Related Experiment Video
Updated: Apr 10, 2026

10:42
Oral Biofilm Sampling for Microbiome Analysis in Healthy Children
Published on: December 31, 2017
18.1K
Developing a propensity score protocol for evaluating the oral health consequences of methamphetamine use
Lauren Harrell1,2, Thomas R Belin1, Vivek Shetty2
1Department of Biostatistics, Fielding School of Public Health, University of California, Los Angeles, Los Angeles, California, USA.
Summary
This study introduces a novel method combining propensity-score matching and stratification for better covariate balance in hard-to-reach populations. This approach improves comparisons between methamphetamine users and nonusers.
Area of Science:
- Epidemiology
- Biostatistics
- Public Health
Background:
- Propensity-score matching (PSM) is crucial for causal inference but struggles with covariate balance when comparing specialized populations to general ones.
- Hard-to-reach populations, such as methamphetamine users, present unique challenges for traditional PSM due to distinct characteristics.
- Achieving balanced covariate distributions is essential for reliable comparisons and valid outcome estimates.
Purpose of the Study:
- To present and evaluate a novel methodology integrating propensity-score matching with stratification.
- To enhance covariate balance when comparing a hard-to-reach subclass (methamphetamine users) with a general population sample.
- To facilitate interpretable and robust between-group comparisons using weighted outcome estimates.
Main Methods:
- Utilized a cross-sectional cohort from the National Health and Nutrition Examination Survey (1999-2004).
- Employed propensity-score-based stratification to create subclasses, improving upon standard propensity-score matching.
- Incorporated weighting strategies between subclasses for outcome estimation.
Main Results:
- The combined propensity-score matching and stratification approach successfully achieved better covariate balance.
- This method demonstrated effectiveness in comparing methamphetamine users against nonusers from the NHANES dataset.
- Weighted outcome estimates provided interpretable comparisons between the defined subclasses.
Conclusions:
- The proposed technique offers a superior alternative to conventional propensity-score matching for studies involving hard-to-reach populations.
- This method enhances the validity of comparative analyses by ensuring better covariate balance.
- Future research should explore optimal parameters (matching ratio, number of subclasses) for this integrated approach.

