Participants were recruited using a purposive sampling method. Two easily reachable geographical areas (originally selected for in-person groups) were chosen providing a mix of inner-city urban (Islington, London; one group), and semi-rural (Hertfordshire; two groups), as well as diversity in ethnicity, educational attainment and employment status. Leaders of preschool parenting networks (employees of Islington Borough Council and Hertfordshire Community NHS Trust) facilitated recruitment through official social networks, for example Facebook groups or weekly meetings, as well as via parent champions who recruited both directly and through snowball sampling until sufficient numbers were reached. It is difficult to identify the exact number of participants who were recruited through snowball sampling because some participants received both information via social media networks and through fellow parent personal contact. Although snowball sampling helps to recruit people with relevant interests to the research project, it may limit diversity in demographics and focus group responses.

Participants were provided with participant information sheets and electronic consent forms, along with an email introducing the study. Author BM answered any questions to ensure fully informed consent was given, and participants signed and filed the consent forms electronically, and received a £20 voucher for their contribution.

Parents’ basic demographic data were gathered, including employment status, ethnicity and parity. Participants were asked to respond in their capacity as parents.

The topic guide listing a priori themes was jointly developed by the researchers, including two with experience in clinical and public health issues related to antibiotic use/AMR (authors LS and AH). The topic guide was structured to provide a snapshot of current understanding around antibiotic use, and both the threat felt (if any), and the sense of responsibility participants held towards their children, and others, in mitigating the emergence and spread of AMR. Topics addressed infection self-care, antibiotic-seeking KABs, the nature and consequences of AMR including sense of individual agency in mitigating AMR for societal good, and COVID-19 influence on views towards infection prevention and management. After piloting by two parents of young children, slight modifications were made to enhance the lay-friendly appeal of content.

Author BM was the focus group facilitator, and only member of the research team attending the sessions. Secure and encrypted MS Teams video conferencing technology was used to collect and record the focus group data. All participants contributions were transcribed and anonymised.

(See Table 1 for a topic guide summary; see extended data for full topic guide).

Transcripts of the recorded discussions were entered into Computer-assisted qualitative data analysis software (CAQDAS, NVivo Version 12 (RRID:SCR_014802)) to organise the data. Alternative open access software to NVivo exist for example, Taguette or RDQA, but these were not used in this study.

Qualitative analysis followed the established Framework Method. This entailed a stepwise process of data familiarisation; mostly inductive analysis consisting of line-by-line coding/sub-coding (using NVivo) and grouping into categories broadly aligned (but adapted to reflect conceptual relationships between comments) with the a priori topics in the topic guide, and this effectively created an analytical framework. This was applied across all transcripts to compare cases (participant comments) both within focus groups and across focus groups until no new codes were generated. This represented the data saturation i.e. no new concepts that made significant contributions were found in the data and indicated a good point at which to terminate coding. By charting the data into a ‘framework matrix’ comprising codes in columns, and cases in rows (see extended data for a table of codes and descriptions, and for an example of the framework matrix), thematic analysis was conducted. This involved the generation of sub-themes initially, and then key themes (synthesis across sub-themes). Ultimately, insights (possible unarticulated explanations) were derived from reviewing the matrix and drawing connections within and between participants and categories to facilitate higher order interpretation, according to a process of thematic analysis. ^{23 – 26}

Coding and themes were checked for consistency and reliability with two co-authors (LS and AH) and to balance any reflexivity of BM, a mother of young children, in analysis of the data. Such similarity may be considered a bias, but also a benefit that may enhance rapport and the richness of data obtained. ²⁷

Consolidated criteria for reporting qualitative research (COREQ) were followed as closely as possible in the reporting of this research. ²⁸ In accordance with these guidelines, lead author BM attended training in both facilitating a focus group and conducting qualitative data analysis. BM had no relationship with participants prior to the study and the participants understood the research formed part of BM’s doctoral research.

Four members of the public (sourced via the Patient and Public Involvement (PPI) facility at University College London Hospitals (UCLH)) were consulted in designing the topic guide and provided input on content and style of questions, as well as proofing and testing of the questionnaire used in the actual focus groups.

The process of analysis using the Framework Method generated 13 codes, divided into four categories that emerged from the data (see extended data for a table of descriptions according to categories and codes; and a table of categories aligned by their codes and sub-codes).

Analysing each transcript according to these codes and selecting verbatim illustrative of them generated insights including drivers of KABs around AMR. Synthesis across the three focus groups generated four key themes, which are presented alongside their sub-themes, codes, and categories/ a priori topics in Table 3.

Key themes 3 and 4 are most novel, while some aspects of key themes 1 and 2 are relatively new but largely reinforce findings from other studies. Table 3 shows the relationship of the original a priori topics (from the topic guide) to sub-themes and key themes (sub-themes and key themes were derived from inductive analysis).

Key theme 1: Uncertainty around the management of symptoms and severity of childhood infection, including when to consult a GP (possibly for antibiotics), including the influence of the pandemic on views around prevention and management of infection.

Most parents remarked on their uncertainty about when the severity of their child’s symptoms warranted a doctor consultation, possibly for antibiotics. Many said they home-managed symptoms for around three days before seeking medical help. Mothers with older children (compared with first-time mothers) suggested greater confidence in managing their child’s illness at home for longer.

One mother recalled giving her 10-month old ‘Calpol’ if he had a fever, but if she felt his ‘heart rate was up and he wasn’t feeding properly’ then she would contact a doctor . (Focus Group (FG)2, participant (p)3)

Such comments highlight the need for improved public/patient engagement around which symptoms and severity justify contacting a GP (potentially seeking antibiotics).

Sometimes non-medical pressures are implicated—economic (employment-related) factors were alluded to.

The COVID-19 pandemic profoundly influenced people’s attitudes and behaviours towards infection prevention and control, with constant public health messages around handwashing and other SARS-CoV-2 mitigation measures. Community antibiotic prescriptions fell during this period. ²⁹

Vaccination, generally, was also raised as a measure to prevent infection and was particularly topical at the time, with potential COVID-19 vaccinations being debated publicly.

Table 4 lists examples of verbatim relevant to this key theme.

Key theme 2: Background including cultural, social, generational, familial and habitual factors influence KABs towards antibiotic use/AMR

Subtle and often unarticulated factors, including generational, familial, cultural, habitual, geographical and social, often influence views towards healthcare, including antibiotic use/AMR, as well as parent/patient expectations from the doctor–patient consultation.

One mother who grew up in Somalia said they expected to receive medication upon visiting the GP, effectively as a validation of their illness, ‘ be it antibiotic or medicine or anything’.

Another mother said that antibiotics were habitually prescribed in her parent’s home country, France.

Older generations lived in a time when antibiotics were considered a cure-all.

A mother from Spain said antibiotics were much more available there than in the UK, and, in her opinion, there was a lack of universal consensus around antibiotic stewardship.

Understanding more about the socio-cultural drivers and experiences underpinning an individual’s KABs towards antibiotic use may illuminate novel, more nuanced, avenues for formulating and communicating AMR messages.

Certain areas of the UK with greater socio-cultural diversity may benefit most from incorporating a more nuanced and tailored approach to the doctor–patient dialogue around antibiotics/AMR.

Key theme 3: Greater understanding of how antibiotics work, and how AMR develops, may impact perception of threat to the individual and society, and consequently individual sense of agency/responsibility towards mitigating AMR

Overall, lay understanding of how antibiotics and AMR work was mixed but largely limited. Most parents knew that antibiotics fight bacteria, not viruses, with one mother understanding that antibiotics can be either narrow or broad spectrum. That antibiotics can target good bacteria too was mentioned. ‘… you don’t know what’s getting killed basically …’ (FG2, p4)

Four participants believed that antibiotic resistance meant the body became tolerant to antibiotics.

The misunderstanding that resistance implies the body becomes tolerant to antibiotics, rather than resistance being a feature of the infecting bacteria, may be instrumental in the sense of threat individuals perceive from AMR, and may help to explain why many people see AMR as a threat not associated with them personally, but only with others who use antibiotics frequently. Four parents commented that the threat of AMR was not very present in their lives today.

AMR is often framed as one of society’s most serious global humanitarian crises, and, as such, participants were asked whether they would personally forego an antibiotic now to prevent future widescale AMR adversity for society. Despite some parents recognising this, all parents said that the immediate threat of infection to their individual child’s health took priority and that they would seek antibiotics if it promised to help their child.

Parents reflected on securing a future free of AMR and the benefit it offered to their child in preference to society. Together, these comments suggest that focusing on benefit to the individual rather than society would potentially resonate more with parents.

Two parents highlighted that the lack of an effective treatment for COVID-19 (summer 2020) echoed the nature of AMR, whereby antibiotics are ineffective against infection.

The mitigation measures taken by individuals to reduce societal spread of COVID-19, e.g. vaccination, isolation, masks and social-distancing, reflect the narrative pertaining to AMR where individual action now (foregoing an antibiotic) helps to avoid future adverse effects for society (emergence of AMR later).

Most parents said they would have a COVID-19 vaccination, when they became available. Of note, the groups were held mid-pandemic and the threat of COVID-19 felt real, such that the risk of future adversity (COVID-19) was considered serious enough to warrant the small inconvenience and/or discomfort of vaccination. By extension, this might highlight the value of imparting knowledge about the real and immediate threat of AMR and how to avert it when people are primed to receive such information, for example, at the point of care, when parents consult a GP with a sick child.

Table 5 lists examples of verbatim relevant to this key theme.

Key theme 4. Strength of the doctor–patient dialogue serves as an opportunity to effect change in KABs relating to antibiotics/AMR at the point-of-care.

Participant comments in this study indicate that the primary care consultation setting is conducive to a constructive interaction around the risks and benefits of antibiotics and AMR. ^{12 – 14} As such, these data both add value to, and reflect, prior reports of antibiotic stewardship leading to a reduction in antibiotic prescribing in primary care over recent years (excepting anomalies related to pandemic prescribing). ^{6 , 30} Such constructive dialogues nurture improved GP patient trust and enhanced compliance, potentially reaping short and longer-term benefit.

That clinicians have a valuable role to play in alleviating misunderstanding around antibiotics, and AMR was widely accepted by parents, most of whom said that they followed the doctor’s advice on antibiotic use.

GP-led antibiotic stewardship efforts and education of specific knowledge at the point-of-care was important to participants.

However, one parent noted that the differing opinions of doctors within practices can be confusing, emphasising the importance of consistency across healthcare points of contact with patients.

Another mother’s comment illustrated how a parent’s instinct about their child sits alongside, and possibly carries equal weight to, a GP’s advice, reinforcing the need to move beyond a more conventional telling or provision of advice by doctor to patient, to the holding of a constructive two-way exchange. A more nuanced dialogue around antibiotics use/AMR may better address any barriers to adopting a mindset where antibiotic use reflects concerns around AMR.

Together with key theme 2 (socio-cultural influences on KABs), the doctor –patient consultation setting may be optimised with a more personalised approach to facilitating and addressing any barriers to appropriate antibiotic use/AMR.

Addressing the topic of parental KABs around antibiotic use and AMR through the concept of individual agency to avert AMR is novel but notably builds on a growing body of relevant, associated literature.

Similar to our findings, another recent study observed that, ‘parents found it difficult to interpret symptoms and signs’, and, ‘[…]need better information and support to manage their child’s illness at home’, adding further emphasis to the continuing need for intervention in this respect. ³⁴ Likewise, the perceived vulnerability of children, has been documented elsewhere. ⁸

Previous studies with families attest to long-held misunderstandings about AMR, for example, parents perceiving that they are at low risk because they infrequently use antibiotics, and that the ‘body becomes immune to them [antibiotics]’. ^{2 , 17} Our study reinforces that such misunderstandings persist and suggests that more novel and potentially more effective means of engagement are needed.

A 2016 study noted the complexity of the doctor –patient interaction and ‘interplay of care seeking’. ³⁵ Our study reinforces and importantly builds on this by noting the nuances of parental background influences on KABs, suggesting that recognising, for example, ethnic, generational and country-of-origin differences may enhance the effectiveness of the doctor –patient dialogue.

Prior studies have referred to the benefits of the point-of-care setting to optimise AMR engagement, but more research is needed on how this may manifest in practice. ^{12 – 14 , 17 , 36 , 37} Pointedly, Cabral’s 2016 study concluded the need for interventions that reduce antibiotic prescribing ‘to address within-consultation communication, prescribing behaviour, and lay beliefs simultaneously’. ³⁶ Our study advances this with specific aspects of ‘how’ these within-consultation communication and lay beliefs may be approached and leveraged to further shape mindset towards antibiotic use/AMR, for example, focus on the risks and benefits to the individual of judicious antibiotic use. It also suggests that making this within-consultation communication more real and relevant to the individual’s immediate situation may be more impactful.

Participant recruitment comprised a broad cross-section of parents from South-East England, which provided a rich demographic diversity due to this area having a high population density, of widely variable backgrounds, across both urban and semi-rural settings. However, there may have been some selection bias, with most parents having very good levels of education and employment, which is largely unrepresentative of the general UK population.

Pandemic restrictions means the virtual (at-home) setting may be viewed as both a strength and a limitation. The format facilitated attendance by parents whose childcare commitments might have prevented attendance at an in-person group. However, this also led to some unavoidable parenting distractions.

Participant numbers were small, and it is not possible to draw generalised conclusions based on the comments of only 14 participants, but insights would draw strength in combination with findings from other studies, with our findings reinforcing and building upon prior findings.

The virtual format also meant the group was shorter in duration, and that the interactions between group members were less natural than an in-person group. In addition, resource constraints and the virtual format meant non-verbal group dynamics could not be recorded.