Weekly Diary, Shared Phones, and a Mid-Study Device Decision
An open-label phase 3 non-small cell lung cancer trial enrolls 62 participants across 14 sites in three countries. The protocol specifies a nine-item weekly ePRO diary completed on a provisioned Android tablet between Sunday 18:00 and 23:59 local time. At week 4, overall compliance is 71%; by week 12 it is 52%. Site 07 (n=11) is at 38%. Participants over 65 make up 41% of enrollment, and 22% do not own a smartphone. Help-desk logs show repeated calls about charging, forgotten PINs, and notifications arriving on Monday. The sponsor proposes switching at week 14 to a bring-your-own-device (BYOD) model with an SMS link, because courier costs are high and two rural sites have no reliable device replacement. The protocol has no BYOD option. The primary endpoint is change from baseline at week 16. The medical monitor asks for a recommendation before the next steering committee.
1. Should the trial switch to BYOD at week 14? Identify the evidence, approvals, and operational controls you would require before any change.
Read the worked answer
A mid-study mode switch is not a logistics fix; it is a measurement change. The first question is whether the compliance problem is device-related or behaviour-related. Charging, PIN, and Monday-notification complaints suggest operational causes that can be fixed without changing mode. If the sponsor still wants BYOD, require a protocol amendment, IRB/EC approval, and any required regulatory notification before any participant uses a personal device. Require documented usability testing on representative personal phones, including the oldest enrolled participants, and evidence that item wording, response options, recall period, and branching are unchanged. Require a validated audit trail, secure authentication, remote wipe, and a rule that a personal device cannot be shared. Plan how data from provisioned and personal devices will be kept distinguishable, and pre-specify a sensitivity analysis because mode may affect reporting. Do not silently pool modes. Operationally, offer a hybrid: keep provisioned devices for participants without smartphones and for site 07, and use BYOD only where approved. A blanket switch at week 14 would create a confounded endpoint and is not justified by the current evidence.
2. What can the team do immediately, without a protocol amendment, to raise compliance at site 07?
Read the worked answer
Stay inside the protocol-defined Sunday 18:00–23:59 window. Do not extend the window, because that changes the recall period and creates a protocol deviation. First, audit the site's device log: are tablets charged, are notifications enabled, are participants locked out, and are PIN resets being handled? Replace chargers and any faulty tablets through the courier process, and set up a site-held loaner pool if courier delays are the barrier. Retrain site staff on device issue, charging routines, and what to say when a participant misses a week; the conversation should support completion, not pressure or coerce. Ask participants what time on Sunday works and whether a family member can help with charging, with consent. Add a Monday check of the ePRO dashboard so the site calls participants before the next window. Document all contacts and device issues. Track compliance weekly by participant and by site, and escalate if site 07 does not improve after retraining. These actions are operational, do not alter the instrument, and can be implemented immediately.
3. How should the 12-week data be handled if 18 participants have no entry for weeks 10–12?
Read the worked answer
Treat the missing entries as missing data, not as data to reconstruct. Confirm from the vendor audit trail whether the entries were never opened, opened but not submitted, or lost through a technical fault; those are different problems. Query the site for documented contact attempts and device failures. Do not backfill from memory, and do not create ePRO entries after the window. If the protocol allows paper backup only for technical failure, and contemporaneous paper exists, use it as source with a documented reason; otherwise leave the weeks missing. Report the missingness by site and by week, because site 07's low compliance may create non-random missingness. The primary analysis should follow the pre-specified method; if complete-case analysis was planned, add a sensitivity analysis that does not assume missing at random. The steering committee should see the pattern before week 16, because a 52% compliance rate threatens interpretability more than a single-site deviation.