Patient-reported symptom monitoring linked to an alert algorithm: Effects on progression-free and overall survival in the SYMPRO-Lung trial.

Abstract

12099 Background: Physician-initiated (active) monitoring of patient-reported outcomes (PROs) improves health-related quality of life (HRQOL) and overall survival (OS) in comparison to care as usual. We recently reported data from the SYMPRO-Lung trial showing for the first time that patient-initiated (reactive) PRO monitoring is equally effective as physician-initiated (active) monitoring in improving HRQOL. Here we investigate whether online PRO symptom monitoring via the active and reactive approach is also associated with improvements in 1-year progression-free survival (PFS) and OS compared to care as usual in lung cancer patients. Methods: SYMPRO-Lung is a Dutch multicenter randomized trial using a stepped wedge design. Inclusion criteria were (non-)small cell lung cancer stage I-IV, starting treatment. Patients in the intervention groups reported PRO symptoms weekly up to 1 year follow-up using the PRO-CTCAE lung cancer subset. If symptoms exceeded a pre-determined threshold, an alert was sent to a physician/nurse (active intervention group) or to the patient themselves (reactive intervention group). Survival was defined as time from start of treatment until progression (PFS), death (OS) or loss to follow-up. Follow-up time of all groups was calculated using the reverse Kaplan-Meier method. Cox proportional hazard analyses were performed to estimate the effect of PRO symptom monitoring on PFS and OS, adjusting for potential confounding (i.e. clinical characteristics and baseline EORTC QLQ-C30 domain scores). Results: 515 patients were included in the study (266 control, 249 intervention). The most pronounced uneven distribution in baseline characteristics was that the active intervention group had significantly more stage IV patients compared to the control group and the reactive intervention group (63% vs. 44% vs. 44%, p = 0.001). The median follow-up duration was 12 months for all groups. Age, sex, cancer staging, ECOG performance status and some baseline EORTC domain scores affected the association and were included as covariates in the analyses. Although not significant, we observed a tendency towards a PFS benefit for the active intervention group compared to care as usual (HR .78, 95%CI: .58 – 1.04) and an OS benefit for both of the intervention groups compared to care as usual (active: HR .80, 95% CI: .55 – 1.15; reactive: HR .69, 95% CI: .42 – 1.15). Conclusions: Although not significant, weekly PRO symptom monitoring linked to an alert algorithm showed a tendency towards a small 1-year survival benefit in lung cancer patients. Since we did not power the trial to demonstrate a survival benefit, the non-significant association might be due to a limited sample size. However, the positive results of symptom monitoring in improving HRQOL within our trial still provide sufficient evidence for implementation of PROMs in standard (lung) cancer care. Clinical trial information: NL7897 .