Why Incomplete Dose Histories Need Careful Reconstruction
A concentration sample is interpreted in relation to the doses that came before it. If part of that history is missing, a dose that was administered may be absent from the dataset, or an expected dose may never have been given. Those possibilities have different meanings for the participant’s actual exposure. A blank dose record does not distinguish them.
Reconstruction therefore means reconciling available evidence about dose events, not filling every gap with the regimen that was intended. The objective is to establish what is supported, preserve what remains uncertain, and avoid representing an unknown history as either a known dose or a known no-dose period.
These statuses should not be collapsed. In particular, a prescription or planned regimen describes intended dosing; by itself, it does not establish that a dose was administered. Likewise, an empty section of a medication record establishes only that no event is recorded there, unless the record’s completeness and meaning are independently known.
Define the Period and the Question First
Before reconciling records, identify the participant and the period that matters. For a predose sample at enrollment, that may include treatment before the study began. For a sample after a visit, it may include administrations between the visit and the sample. The start and end of the period should be explicit, with dates, times, and time zones interpreted consistently with the study records.
Also define what is being reconstructed. Possible questions include whether an administration occurred, what amount was given, when it occurred, and whether it was a bolus or infusion. Evidence may establish some of these details but not others. For example, a medication diary may support a participant’s report that a dose was taken on a particular day without establishing the exact time. Do not let confidence in one detail imply confidence in all the others.
Use the dose-event conventions and event ordering already established for the dataset. Keep the source information and the reconstruction decision traceable to the participant and relevant period. A planned dose can help identify candidate events to investigate, but it is not a substitute for evidence of actual dosing.
A Source-Based Reconciliation Process
Specify the period relevant to the sample or analysis, including whether pre-study treatment is in scope.
Use the documented regimen to identify administrations that may have been expected. Treat this as a search aid, not as a list of actual doses.
Compare appropriate sources, such as administration records, contemporaneous participant records, and medication accountability records. Record what each source can and cannot establish.
Classify the event as confirmed administered, confirmed not administered, or unknown. If only some details are established, retain uncertainty for the unresolved details.
Record the relevant period, sources reviewed, known facts, remaining uncertainty, and reason no further reconstruction is justified.
A source must be interpreted according to its purpose. A clinic administration record may directly document a dose given in the clinic. A prescription can establish the intended regimen but not actual use. Dispensing or returned-medication records can help reconcile supply and use, but may not establish exactly when a dose was taken. A participant report can be relevant evidence of self-administration; preserve that provenance rather than presenting it as a directly observed event.
Worked Examples
Worked Example: A Predose Sample With a Pre-Study Gap
In a hypothetical study, a participant reports taking an oral medication every 24 hours at 08:00. The study’s first documented dose is on March 10 at 08:00, and a blood sample is collected at that time before the study dose. The study dataset contains no pre-study administration records. Can the pre-study history be reconstructed from the regimen?
The regimen identifies candidate dose times, but not whether those doses were taken. For example, if the participant had followed the stated schedule on March 8 and March 9, candidate administrations would have been March 8 at 08:00 and March 9 at 08:00. The absence of records does not establish either that both occurred or that neither occurred.
The supported history is therefore: first study dose documented March 10 at 08:00; pre-study administration status unknown on the available evidence. The first dose’s documented time does not resolve earlier dosing. The pre-study period should be flagged as unresolved unless additional evidence can support particular administrations or non-administrations.
Worked Example: A Confirmed Missed Dose Between Visits
In a hypothetical study, the planned regimen is one 50 mg dose every 24 hours at 08:00. A participant’s records confirm administrations on April 3, April 5, and April 6. A contemporaneous diary states that the April 4 dose was missed, and a study note confirms the report. What should be concluded for April 3 through April 6?
The schedule identifies four candidate administrations: April 3, April 4, April 5, and April 6, each at 08:00. The source records support three actual administrations, each of 50 mg, on April 3, April 5, and April 6. The diary and study note support a confirmed non-administration on April 4.
Thus, there were four scheduled dose opportunities and three confirmed administrations in this period. The April 4 event should not be entered as a 50 mg administration simply because that was the planned dose. It should be represented as a confirmed missed dose or non-administration using the dataset’s established conventions, with the supporting evidence retained.
Worked Example: Evidence Establishes the Dose but Not Its Exact Time
In a hypothetical study, a participant’s diary records taking a 25 mg dose on May 12, but the time field is blank. A later sample is collected at 14:00. No other source establishes when the dose was taken. What can be reconstructed?
The diary supports a reported administration of 25 mg on May 12. It does not support an exact administration time. Therefore the amount and calendar date can be recorded with the appropriate source qualification, while the time remains unknown. Assigning 08:00 because that is the usual schedule would replace missing evidence with an assumption.
The sample time of 14:00 does not resolve the dose time. The elapsed time between dose and sample cannot be calculated exactly from these records. Document that limitation rather than entering a guessed time or a derived elapsed time.
Why Missingness Cannot Be Converted to No Dosing
The logic is simple but important. Suppose the only available record has no entry for a candidate dose. There are at least two possible histories consistent with that observation: the dose was not administered, or it was administered but not recorded. Since both histories produce the same incomplete record, the record alone cannot identify which history occurred.
A schedule does not remove this ambiguity. It tells us when a dose would have been expected under the intended regimen. It does not establish that the participant followed that regimen. Independent evidence can resolve the ambiguity for a particular event, but a rule such as “assume all scheduled doses were taken” is an assumption, not reconstruction.
Similarly, a known dose on one date does not prove that all intervening doses were taken. Each event or period should be assessed using the evidence that applies to it. If sources conflict, retain the conflict for resolution rather than silently choosing the record that best matches the planned regimen.
Documentation, Diagnostics, and Alternatives
A useful reconstruction note makes the decision auditable. It should identify the relevant time window, the candidate regimen if known, the sources reviewed, the facts each source supports, and the final status of each event or unresolved interval. If a dose is confirmed but its amount or time is not, state that specifically. If no reliable source covers the period, document that the history is insufficient rather than filling it with zeros or scheduled doses.
Several signals should prompt review: a long pre-study period with no medication history; unexplained gaps between otherwise documented administrations; a regimen that changes while the event records do not; or a source that covers medication supply but not actual use. These signals identify questions for reconciliation. They do not, by themselves, prove a data error or a missed dose.
There are three common approaches, with different evidential strengths:
| Approach | What it can establish | Main limitation |
|---|---|---|
| Reconcile event-level source records | Specific administrations or non-administrations when the source directly supports them | Sources may be incomplete, imprecise, or limited to part of the period |
| Carry forward the planned regimen | Candidate dose times to investigate | Does not establish actual administration or adherence |
| Infer dosing from concentration patterns | May suggest that the recorded history merits investigation | Concentration patterns alone generally do not uniquely identify a dose history; pharmacokinetic variability and sampling also affect concentrations |
Concentration-based inference may be useful as a diagnostic prompt, but it should not be used to manufacture event records. Multiple dose histories can be compatible with a concentration observation, particularly when other determinants of concentration are not fully known. Any reconstruction should be supported and documented independently of a convenient fit to expected concentration behavior.
The limitation is not merely administrative. An uncertain dose history can affect interpretation of a sample and any downstream analysis that uses dose timing or amount. Marking the uncertainty preserves that limitation for later decisions; assigning an unsupported dose can make the data appear more complete while making the history less reliable.
Common Mistakes
- Treating a blank record as a confirmed hold. A blank means the event is not documented there; it establishes no dose only when appropriate evidence confirms non-administration.
- Copying the prescribed regimen into actual dose records. The planned schedule generates candidate events, not proof that those administrations occurred.
- Guessing a time from the usual dosing time. A usual or nominal time cannot substitute for an unknown actual time when the distinction matters.
- Using a concentration to fill a missing dose event. A concentration may motivate review, but it does not by itself establish a unique dose, amount, or time.
- Recording uncertainty as zero. A zero amount can mean a known absence of dosing in some contexts; it should not be used to represent an unknown history.
Check Your Understanding
Answer each question, then compare with the solutions below.
- A participant’s prescription lists a dose every morning, but there is no administration record for two days. What can be concluded from the prescription and blank records alone?
- A reliable source confirms that a scheduled dose was held. Is the event an unknown dose history or a confirmed non-administration? Explain.
- A diary supports that a dose was taken yesterday but gives no time. Which details may be recorded as supported, and which remain unknown?
- Why should a concentration pattern not be used by itself to create a missing dose event?
Solutions. (1) The prescription identifies candidate scheduled doses, but the blank records do not establish whether either dose was administered. The event status remains unknown unless further evidence resolves it. (2) It is a confirmed non-administration, because reliable evidence establishes that the dose was not given. (3) The reported administration and its supported amount, if recorded, may be retained with the diary as provenance; the exact time remains unknown. (4) A concentration does not generally identify a unique dose history, because different dosing histories and other sources of pharmacokinetic variation can be compatible with the observed concentration.