Association of internet-integrated palliative care pathway assignment with recorded quality-of-life and symptom-burden scores in patients with terminal cancer: a single-center retrospective propensity score-matched cohort study.

BACKGROUND: Internet-supported models may extend palliative care beyond scheduled hospital encounters, but the validity of comparisons from routine terminal-cancer care depends on clear care pathway assignment and outcome ascertainment. OBJECTIVE: This study aimed to evaluate whether initial recorded assignment to an internet-integrated palliative care pathway was associated with recorded global quality-of-life and symptom-burden scores compared with conventional care. METHODS: This single-center retrospective cohort screened 268 potentially eligible patients in 2023. The operational exposure was the archived group field at pathway entry (internet or conventional), analyzed according to initial recorded assignment; actual platform use, crossover, discontinuation, contact frequency, and care intensity were not verifiable. Eligibility was defined at baseline, and 224 patients entered the candidate cohort. The primary composite distance match produced 85 pairs. Recorded EORTC QLQ-C15-PAL global quality-of-life and ESAS total symptom-burden scores were analyzed with linear mixed-effects models including matched-pair and patient random intercepts. Data validation included duplicate-identifier, exposure-code, date, range, and cross-file consistency checks. Missingness analyses and a conventional 1:1 nearest-neighbor propensity-score match were used to assess robustness. RESULTS: The primary matched cohort included 170 patients, and all absolute post-matching standardized mean differences (SMDs) were below 0.10 (maximum, 0.092). At 4 and 8 weeks, recorded global quality-of-life scores were higher by 10.88 points (95% CI, 8.71-13.05) and 16.75 points (95% CI, 14.55-18.96), respectively, and recorded total symptom-burden scores were lower by 9.94 points (95% CI, 8.22-11.65) and 17.78 points (95% CI, 16.04-19.52), respectively (all p < 0.001). Conventional nearest-neighbor matching yielded 88 pairs and are directionally consistent, slightly attenuated contrasts, although covariate balance was weaker (maximum absolute SMD, 0.163). CONCLUSION: Initial recorded assignment to the internet-integrated care pathway was associated with more favorable recorded questionnaire-score trajectories in the selected matched common-support population. Unverified administration metadata, possible differential measurement error, non-random assignment, incomplete implementation data, and inability to reconstruct visit-specific mortality preclude causal attribution to the internet component.