Table 2.

Characteristics of Optimal Versus Minimal Opportunistic Approaches to Pharmacokinetic Sampling for the Study of Tuberculosis Drugs During Pregnancy^a

Key Characteristic	Optimal Approach	Minimal Approach
No. of samples	3–7	1–7
Timing of samples	Determined before study begins based on how many samples can be collected within dosing interval	Late in dosing interval (trough samples are most informative)
Timing in pregnancy	Several time points (2nd and 3rd trimester and postpartum)	3rd trimester better than 2nd trimester better than post partum
No. of womenb	>20 for rich or semi-intensive design; >40 for sparse design	>10 for rich or semi-intensive design; >20 for sparse design
Data to be collected	Dosing time (including previous doses), week of pregnancy, weight and other clinical/demographic variables	Dosing time (including previous doses), week of pregnancy, and weight
Analysis method	Population and/or PK modeling	Population and/or conventional PK modeling

Abbreviation: PK, pharmacokinetic.

^a Opportunistic collection of safety and PK data from women at different stages of pregnancy who are receiving drug(s) of interest as part of clinical care may improve the understanding and management of tuberculosis and human immunodeficiency virus treatment in pregnant and postpartum women. Although such studies can efficiently provide critical PK data, clinical outcome data may be biased since enrollment is limited to subjects who tolerate and have an adequate clinical response to the drug(s) being studied. Examples of such approaches are employed in International Maternal Pediatric Adolescent AIDS Clinical Trials Network (IMPAACT) studies P1026s, P1078, and P2001 (Table 5) and have also been used to assess rifampin, isoniazid, and efavirenz concentrations and interactions.

In general, opportunistic PK sampling should involve the collection of as many samples as possible in a dosing interval (maximum of 5–7 samples spread equally in a dosing interval), but even a single sample can be useful if that is all that is feasible. If it is known ahead of time that opportunistic design is possible, relevant optimal sampling time windows can be determined. Dosing times must be recorded correctly, and pregnancy-related variables must be collected. In general, the more participants enrolled the better, but usually any design including >20 women is informative.

^b Rich, semi-intensive, and sparse designs are defined as >5, 3–5, and <3 samples in a dosing interval, respectively.