-700.
@article{3070672f2a0a43eaaaa47ce24fcbc9e1,
title = "Time scale and adjusted survival curves for marginal structural Cox models",
abstract = "Typical applications of marginal structural time-to-event (e.g., Cox) models have used time on study as the time scale. Here, the authors illustrate use of time on treatment as an alternative time scale. In addition, a method is provided for estimating Kaplan-Meier-type survival curves for marginal structural models. For illustration, the authors estimate the total effect of highly active antiretroviral therapy on time to acquired immunodeficiency syndrome (AIDS) or death in 1,498 US men and women infected with human immunodeficiency virus and followed for 6,556 person-years between 1995 and 2002; 323 incident cases of clinical AIDS and 59 deaths occurred. Of the remaining 1,116 participants, 77{\%} were still under observation at the end of follow-up. By using time on study, the hazard ratio for AIDS or death comparing always with never using highly active antiretroviral therapy from the marginal structural model was 0.52 (95{\%} confidence interval: 0.35, 0.76). By using time on treatment, the analogous hazard ratio was 0.44 (95{\%} confidence interval: 0.32, 0.60). In time-to-event analyses, the choice of time scale may have a meaningful impact on estimates of association and precision. In the present example, use of time on treatment yielded a hazard ratio further from the null and more precise than use of time on study as the time scale.",
keywords = "Acquired immunodeficiency syndrome, Antiretroviral therapy, highly active, Bias (epidemiology), Causal inference, Confounding factors (epidemiology), Proportional hazards model, Survival curve, Survival time",
author = "Daniel Westreich and Cole, {Stephen R.} and Tien, {Phyllis C.} and Chmiel, {Joan S.} and Lawrence Kingsley and Funk, {Michele Jonsson} and Kathryn Anastos and Jacobson, {Lisa P.}",
year = "2010",
month = "3",
doi = "10.1093/aje/kwp418",
language = "English (US)",
volume = "171",
pages = "691--700",
journal = "American Journal of Epidemiology",
issn = "0002-9262",
publisher = "Oxford University Press",
number = "6",

}

TY - JOUR

T1 - Time scale and adjusted survival curves for marginal structural Cox models

AU - Westreich, Daniel

AU - Cole, Stephen R.

AU - Tien, Phyllis C.

AU - Chmiel, Joan S.

AU - Kingsley, Lawrence

AU - Funk, Michele Jonsson

AU - Anastos, Kathryn

AU - Jacobson, Lisa P.

PY - 2010/3

Y1 - 2010/3

N2 - Typical applications of marginal structural time-to-event (e.g., Cox) models have used time on study as the time scale. Here, the authors illustrate use of time on treatment as an alternative time scale. In addition, a method is provided for estimating Kaplan-Meier-type survival curves for marginal structural models. For illustration, the authors estimate the total effect of highly active antiretroviral therapy on time to acquired immunodeficiency syndrome (AIDS) or death in 1,498 US men and women infected with human immunodeficiency virus and followed for 6,556 person-years between 1995 and 2002; 323 incident cases of clinical AIDS and 59 deaths occurred. Of the remaining 1,116 participants, 77% were still under observation at the end of follow-up. By using time on study, the hazard ratio for AIDS or death comparing always with never using highly active antiretroviral therapy from the marginal structural model was 0.52 (95% confidence interval: 0.35, 0.76). By using time on treatment, the analogous hazard ratio was 0.44 (95% confidence interval: 0.32, 0.60). In time-to-event analyses, the choice of time scale may have a meaningful impact on estimates of association and precision. In the present example, use of time on treatment yielded a hazard ratio further from the null and more precise than use of time on study as the time scale.

AB - Typical applications of marginal structural time-to-event (e.g., Cox) models have used time on study as the time scale. Here, the authors illustrate use of time on treatment as an alternative time scale. In addition, a method is provided for estimating Kaplan-Meier-type survival curves for marginal structural models. For illustration, the authors estimate the total effect of highly active antiretroviral therapy on time to acquired immunodeficiency syndrome (AIDS) or death in 1,498 US men and women infected with human immunodeficiency virus and followed for 6,556 person-years between 1995 and 2002; 323 incident cases of clinical AIDS and 59 deaths occurred. Of the remaining 1,116 participants, 77% were still under observation at the end of follow-up. By using time on study, the hazard ratio for AIDS or death comparing always with never using highly active antiretroviral therapy from the marginal structural model was 0.52 (95% confidence interval: 0.35, 0.76). By using time on treatment, the analogous hazard ratio was 0.44 (95% confidence interval: 0.32, 0.60). In time-to-event analyses, the choice of time scale may have a meaningful impact on estimates of association and precision. In the present example, use of time on treatment yielded a hazard ratio further from the null and more precise than use of time on study as the time scale.

KW - Acquired immunodeficiency syndrome

KW - Antiretroviral therapy, highly active

KW - Bias (epidemiology)

KW - Causal inference

KW - Confounding factors (epidemiology)

KW - Proportional hazards model

KW - Survival curve

KW - Survival time

UR - http://www.scopus.com/inward/record.url?scp=77949401712&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77949401712&partnerID=8YFLogxK

U2 - 10.1093/aje/kwp418

DO - 10.1093/aje/kwp418

M3 - Article

C2 - 20139124

AN - SCOPUS:77949401712

VL - 171

SP - 691

EP - 700

JO - American Journal of Epidemiology

JF - American Journal of Epidemiology

SN - 0002-9262

IS - 6

ER -