How to calculate visit windows (with examples)

What a visit window is

Most protocols schedule visits on a study day, such as Day 15 or Week 12, and allow a window around it, such as plus or minus 3 days. The target date is the ideal day. The window is the range of dates on which the visit still counts as on schedule. A visit outside the window is usually a protocol deviation, so getting the arithmetic right matters.

Every window calculation has four inputs: the anchor date, the visit's study day, the number of days allowed early and the number allowed late. The protocol's schedule of assessments gives you the last three. The anchor comes from the participant's record.

Step 1: Find the anchor

The anchor is the event the schedule counts from. It is often randomization or first dose, but some protocols count from the screening visit, from enrollment or from the previous visit. Read the schedule of assessments footnotes carefully; the anchor can change partway through a study, for example after a treatment period ends and a follow-up period begins.

Write the anchor date into the participant's record as soon as it happens. If the anchor is first dose and dosing is delayed, every later target date moves with it.

Step 2: Know whether the anchor is Day 1 or Day 0

This is where most mistakes happen. Protocols use one of two conventions.

Day 1 convention: the anchor date is Day 1. Day N is the anchor date plus N minus 1 days. There is no Day 0; the day before Day 1 is Day -1. This matches how the CDISC SDTM model defines study day, which is why many sponsors use it (CDISC SDTMIG).

Day 0 convention: the anchor date is Day 0. Day N is the anchor date plus N days. Some protocols, particularly those that count from a procedure or dose given on Day 0, use this convention.

The difference is exactly one day, which is often the whole margin of a tight window. If the protocol does not say which convention it uses, ask the sponsor and file the answer as a note to file.

Step 3: Calculate the target date

Add the offset to the anchor date. Count calendar days. Weekends and public holidays count unless the protocol says otherwise.

Step 4: Apply the window

Subtract the early allowance from the target date to get the window open date, and add the late allowance to get the window close date. A window of plus or minus 3 days around Thursday 15 January 2026 opens on Monday 12 January and closes on Sunday 18 January: seven dates in total.

Watch for asymmetric windows such as minus 0 or plus 7 days, and for windows written in weeks. Plus or minus 1 week means 7 days either side.

Worked example

A participant is randomized on Tuesday 3 March 2026. The protocol uses the Day 1 convention, with randomization as Day 1. The Week 12 visit is Day 85 with a window of plus or minus 7 days.

The offset is 85 minus 1, so 84 days. 3 March plus 84 days is Tuesday 26 May 2026. The window opens 7 days earlier, on Tuesday 19 May, and closes 7 days later, on Tuesday 2 June. If your clinic is closed on the Monday holiday of 25 May, that does not change the window; it only removes one bookable day from it.

Common mistakes

• Treating the anchor as Day 0 when the protocol says Day 1, or the other way round.
• Counting from the previous visit's actual date when the protocol counts from the original anchor, or vice versa.
• Skipping weekends. Calendar days include weekends unless the protocol says business days.
• Not re-calculating after a delayed first dose when the anchor is first dose.
• Booking on the last day of the window, leaving no room if the participant cancels.

Book early in the window, not on the edge

Aim for the target date and book no later than the middle of the window. That leaves room to reschedule without a deviation. When a visit does fall outside its window, document it as the protocol and IRB require; ICH E6 expects deviations to be documented and explained (ICH E6(R3)).

You can check any date with our free visit window calculator. In Kyros Trials, each visit is placed on the calendar with its window from the schedule of assessments, and out-of-window visits are flagged on the dashboard.