Observed Cmax and the Sampling Grid

Why the Sampling Schedule Matters

A concentration-time profile can rise and fall between blood draws. If no sample is taken at the exact time of its highest concentration, the largest recorded value may be lower than the profile’s actual peak. Thus, observed Cmax is not automatically the highest concentration that occurred; it is the highest concentration among the measurements collected at the scheduled times.

This distinction matters when interpreting an exposure metric or comparing profiles collected on different schedules. A schedule with a sample near the peak can record a higher maximum than a schedule that misses it, even when the underlying profiles are identical. The measured difference may therefore reflect sampling timing as well as a difference in the profiles themselves.

Definition: Let $C(t)$ denote the underlying concentration at time $t$, and let $t_1,\ldots,t_m$ be the times at which concentrations are sampled. The observed Cmax is the largest recorded concentration on that sampling grid: $$ C_{\max,\mathrm{obs}}=\max_{1\leq j\leq m} C(t_j). $$ The true peak over a specified time interval is the maximum of the underlying profile on that interval: $$ C_{\max,\mathrm{true}}=\max_{t\in [a,b]} C(t). $$ These definitions distinguish a maximum over a finite set of sampling times from a maximum over a continuous time interval.

The notation $C_{\max,\mathrm{obs}}$ makes the observation schedule explicit. In practice, reports often call the largest measured value simply Cmax. That shorthand is appropriate when the context makes clear that the metric is calculated from observed samples, but it should not be mistaken for proof that the true peak was captured.

When Does the Observed Maximum Fall Below the True Peak?

First consider an idealized setting: concentrations are available at the actual sample times, measurements have no assay error, and the sampling times lie within the interval over which the true peak is defined. Every sampled time is then a time in that interval. Since the true peak is the largest concentration over the whole interval, it cannot be smaller than the concentration at any one sampled time.

C(t_j)\leq C_{\max,\mathrm{true}}\quad\text{for every }j \quad\Longrightarrow\quad C_{\max,\mathrm{obs}}\leq C_{\max,\mathrm{true}}.

This inequality is a direct consequence of how a maximum is defined; it does not require a particular pharmacokinetic model. It can be an equality if at least one sampled time has the true peak concentration. If the peak occurs between sample times and no other sampled time has that same maximum concentration, the inequality is strict. For a flat-topped peak, a sample anywhere on the plateau can capture the maximum even if it does not coincide with one unique peak time.

The assumptions matter. If the recorded values include measurement error, the largest measured concentration may be above or below the largest underlying concentration at the sampled times. Also, if the sampling window stops before the underlying peak, the maximum over the observed window is not necessarily a useful estimate of the maximum over the full profile. The simple inequality applies to exact concentrations sampled within the interval used to define the true peak; it is not a guarantee about noisy measurements or an incomplete interval.

How Distance from the Peak Affects the Recorded Value

A short local calculation explains why sample timing can change the maximum. Suppose that, near a smooth peak at time $t_p$, the concentration profile is approximately quadratic:

C(t)\approx C_{\max,\mathrm{true}}-k(t-t_p)^2,\qquad k>0.

If the nearest sample is $d$ time units from $t_p$, the concentration at that sample is approximately $kd^2$ below the peak. This relationship is exact when the local profile is exactly quadratic. It shows that, under this simple shape, moving the nearest sample twice as far from the peak produces four times the concentration shortfall. The calculation is a local illustration, not a universal formula for all concentration-time profiles.

For a regular grid with spacing $\Delta$, if the peak lies within the grid’s covered interval, the nearest grid point is at most $\Delta/2$ from the peak. Under the quadratic profile above, this gives the bound $C_{\max,\mathrm{true}}-C_{\max,\mathrm{obs}}\approx kd^2\leq k\Delta^2/4$. The result depends on the quadratic approximation and regular spacing; it should not be treated as a general bound for an unknown profile.

Worked Example: Finding the Maximum in a Recorded Profile

Suppose a hypothetical participant has quantifiable concentrations recorded at the following times. Assume for this calculation that the recorded values represent the concentrations at the actual sample times.

Time (hours)	Concentration (mg/L)
0.5	1.8
1.0	5.4
2.0	8.7
3.0	7.9
5.0	4.1

Compare the recorded concentrations: $1.8,\ 5.4,\ 8.7,\ 7.9,$ and $4.1$ mg/L. The largest is $8.7$ mg/L, so $C_{\max,\mathrm{obs}}=8.7$ mg/L on this schedule. The sample at 2.0 hours determines the observed maximum because its concentration exceeds every other recorded value.

This calculation establishes the largest value among the five measurements. It does not establish that the true peak was exactly $8.7$ mg/L or occurred at 2.0 hours. For example, the unobserved profile could rise above $8.7$ mg/L between 1.0 and 3.0 hours. Without additional measurements or a justified model, the true peak is not identified by these five values alone.

Worked Example: Calculating a Grid-Related Shortfall

Consider a hypothetical smooth concentration profile, measured in mg/L, described over 0 to 6 hours by $C(t)=10-0.5(t-3.2)^2$. The profile has its maximum at $t=3.2$ hours, where $C(3.2)=10$ mg/L. Suppose samples are taken at 0, 2, 4, and 6 hours.

Substitute each sample time into the profile:

At 0 hours, $C(0)=10-0.5(0-3.2)^2=10-0.5(10.24)=4.88$ mg/L.

At 2 hours, $C(2)=10-0.5(2-3.2)^2=10-0.5(1.44)=9.28$ mg/L.

At 4 hours, $C(4)=10-0.5(4-3.2)^2=10-0.5(0.64)=9.68$ mg/L.

At 6 hours, $C(6)=10-0.5(6-3.2)^2=10-0.5(7.84)=6.08$ mg/L. The largest sampled value is therefore $9.68$ mg/L, so $C_{\max,\mathrm{obs}}=9.68$ mg/L. Relative to the profile’s true peak of $10$ mg/L, the shortfall is $10-9.68=0.32$ mg/L.

The sample nearest to the peak is at 4 hours, $0.8$ hours away. The quadratic calculation gives a shortfall of $k d^2=0.5(0.8)^2=0.32$ mg/L, matching the direct calculation. This agreement is expected because the example uses an exactly quadratic profile.

Comparing Sampling Schedules

The next example holds the underlying profile fixed and changes only the sample times. This isolates the contribution of the grid: different observed maxima need not imply different underlying peaks.

Worked Example: Two Grids, One Underlying Peak

Use the same hypothetical profile $C(t)=10-0.5(t-3.2)^2$ mg/L. Schedule A samples at 0, 1, 3, and 6 hours; Schedule B samples at 0, 2, 4, and 6 hours. Both cover the same 0-to-6-hour interval.

For Schedule A, the values at 0 and 6 hours are 4.88 and 6.08 mg/L, as calculated above. At 1 hour, $C(1)=10-0.5(1-3.2)^2=10-0.5(4.84)=7.58$ mg/L. At 3 hours, $C(3)=10-0.5(3-3.2)^2=10-0.5(0.04)=9.98$ mg/L. The maximum is therefore $9.98$ mg/L.

For Schedule B, the four values are $4.88,\ 9.28,\ 9.68,$ and $6.08$ mg/L, so its maximum is $9.68$ mg/L. The observed maxima differ by $9.98-9.68=0.30$ mg/L, although the underlying profile and its true peak of $10$ mg/L are identical.

Schedule A has a sample only $0.2$ hours from the peak; Schedule B’s nearest sample is $0.8$ hours away. The quadratic shortfalls are $0.5(0.2)^2=0.02$ mg/L and $0.5(0.8)^2=0.32$ mg/L, respectively. The resulting observed maxima are $10-0.02=9.98$ mg/L and $10-0.32=9.68$ mg/L. A denser or better-positioned schedule can reduce this particular grid-related shortfall, but it does not guarantee that every real peak is captured.

Assumptions, Diagnostics, and Alternatives

To interpret a reported observed Cmax as a summary of a concentration-time profile, check what values and times went into it. The basic calculation is only a maximum across recorded values, but conclusions about how well it represents a peak depend on the sampling design and data quality.

Check coverage. Confirm that samples span the period in which the peak is expected, rather than stopping while concentrations may still be rising.
Check timing around the maximum. A large gap between samples near the largest recorded value makes it more plausible that an unobserved higher value could lie between them.
Check actual versus scheduled times. If sample collection deviated from the planned schedule, the grid used to interpret the observed maximum should reflect the actual times.
Check data quality. A single unusually high measurement may reflect error rather than the underlying profile. Measurement error means the largest recorded value is not guaranteed to be below the true peak.
Compare schedules before comparing maxima. If profiles use different sampling times, a difference in observed Cmax can combine a profile difference with a sampling-grid difference.

The most direct alternative is to collect additional samples near the anticipated peak. This can make the observed maximum less sensitive to the placement of a coarse grid, but it increases sampling burden and still may not coincide exactly with the peak. Another option is interpolation or a model-based estimate between samples. Such an estimate can be useful when the interpolation rule or model is scientifically justified, but it is not the same quantity as the maximum of the recorded concentrations. It may exceed the largest observed value, and its validity depends on assumptions about the profile between measurements.

Common pitfall: “The observed Cmax is 9.68 mg/L, so the true peak must be 9.68 mg/L.” This reasoning treats the largest recorded value as if the entire continuous profile had been observed. The correction is to report 9.68 mg/L as the maximum on the stated sampling grid and recognize that the true peak may be higher if it occurred between samples. With measurement error, even the direction of the difference is not guaranteed.

Observed Cmax is practical and transparent: it can be calculated directly from the recorded concentrations without specifying a curve between samples. Its limitation is equally direct: it depends on which times were sampled. Interpolation and model-based methods attempt to use assumptions about the unobserved portions of the profile; a denser schedule gathers more direct information. None of these approaches removes the need to check coverage, timing, and data quality.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions.

A profile has concentrations of 2.1, 6.3, 6.0, and 3.4 mg/L at four sampled times. What is the observed Cmax, and what can you conclude about the true peak?
Under exact measurement and complete coverage of the interval of interest, why can the observed maximum not exceed the true peak?
A smooth quadratic peak has $k=0.4$ mg/L per hour squared. A sample is 0.5 hours from the peak. What shortfall does the quadratic calculation predict?
Two schedules produce different observed maxima for the same underlying profile. Give one reason this can happen and one check that helps interpret the difference.
Does interpolation between samples produce the observed Cmax as defined here? Explain.

Solutions. (1) The observed Cmax is $6.3$ mg/L, the largest of the four recorded values. This alone does not identify the true peak; it may occur between samples. (2) Each sampled time lies in the interval over which the true peak is defined, so its concentration is no greater than the maximum over that whole interval. Taking the largest sampled value preserves that inequality. (3) The predicted shortfall is $kd^2=0.4(0.5)^2=0.4(0.25)=0.10$ mg/L. This is a quadratic-profile calculation, not a general rule for every profile. (4) One schedule may place a sample closer to the peak than the other. Compare the actual sampling times, especially the gaps near the largest measurements, before interpreting the maxima as evidence of different underlying peaks. (5) No. Observed Cmax is the maximum among recorded concentrations. Interpolation estimates values between samples and may yield a different maximum, which depends on the interpolation method.