2. Introduction
• The primary purpose of PK and PKPD analysis
– To individualize drug choice and dosing regimen
• The population approach
– Arose from 1972 to 1977
• An exponential growth in publication
– Since 1985
4. What is a population analysis?
• It is the application of a model to describe
data that arise from more than one individual.
• Allows the use of sparse sampling study
designs
• Quantify the influence of patient
characteristics and any remaining unexplained
variability between patients.
5. Simulation dataset
Gentamicin-like drug
A volume of distribution: 20 l
Clearance: 41/h
IV bolus
Simulate data for 30 pts who received
1. 420 mg single IV bolus
2. Seven blood samples at 0.25, 0.5, 1,
2, 4, 8 and 12 h
6. Three elements of a population model
• A model for the typical response
– This is the response for a typical (average) patient
• A model for heterogeneity
• A model for uncertainty
7. A model for the typical response
A structural model
10. A model for heterogeneity
• Describe the variability between individuals
• Also called
– Between subject variability (BSV)
– Interindividual variability (IIV)
11. Two distinct model
• One model
– Describe predictable reasons why individuals are
different
• Second
– Quantify the remaining source of random
variability
• A statistical model for random variability
12. The range of model predictions
Not only the typical response of the population, but
also to predict the likely range of responses that may
occur
16. A model for uncertainty
• Also called residual error
• Uncertainty arises from
– Process error
• Where the dose or timing of dose or blood samplings are
not conducted at the times that they are recorded
– Measurement error
• Where the response (concentration) is not measured exactly
– Model misspecification
• Too simple equation in reality
– Moment to moment variability within a patient
17. For error
The error for jth observation for the ith individual
This error represents the (residual) difference of the
model prediction from the data.
Usual to consider e to be normally distributed
18. Why are population PKPD analyses
performed?
• Descriptive population analyses
– The best model to describe the study data
• Predictive population analyses
– What dose or interval
– Max effects and min side effects
• Designing clinical trials
• Identification of covariates
– Phenotypic or genotypic
19. Interpretation of population analyses
• Two important questions
– Was the design appropriate to identify a covariate
relationship?
– Was the covariate relationship significant?
20. Design of covariate population
analyses
• Covariates can be assessed based on the
distribution of covariates in the study
population and the number of subjects in the
study
• A minimum of 50 to 100 pts are required to
provide accurate estimates of covariate effects
in a population analysis setting
21. Significance of covariate relationships
• Biological plausibility
– CL increases rather than decreases with increasing weight
• Clinical significance
– 20 % of drug is eliminated renally, but CL as a statistically
significant covariate
• Statistical significance
– Global test
• NONMEN: objective function values (OFV)
• > 3.84 units: p < 0.05
– Local test
• A reduction in unexplained between subject variability