Planned Versus Actual Times

Why Planned and Actual Times Must Remain Distinct

A study schedule describes when a dose or sample is intended to occur. A source record describes when it actually occurred, as documented. These times answer different questions. Planned times support comparisons with the protocol schedule; actual times represent the timing of observed events. When the two are collapsed into one field, a dataset can no longer distinguish a late dose from a dose given on time, or a delayed sample from a sample collected as planned.

This distinction matters when preparing pharmacometric data. The time elapsed between an actual dose and an actual concentration sample can differ from the nominal interval in the schedule. For analyses that depend on the timing of observed events, replacing actual times with planned times can misrepresent when a sample was collected relative to a dose. Conversely, a schedule-based report may specifically need the planned visit time to assess whether events followed the protocol.

The words “actual time” do not guarantee that a timestamp is perfectly precise. A source may record only the nearest minute, or it may contain a transcription error. Retain the source value and its available precision; do not imply finer precision than the record supports. If the actual time is missing, keep that fact visible rather than silently filling the field with the planned time.

Representing Times and Deviations

For a single event, the planned and actual values should be separately identifiable. Depending on the dataset, each may be represented as a date-time or as a time relative to a defined origin. A date-time must include enough date information to identify the correct day, especially for events near midnight. When relative times are derived, the same reference origin and units must be used for all compared events.

A deviation can be reported in minutes or hours. If the timestamps are available as date-times, subtract them after putting them on the same time basis. If both are expressed in hours from the same origin, subtraction gives a deviation in hours. For example, if an event was planned for 10:30 and occurred at 10:12 on the same date, the signed deviation is \(-18\) minutes, or \(-0.3\) hours. The minus sign is meaningful: the event occurred earlier than planned.

For a dose and a sample, distinguish the nominal interval between their planned times from the elapsed interval between their actual times. Let \(t_{\mathrm{planned,dose}}\) and \(t_{\mathrm{planned,sample}}\) denote planned times, and \(t_{\mathrm{actual,dose}}\) and \(t_{\mathrm{actual,sample}}\) denote actual times. The nominal interval is the difference between the planned sample and dose times; the actual elapsed interval is the difference between the actual sample and dose times. All times must use a consistent origin, units, and event-time convention.

This relationship follows directly by substituting the definitions of the deviations:

Thus, a dose that is late does not by itself determine whether the sample-to-dose interval is longer or shorter than nominal. The sample’s deviation matters too. If both events are delayed by the same amount, their elapsed interval is unchanged. If the sample remains at its planned time while the dose is delayed, the actual interval becomes shorter.

Choosing a Time Field for a Dataset Purpose

The appropriate field depends on the question, not on which timestamp is easiest to obtain. For model inputs that represent the timing of observed doses and samples, use the actual event times when those times are available and fit the model’s event-time conventions. The earlier material on dose-time conventions and reconstructing dose histories establishes why dose timing must be represented consistently. For schedule adherence, protocol deviations, or summaries grouped by planned visit, the planned time is often the relevant reference. Keep both when both questions may arise.

A useful dataset design keeps separate planned and actual fields for doses and samples where applicable. Derived deviations or elapsed intervals can then be calculated from those fields, with units and definitions documented. Do not overwrite the source timestamps with derived values. If a particular analysis requires a single time variable, create that analysis-specific variable explicitly and preserve the underlying planned and actual values so the choice can be reviewed.

Sometimes an actual timestamp is unavailable while a planned time is known. These are not interchangeable facts. A planned value may still support a schedule-based description, but it does not establish when the event occurred. If an analysis requires actual event timing, substituting the planned value can make the event appear observed at a time that was never documented. Any imputation or other substitute must be made explicit and distinguishable from a source-recorded actual time.

Diagnostics, Alternatives, and Limitations

A basic timing review should check whether each time has an interpretable date, time basis, and unit; whether planned and actual values have been placed in the intended fields; and whether the calculated deviation has the expected sign. For dose-to-sample intervals, verify that the actual sample and dose times refer to the intended events and that subtraction uses a common time origin. A negative elapsed interval may indicate a date, event-ordering, or data-entry problem, but it should prompt investigation rather than automatic correction.

Also examine whether deviations cluster in a way that suggests a systematic convention problem. For example, if every sample appears to occur at the planned time despite documented delays, actual times may have been replaced by nominal times. If events near midnight yield implausible intervals, dates may have been omitted or assigned incorrectly. A regular difference of a fixed number of hours may warrant checking how local times and other recorded time bases were handled. These are diagnostic clues, not proof of a particular error; compare with source records and document any resolution.

The principal alternatives are not competing calculations so much as different choices of time reference. Using only planned times is appropriate for questions explicitly about the schedule, but cannot describe actual event timing. Using only actual times supports elapsed-time calculations when records are available, but loses the direct comparison with the intended schedule. Retaining both is more informative and allows each derived quantity to be traced to its inputs.

There are limits to what time differences can establish. A recorded timestamp may be rounded, delayed in entry, or otherwise uncertain; arithmetic cannot recover precision absent from the source. A deviation measures a difference from the planned time, not the reason for that difference. Nor does it, by itself, establish the effect of a timing change on a concentration or treatment outcome. Use the event definition required by the analysis, especially where an administration spans time rather than occurring at one instant, and apply the course’s established event-time convention consistently.

Common Mistakes

• Using the planned time as if it were observed. A scheduled 10:00 sample is not evidence that collection occurred at 10:00. Preserve the actual value separately, or leave it missing if unavailable.
• Dropping the sign of a deviation. “18 minutes” does not say early or late. State the signed result and interpret it in context.
• Calculating an interval from only one event’s deviation. The actual dose-to-sample interval depends on both actual event times; a late dose can shorten the interval if the sample remains on schedule.
• Subtracting clock times across dates as though the date did not matter. Include the date or use a consistent relative-time scale, particularly around midnight.
• Writing a rounded decimal as an exact result. For example, \(4 + 8/60 = 4.133333\ldots\), so the value rounded to three decimals is \(4.133\), not an exact equality.