Practical recommendations for population PK studies with sampling time errors.

Area of Science:

• Pharmacokinetics
• Pharmacometrics
• Clinical Pharmacology

Background:

• Population pharmacokinetic (PK) data from clinical practice is valuable.
• Missing accurate blood sample times introduce measurement errors (ME) in observational PK studies.

Purpose of the Study:

• Investigate the impact of using scheduled instead of actual blood sampling times on PK model parameters.
• Develop and evaluate methods to correct for ME in sampling times.

Main Methods:

• Simulation studies assessed the effects of PK profile curvature and ME magnitude.
• Developed Transform Both Sides (TBS) models using Box-Cox transformation and Taylor expansion.
• Compared TBS models against conventional population PK models.

Main Results:

• Bias due to time ME is primarily determined by PK profile curvature and ME size.
• Smaller curvature leads to less bias; larger ME leads to more bias.
• TBS models outperformed conventional models when curvature and ME were substantial.

Conclusions:

• Time ME can bias PK parameter estimates.
• Conventional models are robust to large ME when PK profiles have minimal curvature.
• Proposed TBS methodology effectively reduces bias for moderate to high curvature PK profiles.