Introduction to multilevel modelling in dental research

Area of Science:

• Dental research methodology
• Biostatistics
• Oral health sciences

Background:

• Hierarchical data structures are common in dental research, often involving clustered observations (e.g., patients within clinics, teeth within patients).
• Traditional statistical methods may not adequately account for this nested data, potentially leading to biased results or reduced statistical power.

Purpose of the Study:

• To introduce the concepts and applications of multilevel modeling (MLM) for analyzing complex dental research data.
• To demonstrate how MLM can be applied to both cross-sectional and longitudinal hierarchical dental data.
• To highlight the advantages of MLM over single-level methods in generating novel research hypotheses.

Main Methods:

• Methodological review and illustration using actual and hypothetical dental research examples.
• Application of MLM to periodontal research for cross-sectional and longitudinal analyses.
• Demonstration of multilevel multivariate modeling in orthodontic research.

Main Results:

• MLM provides enhanced detail and insight into dental data, enabling the re-evaluation of existing hypotheses.
• The application of MLM facilitates the generation of new research questions previously unaddressable by single-level methods.
• Discussion includes the impact of study design elements, such as sample size and examiner variability, within the MLM framework.

Conclusions:

• Multilevel modeling has significant potential to advance the understanding of oral disease and health.
• MLM encourages a re-evaluation of established dental research procedures and analytical approaches.
• The adoption of MLM can lead to more robust and nuanced findings in dental research.