Two-year follow-up of neuropsychiatric symptoms in patients with systemic autoimmune rheumatic diseases: longitudinal insights on depression, anxiety, memory and adaptation in the INSPIRE cohort

Summary

Background

To date, studies of neuropsychiatric symptoms in systemic autoimmune rheumatic diseases (SARDs) have largely been cross-sectional. Our longitudinal study with patients with SARDs from the international INSPIRE cohort aimed to advance knowledge of changes in patient-reported neuropsychiatric symptoms and adapting over time.

Methods

We conducted an observational cohort study analysing longitudinal data from the international INSPIRE (Investigating Neuropsychiatric Symptom Prevalence and Impact in Rheumatology Patient Experiences) research project. Self-reported validated survey measures for anxiety (GAD-7), depression (PROMIS SF8), and memory (EMQ-R) were compared between baseline (2022) and follow-up surveys 2 years later (2024). We invited all participants who had provided their contact details and consented for the follow-up survey. Our co-produced “ADAPT” instrument was used to assess and compare changes in how people with SARDs adapt to living with their disease over time. T-tests, Kruskal–Wallis and Mann–Whitney U tests were used to assess differences across groups for changes in study measures. Sociodemographic and disease differences were controlled for by generalised linear modelling in adjusted analyses. Patients with polymyalgia rheumatica (PMR) made up the reference group because the Kruskal–Wallis test established no significant difference between the PMR and the control group for memory, depression and anxiety scores on the baseline dataset.

Findings

The modal within-person change between baseline (BL) and follow-up (FU) for the N = 742 participants was zero for all 3 validated tools. Significant improvements at group level over two years were identified for memory scores (BL:16.14 to FU: 13.96, p = 0.002), and all four ADAPT measures (p < 0.05), whereas mean depression (BL:16.61 to FU:16.21, p = 0.303) and anxiety (BL:5.83 to FU:5.36, p = 0.073) scores were unchanged over time. Significant improvements in memory scores were observed in SLE (p = 0.022) and RA/IA (p = 0.009), and in those diagnosed (at baseline) > 10 years (p = 0.016) and between 6 and 9 years (p = 0.020).

Interpretation

This study provides evidence against the expectation of progressive memory impairment in most cases in SARDs, including SLE. Lack of improvement in mean depression and anxiety scores over time suggests that many patients with SARDs may benefit from more proactive/assertive rheumatologic disease control, psychiatric intervention, and/or psychosocial support. Early and effective SARDs disease control is required, with timely psychiatric intervention and psychosocial support provided, particularly for the newly diagnosed.

Funding

This study was funded by The Lupus Trust.

Research in context.

Evidence before this study

PubMed searches for relevant new articles continued throughout the study, with the first search on 20 Sep 2021 using various combinations of the systemic autoimmune rheumatic diseases (SARDs) (e.g. “SLE”, “lupus”, “Inflammatory arthritis” OR “Rheumatology’’) AND “[neuro]psychiatric” OR mental health (and individual symptoms/diagnoses within that term, e.g. “anxiety”). There were no restrictions by article type, language or time parameter. Additional searches included terms such as “longitudinal”, “adapting”, “self-management” AND “autoimmunity” OR “chronic disease”. The literature reported the high proportion of neuropsychiatric symptoms in SARDs and difficulties in attribution. A gap in the literature was identified with regards to changes over time in neuropsychiatric symptoms including conflicting reports on cognitive changes.

Added value of this study

This longitudinal study found persisting depression and anxiety symptoms in SARDs over time, but significant improvements in memory and a measure of adaptation to illness. Most SARD studies to date have been observational, and/or small, and/or used only objective measures which may be discordant with patient's experiences of their neuropsychiatric symptoms. We used validated patient-reported outcome measures, and a novel co-produced “ADAPT” instrument to ascertain changes over two years in the international INSPIRE SARD cohort.

Implications of all the available evidence

People with SARDs have a high burden of neuropsychiatric symptoms which may relapse/remit over time in line with changes in disease activity. Memory difficulties, although common, distressing and disabling, appear not to be typically progressive, and may even improve over time. Attribution is contested in the literature and impacted by multiple interacting factors including the direct impact of these diseases on the brain, medication side-effects, and the challenges of living with a chronic disease. Many patients with SARDs require more optimal pharmacotherapy and psychosocial support to gain better long-term control of symptoms and improve adaptation to living with the disease(s) and quality of life.

Introduction

High prevalences of neuropsychiatric symptoms have been widely reported in systemic autoimmune rheumatic diseases (SARDs).^1,2 Poor mental health, including cognitive dysfunction, is associated with worse clinical outcomes,³ and reduced quality of life (QoL) in individuals with SARDs.² Some studies suggest that mental health symptoms are often associated with the burden of living with a chronic disease,⁴ or multifactorial,⁵ whilst others report evidence of direct disease involvement of the brain.⁶ Research to date has been largely cross-sectional with limited longitudinal research to ascertain changes in symptoms of anxiety, depression and cognitive impairment over time. Among longitudinal studies, one study reported an increase in prevalence of anxiety and depression symptoms in rheumatoid arthritis (RA) patients over the 10-year period from 2005 to 2014 compared to 1985–1994. However, a similar increase was also apparent among their study participants without RA,⁷ and other population-level research also likewise reported a general rise in anxiety symptoms.⁸ This highlights the importance of contextualising these findings within wider trends, particularly during periods of rapid changes in clinician awareness and diagnostic criteria for illnesses.

Neuropsychiatric symptoms can arise from the direct impact on the central nervous system in some SARDs, most commonly reported in SLE, and also increasingly recognised in other SARDs.¹ In addition, the chronic peripheral inflammation prevalent in rheumatic diseases is associated with increased psychiatric symptoms,^2,7 including cognitive impairment.^2,9 There are also multiple other contributors to anxiety, depression and cognitive symptoms, including the challenges and limitations of living with a chronic disease, and the impact of other common SARD symptoms. For example, insomnia and poor sleep are prevalent in SARDs,¹ and have been found to correlate with reduced QoL and anxiety and depression.¹⁰ Moreover, treatment with corticosteroids is common in SARDs, and is associated with adverse mental health outcomes, particularly at higher doses.¹¹ Cognitive dysfunction is common in SARDs,^1,12 although the literature is inconsistent in terms of changes in cognitive function over time with SARDs. One study found improvement in cognition over 10 years in SLE,¹³ whereas another study noted more cases with cognitive dysfunction over time,¹⁴ and others reported no significant change in cognitive function over time in systemic sclerosis (SSc)¹⁵ and SLE.¹⁶

Most SARD studies of neuropsychiatric symptoms to date have been observational, small and/or used only objective testing which has been found to often be discordant with the patient's self-perceptions,¹⁷ particularly regarding cognitive dysfunction. Our longitudinal study with patients with SARDs from the international INSPIRE¹ cohort was therefore co-designed to further knowledge of patient-reported within-person neuropsychiatric changes over time in SARDs. Self-reported validated measures for anxiety (GAD-7), depression (PROMIS SF8), and memory (EMQ-R) were compared between baseline and follow-up surveys two years later. In addition, our co-produced “ADAPT” instrument was used to assess and compare changes in how people with SARDs adapt to living with their disease over time.

Methods

Study design and procedure

This study is the longitudinal component of the mixed methods INSPIRE (Investigating Neuropsychiatric Symptom Prevalence and Impact in Rheumatology Patient Experiences) research project.¹ Following extensive consultation with patient groups and clinicians (from neurology, psychiatry and rheumatology), and pre-survey interviews, the survey was co-produced to ascertain neuropsychiatric symptom frequency and impact. The baseline survey was made available internationally via Qualtrics through social media and patient charities in 2022, and a shorter follow-up survey (Baseline + 2 years) was sent in 2024 to all participants who had provided their contact details and consented for the follow-up survey. Up to three reminders were sent by email. Informed consent was obtained electronically prior to the start of both surveys. Eligibility criteria included participants aged 18 or over, and with any SARD as reported in clinical correspondence.

Measures

The longitudinal study was designed to ascertain changes in the following validated self-assessment instruments: GAD-7 for anxiety,¹⁸ PROMIS 8-item negatively framed short-form for depression,¹⁹ and the 13-item Everyday Memory Questionnaire–revised (EMQ-R) for subjective memory and attention.²⁰ These instruments all employ Likert scales with higher scores signifying more severe symptoms. While cognition, broadly and inclusively defined, includes many elements of function, we chose to assess memory and attention, as these are more easily assessed with self-assessment measures.

Our exploratory work with patients in choosing appropriate tools for assessing adapting to living with a SARD failed to identify an existing instrument that was deemed to fully capture their priorities and not contain items that some patients found either unnecessary or offensive. This was in line with previous findings that current Patient Reported Outcome Measures (PROMs) had limitations and may not adequately represent patients’ perspectives.²¹ We therefore co-produced a new “ADAPT” instrument, four items of which were included in this study, including: 1) “How do you feel overall in terms of being satisfied with your life?” 2) “How have you adapted to the changes in your life from having a chronic disease?” 3) “How much do you participate in everyday life?”, and 4) “How much control do you feel you have over your disease symptoms?”. For all the ADAPT questions, participants moved a slider from 0 to 100, with scores of zero signifying the worst possible level and 100 the best. The ADAPT instrument is currently undergoing validation in the ADAPT (exercise) trial and was trialled for acceptability and responsiveness in the ADAPT feasibility trial of psychosocial interventions.²² The STROBE checklist can be found in Supplementary Information 2.

Analysis

Following a pre-specified statistical analysis plan, data were analysed using SPSS Software V29 with alpha set at 5%. Due to non-normality of certain variables, Kruskal–Wallis and Mann–Whitney U tests were used to assess differences across groups for changes in validated instruments. Bonferroni corrections were used to control for Type 1 errors. ANOVA and t-tests were used to compare mean within-person changes in scores over time across groups of interest (e.g., disease groups and disease durations). A binary logistic regression was conducted to determine which factors, amongst the validated instrument measures, ADAPT measures, and sociodemographic measures, significantly predicted the odds of completing the follow-up survey.

Generalised linear modelling controlled for disease type, income, age, gender, country of residence, ethnicity, job status, and disease duration, with the outcome variable being the follow-up score for each measure (e.g., memory follow-up scores). Polymyalgia rheumatica (PMR) was used as the reference group because the Kruskal–Wallis test established no significant difference between the PMR group and the control group for memory, depression, and anxiety scores on the baseline dataset (controls could not be used as a comparison group as they were not included in the longitudinal study).¹² Logistic regression investigating predictors of follow-up survey completion was carried out as an additional exploratory analysis due to the relatively high attrition rates observed.

Categories for each level of depression and anxiety (e.g. none, mild) were derived from the validated instruments cut-offs for each level.^18,19 Categories for disease durations and age groups were divided into periods of time that our earlier qualitative research had identified to be commonly associated with degrees of adaptation to living with a chronic disease, as opposed to being divided into even durations. Analyses for anxiety, depression and memory changes were for each individual disease as our previous INSPIRE prevalence project found significant differences among diseases.¹ However, we analysed all diseases together for the changes in the ADAPT measures as adapting to living with an autoimmune disease has been found to be challenging, regardless of specific disease.²² Age and disease duration were used for ADAPT instrument measures to further explore earlier qualitative findings that suggested an association.²³

Ethical approval

The INSPIRE study was pre-registered (https://osf.io/zrehm) with ethical approval obtained from the Cambridge Psychology Research Ethics Committee: PRE.2022.027. Informed consent was taken electronically before both baseline and follow-up surveys.

Role of the funding source

The Lupus Trust had no role in the design, implementation, analysis or write-up of this study, other than in the advertising of the study to its members in line with other rheumatological charities.

Results

Of the 1853 participants completing the first INSPIRE survey, 742 (40%) completed the follow-up survey. Follow-up respondents were mostly female (90%), White (95%) and aged >40 (90%) (Table 1). Differences in validated instrument scores and self-assessed disease activity between responders and non-responders were all non-significant, based on baseline scores. Similarly, all of the ADAPT measures were not statistically significant predictors of follow-up, aside from a small effect (OR = 1.006, 95% CI [1.001, 1.012], p = 0.032) of self-assessment of level of adapting to the disease. All other variables in the model, including disease, sociodemographic groups and psychological measures (e.g., QoL, Anxiety, Depression) were not significant predictors of completing follow-up (all p > 0.05). Characteristics and baseline scores for responders and non-responders to follow-up can be found in Supplementary Table S1.

Table 1.

Sociodemographic characteristics of participants completing both surveys.

Characteristic	n = 742 (%)
Age
18–29	20 (3%)
30–39	52 (7%)
40–49	107 (14%)
50–59	200 (27%)
60–69	232 (31%)
70+	130 (18%)
Prefer not to say	1 (<1%)
Gender
Female	671 (90%)
Male	69 (9%)
Other/undisclosed	2 (<1%)
Country/region
England	547 (74%)
Scotland	64 (9%)
Wales	39 (5%)
N. Ireland or Republic of Ireland	11 (2%)
US or Canada	32 (4%)
Europe	31 (4%)
Asia	2 (<1%)
Latin America	1 (<1%)
Australia or New Zealand	9 (1%)
Other	6 (<1%)
Ethnicity
White	699 (95%)
Asian	13 (2%)
Black	5 (1%)
Mixed	13 (2%)
Other	9 (1%)
Undisclosed	1 (<1%)
Disease
Inflammatory or rheumatoid arthritis	211 (28%)
SLE	183 (25%)
Vasculitis	88 (12%)
Sjögren's disease	71 (10%)
PMR	60 (8%)
UCTD	27 (4%)
Myositis	27 (4%)
Systemic sclerosis	28 (4%)
Mixed/multiple SARDs	43 (6%)
Other	4 (<1%)
Time since diagnosis at baseline
<1 year	49 (7%)
1–2 Years	104 (14%)
3–5 Years	155 (21%)
6–9 Years	137 (19%)
10 + Years	293 (40%)
Unsure	2 (<1%)

Age and gender were the only sociodemographic characteristics that were statistically significant predictors of completing follow-up. The odds of completing the follow-up survey increased with age. For instance, participants in the age group 30–39 had 2.395 times the odds of completing the follow-up survey compared to the reference group (Age group 18–29) (B = 0.873, SE = 0.324, p = 0.007), and the 50–59 age group had 3.532 times the odds of follow-up completion B = 1.262, SE = 0.311, p < 0.001). The odds ratio for males completing follow-up was 0.640, such that males had approximately 36% lower odds of completing the follow-up than females (B = −0.446, SE = 0.205, p = 0.030).

The modal within-person change between baseline (BL) and follow-up (FU) was zero for all 3 validated tools (Fig. 1). The graphs of changes in validated instruments all showed a normal distribution with the majority of changes being minimally increased or decreased. There was a significant improvement in overall memory scores between baseline and follow-up (BL:16.14 to FU: 13.96, p = 0.002, SD = 9.41). There were no significant differences in mean depression (BL:16.61 to FU:16.21, p = 0.303, SD = 5.65) or anxiety (BL:5.83 to FU:5.36, p = 0.073, SD = 3.76) scores between baseline and follow-up.

Fig. 1.

Within-person changes in validated tools over two years, N = 742 maximum.

Univariate analysis of changes in validated scores by time since diagnoses and disease group showed significant improvements in memory scores in SLE (p = 0.022, SE = 0.872) and RA/IA (p = 0.009, SE = 0.675), and in those diagnosed (at baseline) > 10 years (p = 0.016, SE = 0.555) and between 6 and 9 years (p = 0.020, SE = 0.930). All other changes were non-significant and included longitudinal improvements for all disease groups aside from UCTD for memory, and for all disease groups aside from UCTD and Sjögren's for anxiety (Fig. 2A). Comparing changes in scores on validated tools by time since diagnosis, only those who had been diagnosed <1 year at baseline had (non-significant) increases in anxiety (p > 0.05, SE = 0.603) and depression (p > 0.05, SE = 0.700) scores between BL and FU (Fig. 2B).

Fig. 2.

A. Disease group changes from BL to FU, N = 742 Maximum. B. Time since diagnosis (at BL) comparisons of changes from BL to FU, N = 742 maximum. Note: An increase in scores (denoted by a red bar) indicates a worsening in symptoms. ∗ = significant change between baseline and follow-up at p < 0.05. See Table 1 for number of participants in each disease and diagnostic duration group.

Generalised Linear Model analysis (Table 2 supplementary) demonstrated that SLE (B = 0.371, SE = 0.066, p < 0.001), rheumatoid arthritis/other inflammatory arthritis (B = 0.166, SE = 0.064, p < 0.001), Sjögren's (B = 0.304, SE = 0.075, p = 0.009), UCTD (B = 0.354, SE = 0.100, p < 0.001), and vasculitis (B = 0.223, SE = 0.072, p = 0.002) were positively significantly associated with everyday memory scores, after controlling for income, age, gender, country, ethnicity, job status, and diagnosis duration.

Table 2.

Mean change in adapting scores (N-734 maximum).

ADAPT item	Mean baseline score (0–100) (standard deviation)	Mean follow-up score (0–100) (standard deviation)	Mean change [95% CI]	p-value
Satisfaction with life	60.00 (23.79)	61.57 (23.32)	1.57 [0.036,3.110]	0.045
Adapting to living with the disease(s)	66.91 (23.12)	68.94 (20.28)	2.03 [0.447,3.605]	0.012
Participation in life	53.84 (27.63)	57.44 (26.14)	3.61 [1.942,5.272]	<0.001
Feeling of control	44.64 (27.22)	49.72 (25.76)	5.08 [3.06,7.103 ]	<0.001

While disease type was significantly associated with memory, it was not significantly associated with depression and anxiety scores after controlling for other sociodemographic factors, aside from SLE for depression (B = 0.134, SE = 0.066, p = 0.042). Age was significantly inversely associated with validated instrument scores of depression (B = −0.009, SE = 0.004, p = 0.017) and anxiety (B = −0.150, SE = 0.029, p < 0.001), but not for memory (B = −0.004, SE = 0.003 p = 0.290). Income was also found to be significantly positively associated with memory scores (B = 0.024, SE = 0.007, p < 0.001), depression (B = 0.055, SE = 0.007, p < 0.001), and anxiety (B = 0.067, SE = 0.015, p > 0.001). No other sociodemographic or disease factors (gender, country of residence, ethnicity, job status, and duration since diagnosis) were significantly associated with memory, depression, or anxiety scores.

The number of participants who moved to a more severe category of depression and anxiety was smaller than the number who moved to a less severe category (Fig. 3). Movement from no anxiety at baseline into a classified level (of varying levels, most frequently mild) of anxiety (using GAD-7 cut-off levels) two years later at follow-up was observed for 23% of participants. Similar proportions of movement into a classified level of depression occurred for 22% of participants. These figures contrast with 61% and 60% with the severest form of anxiety or depression respectively, moving into a lower classification over the two-year period. The EMQ memory instrument has no defined classification levels so was not used in this analysis.

Fig. 3.

Movement between severity of symptoms between BL and FU (N = 742 maximum). Note: Severity levels determined by published validated instrument cut-off levels.

Changes in the co-produced adapting instrument (“ADAPT”)

The four ADAPT measures all showed significant improvement between the baseline assessment and two-year follow-up. Improvements ranged from +1.57 for life satisfaction, to +5.08 for feeling of control (all p < 0.05) (Table 2). Although the more recently diagnosed (<2 years at baseline) showed greater improvements in all adapting measures over time than the longer-term diagnosed (10 years +) (Fig. 4, Fig. 5), these were not significant aside from in adapting to life with their disease (p < 0.001).

Fig. 4.

Change in ADAPT measure scores by time since diagnosis.

Fig. 5.

Changes in each ADAPT score by age and time since diagnosis (at baseline) N = 734 maximum.

Although the younger age groups generally demonstrated greater increases in all ADAPT instrument scores, these were not statistically significant. Significant associations were found within adapting and gender, f (723) = 7.52, p = 0.006, with men having a higher mean score on change in adapting than women (2.04 vs 3.02). Associations between adapting and ethnicity were not conducted due to small samples of minoritised ethnicities.

Discussion

This longitudinal study of SARDs over two years demonstrated significant improvements in self-reported memory scores at group level, whereas mean depression and anxiety scores were unchanged over time. At the individual level, there was some movement across severity categories for depression and anxiety, with over half of participants in the most severe categories moving to a lower severity category over two years. The extreme differences (both increases and decreases) in some participants’ memory scores between baseline and two years later likely reflect the relapsing-remitting nature of cognitive impairment in SARDs, and impact of treatment. This improvement, even from the severest level of self-reported memory difficulties at baseline, suggests that the cognitive dysfunction is not permanent in most cases. Mean scores on the co-produced ADAPT instrument increased significantly over the two-year period at the group level, although observed changes were small. Our results suggest that, on average, adapting to living with a SARD improves over time. This includes improvements in life satisfaction, sense of control, and participation in daily activities. Relatedly, those who were diagnosed more recently were found to have larger changes in adapting over the two years compared to those with a longer duration since diagnosis. Adapting to a chronic illness is multifaceted, including areas of need which encompass psychological, social and physical factors.

Statistically significant changes within groups were only observed for improvements in memory for SLE, RA/IA and for the disease durations of 6–9 years and >10 years. It was notable that individuals in the earliest stages of diagnosis (<1 year) and those with UCTD and Sjögren's disease diagnoses appeared to show worsening anxiety and depression. Although these findings did not reach statistical significance, likely because of the small sample size in the relevant groups, they may be clinically significant, and indicate increased vulnerabilities. Possible explanations include the relatively limited treatment and support available for UCTD and Sjögren's disease, and the prevailing medical viewpoint that these diseases are “milder”.²⁴ Although “mild” may be used to imply limited or absent major organ involvement, this term can feel dismissive to patients and adversely impact their care by failing to account for the often severe life-changing symptoms. Our previous research found a higher neuropsychiatric symptom burden in both Sjögren's disease and UCTD than other rheumatological diseases such as PMR and IA,¹ and others have similarly called for more psychosocial support for UCTD patients.²⁵ Further exploration is also required as to how UCTDs and Sjögren's disease pathological processes may directly contribute to cognitive, depressive and anxiety symptoms. In addition, future work should establish whether an ‘undifferentiated’ or not otherwise specified type diagnosis may in itself represent, or carry with it, a form of pervading uncertainty (including in diagnosis, prognosis and treatment) that could have additional detrimental effects on mental health. Although the <1 year diagnosis group was the smallest (n = 49), thus affecting statistical power, this being the only disease duration group where results were in the direction of worsening mental health may reflect the initial emotional impact of being diagnosed with an incurable disease. It may also potentially reflect the delay in achieving more effective disease control, with immunosuppression found to improve depression and anxiety.²⁶ This suggests a period of apparent increased vulnerability for depression and anxiety symptoms and difficulties in adapting shortly after disease diagnosis.

There is disagreement as to the frequency and degree that psychiatric symptoms are directly linked to SARD disease activity. In one SLE study, depression persisted over time, even after disease-control, and pain and mood symptom interventions.²⁷ In RA, longitudinal analysis reported a positive association between disease activity and depressive symptoms.²⁸ The literature is also highly variable in relation to cognitive dysfunction. Some note improvement in cognition over 10 years in SLE, with the authors postulating that this can result from improved disease control,¹³ whilst others reported no significant change over time in SSc and SLE,^15,16 and another reported more cases of patients with cognitive impairment defined by objective testing over time.¹⁴ Subjective cognitive function, including memory, may improve over time as patients adapt to disease-related changes, benefit from treatment, and face fewer cognitive demands with changing life circumstances. It is important to highlight the frequent discordance between objective and subjective disease activity,¹⁷ and how this impacts associations of disease activity with depression, anxiety and cognitive symptoms. For example, a longitudinal study of psoriatic arthritis found that changes in depression correlated with subjective, but not objective, measures of disease.²⁹ As we have argued previously, self-assessments are most important in terms of a patient's QoL as they reflect how a patient feels as opposed to what objective tests or clinicians tell them they should be feeling. Even many of the self-assessment instruments were designed in an era of more limited patient involvement in research and care, and thus may not fully reflect patient experiences, priorities, or needs. This highlights the requirement for all patient-reported measures to be co-produced with patients.

Disease control is key in reducing both the direct and indirect effects of the diseases on mental health. Hwang et al. reported longitudinal measures of disease activity and QoL being associated with depression.³⁰ In addition to optimal disease control and mental health support from clinicians and counsellors, patients should also be supported to enact adaptations. Our recent work with people with SARDs on the impact of these diseases on their lives may be useful for clinicians and policy makers in understanding the–often life-changing–repercussions of SARDs.²³ A study using Self-Determination Theory found that the constructs of competency, autonomy, and relatedness were inter-related and an increase could lead to improved QoL.³¹ Another study in inflammatory arthritis found improved mental health by patients engaging in physical activity, making connections socially and developing a positive mindset.³² Support for the complex and multifaceted needs of people with SARDs requires an integrated multidisciplinary approach, with greater involvement of allied healthcare professionals (e.g. psychologists, neuropsychologists). A more holistic and coordinated care model has the potential to optimise disease management and improve patient outcomes and QoL.

The major limitation of this longitudinal study is that only 40% of those completing the INSPIRE survey at baseline were able to be followed up 2 years later. Non-responders included participants who did not provide contact details or permission to be contacted for the follow-up survey, and baseline participants who did not respond to the follow-up survey. Therefore, our data may be skewed, as sociodemographic characteristics and health can differentially influence completion, and findings cannot be generalised to the original baseline population. Age and gender were found to be a significant predictor of follow-up completion, with older age groups and females having significantly higher odds of completion. There were no significant differences between completers and non-completers in terms of baseline validated instrument scores or self-reported disease activity levels The only health/wellbeing measure that was a significant predictor of completing follow-up was the single “adapting to disease” measure within the ADAPT instrument, and the association was small. There are different limitations for studies using objective testing versus those using subjective patient-reported outcome measures (PROMS). Although we greatly value the patient self-assessment, combining evidence from self-assessment with clinical evaluations, disease activity assessments and laboratory tests would provide more conclusive evidence. In terms of inter-study comparability, we only essentially measured memory and (in a contributory way) attention as other cognitive symptoms are not so accessible to self-ratings. Medication use was not evaluated in this study and its association with mental health is being investigated in a concurrent INSPIRE research project.

For this longitudinal study, we chose to focus just on anxiety, depression, and memory specifically as these were identified as areas of high concern and prevalence in our previous research. We acknowledge that this is only a small sample of neuropsychiatric symptoms experienced in SARDs (e.g., not assessing for symptoms including mania or psychosis), and that no assumptions can be made regarding changes over time in other areas of neuropsychiatric function. Additionally, many neuropsychiatric symptoms are inter-linked (e.g., anxiety and depression are commonly associated with memory symptoms) to varying degrees in each individual. The large proportion of White and female respondents reduces the generalisability to other ethnicities and genders. It is also important to consider longitudinal changes in mental health in this patient group in the context of world events, such as the COVID-19 pandemic and global instability. Observed improvements in memory may in part reflect regression to the mean and the influence of measurement variability in self-report scales. A major strength of the INSPIRE study programme is its patient-centred approach, with people living with SARDs setting the research priorities, and co-producing the ADAPT instrument. In addition, an MDT of academics, patients, rheumatologists, psychiatrists, and neurologists was involved at every stage of the research process. Given the often conflicting inter-disciplinary views of attribution and treatment for neuropsychiatric symptoms, this allows any interpretation of data to be balanced by the inclusion of multiple diverse perspectives.

In conclusion, the group-level improvement in memory scores over two years suggests that subjective cognitive impairments in most patients with SARDs are unlikely to be progressive in the majority of cases. Improvements in patient-reported outcomes, such as a sense of control and participation in daily life, also suggest gradual acceptance and adaptation to living with a SARD. However, mean depression and anxiety scores were unchanged and remained high in prevalence, indicating ongoing unmet needs in both disease management and psychosocial support. Pharmacological and non-pharmacological interventions are available to support patients in managing and adapting to these diseases. Early disease control and timely psychosocial support should be prioritised, especially at the point of diagnosis.

Contributors

MS and DD'C conceived of the idea for the INSPIRE project and acquired funding. MS and JB collected the data, with JB managing the data collection and storage. MS, AV, AT and MP analysed and visualised the quantitative data. MS and AT reviewed the relevant literature. MS, AV, and JB (Cambridge data manager) have directly accessed and verify the underlying data reported in this manuscript. LC and ED led the input from team patient partners and advised from the patient perspective, and GL, D’DC, JAB, PH-C, SG, XY, MY, AB and TAP provided rheumatology, psychiatry, neurology and primary care expertise. AT, FN, MP and AA provided academic, psychology, and methodological expertise. MS wrote the original draft with substantial contributions from AV and AT, and input throughout multiple drafts from all co-authors. All authors attended meetings and discussion groups including on analysing, discussing, and agreeing qualitative themes. All co-authors reviewed and edited the final manuscript and earlier drafts.

Data sharing statement

Anonymised data will be available on reasonable request following the completion of the INSPIRE studies.

Declaration of interests

MS reports funding to her research team from The lupus trust, Lupus UK, Vasculitis UK, The NIHR, and SRUK, and consultancy fees from Otoimmune paid to the Department of Public Health. DD’C reports consultancy/speaker fees from GSK, Eli Lilly, Vifor and UCB, and a leadership role on the board of APS support UK. JAB has received speaking fees from Psychiatric Times and Oakstone and receives royalties from American Psychiatric Publishing, Springer International, Lippincott Williams & Wilkins, and Cambridge University Press. All other authors declare no potential conflicts of interest. MS and TAP were partially supported by the National Institute for Health Research (NIHR). The views expressed are those of the study participants and/or author(s) and not necessarily those of the NHS, the funders, the NIHR, or the Department of Health and Social Care.