Sampling Across A Steady-State Dosing Interval

Why Sampling Time Matters at Steady State

A multiple-dose concentration profile is meaningful only in relation to the dosing schedule that produced it. If doses are given every \(\tau\) hours, the interval from one dose to the next is the natural window for planning samples. At steady state, the concentration-time pattern repeats from interval to interval under an unchanged regimen. Sampling must therefore describe the changing concentration throughout that interval, not merely collect a convenient set of clock-time measurements.

A sample taken at the wrong time can be misleading even if its concentration was measured accurately. For example, a concentration collected 2.5 hours after a dose is not a 2-hour concentration. This distinction matters when comparing profiles, locating rapidly changing parts of the curve, or calculating exposure from sampled concentrations. The introductory material on establishing steady state and calculating accumulation provides the context for this tutorial: the focus here is how to schedule and interpret sampling times once steady state is the intended condition.

For a practical schedule, the dose time and collection time should both be recorded. A nominal time such as “2 hours post-dose” describes the plan; the actual elapsed time is calculated from the dose and collection timestamps. If a sample is collected late, its actual time should not be silently replaced by the nominal time.

Planning Coverage Across the Interval

A useful schedule covers the full interval and places samples where they are informative. In general, planners consider a predose concentration, early post-dose times, times around the expected rapid rise and peak region, several later times during decline, and a sample close to the next scheduled dose. The precise schedule depends on the formulation, route, expected absorption and elimination, assay constraints, and the study question. Those expectations should be based on relevant prior evidence or pilot information, not treated as facts known in advance.

The purpose is not to impose a universally optimal grid. A rapidly changing profile needs closer temporal resolution than a slowly changing one. Likewise, the time of a peak may vary across participants, so a schedule that samples only at one presumed peak time can miss the actual peak region in some individuals. Sampling should remain anchored to each participant’s actual dose time, while the planned times provide a consistent framework across participants and periods.

A predose sample is often useful for characterizing the end of the interval. In a repeated regimen, a sample immediately before the next dose is at approximately the same phase of the dosing cycle as the predose sample before the current dose. For an oral dose, concentration is generally continuous across the dosing time; for routes or models in which administration produces an abrupt concentration change, the relevant pre- and post-dose observations must be distinguished rather than treated as interchangeable.

How Sampling Density Affects Exposure Estimates

A common way to approximate exposure between observed concentrations is the linear trapezoidal rule. Between two observations at times \(t_i\) and \(t_{i+1}\), it assumes concentration changes linearly. The area for that segment is the width of the time interval multiplied by the average of its two endpoint concentrations. Adding the segment areas gives an approximation over the sampled interval.

Here \(C_i\) is the measured concentration at time \(t_i\), and \(m+1\) observations span the interval being approximated. If concentration is measured in mg/L and time in hours, the area has units of mg·h/L. The rule is exact for a truly linear concentration segment. When the curve bends between widely spaced observations, however, straight-line interpolation may not reproduce the true area. More frequent measurements can reduce this problem, particularly during rapid changes, but cannot eliminate all measurement or model error.

Actual Timing, Deviations, and Diagnostics

A schedule is only useful if the times it specifies correspond to what happened. Record the actual dose administration time, the actual sample collection time, and deviations such as delayed doses, missed doses, or delayed collections. Calculate elapsed time from those records. If dose timing differs between participants or periods, the same nominal clock time can represent different dose-relative times.

When reviewing a concentration-time data set, plot concentrations against actual elapsed time after dose and mark the dosing interval. Look for unusually long gaps, samples out of order, a missing predose or late-interval sample, and apparent changes that conflict with neighboring observations. Check whether a sample labeled as predose was truly collected before the dose. A sample collected after dosing cannot be repaired by relabeling it as predose.

A delayed sample does not automatically invalidate all data from a participant. Its effect depends on the deviation, the profile’s rate of change at that time, and the analysis method. The key is to retain the actual timing, apply prespecified handling rules, and report material deviations. Replacing actual times with nominal times may cause an interval to be stretched or compressed in calculations and may misrepresent where a concentration lies on the curve.

Choosing Between Sampling Strategies

Rich sampling offers more information about the shape of the profile and supports interpolation with shorter gaps. Its costs include participant burden, staff time, and analytical workload. A deliberately sparse schedule can be appropriate when the study question and analysis method can support it, but sparse sampling makes the result more dependent on assumptions or on information from other sources. It should not be described as equivalent to a fully observed profile without justification.

Uniformly spaced samples are straightforward to organize and can provide broad coverage. A targeted schedule instead places more samples where prior knowledge suggests faster change or where the study question requires resolution. These approaches can be combined: use a basic spread across the interval and add observations in expected high-information regions. Neither strategy compensates for sampling only the first part of the interval or omitting the approach to the next dose.

Sampling should be designed around the intended analysis, but the design should not assume that a chosen interpolation rule recovers unobserved concentrations exactly. Sparse sampling can miss curvature, a narrow concentration feature, or meaningful participant-to-participant variation in timing. Conversely, adding many samples at nearly identical times while leaving large gaps elsewhere may add little useful information. The schedule is strongest when its time points are justified by the expected profile, cover the full interval, and are feasible to execute consistently.

• Coverage failure: Large gaps, especially across a period of rapid concentration change, can make the area estimate highly dependent on interpolation.
• Clock-time substitution: Using the planned time instead of the actual elapsed time can distort interval widths and the placement of observations.
• Inadequate steady-state conditions: If the regimen was not stable or doses were missed, the profile may not represent a repeating steady-state interval.
• Overconfidence in a calculated area: Numerical precision in an estimate does not remove uncertainty caused by sparse observations or incorrect timing.