Alternative Design and Analytical Techniques for Longitudinal Rheumatology Studies: Improved Understanding of Outcomes

https://doi.org/10.1016/j.rdc.2018.01.001Get rights and content

Section snippets

Key points

  • Longitudinal cohort studies (with three or more measurement occasions) enable researchers to examine between- and within-individual variation, providing an improved understanding of disease evolution.

  • Alternative longitudinal study designs, such as the accelerated cohort, two-method measurement approach, and multiform design, increase efficiency of longitudinal designs by reducing time to research output and participant burden, while maintaining statistical power.

  • Longitudinally collected

Alternative longitudinal designs using primary data

Planned missing data (PMD) designs with primary data collection include accelerated cohort, two-method, and multiform designs. These designs are efficient because they rely on strategically placed missing data, meaning that participants do not have the same measurement schedule.3, 6 Missing data in PMD designs are missing completely at random; accordingly, their absence does not result in biased study conclusions.13, 15 By reducing response burden on participants, PMD designs can potentially

Alternative longitudinal design using secondary data

Administrative data are collected for managing and monitoring the health care system and not for research purposes. Examples of administrative data are records of physician billing claims, hospitalizations, and emergency department visits. These data are usually collected by the government to produce official statistics.26 However, they are a potentially valuable resource for observational, longitudinal studies about chronic diseases. Rheumatologists are familiar with the use of administrative

Alternative longitudinal analytical methods

In this section, we discuss several examples of newer (but less commonly used) longitudinal models that enable researchers to make full use of all available data. These models are used in longitudinal cohort studies to address long-term therapeutic outcomes (marginal structural modeling); recurrent events, such as flares of disease (recurrent event modeling); and progression through various stages of disease (multistate modeling). All of these models have in common the ability to evaluate

Summary

We have shown in this review how alternative PMD designs are used to accelerate the time to research output, collect more information, and maintain or sometimes increase statistical power. We have also shown how advanced statistical models are used to provide estimations of unbiased treatment effects through adjustments of time-varying covariates, mediators, and confounders.45 Methods that use all available disease course data allowing the modeling of recurrent events or multistate events were

First page preview

First page preview
Click to open first page preview

References (45)

  • D.P. Farrington

    Longitudinal research strategies: advantages, problems, and prospects

    J Am Acad Child Adolesc Psychiatry

    (1991)
  • L.L. Roos et al.

    Using administrative data for longitudinal research: comparisons with primary data collection

    J Chronic Dis

    (1987)
  • J.D. Singer et al.

    Applied longitudinal analysis

    (2003)
  • S. Galbraith et al.

    Accelerated longitudinal designs: an overview of modelling, power, costs and handling missing data

    Stat Methods Med Res

    (2017)
  • M. Garnier-Villarreal et al.

    Two-method planned missing designs for longitudinal research

    Int J Behav Dev

    (2014)
  • J.W. Graham

    Missing data: analysis and design

    (2012)
  • J.W. Graham et al.

    Planned missing data designs in psychological research

    Psychol Methods

    (2006)
  • W. Wu et al.

    Search for efficient complete and planned missing data designs for analysis of change

    Behav Res Methods

    (2016)
  • L.D. Amorim et al.

    Modelling recurrent events: a tutorial for analysis in epidemiology

    Int J Epidemiol

    (2015)
  • J.M. Robins et al.

    Marginal structural models and causal inference in epidemiology

    Epidemiology

    (2000)
  • R. Sutradhar et al.

    A Markov multistate analysis of the relationship between performance status and death among an ambulatory population of cancer patients

    Palliat Med

    (2014)
  • Z. Guo et al.

    Modeling repeated time-to-event health conditions with discontinuous risk intervals: an example of a longitudinal study of functional disability among older persons

    Methods Inf Med

    (2008)
  • J.G. Hanly et al.

    A longitudinal analysis of outcomes of lupus nephritis in an International Inception Cohort using a multistate model approach

    Arthritis Rheumatol

    (2016)
  • S.Y. Tian et al.

    Comparative effectiveness of mycophenolate mofetil for the treatment of childhood-onset proliferative lupus nephritis

    Arthritis Care Res (Hoboken)

    (2017)
  • R.J. Little

    Methods for handling missing values in clinical trials

    J Rheumatol

    (1999)
  • J.W. Graham

    Missing data analysis: making it work in the real world

    Annu Rev Psychol

    (2009)
  • M. Rhemtulla et al.

    Tools of the trade: planned missing data designs for research in cognitive development

    J Cogn Dev

    (2012)
  • J.R. Nesselroade et al.

    Longitudinal research in the study of behavior and development

    (1979)
  • M. Moerbeek

    The effects of the number of cohorts, degree of overlap among cohorts, and frequency of observation on power in accelerated longitudinal designs

    Methodology

    (2011)
  • R.B. Kline

    Software review: software programs for structural equation modeling: Amos, EQS, and LISREL

    J Psychoeduc Assess

    (1998)
  • R.B. Kline

    Principles and practice of structural equation modeling

    (2015)
  • Disclosure: None of the authors has pertinent conflicts of interest.

    View full text