IADR Abstract Archives
Data Harmonisation of oral health-related data: Exploring Possibilities in National Child Cohort Surveys
Objectives: Data harmonisation is used to improve variable comparability, reduce heterogeneity across multiple studies and increase data discovery. This study describes the process used to evaluate the harmonisation potential of oral health-related variables across and within national longitudinal birth and child cohort surveys.
Methods: An oral health-related data model (three-tier nested structure) was developed using questionnaire data from multiple nationally representative birth and child cohort surveys. The Maelstrom Research Guidelines for harmonisation evaluation potential were followed. An oral health-related DataSchema was created using a consensus approach.
Results: Seven nationally representative child cohort surveys were included, incorporating 50 survey waves. A common data model consisted of three domains and fifteen constructs. An oral health and risk indicator questionnaire DataSchema was compiled with 42 variables. For each study wave, the potential (or not) to generate each DataSchema variable was evaluated. Approximately 50% of the DataSchema variables could be generated across at least 4 cohort surveys, while only 10% (n = 4) of variables could be generated across all surveys. For individual surveys, the DataSchema variables that could be generated ranged between 26% and 76%.
Conclusions: The harmonisation potential of child surveys was found to be low across multiple surveys. More consistency and standardisation in survey questionnaires within and between surveys is advocated to help improve data discovery.
Methods: An oral health-related data model (three-tier nested structure) was developed using questionnaire data from multiple nationally representative birth and child cohort surveys. The Maelstrom Research Guidelines for harmonisation evaluation potential were followed. An oral health-related DataSchema was created using a consensus approach.
Results: Seven nationally representative child cohort surveys were included, incorporating 50 survey waves. A common data model consisted of three domains and fifteen constructs. An oral health and risk indicator questionnaire DataSchema was compiled with 42 variables. For each study wave, the potential (or not) to generate each DataSchema variable was evaluated. Approximately 50% of the DataSchema variables could be generated across at least 4 cohort surveys, while only 10% (n = 4) of variables could be generated across all surveys. For individual surveys, the DataSchema variables that could be generated ranged between 26% and 76%.
Conclusions: The harmonisation potential of child surveys was found to be low across multiple surveys. More consistency and standardisation in survey questionnaires within and between surveys is advocated to help improve data discovery.