S02: An Evaluation of Trust-Building Measures Within Pharmaceutical Brand Websites in Comparison to a Social Media Trust Analysis

Poster Presenter

Angela Greeley Wagner

Biomedical Writing Graduate Student
University of the Sciences
United States

Objectives

The objective of this study was to employ a new model of patient-pharmaceutical brand trust to observe whether a correlation existed between pharmaceutical brand-specific patient trust sentiment on social media and trust-building features of consumer websites for those brands.

Method

For this study, we reviewed trust literature to construct a model applicable to patient-pharmaceutical brand interactions. We used 18 innovator pharmaceutical brand names in a Twitter trust analysis and inspected 18 brand websites for trust-building features. We reviewed the results for correlation.

Results

We created Twitter API calls for two one-week periods in January 2021 using 18 brand keywords, and we procured 1934 unique tweets. Our subsequent manual trust analysis produced a range of trust scores between -0.17 and 0.72 (higher = more trusting). We analyzed 18 brand websites, with a total of 44 indications, for 4 components of the trust model in January and February 2021 and obtained mean scores by brand for each component. These scores were normalized and compared to the Twitter trust scores. The small sample size (n = 17, 1 brand excluded from the final analysis due to insufficient Twitter data) did not merit a full statistical analysis; the correlation was reviewed visually, and outliers were identified using standard measures (eg, 1.5 * IQR method). Preliminary data visualization demonstrated a roughly positive correlation between website trust-building features and consumer trust sentiment toward specific pharmaceutical brands. We observed a closer correlation for some brands than for others, but in general, a certain degree of brand-specific trust observed in the Twitter analysis corresponded to a similar degree of trust-building features observed on the brand website.

Conclusion

This study supports the use of a patient-pharmaceutical brand model of trust for brand-specific data analyses. Employing this model may assist brands in enhancing trust-building website content and functionality. Using this model may also improve the accuracy of trust-based sentiment analysis for social media, which is currently limited to positive, neutral, or negative sentiment but which does not accurately correlate to patient trust in or distrust of a brand. Future applications of this model may include software tools for trust-specific sentiment analysis, trust-based pharmaceutical website usability studies, and social media-based trust analysis campaigns.