Main text: This Analysis is based on work published in: https://royalsocietypublishing.org/doi/10.1098/rsos.201721
Data analysis was conducted by one member of the team (AL) and was descriptive in nature [1]. Analysis was done by populating a table per round of interviews with data from the partial transcriptions, whereby each question, or appropriate groups of questions, represented a row of the table, with each column representing a participant. Once the table had been populated, the researcher summarized each row of answers. In conjunction with this approach, there was some use of quantitative content analysis methods [2] so the data could more easily be used for making decisions regarding risk communication. If multiple participants made similar comments, these comments formed inductive ‘codes’, which were then weighted through the use of descriptive statistics. While undertaking the descriptive analysis, themes were informally identified within the data. As this was a purely qualitative analysis, there was no need to assign descriptive statistics to each theme. A formal thematic analysis [3] was not carried out as time was limited and this was not considered a primary output of the research. Once all the analyses for each round had taken place, it was discussed with two other team members (LF and GL). These two other team members had either partially transcribed the interviews or had carried out their own rapid descriptive analysis using the same populated table. Within these discussions, the researchers identified and resolved any discrepancies between their impressions of the data, as well as adding any missing points to the existing analyses. The alterations and additions made were minimal.
In a descriptive analysis (purely qualitative, with no descriptive statistics), we extracted emerging themes from the interviews.
Emerging theme: infantalization vs empowerment
One of the themes identified within the interviews with primary care physicians was that of infantilization and empowerment. Some physicians spoke strongly about how they felt the public in the UK were being ‘infantilized’ by clinicians who implied that ‘every risk can be managed away’ and by the government for not being honest about the risks that some socializing can bring. They also recounted how patients ask them what they should do, wanting an authority instead of ‘the risk of thinking for themselves’. However, they also felt it was important that patients made their own decision about certain risks, as only the patient’s values can inform that. Additionally, one physician felt such risk information could empower patients to ‘have those conversations with the people who are making the decisions’, for example, employers.
Emerging theme: conflict between politics and medical science
Another theme recognized within the primary care physician interviews was related to conflicting medical advice whereby the participants spoke of experiencing contention between the advice given by the government, which they viewed as political, and the advice they wanted to give, which they viewed as clinical.
Emerging theme: trust
In interviews, we specifically asked the participants if they would trust an outcome from an online tool that showed them a personalised risk of dying from COVID-19, and what would make them trust it more. Sub-themes (which were unprompted) related to trust were identified:
Trust related to the data
Participants questioned whether the data were collected in a rigorous way, how accurate the data were and whether it was being ‘tampered with’ once collected.
Trust related to the source 11
Some participants identified the University of Cambridge logo displayed on materials relating to this study; knowing that the source of the tool was a university institution that undertook research gave them trust in the outcome.
Trust related to the methods behind the results
Suggestions that the risk information was produced through careful research (indicating the high quality of underlying evidence) affected the participants’ perceptions. One said: ‘It’s important that people know it’s not been plucked out the air, like they think it might be, but that it’s actually based on data’.
Trust related to the medium
Referring to increasing incidences of online scams and other malicious online activity, one participant commented: ‘People are careful these days of anything online’, going on to suggest that the inclusion of institutional logos was particularly important in gaining people’s trust when communicating online.
Emerging theme: uncertainty
Uncertainty was identified as a theme within interviews, with many participants commenting that the risk score itself was inherently uncertain. This did not seem to affect their feelings of how useful a tool would be, nor their trust in the guidance it might provide. Some participants indicated that the range around the risk score presented in the mock-ups was superfluous, possibly because they instinctively acknowledged the uncertainty of the result. Some also felt that the data which could be used to calculate the risk score was uncertain. This uncertainty of the data, and whether it was accurate and reliable, did seem to affect trust. These findings were also broadly similar within the primary care physicians, though unlike the public, uncertainty about the risk score itself was attributed to applying population-level data to individuals.
Emerging theme: worry
Throughout the general public interviews, numerous participants spoke about how the tool may worry them or others. Many primary care physicians also commented on how the tool could have a negative effect on the mental well-being of patients. By contrast, one primary care physician detailed how a tool like this would empower them to talk openly with their patients who were unnecessarily concerned, which for some had resulted in deteriorating mental health.
1. Sandelowski M. 2000 Whatever happened to qualitative description? Res. Nurs. Health 23, 334–340. (doi:10.1002/1098-240X(200008)23: 4<334::AID-NUR9>3.0.CO;2-G)
2. Neuendorf KA, Kumar A. 2015 Content analysis. Int. Encycl. Polit. Commun. 1, 1–10.
3. Braun V, Clarke V. 2006 Using thematic analysis in psychology. Qual. Res. Psychol. 3, 77–101. (doi:10.1191/1478088706qp063oa)