Analysing trajectories of a longitudinal exposure: a causal perspective on common methods in lifecourse research

Sarah C. Gadd (Lead / Corresponding author), Peter W. G. Tennant, Alison J. Heppenstall, Jan Boehnke, Mark S. Gilthorpe

    Research output: Contribution to journalArticle

    22 Downloads (Pure)

    Abstract

    Longitudinal data is commonly analysed to inform prevention policies for diseases that may develop throughout life. Commonly methods interpret the longitudinal data as a series of discrete measurements or as continuous patterns. Some of the latter methods condition on the outcome, aiming to capture ‘average’ patterns within outcome groups, while others capture individual-level pattern features before relating these to the outcome. Conditioning on the outcome may prevent meaningful interpretation. Repeated measurements of a longitudinal exposure (weight) and later outcome (glycated haemoglobin levels) were simulated to match three scenarios: one with no causal relationship between growth rate and glycated haemoglobin; two with a positive causal effect of growth rate on glycated haemoglobin. Two methods that condition on the outcome and one that did not were applied to the data in 1000 simulations. The interpretation of the two-step method matched the simulation in all causal scenarios, but that of the methods conditioning on the outcome did not. Methods that condition on the outcome do not accurately represent a causal relationship between a longitudinal pattern and outcome. Researchers considering longitudinal data should carefully determine if they wish to analyse longitudinal data as a series of discrete time points or by extracting pattern features.
    Original languageEnglish
    Article numbere0225217
    Pages (from-to)1-12
    Number of pages12
    JournalPLoS ONE
    Volume14
    Issue number12
    DOIs
    Publication statusPublished - 4 Dec 2019

    Fingerprint

    Glycosylated Hemoglobin A
    research methods
    trajectories
    Trajectories
    glycohemoglobin
    Research
    methodology
    Growth
    researchers
    Research Personnel
    Weights and Measures

    Cite this

    Gadd, Sarah C. ; Tennant, Peter W. G. ; Heppenstall, Alison J. ; Boehnke, Jan ; Gilthorpe, Mark S. / Analysing trajectories of a longitudinal exposure : a causal perspective on common methods in lifecourse research. In: PLoS ONE. 2019 ; Vol. 14, No. 12. pp. 1-12.
    @article{91b1491980a7496caa07a2b9d1ff9dab,
    title = "Analysing trajectories of a longitudinal exposure: a causal perspective on common methods in lifecourse research",
    abstract = "Longitudinal data is commonly analysed to inform prevention policies for diseases that may develop throughout life. Commonly methods interpret the longitudinal data as a series of discrete measurements or as continuous patterns. Some of the latter methods condition on the outcome, aiming to capture ‘average’ patterns within outcome groups, while others capture individual-level pattern features before relating these to the outcome. Conditioning on the outcome may prevent meaningful interpretation. Repeated measurements of a longitudinal exposure (weight) and later outcome (glycated haemoglobin levels) were simulated to match three scenarios: one with no causal relationship between growth rate and glycated haemoglobin; two with a positive causal effect of growth rate on glycated haemoglobin. Two methods that condition on the outcome and one that did not were applied to the data in 1000 simulations. The interpretation of the two-step method matched the simulation in all causal scenarios, but that of the methods conditioning on the outcome did not. Methods that condition on the outcome do not accurately represent a causal relationship between a longitudinal pattern and outcome. Researchers considering longitudinal data should carefully determine if they wish to analyse longitudinal data as a series of discrete time points or by extracting pattern features.",
    author = "Gadd, {Sarah C.} and Tennant, {Peter W. G.} and Heppenstall, {Alison J.} and Jan Boehnke and Gilthorpe, {Mark S.}",
    year = "2019",
    month = "12",
    day = "4",
    doi = "10.1371/journal.pone.0225217",
    language = "English",
    volume = "14",
    pages = "1--12",
    journal = "PLoS ONE",
    issn = "1932-6203",
    publisher = "Public Library of Science",
    number = "12",

    }

    Analysing trajectories of a longitudinal exposure : a causal perspective on common methods in lifecourse research. / Gadd, Sarah C. (Lead / Corresponding author); Tennant, Peter W. G.; Heppenstall, Alison J.; Boehnke, Jan; Gilthorpe, Mark S.

    In: PLoS ONE, Vol. 14, No. 12, e0225217, 04.12.2019, p. 1-12.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Analysing trajectories of a longitudinal exposure

    T2 - a causal perspective on common methods in lifecourse research

    AU - Gadd, Sarah C.

    AU - Tennant, Peter W. G.

    AU - Heppenstall, Alison J.

    AU - Boehnke, Jan

    AU - Gilthorpe, Mark S.

    PY - 2019/12/4

    Y1 - 2019/12/4

    N2 - Longitudinal data is commonly analysed to inform prevention policies for diseases that may develop throughout life. Commonly methods interpret the longitudinal data as a series of discrete measurements or as continuous patterns. Some of the latter methods condition on the outcome, aiming to capture ‘average’ patterns within outcome groups, while others capture individual-level pattern features before relating these to the outcome. Conditioning on the outcome may prevent meaningful interpretation. Repeated measurements of a longitudinal exposure (weight) and later outcome (glycated haemoglobin levels) were simulated to match three scenarios: one with no causal relationship between growth rate and glycated haemoglobin; two with a positive causal effect of growth rate on glycated haemoglobin. Two methods that condition on the outcome and one that did not were applied to the data in 1000 simulations. The interpretation of the two-step method matched the simulation in all causal scenarios, but that of the methods conditioning on the outcome did not. Methods that condition on the outcome do not accurately represent a causal relationship between a longitudinal pattern and outcome. Researchers considering longitudinal data should carefully determine if they wish to analyse longitudinal data as a series of discrete time points or by extracting pattern features.

    AB - Longitudinal data is commonly analysed to inform prevention policies for diseases that may develop throughout life. Commonly methods interpret the longitudinal data as a series of discrete measurements or as continuous patterns. Some of the latter methods condition on the outcome, aiming to capture ‘average’ patterns within outcome groups, while others capture individual-level pattern features before relating these to the outcome. Conditioning on the outcome may prevent meaningful interpretation. Repeated measurements of a longitudinal exposure (weight) and later outcome (glycated haemoglobin levels) were simulated to match three scenarios: one with no causal relationship between growth rate and glycated haemoglobin; two with a positive causal effect of growth rate on glycated haemoglobin. Two methods that condition on the outcome and one that did not were applied to the data in 1000 simulations. The interpretation of the two-step method matched the simulation in all causal scenarios, but that of the methods conditioning on the outcome did not. Methods that condition on the outcome do not accurately represent a causal relationship between a longitudinal pattern and outcome. Researchers considering longitudinal data should carefully determine if they wish to analyse longitudinal data as a series of discrete time points or by extracting pattern features.

    UR - http://www.scopus.com/inward/record.url?scp=85075979872&partnerID=8YFLogxK

    U2 - 10.1371/journal.pone.0225217

    DO - 10.1371/journal.pone.0225217

    M3 - Article

    C2 - 31800576

    VL - 14

    SP - 1

    EP - 12

    JO - PLoS ONE

    JF - PLoS ONE

    SN - 1932-6203

    IS - 12

    M1 - e0225217

    ER -