Understanding attitudes and beliefs regarding COVID-19 vaccines among transitional-aged youth with mental health concerns: a youth-led qualitative study

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • To our knowledge, this article is one of the first to characterise the determinants of vaccine hesitancy in transitional-aged youth living with mental health concerns using a qualitative, youth-engaged approach.

  • This study elicited youth feedback at various steps of data collection, analysis and manuscript writing to ensure final themes emerging from thematic analysis resonated with the study’s target population.

  • This study’s sample included many participants with high levels of postsecondary education and does not capture the perspectives of people who do not speak English—a population that may face unique barriers to COVID-19 vaccine information seeking and vaccine acquisition. Moreover, the use of self-reporting to characterise participant mental health concerns and the qualitative nature of the study limited the exploration of trends in subgroups of individuals with different mental health concerns and severity of illness.

Background

The global spread of SARS-CoV-2 (COVID-19) has had a lasting and detrimental impact on human health and healthcare systems.1 Vaccination is safe, effective and widely considered a key public health strategy for helping to end the pandemic.2–5 Considering data on the gaps in vaccination and the need for ongoing administration of vaccine boosters to protect against future virus variants as well as the important role of vaccination in future pandemics and against other preventable disease (eg, measles and HPV), lower vaccination rates in specific populations are a pressing public health concern.2–5

Among those disproportionately affected by COVID-19 are people with mental health concerns, including youth and young adults. Infection risk is increased by factors such as ability to adhere to public health measures and congregate living settings (eg, psychiatric hospitals or homeless shelters).6 7 One study found that having a psychiatric diagnosis was associated with a 65% increased risk of contracting COVID-19, compared with having no psychiatric diagnosis.8 This population experiences stigma and discrimination and those who contract the virus face barriers to timely care, which increases COVID-19-related morbidity and mortality.7 9 10 Given the broad mental health implications of the pandemic, people with mental health concerns have experienced both worsening and new psychiatric concerns.7 11 Recognising this increased risk profile and the integral role of vaccination in mitigating adverse outcomes, researchers, clinicians and policy-makers have called for targeted vaccine programmes for this population.12 13

Further, transitional-aged youth (16–29 years old; referred to in this paper as ‘youth’) tend to have higher rates of COVID-19 infection and sequelae related to both direct and indirect effects of the SARS-CoV-2 virus.14 This age is a critical period of psychosocial development, identity formation and life transitions, which may also increase the risk for mental illness.15 Thus, youth with mental health concerns—who are situated at the intersection of these risk profiles—may represent a particularly vulnerable population in the context of COVID-19.

Vaccine hesitancy refers to a delay in accepting vaccination, or to a decision to refuse vaccination, despite the availability and safety of vaccines. It is particularly prevalent among young adults as compared with older adults. People between ages 18 and 34 are four to seven times more likely to be vaccine hesitant than those aged 65 and older.16–18 Significantly higher rates of vaccine hesitancy have also been reported in people with mental health concerns.19 These statistics suggest that youth with mental health concerns may have even higher rates of vaccine hesitancy and that specialised interventions and public health messaging may be required to promote vaccination.17

Comprehensive models of vaccine hesitancy consider the role of trust in science, vaccines, healthcare professionals and healthcare systems.20 21 Research has shown that vaccine hesitancy varies according to demographic factors such as race and health status. For some individuals and communities, initial vaccine mistrust may derive from such factors as the historical and ongoing harms inflicted by governments or healthcare systems.21 This interaction between medical mistrust and vaccine hesitancy can often be seen within racialised and marginalised communities and is attributed to their continuing mistreatment within healthcare.22 23 When considering strategies for improving vaccine uptake, these associations are important. For example, rates of vaccine refusal in Black populations have consistently declined throughout the pandemic, whereas resistance among white populations has stayed fairly consistent.24 Such improvements are hypothesised to originate from efforts to promote COVID-19 vaccination in a culturally sensitive way, such as through the recruitment of Black leaders and organisations.24 Another group that may experience vaccine hesitancy is people with mental health concerns, who face stigma and social marginalisation, are questioned about their capacity to make informed decisions, and generally receive poorer medical care.25 These intersecting axes of harm not only create mistrust of healthcare systems, but they also disincentivize people from voicing their health concerns and establishing therapeutic relationships with healthcare providers.26

There is a limited literature on COVID-19 vaccine hesitancy among youth with mental health concerns, particularly regarding attitudes and beliefs.12 This qualitative study addresses that gap by seeking the perspectives of these youth and their family members. It aims to answer this research question: What are the beliefs and attitudes of youth with mental illness about COVID-19 vaccines, and how do these perspectives affect vaccine acceptance?

Methods

Theoretical models and frameworks

Two theoretical models of behaviour serve as frameworks for our study. First, the health belief model consists of six constructs that can help predict health behaviours: (1) perceived susceptibility, (2) perceived severity, (3) perceived benefits, (4) perceived barriers, (5) self-efficacy to engage in a behaviour and (6) cues to action.27 In the context of vaccine uptake, perceived benefits may include a person’s beliefs about the protection offered by vaccination to themselves and their community, while perceived barriers may include the belief that certain psychosocial, physical or financial factors place vaccination out of reach.27 Secondly, the vaccine hesitancy framework incorporates a person’s context, particularly the intersection of their level of commitment to their health and their level of trust in health authorities and mainstream medicine.28

Study sample

We recruited two participant cohorts: (1) 46 youth aged 16–29 years with 1 or more mental health concerns and (2) 6 family members, given their potential influence on youth beliefs. Our study definition of mental health concerns was broad, including mood disorders, anxiety disorders, neurodevelopmental disorders, personality disorders and substance use disorders. We recruited participants through clinical and support networks across Canada (see online supplemental appendix A for full list of recruitment pathways). We chose this strategy to ensure a broad representation of youth with a range of mental health concerns and to encompass diverse perspectives (ie, a range of racialised, ethnic and gender identities). Study candidates self-reported their mental health concerns and had to be able to conduct the interview in English. Participants’ levels of vaccine hesitancy were not included in our approach to recruitment; this information was elicited post-recruitment during interviews.

Supplemental material

Data collection

Between September 2021 and August 2022, we conducted semistructured interviews to explore participants’ beliefs, attitudes and information needs related to the COVID-19 vaccines. This time span encompassed several key events in Ontario, Canada, including the rollout of vaccine certificates in September 2021, the Omicron variant outbreak in December 2021 and subsequent public health protections, and the rollout of COVID-19 vaccine booster doses.29 30

Participants provided informed verbal consent via WebEx video conferencing and/or telephone and written consent via digital signatures of an online form. Interviews were subsequently facilitated and recorded by at least one research team member via the same platforms over approximately 45 min. Participants received a $C30 honorarium via e-gift card or mailed cheque.

Interview questions explored participants’ demographics, including gender and ethnicity as defined by participant self-report; mental health; vaccination status; information needs, beliefs, and attitudes regarding COVID-19 vaccines; levels of vaccine hesitancy or acceptance; and determinants of vaccine hesitancy. We also included a subset of questions informed by the health belief model, including questions on perceived susceptibility to contracting COVID-19, perceived potential severity of COVID-19 illness, benefits of vaccination, barriers to and consequences of vaccine uptake and motivators for vaccination.27 The complete interview guides can be found in online supplemental appendices B and C.

Analysis

Interviews were transcribed and coded by three research team members using Dedoose (Dedoose V.9.0.17, a web application for managing, analysing and presenting qualitative and mixed method research data (2021). Los Angeles, California: SocioCultural Research Consultants. www.dedoose.com). An initial coding dictionary was generated using five randomly selected interviews which were coded by three team members. Subsequent interviews were coded by individual team members with review, discussion and secondary coding by other team members. The coding dictionary was iteratively adapted as needed to reflect emerging codes. We then applied Braun and Clarke’s six-stage approach to thematic analysis to generate major themes and patterns (see online supplemental appendix D).31

Patient and public involvement

A youth-led approach was used throughout the study to ensure findings resonated with our study’s target population. To achieve this, the study team included two youth engagement specialists with lived experience of mental health concerns from CAMH’s Margaret and Wallace McCain Centre for Child, Youth, and Family Mental Health’s Youth Engagement Initiative. The specialists took part in team meetings and provided input on project planning, coding, themes, manuscript writing as well as planning and facilitation of two codesign workshops as part of the knowledge translation activities for this project. We also assembled a youth advisory committee of eight youth advisors with lived experience to help guide the development of the codesign workshops and the cocreated youth-specific vaccine resources generated from these workshops.

Results

Participant characteristics

We conducted and analysed 52 interviews (46 youth and 6 family members) (table 1). Almost half (48%) of the participants were age 20–24 years, 20% were age 16–19 years and 33% were age 25–29 years. The majority were fully vaccinated for COVID-19 (two or three doses), four youth reported one dose and six youth were unvaccinated. The most reported mental health concerns were anxiety (n=18) and depression (n=16) (table 1). No participants reported substance use disorders or concerns.

Table 1

Participant demographics

Themes

Thematic analysis generated two major themes related to factors that contribute to vaccine hesitancy or acceptance. These themes came from analysis of all transcripts and were triangulated by the family member interviews.

Theme 1: Factors affecting trust in COVID-19 vaccines

Trust in and attitudes toward the COVID-19 vaccine related to four areas: trust in the vaccines, trust in healthcare providers, trust in government and the impacts of historical and ongoing institutional racism.

Trust in the vaccines

Several participants discussed their trust in the COVID-19 vaccines as a medical intervention. For instance, participants discussed their trust related to safety, efficacy and quality of the vaccines, as well as to the development and testing process. While some participants indicated strong trust in the vaccines, others expressed mistrust. Lived experiences with the vaccines among those close to the participants influenced trust. Participants who expressed mistrust provided various reasons, including perceptions about how rapidly the vaccines were developed. Table 2 presents quotes reflecting both views.

Table 2

Participants’ reasons for trust or mistrust in COVID-19 vaccines

Another factor contributing to vaccine mistrust was how participants perceived the initial rollout was managed, namely that of the Oxford–AstraZeneca vaccine. Multiple policy changes and evolving recommendations, along with emerging safety concerns, reduced participants’ trust in that vaccine. Some participants explained that issues with the Oxford–AstraZeneca vaccine lowered their trust in all subsequent COVID-19 vaccines, regardless of manufacturer.

Why bother giving people the AstraZeneca vaccine? Because for the Moderna and Pfizer vaccines, where there is genuinely nothing wrong with them … like a lot of the side effects are minimal, extraordinarily minimal… [AstraZeneca] has now tainted those vaccines that can actually save people’s lives, and now no one wants to take the vaccines because of AstraZeneca. (Participant 005; 21y; 3 doses)

Participants’ overall trust in vaccines also affected their trust in COVID-19 vaccines. While we mostly saw concordance between general trust in vaccines and trust in the COVID-19 vaccines, there were instances of discordance—high trust in vaccines overall, but mistrust of the COVID-19 vaccines. Concerns about how quickly the COVID-19 vaccines were developed and approved compared with existing vaccines contributed to this lack of trust.

I know that vaccines do work. [But] I waited it out because I didn’t trust… What if it had side effects that no one was aware of? So then I got it a little bit later… [because] I know they don’t have any research on that because it just came out. (Participant 006; 19y; 3 doses)

Trust in healthcare providers

Most participants trusted healthcare providers as a group. Their main reasons related to education and expertise regarding vaccine safety and effectiveness.

I heard a lot of things [about the vaccine] from people, from friends, and colleagues. But I don’t really trust the information from those sources. I just prefer to get it directly from a doctor because they are obviously medical personnel, an expert in the field. (Participant 002; 25y; 0 doses)

Participants also discussed how their relationship with their healthcare provider influenced trust. For one participant, tying common lived experiences into vaccine discussions built trust:

Representation matters. My doctor, who is Muslim, who is female, represents people like me… So not just someone else who is so different from me, from a different culture and different religion. Not to say I wouldn’t trust them, [but] she was more relatable, and she related it back to my religion, back to me as a female. She said nothing is going to happen to you. Don’t worry about infertility… Because she was a female and because she is Muslim, I think she was able to relate better to me. (Participant 004; 28y; 2 doses)

Trust in government

For several participants, trust in Canadian government vaccine information and policies reflected their general appraisal of how the government handled the pandemic. Participants who were satisfied with the government’s response had more trust in its vaccine information and policies. Conversely, participants with unfavourable views had less trust.

This entire pandemic is a result of a flaw in the government. I strongly believe that viruses are human nature, but pandemics are not… When they tell us or mandate us to get vaccines, or to put things into our body, of course I see the scientific value in it. I support the doctors who are putting hours into research… but when the government tells us to do something…to hear it from the people who are the reason we are in this situation is super frustrating. If anything, that makes me want to not get it. (Participant 007; 21y; 3 doses)

Other participants gave more nuanced evaluations, recognising that the government comprises many individual political agents and civil servants.

I don’t really feel like there is any reason not to trust the government. Obviously some politicians are kind of crooked, but I don’t think it means the government is a terrible thing that’s going to [use the vaccines to] poison us and track us and stuff. (Participant 008; 17y; 0 doses)

Impact of historical and ongoing mistreatment of racialised populations

Another factor that influenced some vaccine decisions was the historical and ongoing mistreatment of racialised populations by government and healthcare systems. These experiences contributed to mistrust of the vaccines, healthcare providers and government.

[Black communities] have a history of vaccines where they were forced to take it… they were the subject of research… There is that lack of trust that is historical, embedded, and intergenerational. Same with Indigenous communities. So when I think about being told to do something, it’s like, well when we did it the last time our community suffered… That history [of mistreatment]… isn’t being accounted for when these messages are being put forward. It’s like everybody should take it, accept it, and just do it. (Participant 007; 21y; 3 doses)

These negative experiences influenced decision-making, even as other factors, such as trusting science and wanting to stay safe, were present:

I have had really bad experiences with healthcare… but I’m not going to allow that to stop me from believing in facts… to keep myself healthy… Yes, there are really crappy doctors, and racial bias does exist, and of course there is a terrible, terrible history of Black women in particular being treated like crap in the field of medicine. At the same time, I will do what is healthy for me, and that means I need to pay attention to the science and the facts. (Participant 005; 21y; 3 doses)

Theme 2: Decision-making: mental health influences and safety considerations

Direct influences of mental health

Although some participants reported their mental health did not affect their vaccine decision-making, there were many participants who reported that acute mental health issues influenced their vaccine decisions. One participant recognised that getting vaccinated was “an impulse decision” to resolve boredom and a desire to “[just] get a shot” while staying in a psychiatric hospital (Participant 009; 22y; 2 doses). Several participants identified trypanophobia (ie, needle phobia) as a barrier to vaccination, despite believing in the vaccines’ effectiveness and safety. They emphasised the need for tailored interventions for needle phobia.

Some participants described how psychosis influenced their decision-making. For some, the experience motivated them to get the vaccine:

While I was having symptoms of psychosis, I remember being really concerned that I did have COVID, and that I would die in my sleep from it… I would think about how [the vaccine] might reduce the chance of me getting it, which would reduce the chance of my family getting it. (Participant 010; 21y; 2 doses)

For others, acute psychosis contributed to negative views about vaccination. One participant explained that these views changed after the psychosis resolved:

When I was in that state of mind, I was like, this is going to be bad for my body. They are putting toxins in me… [After] I was being put on medication and I came back to being aware of what reality is. I was like, oh this is stupid. I should just get vaccinated. It’s not really an infringement on people’s rights or anything. It’s about public safety. (Participant 011; 21y; 2 doses)

Indirect influences of mental health

Beyond acute symptoms, ongoing experiences of mental health challenges often influenced vaccine decisions and tended to promote vaccine uptake. Participants recognised their increased vulnerability to potential sequelae from COVID-19 infection itself, including heightened anxiety related to potential health consequences and the toll of social isolation that comes with self-isolating. This motivated them to accept the vaccine:

I already have a lot on my plate with what I experience, hearing voices. So I think the inclination to get vaccinated was to relieve a bit more. Psychosis is very difficult but manageable, but having COVID as well, and actually being in danger, would be just catastrophic. (Participant 009; 22y; 2 doses)

Another participant connected mental health stigma with the stigma of being unvaccinated, which was a motivator to get the vaccine:

Feeling othered is something that I struggle with, dealing with a mental illness and stuff. So I was like, this is just another kind of stigma – vaccinated versus being unvaccinated. You could tell in that restaurant that I was the only one that was not vaccinated… I was the only one that had to wear a mask. So I was like, you know what, I’m sick of this treatment. I’ll just go get it done and get it over with. (Participant 012; female; 28y; 1 dose)

Discussion

Our analysis generated two main themes related to factors that affect vaccine hesitancy: (1) trust in COVID-19 vaccines and (2) mental health influences in vaccine decision-making. Our findings expand on the existing literature pointing to trust as a central factor influencing vaccination decisions.18 20 21 32 Specifically, our findings point to the relevance of epistemic trust, which is the trust people place in the competence of a person, organisation or institution that promotes science-based knowledge or products.18 For example, our participants reported reduced trust in the safety of Oxford–AstraZeneca vaccine, a finding reflected in the existing literature, which found lower perceptions of safety for this product compared with the Pfizer–BioNTech vaccine.33 Epistemic trust correlates with vaccine confidence, which can be conceptualised through the vaccine confidence model.32 The model consists of three domains: trust in the product, trust in healthcare providers and institutions, and trust in policy-makers.32 Our findings align with this model, showing that trust in the vaccines, trust in healthcare providers, and trust in government were the main factors that influence vaccine acceptance in a group of largely vaccinated youth. Participants who were vaccine hesitant often cited decreased trust in institutions (eg, healthcare organisations, government) and relied increasingly on information from their social networks, a phenomenon which has been characterised in the literature.34 This adds further evidence to support the notion of vaccine hesitancy being a social phenomenon in addition to, and perhaps to a greater degree than a cognitive one.34 Vaccine hesitancy raises complex questions about the role of trust, which means that exploring these factors during clinical conversations and when developing targeted interventions to promote vaccines is an important next step in improving vaccine uptake among youth with mental health concerns.

Addressing the historical and ongoing harms perpetrated by the healthcare system and governments against racialised communities could increase vaccine confidence and uptake. Mistreatment is a main reason for vaccine mistrust in Black and Indigenous communities and was described by several participants in our study.35 Moreover, given that youth with mental health concerns also face mistreatment, stigma and barriers to care, an intersectional lens is required to explore how these health inequities contribute to vaccine hesitancy.26 36 37 For example, among people with opioid use disorder, vaccine hesitancy was found to be highest among Black participants and was likely linked to greater medical and government mistrust.37 Our study has identified a serious gap in current vaccine-related communication and uptake strategies for these populations. Participants described concerns over the lack of acknowledgement of previous harms against these communities when encouraging and enforcing vaccination, and the role of leaders from racialised communities in providing trustworthy vaccine information. Governments and healthcare organisations can demonstrate trustworthiness by acknowledging and addressing these injustices when promoting vaccination. When interpreting study findings and designing interventions, it is crucial to consider the intersecting identities of racialised youth with mental health concerns, who may face compounding forms of discrimination. Applying an intersectional lens also means considering other marginalised identities that may influence vaccine hesitancy (eg, gender identity, sexual orientation, class, disability). Future research should examine the role of these intersectional identities in vaccine attitudes among youth with mental health concerns.

Our findings also provide insight into how lived experiences of mental health concerns directly and indirectly affect decision-making. This study adds to the literature on the influence of acute mental health concerns on vaccine-related decisions. Earlier studies have described the role of anxiety in promoting risk-reducing behaviours such as vaccination.18 Increased exposure and access to medical facilities as part of psychiatric care may also increase vaccine uptake.27 In contrast, other studies have described varying mechanisms through which acute mental health symptoms result in vaccine mistrust.38 The literature cites higher rates of vaccine hesitancy among people with psychotic disorders, often attributing this finding to impaired decision-making skills, paranoid ideation and increased stigma.38 The vaccine-related fears described by our participants with experiences of psychosis resonate with those in the literature, including fears of vaccine adverse events and of disingenuous governmental intentions.38 Other studies have described how symptoms of psychosis and anxiety increase susceptibility to misinformation, resulting in greater vaccine hesitancy.38 These findings are consistent with those of our study in which generalised anxiety about one’s health appeared to motivate vaccine uptake, whereas active psychosis was associated with negative vaccine perceptions.

The literature has established that needle-related fear is most prevalent in children and adolescents and accounts for up to 10% of all cases of vaccine hesitancy.39 Participants in our study discussed strategies to address needle phobia, such as plain language vaccine information, support from trusted loved ones and specialised vaccine clinics offering low-stimulation set-ups and personalised approaches.39 More broadly, participants also discussed the benefits of healthcare providers normalising hesitancy during vaccine conversations, including normalising feelings of doubt. Finally, previous literature has associated mental health stigma with increased vaccine hesitancy and suggests that stigma-reducing interventions may promote COVID-19 vaccine uptake.40 Our study provides an additional perspective, where some participants were motivated to get the vaccine to avoid perceived anti-vaccination stigma and, in interviews subsequent to the rollout of vaccine passports, to access documentation that would allow them to engage with others in social spaces.

This study’s strengths included the use of a youth-engaged approach to data collection, analysis and manuscript writing, wherein youth engagement specialists provided feedback at each stage. The input of the youth team with respect to which themes and strategies were personally relevant helped to ensure that the most resonant themes were analysed. Importantly, the input of the youth team helped to ensure that final themes synthesised by the research team were representative of the initial interviews and overall perspectives of individuals in this population. Interviews began at a time point in Ontario vaccine rollout where vaccines were widely available and approximately 73% of the general population had received at least one dose.41 This was reflected in our sample, where most youth included in this study had received at least one dose. As such, the perspectives captured surrounding determinants of vaccine hesitancy may not generalise to youth who are extremely vaccine hesitant. Moreover, due to the qualitative nature of the study, we did not quantitatively assess vaccine hesitancy or include it in our analysis. Future studies should seek to elicit perspectives from youth with clearly defined levels of vaccine hesitancy to gauge whether certain concerns are more prevalent in groups of certain levels of hesitancy. This study was limited to English-speaking participants and does not capture the perspectives of people who do not speak English—a population that may face unique barriers to vaccination. Although efforts were made to recruit a diverse sample, demographic factors such as ethnicity and gender were not always concordant between participants and interviewers. Given the sensitive nature of interview topics, often directly involving these factors and their influence, it is possible this may have facilitated or impaired information sharing from certain participants. Many participants were noted to have high levels of postsecondary education which often facilitated their ability to digest vaccine information, which may be reflective of some of our recruitment occurring at a postsecondary institution, the University of Toronto. Moreover, although we interviewed six family members to help triangulate our findings, the small sample size means that the results do not comprehensively summarise the attitudes of family members toward the vaccines. Our study relied on self-reporting of mental health concerns and diagnoses. With the aim of minimising barriers to participation, we did not collect data detailing the severity of participant mental health concerns, which may influence their vaccine hesitancy (eg, use of medications, need for hospitalisations). Future studies should investigate whether there are distinct trends among certain subgroups, such as those living with severe mental illness or with multiple diagnoses. Similarly, while our study provides insights into vaccine hesitancy among youth with the mental health concerns identified in our sample, transferability of these findings should be thoroughly assessed for other youth populations, including for youth with other mental health concerns which may not have been represented in our study.

Conclusions

This study’s findings resonate with the literature on key factors related to vaccine trust, including those outlined in the vaccine confidence model. They provide novel insights into the model’s relevance for youth with mental health concerns, which is highly relevant to future pandemics and vaccination efforts for other communicable diseases. Further, these findings can inform the development of public health campaigns seeking to increase vaccine uptake more broadly in this vulnerable population.

Given the complex reasons and motivations that youth have for getting, or not getting, the COVID-19 vaccine, the lived experiences of youth must be embedded in the development of policies, education and resources.26 Our findings will inform the development of youth-specific vaccine resources, including a clinical conversation guide, that will be cocreated by youth and researchers through interactive, youth-led workshops as part of our knowledge translation activities.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by (1) Centre for Addiction and Mental Health (CAMH) Research Ethics Board (CAMH REB Ref #: 078/2021) and (2) University of Toronto Research Ethics Board (Protocol #31600). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

The authors would like to express gratitude to the youth and family members who participated in these interviews. The authors would like to thank CAMH’s Slaight Centre Early Intervention Service and the Margaret and Wallace McCain Centre for Child, Youth and Family Mental Health for recruitment and youth engagement support. The authors would like to thank Jejociny Consalas for background research assistance, Aloha Narajos for helping ensure that the youth perspective was embedded throughout the process and to Hema Zbogar for editing the manuscript.

This post was originally published on https://bmjopen.bmj.com