585 - Predicting Lead-time Respiratory Syncytial Virus-related Pediatric Hospitalizations from Historic Google Trend Search Activity in the United States
Monday, April 25, 2022
3:30 PM – 6:00 PM US MT
Poster Number: 585 Publication Number: 585.417
Sahithi Pemmasani, Icahn School of Medicine at Mount Sinai, Elmhurst, NY, United States; Mercy Danso-Aboagye, None, Worcester, MA, United States; Jacob H. Umscheid, University of Kansas School of Medicine, Wichita, KS, United States; Ernestina H. Bioh, Ghana Health Service- Western Region, Oakland, CA, United States; Narendrasinh Parmar, East Tennessee Children's Hospital, Knoxville, TN, United States; Rhythm Vasudeva, University of Kansas School of Medicine, Wichita, KS, United States; Keyur Donda, University of South Florida, Tampa, FL, United States; Harshit Doshi, Pediatrix Medical Group of FL, Venice, FL, United States; Tarang Parekh, George Mason University, Elmhurst, NY, United States; Parth J. Bhatt, United Hospital Center, Bridgeport, WV, United States; FREDRICK Dapaah-Siakwan, Valley Children's Hospital, Madera, CA, United States
Resident Icahn School of Medicine at Mount Sinai Elmhurst, New York, United States
Background: Respiratory syncytial virus(RSV) is responsible for more than 120,000 hospitalizations in the United States each year. Since collecting real-time data on the timing and intensity of RSV is resource-intensive, it is imperative to explore novel approaches to monitoring and forecasting RSV-related hospitalizations.
Objective: To examine the utility of Google Trend (GT) search activity on RSV to predict changes in RSV-related hospitalizations in children in the USA, 2019
Design/Methods: We performed a retrospective cross-sectional study of RSV-related hospitalizations within the HCUP’s Kid’s Inpatient Database in 2019. GT search activity for RSV in US was obtained for each month in the calendar year 2019. We applied the Finite Distributed Lag(FDL) model controlled for race, sex, and hospital regions to estimate the causal effect over time of how historical relative search activity might influence the current RSV-related hospitalization rate. A two-lag period model was selected based on the Akaike Information Criteria. We also calculated the Long Run Propensity(LRP), which is reported as the cumulative effect of changes in relative search activity on hospitalization rate. The change in the number of hospitalizations with differences in relative interest is reported as marginal differences in RSV-related hospitalization.
Results: The average monthly RSV-related hospitalizations were 3,044 (SD= 3220.7) and the average relative interest score for RSV-related search [scaled from 0 (lowest interest to 100 (highest interest) units by GT] was 24.9(SD= 23.1) in 2019. Across two successive months, the cumulative effect of a 1-unit score increase in relative interest was associated with an increase of 140.7(95%CI: 96.2–185.2, p< 0.05) RSV-related hospitalizations in 2019. Figure 1 shows RSV-related hospitalization rate differences with lagged relative search over two months. The average difference of decline in relative interest score from January to March predicted marginal difference of hospitalizations declined by 1,885.6 in March. Similarly, the average difference of increased relative interest score from September to November predicted increased marginal difference by 3,760.2 RSV-related hospitalizations in November.Conclusion(s): Google Trend search activity correlated closely with changes in RSV-related hospitalizations in the USA. A forecast model from GT activity on RSV can potentially serve as an early warning system to any surge in RSV-related hospitalizations, allowing policymakers and healthcare systems to plan and ensure the availability of adequate resources for impending RSV outbreaks. Predicting Lead-time Respiratory Syncytial Virus-Related Pediatric Hospitalizations from Historic Google Trend Search Activity in the United StatesCV_January2022_SahithiP.pdf