Autosaved Incident Report Drafts: Log-Based Evidence of Completion, Return-to-Complete Behavior, and Reminder-Linked...

BACKGROUND: Patient safety reporting studies usually examine intention, submitted reports, or reporting rates, leaving unclear what happens after a report is started but before it is submitted, deleted, or left unfinished. We evaluated whether autosave and fixed-time reminders make this hidden stage observable.

OBJECTIVES: This study aimed to quantify draft outcomes, interaction burden, completion timing, return-to-draft behavior, and reminder-linked submission among autosaved incident report drafts.

METHODS: This was a retrospective log-based study of draft header records and action logs from March 2024 through March 2026. Reminder exposure was defined as a draft remaining incomplete at the daily 05:31 batch; return as a load action within 24 hours after first eligibility. Analyses used descriptive statistics and site-adjusted logistic regression with reporter-clustered standard errors.

RESULTS: Among 6,835 drafts, 5,397 (79.0%) were submitted, 1,268 (18.6%) were deleted, and 170 (2.5%) were pending. Completed reports were usually rapid (median: 15.0 min; 81.7% within 60 min), but 1,961 drafts (28.7%) had at least one return/load action. Among submitted drafts with action logs, the median interaction burden was 2 actions (interquartile range: 1-3). Overall, 1,617 drafts (23.7%) were reminder-exposed; 337 (20.8%) returned within 24 hours. Return within 24 hours was strongly associated with submission (74.2 vs. 19.8%; adjusted odds ratio: 11.16, 95% confidence interval: 7.29-17.09).

CONCLUSION: Draft-level logs expose the previously unmeasured middle of incident reporting. Autosave captured interrupted work, and reminder-linked return identified a high-yield recovery point for at-risk drafts. These findings show how interruption-resilient design can support actual report completion, not only reporting intention or submitted-report counts.