Background

Rheumatoid Arthritis (RA) is a progressive inflammatory disease, which causes pain, joint damage and disability, and affects 0.5–1% of the adult population [1]. Patients with RA commonly experience psychological distress in addition to their physical symptoms, and increased prevalence of depression and anxiety has been reported in RA populations [2,3,4].

Disease-specific distress (DSD) is one form of psychological distress, which has been identified in several different long-term conditions such as cancer [5,6,7], type 1 and 2 diabetes [8], and Inflammatory Bowel Disease (IBD) [9]. DSD refers to the distress or burden of living with a particular long-term condition, its symptoms and/or treatments. People with DSD exhibit signs of subjective stress that are not necessarily consistent with those of a diagnosable mental health condition [10]. This means that patients can have DSD without being diagnosed with depression or anxiety [6, 9, 11].

In diabetes, where the evidence is most established, higher levels of DSD, not depressive symptoms or clinical depression, are associated with out of target blood glucose levels [12], and interventions that effectively reduce diabetes distress can improve patients’ glycaemic control [13, 14]. Diabetes distress is an expected part of living with a complex long-term condition. It is not psycho-pathologised nor viewed as diabetes comorbidity [15]. It can be effectively addressed as part of routine disease-specific clinical care by diabetes focused practitioners [13, 15].

DSD in long-term rheumatological conditions has not yet been described in the literature. Given the progress in the detection and management of diabetes distress, and the positive clinical outcomes following effective interventions [12, 13], this current study is important for patients who experience distress when diagnosed with RA, and for their clinicians to offer evidence-based based treatment. The development of a Patient Reported Outcome Measure (PROM) to identify DSD in people with RA seems therefore timely as a first step towards comprehensive care and management.

The aims of this study were to 1. determine if there is any evidence of DSD in patients with RA and 2. if so, to develop a PROM to identify DSD in people with RA. The specific objectives delivered in five phases were to:

  1. 1

    Identify the presence, or otherwise, of DSD in four existing qualitative rheumatological diseases datasets.

  2. 2

    Validate findings with a Patient and Public Involvement (PPI) group of patients living with RA.

  3. 3

    Use the evidence from phases 1 and 2 to generate items for a PROM assessing RA distress.

  4. 4

    Assess the face and content validity and redundancy with a PPI group of patients with RA.

  5. 5

    Establish face and content validity of the PROM with scale naive participants.

Methods

Study design

This was a five-phased qualitative research study, which aimed to follow previous research methods in developing PROM’s [16,17,18,19,20] and use criteria for reporting qualitative research [21].

In Phase 1, a secondary thematic qualitative data analysis was undertaken retrospectively from 61 audio-recorded 1:1 interview transcripts from four existing data sets, with interview studies dating 2004–2015 [22,23,24,25]. Primary research questions focused on the impact of fatigue and inactivity in Idiopathic Inflammatory Myositis (IIM) and Antiphospholipid Syndrome (APS), and the experiences, expectations and needs of patients with RA about their disease management [22,23,24,25]. In Phase 2, a PPI group of people with RA (n = 4) was set up to explore and validate the findings of the thematic analysis. In Phase 3, items for a PROM were generated from the identified themes of rheumatological disease distress. In Phase 4, a PPI group of people with RA (n = 3) was consulted with the aim of establishing face and content validity of the measure and performing initial item reduction. In Phase 5, the PROM was presented to people with RA and individual cognitive interviews (n = 9) were conducted to further establish face and content validity, refine items where necessary and ensure the PROM ‘made sense’ to participants. The final draft was presented at a Patient Education Evening for patients with long-term rheumatological conditions, including RA, and carers.

Ethical considerations

All four anonymized original data sets used in Phases 1–3 had received written ethics approval [22,23,24,25]. Author H.L. was an investigator in the primary research for all four studies. Following review of the original protocols it was determined that the aims of the secondary analysis were closely aligned with the original studies’ overall aims for which written consent was previously obtained. Original participants were not therefore approached to re-consent. Transcripts from the original data sets were previously de-identified prior to the original analysis and remained that way for the further analysis. This current research was conducted prior to the stated disposal date for data collected between 2004 and 2015 [22,23,24,25].

The local University Ethics Committee provided ethics approval for Phases 3–5 on 24.04.2018 (REC Number: MRS-17/18-6443). Participants undertook written informed consent procedures. Cognitive interviews were transcribed, following removal of identifiable data, by L.S. and a professional transcription service. All audio files were deleted following transcription. All face-to-face interviews and PPI groups were held in a private room in a Medical School.

Study sample

The views and perspectives of a total of 71 people with rheumatological disease plus five carers of people with rheumatological disease were involved in phases 1–5 with specific sample detail presented in Fig. 1 and Table 1.

Fig. 1
figure 1

Flow chart of phases of development of the RADS

Full size image
Table 1 Study sample characteristics
Full size table

Phase 1 demographic data was previously reported [22,23,24,25] and not formally collected during Phases 2–5.

We aimed to recruit 3–5 participants for the Phase 2 and 4 PPI groups and 8–10 participants for the Phase 5 cognitive interviews, reflecting current research practice [26, 27].

Data collection and analysis

An overview of the five phases of scale development is shown in Fig. 1.

Phase 1: secondary analysis

Using the computer software program NVivo 12, a thematic secondary qualitative data analysis was undertaken by author O.K. retrospectively from 61 audio-recorded interview transcripts from four existing data sets [22,23,24,25]. Through three codeing phases, codes were generated which identified six broad key domains: pain, fatigue, physical consequence of disease, mood, social impact and healthcare related issues. Relevant accounts from the interview transcripts were linked to the six domains, reviewed in detail and further refined looking specifically for DSD, with authors H.L. and J.S. Data that included distress was then flagged and grouped into a common ‘theme’ (see Figs. 1, 2).

Fig. 2
figure 2

Themes of rheumatological disease distress

Full size image

Phase 2: validation of secondary analysis

A PPI group with patients diagnosed with RA (n = 4) was established to confirm whether the findings of the thematic analysis resonated with patients’ experiences of living with rheumatological conditions. The PPI group deliberations were audio-recorded and summarized by O.K.

Phase 3: initial item generation

Items for the Rheumatoid Arthritis Distress Scale (RADS) were generated from statements based on the anonymised transcripts used in the secondary data analysis, and the five themes of distress identified from that analysis (see Fig. 2). The initial list of items generated was refined by L.S. and H.L. Item order was considered at this stage by reviewing the literature [28, 29]. At the end of Phase 3, 44 items formed the first draft of the RADS.

Phase 4: PPI group

A unique group of patients with RA (n = 3) (who had not participated in earlier research phases) were invited and took part in a PPI group. The 44-item first draft of the RADS was presented to participants. Questions about the content, phrasing, structure, and layout of the PROM were explored. Participants were presented with alternative versions of the RADS to assess their preferences for scoring (numbers versus words). Field notes were taken from the PPI group deliberations, which were audio recorded and transcribed verbatim by L.S.

The same participants were contacted one week later via email to confirm changes made to the PROM [30]. At the end of Phase 4, a 35-item RADS was drafted with three different layout versions.

Phase 5: cognitive interviews

Individual face-to-face (n = 4) or telephone (n = 5) cognitive interviews were conducted with participants with RA, by L.S., to identify any difficulties in understanding PROM instructions/content or scoring. Participants were asked to ‘think aloud’ as they read each item about what they understood it to mean, and if they thought the item was relevant [31, 32]. Participants were asked to clarify how they ‘came up’ with their answer, to assess scoring discrimination [33]. These cognitive interview techniques have been described elsewhere and used in PROM development [31,32,33].

By Phase 5, three versions of the RADS had emerged; one scored by word phrases one by numbers and one had supplementary condition-related contextual questions about time since diagnosis and level of disease activity. The versions were alternated during cognitive interviews after every second participant, to assess for preference to score using numbers versus words.

At the end of each interview, participants were asked which version they preferred, and if they thought the supplementary questions were relevant. Field notes were taken for the interviews, which were audio recorded, and transcribed verbatim by L.S. (interviews 1–2) and professional transcriber (interviews 3–9). To avoid participant burden, these transcripts were not sent for checking [34].

Data saturation [26] was continuously assessed by L.S. through monitoring the quality and level of new information emerging from subsequent interviews.

Phase 5 transcripts were uploaded to NVivo 12. Codes were created for each item (1–35). Sub-codes within these were generated to group the qualitative data into ‘Discard’, ‘Keep’, ‘Rephrase’ and ‘Unsure’. Additional main codes created are as follows; ‘Scoring’ to analyse scoring preferences; ‘Supplementary Questions’ to analyse preferences on the additional questions; ‘Instructions Clarity’ to analyse if participants reported instructions were clear or not; ‘Instructions Time Recall’ to analyse preferences for the time recall period of the scale; ‘Concept of DSD’ to identify if participants validated the concept of DSD in RA. Data was coded by L.S. and reviewed by H.L.

Modifications to the RADS were finalised based on the data analysis of Phase 5.

Psychometrician consultation

An academic psychometrician was consulted twice at the University to advise on the study design, scoring, numbering of items and modifications to the RADS, and recommendations for the future steps in the PROM development.

Results

Phase 1

Figure 2 shows the five themes of rheumatological disease distress formed from the secondary data analysis. Appendix 1 shows examples of relevant accounts from original transcripts illustrating identified themes of distress.

Phase 2

Overall, participants in the PPI group agreed that the themes of distress resonated with their experiences of living with RA (see Appendix 2).

Phase 3

Initial item generation yielded 94 items, categorised in order of the five themes of distress. After removal of repetitive statements, 44 items formed the first draft of the RADS. Items remained grouped in the themes of distress, but those relating to emotional distress were placed before items of physical distress to reflect the journey of patients with RA, following diagnosis, acceptance and impact of the disease. No other item order changes were made, informed by a literature review that indicated that there was no consensus for ordering of items [28, 29].

Phase 4

Appendix 3 shows examples of relevant accounts from Phase 4 PPI group. Participants unanimously reported that the themes and concept of DSD in RA resonated with their experiences of living with the disease. An additional theme of ‘Work-related distress’ was identified, added as a sub-theme of Social Distress (see Fig. 2). A corresponding new item to this was suggested, as was a new item regarding infections.

PPI group feedback was to use word-based scoring and reduce the recall period in the PROM instructions from two months to two weeks, as a shorter duration was seen as the most reasonable amount of time one could remember accurately for scoring. Three additional supplementary questions based on time since diagnosis and disease activity were suggested, considered as contextually important for interpreting the distress score.

After the PPI group, a total of two new items had been created, 11 had been removed, and 23 had been re-phrased.

By the end of Phase 4, three versions of the RADS existed, each with the same 35 items, but differed based on word or number scoring, and the presence of supplementary questions.

Phase 5

Thirteen members of a national RA charity expressed interest, with nine participating. Four could not take part due to scheduling conflicts.

Phase 5 interview data that was uploaded onto NVivo 12 yielded the results presented below. See Appendix 4 for relevant accounts from interview transcripts.

  1. (i)

    The concept of DSD in RA

Five participants clearly validated DSD as an entity in RA and three inferred it was a valid concept when discussing certain items, for example in relation to pain.

One participant did not seem to have experienced DSD but verbalised she could see how the items were relevant and distressing for others living with RA.

  1. (ii)

    Item relevancy and rephrasing

Participants (n = 9) reported that most of the items were relevant to their personal experiences living with RA or would be to others living with RA. Participants suggested minor rephrasing of 19 items.

  1. (iii)

    Item creation

New items were suggested in relation to the theme of treatment-related, physical-related, and healthcare-related distress (n = 5).

  1. (iv)

    Item removal or combination

Five participants reported that two items regarding RA pain were similar and could be combined, and that the word ‘distress’ better explained feelings towards their pain than ‘anger’. Two disagreed.

Seven participants reported that two items concerning test results and healthcare professionals’ explanations of treatments were similar and could be combined.

  1. (xxii)

    Item order

Participants had few strong opinions on the order of items on the PROM and stated that the current sequence made sense. However, six suggested that item ‘I feel overwhelmed living with RA’ could move to the end of the scale.

  1. (vi)

    Supplementary questions

Nine, seven and six participants preferred to keep supplementary questions 1, 2 and 3 in the scale, respectively. Participants reported that the questions gave context to scoring the PROM. Data regarding participants’ specific answers for supplementary questions and disease activity scoring was not collected.

  1. (vii)

    PROM scoring

Six participants reported preferring to answer the PROM using word scoring, two preferred numbers, and one held no preference.

Participants who preferred word labels reported being able to ‘relate’ more to word scoring and helped them to consider each item more carefully than if scoring on a numerical scale.

  1. (viii)

    PROM instructions

Overall (n = 9), participants reported that they understood PROM instructions and they were clear. Regarding the timeframe mentioned in the instructions, they reported that due to the variability of their condition, two weeks may not accurately capture distress levels.

Some participants (n = 3) suggested to extend the timeframe to one month, some (n = 1) to three months, some (n = 2) were unsure, while others (n = 3) agreed with two weeks.

Demographic data

Although demographic data was not formally collected for Phases 2, 4 and 5 participants, some information unintentionally was disclosed; Most participants were female, Caucasian, and revealed their occupations were in the banking, business, or healthcare-sectors. Disease duration varied from two to 40 years in Phase 5 participants.

Psychometrician consultation

The psychometrician validated the study design and confirmed word-based scoring to be appropriate for a PROM so long as presented in ordinal format for summation of scoring. The psychometrician also confirmed the supplementary questions as relevant and suggested these to be inserted at the end of the PROM to avoid potentially influencing participants scoring. Finally, the psychometrician suggested consideration of the time recall period in the instructions to ensure consistency and validity of participant answers.

Final changes to the RADS

Final changes were made to the RADS based on the results and analysis from Phase 5 data presented above (See Table 2 for final item list and RADS Final Version in Appendix 5).

Table 2 Rheumatoid Arthritis Distress Scale (RADS) items and response options
Full size table

A preliminary decision was made, following PPI -group discussions, for instructions to score based on the previous three months rather than two weeks, ensuring enough timeframe to capture different aspects of distress. Word-based scoring was chosen as the predominant preference from Phase 4 and 5 participants. No major differences were noted from gross observation of scoring patterns when alternating the different versions in Phase 5.

In total, 19 items on the PROM were rephrased, two were removed, and six new items were created. The supplementary questions were retained in response to the participants’ preference, and the psychometrician’s advice.

For item order, new items were inserted into the PROM alongside those from their corresponding theme. Item ‘I feel overwhelmed living with RA’ was moved to the end to become item 39 on the RADS as per the participants’ (n = 6) suggestions. Therefore, after analysis of the data in Phase 5 of the RADS development, the final version consisted of a 39-item scale with three supplementary questions (See Appendix 5).

Discussion

Presentation of principle findings

This study described the five initial phases of development of the RADS, a PROM to identify DSD in people with RA. The PPI group and cognitive interviews confirmed empirically the findings from the secondary data analysis that RA distress does exist as an important entity and appears distinct from other conditions like clinical depression. The 39-item RADS demonstrates initial face and content validity with people with RA. Conceptually, the 39 items link to one of five themes of RA distress. The RADS is now available for further psychometric evaluation in clinical and research populations.

Comparisons of evidence with the wider literature

To the authors’ knowledge, this is the first PROM developed to identify DSD in people with RA. The Rheumatoid Arthritis Impact of Disease (RAID) questionnaire measures seven domains of disease impact [35]. However, the RADS focuses in more detail on the emotional distress and cognitive impact that is commonly experienced in RA, specifically on the symptoms, burden and treatment.

The themes of RA distress (Fig. 2) from this study are similar to those identified in type 1 and 2 diabetes [19, 36], IBD [9] and to a lesser extent cancer distress scales [37]. These long-term diseases share constructs of emotional, healthcare-related, treatment-related, and social distress. Emotional distress items from two validated diabetes distress measures [19, 36] echo the RADS, describing difficulties of illness acceptance and worries about future complications. In this study the predominant emotion for people with RA seemed to be of frustration, although minor themes of anger also emerged, with some participants reporting that anger was not an emotion they were ‘allowed’ to feel or express, resonating findings from previous qualitative studies [38]. In diabetes emotional health PROMs, distinctions are drawn between appraisal of emotional impacts such as distress and assessment of quality of life [39]. In assessing distress people talk about feelings such as worry and frustration. In cognitive assessments people use terms such as “I think” or “being concerned about” something. Three of the 39 items [6, 10, 24] use cognitive appraisal terms with the remainder using emotional phrases consistent with the diabetes distress PROMs [19]. Further item redundancy may be identified in future psychometric evaluation.

Cancer [37], diabetes [19, 36], IBD [18] and RA scales have all included items about healthcare-related distress. In diabetes and IBD ‘concerns not taken seriously’ by healthcare professionals emerged, while in RA lack of clinicians’ time spent to assess the condition and provide emotional support caused distress. In both cancer and RA, distress over lack of adequate information from healthcare professionals has been described [37].

Treatment-related distress in IBD and RA included concerns over side effects, while in diabetes lack of confidence/motivation in self-management, and guilt about failing with treatment regimens, seems predominant. Social-related distress common to all four long-term conditions highlighted a lack of understanding from others about their illnesses.

People with RA, IBD or cancer, but not diabetes, demonstrate symptom-related distress. People with RA are distressed about their joint pain, stiffness, and fatigue. Pain may explain the difference in degree of depression between RA and healthy controls [40] and there is some evidence that psychological distress in RA can be secondary to pain rather than vice versa [41]. The link between pain and DSD in RA, however, has not been previously investigated.

Elevated diabetes distress is prevalent in 20–40% of people with type 1 [42, 43] and in 36% of people with type 2 [44] diabetes. Four systematic reviews have explored psychological interventions in diabetes. In general, reviews found psycho-educational treatments resulted in a low to moderate effect on DSD reduction and more intensive and longer duration interventions seem to achieve a greater effect [13, 14, 45, 46]. One systematic review demonstrated that motivational interviewing significantly reduced diabetes distress and improved glycaemic control [13]. Furthermore, diabetes-tailored interventions, as opposed to general mindfulness interventions, showed to most likely improve both DSD and glycaemic control in people with type 1 and 2 diabetes [14].

In a cross-sectional study of 189 cancer patients, 58% demonstrated elevated DSD [6]. Psychosocial interventions, such as cognitive behavioural therapy and meaning-centred psychotherapy may improve quality of life and alleviate anxiety in people with cancer, although systematic reviews have not focussed on DSD as a primary outcome [47, 48].

In IBD and RA, the prevalence of DSD is not yet known, nor what effective treatment may entail. However, psychological interventions such as supportive counselling can result in small to moderate improvements in biopsychosocial outcomes for patients with RA [49].

Strengths and limitations of the study

The strengths of this research are its systematic study design, which was validated by a psychometrician, and followed recommended guidelines of PROM development [20] and reporting [21]. The secondary data analysis was an ethically efficient way to make good use of anonymised data. People with RA were involved throughout the process of development. Changes to the PROM after Phase 4 were endorsed by PPI participants, ensuring accurate interpretation of data [30]. The robustness and validity of data was further strengthened by inclusion of diverse cases/minor themes [21]. One notable strength of our demographics was the mixture of early and established RA.

The PPI groups in this study had taken part in several similar research projects in the past and embraced the topic of RA distress. The data generated therefore led to high ‘information power’ [50]. The final two cognitive interviews in Phase 5 did not yield as much ‘new information’ indicating close ‘data saturation’ [26].

This research has limitations. Our study would have benefited from a larger sample size for the PPI groups and cognitive interviews, including more males and people from diverse socioeconomic and ethnic backgrounds. RA incidence can be 4–5 times higher in females [51], hence challenges in recruiting males for this study. Cognitive interview participants were recruited from one national charity, which could explain the homogeneity of this sample. The team did not send participants transcripts for fact/content checking to minimise research burden and/or fatigue [34].

We included data in our analysis from participants who were carers (n = 5) to people with RA and diagnosed with APS (n = 9) and IIM (n = 12), in addition to people with RA (n = 50). Therefore, it is possible our data captured evidence of rheumatological disease distress rather than isolated RA distress. As 60% of our participants had a confirmed diagnosis of RA, the authors concluded to present this PROM as an RA distress scale, while acknowledging that there is likely overlap in distress measures across rheumatic diseases. We cannot exclude similar overlap in causes of distress amongst patients who may have un-diagnosed co-morbid Fibromyalgia or central sensitisation disorders.

Finally, participants in Phase 1 were not specifically asked about RA distress and since the 39 items were generated during secondary analysis of this data, some items areas may be missing from the RADS.

Recommendations for practice, policy and research

DSD in RA appears to be a new important concept and its identification comes at a pivotal time; the 2018 updated NICE guidelines for RA management in adults recommended clinicians to assess the effect RA has on the patients’ life [52]. Furthermore, patients with RA report that the most frequent reasons for flare up of joint symptoms is psychological stress [53]. It therefore is important to identify this subset of psychological distress.

With the development of the RADS it is hoped that DSD can be identified in RA to build on interventions to reduce distress, as has been achieved in type 1 and 2 diabetes [13], whereas the RAID appears to be useful more in monitoring the impact of RA on patients’ lives.

There is a need to establish face and content validity of the RADS in a more diverse patient population. Cut-off thresholds for severity need to be explored as currently the measure does not have a scoring range or item weighting. Item order could be investigated for reliability differences between item grouping or intermixing. Future studies need to focus on test–retest reliability, cross-sectional and longitudinal construct validity. Feasibility needs to be evaluated, including scoring time; The RADS is long with 39 items to record. Further psychometric evaluation will clarify additional potential item redundancy.

Currently no gold standard for RA distress exists. Therefore, the RADS could be evaluated against modified validated diabetes and/or IBD distress scales, considering the similarities in distress domains.

Following detailed psychometric evaluation, the RADS has the potential to be employed in larger longitudinal studies to identify the prevalence of RA distress. It would be worth investigating whether DSD is associated with clinical outcomes in RA. Effective psychological interventions for DSD, as implemented in type 1 and 2 diabetes patients [13], would be welcome by people with RA and their carers.

Conclusions

In summary, this study offers evidence for DSD as an important entity experienced by people with RA. It appears RA distress shares some domains with other long-term condition specific anguish. The 39-item RADS has acceptable first phase face and content validity in people with rheumatological long-term conditions specifically RA. RA distress appears to be distinct from clinical depression or anxiety disorders. The RADS has the potential to be a useful tool for identifying RA distress.