Article Text

How can clinical practice guidelines be adapted to facilitate shared decision making? A qualitative key-informant study
  1. Trudy van der Weijden1,
  2. Arwen H Pieterse2,
  3. Marije S Koelewijn-van Loon1,
  4. Loes Knaapen3,
  5. France Légaré4,
  6. Antoine Boivin5,
  7. Jako S Burgers6,
  8. Anne M Stiggelbout2,
  9. Marjan Faber7,
  10. Glyn Elwyn7,8
  1. 1Department of General Practice, CAPHRI School of Public Health and Primary Care, Maastricht University, Maastricht, The Netherlands
  2. 2Department of Medical Decision Making, Leiden University Medical Centre, Leiden, The Netherlands
  3. 3Department of Social Studies of Medicine, McGill University, Montreal, Canada
  4. 4Department of Family Medicine and Emergency Medicine, Université Laval, Québec, Canada
  5. 5Department of Family Medicine, Université de Sherbrooke, Longueuil, Canada
  6. 6Department of Clinical Practice Guidelines, Dutch College of General Practitioners, Utrecht, The Netherlands
  7. 7Department of IQ Healthcare, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
  8. 8Department of Primary Care and Public Health, Cardiff University, Cardiff, UK
  1. Correspondence to Professor Trudy van der Weijden, Department of General Practice, CAPHRI School of Public Health and Primary Care, Maastricht University, PO Box 616, Maastricht 6200 MD, The Netherlands; trudy.vanderweijden{at}maastrichtuniversity.nl

Abstract

Background To explore how clinical practice guidelines can be adapted to facilitate shared decision making.

Methods This was a qualitative key-informant study with group discussions and semi-structured interviews. First, 75 experts in guideline development or shared decision making participated in group discussions at two international conferences. Next, health professionals known as experts in depression or breast cancer, experts on clinical practice guidelines and/or shared decision making, and patient representatives were interviewed (N=20). Using illustrative treatment decisions on depression or breast cancer, we asked the interviewees to indicate as specifically as they could how guidelines could be used to facilitate shared decision making.

Results Interviewees suggested some generic strategies, namely to include a separate chapter on the importance of shared decision making, to use language that encourages patient involvement, and to develop patient versions of guidelines. Recommendation-specific strategies, related to specific decision points in the guideline, were also suggested: These include structuring the presentation of healthcare options to increase professionals' option awareness; structuring the deliberation process between professionals and patients; and providing relevant patient support tools embedded at important decision points in the guideline.

Conclusions This study resulted in an overview of strategies to adapt clinical practice guidelines to facilitate shared decision making. Some strategies seemed more contentious than others. Future research should assess the feasibility and impact of these strategies to make clinical practice guidelines more conducive to facilitate shared decision making.

  • Clinical practice guidelines
  • Shared decision making
  • Quality improvement

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

There is a purely patient-driven argument for considering patients' preferences in guideline use related to ethical considerations about patient autonomy. Shared decision making (SDM)—an approach that aims to share the evidence of risks and benefits of healthcare options with the patient and to deliberate about patient preferences1 ,2—is advocated in this respect,3 but seldom accomplished.4–6 Clinical practice guidelines (CPGs) are systematically developed to support healthcare professionals, and sometimes patients, in making decisions about appropriate healthcare in specific circumstances. Despite the fact that explanation of pros and cons of all available diagnostic and treatment options including doing nothing—the fundamentals of SDM—is legally prescribed in some countries, it has not been broadly adopted yet in CPGs. Although some exceptions have been described, for example, in French oncology guidelines,7 CPG recommendations are usually stated in the format of a one best option.

Promoting the use of CPGs can, however, unintentionally result in healthcare professionals feeling pressured to adhere to the one-option guideline recommendation and thus inhibit the process of SDM.8 Also, the complexity of current CPGs and related time investment needed to use them can hamper the process of SDM. At the same time, there are many reasons why the recommendations in guidelines may not always be applied uniformly in clinical practice,9 with clinicians following their own routines and preferences, introducing unwarranted practice variation. SDM is particularly important when trade-offs between options strongly depend on individual preferences. This includes recommendations within CPGs for which the evidence is scarce or conflicting or for which there is more than one relevant treatment option that different individuals may value differently.10 These are also known as weak or conditional recommendations in the discourse of evidence-based medicine, the GRADE system.11

CPGs may foster SDM if they help professionals in being well informed, not only about the one best option, but also about other available options, or medically equivalent options in specific decision contexts. It is in this context that we explore how CPGs could be adapted to assist healthcare professionals and patients in SDM.

Lately, there have been initiatives to involve the public and patients in CPG development and implementation, which can improve the consideration of patient preferences and values in the formulation of CPGs.12 ,13 Even when guidelines are designed for more specific patient populations, they necessarily remain a ‘one-for-all’ type of recommendation; they cannot take into account all the individual differences in patient characteristics and preferences. Some authors suggested possible ways forward, including integrating preference-related evidence in recommendations and publishing a patient version of the guideline or a patient decision aid to increase the patient's option awareness.14 ,15 However, these approaches have not yet been systematically explored, let alone structurally embedded in current guidelines.16–19

We belief that CPGs can be adapted to increase the chance of patients becoming involved in medical decision making, for example, by clarifying that options exist and that individual healthcare providers and patients need to consider the risks and benefits carefully in relation to patient preferences. We feel it is more preferable to adapt existing tools than to develop new tools and implementing these independently. This poses less challenge to normalised patterns of working and can facilitate the integration of SDM into existing work streams. To facilitate this next step in CPG development, we present the results of an international key-informant study to identify promising strategies for adapting CPGs for SDM.

Methods

Study design

We performed a two-stage qualitative study consisting of group discussions and semi-structured interviews. The details of the protocol have been described elsewhere.20

Stage 1: group discussions

We organised two structured group discussions to build the interview scheme for the semi-structured interviews in stage 2. Seventy-five international researchers in the fields of guideline development or SDM took part, representing 15 countries. We convened the groups at two workshops; one during the International SDM (ISDM) conference in Boston in 2009 (42 participants) and the other during the Guidelines International Network (G-I-N) in Lisbon in 2010 (33 participants). The group discussions have not been described in the study protocol, as it was not foreseen at the time of writing the protocol that we would be allowed to actually conduct the workshops at international conferences. At these workshops, the participants first individually reported their ideas about how CPGs could be adapted to facilitate SDM on self-adhesive notes (post-its), and then discussed their ideas in small groups. Posters made in the small groups, again with each idea written on a self-adhesive note, were presented and discussed plenary. One of the authors (MK) made field notes of these plenary discussions. After the workshop, we analysed the ideas on the posters and the field notes by concept mapping. This resulted in four clusters of ideas. These were reported back to the participants by email for member checking, resulting in one additional idea. On the basis of the findings we wrote the interview scheme (see online supplementary appendix 1).

Stage 2: semi-structured interviews

We selected a purposive sample of key informants known as opinion leaders from three groups: health professionals using CPGs for depression or breast cancer; developers, researchers, and policy makers specialised in CPGs or SDM; and patient representatives who closely collaborate with relevant national patient and quality improvement institutions, such as the Dutch Institute for Healthcare Improvement. This led to a list of 49 eligible key informants. We purposefully sampled 25 key informants from this list with the aim of covering the three groups in a balanced way, and to reach as much heterogeneity as possible with regard to country and gender. They were invited by email for a 30 min face-to-face or telephone interview: five clinical specialists, five guideline developers (two of whom were members of the G-I-N Patient and Public Involvement Working Group, or G-I-N PUBLIC, one was also a patient representative), three members of the International Patient Decision Aids Standard (IPDAS) Collaboration, seven other specialists in the field of SDM or guideline development, one member of the Cochrane Consumer Group, and four patient representatives of whom two were members of G-I-N PUBLIC and two of the Dutch Patient and Consumer Federation. We purposively included patient representatives who are familiar with CPG development and policy issues on patient participation because of the complexity of the subject. We sent each non-responder one reminder. It was not necessary to invite other key informants in addition to these first 25 because saturation of data was reached—no new relevant knowledge emerged after 12 interviews—after this first round of interviews.

At the start of the semi-structured interviews we asked for the participant's consent for audio recording. We then asked the participant's view of how CPGs can be adapted to enhance SDM. Online supplementary appendix 1 provides the interview scheme. To guide the interviews, we used one of two examples of what we consider to be preference-sensitive decisions: the choice between an antidepressive drug and/or cognitive behavioural therapy to treat a patient with a depression, or mastectomy or lumpectomy with radiotherapy to treat a woman diagnosed with early-stage breast cancer. Three experienced interviewers conducted the interviews. The interviewers took field notes during the interviews and the audiotapes of the interviews were transcribed verbatim.

Data analysis

We analysed the interviews using the qualitative content analysis approach.21 We collected and analysed the data concurrently, incorporating expected and emergent themes and ideas. Then we explored emergent themes and ideas in subsequent interviews. The coders read the transcribed interviews completely several times to become immersed in the data. We started coding in the organising phase with a coding list based on the ideas generated in the workshops. New ideas received new codes. Two independent coders (MK and AP or LK) coded six interviews, which they selected for the rich variety of strategies and suggestions. They discussed the codes until they reached consensus. A single coder (MK) coded all the other interviews on the basis of this coding scheme. We reported back on the strategies to each interviewee as a member check. Corrections and clarifications of these reports did not lead to new strategies. The coders and TvdW and GE contributed to the analyses in iterative comment rounds. We grouped and categorised the strategies into main clusters to report the results.

Results

Twenty key informants from seven countries accepted our invitation for an individual interview. The interviews took about 30 min each. Table 1 shows the characteristics of the key informants.

Table 1

Characteristics of the N=20 key informants

We classified the ideas about how the CPGs could be adapted to facilitate SDM into generic strategies—strategies that were related to the entire guideline document—and recommendation-specific strategies—strategies that were related to one specific recommendation. We also identified whether strategies were aimed at professionals or at patients. Figure 1 shows the classification.

Figure 1

Classification of strategies of how clinical practice guidelines  can be adapted to facilitate shared decision making. CPG, clinical practice guideline; SDM, shared decision making.

Generic strategies to facilitate SDM, not related to one specific recommendation

The generic strategies are typically related to the entire guideline document and not to one specific recommendation. These may include the addition of a separate CPG chapter on the importance of SDM; using language that encourages patient involvement throughout the whole CPG document and throughout the communication in the clinical pathway; and the creation of a patient version of the CPG. We detail each of these strategies below:

Separate chapter

The interviewees suggested three generic approaches to facilitate SDM that could be described in a separate CPG chapter. First, professionals could be generally alerted to the importance of involving the patient in decision making. For example, the CPG could describe the value of SDM and of patient centeredness, or might provide relevant variables about reasons for adherence or non-adherence to recommendations and their correlation to disease characteristics (history of the disease, comorbidity), patient socio-demographic characteristics (age, gender, educational level, socio-economic status, family environment, culture, religion), and the health system context. Second, the necessary and facilitating conditions for SDM at the micro (professional–patient interaction), meso (healthcare team) and macro (organisation) level could be described. At the micro level, for example, the CPG could explain that discussing patient preferences should be timely and be tailored to patient characteristics (eg, to patient's health literacy and numeracy level, to his/her illness perceptions) and that a follow-up consultation should be offered when desired. Third, the CPG could contain examples of patient-centred questions for suggesting ways to elicit the patient's preferences, such as: ‘How can I help you to improve your quality of life?’, ‘What is important to you?’, or ‘How do you see this decision?’.

Language

The CPG could use wording that makes the patient's involvement in decision making explicit, for example by phrasing ‘offer the patient a statin prescription’ instead of ‘prescribe statins to the patient’ in a guideline on cardiovascular risk management. Also, the CPG could encourage professionals to use the same plain language they use with patients in referral letters to colleagues, to facilitate patient-centred communication during the entire clinical pathway.

Patient version

A patient version of the CPG could be provided to promote patient involvement in decision making. Interviewees suggested different formats. Patient versions could be entire CPG documents translated into lay terminology that would be useful to professionals and patients. Alternatively, patient versions could be short documents focusing only on specific recommendations and explaining the decision and related diagnostic and treatment options in lay terms. This design closely resembles the structure of fact sheets or summaries of CPGs in table format. Additionally, patient versions could suggest generic questions that patients can ask to their provider, such as ‘what are my options, what are the pros and cons per option?’ and ‘how likely are these?’. What we do as well … in each of the versions for patients, is that we include some questions that people might want to ask their health professionals … That's one of the most powerful ways you can encourage shared decision making. (Interviewee 6)

Recommendation-specific strategies to facilitate SDM

Table 2 lists the strategies related to a specific recommendation. They can be categorised into three clusters: those that improve the presentation of options to increase option awareness of healthcare providers; those that describe the deliberation process to improve deliberation about options; and those that provide patient support tools to facilitate patients' option awareness, elicit their preferences and/or foster deliberation. We detail each of these strategies below.

Table 2

Recommendation-specific strategies for promoting shared decision making

Cluster 1: structuring the presentation of healthcare options

This cluster contains five strategies to structure the presentation of the options relating to a specific decision point to increase option awareness among healthcare providers. First, more details about decision options should be reported (table 2, strategy 1.1). This includes explicitly presenting the option of doing nothing. Also, information about the benefits and harms of each option in terms of patient-oriented outcomes, such as effect on survival and treatment burden, should be provided. If available, outcomes that patients consider especially relevant, such as quality of life, the duration of hospitalisation, the number of revisits, or recovery time, should also be specified. Short-term and longer-term follow-up data should be given. Reasons for grading a recommendation as conditional should be given. This is especially important if a recommendation with high-level evidence is downgraded, according to GRADE, due to large variation in patients' preferences. The assumed values on which the recommendation was based should be made explicit. Try to make sure that it goes beyond the typical risks of metastasis, risk of recurrence, and risk of death or survival. And try to get at some of the other things, you know … “convenience”. How often do you have to go back in for a follow-up visit when you have had either surgery or radiation or whatever treatment. Getting at some of the other dimensions that will affect [the decision], whether patients have a car, have access, have to deal with babysitting, that kind of thing. (Interviewee 8)

Second, evidence about patients' preferences and patients' actual decisions and important attributes of the decision could be displayed (strategy 1.2). Both qualitative and quantitative studies could generate this evidence. Also, evidence for professionals' biases could be displayed to show what professionals think patient prefer or would decide compared with actual patient preferences and decisions. Third, illustrations of how patients deliberate, for example using narratives, could be powerful tools to demonstrate the importance of patient involvement in decisions (strategy 1.3). A vignette describing how a patient and a provider discuss the options to reach a shared decision can be used. Stories are often more powerful in getting the message across … I would choose an ordinary patient, who for good reason, opts for something other than the first recommended option, say, for specific reasons because of his family life, work, or social environment. (Interviewee 3)

Fourth, a ‘second best’ or alternative option, such as watchful waiting, could explicitly be included even for strong recommendations (strategy 1.4). Finally, recommendations could be flagged to indicate those decisions for which elicitation of patient preferences is urgent (strategy 1.5). In a guideline there should be, like, you know, a star, a symbol, reminding the clinician that, at this special point of the clinical reasoning or clinical decision, it is very important to share this part with the patient. (Interviewee 12)

However, some interviewees objected to this latter strategy on principle. To them, every recommendation and decision should be deliberated with the patient because the patient always has a choice, as illustrated by the following quote: Each intervention that the clinician proposes … The advice of anti-allergenic measures in your house because of your allergy must have the same value as advice for surgery. And we just have to start with the uncomplicated things. That will create a good relationship between the physician and the patient. (Interviewee 17)

Others pointed out that opinions on whether or not a decision is preference sensitive or not can vary and that it is unclear which recommendations should be flagged and how. In other words, ‘preference sensitivity’ is not a discrete and universal category that can easily be designated within a guideline to be used in all situations for all patients. The interviewees offered the following examples of preference-sensitive decisions:

  1. When options have very different benefit and harm profiles and implications in terms of delivery mode, for example, and the specific trade-offs between benefits, harms and implications can lead to large inter-individual variability.

  2. When options are generally equally attractive or when their outcomes are generally equally desirable.

  3. When there is insufficient evidence about the risks and benefits of one or more options.

  4. When the potential risks of the option are high regardless of its benefits (there is an impressive (low) number needed to harm, even though the number needed to treat is also low).

  5. When the effect of the option depends on the patient's adherence or if the option intervenes with the patient's lifestyle, even in case of strong recommendations according to GRADE.

Cluster 2: structuring the deliberation process between professionals and patients

This cluster of strategies aims at facilitating the deliberation between providers and patients about recommendations. Most strategies focus primarily on professionals and one primarily on patients. First, a CPG could describe who does what in guiding the patient through the deliberation process (table 2, strategy 2.1). This includes describing the delegation of deliberation tasks to specific health professionals. The CPG must state that, when the diagnosis is made, a multidisciplinary team of physicians must carefully weigh a treatment plan, and in my opinion, social workers or psychosocially educated people should pay attention to what kind of patient we have here. These people can also stop the [decision making] process when someone needs extra help during the process. (Interviewee 18)

Also, the CPG could recommend offering the patient a copy of the personalised treatment plan (strategy 2.2) to give the patient insight into the treatment and the available options at different stages for a specific decision point. The CPG could further recommend the professional to encourage the patient to engage a proxy in the deliberation, such as a partner, parent, another family member or friend (strategy 2.3). The CPG could provide specific questions for the professional that he or she can use to question patients about their values to prepare them for the deliberation (strategy 2.4). In addition, the CPG could refer to relevant patient decision aids that may be published independently of the CPG and indicate at what times these could be used during the deliberation process (strategy 2.5). The CPG could also provide a list of topics and arguments that the healthcare providers should at least discuss with the patient (strategy 2.6). Possibly, professionals should be encouraged to use such a list sensibly, as illustrated by the following quote: Prophylactic mastectomy is such an important decision that you want to discuss all the pros and cons before the actual decision is made. I do envision the topics; you could provide a quite complete list of topics that you should discuss with the patient … There is a danger of cookbook medicine: if I complete the topic list, I have completely adhered to the guideline recommendation. But if this is done unemotionally and without interpretation and feeling for the context, it doesn't make sense. (Interviewee 4)

The CPG could further describe stepwise communication plans tailored to subgroups of patients (strategy 2.7) based on patient characteristics, such as clinical symptoms or health literacy. To apply the appropriate communication plan, professionals would first need to determine to what subgroup the patient belongs to. A final strategy aimed primarily at professionals is to define performance indicators and benchmarks for the decision-making process (strategy 2.8). This strategy would facilitate the evaluation and subsequent improvement of decision-making processes, but would also possibly encourage the use of strategies and tools to facilitate SDM. One of the ways in which it [implementation] can be done is, if you can frame the decision aid, the use of the decision aid as a quality parameter for instance … it becomes something that people can be held accountable for, right? So that may provide some impetus for implementation. (Interviewee 9)

Some interviewees objected to the general idea of including strategies recommending a deliberation process or specifying particulars of such a deliberation process: We can't prescribe what needs to be done for everybody. The tools that are effective for one person are not effective for another. So I think dictating that you have to use particular tools is wrong. (Interviewee 14)

Cluster 3: providing decision support tools within the CPG

Three strategies aimed at patients were suggested which all related to providing, or closely linking, to relevant patient decision support tools for specific decisions. First, patient support tools such as risk communication tools, patient values clarification tools, and complete patient decision aids could be integrated in specific recommendations (table 2, strategy 3.1). Risk communication tools, for example population charts and pictograms, were seen as helpful in presenting probabilities, including absolute risks and absolute and relative risk reduction data. The tools could be developed in close collaboration with guideline developers and be fully integrated within the CPG, or be linked to the CPG while remaining separate documents. Keeping the tools separate from the actual CPG can be advantageous: One is just about the length of the guideline … anything that I think adds to what looks like the guideline may be off-putting to practitioners … Having a separate document may make your guideline slightly less bulky. And the second thing is that I think that, in day-to-day practice, it is relatively unusual for practitioners to go and look at guidelines … and some of these decision tools are things that patients may want to take home with them. (Interviewee 13)

Second, the specific CPG recommendation including the relevant patient decision aid could be incorporated into electronic health records to give patients access to online decision aids when needed, to register the patient's preferences in the health record, and to have these preferences available for SDM in follow-up visits (strategy 3.2). Third, a decision tool could be delivered that gives insight into the current quality of care for the specific preference-sensitive recommendation, using an overview of scores on relevant performance indicators and benchmarks. This tool could make patients aware of conditions that may facilitate SDM, for example, whether there is a breast cancer nurse in the hospital (strategy 3.3).

Discussion

To foster SDM we hypothesised that it could be a more fruitful approach to adapt CPGs, rather than developing new tools and implementing these independently. Our qualitative key-informant study demonstrated that opinion leaders on guideline development and on SDM could propose a variety of strategies to adapt existing CPGs that should facilitate SDM. Some strategies that were suggested related to the complete CPG document, such as carefully choosing language, while other strategies related to specific recommendations which are included in the CPG. This article summarises the views of these key informants from a variety of backgrounds and including patient representatives on strategies for the adaptation of CPGs to facilitate SDM.

Strategies affecting complete CPGs, or generic strategies, included alerting professionals to the necessity of and best conditions for SDM, ways in which to elicit patient preferences, and promoting the use of language that encourages patient involvement. Generic strategies primarily aimed at patients related to developing patient versions of CPGs. Recommendation-specific strategies were also primarily aimed at professionals or patients. Specific strategies aimed at professionals focused on raising awareness of multiple options by providing more complete information about options and providing support for the process of deliberation. Specific strategies aimed at patients mainly focused on increasing their understanding of options and helping them to clarify their preferences.

Our study resulted in an extensive list of strategies, of which some seemed to be more contentious than others. Clearly, our study did not aim at achieving consensus, but it appeared that opinions especially diverged regarding strategies to alert professionals that specific decisions are preference sensitive and those that affect the deliberation process. Increasingly it is acknowledged that individual preferences are relevant to many, if not all, health-related decisions. Also, how best to support the deliberation process is not yet known.22

As far as we know, this is the first study of pragmatic strategies to make CPGs more supportive of SDM. In a recent knowledge synthesis of existing programmes to involve patients and the public in CPGs that was focused on the development and implementation of CPGs, no strategies were found to use CPGs to promote the involvement of individual patients in clinical decision making.12

We assured the credibility of this study by using multiple methods of data collection, such as group discussions, semi-structured interviews and member checks.23 To validate the findings, interested professionals with different backgrounds and nationalities were involved, and data collection continued until saturation was reached. Five researchers analysed the data to assure investigator triangulation. Saturation was found after 12 interviews. The number of patient representatives was limited and they were all purposively recruited from patient representatives who are familiar with policy matters. It is possible that a larger and more diverse patient interview sample would have offered different ideas. Another limitation is that the participants were all from Anglo-Saxon background, so we did not cover cultural diversity perspectives.

We are aware that our study focused only on one level for change: how CPGs can be used to facilitate SDM. For the successful implementation of change, we recognise that we must also target other levels of the healthcare system, such as the individual professional, the patient, and the social, organisational, political and economic contexts.24 Even the design of patient studies can be changed to facilitate the implementation of SDM. For example, details of treatment options that seem to be important in decision making are often not captured by the outcomes of current randomised controlled trials.25 However, we think that CPGs are an important part of the implementation of SDM. Their use is expected to increase with time, and they are the basis of many educational, internal and external accountability activities for healthcare professionals. Although we focused our interviews on preference-sensitive decisions for depression and breast cancer, many interviewees came up with their own examples. Therefore, we think that the strategies we collected can be useful for other preference-sensitive decisions as well.

This study describes strategies to adapt CPGs that can facilitate SDM between healthcare professionals and patients and thus to facilitate the process of involving patients in medical decision making. Guideline developers, including patient representatives in guideline development groups, can use our overview of pragmatic strategies to generate ideas on how to adapt CPGs to facilitate SDM, although further research is needed to assess the effectiveness and feasibility of the proposed strategies.

Acknowledgments

We would like to thank the interviewees for their contributions.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors MK, the project coordinator, conducted eight interviews and was involved in the group sessions, developing the interview scheme, coding and analysis of all interviews. She also wrote the draft paper. GE was coapplicant. He contributed to the methods and the interview scheme. He advised in the analysis phase and assisted in writing as a coauthor. LK conducted seven interviews and took part in designing the study, developing the interview scheme, selecting the interviewees, and coding the interviews. She also commented on the draft of this paper. AP conducted five interviews, took part in coding the interviews, and commented on the draft of this paper. FL, AB, JB, AS and MF were involved in writing the research protocol and commented on the draft of the paper. TW was the project leader and the applicant. She was involved in all phases of the study other than interviewing and coding. All authors have approved the final version of the paper to be submitted.

  • Funding The Netherlands Organisation of Health Research and Development (ZonMW) funded this study (grant number 150040004).

  • Competing interests Dr Arwen Pieterse is currently employed and has received grants from the Dutch Cancer Society. Dr Loes Knappen is currently employed at the McGill University and has received grants from Provincial and Federal funding agencies (FRSQ, CIHR).

  • Ethics approval The Maastricht Medical Research Ethics Committee approved the study.

  • Provenance and peer review Not commissioned; externally peer reviewed.