Confounding in exposure–response analysis
An exposure–response association can reflect factors that influence both exposure and the clinical outcome.
An exposure–response association can reflect factors that influence both exposure and the clinical outcome (FDA. Exposure–Response Relationships: Study Design, Data Analysis, and Regulatory Applications. 2003.).
If patients with more advanced disease have higher antibody clearance and worse outcomes, lower exposure can appear to explain worse outcomes. Check the timing of exposure and outcome measurements, adjust for measured baseline factors, and compare with randomized dose-group evidence when available. A flat fitted association over a narrow exposure range leaves effects outside that range unresolved.