Beyond one-time PROMs: high-frequency single-item mobile ePROM tracking of postoperative recovery in spine surgery-a...

Between lumbar decompression and the first follow-up visit (~ 6 weeks post-op), the day-to-day course of recovery is not routinely captured in detail. We tested whether a single-item, smartphone well-being ePROM ("How are you today?", 0-100 VAS; higher = better), sent every 2-3 days, could be deployed in routine care, what level of endpoint completion was achieved, whether it showed preliminary convergent validity when measured at the same visit as clinic PROMs (ODI, PHQ-9), and whether serial responses could illustrate postoperative recovery trajectories.

Consecutive patients with degenerative lumbar disease completed a digital baseline survey (Feb 1-Aug 31, 2025; n = 429). Patients with an operative indication for lumbar disc herniation or spinal canal stenosis who underwent lumbar decompression entered postoperative high-frequency well-being monitoring for 6 weeks. Patients were not included in the digital follow-up pathway when web-based participation was not feasible in routine care. The pathway generated electronic health record-accessible submissions but did not include automated alert thresholds or a mandated alert-to-action protocol.

Of 183 operated patients invited to postoperative monitoring, 70/183 (38.3%) contributed an evaluable clinic ODI at ~ 6 weeks, 68/183 (37.2%) had a same-day PHQ-9, and 48/183 (26.2%) were endpoint-complete for both visit-day measures; this low endpoint-completion rate represents a major implementation barrier. In the pair-level timing dataset, same-day well-being versus reverse-coded ODI (100 - ODI; higher = less disability) showed moderate concordance (ρ = 0.54, 95% CI 0.37-0.68; n = 107 pairs), with similar moderate estimates at short lags (≤ 2 days ρ ≈ 0.46; ≤4 days ρ ≈ 0.42), while longer-lag estimates were exploratory and non-monotonic. On the visit day, well-being related inversely to depressive burden (PHQ-9 ρ = -0.25; n = 68). In a joint model using raw ODI and PHQ-9, higher disability and higher depressive burden were associated with lower concurrent well-being (standardized β: -0.35 for ODI; -0.20 for PHQ-9), although explained variance was limited (adjusted R² = 0.151).

Routine-care deployment was technically possible, and time-aligned measures showed preliminary convergent validity, but endpoint completion was low (26.2%) and prevents any claim of broad feasibility or universal scalability. The ePROM should be considered a hypothesis-generating trajectory signal that requires equity safeguards and prospective closed-loop alert-to-action testing before implementation as a clinical decision tool.