Themes, communities and influencers of online probiotics chatter: A retrospective analysis from 2009-2017

Abstract

We build on recent examinations questioning the quality of online information about probiotic products by studying the themes of content, detecting virtual communities and identifying key influencers in social media using data science techniques. We conducted topic modelling (n = 36,715 tweets) and longitudinal social network analysis (n = 17,834 tweets) of probiotic chatter on Twitter from 2009–17. We used Latent Dirichlet Allocation (LDA) to build the topic models and network analysis tool Gephi for building yearly graphs. We identified the top 10 topics of probiotics-related communication on Twitter and a constant rise in communication activity. However the number of communities grew consistently to peak in 2014 before dipping and levelling off by 2017. While several probiotics industry actors appeared and disappeared during this period, the influence of one specific actor rose from a hub initially to an authority in the latter years. With multi-brand advertising and probiotics promotions mostly occupying the Twitter chatter, scientists, journalists, or policymakers exerted minimal influence in these communities. Consistent with previous research, we find that probiotics-related content on social media veers towards promotions and benefits. Probiotic industry actors maintain consistent presence on Twitter while transitioning from hubs to authorities over time; scientific entities assume an authoritative role without much engagement. The involvement of scientific, journalistic or regulatory stakeholders will help create a balanced informational environment surrounding probiotic products.

Recommendations to Improve Quality of Probiotic Systematic Reviews With Meta-Analyses

“…Goodman et al reviewed 42 RCTs of 26 different probiotics, concluding that L casei probiotics were “effective to prevent AAD,” but a closer examination of their data revealed that, of 5 different strains of Lacticaseibacillus (formerly Lactobacillus ) casei studied, only 1 (a 3-strain blend of L acidophilus CL1285, L casei LBC80R, and L rhamnosus CLR2) prevented AAD, whereas the other 4 were ineffective. This type of imprecision continues to occur in the published literature, despite being firmly discouraged …”

“…This type of imprecision continues to occur in the published literature, despite being firmly discouraged. 25,31 When MAs were limited to 1 probiotic strain or used subgroups comprising identical strains (or the same strains in multistrain blends), it was possible to discern which strains might be effective for a given disorder. 35,39,47,50,[52][53][54]56,65 For example, Farahmandi et al 66 reviewed 13 RCTs for allergic rhinitis and found that conclusions could only be drawn on 2 of the 9 strains eligible for analysis (ie, among strains for immune marker responses.…”

“…Guidelines published by large organizations have also provided conflicting recommendations for probiotics because of differences in methods . These issues have led to confusion for the general public and health care professionals when attempting to choose a probiotic for clinical use …”