Search strategy

Qualitative studies using methods involving direct patient contact, such as interviews and focus groups, and seeking to understand the patient experience of treatment burden across all conditions and treatments were sought. Searching and screening were conducted according to the PRISMA statement (See S1 PRISMA Checklist) [15]. The data bases Medline, CINAHL, Embase, and PsychINFO were searched electronically. No date limitations were set but “language” was restricted to English or Portuguese as there were no resources for translation. We aimed to identify all papers that used the terms “treatment burden” or “burden of treatment” in their title or abstract. Given that the systematic identification of qualitative research is problematic [12, 16] we did not limit our initial search by research method. Rather, identification of qualitative papers was undertaken during the blinded screening process. Initial searches were conducted in June 2012 and were updated in April 2014.

Data screening, extraction and analysis

Title, abstract and full paper screening were undertaken independently by three researchers (AJM; ACG; CA) using a data-extraction proforma designed and piloted by the team. Inclusion was accepted by concordance; a third party (SD or KH) resolved any disagreements. Duplicates and any papers not addressing treatment burden at the level of the patient were excluded (e.g. global economic treatment burden; treatment burden on services). Papers were included only if they utilised recognised inductive qualitative data collection and analysis methods. Quantitative research, systematic reviews, qualitative syntheses, opinion pieces and papers reporting qualitative methods but containing no qualitative data (e.g. quotations or thematic frameworks) were excluded.

Qualitative research is interpretative: data therefore included verbatim quotes and authors’ interpretative comments and were extracted from the findings/results and discussion sections of papers [11]. Data were analysed using framework synthesis [17, 18] using a coding framework informed by the Cumulative Complexity Model (CCM) [19]. The CCM proposes that the balance between patient workload (treatment, ‘everyday’ and occupational tasks) and their capacity to undertake that work influences access and adherence to treatments and consequent health outcomes. This model was appropriate to our focus on understanding how BoT impacts on “patient functioning and well-being” [3] and the factors contributing to and shaping these experiences. We used Shippee et al’s categories (e.g. capacity, workload, adherence etc.) to generate our framework but did not pre-determine the existence of their proposed inter-category relationships.

Framework synthesis uses a two staged approach; data extraction and management into pre-determined categories and then thematic analysis to identify patterns of data within and between categories [17]. A data management framework, with 4 well defined categories (workload, capacity, treatment impact and engagement/adherence) was developed. Data assignment was undertaken by pairs of researchers (SD/CA: papers 1–4; KH/RO: papers 5–8; SD/AM: papers 9–11) who coded blind and then met in pairs to discuss and agree categorisation. Finally, each pair presented their coding to the other and any issues of contention were discussed and agreed by the whole team.

The second phase of framework synthesis involved thematic analysis of data categorised within the ‘treatment impact’ category. Two researchers (SD/KH) conducted this analysis separately, and then through collaborative discussion, using paper based labelling, fragmenting, comparing and grouping [20] until a clear taxonomy of the components of treatment impact were agreed. Finally, relationships between the treatment impacts and those proposed previously (e.g. capacity, workload and adherence) were identified using constant comparison, team brainstorming and diagrammatic modelling. Our synthesis generated both second-order (interpretations offered by the original researchers) and third-order constructs (new interpretations beyond those offered in individual studies) [21].

Our emerging synthesis indicated that people seek healthcare not simply to relieve physical or emotional symptoms but because those symptoms stop them from doing what they want (e.g. running or hiking) and being who they want to be (a professional athlete, a member of the rambling club or an optimistic person). In finalising our synthesis model we therefore called upon Sen’s capability approach [22] which considers the genuine opportunities (capabilities) people have to achieve the kind of lives they value: to feel like, do what and be who they want to be. Treatment burdens were therefore considered to be any treatment generated disruption in people’s ability to feel, do or be who they wanted to be.

Quality appraisal

Quality appraisal was undertaken using the Critical Appraisal Skills Programme [23] criteria for qualitative studies. Quality appraisal was independently conducted by two researchers (SD and KH) and answers compared and discussed. Studies were not, however, excluded on the basis of quality.

Enhancing analytic credibility through expert patient review

Analytic relevance and credibility was enhanced by inclusion of a lay representative in the study team (RP). RP brought the following expertise to the team: i) experience of a long-term condition (stroke); ii) pre-stroke expertise in analysing complex systems and processes iii) leadership of local and national patient organisations. RP critiqued and challenged the emergent analysis, reflecting on its relevance to his own experience and those of other people with long-term conditions that he worked with.

Retrieved studies

The initial (April 2012) and updated (April 2014) searches identified a total 1177 papers; after removal of duplicates, 774 titles and abstracts were screened for relevance; 368 full text articles were assessed for eligibility; 11 papers which used qualitative methods and analysis and which presented data on patients’ perspectives of treatment burden were included. Fig 1 presents the PRISMA flow diagram indicating the inclusion and exclusion of papers at each stage of the screening process.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow diagram indicating inclusion and exclusion criteria of papers at each stage of screening.

https://doi.org/10.1371/journal.pone.0125457.g001

Study details

Details of included studies are presented in Table 2. A range of qualitative methods were reported: ten used semi-structured interviews, either alone [3, 7, 24–28], with focus groups [29, 30] or with structured measures [8]; one conducted secondary analysis of existing qualitative data [11]. Included papers addressed a range of chronic conditions across the life-course (aged 8–96 years) including: adults with spasmodic dysphonia [7]; chronic heart failure (CHF) [11]; conditions requiring percutaneous endoscopic gastrostomies (PEG) [8]; chronic kidney disease (CKD) [24, 28]; people with limited life expectancy due to various conditions [29]; adolescents/young adults with cystic fibrosis (CF) [25] and tuberculosis (TB)[30]; children with Primary Ciliary Dsykinesia (PCD)[26] and, a range of chronic conditions [3, 14]. The conditions varied in terms of the severity and impact of disease, the likelihood and immediacy of life threat, and the invasiveness and criticality of treatments. Studies were undertaken in: the UK [8, 11, 24, 26], US [3, 7, 25, 29], Nepal [30], Australia [27], and Greece[28]. All of the papers included a mix of genders. A variety of qualitative analysis methods were used; all sought to identify common themes raised by participants.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Details of included papers.

https://doi.org/10.1371/journal.pone.0125457.t002

Quality appraisal

Papers were moderate to high quality. The sampling strategy, relationship between researchers and participants and detailed consideration of ethical issues were the weaker elements of these papers.

Thematic findings

The synthesis generated eight second-order constructs related to the “negative impacts of treatment on functioning and well-being” which we collated into three third-order constructs “biographical, relational and biological treatment disruptions”. Table 3 presents these 2^nd and 3^rd order constructs and identifies where evidence for each can be found. The synthesis generated a further two third-order constructs related to the strategies employed by patients to minimise the disruptions to their valued capabilities: “adaptive treatment work” and “rationalised non-adherence”. This is also presented graphically in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The biographical, relational and biological disruptions generated by treatment burdens and the strategies of adaptive work and rationalised non-adherence which patients employ to minimise these.

https://doi.org/10.1371/journal.pone.0125457.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Definitions of the 2nd and 3rd order constructs identified, how they inter-relate and where evidence for each can be found.

https://doi.org/10.1371/journal.pone.0125457.t003

Strengths and limitations

This review is the first to explore and understand how treatments generate biological, biographical and relational disruptions in a range of conditions across the life course and illness trajectory. Moreover, we have been able to identify additional secondary treatment work required to minimise those disruptions: the combined strategies of adaptation and rationalised non-adherence. Eleven empirical studies were included, reporting on the perspectives of treatment burden from a total of 294 patients, across a range of age groups and countries. However, there are a number of limitations to our review. For instance, although we adopted a thorough and comprehensive search strategy, some relevant studies may not have been identified. We wanted to focus on gathering emergent data from patients’ own perspectives (rather than more researcher-led deductive methods) and therefore included only papers which used inductive qualitative methods suitable for generating depth data. This may have resulted in the exclusion of potentially relevant data gathered using structured postal, on-line or telephone surveys. Further, we restricted our search to English and Portuguese reports as we had no resources for translation. However, we consider our analysis to have produced an integrated model with sufficient explanatory power to explain the relationships between treatment workload, capacity, disruptions and adherence. The quality of included studies may affect the validity of our findings: although we undertook quality review of all relevant studies, in the absence of consensus on the best way to appraise qualitative research [32], and because we wanted to maximise the reach and comprehensiveness of our findings, we did not exclude on this basis. All aspects of data extraction, quality appraisal and data analysis were conducted by teams of two researchers, with a third party for disagreements. This minimised researcher bias and enhanced analysis. Framework analysis provided a robust theoretical underpinning, using existing models of treatment burden such as NPT and the cumulative complexity model to inform analytical development. We consider this an effective approach but acknowledge that a priori frameworks risk forcing data inappropriately. We sought to minimise this risk, deliberately moving from our initial deductive use of these a priori models (to fractionate and manage the data) into inductive thematic analysis specifically focussed on the perceived impact of treatment burden and the strategies employed to manage or reduce this. A further strength of this work is the inclusion of an expert patient researcher who contributed to the data analysis and writing of the paper, challenging underdeveloped conceptions and confirming the saliency of the findings to his own experiences and those of other people with chronic illness that he works with.

Treatment generated disruptions

Over recent years great strides have been taken to clarify conceptual and operational definitions of treatment burden. To date, much of this research has focused on the burden arising from treatment workload and the impact that workload has on the maintenance of normal activity. For instance, Gallacher’s work [11, 12] has highlighted the steps people take to embed treatments into their daily lives; Eton’s work [3] has considered the impact of patient functioning on wellbeing; and Sav’s work [14, 27] has indicated that treatment burden consists of both objective elements, such as total workload and work complexity, and subjective, patient specific, elements. By synthesising the evidence across qualitative studies of treatment burden, our review has been able to expand further on these subjective elements and impacts.

Our findings support the theory proposed by May, Eton [33], by showing that there are important factors, in addition to the complexity of treatments and the time involved in enacting them. This means that treatment burden is brought about by both the workload associated with treatment, and the impacts that workload and treatment complexity have on everyday life, valued daily activities and patient identity. In this way, we have shown that treatments and their total workload cause disruptions to a person’s biological, biographical and relational capacity. Despite finding, in line with previous literature [4–6], that treatments lead to biological disruptions in the form of physical symptoms and side-effects (such as pain and nausea), our findings also indicate that it is often not the severity of symptoms that determines how burdensome treatments are: rather, that the biographical and relational disruptions arising from those symptoms and side-effects have important impacts for patients. For instance, treatments had effects on identity, interaction with others and, in many cases, were associated with negative affective states. These affective states include anxiety, fear, anger, and frustration. In some cases these symptoms were severe and debilitating, further impacting on independence, relationships with others and ultimately, adherence to treatment regimens. Whilst previous work in this field has highlighted consequences such as fatigue [8, 27] and frustration [3, 7, 8, 11, 24, 27, 30], the results of this review indicate that the psychological and biographical consequences may be more far reaching and severe than initially considered. Further research is required to explore the impact and severity of negative affect and biographical disruption arising from treatment burden, in order to investigate relationships with quality of life, treatment adherence and outcome.

Strategies for minimising treatment disruptions: adaptation

This review also adds to the conceptual armoury of treatment burden theory by identifying two strategies for minimising treatment burden and disruptions: the first of these is adaptation. Building on previous research [1, 2, 11, 12, 31, 34] we identified three forms of adaptive work:

1. Patients engaged in their own form of Sentimental work to manage the negative affective states associated with treatment burden. This involved self-soothing behaviours, managing contact and interactions with friends and family, and developing other strategies to minimise distress. This builds on the work of Corbin and Strauss [1] by extending the concept of ‘sentimental work’ to work within a patient’s scope, rather than work limited to the remit of health professionals. Indeed, in a self-management care model where much of the day-to-day work of illness management is delegated to patients, sentimental work is an important part of that delegated work.
2. Biographical work was employed by patients to manage changes in their identity brought about by the burden of treatment. This builds on Bury’s description of illness work [2] by proposing that, in a healthcare model that has added treatment work to a patient’s total self-care workload, patients need to engage in biographical work to maintain existing or adapt to new identities caused by treatment-generated changes.
3. Relational work was carried out to maintain relationships. Earlier burden of treatment research, which was more focussed on the planning, doing and monitoring work of treatment [3, 11, 12], has discussed the concept of ‘treatment related relational work’. However, this concentrated on the strategic mobilisation of others to facilitate treatment. Our review extends this research by addressing the work that patients do to minimise the impact of treatment on valued relationships with self and others. In this way, patients use relational work to draw on a wider workforce [11, 12, 34] but also to minimise relational disruption and maximise the work output from that workforce by maintaining optimal, productive and agreeable relationships.

Strategies for minimising treatment disruptions: rationalised non-adherence

We identified a second strategy to minimise treatment burden: rationalised non-adherence. We found that non-adherence was often not an arbitrary act or sign of personal moral failure as it is sometimes viewed [35, 36] but a rationalised process, used actively and mindfully, as a way of minimising the burden of treatment. Karamanidou, Weinman [28] discusses a similar phenomenon in patients undergoing haemodialysis, which they called ‘active non-adherence’. They attributed this non-adherence to beliefs about the importance (or not) of treatments, however, we noted non-adherence associated with rationalised decisions based on a desire to minimise the disruptions associated with treatment. As a result, we have termed this strategy ‘rationalised non-adherence’. Rationalised non-adherence occurred when patients, having appraised the impact of treatment, deliberately decided to cease, modify or reduce their treatment regime.

We identified two distinct forms of rationalised non-adherence. We found that in studies which included populations approaching the end of life, non-adherence decisions were frequently sanctioned and supported by clinicians [24, 29]. Indeed, recent policy and practice initiatives recommend open discussion about end of life treatment decisions and include directives to support and endorse patients’ decisions, even where these are decisions not to treat [37–39]. This is in contrast to how non-adherence was discussed in studies exploring less critical situations [14, 25–28]. We found that rationalised non-adherence at other points in the illness trajectory or life course was a ‘secret-act’ that must be hidden from others. Such (non)-treatment decisions were undertaken covertly, without the knowledge, approval or guidance of professionals. This suggests a lack of concordance between patients’ and clinicians’ perspectives, which according to our data, results in feelings of guilt. It may also affect patient outcome, because important information is withheld that could affect clinical-reasoning and future treatment recommendations. Adopting some of the principles of the palliative care approach would facilitate open discussion about the impact of treatment burdens and disruption on adherence to treatment, thus providing healthcare professionals with an opportunity to consider the balance between treatment burden, adaptation and adherence. Links between treatment burden and treatment adherence have also been discussed by May, Montori [40] in their seminal text on Minimally Disruptive Medicine. They propose that treatment burden can lead to structurally induced non-compliance as a result of increasingly complex and abundant treatment regimens. In this way, structurally induced non-compliance might be viewed as an overarching term linking treatment burden with adherence. Our analysis allowed us to explore these links in greater depth and we have shown that rather than being solely a response to treatment overload, non-adherence is often an adaptive and rationalised process employed to minimise treatment disruption.

Future developments for BoT research

Whilst we have been able to partially explain the link between treatment burden, disruption and adherence, we have been unable to explore the impact of this non-adherence on relationships with healthcare professionals and family. The data suggest that rationalised non-adherence necessitates further work in the form of concealment, persuasion, and the recruitment of allies among family and health professionals. However, further empirical research is required to explore this work in more depth; to consider what this work entails, the severity of its consequences for relationships with others and its impact on outcomes.

In addition to extending our understanding of the burden of treatment by contributing to a body of work in the field, this review also builds on and adds to recent work on how capacity is expressed by patients. May, Eton [33] suggest that in order for functional performance (the potential to do the treatment work that needs to be done), patients must mobilise resources from their social capital and secure the cooperation of others in their formal and informal social network. Doing so is said to allow patients to develop structural resilience. In other words, patients must adapt in order to absorb, embed and minimise treatment related burden and disruption. This adaptation or ability to absorb adversity is an expression of capacity. Our findings support this view by identifying biographical, relational and biological capacity, and outlining a number of strategies patients use to adapt to disruptions to these components of their capacity. However, we note that these strategies involve work that can generate further burdens and disruptions. This may explain why we also identified that in some cases, functional performance is deliberately sacrificed for quality of life: rationalised non-adherence means ceasing or modifying some of the work of treatment in order to minimise disruption without the additional work of adaptation. Of course, rationalised non-adherence is still a form of adaptation, albeit one of circumvention.