Related Experiment Videos
Propensity scores with misclassified treatment assignment: a likelihood-based adjustment
Danielle Braun1, Malka Gorfine1, Giovanni Parmigiani1
1Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA; Department of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston, MA, USA; Department of Statistics, Tel Aviv University, Tel Aviv, Israel and St. Luke's Radiation Oncology Associates, St. Luke's Regional Cancer Center, and Whiteside Institute for Clinical Research / University of Minnesota Duluth, Duluth, MN, USA.
Treatment misclassification in claims data biases propensity score analyses. A new method adjusts for this bias, improving comparative effectiveness research using validation data.
Area of Science:
- Health Services Research
- Biostatistics
- Epidemiology
Background:
- Propensity score methods are crucial for comparative effectiveness research (CER) with claims data.
- Claims data often contain misclassified treatment assignments due to inaccurate billing codes, introducing measurement error.
- This treatment misclassification can bias propensity score analyses at multiple stages.
Purpose of the Study:
- To investigate the impact of treatment misclassification on propensity score estimation, implementation, and outcome analysis.
- To propose and evaluate a novel statistical method for adjusting treatment misclassification bias in propensity score analyses.
- To apply the developed methods to real-world healthcare claims data for a specific clinical scenario.
Main Methods:
- Examined the effects of treatment misclassification on subclassification, matching, and inverse probability of treatment weighting (IPTW).
- Developed a two-step likelihood-based approach to adjust for treatment misclassification bias, assuming non-differential measurement error and transportability.
- Validated the proposed method using simulation studies and applied it to Medicare Part A claims data for brain tumor treatment effectiveness.
Main Results:
- Treatment misclassification significantly impacts propensity score estimation, implementation, and subsequent outcome analyses.
- The proposed two-step likelihood-based method effectively adjusts for treatment misclassification bias under subclassification.
- Simulations demonstrated the method's performance, and the real-world application provided estimates for resection versus biopsy in brain tumor patients.
Conclusions:
- Treatment misclassification is a critical issue in claims-based CER that requires careful consideration and adjustment.
- The developed adjustment method offers a robust approach to mitigate bias, enhancing the reliability of findings from propensity score analyses.
- This research provides valuable tools and insights for researchers utilizing claims data in comparative effectiveness studies.