Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Comparison of methods for analyzing longitudinal binary outcomes: cognitive status as an example.

M Kuchibhatla1, G G Fillenbaum

  • 1Center for Study of Aging and Human Development, Duke University Medical Center, Durham, NC 27710, USA. mnk@geri.duke.edu

Aging & Mental Health
|October 28, 2003
PubMed
Summary

Ignoring correlated data in longitudinal studies can lead to incorrect inferences. Generalized estimating equations (GEE) and random-intercept models offer accurate analysis of cognitive impairment in elders, unlike standard logistic regression.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Predicting bone scan positivity after biochemical recurrence following radical prostatectomy in both hormone-naive men and patients receiving androgen-deprivation therapy: results from the SEARCH database.

Prostate cancer and prostatic diseases·2014
Same author

Trajectory classes of depressive symptoms in a community sample of older adults.

Acta psychiatrica Scandinavica·2011
Same author

Increased health burden associated with comorbid depression in older Brazilians with diabetes.

Journal of affective disorders·2011
Same author

A prospective study of religion/spirituality and depressive symptoms among adolescent psychiatric patients.

Journal of affective disorders·2009
Same author

The effect of spousal caregiving and bereavement on depressive symptoms.

Aging & mental health·2008
Same author

The predictive value of self-report scales compared with physician diagnosis of depression in hemodialysis patients.

Kidney international·2006

Area of Science:

  • Biostatistics
  • Epidemiology
  • Gerontology

Background:

  • Longitudinal data analysis requires accounting for correlated observations.
  • Ignoring data correlation can lead to inaccurate standard error estimation and parameter inference.
  • Cognitive impairment in community-dwelling elders is a significant public health concern.

Purpose of the Study:

  • To compare the performance of standard logistic regression, population-averaged (PA) models using generalized estimating equations (GEE), and random-intercept models.
  • To assess how these models handle correlated binary outcomes in longitudinal studies of cognitive impairment.
  • To evaluate the impact of model choice on the estimation of time-invariant and time-varying covariates.

Main Methods:

  • Longitudinal binary outcomes (cognitive impairment) were modeled at baseline, 3, and 6 years.

Related Experiment Videos

  • Standard logistic regression, GEE (PA model), and random-intercept models were applied.
  • Time-invariant (age, gender) and time-varying (time, interactions) covariates were included.

    • Main Results:

    • Standard logistic regression overestimated standard errors for time-varying covariates and underestimated for time-invariant ones compared to GEE.
    • Random-intercept models yielded larger absolute estimates than logistic regression and GEE models.
    • Standard errors from random-intercept models were larger than those from logistic regression and GEE.

    Conclusions:

    • The choice between GEE and random-intercept models depends on the research question and covariate nature.
    • Population-averaged methods are suitable for between-subjects effects.
    • Random-effects models are appropriate for subject-specific effects in longitudinal cognitive impairment studies.