OBJECTIVES: Surrogate endpoints are utilized to determine vital status of patients in the real-world setting. PrecisionQ utilizes a fully de-identified clinical data set, which has been harmonized with date of death. This study aimed to understand the survival measures obtained by calculating the vital status based on the proof of life method (surrogate) and by using date of death.

METHODS: A retrospective study including adult metastatic non-small cell cancer(mNSCLC) patients identified in the IntegraConnect database who initiated treatment between 01-Jan-2018 and 30-Sep-2021 were included and followed up through 28-Feb-2022. The index date was 1L treatment initiation. For apparent survival (AS), patients were considered dead if the last visit (office visit, labs, treatment, contact) happened two months or more before the study end date with the apparent date of death imputed as last visit date+30days. Overall Survival (OS) was calculated using the actual month of death.

RESULTS: A total of 3,949 mNSCLC patients who initiated treatment were included. The median age was 70 (27,89) years with 50.7% being male and 57% being white. The median (95% CI) AS was 9.1 (8.6,9.8) months and OS was 12.2 (11.3,12.9) months. The median AS underestimated the survival when compared to the OS, even when stratified by gender. Interrater agreement between the survival status by both methods was substantial by kappa statistic (0.61, 95%CI:0.58,0.63) with the sensitivity and specificity values (95% CI) being 0.98 (0.975-0.987) and 0.59 (0.56,0.61), respectively.

CONCLUSIONS: Patients who may have been transferred to hospice or a different practice would have been considered as dead by AS. Further, AS estimates may be similar to the OS estimates in aggressive cancer types. This study highlights that OS calculated using death dates remains a gold standard and continued research into AS is warranted given the difficulty to obtain OS in some data sources.