Population Pharmacokinetics and Pharmacodynamics

Population pharmacokinetics and pharmacodynamics, often abbreviated popPK/PD, is the study of how drug exposure and response vary across a population and what explains that variation. Using nonlinear mixed-effects models, it estimates typical parameter values, the influence of covariates, and the random variability between and within individuals from data pooled across many subjects.

Definition

Population pharmacokinetics-pharmacodynamics is the analysis of pharmacokinetic and pharmacodynamic data from a population using nonlinear mixed-effects models that simultaneously estimate typical (fixed-effect) parameters, covariate relationships, and between-subject and residual (random-effect) variability.

Scope

The entry covers the structure of population models, the separation of fixed effects, covariate effects, and random effects, and the role of these models as the quantitative foundation for individualized dosing. It is a methodological topic and does not give drug-specific exposure targets or treatment advice.

Core questions

• What is the typical value of a pharmacokinetic or pharmacodynamic parameter in a population?
• Which covariates explain part of the variability in exposure or response?
• How much variability remains unexplained between and within individuals?
• How can sparse data from many subjects be combined into a single model?

Key concepts

• Fixed effects and typical parameters
• Covariate relationships
• Between-subject variability
• Residual variability
• Structural and statistical models
• Sparse-data estimation
• Model evaluation and qualification

Key theories

Nonlinear mixed-effects modelling

A statistical framework that estimates fixed effects (typical parameters and covariate relationships) and random effects (between-subject and residual variability) simultaneously, allowing population structure to be learned even from sparse, unbalanced data.

Mechanisms

A population model has a structural component describing the typical time course of concentration or effect, a covariate component relating parameters to characteristics such as weight, age, organ function, or genotype, and a statistical component partitioning unexplained variation into between-subject and residual terms. Parameters are estimated by fitting the model to data pooled across subjects, which allows even sparse samples per person to contribute. Diagnostic plots, simulation-based checks, and quality-control practices are used to evaluate whether the model adequately describes the data. The resulting model provides the population prior that individualized and Bayesian dosing methods build on.

Clinical relevance

Population PK/PD models underpin how variability in drug exposure and response is quantified for research, drug development, and the design of dosing strategies. This entry describes the modelling methodology; it characterizes how exposure varies and is not a source of specific dosing targets or individual treatment decisions.

Evidence & guidelines

Best practices for building and reporting population analyses are described in quality-control guidance for popPK/PD analyses and in tutorials for the estimation software widely used in the field; these complement the foundational methodology established for population pharmacokinetics and dynamics.

History

Population pharmacokinetics arose from Sheiner and colleagues' work in the 1970s on estimating individual pharmacokinetics from routine data, and the mixed-effects approach was formalized and popularized through dedicated estimation software. By the early 1990s the population PK/PD framework was consolidated, and later decades added structural refinements, quality-control standards, and broad use in regulatory and clinical pharmacology.

Debates

How should covariates be selected and validated?

Stepwise covariate selection can overfit and produce spurious relationships, so the field has debated more principled selection and external validation to ensure that covariate effects, including genotype, are real and transportable.

Key figures

• Lewis Sheiner
• Stuart Beal
• Mats Karlsson
• Peter Bonate

Seminal works

• sheiner1972
• sheiner1992

Frequently asked questions

What does the 'mixed-effects' part of population modelling mean?

It means the model contains both fixed effects, which describe typical parameters and covariate relationships shared across the population, and random effects, which capture how individuals vary around those typical values.

Why pool data across many subjects?

Pooling lets the model learn population-level structure and variability even when each individual contributes only a few samples, which is common in clinical settings.