Clinical Scorecard: Integrating Targeted Surveys of Unique Populations with Broader Surveys for Generalized Insights: A Cross-Sectional Study
At a Glance
Category
Detail
Condition
Surveillance and inference of drug use behaviors, focusing on psychedelic drug use
Key Mechanisms
Data fusion and transport weighting to combine targeted and general surveys for generalizable estimates
Target Population
General population and rare subpopulations using psychedelic drugs in the United States
Care Setting
Epidemiological and public health surveillance settings
Key Highlights
Large representative surveys prioritize breadth over depth, limiting detailed drug-specific insights.
Targeted surveys provide detailed drug-specific data but lack generalizability to broader populations.
Data fusion with transport weighting can correct selection bias and enable generalizable inference for rare subpopulations.
Guideline-Based Recommendations
Diagnosis
Use large anchor surveys with well-established sampling frames to provide population benchmarks.
Conduct smaller enriched surveys targeting specific subpopulations for detailed data.
Management
Apply a two-step data fusion approach combining anchor and enriched surveys.
Use generalized raking calibration to adjust for selection biases between surveys.
Monitoring & Follow-up
Verify assumptions of calibration by examining internal consistency between fused and anchor surveys.
Assess external validity by comparing fused survey estimates with benchmark survey metrics.
Risks
Unmeasured factors inducing selection bias between surveys can lead to inaccurate estimates.
Differential measurement errors between surveys may affect validity of fused data.
Patient & Prescribing Data
Individuals using psychedelic drugs, including rare subpopulations within the general US population
Fused survey data can inform policy and health surveillance by providing generalizable insights into motivations, experiences, and health outcomes related to psychedelic drug use.
Clinical Best Practices
Combine large representative surveys with targeted surveys to balance breadth and depth of data.
Use transport weighting and generalized raking to correct for selection biases in fused datasets.
Explicitly test and verify assumptions underlying data fusion methods to ensure valid inference.
Leverage anonymous online surveys for stigmatized behaviors to improve reporting accuracy.
Apply data fusion methods to study rare subpopulations where prevalence is low.
