(174) Evaluation of Changes in Lung Cancer Care Patterns During The COVID-19 Pandemic

Publication Number: 1219

Background: The COVID-19 pandemic has affected US cancer care, potentially impacting studies using real world data (RWD) from this time.

Objectives: To evaluate changes in cancer care of lung cancer patients during COVID era, which may introduce confounding or bias in RWD studies.

Methods: We used the nationwide Flatiron Health electronic health record (EHR)-derived de-identified database to compare patients diagnosed with advanced non-small-cell lung cancer (aNSCLC) in (latent) COVID era [January 1, 2020 - September 30, 2021 (most recent data with ≥ 60 days follow-up after diagnosis)] with those diagnosed in pre-COVID era (January 1 - December 31, 2019). We calculated proportions of patients diagnosed in each month with following variables of interest: (1) timely visit: having an oncology clinic visit within 60 days of diagnosis; (2) timely systemic treatment: initiating first-line (1L) treatment within 60 days of diagnosis; (3) among those with 1L, baseline missingness pattern and distributions of demographic (sex, age, race/ethnicity, region) and clinical characteristics (practice type, body mass index, ECOG, stage, histology, smoking history & routine lab tests); and (4) baseline biomarker testing status. Baseline data accrued from 30 days prior up to 7 days past the 1L initiation. Means and 95% confidence intervals of each variable of interest across all months in 2020 & 2021 were compared separately to those in 2019.

Results: In 2019, an average of 617 new aNSCLC cases were recorded monthly compared to 556 in 2020 and 488 in 2021, a drop of 10% and 21% respectively. However, once diagnosed and despite COVID, patients in 2021 & 2020 were as likely or slightly more likely to have timely visits and receive timely 1L than 2019 patients [% with timely visit: 91% (91-92%) in 2019, 92% (91-92%) in 2020, 95% (95-96%) in 2021; % with timely 1L: 62% (61-63%) in 2019, 64% (62-65%) in 2020, 64% (62-65%) in 2021]. Missingness pattern and distribution of baseline characteristics were similar across years except stage: among 1L patients, compared to 2019 [60% (59-62%)], the proportion of de novo stage IV was similar in 2020 [60% (59-62%)] but higher in 2021 [65% (63-66%)]. The likelihood of biomarker testing slightly increased across years overall (68 to 73% from 2019-2021) and for individual biomarkers (e.g., PDL1: 52 to 55%, EGFR: 56 to 63% from 2019-2021).

Conclusions: Some shifts in aNSCLC care during COVID era were seen via fewer diagnoses but more advanced stages at diagnosis, although the impact from the EHR network composition cannot be ruled out. Future work with other data sources & longer follow-up time is necessary to confirm this finding. Other aspects of aNSCLC care in this oncology network remains stable, suggesting limited COVID impact from these aspects on RWD studies.