ASPO Abstracts
Use of Google Trends to Understand State-level Differences in Two-year Mammography Rates: An Ecological Infodemiology Approach
Category: Behavioral Science & Health Communication
Conference Year: 2023
Abstract Body:
Purpose: To examine the utility of Google Trends (GT) as a surrogate for breast cancer awareness and screening prevalence in U.S. states. Methods: GT is an online tracking system of Google searches, presented as relative search volumes (RSV) from 0 (lowest) to 100 (highest) that are indirectly corrected for population size and internet access. We obtained state level GT data on search terms related to 'mammography', 'breast cancer', 'breast cancer awareness', 'BRCA testing', and 'breast self-exam' from January 1, 2015 to December 31, 2020. We calculated average RSV by state for all of the included search topics, including non-English terms. The outcome was state level age adjusted breast cancer screening prevalence among adults aged 50-74 years who reported receiving a mammogram within 2 years from the Behavioral Risk Factor Surveillance System (BRFSS) 2020. We calculated the Pearson correlation coefficient between state level GT average RSV and BRFSS state level breast cancer screening prevalence. Additionally, we performed a qualitative analysis of the specific search terms used within the above search topics. Lastly, we categorized average RSV according to its median as low (<67) and high (>=67) and determined its predictive value in correctly identifying whether states met the Healthy People 2020 objective for mammogram prevalence within two years (81.1%). Results: Mississippi had the highest average breast cancer screening RSV while Nevada had the lowest average RSV. Nine (18%) states met the Healthy People 2020 mammography objective. Average breast cancer screening RSV was significantly associated with 2-year mammography prevalence (r=0.47, p=0.0006). The most searched inquiries included: 'what is mammogram', 'mammogram age', 'mammogram near me', 'mammogram ultrasound', and 'mammogram cost'. A median RSV threshold resulted in low sensitivity (54.8%) and negative predictive value (26.9%), but high specificity (77.8%) and positive predictive value (92.0%) in detecting states that met the Healthy People 2020 mammography objective. Conclusion: GT may be a useful real-time epidemiological surveillance tool to supplement state wise estimates of mammogram prevalence, although longitudinal studies are needed to determine its predictive accuracy.
Keywords: Google trends, breast cancer, mammography