Evaluation of studies with commonly used real world data sources submitted to ASCO journals: Initial steps toward furthering transparent reporting.

Abstract

6630 Background: Real world data (RWD) are used to understand treatment, care, and outcomes of patients with cancer. Reporting guidelines for observational studies, such as RECORD (health), RECORD-PE (treatment) and TRIPOD (prediction), were developed to increase the transparency of studies using RWD. These guidelines are not required for publication in ASCO Journals (JCO- Journal of Clinical Oncology, PO- Precision Oncology, OP- Oncology Practice, CCI- Clinical Cancer Informatics, GO- Global Oncology). Our objective was to describe the use of RWD among studies submitted to ASCO Journals and evaluate the applicability of guidelines and adherence to guidelines. Methods: ASCO submission system was used to identify all studies submitted to ASCO Journals between 2018-2022. Based on the abstract, the studies were categorized on whether selected RWD were used: SEER, Medicare, National Cancer Database [NCDB], Flatiron, CancerLinQ, IQVIA, Optum, MarketScan. Through manual review of the published manuscripts, we calculated positive predictive value (PPV) of our algorithm to identify the RWD. We estimated acceptance rate. For each study we evaluated the following: (1) linkages with other RWD; (2) applicability of reporting guidelines RECORD, RECORD-PE, TRIPOD; (3) use of flow diagram (RECORD) or study design figure for treatment and outcome assessment (RECORD-PE). Results: The PPV of our search algorithm was 79% (95%CI 73-84) but varied by journal (PPV 73-100; p=0.02). Based on our algorithm, among the 16,133 original reports submitted, 1,156 (7.2%) used one of the RWD sources. Overall acceptance rate among RWD sources was 18%. There was variability in submissions and acceptance rate of RWD by journal. OP and GO most and least frequently received a manuscript with the selected RWD 13% and 1%, respectively. GO accepted 77% of manuscripts that used selected RWD (88% of which were SEER). JCO accepted 3.5% of RWD manuscripts. Medicare, SEER, and NCDB were frequently identified data sources (61%, 31%, and 11%, respectively). SEER was linked to Medicare claims, surveys, or other sources in 77% of studies. The other data sources accounted for <5% each, although submission of EHR-linked to claims/registry increased over time (OR 1.26: 95%CI 1.07-1.49). RECORD, RECORD-PE and TRIPOD were applicable to 100%, 10% and 8% of the studies. RECORD-PE and TRIPOD were most applicable to PO/OP/CCI and CCI, respectively. A flow diagram demonstrating selection of participants was present in 31% of the studies. A figure to visualize study design was used in <5% of RECORD-PE relevant studies. Conclusions: RWD are commonly submitted and published within ASCO Journals. Reporting guidelines are applicable and particularly important when linkages occur. Studies using oncology RWD may benefit from reporting guidelines tailored to the unique qualities of these data sources and populations.