Facebook Ratings of Hospitals in Switzerland - Coverage of Hospitals, Rating Scores, and Factors Influencing Them
Background: Facebook ratings provide relevant information and data on hospital quality for individual patients, the hospitals themselves and the institutions responsible for the quality assessment of hospitals. However, there are still considerable doubts in the healthcare sector about the extent to which Facebook ratings can actually be used as a valid instrument or input for the quality assessments of hospitals.Objective: The aim of this study was, on the one hand, to make transparent the extent and representativeness of the coverage of the Swiss hospital sector by Facebook ratings and to show which applications are possible with this data. On the other hand, it pursued the goal of systematically identifying possible biases based on the motivation and behavior of individual users, the decisions of hospitals, and the design of the platform.Methods: We collected empirical data on all hospitals in Switzerland regarding their characteristics and Facebook pages. Furthermore, we collected data on dates, types, and rating scores for all 7,103 Facebook ratings on Swiss hospitals on Facebook between 2013 and 2019. We applied cross-sectional and longitudinal descriptive analyses to explore the characteristics of Facebook ratings and their development during the study period. In addition, we applied robust regression analyses to identify the factors influencing hospitals' decisions to disable the rating feature and their average Facebook rating scores.Results: For 39 (14.2%) of the 274 hospitals finally included in this study, no ratings were possible (no Facebook page or rating feature turned off). A total of 173 hospitals (63.1%) received at least one Facebook rating during the study period. For the remaining 62 hospitals (22.6%), receiving ratings was possible, but they did not receive any. Psychiatric hospitals are more likely to disable the rating feature on their official Facebook pages, whereas birth centers have a very low propensity in this regard. The individual Facebook ratings on hospitals have a J-shaped distribution with a very strong preponderance of top ratings (76.1% 5-star rating scores vs. 6.2% 1-star rating scores). Hospitals with higher patient volumes or psychiatric hospitals tend to have lower average rating scores, whereas hospitals with a higher share of patients with voluntary health insurance or birth centers tend to have higher average rating scores.Conclusions: The representativeness and coverage of the Swiss hospital sector by Facebook ratings are reduced to some extent and the ratings are distorted based on certain biases in the rating behavior of individuals, the decisions of hospitals, and the design of the Facebook platform. Considering the patterns of biases found in this study, Facebook ratings can be used beneficially for certain applications on hospital quality assessments. However, primary use should only be complementary to other commonly used hospital quality assessment tools.