IADR Abstract Archives
Methods for Restrospective Harmonisation of Longitudinal Cohort Data.
Objectives: To demonstrate methods for retrospective harmonisation of longitudinal cohort survey data describing oral health trends in Irish children.
Methods: Data from participants of all three waves (2008-2016) of the longitudinal survey of children in Ireland (child cohort ’98 at 9, 13 and 17/18 years of age, N=6,039) were converted, joined and standardised using the R language. Variable selection and renaming were completed to facilitate their use programmatically. Selected missing variables were added by computing to ensure comparison across waves. Common data wrangling problems with categorical variables including changing the value or order of response levels were completed using the forcats package.
Results: Barplots of categorical variables and density curves of continuous variables were created to illustrate the effect of remapping variable levels or renaming-creating variables for longitudinal comparison. Visualisation of selected oral health variables before and after remapping demonstrated the effects of data harmonisation. A consistent traceable mapping record facilitated further utilisation of data when generating dashboards.
Conclusions: Data harmonisation greatly improves the ability to compare oral health trends in longitudinal surveys.
Methods: Data from participants of all three waves (2008-2016) of the longitudinal survey of children in Ireland (child cohort ’98 at 9, 13 and 17/18 years of age, N=6,039) were converted, joined and standardised using the R language. Variable selection and renaming were completed to facilitate their use programmatically. Selected missing variables were added by computing to ensure comparison across waves. Common data wrangling problems with categorical variables including changing the value or order of response levels were completed using the forcats package.
Results: Barplots of categorical variables and density curves of continuous variables were created to illustrate the effect of remapping variable levels or renaming-creating variables for longitudinal comparison. Visualisation of selected oral health variables before and after remapping demonstrated the effects of data harmonisation. A consistent traceable mapping record facilitated further utilisation of data when generating dashboards.
Conclusions: Data harmonisation greatly improves the ability to compare oral health trends in longitudinal surveys.