Related Experiment Videos
Specimen allocation in longitudinal biomarker studies: controlling subject-specific effects by design
Shelley S Tworoger1, Yutaka Yasui, Lilly Chang
1Channing Laboratory, Harvard Medical School and Brigham and Women's Hospital, Boston, Massachusetts, USA.
Summary
Assaying samples from the same individual in different batches can skew biomarker results in longitudinal studies. Ensuring all samples are processed together is crucial for reliable intervention effect estimates.
Area of Science:
- Biomedical research
- Clinical trials
- Biomarker analysis
Background:
- Longitudinal studies require reliable biomarker measurements for accurate within-person change assessment.
- Specimen batch allocation can significantly impact the validity of results in biomarker studies.
Purpose of the Study:
- To investigate the impact of assaying samples from different time points in separate batches on intervention effect estimates.
- To quantify the bias introduced by split-batch sample processing in a randomized clinical trial.
Main Methods:
- Utilized data from a randomized clinical trial involving 136 postmenopausal women.
- Compared intervention effect estimates using split-batch samples versus excluding or reassaying them in a single batch.
- Analyzed serum concentrations of estrone, estradiol, testosterone, androstenedione, and dehydroepiandrosterone.
Main Results:
- Median differences in intervention effect estimates were 59.6% (split vs. excluded) and 74.6% (split vs. reassayed).
- Coefficients from split-batch data were closer to zero and less statistically significant.
- Bias was artificially introduced into intervention effect estimates when samples were not assayed in the same batch.
Conclusions:
- Assaying samples from the same subject in different batches can introduce significant bias into longitudinal study results.
- Consistent batch processing of all samples from an individual is essential for maintaining the integrity of biomarker data and intervention effect estimates.