(252) Impact of COVID-19 Pandemic on Administrative Data – Examples using Mammography Screening and Healthcare Utilization after Cancer Diagnosis

Sunday, August 28, 2022

8:00 AM – 1:45 PM CEST

Location: Hall C

Publication Number:

Presenting Author(s)

Paul V. Petraro, DSc
– Executive Director, Real World Evidence Analyic Center of Excellence, Boehringer Ingelheim International GmbH, Ridgefield, Connecticut, United States

Co-Author(s)

Sabina O. Nduaguba
– Dr, University of Florida College of Pharmacy, Morgantown, West Virginia, United States
Haikun Bao
– Boehringer Ingelheim Pharmaceuticals, INC, United States
Montse Soriano Gabarro, MD, MSc, FISPE, Bayer Distinguished Science Fellow
– Head Partnerships and Integrated Evidence Generation Office, Bayer AG, Berlin, Berlin, Germany
Fidelia Ida
– MCPHS University, King of Prussia, United States
Dana Yael Teltsch, PhD
– Takeda Pharmaceuticals International, Lexington, Massachusetts, United States

Background: The COVID-19 pandemic has resulted in behavioral and policy changes that impact health services utilization, which form the basis of administrative data.

Objectives: To assess changes in health services utilization following the onset of the pandemic in IBM Market Scan Commercial & Medicare Supplemental Database using cancer as a case study and March 2020 as the pandemic start date in the US.

Methods: Commercially insured women 40-64 years were identified each month from January 2017 to March 2021. Monthly proportion of mammography screening (CPT codes: 77057, 77063, 77067, G0202) was determined, overall and by US region.
Three cohorts were identified for patients ≥18 years diagnosed with a malignant cancer (ICD 10: C00-D48) during 03/01/2018 – 08/31/2018 (C1, N=295,922), 03/01/2019 – 08/31/2019 (C2, N=221,261), and 03/01/2020 – 08/31/2020 (C3, N=163,566). The average number of office visits, hospitalizations, emergency department visits, and prescription fills per patient in each cohort six months pre- and post-diagnosis was determined.

Results: Mammography screening proportions ranged from 3.1%-4.4% except for March-May 2020 where it dropped to 0.2%. Regional trends were similar. Among cancer patients, pre-diagnosis office visits (C1: mean (SD)=4.06 (5.39) vs 5.98 (6.49), C2: 4.13 (5.42) vs 5.97 (6.46), C3: 3.73 (5.03) vs 5.98 (6.59)), hospitalizations (C1: 0.05 (0.27) vs 0.1 (0.43), C2: 0.05 (0.27) vs (0.1 (0.42), C3: 0.06 (0.29) vs (0.11 (0.45)), ED visits (C1: 0.11 (0.44) vs 0.13 (0.48), C2: 0.12 (0.43) vs 0.13 (0.48), C3 (0.12 (0.47) vs 0.13 (0.51)), and prescription fills (C1: 7.81 (10.7) vs 9.96 (12.79), C2: 7.81 (10.72) vs 9.82 (12.57), C3: 8.15 (10.87) vs 10.55 (13.05)) increased slightly after diagnosis but post-diagnosis utilization remained stable during the pandemic (C3).

Conclusions: Findings show an expected disruption in cancer screening behavior with a quick return to baseline rates that warrant further investigation. It further suggests that once a cancer diagnosis was made, patients utilized health services despite the pandemic situation. However, treatment delays coinciding with the screening disruption period (March-May 2020) cannot be ruled out. Epidemiologists using administrative data may want to consider methods that address such disruption e.g. restricted time periods, rolling averages.