Identifying ways to improve diabetes management during cancer treatments (INDICATE): protocol for a qualitative interview study with patients and clinicians

Introduction A large and growing number of patients with cancer have comorbid diabetes. Cancer and its treatment can adversely impact glycaemic management and control, and there is accumulating evidence that suboptimal glycaemic control during cancer treatment is a contributory driver of worse cancer-related outcomes in patients with comorbid diabetes. Little research has sought to understand, from the perspective of patients and clinicians, how and why different aspects of cancer care and diabetes care can complicate or facilitate each other, which is key to informing interventions to improve diabetes management during cancer treatments. This study aims to identify and elucidate barriers and enablers to effective diabetes management and control during cancer treatments, and potential intervention targets and strategies to address and harness these, respectively. Methods and analysis Qualitative interviews will be conducted with people with diabetes and comorbid cancer (n=30–40) and a range of clinicians (n=30–40) involved in caring for this patient group (eg, oncologists, diabetologists, specialist nurses, general practitioners). Semistructured interviews will examine participants’ experiences of and perspectives on diabetes management and control during cancer treatments. Data will be analysed using framework analysis. Data collection and analysis will be informed by the Theoretical Domains Framework, and related Theory and Techniques Tool and Behaviour Change Wheel, to facilitate examination of a comprehensive range of barriers and enablers and support identification of pertinent and feasible intervention approaches. Study dates: January 2021–January 2023. Ethics and dissemination The study has approval from National Health Service (NHS) West Midlands—Edgbaston Research Ethics Committee. Findings will be presented to lay, clinical, academic and NHS and charity service–provider audiences via dissemination of written summaries and presentations, and published in peer-reviewed journals. Findings will be used to inform development and implementation of clinical, health services and patient-management intervention strategies to optimise diabetes management and control during cancer treatments.


INTRODUCTION
A large and growing number of patients with cancer have comorbid diabetes The incidence and prevalence of diabetes mellitus are high and increasing worldwide. 1 2 In high-income countries, it is estimated 87%-91% of all people with diabetes have type 2 diabetes, 7%-12% type 1 and 1%-3% other rarer types of diabetes. 1 In the UK, it is estimated that 4.8 million people are living with diabetes, projected to increase to 5.3 million by 2025. 3 Type 1 and type 2 diabetes are associated with an increased risk of cancer, including, collectively, cancers of the breast, colorectum, endometrium and oesophagus. [4][5][6][7] In the UK, one in two people Strengths and limitations of this study ► Largest, and first UK-based, qualitative study to date of patients' and clinicians' views on barriers and enablers to effective diabetes management during cancer treatments. ► Most in-depth qualitative examination of this topic to date, with a focus on both barriers and enablers, and ways to address and harness these, respectively, for both patient and clinician diabetes management during cancer treatments. ► Will extend previous research by interviewing a wider range of clinicians (crucially including diabetologists) and considering other cancer treatments in addition to chemotherapy. ► Data collection and analysis will be informed by the Theoretical Domains Framework, and related Theory and Techniques Tool and Behaviour Change Wheel, to facilitate examination of a comprehensive range of barriers and enablers and support identification of feasible intervention approaches. ► Recruitment is limited, in the main, to just two sites, both Yorkshire based.
Open access now develop cancer in their lifetime. 8 Figures on the prevalence of pre-existing diabetes in newly diagnosed patients with cancer vary (eg, by country, cancer type) but studies typically report rates ranging from 10% to 20%. [9][10][11][12] .
Cancer and its treatment can adversely impact glycaemic management and control For people with diabetes, maintaining good glycaemic control can be a significant challenge. 13 14 This challenge may be exacerbated by cancer and its treatment, which has high potential to complicate diabetes management and glycaemic control. The psychosocial sequelae of a cancer diagnosis (eg, distress, anxiety, depression [15][16][17] and the side-effects of some cancer treatments (eg, vomiting, fatigue, pain, 18 19 ) could impede diabetes management behaviours such as healthy eating and blood glucose monitoring. Furthermore, some cancer and supportive treatments can directly impact blood sugar levels increasing the risk of hypoglycaemic and hyperglycaemic episodes (eg, somatostatin analogues, high-dose steroids). Though results are mixed, studies collectively indicate a deleterious effect of cancer and its treatment on diabetes management and control, which for many people is suboptimal even before diagnosis of cancer. 13 14 Research shows that during cancer treatment many people with diabetes have reduced adherence to diabetes medications and self-care behaviours that contribute to glycaemic control (eg, blood glucose monitoring, eating healthily, exercising), and have poorer glycaemic control (eg, increased HbA1c levels, diabetes treatment escalations), than pre-cancer diagnosis. 20 21 Studies have also noted that, following a cancer diagnosis, some people with diabetes undergo less diabetes-related screening aimed at mitigation of diabetic complications (eg, retinal screening, low-density lipoprotein tests). [22][23][24] Suboptimal diabetes management and control during cancer treatment is associated with worse outcomes Some studies have found patients with cancer with preexisting diabetes are more likely to experience toxicities and complications (eg, infections) during cancer treatment, [25][26][27][28][29][30][31] which can result in costly hospitalisations and compromise treatment completion. Moreover, numerous studies show that, compared with other patients with cancer, those with pre-existing diabetes have higher perioperative and longer term mortality. 6 25-27 30-34 Though findings are not uniform, there is accumulating evidence that suboptimal glycaemic control during cancer treatment is a contributory driver of worse cancer-related outcomes in patients with comorbid diabetes. 28 35-41 Retrospective studies of patients with cancer with comorbid diabetes have shown, for example, that good perioperative glycaemic control is associated with reduced risk of morbidity and death following colectomy for colon cancer 35 and good glycaemic control prior to neoadjuvant chemotherapy for cervical cancer is associated with superior tumour response and survival. 37 A small 12-week prospective study of patients with cancer with pre-existing diabetes found suboptimal glycaemic control at chemotherapy outset predicted increased risk of developing an infection, hospitalisation and chemotherapy reduction or discontinuation. 41  Also, current studies are exclusively USA based, and findings may to some degree be context specific, given differences in the organisation and financing of heathcare systems globally, even within higher income countries. We aim to extend, and address some of the limitations of, this previous qualitative work.
Our study aims to extend previous qualitative work in this area and help inform intervention development The current qualitative interview study aims to identify and elucidate challenges and enablers to diabetes management and control during treatment for cancer, based on the experiences and perspectives of people with diabetes and comorbid cancer and healthcare professionals involved in their care. To facilitate examination of a comprehensive range of individual-level and servicelevel barriers and enablers, and to support the identification of potential pertinent intervention approaches to address and harness these, respectively, we will use the Theoretical Domains Framework (TDF), 48 49 and related Theory and Techniques Tool (TTT) [50][51][52][53] and Behaviour Change Wheel (BCW), 54 55 to inform data collection and analysis, as detailed in the Methods section.
Research questions 1. What are the patient-perceived challenges and enablers to effective self-management and control of diabetes during cancer treatments? 2. What are the clinician-perceived challenges and enablers to effective clinical management and control of diabetes during cancer treatments? 3. What are patients' suggestions for ways to support and improve self-management and control of diabetes during cancer treatments? 4. What are clinicians' suggestions for ways to support and improve clinical management and control of diabetes during cancer treatments? 5. What are potentially promising intervention targets and strategies for consideration in future research to optimise patient and/or clinician management and control of diabetes during cancer treatments?

METHODS AND ANALYSIS Participants
Eligibility criteria are detailed in table 1. We will include patients with type 1 or type 2 diabetes, and clinician interviews will enquire about differences between these patient groups. This will enable us to examine in the analysis similarities and differences in patient and clinician reported challenges and enablers to diabetes management during cancer treatments on the basis of diabetes type, and thus help inform to what extent different future interventions in this area could address both diabetes types or may need to target one or both of the types separately. We will restrict recruitment to comorbid breast, prostate or colorectal cancer; in the UK, these are three of the four most common cancers and the largest survivor groups. 56

Recruitment sites and procedures Patients
Hospital-based recruitment Patients will be primarily recruited from cancer centres in two Yorkshire-based National Health Service (NHS) Research Network support, will identify eligible patients and first approach them about the study, providing a patient information sheet. Interested patients will contact our research team directly, or, if a patient requests it, the clinical team will pass onto us patient-provided contact information and we will initiate correspondence.

Community-advertisement recruitment
We will also recruit patients via an advertisement flyer calling for people aged 18+ 'with diabetes (type 1 or type 2), who are being treated for breast, bowel or prostate cancer, or have been in the past 3 years' to 'tell us about your experiences of managing diabetes during cancer treatments'. The flyer will be disseminated via social media (eg, Twitter and Facebook accounts of the research team members), and by relevant willing UK-based charities and organisations (eg, in newsletters). Patients who see the flyer and are interested will contact our research team directly.

Clinicians
Hospital-based recruitment Clinicians will be recruited from the cancer centres, as well as other relevant hospital departments and specialties (eg, endocrinology, anaesthesia, pharmacy, dietetics) within the participating hospitals. Our research team includes oncology and diabetes clinicians working at the participating hospital trusts. Potentially eligible clinicians will be identified and emailed a staff information sheet by a member of the research team, or by a gatekeeper colleague who has agreed to disseminate study information. Eligible and interested clinicians will contact the research team directly.

Primary care recruitment
We will also recruit general practitioners (GPs) and practice nurses working at general practices in Yorkshire. GPs and practice nurses will be informed about the study by local primary care R&D teams who are willing to disseminate study information (eg, in a CCG-newsletter), or will be emailed a staff information sheet by a member of the research team (eg, GPs with a part-time academic post known to the research team). Eligible and interested clinicians will contact the research team directly.

Sample size and sampling strategy
We will recruit and interview 30-40 patients and 30-40 clinicians. Based on our experience, this sample size will enable adequate sample diversity on key participant characteristics and allow us to reach sufficient data saturation. We aim to recruit comparable numbers of people with breast, prostate and colorectal cancer (n≈10-13 each cancer type), with both diabetes types represented in each cancer subgroup (2-3 patients in each cancer subgroup with type 1 diabetes, which accounts for 7%-12% of all diabetes cases in high-income countries 1 ). We also aim for some diversity in the sample as a whole with regard to types of cancer treatment, sociodemographic characteristics (gender, age, ethnicity), and the extent of experienced difficulties with diabetes management during cancer treatments. We aim to recruit clinicians from a wide range of relevant professions and specialities (eg, oncologists, surgeons, diabetologists, specialist nurses, dieticians, GPs) and for some diversity in the sample as a whole with regard to professional seniority. The composition of the sample will be monitored during recruitment and, if necessary and possible, targeted recruitment of underrepresented groups will be undertaken.
Theoretical framework informing data collection and analysis We will use the TDF 48 49 and related resources to inform data collection and analysis. The TDF synthesises key constructs in numerous theories of behaviour and behaviour change, and thus 'provides a theoretical lens through which to view the cognitive, affective, social and environmental influences on behaviour' (p2). 57  MoAs, which include the 14 TDF domains and the 12 most frequently occurring MoAs which did not overlap with these identified in a review of behaviour change theories. 52 The BCW 54 55 is a 'theory and evidence based' intervention development approach 58 that provides guidance on considering and identifying the function of interventions (eg, education, persuasion, enablement) and policy categories that may support the delivery of these functions (eg, guidelines, service provision, environmental planning) and how these both link to MoAs and/or BCTs. The TDF and related resources have been used to inform the development and evaluation of health-focused behavioural interventions, 59-61 including informing the data collection and/or analysis of qualitative studies forming an early stage in the process of intervention development. [62][63][64][65] Data collection: interview content and procedures Participants will take part in one semistructured qualitative interview lasting around 45 min, though duration is likely to be variable depending on how much a participant wishes to say and their preferred interview pace.

Open access
Participants can choose their interview date/time and mode (eg, telephone, videocall, in person) provided arrangements adhere to current relevant government and workplace rules around COVID-19 social distancing. Interviews will examine participants' experiences of and perspectives on diabetes management and control during cancer treatments. Early interview questions will enquire about key sociodemographic characteristics and relevant clinical (eg, diabetes and cancer type and treatments) or professional (eg, job-title, workplace) details. Interviews will seek to identify and elucidate patient-perceived and clinician-perceived challenges and enablers to effective diabetes management and glycaemic control in the context of cancer diagnosis and treatment, and ways to address and optimise these, respectively. Interview guides were developed, informed by: (1) previous research [44][45][46][47] ; (2) advice and feedback from the study patient and public involvement (PPI) and steering groups; and lastly (3) the MoA covered by the TTT 50 which, as previously discussed, includes the TDF domains. We were mindful interviews do not become dominated by examining MoAs, especially given multiple behaviours are involved in managing diabetes (eg, prescribing medications, following dietary advice, blood glucose self-monitoring) and to systematically examine each MoA in relation to each different behaviour that may be discussed would make for an overly long, repetitive, and granular-level interview. Also, we did not want interviews to be restricted to examining only the MoAs, thereby potentially overlooking other influential factors. Thus, in an approach consistent with recent recommendations for using the TDF in qualitative studies, 66 the interview guides contained at least one question or follow-up question likely to encourage discussion of challenges or enablers relevant to each MoA, rather than a highly structured list of MoA-focused questions, one per MoA per different behaviours. Pilot interviews were undertaken with three clinician coapplicants and three PPI group members, to refine the interview guides and hone interviewer technique. Box 1 shows sample questions from the patient and clinician interview guides. Interviews will be audio recorded.

Data analysis Primary
The interview data will be analysed to identify and elucidate patient and clinician perceived challenges and enablers to self and clinical diabetes management and control during cancer treatments, and suggested ways to overcome and optimise these respectively. We will use the framework method, 67 guided by the stages and recommendations set out by Gale et al 68 : (1) verbatim transcription; (2) familiarisation with the data; (3) coding; (4) developing a working analytical framework; (5) applying the framework; (6) charting data into a framework matrix (a summary of the data by analytic code/category per participant); and (7) interpreting the data. Analysis will overlap with data collection. To ensure rigour, analysis will be an iterative, collaborative process led by the core  63 69 70 with input and feedback from other members of the research team and the PPI and steering groups at key points. We will use QSR-NVivo software to support analysis, and document substantive decisions during the analytic process. Coding will use both deductive codes (ie, based on the research questions, interview guides) and inductive codes (ie, based on reading a sample of transcripts), which will be generated through iterative rounds of independent work and subsequent group discussion. In line with recent recommendations for using the TDF in qualitative studies, 66 we do not intend to include the MoAs as a priori codes, but to subsequently consider inductively generated findings about challenges and enablers against the MoAs (see below). This will guard against overlooking challenges and enablers discussed in the interviews not covered by the TTT and, as the TTT includes a substantial number of MoAs (n=26), guard against other key focuses of the analysis being overshadowed (eg, interviewees suggestions for tackling challenges). A working analytical framework will be iteratively developed through agreeing by consensus on a final set of codes, and their initial organisation into categories, informed by input from the PPI group and wider research team. Members of the PPI group will each read a different sample of transcripts and provide coding-relevant feedback (eg, feedback on sections that stood out to them, on barriers and enablers discussed in the interview). Other members of the research team and steering group will be asked to review and feedback on a draft(s) of the working analytical framework. The analytical framework will be applied to all interview transcripts, and subsequently the data will be charted into a framework matrix, including references to illustrative quotations. A proportion of transcripts will be double coded and double charted (≈10% at both stages), and compared and discussed, to ensure consistent application of the framework and data charting. We will identify and develop themes and subthemes pertinent to the research questions, using the matrix to facilitate comparison within and between codes/categories and participants, and thus the identification of patterns and deviant cases in the data, including on the basis of diabetes and/ or cancer type and/or treatments. Members of the PPI group, wider research team and steering group will input into data interpretation by reviewing and providing feedback on the matrix and a related working draft(s) of themes and subthemes. We envisage the patient and clinician data will initially be coded and charted separately, but subsequently synthesised as much as possible during interpretation and theme development. However, decisions about this aspect of the analysis will be made during analysis, after familiarisation with the data.

Secondary
In a second phase of analysis we will: (1) map the challenges and enablers identified in the framework analysis phase to the MoAs; (2) prioritise MoAs for targeting in interventions; and (3) determine pertinent intervention functions and BCTs to deliver these. We will map to the MoAs using current definitions, 50 and by independent work followed by group discussion to achieve a consensus. Target MoAs will be prioritised through consensus discussion, on the basis of the mapping results, the contextualised understanding of barriers and ways to address these provided by the framework analysis of the interview data, and views of the study PPI group and wider research team and steering group (which include multidisciplinary clinicians). For each prioritised MoA, we will determine, through consensus discussion, potential intervention functions and policy categories using the BCW 54 55 and pertinent BCTs using the TTT. 50 These processes are consistent with those undertaken in previous qualitative studies, including by members of our team, which have similarly sought to understand, and inform interventions to address, healthcare challenges (eg, gestational diabetes 62 ; deprescribing 63 ; smoking in pregnancy 65 ).

Patient and public involvement
Six PPI representatives, with personal experience of diabetes, cancer or both conditions as a patient or carer, were involved in developing this research and helped shape the design and methods of the study. One of these PPI representatives was a coapplicant on our grant application for funding and is a coauthor of this protocol paper. We have established a study PPI group with six members, one of whom, coapplicant and protocol coauthor MM, was involved in the previous study development -what about resources/recording and reporting systems/workflows and processes/culture? Example cross-topic follow-up questions ► is that something you know how to do? ► is that something that would be easy or difficult to do? ► is that something you feel confident to do? ► is that something you always/usually/often do? ► how would you know if that had been done? ► and do you think your colleagues tend to think/do similarly? ► what would need to happen for you to do that/be able to do that?
Open access stage. As well as PPI-group meetings, MM will also attend study steering group meetings, helping ensure effective communication between these two groups. The PPI group will collaborate, advise and feedback on all stages of the research including: design and piloting of the interview guides; data analysis including coding and interpreting the data; and dissemination outputs including lay summaries, presentations and journal papers. We will discuss with PPI group members the research activities they wish to be involved in, their relevant prior experience (if any), and therefore what training they may require (eg, practising data coding using transcripts from the pilot interviews), and arrange as and when appropriate. We will be guided by the UK Standards for Public Involvement, 71 and PPI representatives will be paid following UK National Institute for Health Research guidance 72 and reimbursed any expenses.

ETHICS AND DISSEMINATION Approvals and ethical considerations
The study has approval from the NHS West Midlands-Edgbaston Research Ethics Committee (20/WM/0310), NHS Health Research Authority (IRAS-ID:276694) and the Leeds Beckett University Psychology Research Ethics Committee. Participants will receive written study information which will include contact details of organisations providing diabetes-related and cancer-related information and support (eg, Macmillan helpline). Consent will be obtained in writing or verbally, depending on participant preference; verbal consent will be recorded immediately prior to the interview and stored on a separate audio file to the interview. Participants will be informed of their right to choose, without needing to provide a reason why, to not answer an interview question(s), to take a break or stop the interview at any time, and to request withdrawal of any/all of their data. Research and personal data will be handled and stored confidentially and securely in line with government and Leeds Beckett University data protection requirements and guidelines. We will protect participant anonymity and make unidentifiable all illustrative quotes used in reports of the findings. Participants will not be offered any incentives.

Dissemination
Findings will be presented to lay, clinical, academic and NHS and charity service provider audiences via dissemination of written summaries and articles, infographics and presentations. Findings will be published in a peerreviewed journal(s) following Consolidated criteria for Reporting Qualitative research and Standards for Reporting Qualitative Research reporting guidelinesonline supplemental file 1 . 73 74 Study data may be made available on reasonable request to the corresponding author. This UK-based study will be the first published study of its kind outside the USA, and will provide the most in-depth qualitative examination of this topic to date, with a focus on both barriers and enablers, and ways to address and harness these, respectively, for both patient and clinician diabetes management during cancer treatments. Our study will extend previous work by also interviewing a wider range of clinicians, crucially including diabetologists, and considering other cancer treatments in addition to chemotherapy. Furthermore, while previous work in this area has not been theoretically informed, at least explicitly, we will use the TDF, TTT and BCW [48][49][50][51][52][53][54][55] to facilitate examination of a comprehensive range of barriers and enablers and support identification of pertinent and feasible intervention approaches. There is increasing interest worldwide in improving glycaemic control during cancer treatments, evident in the recent development of interventions for people with diabetes having treatment for cancer, such as clinical guidelines in UK, 75 new integrated care pathways in Italy, 76 and a clinical pharmacy intervention and counselling programme in Turkey, 77 though intervention research in this area is in its infancy. Findings of this study will be used to inform development and implementation of clinical, health services and patient-management intervention strategies to optimise diabetes management and control during cancer treatments.

Timeline
Data collection commenced February 2021 and is projected to close August 2022.
Weston Park Cancer Centre, Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, UK Twitter Laura Ashley @Dr_Laura_Ashley, Lisa Kidd @lisakidd22 and Frances Mair @FrancesMair Contributors LA conceived the study, led on its development and design, and is chief investigator and lead applicant of the grant funding this study. SK, IK, LK, FM, MM, DS, JT, GV, and JW contributed to the development of the study design and protocol and are coapplicants of the funding grant. MP is the appointed study research assistant and will undertake the interviews. The roles of team members in analysis are specified in the manuscript. LA drafted this manuscript and all coauthors reviewed, provided feedback and approved it for submission. Disclaimer The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Patient consent for publication Not applicable.
Provenance and peer review Not commissioned; peer reviewed for ethical and funding approval prior to submission.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/ licenses/by/4.0/.