Doseresponsanalyse

Doseresponsanalyse is the term used in some Scandinavian languages for dose–response analysis. It is a statistical approach to characterize how a biological response changes with varying doses of a substance. It is central to toxicology, pharmacology, and environmental health risk assessment, enabling estimation of thresholds, potency, and risk at different exposure levels. The analysis can be conducted on experimental data from animal studies, clinical trials, or observational human data.

Responses can be continuous or binary. For continuous endpoints, linear or nonlinear regression is used; for binary endpoints, probit or logit models are common. Dose–response models include linear, Hill (sigmoid), four-parameter logistic, exponential, and Emax models. The choice depends on the endpoint and data. Dose levels may be fixed or dose-ranging and may require transformation.

Key outputs include the estimated dose producing a predefined response (EDx), the benchmark dose (BMD) approach, and regulatory reference values such as NOAEL, LOAEL, and derived reference doses. Model selection and uncertainty are addressed with goodness-of-fit metrics, confidence intervals, and sensitivity analyses. Non-monotonic dose–response relationships can complicate interpretation and may require alternative modeling or stratified analyses.

Practical considerations include data quality, study design, cross-species extrapolation, and extrapolation to humans. Software commonly used includes R packages such as drc, as well as SAS and other nonlinear modeling tools. Doseresponsanalyse supports risk assessment by quantifying potency, variability, and uncertainty in dose–response relationships.