Electronic Migration of a 24-Hour Recall Scale

The starting principle is that an electronic PRO is a new mode of administration, even when the words are identical. Before launch, require documented cognitive debriefing of the electronic layout with the target population, including older participants and at least one language other than the source language. The debriefing should confirm that each item is read as intended, that scrolling does not hide response options or change the order in which items are seen, and that the font-size control is discoverable. Require a usability test of the actual device class and operating system, with observed completion and error recovery. The sponsor must document that wording, response options, recall period, and scoring are unchanged from the validated paper version. If the 'usual activity' item is genuinely ambiguous, do not rewrite it in the electronic version without a validated amendment; instead, test whether the ambiguity is increased by the screen layout. Because no equivalence study exists, the protocol and analysis plan should either include a mode-equivalence substudy or explicitly justify why the electronic version is considered a migration rather than a new instrument. Four weeks is tight, but launch without this evidence is not defensible.

The 23:59 cutoff is part of the measurement, not an arbitrary technical setting. A 6-hour late-entry window would let a participant complete at 05:59 for the previous day, when the 24-hour recall no longer matches the intended reporting period. It would also create ambiguous day boundaries: does an entry at 00:30 belong to yesterday or today? Disable the vendor default unless the protocol is amended with a clear scientific rationale and the change is validated. If the team believes late completion is needed, the safer design is to define a new window in the protocol, test it, and keep the timestamp rules explicit, for example local time versus server time and which diary day is credited. Do not let a participant 'think about yesterday' as an informal instruction, because that introduces uncontrolled recall variation. For missed days, record the day as missing and follow the protocol's missing-data rules. The site should remind participants to complete before bedtime and should not tell them that late entry is acceptable when the protocol says otherwise.

Separate presentation problems from content problems. Scrolling on a 5-inch screen is a presentation risk if it hides response options, separates an item from its scale, or changes the visual context. Redesign the screen so each item and its full response scale are visible together, or use one item per screen; then retest with the same type of participants. Do not change item wording to solve a layout problem. The 'usual activity' item is a content ambiguity: if the paper version was validated with that wording, changing it electronically would create a new instrument. Test whether participants interpret it consistently in context; if not, escalate to the instrument owner and consider a protocol-level decision, not an ad hoc edit. The font-size control must be findable without training; add it to onboarding and test it. For eight languages, use a documented translation and reconciliation process, including back-translation where required, and test the longest translated strings on the smallest supported screen. Delay launch if these issues cannot be resolved and documented before first patient in.