Patient preferences and current practice for adults with steroid-resistant ulcerative colitis: POPSTER mixed-methods study

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 17/72/02. The contractual start date was in November 2018. The draft report began editorial review in May 2021 and was accepted for publication in March 2022. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Disclaimer

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

Copyright © 2022 Coates et al. This work was produced by Coates et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Coates et al.

Ulcerative colitis

Rationale

Ulcerative colitis runs a relapsing and remitting course, resulting in debilitating symptoms, impaired quality of life and severe attacks necessitating hospitalisation. For moderately active UC not requiring hospitalisation, oral corticosteroids may induce remission in those refractory to aminosalicylate therapy. 21 Corticosteroids are recommended as first-line therapy for a severe relapse of UC. 22 This programme of work applies to patients with moderately severe UC treated as outpatients.

Approximately 50% of patients with moderately severe flares do not respond fully to corticosteroids. 23,24 There are several subsequent treatment options in the National Institute for Health and Care Excellence (NICE) treatment pathway,21 and these have been the subject of randomised controlled trials (RCTs) comparing individual agents with placebo or steroid comparators. To the best of our knowledge, direct ‘head-to-head’ comparisons are not available, except in the VARSITY trial, which compares adalimumab and vedolizumab. 25 NICE has recommended that the risks and benefits of methotrexate, ciclosporin, tacrolimus, adalimumab and infliximab be assessed for the induction of remission in steroid-refractory UC. 21 Subsequently, vedolizumab and tofacitinib have demonstrated efficacy in RCTs and have been recommended for use in this population. 26,27 Ustekinumab has also been shown to be effective and has been recommended for use in UC, but the guidance on this was issued after the current study was under way. 28 Other options, including surgery and the use of intravenous (i.v. ) steroids, have not been the subject of RCTs. These options vary in cost, availability, mode of administration, patient acceptability and utilisation of health-care facilities, as well as clinician experience in their use, especially for newer agents. There is insufficient information to inform patients with steroid-resistant UC about the optimum choice of agent, concomitant immunosuppression and the timing of surgery.

There is no universally adopted definition of steroid-resistant UC. Current definitions might include clinical, endoscopic and quality-of-life dimensions. There is also an overlap between patients with ongoing symptoms or endoscopic inflammation despite corticosteroids (i.e. ‘steroid resistant’) and patients who initially respond and then relapse on reducing the steroid dose (i.e. ‘steroid dependent’). Both groups of patients have been included in clinical trials of agents used for steroid-resistant disease. Pivotal trials, particularly of biologic agents, have included patients in both groups, and the results of treatment in each group have been reported together. Both situations can lead to the prolonged use of systemic corticosteroids and the attendant consequences. The prolonged use of systemic corticosteroids is considered a disutility of care. 29 In addition, these patients may be taking concomitant immunomodulator drugs.

Therefore, although national and international guidance21,26,30 reflects this range of options, there remains a lack of clarity about:

• the definition of steroid resistance

• the specific applicability of current evidence to a population of patients with UC resistant to corticosteroids

• the optimum choice of treatment for this group and the importance of factors such as patient and clinician preferences, concomitant immunosuppression and prior biological therapy.

Review of existing evidence

Ulcerative colitis is associated with disability, psychological morbidity and distress. 31–33 UC also has a significant socioeconomic impact arising from disrupted education and employment,31 with 20 days of household and recreational activities per year typically lost to illness. 34 In 2000, before the widespread use of biological agents, the costs of treating UC were estimated at £3021 per patient-year in the UK35 and at €1524 per patient-year in an European Collaborative inception cohort. 36 Substantial costs relate to hospitalisation and surgery, both of which are more common in those aged < 40 years, and to drug-refractory patients. 37,38 Costs may be reduced by more effective treatment, but it is unclear whether or not newer agents are cost-effective in this population. 37

Oral corticosteroids are associated with significant side effects, which preclude long-term treatment. Therefore, it is important to escalate treatment in patients who do not respond adequately to systemic steroids. A number of treatments are available for patients thought to be resistant to systemic steroids. The anti-tumour necrosis factor (TNF) agents infliximab, adalimumab and golimumab target the pro-inflammatory cytokine TNF-alpha. Infliximab is a human–murine chimeric monoclonal antibody and adalimumab and golimumab are humanised antibodies. Infliximab is most frequently administered intravenously, but a preparation for subcutaneous administration became available in 2021. Adalimumab and golimumab are administered by subcutaneous injection. Vedolizumab is a monoclonal antibody that targets α₄β₇ integrin, an adhesion molecule on the surface of T lymphocytes, preventing it from binding to endothelial MAdCAM-1 (mucosal addressin cell adhesion molecule-1) and preventing these lymphocytes from migrating into gut mucosa. 39 Therefore, it is suggested that this activity is gut specific. Vedolizumab is most often administered by i.v. infusion, but a preparation for subcutaneous injection became available in 2021. Tofacitinib is a ‘small-molecule’ agent that inhibits janus kinase (JAK1 and JAK3)-mediated transcription pathways, preventing inflammatory cytokine production. 40 Tofacitinib is administered orally.

Data about these agents are available from trials and also from meta-analyses. Clinical trials in patients who have not responded or responded inadequately to systemic corticosteroids show benefit of anti-TNF agents (e.g. infliximab,41 adalimumab42,43 and golimumab44,45), vedolizumab,46 tacrolimus, tofacitinib47 and ustekinumab. 48 Clinical benefit is seen in terms of inducing remission, inducing response (i.e. improvement), maintaining remission and healing the inflamed intestinal lining (i.e. mucosal healing).

In clinical practice, supported by guidance from NICE, the British Society of Gastroenterology (BSG) (London, UK) and the European Crohn’s and Colitis Organisation (ECCO) (Vienna, Austria), additional approaches are available, including inpatient i.v. steroids and surgery. 21,26,30,49,50 However, the preferred treatment or approach is not clear.

Guidance from NICE recommends the anti-TNF agents infliximab, adalimumab and golimumab as options for adults whose disease has responded inadequately to conventional therapy, including corticosteroids and mercaptopurine or azathioprine, because they either cannot tolerate them or have contraindications to them. Vedolizumab is also recommended for treating moderately to severely active UC. Tofacitinib is recommended when conventional therapy or a biological agent cannot be tolerated or when the disease has responded inadequately or lost response to treatment.

Definitions of steroid resistance also differ in the literature. Widely used guidelines define steroid resistance as a failure to aachieve symptomatic remission following treatment with systemic steroids (prednisolone at 0.75 mg or 1 mg/kg per day for 4 weeks). 26,51 Creed and Probert52 regard steroid-resistant UC as failure to respond to treatment with high-dose oral steroids within 30 days. However, consideration of escalation to additional treatment after 2 weeks is also discussed. 51 UK guidelines describe steroid dependence as the inability to wean systemic steroids below a prednisolone dose of 10 mg per day within 3 months without the development of active disease, or symptomatic relapse within 3 months of stopping steroids.

Prior steroid response or exposure is not uniform in pivotal trials. Not all patients included in these RCTs are steroid resistant, and results may not be reported separately for patents with steroid-resistant disease and those with steroid-dependent disease. For example, a minority of patients included in pivotal trials of infliximab in active UC were taking ≥ 20 mg of prednisolone per day at trial entry. 41 In addition, in trials of adalimumab and vedolizumab, remission rates were not significantly different from placebo in patients on corticosteroids42 or failing corticosteroids alone. 42,46

Side effects are also important when considering the use of drugs. The adverse effects of systemic steroids are well established, and additional treatments also have important side effects. The risk of infection, serious infection and cancer (including lymphoma) is related to the immunosuppressive properties of drugs. These risks may be low, and randomised trials involving small samples are often unable to detect risks of this size, resulting in the need for evidence from larger, longer-term studies. Therefore, clinicians and patients have to use the evidence in the forms available.

Although qualitative research highlights the diverse perspectives of medical and nursing staff,53 there are limited published patient perspectives, which are needed to inform the design of future clinical trials as well as clinical practice. Survey data suggest that patients’ ideal therapy would be an effective oral formulation that requires them to take few tablets infrequently, with minimal side effects. 54

The role of surgery also needs to be evaluated, as there is limited evidence to determine its ideal position in the treatment pathway. Emergency surgery is positioned in NICE guidance for acute severe colitis,55 but a position is not specifically defined for elective surgery for chronic relapsing disease. The NICE guidance indicates that surgery can be effective to remove symptoms of UC that recur rapidly, but does not describe its optimum timing. 55 The NICE guidance also does not take into account the differing impact of relapses of varying frequency and severity on an individual, nor the impact of drug administration regimes or side effects. Guidelines developed by the Association of Coloproctology of Great Britain and Ireland (London, UK) advocate surgery for relapsing and remitting disease but do not say where in the pathway this should go. 56 Optimum timing of surgery is, therefore, likely to be influenced by a number of disease factors, personal patient preferences and clinician perspectives and, therefore, will differ between patients. We have previously demonstrated that patients wish to undergo surgery when faced with severe restrictions on quality of life. 57 Clinician and patient preferences are, therefore, of central importance in understanding the position of surgery in the treatment pathway.

In a health economic assessment from our group58 (multitechnology appraisal TA329 of anti-TNF agents49) for patients in whom surgery is an option, colectomy was expected to dominate all medical treatment options. For patients in whom colectomy is not an option, infliximab and golimumab were ruled out because of dominance, with the incremental cost-effectiveness ratio for adalimumab compared with conventional treatment expected to be approximately £50,278 per quality-adjusted life-year gained. However, there remains debate with regard to whether surgery should be considered a comparator or an end point and as to when surgery would not be an option. Indeed, in NICE technology appraisal guidance TA342,27 neither surgery nor tacrolimus was thought to be a suitable comparator for vedolizumab.

These issues illustrate the pressing need for a trial with a clearly defined population of steroid-resistant patients to evaluate clinical effectiveness and cost-effectiveness. To develop such a trial, a detailed understanding of how best to describe steroid resistance, and of patients’ and clinicians’ views of treatment options and objectives, is required. This will allow appropriate identification of equipoise and acceptable invention and comparator arms in a trial.

Statement of the problem

In commissioning brief 17/72, the Health Technology Assessment programme indicated its interest in improving the understanding of how adults with steroids-resistant UC are managed, with a view to informing future commissioning briefs addressing which treatments require further evaluation in steroid-resistant UC.

Aims and objectives

The overarching research question for this study was ‘How are adults with steroid-resistant UC being managed in secondary care, and how does current practice compare with patient and clinician preferences?’.

To answer the research question, this mixed-methods study had five objectives:

1. to describe current practice in the management of adults with steroid-resistant UC and to describe how medical resistance to steroids is defined

2. to understand how treatment pathways and definitions of steroid resistance are operationalised in practice

3. to understand patient experiences of different treatment options and approaches to decision-making

4. to estimate the relative utility of different treatment options and to elicit patient and clinician preferences for these and their willingness to trade between them

5. to make recommendations about future research and treatment options.

The five objectives aligned with five corresponding work packages (WPs) delivered as part of this mixed-methods study. To understand the management and treatment preferences of patients, a survey of clinical practice (WP1) and qualitative interviews with adults with steroid-resistant UC (WP3) were carried out. A survey and qualitative interviews with a purposive sample of health-care professionals were designed to explore how health-care professionals define and treat steroid-resistant UC (WP2). A discrete choice experiment (DCE) with patients and health-care professionals was designed to quantify their preferences for, and estimate the relative utility of and willingness to trade between, different treatment options (WP4). Finally, the aim of a multidisciplinary workshop with patients and health-care professionals (WP5) was to present and discuss the study findings to generate recommendations about optimum treatment and future research.

Overview of methodological approach

The PoPSTER (Patient preferences and current Practice in STERoid resistant ulcerative colitis) study was a mixed-methods study and was composed of five WPs, using a combination of qualitative and quantitative research methods to help achieve each of the five corresponding objectives. This included a cross-sectional survey of health-care professionals (objective 1, WP1), qualitative interviews with health-care professionals (objective 2, WP2), qualitative interviews with people with UC (objective 3, WP3), DCEs with health-care professionals and patients (objective 4, WP4) and a multistakeholder workshop (objective 5, WP5). The details of the study design, setting, eligibility criteria, sampling, recruitment, data collection and analysis for each of the WPs are described in this chapter in Health-care professional survey, Qualitative interviews, Discrete choice experiments and Multistakeholder workshop. This chapter also includes details of patient and public involvement (PPI) (see Patient and public involvement), ethics approval (see Ethics approval) and protocol management (see Protocol management and version history).

Health-care professional survey

Qualitative interviews

Discrete choice experiments

Health-care professionals

Study design

This WP involved an online DCE with health-care professionals in the UK with expertise in IBD. The DCE involves a series of tasks in which respondents select a preferred treatment option when presented with two alternative treatment profiles. These treatment profiles are constructed using a set number of attributes and levels that differ across the alternatives. An attribute is a treatment characteristic that is important to the treatment decision, and a level is a clinically plausible range for each attribute. 66

Identification of treatment characteristics

All relevant attributes and levels were identified using three approaches: (1) reviewing the literature, (2) conducting qualitative interviews and (3) consulting clinicians to select the most important attributes and levels for the DCE. As reported in WP2, qualitative interviews (n = 20) were conducted with health-care professionals with expertise in IBD to understand how patients with steroid-resistant UC are treated in the UK. From these interview transcripts, we identified 16 key themes that health-care professionals considered important when making treatment choices. To convert these themes to possible attributes and levels, we convened a panel of four IBD specialist clinicians (AL, ML, CP and SS). Using iterative rounds of discussion, the panel helped to consolidate and select the most important attributes and also refined the phrasing of attributes.

This process of selecting the key attributes is important for the structural reliability of a DCE, as evidence suggests that the DCE tasks can become cognitively challenging if attribute and level selection is not optimised. 67 When deciding the appropriate levels for each attribute, we carefully selected clinically meaningful values from the clinical trials literature that are sufficiently wide that respondents will be encouraged to trade treatment profiles. The final five attributes (each with three levels) focused on treatment efficacy and safety (Table 1).

TABLE 1 - Attribute descriptions and levels of health-care professional DCE

Attribute	Level
The likelihood of induction therapy successfully leading to a clinical response (i.e. significant improvement in clinical symptoms)	40% 50% 60%
The likelihood of a treatment achieving mucosal healing (i.e. a Mayo endoscopic subscore of ≤ 1)	40% 50% 60%
Remission: efficacy as a maintenance treatment – likelihood of achieving clinical response at 12 months	35% 50% 70%
Risk of lymphoma	3 in 10,000 patient-years 5 in 10,000 patient-years 8 in 10,000 patient-years
Risk of serious infection (the baseline risk in patients unexposed to immunosuppressive medication is approximately 1–2 per 100 patient-years)	1 in 100 patient-years 5 in 100 patient-years 10 in 100 patient-years

Questionnaire design

An online survey was developed that contained DCE questions, questions about respondent demographics and feedback questions about the survey. The DCE questions were generated using Ngene software (ChoiceMetrics, Sydney, NSW, Australia). Five attributes with three levels each produced 243 possible treatment profiles and, therefore, to create a manageable DCE task a fractional factorial design was used. The D-optimal design generated 12 choice questions, which followed the principles of orthogonality, minimum overlap and level balance. 68 Each DCE question contained two unlabelled treatment profiles (Figure 1). Opting out of treatment choices was not given as a response option because patients with steroid-resistant UC need treatment and doing nothing can be detrimental to their quality and length of life.

FIGURE 1.

Example of a DCE choice question from the health-care professional DCE.

At the start of the survey, a series of screens displayed instructions, including detailed descriptions of the attributes, the levels and the choice context in which the respondents should address the DCE tasks. For each question, the respondents were asked to select the treatment they would choose to offer a patient with steroid-resistant UC (see Figure 1). In addition to the 12 choice questions, a dominant choice question, where one treatment was logically better, was included in the DCE section to test the internal validity of the survey. The final version of the survey contained 13 choice questions that were displayed in a random order to the respondents. A soft launch of the survey was undertaken (n = 50) to assess any problems, including comprehension and dropouts.

Sampling and recruitment

Clinicians and specialist nurses with experience of treating patients with UC in NHS trusts were invited to take part in the online survey through the Qualtrics platform. A link to the survey was sent to health-care professionals in an e-mail (from the IBD section of the BSG and the RCN IBD Nurses Network), via social media and in newsletters during the data collection period between June and August 2020. Participants provided informed consent to participate in the study and were not given any financial incentives to complete the survey. The target sample size was 100 survey respondents, which was based on the minimum sample size required (n ≥ 62.5) to estimate a model using the rule-of-thumb approach and also on the literature that reports the average sample of a DCE ranging between 100 and 300 participants. 69,70

Data analysis

Descriptive statistics were used to summarise respondents’ characteristics and survey feedback questions. Responses to the DCE questions were analysed using a conditional logit model. Attributes were first included as categorical variables using dummy coding; however, after the linear relationship was confirmed through visual inspection and model fit, all attributes were included as continuous variables in the models. 71 The models with the full sample report only main effects, and no interaction terms were included. All statistical analyses were conducted using Stata^® v16 (StataCorp LP, College Station, TX, USA). The utility function for estimating the probability of choosing the preferred treatment profile takes the following form:

⁽¹⁾ $$\begin{gathered} V = β 1 response + β 2 mucosal healing + β 3 remission + β 4 lymphoma risk \\ + β 5 serious infection risk + ε , \end{gathered}$$

where V is a binary variable (1 = the preferred treatment profile is chosen, 0 = the treatment profile is not chosen), β is the estimated coefficient for each of the treatment attributes and ε is the unobserved error term.

Exploratory analyses were also conducted to evaluate preferences across subgroups according to hospital type. This was achieved by estimating a model with interactions for all main effect variables for health-care professionals working at secondary or tertiary hospitals to identify any significant differences in their preferences.

Using the results of the conditional logit model, marginal willingness to trade benefit and risk was calculated to find the rate at which health-care professionals are willing to trade levels of one attribute for the preferred levels of another attribute. Benefit–risk trade-offs were calculated by dividing the estimated parameter coefficient for benefit attributes (i.e. induction of response, mucosal healing or maintenance of remission) by the estimated parameter coefficient for the risk attributes (i.e. risk of lymphoma or serious infection).

Parameter estimates from the conditional model were also used to predict uptake rates of a selected number of drugs currently commonly prescribed to patients with steroid-resistant UC. 72 The probability of choosing alternative i is:

⁽²⁾ Pr ( alternative i is chosen ) = e V i Σ ∈ j e V j ,

where V_i is the estimated utility associated with alternative i and V_j is the sum of utility of j alternatives (see example calculations in Appendix 5).

Moreover, parameter estimates from the conditional model were also used to calculate the change in probability of uptake from a baseline scenario where all attributes are set to their worst level and then improving each attribute one at a time (see example calculations in Appendix 5).

Patients

Study design

This WP involved an online DCE with adults living with UC in the UK.

Identification of treatment characteristics

The development of attributes and levels was informed by a combination of reviewing the literature, conducting qualitative interviews with patients and consulting Patient Advisory Group (PAG) members to find out what they considered the most important attributes. As reported in WP3, we conducted 33 interviews with patients to identify key characteristics that patients consider important when selecting a treatment. Thematic analysis of the qualitative interviews generated eight themes, which were ranked by the four PAG members using a dot-voting technique. 73 The patients were given 16 dots to distribute across the eight themes, with themes with the most dots indicating the most desirable treatment characteristics. Through this process the eight themes were converted and reduced to attributes. After discussions with the PAG members, one theme around the need for regular monitoring was dropped, as this was deemed the least important of the eight. Similar themes were merged; for example, route and frequency of administration were merged to create one single attribute, and quality of life and inducing a treatment response were merged to create another single attribute. The final five attributes focused on effectiveness, remission, speed of response, treatment administration and safety of treatment (Table 2).

TABLE 2 - Attribute descriptions and levels of patient DCE

Reproduced with permission from Wickramasekera et al. 74 © 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The table includes minor additions and formatting changes to the original table.

Attribute	Level
How effective the drug is at treating your symptoms: the drugs may improve or settle your symptoms (e.g. in reducing stool frequency and bleeding, or returning these to normal), improve your quality of life and make you feel better	40% 50% 60%
Speed of response to treatment: some drugs take longer than others to take effect	6 weeks 8 weeks 14 weeks
Chance of your symptoms remaining improved after 12 months: after your initial symptoms improve, the drugs can help to control your symptoms over time; however, there is also a possibility that you may lose the improvement and develop a flare of your symptoms	35% 50% 70%
Route and frequency of administration: how and where the medication would be taken is different according to which drug you take	A pill taken daily at home A self-administered injection under the skin administered every 2 weeks at home A self-administered injection under the skin administered every 8 weeks at home An i.v. infusion (drip) administered every 8 weeks at hospital
Chance of experiencing side effects: drugs can cause unwanted side effects. Common side effects include nausea, headache, skin rashes and mild infections. These side effects often settle without treatment, can be easily treated or are reversed if the drug is stopped. In rare cases, the drugs may cause severe side effects over a longer period of time. These severe side effects include more severe infections (e.g. tuberculosis and viral infections, including the shingles virus), some cancers, including lymphoma (i.e. lymph gland cancer), blood clots in the leg (i.e. deep-vein thrombosis) or lung (pulmonary embolism) and nervous system problems. The chance of experiencing severe side effects is very low for all treatments	Very common (may affect more than 10 in 100 people) Common (may affect up to 10 in 100 people) Uncommon (may affect 1 in 100 people or fewer) Very rare (may affect up to 1 in 10,000 people)

The PAG and PPI co-applicants also helped to refine the language explaining the DCE task to respondents by piloting the draft questionnaire. For example, to aid understanding, we framed the risk attribute quantitatively in the DCE task (i.e. 60/100, 60%); however, in the introduction to the DCE, risks were framed qualitatively (i.e. ‘a drug that is 60% effective means that, if 100 people had the same drug for UC, for 60 people the treatment would be effective but for 40 people treatment would not be effective’). 71 The levels for the attributes were selected to reflect plausible values from the published clinical trials literature. For the side effects attribute, we expressed the levels modelled in accordance with the categories described in printed information leaflets included with drugs that patients read and are familiar with.

Questionnaire design

The first section of the survey contained the DCE tasks, where participants were shown a series of side-by-side comparisons of competing treatment profiles and were asked to select the preferred treatment profile (Figure 2). We used Ngene software to create 12 DCE tasks using the D-optimal method to maximise statistical efficiency, and followed the principles of orthogonality, minimum overlap and level balance. One additional dominant question that was logically better was also included to test whether or not participants understood the DCE task. 66 To make the choice realistic, participants were not given an opt-out option because treatment is necessary to improve their length and quality of life.

FIGURE 2.

Example of a DCE choice question from the patient DCE.

The second section of the survey involved a ranking exercise in which patients were asked to rank four commonly used treatments (i.e. adalimumab, infliximab, tofacitinib and vedolizumab) in order of preference from 1 to 4 (1 = best preferred treatment and 4 = least preferred treatment). To aid this task, we provided comprehensive details of the treatments, which included the effectiveness of the drug, speed of response to treatment, route of administration, side effects and whether or not concomitant medication is needed (see Report Supplementary Material 3). These treatment descriptions were developed using published literature, with clinical input from the study team, and were presented to participants in a randomised order to reduce question order bias.

In the third section of the survey, we gathered sociodemographic details and information on the respondents’ personal history and severity of UC. The survey included two validated instruments, the IBD Control-8 questionnaire and the EuroQol-5 Dimensions, five-level version (EQ-5D-5L). The IBD Control-8 questionnaire captures disease control and impact from the respondents’ perspective and generates a summary score, ranging from 0 (representing worst control of disease) to 16 (representing best control of disease). 75 The EQ-5D-5L instrument captures respondents’ overall quality of life, generating a summary score of between –0.59 and 1, where higher scores represent better quality of life (and 1 represents perfect health). 76 The four survey sections contained feedback questions about the survey. A soft launch of the survey was undertaken (n = 50) to assess any problems, including comprehension and dropouts.

Sampling and recruitment

The study population included adults aged ≥ 18 years who had a diagnosis of UC. Participants were recruited primarily through two NHS trusts where staff working in IBD services advertised the study by sending participants invitation letters. The study was also advertised on social media (via Twitter and Facebook) to recruit further participants from across the UK. If participants decided to take part, then they were able to access the online survey via the Qualtrics platform and complete the survey after providing informed consent. Respondents were not offered any financial incentives for completing the survey. We hoped to recruit up to 300 survey participants as the literature shows that DCE sample sizes can range from 100 to 300 participants. 70 However, the minimum sample size required was n ≥ 83.3 to estimate a model using the rule-of-thumb approach. 69

Data analysis

Descriptive analyses were performed to analyse the demographic data and IBD characteristics of the respondents and to rank medications in order of importance. We performed conditional logistic regression models to analyse the DCE task data. All of the models report main effects, and no interaction terms were included. Attributes were first included as categorical variables using dummy coding; however, after confirming the linear relationship through visual inspection and model fit, speed and remission attributes were treated as continuous variables and effectiveness, administration and side effects as categorical variables in the main model. All statistical analyses were conducted using Stata.

The utility function to estimate the probability of choosing the preferred treatment profile takes the following form:

⁽³⁾ V = β 1 response + β 2 speed + β 3 remission + β 4 administration + β 5 side effects + ε ,

where V is a binary variable (1 = the preferred treatment profile is chosen, 0 = the treatment profile is not chosen), β is the estimated coefficient for each of the treatment attributes and ε is the unobserved error term.

Parameter estimates from the conditional model were also used to calculate the change in probability of uptake from a baseline scenario in which all attributes are set to their worst level and then one attribute is improved at a time72 (see Appendix 6).

Multistakeholder workshop

Patient and public involvement

The PoPSTER study was developed in collaboration with people with UC. PPI activity is reported here and in Chapter 9 in accordance with the GRIPP2 (Guidance for Reporting Involvement of Patients and the Public 2). 78

At the initial grant application stage, we invited Sue Blackwell to join the co-applicant team and, as part of this, she reviewed the stage 1 application (leading to changes to the design of WPs 2 and 3 and to the content of the Plain English summary). At the second stage of grant application, clarifications were made to the dissemination strategy based on Sue Blackwell’s feedback. Sue Blackwell has expertise in digital marketing, which was helpful to study promotion throughout the grant.

During the preparation of the second-stage application, a local group of patients from Sheffield Teaching Hospitals NHS Foundation Trust were convened to seek wider feedback on the design and scope of the study. The group of patients varied in age (ranging from 20 to 80 years), gender and ethnicity, but they shared a positive response to the research. The group provided encouraging feedback on the study aims and methods, highlighting that these were easy to understand and appropriate to the aims of the research call. This group also made some specific recommendations for improving the study design, including making explicit the maximum interview duration for patients (i.e. 60 minutes), offering breaks to encourage and support those people experiencing a UC flare and including a measure of health-related quality of life in WP4, all of which were introduced. Through this group, Hugh Bedford was identified and was invited to join the team of co-applicants on the grant.

In addition to this, Alan Lobo and Elizabeth Coates worked closely with Sue Blackwell to organise a remote feedback session with patient representatives across the UK. Sue Blackwell distributed an appeal for patient feedback via her Facebook networks and nine people with UC shared their views on the proposed research. Again, this group welcomed the focus on research addressing treatment for steroid-resistant UC. As with the Sheffield group of patients, this group also highlighted the importance of offering breaks during interviews and providing flexible timings for the patient interviews. This group also suggested that limited patient knowledge about the treatment options could make participation in the DCE challenging (highlighting the importance of ongoing PPI input to study documentation and data collection materials). Again, all of these suggestions were incorporated into the final grant application. In addition to this, Nicola Dames was identified and was invited to join the team as a third PPI co-applicant.

During the conduct of the PoPSTER study, the three patient co-applicants (SB, HB and ND) were integral members of the Study Management Group. Therefore, the three patient co-applicants were invited to all Study Management Group meetings and asked to provide input to relevant study documentation, outputs and key issues arising during the delivery of the research, alongside the clinical and methodologist co-applicants. This ensured that the patient perspective informed the design, delivery and reporting of each stage of the studies.

In parallel with this, Nicola Dames and Sue Blackwell helped the research team to establish a separate PAG of four people living with UC (please refer to the Acknowledgements for information on the membership of this group) to get a wider perspective. This group was convened three times to coincide with key stages of the study. At the first meeting (April 2019), the members gave detailed feedback on the content of the qualitative interview schedules for patients. This feedback included comments on the importance of mental health and trauma in the lived experience of UC, comments on the influence of health-care professionals on treatment choices and the significance of that relationship to informing those decisions, a number of helpful clarifications to question wording and practical suggestions, such as sending the list of questions to patients in advance. At the second meeting (November 2019), the PAG members were presented with the headline findings from the WP2 and WP3 interviews and were asked to help with prioritising the long list of attributes for the patient DCE. Subsequent to the meeting, the PAG and the patient co-applicants piloted the DCE questionnaire and gave helpful feedback to improve content and presentation. At the final meeting (March 2021), the headline findings from the patient DCE were presented and the group gave feedback on those, as also helped to interpret the results.

Other PPI activity included promotion of patient DCE on social media by SFB to help support recruitment and ongoing support for study recruitment on Twitter by Sue Blackwell and Nicola Dames. Sue Blackwell, Hugh Bedford and Nicola Dames also gave valuable input by reviewing the Plain English summary and PPI sections of this report.

Ethics approval

Ethics approval for the PoPSTER study was granted by the NHS East Midlands – Derby Research Ethics Committee on 10 January 2019 (reference 19/EM/0011) and governance approval was granted by the Health Research Authority on 17 January 2019 (Integrated Research Application System reference 255616).

Protocol management and version history

See study protocol version 5.0, which is available online at URL: www.fundingawards.nihr.ac.uk/award/17/72/02 (accessed 26 April 2022). A protocol version history is provided in Appendix 1.

Participants

There were 387 unique visitors to the online survey, and 168 of these visitors consented to take part in the survey (i.e. a 43% participation rate). Of these 168 participants, 145 started the survey and 88 completed every survey question, giving an overall completion rate of 52%. The denominator for individual questions varied from 88 to 145, and is reported for each question to help facilitate interpretation of the results.

The characteristics of participants are summarised in Table 3 (see also Appendix 2, Table 24). The majority of survey respondents were medics (n = 99, 68%) and 44 (31%) were nurses. Fifty-five (38%) participants were working as consultant gastroenterologists with a special interest in IBD. On average, the participants were appointed 9.6 years ago. Eighty-one per cent of participants had no personal experience of IBD. Eighty-eight (62%) participants were from secondary referral centres, 48 (34%) from tertiary referral centres and five (3%) from quaternary referral centres. All UK regions were represented in the survey. Most participants stated that some (n = 59, 41%) or the majority (n = 53, 37%) of their clinical time was dedicated to working with IBD. The majority of participants were part of a MDT (n = 133, 93%) containing, on average, 14 health-care professionals. Most participants reported that they were using national and European guidelines for managing UC (77% and 87%, respectively), as well as NICE guidance for specific treatments (66–83%).

Definitions of steroid resistance

As shown in Table 4, definitions of steroid-resistant UC varied, with 68% (92/135) of participants agreeing or strongly agreeing that this is indicated by an incomplete response to prednisolone at 40 mg per day (or equivalent) after 2 weeks. Of the remaining participants, 58% (25/43) agreed or strongly agreed that steroid resistance was indicated by an incomplete response after 4 weeks. If participants did not agree with either of the definitions presented (n = 13), then they were given the opportunity to design their own definition of steroid-resistant UC using pre-set response categories (see Appendix 2, Box 1).

Table 5 shows that 77 (57%) of participants agreed that steroid resistance did not include those patients who relapse after initial remission with corticosteroids (i.e. steroid dependent). A greater proportion of participants felt that treatments should be different for steroid-dependent and steroid-resistant disease at each time interval from 2 weeks to 3 months. Whereas, at 6 months, more participants felt that the treatment options should not differ. Only 13% of participants felt that steroid-dependent and steroid-resistant disease should be treated identically, regardless of the interval between remission and subsequent relapse.

Treatment options for steroid-resistant and steroid-dependent clinical scenarios

Treatment options for steroid-resistant and steroid-dependent clinical scenarios

The results are shown in Table 8 and Figure 3, with additional information on treatment preferences given in Appendix 2, Tables 25–28. In steroid-resistant patients, anti-TNF agents (i.e. infliximab, adalimumab and golimumab) were the most frequently suggested treatments for both those exposed to thiopurines and those who were thiopurine naive [95% (n = 114) and 87% (n = 104), respectively]. In steroid-dependent patients, anti-TNF agents remained the most frequently offered [88% (n = 105) in thiopurine-exposed patients and 75% (n = 90) in thiopurine-naive patients]. In all scenarios, infliximab was the most frequently suggested treatment, suggested by 94% (n = 113), 73% (n = 88), 86% (n = 103) and 67% (n = 80), respectively (and albeit with thiopurine or methotrexate in the thiopurine-naive groups).

TABLE 8 - Summary of treatment options for steroid-resistant and steroid-dependent patients (N = 120)

Treatment option	Number (%) of responses
	Steroid resistant: thiopurine exposed	Steroid resistant: thiopurine naive	Steroid dependent: thiopurine exposed	Steroid dependent: thiopurine naive
Anti-TNF agent	114 (95)	104 (87)	105 (88)	90 (75)
Admit for i.v. steroids	78 (65)	76 (63)	23 (19)	23 (19)
Thiopurine		64 (53)		70 (58)
Vedolizumab	83 (69)	71 (59)	92 (77)	73 (61)
Tofacitinib	51 (42)	34 (28)	50 (42)	33 (28)
Other	87 (72)	73 (61)	74 (62)	82 (68)

FIGURE 3.

Treatment options for steroid-resistant and steroid-dependent patients.

Vedolizumab was more frequently offered to steroid-dependent patients on thiopurines than to those with steroid-resistant disease on thiopurines, but this difference was not statistically significant [n = 92 (77%) vs. n = 83 (69%); p = 0.137]. As would be expected, admission for i.v. steroids was less frequently offered in steroid-dependent scenarios than in steroid-resistant scenarios [n = 23 (19%) vs. n = 76/78 (63–65%); p < 0.001]. Across all four scenarios, tofacitinib would be more likely to be offered in patients already on thiopurines than in patients naive to thiopurine [n = 101 (42%) vs. 67 (28%), χ² = 10.586(1); p = 0.001].

Treatment options for clinical scenarios with steroid-dependent disease and relapse at different doses

The results are shown in Table 9 and in Figures 4–6, and in full in Appendix 2, Tables 29–34. Anti-TNF drugs were, again, the most frequently suggested treatment for steroid-resistant disease, with significantly more patients on thiopurines being offered anti-TNF agents than patients who were thiopurine naive [n = 75 (81%) vs. n = 58 (62%); p < 0.001]. For patients receiving prednisolone at 25 mg per day, on relapse, 78% (n = 73) of clinicians would offer an anti-TNF agent to thiopurine-exposed patients and 49% (n = 46) of clinicians would offer an anti-TNF agent to thiopurine-naive patients (p < 0.001). Forty-four (47%) clinicians would offer thiopurine-naive patients a further increase in steroids to allow thiopurine or methotrexate introduction and 46 (49%) clinicians would introduce thiopurines alone. Sixty-seven (72%) clinicians would offer either of these options in thiopurine-naive patients. For patients receiving prednisolone at 5 mg per day, on relapse, 43 (46%) clinicians would offer anti-TNF agents for a thiopurine-naive patient and 65 (70%) clinicians would introduce a thiopurine. In an individual on thiopurine, 79 (85%) clinicians would offer an anti-TNF agent, 43 (46%) clinicians would offer vedolizumab, 23% (n = 21) clinicians would offer tofacitinib and 20 (22%) clinicians would increase the steroids alone.

TABLE 9 - Summary treatment options for steroid-resistant and steroid-dependent (relapsing at different doses) patients (N = 93)

*p < 0.001 compared with the thiopurine-naive steroid-resistant group; **p < 0.001 compared with patients relapsing at 25 mg/day who are thiopurine naive; ***p < 0.001 compared with patients relapsing at 5 mg/day who are thiopurine naive.

Treatment option	Number (%) of responses
	Steroid resistant		Steroid dependent with relapse at 25 mg/day		Steroid dependent with relapse at 5 mg/day
	Thiopurine naive	Thiopurine exposed	Thiopurine naive	Thiopurine exposed	Thiopurine naive	Thiopurine exposed
Anti-TNF agent	58 (62)	75 (81)*	46 (49)	73 (78)**	43 (46)	79 (85)***
Admit for i.v. steroids	30 (32)	36 (39)	11 (12)	15 (16)	10 (11)	7 (8)
Thiopurine (including increase in steroid dose and addition of thiopurine)	39 (42)	6 (6)	67 (72)	11 (12)	65 (70)
Vedolizumab	27 (29)	41 (44)	25 (27)	44 (47)	36 (39)	43 (46)
Tofacitinib	20 (22)	23 (25)	19 (20)	20 (22)	14 (15)	21 (23)
Other	44 (47)	50 (54)	52 (56)	45 (48)	42 (45)	49 (53)

FIGURE 4.

Percentage of respondents who would offer an anti-TNF agent according to steroid dose at relapse (thiopurine naive and exposed).

FIGURE 5.

Percentage of respondents who would offer vedolizumab according to steroid dose at relapse (thiopurine naive and exposed).

FIGURE 6.

Percentage of respondents who would offer tofacitinib according to steroid dose at relapse (thiopurine naive and exposed).

The difference in suggested use of anti-TNF agents, depending on thiopurine use and dose of steroid at relapse, is shown in Figure 4. In those patients who are naive to thiopurines, there is reducing likelihood that anti-TNF agents will be offered when steroid-resistant patients are compared with patients who relapse at a prednisolone dose of 25 mg and 5 mg [62% vs. 49% (p = 0.014) and 46% (p = 0.009), respectively]. Anti-TNF treatment would be used more frequently in patients on thiopurines, but with no difference at the differing prednisolone doses at relapse (78% vs. 85%; p = 0.077).

As shown in Appendix 2, Tables 29–34, across the scenarios, infliximab was the most frequently offered treatment (albeit with thiopurine or methotrexate in the thiopurine-naive steroid-resistant group), apart from in thiopurine-naive steroid-dependent patients who relapse at a prednisolone dose of 25 mg or 5 mg, where thiopurine was most preferred [at 49% (n = 46) and 70% (n = 65), respectively].

The frequency with which vedolizumab was suggested (see Figure 5) did not change with flares as a consequence of steroid dose reduction (range 44–47%) in thiopurine-exposed patients; however, vedolizumab was most frequently offered (albeit not statistically significant) in thiopurine-naive patients who flared at 5 mg, than when flares occurred at 25 mg daily (p = 0.003) or in a resistant scenario (p = 0.006). Offering tofacitinib occurred less frequently (at 15–23% across all scenarios) and the frequency did not change according to the steroid dose or with thiopurine use (see Figure 6).

Factors influencing treatment choices

To understand the importance of a range of factors influencing treatment decisions for patients with steroid-resistant UC (both thiopurine naive and exposed), participants were asked to rate factors on a five-point Likert scale (i.e. very important, important, neutral, low importance and not at all important). These responses were converted from five to three categories (i.e. important, neutral and not important). The results for thiopurine-exposed patients are displayed in order of most importance in Table 10.

Whether or not patients had been exposed to thiopurines, treatment efficacy and effect on quality of life were most frequently rated as important, at 96% (n = 99), 95% (n = 98) and 88% (n = 91), respectively. Although the frequency of ratings was often lower for patients who are naive to thiopurines, more than two-thirds of respondents rated most of the factors as being important to treatment decisions for steroid-resistant patients (ranging from 67% to 92% positive ratings across both groups). Differences between the groups were statistically significant for effect on fertility/pregnancy only (p = 0.010).

Some factors were less frequently rated as important for both thiopurine-exposed and thiopurine-naive patients, that is, effect on intimacy, infusion bay or service capacity, cost and hospital inpatient bed use, with lower importance scores (32–55%) and greater neutral (20–31%) or negative ratings (8–31%) relative to the other factors. Taken together, these findings highlight the wide range of factors that potentially influence treatment decisions in this patient group.

Referral for surgery

Participants were asked ‘In general terms, in which situations would you consider referring a patient with corticosteroid-resistant UC, of moderate severity, for surgery?’. As shown in Table 11, the majority (n = 57, 48%) of participants stated that they would offer surgery ‘at any time’, whereas 12% (n = 14) would wait to offer surgery only once all medical options had been used. Only two respondents stated that they would refer a patient to surgery once they had been deemed resistant to steroids. Six per cent (n = 7) of participants would offer surgery after unsuccessful use of one biologic and 9% (n = 11) of participants would do so after unsuccessful use of two biologic therapies.

Participants who would offer surgery ‘at any time’ were asked to clarify which factors would influence their decision to refer patients. Sixty-one per cent (n = 35) of respondents cited patient preferences for treatment options as the reason to refer patients. Over half (n = 31, 54%) of respondents indicated that patients were jointly managed by surgeons in the local MDT and, therefore, surgery was presented early in the treatment pathway for UC. Both failure to respond to medical treatment options (n = 18, 32%) and disease severity (n = 14, 25%) were also important reasons. The four respondents who selected ‘other’ for the timing of surgery referral also gave similar reasons.

Differences in practice around surgery referral between key subgroups (i.e. profession, centre type and clinical time devoted to IBD care) were explored using chi-squared tests. However, there were no statistically significant differences in timing of surgery referral between professions (medics vs. nurses; p = 0.28), centre type (tertiary and quaternary vs. secondary; p = 0.14) or clinical time devoted to IBD care (majority/all of your time vs. some/little; p = 0.49).

Use of endoscopy

Participants were asked about how they use endoscopy with steroid-resistant and steroid-dependent patients (Tables 12 and 13). In a patient with ongoing symptoms, resistant to what the respondent felt was their maximum duration and dose of systemic steroids, 51% (n = 45) of participants would always carry out endoscopic assessment to confirm disease activity, whereas 43% (n = 38) of participants stated that they would never do so. For steroid-resistant patients, a moderate degree or more severe symptoms/unchanged from the last examination were most frequently reported as being indicative of a need for treatment change, at 42% (n = 37) and 31% (n = 27), respectively.

In contrast, in a patient whose symptoms rapidly recur on reduction of systemic steroid dose (at or before reaching 15 mg/day), only 31% (n = 27) of participants would always undertake endoscopic assessment and 58% (n = 51) of participants reported that they would never do so. As with steroid-resistant patients, a moderate degree (n = 33, 38%) or more severe symptoms/unchanged from the last examination (n = 23, 26%) were most frequently reported as being indicative of a need for treatment change.

Participants

A total of 25 health-care professionals were approached for interview, but five health-care professionals did not reply (n = 3) or there was loss of contact, despite initial expression of interest (n = 2).

Twenty health-care professionals participated in the qualitative interviews (Table 14). Twelve (60%) participants were consultant gastroenterologists and eight (40%) were IBD nurses. The median time since appointment was 14 (minimum–maximum 2–21) years. There was an equal split in the gender of participants. Participants were based in all regions of England and Wales, and there was a 50 : 50 split between secondary and tertiary referral centres. The median length of individual interviews was 33 (minimum–maximum 18–60) minutes.

Analysis

Through familiarisation with the interview transcripts, an initial list of codes was developed. After several iterations, it was possible to agree on a coding framework that included 32 codes, which were grouped under five categories (see Appendix 3, Table 35). The coding framework was then applied to the whole data set and summaries of each code were produced to help identify themes. The presentation of the findings in this chapter is structured around these themes and subthemes, as follows.

Definitions of steroid resistance

Treatments for steroid-resistant ulcerative colitis

Health-care professionals were asked to describe the treatments that they use with patients whose UC is resistant to steroids. Overall, health-care professionals consistently explained that they followed a treatment pathway that started with optimisation of aminosalicylates for mild to moderate disease, before escalating to thiopurines (e.g. azathiopurine or mercaptopurine) for patients able to tolerate this treatment, followed by biologic therapies, such as anti-TNF treatments, vedolizumab or tofacitinib. For example, as this IBD nurse succinctly explained:

So we would offer them thiopurines, so azathioprine and mercaptopurine, we usually use azathioprine first line and then mercaptopurine as second line and then if patients fail on that, then we would go at biologics and our trust we generally use infliximab as first line depending on the reasons for any loss of response we’d then go to adalimumab. Obviously if it’s the loss of response because they’ve built up antibodies, then we would switch out of class and use vedolizumab but and then now we’d probably do tofacitinib, but that is more of like a third line at the moment, unless there’s a reason why we can’t use any of the others.

IBD nurse 1

Some health-care professionals explained that they may not always use thiopurines and prefer instead to move straight to biologic treatments as first-line agents when steroids are not working. When considering anti-TNF treatments, infliximab was commonly described as the first-line treatment option, with adalimumab or golimumab used to a lesser extent. During the interviews, it was clear that practice around the use of tofacitinib varied, which is not surprising, given the timing of this study. It was evident that some health-care professionals were already using this on a parity with other biologics, whereas other participants were more hesitant and have not yet tried this with any or many UC patients:

At the moment in most patients we probably still use an anti-TNF first line in that patient population because of familiarity and relatively straight-forward access. In some patients I would consider using tofacitinib, we’ve got reasonable access to that; locally we’ve treated 36 patients so far, some of whom were biologic naive, which is unusual. I think most UK sites are using tofacitinib and have mainly used it in patients who have been refractory to other biologic or to biologics.

Consultant gastroenterologist 4

. . . so the things that changed relatively recently, which is interesting, so for, so if we just start with steroid-refractory disease, so people who have got active disease that hasn’t responded to steroids, and you need an induction agent, we would probably use, here tofacitinib first line now.

Consultant gastroenterologist 5

But those would be our sort of two, the erm TNF-alpha blockers, plus the vedolizumab, we haven’t yet had to use tofacitinib acutely yet.

Consultant gastroenterologist 9

And we are looking into a new drug that’s out there, tofacitinib, but obviously there is quite a lot of things to consider when putting patients on, you know, using the tofacitinib.

IBD nurse 7

Health-care professionals also explained how their treatment decisions are made on a patient-by-patient basis and so their relationships with patients were paramount to their understanding of the most appropriate next treatment. More broadly, health-care professionals explained that their decisions are informed by their knowledge of different treatments, which is gathered from clinical guidelines, research, conferences and their IBD colleagues.

Factors influencing treatment choices by health-care professionals

Health-care professionals were also asked about the factors that they take into account when offering treatments to people with UC. As in the survey results, a broad range of inter-related issues were identified as important considerations, and are summarised below. It is also important to note that most interview participants were keen to clarify that they consider treatment choices on a case-by-case or patient-by-patient basis, taking into account different factors as necessary, and, as explained in Patient preference, patient preferences are key.

Referral for surgery

Most health-care professionals explained that they typically introduce the possibility of surgery for UC to patients at an early stage, usually at, or shortly following, diagnosis. Although surgery was generally seen as a longer-term treatment option for UC, health-care professionals explained that it is important to present this as one of the available options alongside medical treatments from the outset:

We do use surgery as a long resort but we should be talking about it earlier so actually if it does come up it’s not quite as scary and because we do have more medications. But even still there will still be a certain amount of patients who will end up having surgery and it’s important that we prepare them for that.

IBD nurse 5

I talk about it really quite early on erm in fact when I see people in clinic for the first time at diagnosis, I will mention that surgery, so what I tend to do is I tend to talk to them about a therapeutic ladder where they are starting with 5-ASAs [5-aminosalicylic acids] then steroid then immunosuppressant and biologicals. But I say to them ‘but you can take two steps on the ladder or three steps on the ladder without you know don’t worry about that, and if you fall off the ladder that’s surgery’.

Consultant gastroenterologist 2

Central to this was the importance of carefully managing patients’ expectations about possible surgical options to help with acceptance in the longer term. Health-care professionals generally said that early discussions about surgery facilitated patient acceptance and helped to reduce fear. Providing verbal explanations or written information about the types of surgery available [e.g. Crohn's & Colitis UK (CCUK) leaflets or booklets developed locally in their NHS trust] were the typical approaches at this stage:

We give them all treatment options, we would even discuss surgery as well obviously, I haven’t mentioned that. But it’s always discussed with patient and the first-line treatment, if they want it, we always say that surgery should be done as a managed therapy basically rather than it being an acute reaction, it is better for patients, that’s what we try and do, doesn’t always work . . .

IBD nurse 2

I mean biologics has changed it an awful lot but you have to try and plant the seed that surgery might be required, quite early I think and so we do try and do that but I think if a patient is sort of heading in that direction I will sort of say that you know look this is a possibility, you need to be filing it in your head so yes.

IBD nurse 3

Health-care professionals also described situations in which surgery may become more appropriate or necessary, such as when patients had tried one or two biologic therapies and their symptoms had not improved, and this was often described as having exhausted all medical treatments or ‘running out of options’. Some health-care professionals also talked about presenting surgery as an option specifically for patients who are resistant to steroids. At this point, many health-care professionals explained that they would refer patients either to surgeons as part of a joint medical–surgical clinic or for more in-depth individual discussions with a local surgeon or stoma nurse. A minority of health-care professionals also explained that their IBD services maintain a list of patients who have previously had surgery for UC and are willing to discuss their experiences with other prospective candidates:

There’s always, there’s, you know, there’s nearly, I mean sometimes when you’re through everything, then there isn’t a choice, but I think we’re quite proactive about surgery, we offer surgery relatively early in addition to all of the other things.

Consultant gastroenterologist 5

We have a joint medical–surgical clinic that patients come to and they are also seen individually by the surgeon to see and talk them through the logistics of it, and they also get to see a stoma nurse and they also get to see a patient who’s had a similar operation, so they can speak to somebody who has had the experience of it.

Consultant gastroenterologist 8

In keeping with their perspectives on medical treatment options, health-care professionals also frequently mentioned the importance of respecting patient preferences and patient choice, and what patients are able to tolerate, in terms of their symptoms, possible toxicity from medication, quality of life and socially (i.e. post surgery). Patient age was another important factor influencing referral for surgery, given that younger people often have more concerns about living with a stoma. Likewise, for women who want to start a family, health-care professionals would encourage them to go through pregnancy before opting for surgery. Ensuring that patients have the necessary information about surgery and are comfortable with the decision they are making was important to health-care professionals:

I think one of the things that I would say is I wouldn’t, I can advise people to have an operation but I wouldn’t bully someone into an operation because if you twist somebodies arm into having an operation when it’s uncomfortable they will always regret it.

Consultant gastroenterologist 2

. . . their choice really, sometimes it depends, patients have got different threshold to what they will accept and some patients have really very mild symptoms but really want to go ahead for surgery. You know it’s kind of a combination really, it’s not really just one thing.

IBD nurse 1

Participants

A total of 53 patients were approached for interview; however, it was not possible to establish contact with 11 people, contact was lost with three people (despite initial interest), four people failed to attend their prearranged interview, one person was ineligible at screening and one person was too busy to participate in an interview.

Therefore, a total of 33 people with UC participated in the qualitative interviews (Table 15). There was an equal split in the gender of participants (51.5% female) and the majority of participants were white British (81.8%). The median age of participants was 39 years and the median time since diagnosis of UC was 6 years. The participants were also sampled on self-reported current and previous treatments (in addition to steroids, which all patients had previously received). At the time of the interview, 13 patients were receiving more than one treatment, and the median number of previous treatments was 2 (minimum–maximum 0–5). The median length of individual interviews was 31 (minimum–maximum 16–74) minutes.

Analysis

Through familiarisation with the interview transcripts, a preliminary list of codes was developed. After several iterations, it was possible to agree on the coding framework. The coding framework included 49 codes that were grouped under 10 categories (see Appendix 4, Table 36). The coding framework was applied to the whole data set and summaries of each code were produced to help identify themes. The presentation of the findings in this chapter is structured around these themes and subthemes, as follows.

Lived experiences of ulcerative colitis

Experiences of steroids

Most participants explained that they had become steroid resistant or dependent during their treatment, with steroids not relieving their symptoms at all or relieving them only at higher doses and becoming less effective as the dose was tapered off. Some participants also said that, even when they felt that the steroids ‘do work’ for them, their symptoms were only dulled, but still present:

I think they stopped, I felt like they’d stopped working because I was getting symptoms again and I wasn’t, my symptoms wasn’t improving, so when I had like low symptoms, I’ve had the diarrhoea and stuff, I expected to take steroids and it be OK again but it didn’t it just sort of carried on. So, I just sort of knew they wasn’t really working.

P1

I do appear to be quite steroid dependent; when the dose starts to reduce, the symptoms start to come back.

P3

It stopped working, just my symptoms came back so when I went on the lower dose the blood, so the mucous stools, they’d stopped on the higher dose but the second I’d start tapering down, so from, I think I started on eight tablets twice, second I get down to about six or five, I’d start seeing the symptoms coming back again, so the mucousy stools and stuff like that so.

P30

In addition to this, most participants reported the side effects or adverse events they experienced while taking steroids. These included increased hunger and weight gain (i.e. ‘moon face’), altered mood, anxiety, joint pain and sickness. A small number of participants also said that because of these side effects they felt that they would rather manage their UC without steroids:

Well, I suffer with anxiety and depression and I’m medicated for that and it [steroids] made me even worse.

P28

I was like I’m sick of being on these tablets cause for me the worse side effects came from steroids. I didn’t really feel it the other side but my hair was falling out a lot but steroids as soon as I went on steroids I knew I was going to sleep, I’m going to be achy, I’m going to be hungry, I’m going to have a fat face so they were like devil tablets in the end.

P32

I still have issues but because I don’t want to go to another medication, I’m happy to take those you know conditions you know, rather than to go on a steroid.

P13

Reasons for treatment changes

All participants had tried several different treatments in addition to, or subsequent to, steroids and so were able to reflect on those experiences and the main reasons why they had changed treatment. In general, the decision to move from steroids to other treatments was not sharply distinguished from other treatment changes during their disease course for participants in this study, and participants reported three main reasons for changing treatment: (1) reduced effectiveness, (2) lack of response and (3) side effects.

Factors influencing treatment choices

Participants were able to identify and explain factors that influenced their choices of treatment for UC over time. With the exception of decisions about surgery, treatment choice was described in general terms and participants did not necessarily differentiate between the treatments following steroid resistance or dependence, relative to other decision points. The main factors influencing treatment choices reported by participants were treatment effectiveness, guidance from health-care professionals, route of administration and treatment convenience, safety and side effects, and ‘running out of options’. Participants also raised a number of other issues, which are summarised in this section.

Perspectives on surgery

Most participants referred to surgery as a ‘last resort’ treatment for UC, as they wanted to try other medical treatments first. Following diagnosis of UC, and often towards the beginning of their treatment journey, participants typically did not consider surgery a viable treatment option because they were aware that medical options were available. Some participants had negative views of surgery due to the stigma of having a stoma, as well as concerns about the cosmetic appearance of this. For this reason, some participants explained that they preferred the idea of an ileo-anal pouch, as opposed to a permanent stoma:

The last one was the colostomy bag, you see. So I wanted to try everything else before I go down that route and he [consultant] was quite happy with that but yeah.

P10

I just didn’t want to go down that road, I just didn’t want to feel that I was gonna be carrying my toilet around on the side all my life.

P4

Oh yes, I would probably jump off a building before I had a stoma bag. Like, my mental health would not cope with having a stoma bag . . .

P29

However, participants reported these concerns to be less important as surgical options became more likely (because of the severity of their condition), or indeed post surgery. Participants who had discussed the possibility of surgery with their IBD team or had already had surgery explained how they had become less resistant to and more accepting of surgery as their medical treatment options were ‘running out’. Talking to people who had already had surgery, discussing this with their health-care professionals and family and friends, and reading information from CCUK or NHS information leaflets were all helpful in facilitating the decision to have surgery:

As I say, looking back, you know, I think I’m the kind of person that was glad that we tried all the alternatives, you know, surgery is not something I ever want, but at the same time, you know, I can look back and say, there was no choice.

P16

Initially it was ‘oh no! you know I don’t want surgery!’ you know, that’s a bit drastic, and you know, now I would consider it, so yeah, it’s definitely changed.

P8

I had a fear of that [surgery] at one time and then I read up a lot of people that had had it done on the internet and everything, and also with the Crohn’s [and Colitis UK] leaflets and pamphlets about it all and I sort of got it in my head, I’d accepted it and I was quite happy with it.

P2

Participants explained that, when surgery had been mentioned as an option, their consultants had given explanations of the different types of surgery available and some had been referred for meetings with stoma nurses to find out more information about how to manage their UC post surgery. Participants also said that surgery was mentioned at different time points. Some participants had been made aware of surgery soon after their diagnosis and so they had always been aware of this option, whereas others had become aware of the surgical options only when the pharmaceutical options had been exhausted and still others (usually patients with less severe UC and early in the disease course) were not aware of the surgical options. This variety of experiences is in keeping with the variation in duration of disease and timing of diagnosis, as well as, presumably, divergent clinical practice even within the three IBD centres:

. . . she did mention it [surgery] a couple of times but what she would just say is, you know, ‘if your options run out, that’s what we’d do, but we’ll always try every options first’.

P6

They don’t just stick leaflets in your face; they do give you other areas to get advice and support.

P28

No, no as my nurse says, we’ll just take one step at a time, let’s try all options and then it’s just only ’cos I’ve read up on it, and that’s how I know it could be a possibility.

P19

Other issues

Patients raised several other issues during the qualitative interviews, including reflections on their relationships with their GP, how they involve their families in decision-making about treatments, use of alternative treatments and longer-term views on treatment choices.

Sample characteristics

A total of 116 participants completed the survey. The response rate was 63.7%, based on the respondents who visited the survey platform and not on the number of who saw the e-mail or social media invitation. The median age of respondents was 46 years, and 57% of the respondents were male (Table 16). Health-care professionals had a median of 9 years of experience in practice. Approximately 60% of the respondents were consultant IBD specialists or gastroenterologists and 64% were primarily working in secondary referral NHS hospitals. All regions of the UK were represented in the DCE, except Northern Ireland.

Understanding and engagement

All participants completed all the choice tasks, and for the remainder of the survey the number of missing data was small and did not exceed 5%. The median time taken to complete the survey was 8 minutes. All survey participants answered the dominant choice question correctly, and the majority (94.64%) reported that they agreed or strongly agreed that they understood the DCE tasks (Table 17). However, half of the sample said that they needed more information than was provided in the DCE tasks, for example further details about patient preferences, patient compliance, patient history and patient demographics.

Modelled preferences

We found that all five attributes in the model were statistically significant at predicting the probability of choosing a preferred treatment, implying that health-care professionals consider all five important when choosing a treatment to offer to patients with steroid-resistant UC. Figure 7 depicts the coefficients of the conditional logit model and shows that induction of response, mucosal healing and maintenance of remission have positive coefficients, which implies that higher levels of these attributes will increase the uptake of a treatment. By contrast, risk of lymphoma and risk of serious infection have negative coefficients, which implies that higher levels of these attributes dissuade health-care professionals from choosing a treatment. Figure 7 shows that a 1-unit increase in both lymphoma risk and serious infection risk attributes were strongly considered when health-care professionals were choosing a treatment compared with a 1-unit increase in the symptom improvement attributes. However, as the attributes were not measured in the same units (i.e. a 1-unit increase in lymphoma is measured in 10,000 patient-years, but a 1-unit increase in response is measured in percentages), the coefficients are not directly comparable.

FIGURE 7.

Model results of attributes for health-care professional DCE. The figure shows the coefficients and standard errors from the conditional logit model. All attributes are significant (p < 0.01). The raw coefficients can be found in Appendix 5, Table 37.

The willingness to trade benefit and risk

The benefit–risk trade-offs that health-care professionals were willing to make are presented in Table 18. The larger trade-offs in risks indicate that health-care professionals viewed that attribute as relatively more important than other attributes in their decision-making. Table 18 shows that maintenance of remission was the most important symptom improvement attribute, followed by induction of response and mucosal healing. On average, health-care professionals were willing to accept a lymphoma risk of 4 cases per 10,000 patient-years if the medication would increase patients’ induction of response by 10%. The lymphoma risks that health-care professionals are willing to accept for a 10% improvement in remission is 5 cases per 10,000 patient-years, whereas the lymphoma risk that health-care professionals are willing to accept for a similar increase in mucosal healing is 2 cases per 10,000 patient-years. This shows that the maintenance of remission is approximately twice as important as mucosal healing when deciding which treatment to offer to patients with steroid-resistant UC. Similar results were found with risk trade-offs calculated using serious infection risk, as the regression coefficients for lymphoma risk and serious infection risks were similar.

Exploratory analyses were also conducted to evaluate preferences across subgroups, such as health-care professional type (consultants vs. specialist nurses) or place of work (secondary vs. tertiary hospitals) (see Appendix 5, Table 37). Similar to the full sample results, in the secondary and tertiary hospitals subgroup analyses we found that maintenance of remission was the most important symptom improvement attribute, followed by induction of remission and, finally, mucosal healing. Data also suggest that health-care professionals who work at tertiary hospitals have a higher risk tolerance than those who work at secondary hospitals. For example, for a 10% improvement in remission, the lymphoma risks that health-care professional at a secondary hospital are willing to accept is 4 cases per 10,000 patient-years, whereas health-care professionals at a tertiary hospital are willing to accept a higher risk of 7 cases per 10,000 patient-years. This suggests that health-care professionals working at tertiary centres are willing to accept higher risks in exchange for improvements in all three symptom improvement attributes.

Probability of uptake

When we applied the results from our model to predict how the respondents would prescribe four currently used treatments for patients with steroid-resistant UC, we found that uptake rates were highest for infliximab (62%), followed by tofacitinib (18%) and vedolizumab (15%). The probability of uptake was lowest for adalimumab (5%) (Table 19).

Figure 8 shows the change in uptake rates of improved treatment profiles compared with a baseline treatment that contains the worst possible levels for each attribute. Larger changes in uptake rates indicate the attributes that health-care professionals prioritised more when choosing a treatment. For example, a treatment with 70% remission rate was ranked very favourably, with a change in uptake rate of 92%, followed by a treatment with a risk of infection of 1 case per 100 patient-years, with a change in uptake rate of 69%. When comparing the attribute levels, some were interspersed; for example, a 60% response rate was at the top end of the priority list, but a 50% response rate was at the bottom end. However, the data show that remission attribute and risk of serious infection attributes were prioritised over risk of lymphoma and mucosal healing attributes when health-care professionals were choosing a treatment.

FIGURE 8.

Uptake rates of a baseline and improved treatment profile. Note that each vertical line represents the change in uptake rate of moving from a baseline (worse) treatment profile to an improved treatment profile. The baseline treatment profile was selected to represent the worst possible treatment profile, where induction of response was 40%, mucosal healing was 40%, remission rate was 35%, risk of lymphoma was 8 cases per 10,000 patient-years and risk of serious infection was 10 cases per 100 patient-years.

Sample characteristics

A total of 720 invitation letters were sent to eligible participants. Of the invited participants, 166 visited the Qualtrics survey platform, 115 completed the survey and 5 declined to complete the survey. Based on the respondents who visited the survey, the response rate was 69%. Table 20 presents the sociodemographic characteristics of the respondents. The median age of respondents was 41 years, and 52% of the responders were female. The majority of patients were white, employed and educated to secondary General Certificate of Secondary Education level or above.

Table 21 presents the clinical characteristic of the sample, including current and previous medical treatments. Among the 115 patients, the median time since diagnosis was 10 years. The median IBD Control-8 score was 13, with 45% of the sample reporting poor control of the disease at the time of the survey. The median quality-of-life score was 0.77, which is lower than the UK general population mean quality-of-life score of 0.85. 80 The majority of patients had previously received steroids (93%), thiopurines (70%) and biological therapies or tofacitinib (70%). The most commonly used biological therapy was infliximab (51%). At the time of the survey, one-quarter of the sample was taking a combination of treatments [i.e. biologics/thiopurines/5-aminosalicylic acids (5-ASAs)].

Understanding and engagement

All participants completed all choice tasks. For the remainder of the survey, missing data were few and did not exceed 5%. The median time taken to complete the survey was 21 minutes. Ninety-seven per cent of survey participants answered the dominant choice question correctly and only four participants failed this internal validity test. Overall, the majority of participants understood the DCE (92%) and ranking tasks (91%), and agreed that the information provided in the DCE task was appropriate (70%) (Table 22).

Modelled preferences

All attributes have a significant influence on the choice of treatment patients preferred. Table 23 and Figure 9 show the relative changes that make respondents more likely (i.e. a positive coefficient) or less likely (i.e. a negative coefficient) to take a treatment. Higher levels of the remission attribute strongly influenced respondents’ choice of treatment (β 0.065; p < 0.01). Similarly, a treatment having a lower likelihood of side effects [i.e. a change from very common to very rare (β 2.937; p < 0.01) or from very common to uncommon (β 2.260; p < 0.01) or from very common to common (β 1.417; p < 0.01)] strongly increased the likelihood of the respondent choosing it. An increase in the induction of response from a lower level (40%) to a higher level (i.e. 50% or 60%) made respondents more likely to take the treatment. However, patients are less likely to choose a treatment that takes longer to improve their symptoms (β –0.145; p < 0.01). In addition, taking a pill daily at home (β 0.848; p < 0.01) and injections at home every 8 weeks (β 0.541; p < 0.01) were preferable to receiving infusions at hospital every 8 weeks. Notably, a change from infusion at hospital to injection every 2 weeks at home was not significant (β –0.029; p = 0.85), suggesting that patients found these two ways of administering the treatment comparable.

FIGURE 9.

Conditional logit model results reproduced in a diagram. Note that the figure shows the coefficients and standard errors from the conditional logit model. All attributes are significant (p < 0.05), except injection every 2 weeks at home. Light-blue points represent the reference category of categorical variables. Speed of response and maintaining remission are continuous variables and represented by a single coefficient.

Predicted probabilities

To compare the relative importance of treatment attributes, we calculated the change in predicted probabilities in a baseline scenario compared with a new scenario. In the baseline scenario, the treatment profile consisted of the worst possible levels for all of the attributes and in each new scenario only one attribute level was improved at a time to find out how important that attribute is when selecting a treatment (see example calculation in Appendix 6). Figure 10 shows the change in uptake rates of the improved treatment profiles. Larger changes in uptake rates indicate more favourable treatment attributes. For example, a treatment with very rare side effects was ranked very favourably, with a change in uptake rate of 90%, followed by a treatment with a 70% remission rate that had a change in uptake rate of 81%. When comparing the change in probabilities across these scenarios, a treatment with very rare side effects was ranked very favourably, followed by a treatment with a remission rate of 70% and treatments with high induction of response rates. Conversely, route and frequency of administration were seen as less important.

FIGURE 10.

Uptake rates of a baseline and improved treatment profile. Note that each vertical line represents the change in uptake rate of moving from a baseline (worse) treatment profile to an improved treatment profile. The baseline treatment profile was selected to represent the worst possible treatment profile, where induction of response was 40%, speed of response was 14 weeks, remission rate was 35% and mode of administration was injection every 2 weeks at home, with very common side effects.

Ranking treatments

When participants were asked to rank the four biological treatments in order of preference, where ‘1’ equated to most preferred and ‘4’ equated to least preferred, the most preferred treatments were infliximab (38%) and tofacitinib (38%), followed by vedolizumab (17%) and adalimumab (6%) (Figure 11). The least preferred medication was adalimumab (54%), followed by vedolizumab (26%), infliximab (11%) and tofacitinib (9%). When respondents explained their choices, the most common reasons were the effectiveness of the drug at inducing and maintaining remission, the side effects, convenience and speed of response. Further analyses also showed that if participants have prior experience of taking any one of the four treatments, then they are more likely to rank that treatments more favourably than those who have no experience of taking the treatment (see Appendix 6). Nevertheless, a subgroup (n = 35) of biologic-naive patients were analysed separately to see if their ranking differed from that of those who were taking biologics. Among the biologic-naive subgroup, the most preferred treatment ordering changed slightly, with tofacitinib ranked first (46%), followed by infliximab (37%), vedolizumab (11%) and adalimumab (6%). The least preferred treatments were adalimumab (60%), vedolizumab (26%), infliximab (9%) and tofacitinib (6%), and this ordering was similar to the full sample.

FIGURE 11.

Biologic therapies ranked in order of preference.

Participants

A total of 36 people (medics, n = 16; nurses, n = 10; patients, n = 10) were invited to take part in the multistakeholder workshop. It was not possible to make contact with 15 people and seven confirmed that they were unable to attend because they were too busy with work commitments (whether clinical for health-care professionals or otherwise for patients), and five health-care professionals confirmed their places at the workshop but were unable to attend on the day because of urgent clinical pressures.

Nine people (two people with UC, three consultant gastroenterologists and four IBD nurses) participated in the multistakeholder online workshop on 11 March 2021. Further demographic information was not collated from workshop participants, as the focus was on sharing and discussing the research findings, as opposed to further understanding their experiences as health-care professionals or as patients.

Summary findings

The purpose of the online multistakeholder workshop was to bring together patient representatives and IBD health-care professionals to present them with the key findings from the PoPSTER study, to seek their feedback on the results and to work with them to generate recommendations for research and practice about steroid-resistant UC. To that end, a presentation summarising the methods and results from WPs 1–4 was generated by the team and shown to participants at the beginning of the workshop to orient them to the key findings. In addition to this, prior to the workshop, the key findings from each WP were reviewed by Elizabeth Coates, Nyantara Wickramasekera and Alan Lobo to identify the main areas of interest and convergence between the different studies. Triangulating the findings in this manner helped to provide a focus to the overall findings from the PoPSTER study and these findings were converted into summary statements that could then be considered by participants during the small group discussions session of the workshop. The findings were summarised as bullet points under four headings, as follows, and as shown in the first column of Table 39 in Appendix 6.

Outputs from the workshop

Group 1 was tasked with considering the findings on topics 1 and 2 (i.e. definitions of steroid resistance and factors influencing treatment choices).

Discussion

Taken as a whole, the PoPSTER study provides a number of helpful insights into practice and patient and health-care professional preferences for treatments for steroid-resistant UC. The findings of this programme of work have a number of important implications for practice. We have demonstrated variation among health-care professionals in terms of definitions of steroid-resistance in UC, the importance of differentiating steroid resistance from steroid dependence, and preferences for different treatments for steroid-resistant disease. Current literature indicates either 2 or 4 weeks as the treatment duration required to determine steroid resistance. 51,52 The majority of respondents in the survey indicated that they considered 2 weeks the required duration, but a minority felt that treatment should continue for 2 further weeks before a patient is considered steroid resistant. The efficacy of a therapeutic intervention, both in practice and in clinical trials, may depend on this timing. Efficacy and safety in those resistant at 2 weeks may be different from efficacy and safety in those who have received 4 weeks of steroids, and greater steroid use is considered a disutility of care and should be avoided. 29

However, the qualitative interviews with health-care professionals also provided insight into how such guidelines might be interpreted at the level of the individual patient, for example when time thresholds might be extended or reduced based on the clinical picture. This form of variation can be captured in guidance, and the findings from this research help to inform this from the perspectives of both clinicians and patients. More generally, research shows that clinicians often deviate from practice guidelines for legitimate biomedical, patient-centred and contextually specific reasons. 81,82 Understanding these reasons for deviation can inform the development of guidelines that are better able to accommodate tailoring of management.

Anti-TNF agents were the most frequently preferred agent for steroid-resistant and steroid-dependent scenarios. However, there are few data to compare agents ‘head to head’. It is not clear whether this is a result of how long anti-TNF agents have been available and, therefore, whether health-care professionals are familiar with them or perceive benefits over other agents. However, in an assessment of the uptake probability of individual drugs in which drug attributes were matched to clinician preferences as part of the DCE, infliximab was the most preferred drug, followed by tofacitinib. Whether or not this pattern of use represents best practice is unclear, and the role that agents such as vedolizumab or tofacitinib should play needs to be further explored through additional direct comparisons in the steroid-resistant cohort as a whole or through clarification of when these agents might be the preferred option (e.g. in higher-risk subgroups, including elderly people, those with previous cancer or those with concomitant immunosuppression with thiopurines, who would normally be treated with anti-TNF agents). Again, the qualitative study provided insight into this, with some clinicians advocating treatment with biologics before the use of thiopurines, suggesting that they perceive tofacitinib and biologic treatments to be equivalent.

In the DCE, patients also selected infliximab as the preferred option based on its attributes, alongside tofacitinib. This may be because infliximab is the agent that has been available the longest and, therefore, the agent of which most patients have had experience. Among biologic-naive patients, tofacitinib was most preferred option. Clinicians should combine this evidence on patient and professional preferences with trial and ‘real-life’ data on these agents when choosing treatment for an individual.

Steroid dependence (i.e. the recurrence of symptoms on steroid dose reduction) was differentiated from steroid resistance by health-care professionals in both the survey and the qualitative studies. The distinction may not change treatment in some circumstances, with anti-TNF therapy being the most frequently chosen treatment option in both steroid-resistant and steroid-dependent disease. For those patients on thiopurines, during tapering the steroid dose at which symptoms recur did not affect the preference for anti-TNF, suggesting that, in this situation, the distinction did not affect treatment choice. However, among those patients who were naive to thiopurines, the uptake was less at lower steroid doses at the time of symptom flare, and this is presumably because clinicians would increase the dose of steroids again so that a thiopurine could be introduced. The potential for increased steroid use needs to be considered in this approach.

In addition, we have demonstrated that prior immunosuppression has an impact on the frequency with which biologics and tofacitinib are used, indicating that clinicians seem likely to want to have tried conventional immunosuppression with thiopurines before using biologics (i.e. a ‘step-up approach’ to treatment). The earlier use of biologics, without prior use of thiopurine immunosuppression, might reduce side effects relating to thiopurines and reduce long-term tissue damage. An approach in which previously thiopurine-naive patients receive thiopurines before receiving a biologic or tofacitinib may also encourage greater steroid use, that is, as a bridge to cover the 6 weeks until thiopurines are effective. Recognising this influence is, therefore, important in considering whether or not a paradigm shift away from thiopurine use is needed. Clinical guidelines might address the optimum approach, but evidence to support recommendations may be missing at this time.

Detailed understanding of patients’ preferences in their treatment and their decision-making has also been obtained. In the qualitative study, treatment effectiveness was cited as the primary concern of all participants when choosing a new treatment. The participants explained that the capacity of a treatment to alleviate symptoms and improve their quality of life was the most important driver of their preferences. Although participants explained that the route of administration and side effects were important influences on their treatment preferences, they generally placed limited value on this relative to treatment effectiveness when describing their experiences. It was also evident that patients’ preferences change over time and there was an increased willingness to try alternative treatments and, eventually, surgical options for UC according to the severity and duration of symptoms and, crucially, as medical treatment options were exhausted. This is the first qualitative study to explore patient preference for treatments in steroid-resistant UC and so the findings are unique in their qualitative perspective on this, but are in keeping with a recent DCE. 83 More broadly than this, the qualitative study also highlighted the wider concerns of the patients and the stigma and challenges they face in daily life with UC, which were very much in keeping with previous qualitative research in this area. 84–86

In contrast to findings from a recent systematic review,84 the importance of the relationship with the IBD team was clear in the patient qualitative study, with participants explaining how IBD health-care professionals guided their treatment discussions and choices. Most participants in the qualitative study described valuable relationships with their IBD nurses and medics, and said that they trust and respect the clinical expertise of these professionals. This finding accords with recommendations and standards regarding the role of multidisciplinary care, emphasising the need for such teams to supervise care and for clinicians to be part of these teams. 51,55

Several previous studies have quantified patient preferences for treatment through the use of DCEs with IBD patients. 87 The results of different studies, however, are not always comparable because different DCEs include diverse attributes for different severities of UC (e.g. 5-ASA vs. biologics vs. surgery). 87–91 To the best of our knowledge, our patient DCE is the only study that focuses on a steroid-resistant scenario conducted in the UK, but close comparative scenarios have evaluated treatments for moderate to severe UC. Almario et al. 83 conducted a DCE of patients with IBD selecting biologics as a treatment option in the USA and found that efficacy was the most important attribute for patients with UC and that side effects were the key priority factor for patients with Crohn’s disease. 83 In a DCE study by Boeri et al. conducted in the USA,92 for patients with moderate to severe disease, the authors reported that symptom improvement and risk of malignancy were the two most important attributes. Similar to our study, Boeri et al. 92 also reported that patients preferred oral route of administration to subcutaneous or i.v. administration.

The DCE studies also gave important information about the trade-offs that both clinicians and patients are prepared to make when choosing treatments. The importance that both groups placed on treatment risks should be taken into account and is perhaps greater than expected. Even low absolute risks and small differences in risk have been found to be important. Health-care professionals, therefore, need to be clear in communicating these risks to patients.

Health-care professionals expressed the importance of considering surgery as an elective option at any point in the treatment pathway. In the multistakeholder workshop, all participants supported this view and highlighted the importance of having clear information available to help inform the decision about whether or not to have surgery and its timing, and it might be helpful if this was in the form of an evidence-based patient decision aid. 93

The decision about treatment of steroid-resistant UC is necessarily preceded by an assessment of response to steroids. In addition to using a uniform and accepted definition, the process needs a mechanism to ensure that patients receive an assessment at an appropriate time point and agree about the nature of the assessment.

Route of treatment administration was also addressed in each of the PoPSTER studies. Unsurprisingly, oral administration was preferred by most patients in the DCE. However, perhaps surprisingly, the DCE showed that self-administered subcutaneous injections every 2 weeks were as acceptable as i.v. infusions every 8 weeks at hospital. The PAG explained that the injections, even at home, can be problematic for some people with UC, acting as a more frequent reminder of their illness.

Health-care professionals and patients at the multistakeholder workshop discussed a process for assessment. Services need to ensure that patients are assessed 2 weeks after starting a course of systemic steroids. This might be arranged by the IBD service, but there was also discussion of the role patients could have in self-reporting response. The process is also made more complex by the need to capture the responses of patients started on steroids in other parts of the health service (e.g. in primary care or emergency departments).

Assessment might also require objective evidence of ongoing inflammation beyond patients’ symptoms. This may be regarded as particularly important prior to escalating treatment and introducing biologic agents or further immunosuppression. Endoscopy provides direct and accurate assessment but is invasive. Measurement of faecal calprotectin is non-invasive but less sensitive and less specific. Therefore, guidance is needed on (1) the circumstances under which treatment can be escalated and a definition of steroid resistance given based on symptoms alone, (2) when additional non-invasive assessment by faecal calprotectin would suffice and (3) when endoscopic assessment is required.

The use of a multidisciplinary workshop using the ‘so what/now what’ approach was a useful method to synthesise questions. By the end of the study, the research team were quite familiar with the findings and were able to distil these into key points for discussion, which meant that preparation was not onerous. Breaking down the topics into manageable sections meant that we were able to hold focused discussions and could complete these in the time allocated. In addition, by ensuring a mix of health-care professionals and patient representatives in the group, the recommendations were kept relevant to all stakeholders. We believe that this method could be used in similar work to set priorities.

Strengths and limitations

Recommendations for future research

Following discussions with health-care professionals and patients at the multistakeholder workshop, several areas of further research are recommended as follows:

• A parallel-group individually RCT comparing infliximab and tofacitinib for people with steroid-resistant UC would be acceptable to both patients and health-care professionals. The primary outcome of interest would likely be the presence of clinical remission, as used in previous similar research. 41,44,47 Other outcomes, such as the IBD-specific health-related quality of life (e.g. the IBD Control-8 or IBD Questionnaire),75,104 would also be included.

• Further work to reach a consensus on the clinically meaningful definition of steroid resistance in UC is recommended to build on the findings from this study. A nominal group technique or Delphi study with health-care professionals would be advisable to address the uncertainty and to clearly define steroid resistance.

• Greater understanding of how tofacitinib can be used optimally to treat patients with steroid-resistant UC is still needed. There are several unanswered questions: can tofacitinib be used alone without the need for concomitant steroids? Can tofacitinib be used in the treatment paradigm where azathiopurine currently sits? A detailed practice survey of health-care professionals would provide a useful first step to exploring these questions.

• Further understanding of the optimal position of surgery within a pathway for steroid-resistant UC is needed. The multistakeholder workshop recommended several related research questions: do patients have better outcomes if they have earlier discussions with the surgical team? What are the decision support needs of patients considering surgery? What are patients’ perceptions of surgery in UC? Do gastroenterologists and IBD nurses feel comfortable in talking about surgery and where do they get their surgical education?

• During the multistakeholder workshop, there was also a desire for a RCT comparing surgery and medical therapy for steroid-resistant UC; however, patient attitudes and experience might make this difficult to deliver. Therefore, a cohort-based assessment may be more useful and feasible initially.

Conclusions

To the best of our knowledge, this is the largest programme of work to address both clinician and patient preferences and attitudes to steroid-resistant UC. This study has been carried out using quantitative and qualitative methods to provide a detailed and nuanced picture to inform practice and research.

Variation in the very definitions used for steroid-resistant and steroid-dependent disease have been highlighted, as well as the variations in drugs used, the influence of previous treatment, methods of assessment and the role of surgery. These findings have been set in the context of individual health-care professionals’ and patients’ experience and values in qualitative work to better understand how principles and guidance are best operationalised. The consistent influence of prior thiopurine immunosuppression on treatment choices is of particular importance in terms of future treatment paradigms. More general principles that enhance quality of care and patient experience are also evident, including information, support and access to a MDT.

This research has also gone further to use, to our knowledge for the first time, DCEs to understand the trade-offs that health-care professionals and patients are prepared to make in choosing treatment. These DCEs have highlighted the importance that both groups place on risks of treatment and perhaps more than is widely appreciated, even when absolute risks are low and differences in risk are low. The responses have allowed extrapolation to specific drugs that might be preferred and, although there are clear factors that confound those results, the information can usefully form the basis for choosing treatment in clinical practice and for designing trials, both in terms of choice of intervention and outcomes.

Future research and clinical practice developments in the management of steroid-resistant UC can, therefore, utilise the methods used in this study, as well as its findings.

Co-applicants

Professor Alan Lobo (chief investigator, Consultant Gastroenterologist and Professor of Gastroenterology, Sheffield Teaching Hospitals NHS Foundation Trust), Professor Christopher Probert (Professor of Gastroenterology, University of Liverpool), Professor Shaji Sebastian (Professor of Gastroenterology, Hull University Teaching Hospitals NHS Trust), Dr Elizabeth Coates (Research Fellow, University of Sheffield), Dr Aoife Howard (Postdoctoral Researcher, National University of Ireland, Galway), Professor Daniel Hind (Professor of Evaluation and Assistant Director, University of Sheffield), Dr Phil Shackley (Reader in Health Economics, University of Sheffield), Mr Matthew Lee (National Institute for Health and Care Research Clinical Lecturer in General Surgery, Sheffield Teaching Hospitals NHS Foundation Trust and University of Sheffield), Nicola Totton (Medical Statistician, University of Sheffield), Sue Blackwell (patient representative), Nicola Dames (patient representative) and Hugh Bedford (patient representative).

Study Management Group

Professor Alan Lobo (chief investigator, Consultant Gastroenterologist and Professor of Gastroenterology, Sheffield Teaching Hospitals NHS Foundation Trust), Professor Christopher Probert (Professor of Gastroenterology, University of Liverpool), Professor Shaji Sebastian (Professor of Gastroenterology, Hull University Teaching Hospitals NHS Trust), Dr Elizabeth Coates (Research Fellow, University of Sheffield), Amy Barr (Research Assistant, University of Sheffield), Nyantara Wickramasekera (Research Associate, University of Sheffield), Professor Daniel Hind (Professor of Evaluation and Assistant Director, University of Sheffield), Dr Phil Shackley (Reader in Health Economics, University of Sheffield), Mr Matthew Lee (National Institute for Health and Care Research Clinical Lecturer in General Surgery, Sheffield Teaching Hospitals NHS Foundation Trust and University of Sheffield), Nicola Totton (Medical Statistician, University of Sheffield), Sue Blackwell (patient representative), Nicola Dames (patient representative), Hugh Bedford (patient representative) and Zoe Furniss (Trial Support Officer, University of Sheffield).

Staff in participating centres

Jemima Clarke (Research Co-ordinator, Sheffield Teaching Hospitals NHS Foundation Trust), Luke Barron (Directorate Research Co-ordinator for the Academic Directorate of Gastroenterology, Sheffield Teaching Hospitals NHS Foundation Trust), Rebecca Bolton (Medical Secretary, Sheffield Teaching Hospitals NHS Foundation Trust), Sally Myers (Senior IBD Clinical Research Nurse, Hull University Teaching Hospitals NHS Trust) and Martina Lofthouse (Gastroenterology and Hepatology Research Nurse, Royal Liverpool and Broadgreen University Hospitals NHS Trust).

Patient Advisory Group

Sahara Fleetwood-Beresford, Joseph Greenly, Emma Greenwood and Stephie-Jayne Simpson.

Study Steering Committee

Professor Tariq Iqbal (chairperson of committee and Consultant Gastroenterologist, University Hospitals Birmingham NHS Foundation Trust), Jayne Kranat (public representative), Cath Stansfield (Nurse Consultant Gastroenterology, Salford Royal Hospital NHS Foundation Trust), Dr Laura Ternent (Senior Lecturer in Health Economics, Newcastle University) and Gabrielle Thorpe (Lecturer, University of East Anglia).

Contributions of authors

Elizabeth Coates (https://orcid.org/0000-0002-2388-6102) (Research Fellow) contributed to the conception, design and delivery of the overall project, the acquisition and management of data in all WPs, analysis and interpretation of data in WPs 1–3 and 5, and drafted the report.

Nyantara Wickramasekera (https://orcid.org/0000-0002-6552-5153) (Research Associate) contributed to the acquisition of data in WPs 2–4, with particular responsibility for design, delivery and analysis of the DCEs, drafted the DCE sections of the report and commented on drafts of the report.

Amy Barr (https://orcid.org/0000-0002-7990-7451) (Research Assistant) contributed to the delivery of WPs 1–3, to the acquisition and management of data in WPs 1–3 and to the analysis and interpretation of data in WPs 2 and 3, and commented on drafts of the report.

Phil Shackley (https://orcid.org/0000-0002-1862-0596) (Reader in Health Economics) contributed to the conception, design and delivery of the overall project, supervised the DCEs and commented on drafts of the report.

Matthew Lee (https://orcid.org/0000-0001-9971-1635) (National Institute for Health and Care Research Clinical Lecturer in General Surgery) contributed to the conception, design and delivery of the overall project, supported analysis and interpretation of data, and commented on drafts of the report.

Daniel Hind (https://orcid.org/0000-0002-6409-4793) (Professor of Evaluation) contributed to the conception, design and delivery of the overall project, supervised analysis and interpretation of data, and commented on drafts of the report.

Christopher Probert (https://orcid.org/0000-0003-4550-0239) (Professor of Gastroenterology) contributed to the conception, design and delivery of the overall project, acted as local principal investigator for the NHS components of WPs 3 and 4, and commented on drafts of the report.

Shaji Sebastian (https://orcid.org/0000-0002-3670-6545) (Professor of Gastroenterology) contributed to the conception, design and delivery of the overall project, acted as local principal investigator for the NHS components of WPs 3 and 4, and commented on drafts of the report.

Nikki Totton (https://orcid.org/0000-0002-1900-2773) (Medical Statistician) contributed to the conception, design and delivery of the overall project, contributed to analysis and interpretation of data in WP 1, and commented on drafts of the report.

Sue Blackwell (https://orcid.org/0000-0002-2819-3727) (Patient Representative) contributed to the conception, design and delivery of the overall project, and commented on drafts of the report.

Hugh Bedford (Patient Representative) contributed to the conception, design and delivery of the overall project, and commented on drafts of the report.

Nicola Dames (Patient Representative) contributed to the conception, design and delivery of the overall project, and commented on drafts of the report.

Alan Lobo (https://orcid.org/0000-0001-8037-793X) (Professor of Gastroenterology) contributed to the conception, design and delivery of the overall project, supervised the project as chief investigator and acted as local principal investigator for the NHS components of WPs 3 and 4, and drafted the report.

Publication

Wickramasekera N, Coates E, Barr A, Lee MJ, Blackwell S, Bedford H, et al. Patient preferences for treatment in steroid resistant ulcerative colitis – a discrete choice experiment. Scand J Gastroenterol 2022;57:797–806.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the HTA programme or the Department of Health and Social Care.

Examples of predicted treatment uptake calculations

Examples of change in predicted uptake rate calculations

Examples of change in predicted uptake rate calculations