Editorial: Leaving no stone unturned, or extracting blood from stone?

  • Anderson B
  • Holford N
  • 9


    Mendeley users who have this article in their library.
  • 3


    Citations of this article.


Models describe systems in simple terms, although some models may be quite sophisticated. They are used to describe, predict, and explain observations. Pharmacokinetic (PK) and pharmacodynamic (PD) models are used to improve pediatric anesthetic management. They quantify the exposure–response relationship, often providing clarity and insight into complex systems as well as a mechanistic under-standing of the drug effect. Dosing can be rational-ized. Models may enable extrapolation beyond observed data. Modeling is a knowledge manage-ment tool; it captures and integrates data from all studies. Models can also be used for hypothesis testing and can drive decisions about the best way to interpret observations in clinical practice. A population approach, achieved through nonlin-ear mixed effects models, is now an established method for investigating PK and PD data. This population approach has implications beyond phar-macology and can be used to describe physiology (1) and has also been advocated for interpreting anal-gesia trials (2). The approach was used by Standing et al. (3) in the current issue of Pediatric Anesthesia to examine the PK–PD relationship between remifen-tanil and blood pressure changes. The authors used complex modeling to rend this relationship from only five subjects. Can we really believe such results or are they taking things too far? Mixed effects models Mixed effects models are used for the population approach. They provide a means to study vari-ability in drug responses among individuals representative of those in whom the drug will be used clinically. Traditional approaches to the inter-pretation of time–concentration profiles (e.g., nave and standard two-stage approaches) relied on ÔrichÕ data from a small group of subjects. In contrast, mixed effects models can be used to analyze ÔsparseÕ (2–3 samples) data from a large number of subjects. These models are ÔmixedÕ because they describe the data using a mixture of fixed and random effects. Fixed effects predict the average influence of a covariate such as weight as an explanation of part of the between-subject vari-ability in a parameter like clearance (CL). Explan-atory covariates (e.g. age, size, renal function, sex, temperature) can be introduced, which explain the predictable part of the between-individual vari-ability. Random effects describe the remaining variability between subjects that is not predictable from the fixed effect average. Interpretation of truncated individual sets of data or missing data is also possible with this type of analysis, rendering it useful for pediatric studies. The appropriate number of patients for a population study is difficult to determine and will depend on the number of covariates under examination (4). Approximately 50 subjects are often used in a population study, but covariate investigation in a study investigating children ranging from neonates to adults requires large numbers. Fewer subjects are typically used in a discrete population such as neonates, but covariate relationship inter-pretation may be limited. Population modeling also allows pooling of data across studies to provide a single robust PK analysis rather than comparing

Get free article suggestions today

Mendeley saves you time finding and organizing research

Sign up here
Already have an account ?Sign in

Find this document


  • Brian J. Anderson

  • Nick H G Holford

Cite this document

Choose a citation style from the tabs below

Save time finding and organizing research with Mendeley

Sign up for free