This paper presents a 27 month time-series analysis of Twitter messaging related to chiropractic/SMT and immunity. From January 1, 2020, Twitter activity increased until March 31, 2020 when it peaked then declined steadily over the remaining 24 months. The ratio of tweets promoting a relation between chiropractic/SMT and immunity and those tweets refuting the same relation, remained relatively constant over the 27 months. Metrics of engagement overwhelmingly supported tweets that refuted a relation between chiropractic/SMT and immunity. Following peak Twitter activity, tweets promoting a relation between chiropractic/SMT and immunity decreased at a rate that was higher than expected. Possible reasons for this observation include efforts by chiropractic organizations and regulators to address misinformation early in the pandemic. There remains a divide between the geospatial origin of tweets promoting a relation between chiropractic/SMT and immunity (United States of America) and tweets refuting this relation (Canada, Europe and Australia).
It should be noted that in our previous paper, chiropractic was mentioned most often in tweets associated with immunity (21%) followed by naturopathy (6%). As a result, this two-year follow-up was limited specifically to the chiropractic profession as it was clearly most often associated with SMT and immunity.
In this two-year follow-up study, our first hypothesis was supported; Twitter activity reached its peak just 20 days following the pandemic declaration. Once reaching its peak, Twitter activity declined steadily without any sign of rebound (Fig. 1, Table 1).
Our second hypothesis was also supported. Through the 27 months of data collection, the ratio of promoting versus refuting tweets remained constant at ~ 50% (Table 2, Fig. 2). This constant ratio suggests that authors of promoting or refuting tweets tend to counter-post in response to tweets of opposing viewpoints thereby balancing out the ratio over time.
Table 2 Engagement of tweets (absolute counts and monthly percentages) stratified by period and content Full size table
Interestingly, engagement and reach of promoting versus refuting tweets were far from equivalent. As was the case in our first paper [8], the total likes, retweets and followers of refuting tweets were orders of magnitude greater compared to promoting tweets. The result was that refuting tweets were much more impactful (Table 1, Fig. 3).
While the decline of Twitter activity after March 31, 2020 could be explained by social media fatigue, confusion, or dilution of attention by other media sources [28], we also observed a parallel decrease in tweets with messaging that SMT boosts immunity (Tables 2 and 3. Figure 4). Interestingly, these tweets declined at a higher-than-expected rate (Table 3) compared to tweets suggesting that chiropractic care or nonSMT interventions improve immunity (both of which declined at a less-than-expected rate).
Table 3 Overall tweet rates with the expected and actual tweet rates between time periods Full size table
The decrease in Twitter activity following March 31, 2020, combined with a coincident decrease in controversial tweets with messaging that SMT boosts immunity, strongly suggests the appearance of some external factor driving these parallel changes. While we cannot confirm the chiropractic announcements plotted in Fig. 6 caused a parallel decline in Twitter activity and SMT messaging, the intended effect was observed; there is quantitatively less misinformation on Twitter regarding SMT and immunity. Interestingly, the timing of these contrary efforts was not associated with any contrary rise in Twitter activity nor contrary increase in SMT messaging.
The resulting decrease in Twitter activity, together with the parallel decrease in tweets linking SMT to improved immunity, may have been sustained by other activities occurring weeks or months after peak Twitter activity and include:
A unified statement from more than 150 chiropractic researchers against the claim that chiropractic care boosts immunity [29].
An interview with a prominent chiropractic vitalistic researcher who stated that “because we have no studies yet that look at would chiropractic care prevent you from getting sick or would chiropractic care reduce the symptoms of being sick or the frequency of getting sick? Those studies haven’t been done yet.” [30].
The emergence of interventions over the course of the pandemic (social distancing, vaccines and anti-viral medications) that mitigated infection and/or serious consequences of covid infection (hospitalization, long-covid, death), acting to make messaging about boosting immunity less relevant, urgent or attention-grabbing.
Changes in Twitter policy designed to target misinformation, and account owners who distribute misinformation.
Although the overall decline in tweets promoting SMT as a positive influence on immunity is a desirable development, we note that the remaining proportion of tweets extolling a positive benefit of chiropractic care on immunity is no less of a concern. Although we cannot know the intent of those posting to social media, we suspect that given the factors listed above (especially increased regulatory oversight), some tweet authors may have consciously or unconsciously developed a Trojan Horse strategy by de-emphasizing controversial messaging about SMT while alternatively promoting the profession that provides it. It must be emphasized here that replacing SMT with chiropractic care to suggest a positive effect on immunity, is also misinformation. As is the case with SMT, there is no evidence that chiropractic care, however it may be defined, generates a clinically meaningful improvement in human immunity compared to those withheld from the same intervention. Importantly, we acknowledge studies that report changes in immune parameters following SMT, but these studies have not shown clinical significance in humans. They join an almost endless list of other studies showing any number of changes in anatomy, physiology, various biomarkers and neurology post-SMT. The critical point in the evolution of this body of literature is that for any of these observed changes to be meaningful, these changes must result in a clinically important improvement in human health compared to persons who do not receive the same intervention(s) [11].
Our observation that the majority of promoting tweets originate in the United States is in agreement with the data from our prior paper. While it is difficult to know the global extent of all prompting and refuting messaging outside of Twitter, we also note that announcements from chiropractic organizations that promoted the idea of chiropractic/SMT improving immunity also came primarily from the United States. Explanations for this geographic separation are not readily available, but possible avenues of future investigation may include comparing the proportion of senior versus early career chiropractors in various countries and the location of chiropractic schools that emphasize conservative or dogmatic chiropractic views [31].
Limitations
It is important to emphasize that it is not possible to confirm the occupation or affiliation of those who author tweets. In addition, Twitter was selected for this study as its entire corpus is searchable. While there is evidence that other social media outlets such as Facebook have many more posts regarding this issue, the majority of these posts occur within private groups and are therefore inaccessible to systematic searching and analysis. The ability to search and track content of Tweets is an advantage and is indicative of the volume of activity in a specific topic. It is also a common way of measuring impact of social media and it is done in many different ways including through Altmetric (www.altmetric.com). However, it must be remembered that measures of engagement do not guarantee that tweets, like any other written content, influences actions or public opinion.
In rating Tweets, the TTI is a new tool that is not used widely, therefore, we took great care discussing its development in our prior paper and then how it was applied here by the same investigators using the same processes to resolve any disagreement.
Finally, it is important to note that we do not infer causal relationships between events and twitter claims; we simply describe what was observed and suggest there is a striking pattern between these events.