The Qualitative Methods Journal Club discussed a paper on the issue of ‘non-carriage’ of take home naloxone among people who use opioids in New York City. Stigma, indifference toward overdose, fear of negative consequences of carrying naloxone, and fear of misrecognising the need for naloxone were identified as barriers to carriage. This was considered a timely issue by the group.

About this month’s Journal Club

The article (available here) was discussed online within the National Addiction Centre, Institute of Psychiatry, Psychology & Neuroscience at King’s College London with 6 attendees from King’s College London and the University of Exeter. Article presented and discussion led by Anya Aggarwal (National Addiction Centre, KCL). The meeting was chaired and summarised by Stephen Parkin (National Addiction Centre, KCL).

Summary of the article

This article is published in an online journal called JMIR Formative Research and is concerned with the issue of non-carriage of take home naloxone (THN) amongst people who use opioids in New York City (NYC). The study is contextualised by an increase in availability of THN in NYC and by a concomitant increase in Overdose Education and Naloxone Distribution (OEND) programmes throughout the same location. As such, the article presents a qualitative study regarding the reasons why people who use drugs may choose not to carry THN in a location where availability and access to THN have increased as a direct public health response to a local/national opioid epidemic (and associated fatal overdose).

As a result of the increased availability of THN, in turn set against the backdrop of increasing overdose fatalities, Bennet and colleagues aimed to identify the reason why people who use opioids ‘may be reluctant to accept, carry, and administer naloxone’. In addition, they aimed to use formative qualitative research as the key approach for informing ‘the development of messaging content that addresses barriers to its acceptance and use’. The formative qualitative research was informed by methods that involved two researchers visiting participants across 5 boroughs of NYC and conducting interviews in semi-public / semi-private locations within public areas (such as green areas/parkland). Semi-structured interviews with 20 participants ‘explored why people decline naloxone kits when offered, why people might be reluctant to use naloxone to reverse an overdose, and what types of messaging content could increase uptake, possession, and the likelihood of using naloxone to reverse an overdose’.

Findings from these interviews identified four barriers relating to carriage, possession, uptake and use of THN. These being: (1) stigma related to substance use, (2) indifference toward overdose, (3) fear of negative consequences of carrying naloxone, and (4) fear of misrecognising the need for naloxone.

In conclusion, Bennet and colleagues applied these findings to advocate ways in which the messaging content of OEND may be amended in order to improve THN uptake, carriage and use. Their suggestions for improved message content involved: (1) naloxone is short-acting, and withdrawal sickness does not have to be long-lasting; (2) it is critical to accurately identify an opioid-involved overdose; (3) anyone can overdose; (4) naloxone cannot do harm; and (5) the prompt administration of the medication can help ensure that someone can enjoy another day. Such messages, when embedded within existing OEND, would aim to assist the national public health emergency in expanding naloxone uptake/use to those best placed to respond to overdose events.

Discussion of the article

The article formed the basis of a very lively discussion in which all present raised a number of topics that included THN carriage, formative qualitative research, ethics of payments to participants in qualitative research, the notion of data saturation and cohort size / recruitment of participants in qualitative studies. In short, this article generated a considerable interest in applied qualitative research (beyond the location described within the article) and the key points raised during discussion are summarised below.

Take home naloxone carriage vs non-carriage

The selection of an article with a focus upon THN carriage was considered timely by members of the group. This was due to the prominence of this issue in recent debates within the research literature in which various researchers (see Burton, McAuley et al 2021 and McDonald, Parkin et al 2021) have begun to rethink the way in which ‘carriage’ is understood (and misunderstood) in studies of THN uptake. Similarly, a national poster campaign, during Spring 2021, by a UK rights-based advocacy group (Release 2021) highlighted the importance of carrying THN to prevent fatal opioid overdose. As such, the group felt that this article provided much needed insights into the various reasons why people may not carry naloxone, in a time when public health messaging emphasises carriage, which in turn is interpreted differently by different researchers.

Formative research

The article appears in a journal dedicated to formative research and the authors describe the application of formative qualitative research in their study of THN carriage. The concept of formative research greatly interested the group when it became apparent that studies adopting this approach seek to improve the design and development of existing initiatives. That is, formative qualitative research does not seek to design new intervention, but is typically embedded within existing intervention with the goal of improving and developing current practice. In the context of this study, the group discussed the way in which Bennet and colleagues embedded their study with existing intervention (OEND programmes) with the aim of improving the message content of these programmes (and providing explanations for, and appropriate responses to, non-carriage of THN). The formative research design appears to have contributed to the development of 5 key messages to include in OEND that respond specifically to barriers to THN carriage. This approach was then considered in the context of the aforementioned UK naloxone poster campaign (Release 2021). The discussion focused on how those posters had been designed (without conclusion) including a question concerning whether or not the message content had been informed by a formative research design (and if not, would the content have been any different). The conclusion of the discussion point was that formative qualitative research provides opportunities for researchers to develop the reach of interventions via applied research in a way that perhaps differs from evaluative research (in which the latter typically focuses of ‘what works, how it works’ rather than ‘how it could work better’).

‘Ethnographic’ methods in public places

Associated with the formative qualitative research design was the main method of data generation. This involved conducting semi-structured interviews in semi-public / semi-private locations in green areas throughout NYC. The group found this to be a unique way of obtaining data, in talking to people in more naturalistic environments than, for example, a Needle and Syringe Programme or a treatment centre/methadone clinic. People in the group thought this method allowed for more open and honest discussion and would give the outward appearance of two people sitting on a bench/seat engaged in a personal conversation. (However, some members of the group did wonder how background noise, or the co-presence of the general public may have affected what was said and if the overall clarity of the audio recordings was affected by other people, machinery, birdsong etc). This method was considered by some as drawing upon established methods used by ethnographers in studies of behaviour/cultural practice. This led the group to wonder if any of the authors had academic backgrounds in anthropology, especially given their use of the term ‘key informants’ in the article to describe those with ‘gatekeeping’ roles. As this term is perhaps more synonymous with anthropology than any other of the social sciences, it led to a conclusion that the research team had also (inadvertently?) provided a case study in ‘applied anthropology’ to the study of THN and its carriage. A further discussion point on this issue – although acknowledged briefly in the final sentences of the article – was the way in which COVID-19 and social distancing would currently prevent such ‘ethnographic focus’ in any study of substance use and/or related intervention.

Cash payment as ‘research method’

In the article, Bennet et al describe the way in which research participants were given $20 for completing an interview for the study. This led to a broader discussion of the merits of paying research participants with cash (in recognition of the time, effort, commitment, participation of individuals who are prepared to have social scientists scrutinise aspects of their lives) rather than by shopping vouchers / gift tokens. A further consideration was that the values and culture of organisations hosting a research project (for example, when affiliated to a government department) would invariably determine whether or not cash payments could be provided – or not. Nevertheless, the article by Bennet and colleagues inspired a lively discussion on this issue including the various experiences (including dilemmas and difficulties) of those in the group in providing ‘payments’ for research participants. The conclusion of this topic was that, wherever/whenever possible, researchers should follow the example of Bennet et al and provide cash as part of a reciprocal arrangement for research participation.

Data saturation and cohort size

Bennet et al state that 20 people took part in the research and that this was sufficient to provide data saturation. This inspired a lengthy discussion of the contemporary significance of data saturation in qualitative research per se (as in, is this still a valid concept? Why do researchers feel the need to reach data saturation? Does saturation mean that a structured schedule has been applied to the interview?) and if data saturation was more necessary given the formative research design of the study. Similarly, for some of the group, the number of participants in the study seemed comparatively small to other qualitative studies. This, in turn, led to discussions of the sampling strategy of the research team. For example, the group was interested to learn if a purposive sample strategy employed (recruiting those who did not accept/carry THN) or a convenience sample (whoever was available) or if other strategies were employed. (Relatedly, another discussion point focused upon the limited ethnic diversity amongst the 20 participants, in the multicultural setting of NYC, which led to discussions on the transferability of the findings to those from more diverse ethnic backgrounds).

The article definitely inspired interest in formative research as well as fostered discussion on established debates within the qualitative research paradigm (ethics of payments, data saturation and sampling strategies). However, for some of the topics discussed, the group could only hypothesise on possible reasons underlying the researchers’ rationale for the design and application of the study. This was because of the detached nature (and cultural setting) of the meeting participants from the geographical setting of the study/research team. For these reasons, this summary provides an opportunity for the lead author (Dr Alex Bennet) to provide a post-script response to any of the specific discussion points in the space below (or to provide comment on the Discussion summarised above).

About the article

Bennett, A.S., Freeman, R., Des Jarlais, D.C., Aronson, I.A., (2020). Reasons people who use opioids do not accept or carry no-cost naloxone: qualitative interview study. JMIR Form Res 2020, vol. 4, Iss. 12 | e22411 |

The article can be accessed here.

Lead author’s response to discussion (added as post-script)

Thank you to the study group for selecting our article “Reasons for declining no-cost naloxone” for discussion and for providing very provocative comments and questions and the opportunity to respond. As the group points out, while carrying naloxone is critical, there are very real reasons why people choose not to carry naloxone, and we sought to speak to members of the New York City community where there is a robust naloxone distribution infrastructure and widespread naloxone distribution in low-threshold settings.

Our decision to interview people who use/inject opioids about reasons for declining naloxone was informed by Dr. Bennett’s prior and ongoing work with various communities of people who may be at risk for, or in a position to respond to, an opioid-involved overdose. In a prior study, for example, my team and I worked with veteran health outreach workers, peers with histories of or current substance use, to engage other veterans who may be at risk for overdose in community settings, and offer them free naloxone. Despite the offer of no-cost naloxone, roughly a quarter of all individuals who were approached declined the naloxone. This compelled us to look into the reasons, processes and mechanisms shaping these decisions. As in our other work using qualitative methodologies, at times embedded in mixed-methods designs, we sought to learn from people who use drugs (the experts!) about barriers to naloxone carriage and strategies they deploy for overcoming them. As the study group points out, the research process is a collaborative effort and the knowledge co-constructed. For research to be useful, it must be grounded in the experiences of the people whom it is designed to help in the first place.

In response to your questions about ethnographic methods and the anthropological language used in the paper (i.e., the term “key informants”), my research has generally been community-based, with different groups of people who use drugs, especially opioids, many of whom are economically disadvantaged, such as substance-using military veterans in New York City. My research is grounded in my interdisciplinary graduate and post-doctoral training in history, policy, anthropology, and sociology, and more recently, working with psychologists and epidemiologists. This work aims to capture people’s agency and resilience in light of the myriad social, economic and structural constraints they face on a daily basis. To get at the day-to-day, we have always tried to meet with people where they are most comfortable, often in homes, parks or coffee shops. For this study, my colleague and I met many participants near or in parks, settings that, in many cases, were also close to drug markets and syringe exchange programs. Parks provided adequate space for us to find a spot on the grass or a park bench that was sufficiently quiet, private and away from other people for us to conduct an interview. Also, because a park is a more causal, and for some, more familiar, setting than an office, interviewees seemed to be more comfortable in these places.

With regard to recording quality with interviews conducted outdoors, a high-quality recorder is essential; I use a SONY recorder that allows background noise to be filtered out. Also, I always test my recorder to assess the sound quality of a recording before using it for an interview in a new setting. Other issues we encountered while conducting interviews in public parks include cases where someone sat too close to us, requiring us to move (to insure the interviewee’s privacy). The weather also has to cooperate and in the winter we find that second floors of delis or bodegas (where there are tables and chairs for patrons) often provide the most privacy.

The group raised very good questions about data saturation and sample size. First to provide additional context, I can explain the process by which our team determined that thematic saturation had been reached in a bit more depth than was possible in the article. As the qualitative interviews for this study were being conducted, each completed interview was promptly transcribed by an outside service. On an ongoing basis, our group conducted preliminary analysis of the interview transcripts to identify broad thematic categories addressed in each interview. Our team met weekly to discuss these identified thematic categories which consisted of both a priori constructs (based on the aims of the study and the interview guide) and emerging themes (that were related to the study aims but not specifically anticipated). Once we reached a point where no new thematic areas were identified in newly conducted interviews, we determined by consensus that saturation had been reached; at this point, we completed scheduled interviews and stopped recruiting new interviewees. This process resulted in a total of 20 interviews. As to our rationale for using thematic saturation as a rubric for determining sample size, we see this as a heuristic that is both useful on a practical level (why continue to conduct interviews if new interviews are not generating substantially new information about one’s topic(s) of interest?) and an outgrowth of the interpretive/social constructionist roots of qualitative research methodologies. Whereas more positivistic, quantitative research methods, such as survey research, prioritize samples that are representative of a larger population, and determine appropriate sample size on the basis of the statistical power needed to detect certain relationships between variables, qualitative research traditions typically make no claims regarding the representativeness of a study sample. In the absence of concerns about statistical power, qualitative researchers typically determine appropriate sample size on the basis of the number of participants that is sufficient to provide a thorough understanding of the phenomenon of interest – and techniques such as theoretical sampling and thematic saturation are systematic ways of determining this.

A related question was about sample size and lack of diversity among interviewees. It is very possible that the relative demographic homogeneity of participants, with the majority being socioeconomically disadvantaged white or black persons who inject drugs and utilize a syringe exchange program, had the unintended effect of allowing us to reach saturation with a relatively small sample size (n=20); as a result, we may have inadvertently missed the experiences and reasons for declining or not carrying naloxone that are salient among other demographic groups (e.g., more affluent people who use opioids). As we note in the Methods section of the paper, interviewees were recruited via referral from key informants and chain referral, and all participants had experience using syringe exchange programs, which further limits the generalizability of the findings. That said, what this, and indeed all, qualitative research lacks in generalizability, it makes up for in terms of detail, in this case about the sociocultural contexts and processes underpinning naloxone access, use, and carriage, and the related health risks that are created or exacerbated when barriers impede these outcomes.


Burton, G., McAuley, A., et al (2021). A systematic review and meta-analysis of the prevalence of take-home naloxone (THN) ownership and carriage.

McDonald, R., Parkin, S., et al (2021) Rethinking ‘carriage’ of take-home naloxone.

Release (2021). National Overdose Awareness and Naloxone campaign launches in the UK

The opinions expressed in this post reflect the views of the author and do not necessarily represent the opinions or official positions of the SSA or the author’s academic institution.

The SSA does not endorse or guarantee the accuracy of the information in external sources or links and accepts no responsibility or liability for any consequences arising from the use of such information.


Share this story