In 2014 the Ebola outbreak in West Africa became a global health emergency, spurring an international effort to trial candidate Ebola vaccines and other potential prevention and control measures. The European Union Innovative Medicines Initiative (IMI) funded several projects to help combat the epidemic, one of them EBODAC (EBOla Vaccine Deployment, Acceptance and Compliance).

Within its broader communication and community engagement mandate, EBODAC addressed the risk of misinformation and controversies undermining the trial of a prime-boost Ebola vaccine regimen, developed by Janssen Pharmaceutical Companies of Johnson & Johnson (Janssen). Rumours were already hindering the Ebola response in the affected countries [1], [2] and the early detection and analysis of emerging rumours around the vaccine trials was considered a priority for EBODAC [3].

In February 2015, the EBODAC consortium launched a system for real-time monitoring and analysis of online newspapers in Sierra Leone, and in April 2015 the media monitoring system was extended to include Ghana, Uganda and Kenya, where Phase I trials for Ebola vaccines were also being planned.

In late May 2015, a controversy was reported in Ghanaian newspapers and became inflamed into a public debate, finally resulting in the suspension of both the planned Phase 1 trial for the Janssen Ebola vaccine at the University of Health and Allied Sciences (UHAS) in Hohoe, Volta region, and a planned Phase 2 trial for the GlaxoSmithKline (GSK) Ebola vaccine in Hohoe and in Kintampo. This controversy was exacerbated by the fact that, when these critical articles first became public, both the Minister of Health and the trial lead investigator were travelling outside Ghana – a fact which should be borne in mind while reviewing this analysis.

At the time the controversy broke out, the Ghana Food and Drugs Authority (FDA) was considering both vaccine trials for approval. Five months after the suspension, both vaccine trials were finally approved by Parliament [4]. To date, however, neither trial has proceeded since receiving this approval.

This article examines the messages and perspectives put forth by different stakeholders in the controversial vaccine trials in Ghana through the use of a media monitoring and analysis system. The study analyses the stakeholder concerns and incentives, and the phases and dynamics of the dispute. It provides a unique methodological approach to studying complex interactions between public health authorities, political actors, and public citizens, offering lessons for future research.

To monitor news in each country with planned vaccine trials – Sierra Leone, Ghana, Uganda, and Kenya – the project team selected online newspapers in each country. The selection was based on a qualitative assessment of expected impact and maximum variation to ensure detection of any news regarding the planned trials. Services like Google News Alert¹ were considered for monitoring new articles meeting specific criteria, but in several of the target countries the major search engines indexed only a few of these news outlets, and these services were therefore not appropriate. Instead a PHP-based² system was set up to monitor the front page of the outlets for new articles every hour, downloaded any new articles and scanned the articles for variants of the keywords “Ebola” and “Vaccination”. The system also forwarded all articles containing relevant keywords to the Ebola vaccine program communication crisis team.

Using a thematic analysis [16] approach, three independent coders (LL, WS, PEK) reviewed a subset of articles and were assigned to identify between 5 and 15 themes. Findings from a previous study [17] investigating online vaccine sentiments were used as a starting point for the codebook, which was further adapted to the Ghana context. The team met to compare their proposed themes and 24 were initially agreed upon, but after a change in coders (ES, WS, PEK) an additional three themes were added, for a total of 27 themes [see: Appendix 1].

Theme inclusion criteria were: a) they should be repeated in at least two different news outlets; and b) they should be able to be reduced to declarative statements about the trials or their proponents/opponents.

Two overarching types of themes were defined: (1) “critiques” which directly or indirectly argue that the trials should not take place, and (2) “rebuttals” refuting the critiques, sometimes introducing new arguments in favour of the trials.

Based on the codebook, a web-based coding interface was developed allowing the coders to mark example text and attach relevant coding tags to the text [see: Appendix 2]. When all coders had completed the first round of scoring, the coders unanimously agreed on 87% of the coding decisions. This gave an inter-rater reliability (IRR) score of 0.62, calculated using Fleiss’ kappa, which can be considered a high level of agreement [18]. The main reason for calculating IRR here is to improve reliability. Inter-rater disagreement can arise from actual disagreement, but when dealing with large amounts of data, it is often based on misreading or overlooking themes that are present.

In the second round, where one coder disagreed with the other two, s/he was presented with the other raters’ coding, accompanied by a relevant text extract from the article, and asked whether s/he agreed. If s/he still disagreed it was sent back to the others. This was repeated, and virtually all discrepancies were reconciled. After two rounds of the reconciliation process there were only five disagreements left, giving an IRR which was very close to 1.0. In these remaining five cases the coders agreed that the coding decisions were fundamentally subjective [see: Appendix 3 for more details on the coding procedure].

A total of 27,460 articles from nine different newspapers were collected on-line from Ghana in the time period from May 1st to July 30th 2015. Of these, 150 articles in six different newspapers contained the keywords. Eleven of the 150 articles were excluded because they referred only indirectly to the trials, or not at all. The final dataset included 139 articles (Fig. 1) [see: Appendix 3 for a list of analysed articles].

The 139 reports analysed were collected from six news outlets (Fig. 2) in which we identified 385 instances of critiques and 279 instances of rebuttals.

Because the codebook specified more negative themes than positive themes, no conclusions should be drawn about overall ratio of positive versus negative sentiment. Instead, we gain insights by tracking groups of related themes over time, how they evolved, identifying which critiques were more prevalent, and which rebuttals were issued in response. There were three key areas of critique: (1) the allegation that the trials were secret; (2) the belief that the trials would lead, in one way or another, to Ebola cases in Ghana; and (3) the argument that the trials were providing inappropriate incentive packages for participants.

Figure 3 shows the major organisations, listed in order from most-mentioned to least-mentioned, referenced in the media reports about the trials. Blue bars indicate the number of articles mentioning a given organisation shown against grey background bars representing the total number of included articles published on a given day. The Ghana FDA was named in the very first articles in mid-May and remained a key actor throughout the debate, whereas the Ministry of Health and World Health Organisation (WHO) became more involved in mid-June. Early and ardent critics of the trials, such as the Coalition for Ghana’s Independence Now (CGIN) and the National Democratic Congress (NDC), were most involved in early June, whereafter their salience faded. Meanwhile the Ghana Academy of Arts and Sciences (GAAS) maintained a low but persistent media presence through to early July.

Figure 4 shows the prevalence of positive and negative themes over time. The first article in our dataset was published 21 May 2015 by Starr FM Online, the website for the Ghanaian radio station Starr FM. This article broke the story, presented as an “undercover” investigation into a secret trial, which remained a common theme throughout the controversy (see section "Secret trial and insufficient public information"). It also drew attention to the incentives offered for participation in the trial, which likewise remained relevant for many weeks (see section "Incentives").

The story went seemingly unnoticed for more than a week, until a press release was issued by the CGIN.³ The statement (see Additional file 1), first published by Starr FM Online (30.05), claimed that the trial researchers would infect healthy Ghanaians with Ebola in order to test the vaccine (see section "The fear that vaccine trials will bring Ebola").

The story received increasing coverage over the following week, until the Volta regional office of the NDC political party issued a statement opposing the trials, which was then raised in Parliament. At this point, critiques spiked dramatically, and rebuttals followed.

As shown in Fig. 5, a variety of themes were prominent in the ensuing weeks, including arguments that the trials were “unnecessary” because Ghana was unaffected by the Ebola outbreak, that other diseases were more important (Starr FM Online, 08.06a), and that the trials should be conducted in a country other than Ghana.

Rebuttals tended to be broader statements, making the case that the risk to participants was small and that there had been no deviation from established protocols (GhanaWeb, 08.06a), emphasising the global importance of developing an Ebola vaccine (GhanaWeb, 08.06a), and also that Ghana had an interest in the publicity and economic benefits of hosting such research (Modern Ghana, 16.06b). They also pointed out that the vaccines had already undergone some safety checks as they had been tested in other countries (GhanaWeb, 08.07).

The issue of “ignorance” emerged in a prolonged back-and-forth between Professor Alex Dodoo and the Ghanaian Parliament, during which Dodoo was summoned to explain his alleged criticism of Members of Parliament (MPs) as “ignorant” regarding the procedures of clinical trials.⁴ While this prompted considerable reaction, most of the related articles were as much about Parliament itself as they were about Professor Dodoo’s comments.

A chronological overview of the major events can be found in Appendix 1. To give a visual summary of the coded data, a timeline of theme density is shown in Fig. 6.

The rumours identified in this dataset highlight the vital importance of understanding the emotional context of public concerns: a story of secret vaccine trials in Ghana, or scientists intentionally infecting Ghanaians with Ebola, is only credible in the context of deep mistrust. Indeed, the trials debate is marked, from beginning to end, by the overwhelming fear engendered by the Ebola outbreak, as well as a pervasive anxiety to distance Ghana from the disease, even at the expense of research that could help protect against it. Since the trials themselves were partly sponsored by Western pharmaceutical companies, opposing them could be portrayed as a show of independence from the influence of former colonial powers. In this way, critics tapped into long-running historical themes, as well as the visceral fear of the Ebola epidemic.

By analysing the debate from its inception, we can track the evolution of issues from their initial appearance in alarming news articles, to examples of extreme claims becoming more sober, as seen in the transition from claims of a “secret trial” towards more moderate complaints that too little sensitisation and stakeholder consultation had been done. As early opponents like the CGIN dropped out of the news, attention focused on high-profile figures, and in this process, these public figures may have felt that they needed to justify their initial position to avoid embarrassment. This could explain the overall increase in the sophistication of arguments against the trials, while the initial emotions remained at the heart of the debate. As shown in this analysis, even marginal critics can mobilize a widespread and sophisticated opposition movement to derail important public health programmes, if they are taken up by high-profile opinion leaders. The situation in Ghana demonstrates the importance of early and constant engagement with both supporters and critics of trials, bringing them into conversations and ensuring that their fears, concerns and perspectives are heard and considered from the very beginning of the process. We recommend early engagement as a way to minimise the risk of individuals and communities feeling excluded, or their concerns dismissed, provoking perceptions of secrecy and potentially leading high-profile opinion leaders to take up their arguments on a wider scale.

It is noteworthy that although some of the initial critiques remained major themes throughout the debate, others, such as the conspiratorial allegation that “Ebola is a business,” did not last as long. Not unlike an infectious disease outbreak, beginning with an isolated index case and followed by a larger flare-up, some strains of these rumours proved more contagious than others. New arguments and angles also emerged as more people joined the debate. For example, the controversy clearly resonated with racial and colonial themes, as evidenced by the repeated references to historical instances of unethical experimentation on African bodies. This should serve as a stark reminder that past injustices can still undermine trust in the present. In addition, researchers should work to build trusting relationships within the communities where they are working by being consistently open and transparent as early as possible. This is particularly vital when considering sensitive issues or protocol requirements such as taking blood or the use of incentives that may be particularly sensitive given histories of extraction in the region.

Although regulations sometimes prevent trials from sharing public information about a specific trial prior to receiving ethical approval, this doesn’t necessarily preclude the possibility of a more general, government-led national discussion on the need for an Ebola vaccine, the strong value that such a vaccine would have for Ghana’s national security, and the long-term economic and diplomatic benefits of taking part in the international effort. Such a generalised national dialogue - avoiding the specifics of any one candidate vaccine trial - should be considered, especially when the disease in question is so emotionally fraught, and so relevant to the whole nation. We also recommend the involvement of community representatives in the development of clinical trial protocols, in order to ensure that trial designs are both scientifically valid and also acceptable to the communities within which they are occurring.

One of the other major challenges is making the collected information comprehensible and useable to decision makers. To achieve this, effective visualisation techniques are vital. Some of these techniques are demonstrated in this article, and many of these can be generated on-demand in real time. For stakeholders involved in a major health crisis, real-time insights can be vital, given the limited amount of time available for decision-making. It is our hope that these techniques can be incorporated into a reliable tool for assessing public information needs in future epidemics and similarly emotive national crises.