Panel Paper: Social Media, Advertising, and Hospital Quality

Friday, November 9, 2018
Wilson A - Mezz Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Victoria Perez and Seth Freedman, Indiana University


Over the last 7 years, hospitals have increased their presence on social media and crowd-sourcing quality rating sites, such as Facebook, Google and Yelp. In a recent consumer survey, 40% of people identified social media as a factor in their provider choice. By engaging with the public on these sites, hospitals raise their community profile and attempt to market themselves along a number of clinical and non-clinical quality measures. The extent to which these online profiles influence patient choice and outcomes is not well understood. Hospitals also attempt to influence patient choice through more traditional advertising. In this paper, we estimate the impact of hospital-initiated online content, consumer engagement with that content, and advertising investments on patient hospital choice. We ask whether these types of outreach lead patients to choose hospitals with better or worse quality along dimensions of both clinical and non-clinical quality.

This study draws on a unique dataset of hospital-initiated online Facebook content and consumer responses to that content, as of July 2017. We then supplement this data with discharge data from the Hospital Cost and Utilization Project, advertising data from Kantar Media, which tracks advertising across media types, including internet, print, and television, and Hospital Compare measures of clinical quality and patient satisfaction. We first measure the effect of social media engagement and advertising on patient choice of hospitals to establish whether patients respond to these sources of information. We then test whether changes in patient volume are consistent with patients being more or less likely to visit hospitals of higher clinical quality based on risk-adjusted mortality and readmission rates and non-clinical quality based on patient satisfaction surveys. We separately consider effects for non-emergent conditions, which allow patients to choose a site of care in advance, relative to emergent conditions.