Combining Q methodology and interviews using mixed methods integration: an exemplar study exploring over-the-counter codeine misuse in Australia

ABSTRACT

Q methodological studies often incorporate post-Q sort interviews to facilitate and enrich the interpretation of identified factors. This article describes a novel approach in which Q methodology and interviews comprise separate strands which are analyzed separately then converged for further analysis. The new approach is exemplified in a study exploring over-the-counter codeine misuse in Australia. The Q methodology and interview results are summarized, integrated and then discussed, including methodological challenges. This article contributes to scholarly literature by introducing and illustrating the applicability of using mixed methods integration to combine Q methodology and interviews; providing new insights into over-the-counter codeine misuse; and proposing ‘unique contribution’ as an alternative descriptor to the integration fit of ‘silence’.

KEYWORDS:

• Mixed methods research
• combining data
• mixed methods integration
• Q methodology
• drug misuse

Introduction

Q methodology provides an approach to explore the diversity of accounts about a topic, identifying ‘the different perspectives within a group of people, rather than the aggregate, singular view’ (Ramlo, 2015, p. 73). Although the purpose, to study subjectivity, aligns with qualitative research, this is achieved using a mix of qualitative and quantitative techniques. Whilst there is some debate on the classification of Q methodology, it has been recognised as a mixed method approach (Ramlo, 2016, 2021), with the term ‘qualiquantology’ coined to describe its unique hybridity (Stenner & Stainton Rogers, 2004, p. 101).

The first step in a Q methodological study is to identify existing opinion statements about a particular topic, known as the concourse, often via literature review or interviews with potential participants. A representative sample of these statements, the Q sample, is then selected. Next, participants rank the Q sample statements according to their personal views, typically from ‘most disagree’ to ‘most agree’ using a fixed normal distribution grid, in a process known as the Q sort. By-person factor analysis of the completed Q sorts is then undertaken to statistically identify factors, representing shared viewpoints, which are interpreted qualitatively. Participant comments, obtained via responses to post-Q sort questions or interviews, are often used to aid in factor interpretation (Brown, 1980; Watts & Stenner, 2012).

The post-Q sort interview has been described as ‘an important step often overlooked in Q studies’ (Brown, 1980, p. 200). Indeed, to date it has received little attention in the Q methodology literature and few researchers have explicitly detailed this component. Exceptions include articles by Gallagher and Porock (2010) and Wolf (2014). Traditionally, post-Q sort interviews form the final stage of data collection in a Q methodological study. They are used to explore the reasoning behind statement ranking decisions, particularly those statements that participants felt most strongly about. This is used to assist in interpretation of the factors, with participant quotations then used to provide illustrative support (Gallagher & Porock, 2010). Less commonly, interviews have been conducted after factor interpretation as a process of ‘ex-post verification’ (Van Exel & de Graaf, 2005, p. 10), in which participant exemplars are identified to ‘negotiate and flesh out summaries’ (Stainton Rogers, 1991, p. 131) of the factors. It has also been proposed that rather than focusing on the meaning of the Q sort and statements, interviews may be used to more broadly explore participant experiences (Wolf, 2014), and their ‘wider understanding of the issue at hand’ (Watts & Stenner, 2012, p. 82). The possibility that interviews may be considered as a related, but separate, qualitative study has also been suggested (Watts & Stenner, 2012).

Methodological purpose

This article extends the idea of using post-Q sort interviews to explore participant experiences rather than focusing solely on the Q sort, presenting a novel approach in which the Q methodology and interview findings are combined using mixed methods integration. This new approach is exemplified in a study exploring over-the-counter (OTC) codeine misuse in regional Tasmania, Australia.

Study specific purpose

The aim of the exemplar study was to explore Tasmanian OTC codeine misuser accounts of their misuse. Codeine is an opioid pain reliever with addictive potential which may be purchased without a doctor’s prescription (OTC) in a number of countries, including Canada, Ireland, South Africa, the United Kingdom and until recently, Australia. Reports of misuse, serious morbidity and even death have resulted in increasing international public health concern about its availability OTC (Cooper, 2011; Nielsen et al., 2010; Van Hout et al., 2014). Few studies have explored OTC codeine misuse from the perspective of the misuser; this study addresses this gap.

Rationale for the mixed methods approach

The rationale for using mixed methods integration to combine Q methodology and interviews was driven by the study aim, enabling a more comprehensive exploration and understanding of the OTC codeine misuse experience than could have been achieved using a single method. Over-the-counter codeine misusers have previously been described as ‘respectable addicts’ (Cooper, 2011; Van Hout, 2013), based on Cooper’s (2011) qualitative study finding that amongst 25 help-seeking OTC misusers in the UK, many considered themselves to be professional, educated, well-presented, and therefore different to illicit drug users. The exemplar study sought to explore the possibility of alternative accounts of misuse, using mixed methods to allow greater opportunity for these to be captured. Q methodology, to our knowledge not previously reported within the OTC codeine misuse literature, was chosen to facilitate the identification of shared accounts of misuse. The standalone qualitative interviews allowed for a more nuanced, in-depth exploration of individual misuser experiences.

Methods

Design

This study utilized a complex convergent mixed methods design (Appendix A). This type of design consists of two strands; one quantitative and one qualitative, the results of which are merged and compared. The reasoning for the use of a convergent design was ‘to develop integrated results and interpretations that expand understanding, to provide comprehensive results, and/or validate and confirm results’ (Creswell & Plano Clark, 2018, p. 222). The complexity of this study’s convergent design relates to the quantitative strand, which utilized an additional research approach, Q methodology, and also incorporated the use of a Delphi technique in the construction of the Q sample. Although Q methodology itself is considered a mixed methodology, it is considered to represent the quantitative strand for the current study due to the collection of numerical data and use of statistical factor analytic techniques. The second strand is comprised of qualitative interviews. In keeping with convergent design principles, the Q methodology and the interview strands were conducted at roughly the same time, weighted equally, analyzed separately, with the results then integrated to compare and expand on the findings.

To complement the exploratory drive of the research and in line with the philosophical underpinnings of both Q methodology and interviews, a ‘qualitative dominant’ approach was used, taking ‘a qualitative, constructivist-poststructuralist-critical view of the research process, while concurrently recognizing that the addition of quantitative data and approaches are likely to benefit most research projects’ (Johnson et al., 2007, p. 124). This is not to say the qualitative interview strand was given greater priority, rather that both strands were approached with a similar worldview, or philosophical position.

Setting, ethics, participants, instruments, data collection, analysis

The study was conducted in regional Tasmania, Australia, with ethical approval provided by the Social Sciences Human Research Ethics Committee of the University of Tasmania (Reference number H0015843). A parallel sampling model was used, in which two different participant samples were drawn from the same population of OTC codeine misusers for the quantitative and qualitative strands. There was, however, some overlap of participants between the two samples. The methods for sampling and recruitment of OTC codeine misusers, instruments and measures used, and data collection and analysis processes for the separate Q methodology and interview strands are detailed in Kirschbaum et al. (2019a, 2019b, 2020). A brief summation is provided below.

Ninety-nine participants who self-identified as long-term OTC codeine users living in North or North West Tasmania completed the online 46-statement Q sort in response to Facebook or Gumtree advertising, posters or word of mouth, between October and December 2017. Responses of the 26 participants classified as dependent on OTC codeine (Severity of Dependence Scale score of 5 or above) were included in the Q analysis and interpretation. The 26 Q sorts were statistically analysed using Principal Components Analysis followed by varimax rotation, using the program PQMethod version 2.35 (Schmolck, 2014). Factors were interpreted using the crib sheet method described by Watts and Stenner (2012) and via identification of statistically derived distinguishing and consensus statements. Participants’ written comments about their most strongly positioned statements were used to confirm the interpretation.

For the interviews, 15 of the self-identified long-term OTC codeine users who completed the online Q sort were recruited. Semi-structured telephone interviews were conducted between October and December 2017. The interview guide was developed after review of existing literature and included questions such as ‘Can you tell me about how you came to regularly use OTC codeine’ and ‘How has your codeine use impacted your life’? Interviews lasted on average 48 minutes. Transcripts were analyzed abductively using qualitative content analysis.

Mixed methods integration

The Q methodology and interview findings were integrated by comparing the results side-by-side in a joint display (Creswell & Plano Clark, 2018). Common concepts or content areas represented in both sets of findings were identified (Creswell & Plano Clark, 2018, p. 70, 224). This was achieved using the inductive approach of ‘following a thread’ (Moran-Ellis et al., 2006, p. 54); selecting a theme from one dataset and following it across the other. Initially, the Q methodology findings were followed across the interview results, in keeping with the chronology of the strands. This process was then repeated in reverse; following the interview findings across to the Q methodology results. The results were then compared side-by-side to draw metainferences describing the ‘fit’ (Fetters et al., 2013) or coherence of the integrated findings: ‘congruence’ where the qualitative and quantitative results led to similar conclusions; ‘discrepancy’ where the two sets of results conflicted; and ‘unique contribution’ where a finding from one strand did not arise in the other strand. In addition to assessing the quality of the individual qualitative and quantitative strands, the overall project was evaluated according to the core criteria for appraising mixed methods research quality suggested byFàbregues and Molina-Azorín (2017).

Results

This section summarizes the findings from the Q methodology and interview strands and presents the integrated findings in a joint display. The integrated findings are then interpreted in the Discussion. For a full presentation of the Q methodology and interview data, see Kirschbaum et al. (2019a, 2020).

Q methodology findings

Statistical analysis of the 26 Q sorts produced a two-factor solution, with 12 participants loading significantly onto Factor One and 13 on Factor Two. The two distinct factors, or accounts of misuse, were allocated the titles: ‘For pain, no shame’ (Factor One) and ‘My secret solace’ (Factor Two). The Q sorts exemplifying each factor are shown in Appendices B and C.

Factor One: ‘For pain, no shame’ respondents were not ashamed of their codeine use; believing that they used it responsibly and legitimately for the relief of physical pain. Codeine use was attributed to external causes, such as the underlying condition causing pain. Participants representing this factor were strongly opposed to codeine up-scheduling, supporting OTC availability to allow timely access to pain relief. They believed that people dependent on OTC codeine are different to those dependent on injected drugs, and that recovery is a lifelong process requiring professional intervention.

Shame and secrecy featured strongly in the accounts of participants aligning with Factor Two: ‘My secret solace’. These respondents were fully aware they were misusing codeine and attributed this to personal choice rather than external factors; they intentionally took more codeine than is recommended for mood effects rather than pain relief. The responses of Factor Two participants suggested they were not strongly opposed to codeine up-scheduling. Like Factor One participants, they considered people dependent on OTC codeine to be different from those dependent on injected drugs, and recovery to be a lifelong process requiring professional intervention.

Interview findings

Thirteen categories emerged from the OTC codeine misuser interview accounts, which were organized into three broad temporal phases: transition to misuse; growing awareness; and towards recovery (Appendix D).

Transition to misuse

Initiation of OTC codeine was almost always for self-treatment of legitimate physical pain. Two distinct typologies of misuser were identified; those who reported ongoing use for self-medication of physical pain and those who acknowledged use for the calming mood effects to self-manage stress and mental health conditions. Progression to misuse was often described as accidental, with misusers frequently stating that codeine was helping and/or that dosing directions were being followed. The perception that OTC codeine was safe, particularly when compared to illicit drugs, also influenced the transition to misuse.

Growing awareness

Initially, the benefits of codeine use outweighed concerns. Many participants were ashamed of and attempted to hide their use. In contrast, some who used codeine for pain relief did not keep their use a secret. Several misusers described having workmates, family or friends who were also using OTC codeine, who they felt they could talk to. Significant others were sometimes involved in purchasing OTC codeine. Misuse was reported to have negatively affected some relationships. Challenges with obtaining codeine, the manifestation of side effects, or information accessed via the internet often prompted acknowledgement of the existence of a problem. Participants blamed themselves, their health condition and/or the health system for their misuse.

Towards recovery

Many participants in this study had successfully self-managed their path to recovery without formal intervention, with support provided by family and information accessed via the internet. Recovery, described as difficult for some, and easy for others, was considered to be an ongoing process, often involving cautious occasional use or continuing temptation.

Integrated findings

Integration of the Q methodology and interview findings revealed areas of congruence and discrepancy, as well as ‘unique contributions’ arising from a single dataset. The integrated findings are presented in Table 1.

Table 1. Joint display table integrating findings from the Q methodology and interview strands.

Metainferences (Fit of integration)	Concept/content area	Q methodology findings	Interview findings	Explanation of integration fit
CONGRUENCE	Disclosure	Factor One – not ashamed of codeine use, did not feel the need to hide it (36: +2; 38: −1; 39: −3; 42: −4) Factor Two – shame and secrecy, desire to hide codeine use (36: −5; 38: +5; 39: +3)	Category – Feelings regarding disclosure polarized; shameful hiding or open use	Two distinct accounts identified.
	Primary reason for long-term use	Factor One – for pain relief (43: +5) Factor Two – to relax and relieve stress (37: +4; 46: +4)	Category – Use continued to relieve physical pain or for calming effects on mood	Two distinct typologies identified. Interview results provided more detail e.g. Factor One occasional use for pain prophylaxis rather than relief, Factor Two descriptions of the calming feeling, self-management of mental health conditions.
	Attributions	Factor One – attributed codeine use to external causes e.g. circumstances (22: +4), dependence is accidental (30: +2) Factor Two – codeine use was considered to be a personal choice, not attributed to circumstances (22: −4) (but possibly believed there was no choice once addiction was established: ‘It was my choice, but now it’s more of an addiction instead of a choice’ (P9)	Category – Blame attributed to self, health conditions, health system	Both internal and external attributions identified. The Factor Two Q results suggested that attributions may change over time, which was not explored in the interviews (see Unique Contributions).
CONGRUENCE	Perception of safety	Factor One – believed people dependent on codeine are different to people dependent on injected drugs (31: −1), and that OTC codeine is a less serious drug of dependence (12: +2) Factor Two – believed people dependent on codeine are different to people dependent on injected drugs (31: −3) possibly due to perceived differences in safety and legality: ‘Most injected drugs are both illegal and made using who knows what chemicals in backyard labs’ (P20)	Category – Perceived as safe; available from pharmacies, legal not injected, absence of noticeable side effects, doctors not concerned (Many participants specifically contrasted OTC codeine use with illicit and/or injected drug use: ‘It’s not made out of the back of bloody, you know, Huon Valley or somethin’ in someone’s caravan … you know it’s cleaner, you know it comes in a packet, it’s done by you know pharmacists’ (R6))	Perception of safety of codeine, particularly compared to injected and illicit drugs. Interview findings provided more detailed reasoning as to why OTC codeine was perceived to be safe.
	Codeine accessibility	Factor One – opposed to up-scheduling; believed it is a right to access codeine OTC (18: +5), considered recording sales on a national database to be a suitable solution (19: +4) Factor Two – possible support for up-scheduling, did not believe that up-scheduling would deny patients the right to timely pain relief (18: 0) or that a national database of sales was a suitable solution (19: −2)	Category – Mixed views about OTC codeine accessibility	Two distinct accounts identified. Interviews provided insight into the possible reasoning behind the finding that some OTC codeine misusers support up-scheduling – public protection.
	Recovery	Both Factors – recovery is a lifelong process Factor One – (8: +3) Factor Two – (8: +2)	Category – Recovery is ongoing; cautious occasional use, continuing temptation	Recovery considered to be an ongoing process. Interview findings expanded on this, elucidating reasons for this belief.
DISCREPANCY	Recovery	Both Factors – recovery requires professional intervention Factor One – (8: +3; 11: +3) Factor Two – (8: +2; 11: +2)	Category – Support for recovery provided by family and information access via the internet	Q findings indicated that recovery requires health professional intervention, while interview findings suggested recovery via self-change.
	Level of awareness of misuse	Factor One – did not identify as misusers: denied use for reasons other than pain relief (37: −5; 46: −3; 21: −4); did not consider they were taking more tablets than recommended (1: −1); believed they followed directions on box (6: −3) Factor Two – acknowledged their misuse: aware that they were taking more than recommended (1: +5) for reasons other than pain relief (37: +4; 46: +4); dangers and directions on box ignored (6: +3; 2: +4)	Category – Insidious; accidental, gradual, easy to justify use Category – Self-awareness of misuse prompted by side effects, difficulty accessing codeine, information from the internet	Interview results indicated a gradual awareness of misuse over time, while the Q results suggested a more polarized view that those using codeine for calming mood effects are aware of misuse while those using codeine for pain relief are not.
UNIQUE CONTRIBUTIONS			Categories: − Initial use for physical pain relief − Benefits initially outweighed concerns about dependence and long-term side effects (no specific Q statements about this, although Q Participant 4 commented that ‘once addiction took place I ignored the dangers of regular codeine use. The positive qualities outweighed physical dangers in my mind’) − Relationships − Recovery was difficult for some, easy for others − Support for recovery (no Q statements regarding family support, internet) − Insidious; accidental, gradual, easy to justify use (no Q statements regarding transition to misuse)	No comparable Q findings as the Q sample contained no statements covering these concepts.
			− Self-awareness of misuse prompted by side effects, difficulty accessing codeine, information from the internet (no Q statements regarding awareness prompts)
		Possibility that attributions may change over time: Factor Two – ‘It was my choice, but now it’s more of an addiction instead of a choice’ (P9) Factor Two – believed that treatment for OTC codeine dependence should be provided together with people dependent on other types of drugs (16: −4)		No comparable interview findings; not specifically explored in the interview schedule.

Discussion

To the extent of our knowledge, this is the first study to utilize mixed methods integration techniques to combine Q methodology and interview findings. The fit of integration, including choice of terminology, is discussed below, followed by the content-specific findings and methodological contributions of this study, its limitations and areas for further research.

Fit of integration of the current study results

The integrated findings from the Q methodology and interview strands were mostly congruent. Furthermore, the interview results often illuminated the Q methodology findings. This reflects the nature of qualitative research and its ability to capture rich and detailed participant responses. For example, both the interview and Q methodology data indicated that some OTC codeine misusers were supportive of up-scheduling. The qualitative interview data provided expansion of this common finding by suggesting the reasoning behind this view; protection of the public. There was also an example of the Q methodology results expanding a congruent finding. Each dataset indicated that both internal and external attributions exist amongst codeine misusers. The Q sort results suggested that these attributions may change over time. Though not explored in the interviews, the notion that attributions may change presents an opportunity for further research.

The integrated results revealed two areas of discrepancy, for which there are plausible explanations. Nevertheless, it is also possible that the inconsistencies may be due to the complexity of the codeine misuse experience or may represent a third account of misuse not identified in the Q methodological strand.

Firstly, the interview findings indicated that misusers gradually became aware of their misuse over time, while the results of the Q methodology strand suggested that those using codeine for calming mood effects were aware of their misuse and those using codeine for pain relief were not. A likely explanation for this discrepancy is that a gradual process of awareness was not able to be expressed by Q participants as no Q sample statements covered this. It is also possible that Q participants representing Factor One, ‘For pain, no shame’, may have been at an earlier stage of change and not yet cognizant of their misuse.

A second discrepancy concerned the findings regarding the process of recovery from OTC codeine misuse. The Q methodology data indicated that recovery requires professional intervention, while the interview results suggested that recovery occurs via self-change with support provided by family and/or the internet. It is possible that the lack of awareness of self-change as a recognised form of recovery (Bishop, 2018) may have influenced responses in the Q sort; participants may have felt that even if they were able to recover without professional intervention, that others would require assistance. Another possibility is that Q sort participants believed that professional intervention was required for recovery as they may have had a higher level of codeine dependency than the interview participants; Q sort participants all had a Severity of Dependence Scale score of 5 or above, whereas the interview participants did not have a Severity of Dependence Scale score inclusion criterion and their scores ranged from 2 to 14.

In this study, ‘unique contributions’ arising from a single dataset were considered to be important findings enhancing understanding of the OTC codeine misuse phenomenon. They may have arisen from the differing strengths and limitations of the two strands. For example, seven of the 13 categories resulting from the interview analysis had no comparable statements in the Q sample. This was because it was possible to explore more concepts in the interviews compared to the Q sort, as the number of Q statements able to be presented was limited to avoid participant fatigue.

The ‘unique contributions’ of the interview strand were investigated to see if they had been represented in the initial concourse of Q statements. Indeed, all were represented in some way in the overall concourse and in retrospect, it would have been useful to include some of these statements in the Q sample to verify the interview findings. Researchers wishing to replicate this study may wish to consider this as an option. However, the optimal size for a Q sample is another identified methodological issue, with a lack of consensus within the Q methodology literature. The completion rate for this study’s Q sort was 55%; 99 completed responses out of 180 commenced surveys, which could indicate fatigue, difficulty with comprehension or task complexity.

Terminology used to describe integration fit

The terminology used by mixed methods researchers to describe the fit of the integrated results is not universal or consistently defined. Fetters and Molina-Azorin (2017, p. 302) describe four possible outcomes: complementarity; ‘the qualitative and quantitative data when compared, tell different but nonconflicting stories’, confirmation; ‘the qualitative and quantitative data give cause for drawing the same conclusion’, expansion; ‘combining of the qualitative and quantitative data provide a broader but overlapping understanding’, and/or discordance; ‘the situation when the data may conflict with each other’. Fetters, in collaboration with Curry and Creswell (2013), had previously outlined only the latter three. Creswell and Plano Clark (2018) later reference multiple terms within the same book, including: confirm, disconfirm, expand, convergence, divergence, contradictions, relationships (p. 222) complement (p. 226), consistencies, inconsistencies, conflicts, contradictions, complexities, discrepant and congruent (p. 233). Other terms used by mixed methods researchers include: agreement, partial agreement, dissonance and silence (O’Cathain et al., 2010).

For this study, the terms congruence and discrepancy, recently favoured by Creswell and Plano Clark (2018, p. 233), were chosen. In addition, the term ‘unique contribution’ was used to describe the fit of integration in which a finding from one strand did not arise in the other strand. This latter type of integration fit is rarely considered in scholarly discussions about mixed methods integration, including those by Creswell and Plano Clark (2018), Fetters et al. (2013), Fetters and Molina-Azorin (2017) and Morgan (2019). An exception is the triangulation protocol, which involves consideration of areas of ‘silence’; where ‘one set of results covers the theme …, whereas the other set of results is silent on the theme’ (Farmer et al., 2006, p. 383).

The triangulation protocol was originally designed for the integration of multiple qualitative methods, but was suggested by O’Cathain et al. (2010) to be relevant to mixed methods integration. The protocol has since been utilized in a number of mixed methods studies, including those by Dodd et al. (2020), Harrison et al. (2020), Heslehurst et al. (2015), Racine et al. (2020), Taylor et al. (2020), Tonkin-Crine et al. (2015) and Toomey et al. (2017). The term ‘silence’ was not used in the current study to avoid confusion with the more traditional usage of the term within Discourse Analysis, where it is used to describe ‘individual speakers when they make a more conscious and intentional choice about what (not) to say – when they choose to say nothing, but instead could have said something’ (Schröter & Taylor, 2018, p. 7).

Content-specific findings

Four key content-specific findings were identified amongst the integrated results which are relatively underrepresented in the existing body of knowledge about OTC codeine. They provide new insights into the misuse experience and are likely to make a difference to the way in which OTC codeine misuse is understood. The four key content-specific findings are: (1) the identification of two distinct accounts of misuse; (2) use to self-treat physical pain, stress and mental health conditions; (3) a gradual transition from use to misuse and growing awareness of a problem; and (4) the ability to recover via self-change.

Contribution to the field of mixed methods

This article has described and illustrated the successful application of a mixed methods approach to combine Q methodology and interviews. This approach is novel for several reasons. It extends the post-Q sort interview beyond its traditional purpose focussing on Q statement ranking decisions (Gallagher & Porock, 2010), to a standalone, broader exploration of participant experiences. In addition, the use of mixed methods integration techniques to combine the rich participant responses afforded by the interviews, with the statistically based accounts generated by the Q sort, has not been previously reported. Furthermore, Q methodology itself has only rarely been used within mixed methods study designs.

Mixed methods integration of the Q methodology and interview findings has identified areas of congruence and discrepancy, as well as ‘unique contributions’ arising from a single strand. The integrated findings contribute a greater understanding of OTC codeine misuse compared to a single method approach and highlight the complexity of the OTC codeine misuse phenomenon, as well as suggesting areas for further research.

This article has also highlighted the lack of consensus in terminology used to describe the fit of integration in mixed methods research. While there is a necessity for the evolution of new language in ‘our fast-changing mixed methods world’ (Fetters & Molina-Azorin, 2017, p. 5) and non-fixed terms and flexible definitions ‘help acknowledge and even embrace the “messiness” of doing mixed methods research’ (Åkerblad et al., 2021, p. 167), standardization of terminology may reduce definitional confusion and facilitate comparison between studies, particularly for researchers new to the field.

In this article, we propose a new term ‘unique contribution’ as an alternative to ‘silence’ to describe the fit of integration in which a finding from one strand did not arise in the other. Use of this new descriptor satisfies a number of the principles for adding new and divesting of old language as proposed by Fetters and Molina-Azorin (2017). Namely, ‘unique contribution’ provides greater clarity, being specific to mixed methods without having another meaning in another paradigm. The new term may be useful for other mixed methods researchers wishing to describe this less commonly reported type of integration fit.

Limitations

This research has a number of strengths, including originality and rigour in its design and conduct. However, it is not without limitations, including the small and not necessarily representative sample of misusers and the possibility of recall bias amongst past codeine misusers. The findings also represent the views of the sampled participant group at a specific time point, which may not be enduring. With regard to integration of the two strands, it would have been ideal if the same Severity of Dependence Scale inclusion criterion had been used for both the Q methodology and qualitative interview participant groups, to facilitate the mixed methods analysis. Furthermore, a larger Q sample with closer alignment to the interview guide, using parallel questions, may also have assisted the integration process. It is acknowledged that the mixed methods design of this study could also potentially be viewed as explanatory sequential; use of the qualitative data to further explain the Q factors. However, the interviews were not designed or analysed based on the Q methodology findings, therefore the study is more closely aligned with convergent design principles.

Areas for future research

Further exploration is needed of the areas of discrepancy and the ‘unique contributions’ identified during mixed methods integration, such as whether codeine misuser attributions change over time and whether other accounts of misuse exist. Researchers wishing to replicate this study should consider the addition of Q sample statements and/or interview guide questions to explore these areas, being mindful of the possibility of participant fatigue. The establishment and standardization of the most appropriate terminology to describe the fit of mixed methods integration, including assessment of the utility of the newly proposed term ‘unique contribution’, is indicated. Finally, further studies are needed to confirm mixed methods integration as a useful approach to combine Q methodology and interviews.

Conclusion

This article has illustrated the applicability of a mixed methods research approach to combine Q methodology and interviews. The exemplar study, exploring OTC codeine misuser accounts, contributes new insights into the experience of OTC codeine misuse. The terminology used to describe mixed methods fit of integration was discussed and advanced, with the term ‘unique contribution’ proposed as a replacement for the integration fit of ‘silence’.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Melissa Anne Kirschbaum

Melissa Kirschbaum is a pharmacist who completed her PhD at the University of Tasmania Centre for Rural Health in 2022. Her research interests include drugs of addiction, health professional education and clinical training, and research methodologies.

Tony Barnett

Tony Barnett was the Director of the Centre for Rural Health at the University of Tasmania. He has published over 80 refereed research papers in international journals and been awarded a significant number of project and research grants. Tony maintains research interests in rural health services, workforce and clinical education; health program evaluation; ageing well; interprofessional learning, education and practice; and oral health care.

Merylin Cross

Merylin Cross has a dual background in nursing and sociology. During her academic career, she has developed a particular interest in qualitative and mixed methods research designs and their utility in addressing health and social issues from multiple perspectives and at the interface between systems and individuals where policy affects people.

Appendices Appendix A.

The convergent mixed methods study design utilized for this study

Appendix B.

Factor-exemplifying Q sort for Factor One ‘For pain, no shame’ ^1.

1. An estimation of the response of a participant who would load 100% onto Factor One, calculated using the weighted average of all the Q sorts that load significantly onto Factor One.

Source. Kirschbaum et al. (2019a)

Appendix C.

Factor-exemplifying Q sort for Factor Two ‘My secret solace’ ^2.

2. An estimation of the response of a participant who would load 100% onto Factor Two, calculated using the weighted average of all the Q sorts that load significantly onto Factor Two.

Source. Kirschbaum et al. (2019a)

Appendix D.

Categories that emerged from misuser accounts

Table

Temporal phase	Category
Transition to misuse	Initial use for physical pain relief; directly OTC or following use of prescribed analgesics
	Use continued to relieve pain or for calming effects
	Perceived as safe; available from pharmacies, legal, not injected, absence of noticeable side effects, doctors not concerned
	Insidious; accidental, gradual, easy to justify use
Growing awareness	Benefits initially outweighed concerns about dependence and long-term side effects
	Feelings regarding disclosure polarized; shameful hiding or open use
	Relationships; negative impact, talked to family/friends who were also misusing, others used to purchase OTC codeine
	Mixed views about OTC accessibility; easy, depends on pharmacist, became more difficult over time
	Self-awareness of misuse prompted by side effects, difficulty accessing codeine, information from internet
	Blame attributed to self, health conditions, health system
Towards recovery	Support for recovery provided by family and information accessed via internet
	Recovery was difficult for some, easy for others
	Recovery is ongoing; cautious occasional use, continuing temptation

Source. Kirschbaum et al. (2020).