Fatigue in adults with primary antiphospholipid syndrome: findings from a mixed-methods study

Objective This study aimed to explore the experience and impact of fatigue in adults with primary antiphospholipid syndrome (pAPS). Methods This sequential, explanatory mixed-methods study enrolled adults with a six-month or more history of pAPS. Consenting participants completed the Functional Assessment of Chronic Illness Therapy–Fatigue subscale (FS), Multi-Dimensional Perceived Social Support Scale, Patient Health Questionnaire (PHQ9), Pittsburgh Sleep Quality Index (PSQI), International Physical Activity Questionnaire (IPAQMETS). Relationships between FS and other variables were explored with multiple linear regression. Interviews were conducted with a subgroup of participants, and the data were analysed thematically. Results A total of 103 participants were recruited (Mage = 50.3 years; standard deviation = 10.1 years; 18 males). Of these, 62% reported severe fatigue. Greater fatigue was associated with lower mood, physical inactivity, poorer sleep quality and lower perceived social support. The best-fit model explained 56% of the variance in FS (adjusted R2 = 0.560, F(3, 74) = 33.65, p > 0.001) and included PHQ9 and IPAQMETS as significant predictors, and PSQI as a non-significant predictor. Twenty participants completed interviews. Three key themes were identified: characteristics of fatigue, impact on life and coping strategies. Conclusion Fatigue was a common symptom of pAPS and challenging to manage. Other factors, particularly mood and physical activity, influenced fatigue. Evidence-based self-management interventions are needed.


Introduction
Antiphospholipid syndrome (APS) is an autoimmune prothrombotic disorder which occurs as a distinct clinical syndrome in isolation (primary APS (pAPS)) or with other rheumatic and musculoskeletal diseases (RMDs) such as systemic lupus erythematosus (SLE). 1 The incidence of APS is approximately 5/100,000 people annually, and the prevalence is 40-50/100,000 people. 1 More women than men are affected (female:male ratio 5:1), and there is no racial prevalence for pAPS. 2,3 Its clinical spectrum includes venous and/or arterial thromboses and pregnancy morbidity in the presence of antiphospholipid antibodies (e.g. lupus anticoagulant, anticardiolipin antibodies and anti-b2 glycoprotein-I antibodies). People with pAPS also experience other symptoms, such as fatigue and low mood, and have difficulty keeping physically active. 4,5 Chronic fatigue is an unpleasant, abnormal or excessive whole-body tiredness, disproportionate to or unrelated to activity or exertion and present for more than one month. It is not relieved easily by sleep or rest. 6,7 Fatigue is a common symptom of RMDs and adversely affects an individual's health status, as well as physical and social function. Studies in RMDs (e.g. rheumatoid arthritis (RA) 8 and SLE 9 ) and other long-term conditions (e.g. multiple sclerosis 10 ) have highlighted the impact and importance of fatigue. [11][12][13] The aetiology of fatigue is poorly understood. Theoretical models propose fatigue is a multi-causal, multidimensional symptom where disease-related (e.g. disease activity), cognitive/emotional (e.g. mood), physical and social factors (e.g. physical activity and social support) interact with each other. [14][15][16] In other RMDs, aspects outside the direct disease effects account for more of the variation of fatigue than condition-related features. 15,17,18 For this study, key factors which influence fatigue in other RMDs were identified. Low mood is common in people with RMDs and is consistently correlated with higher fatigue. 16,18,19 Low mood is also linked with sleep disturbance, and evidence suggests that poor sleep quality influences fatigue levels. 15,20,21 Physical activity tends to be low in people with RMDs, 22,23 and physical inactivity is associated with higher fatigue. 15,24,25 This relationship may be mediated by mood, sleep or obesity. 15,24,25 Social support from family, friends or significant others (e.g. health-care professionals) may also influence fatigue. 16,26 Few studies have explored the experience and impact of fatigue in people with pAPS, and it is rarely acknowledged in evidence-based management recommendations. 27 To inform the development of non-pharmacological interventions, a mixed-methods approach is needed that comprises quantitative data to investigate the relationships between fatigue and key variables and qualitative data in order to gain greater insight into these interactions and the experience and impact of fatigue. This mixed-methods study aimed to explore the experience and impact of fatigue in adults with pAPS.

Study design
This mixed-methods study adopted a sequential, explanatory design informed by the priority sequence model. 28,29 In this model, an initial decision about the priority of a quantitative or qualitative method is taken. Next, the sequence determines whether the complementary method serves either as a preliminary or a follow-up phase. For this study, a cross-sectional survey was conducted to investigate relationships between fatigue and key psychosocial variables. Then, follow-up qualitative data were collected to explore the interactions identified in the quantitative phase and the direct experience and impact of fatigue in people with pAPS. The

Quantitative phase
Participants and data collection. Patients were eligible to be enrolled onto the study if they were aged !18 years with pAPS (Sydney classification criteria 30 ) for six months or more and had adequate verbal and written English language. Patients who had other autoimmune rheumatic or inflammatory co-morbid conditions, malignancy, active chronic infections, current alcohol and/or drug abuse or dependence, a body mass index > 30 kg/m 2 recorded in their medical records and/or were pregnant or breastfeeding were excluded. Patients with positive ANA were also excluded, as this may indicate APS secondary to other RMDs, such as SLE. 31,32 Members of the direct care team reviewed the medical records of patients attending routine clinical appointments at a tertiary health-care centre for the management of APS in the UK. Potentially suitable patients were identified and approached by the members of the direct care team to gauge their interest in the study and to confirm their eligibility. Interested patients received a questionnaire pack, which comprised study information, a consent form and six questionnaires to self-complete after their appointment. Alternatively, patients completed the questionnaires at home and returned them to the researchers in a prepaid envelope. A researcher offered to support questionnaire completion after the clinical appointment or via telephone at a mutually convenient time. Patients who did not return the questionnaires were sent a second pack four to six weeks later. No further reminders were issued to non-responders.
Sociodemographic and disease characteristics. Sociodemographic and disease characteristics, including age, sex, birthplace, ethnicity, employment status (full-time, part-time, retired, unemployed, higher education or other), disability registration (yes/no) and duration of APS, were collected using a bespoke, selfadministered questionnaire.
Variables. Participants completed five self-completed, validated and standardised questionnaires.
Fatigue. Fatigue over the past week was measured using the 13-item Functional Assessment of Chronic Illness Therapy (FACIT)-Fatigue Scale (FS; a four-point Likert scale ranging from 'not at all fatigued' to 'very much fatigued'), with a lower score indicating greater fatigue (range 0-52). A score of <30 indicates severe fatigue. 33,34 Mood. Mood was assessed with the nine-item Patient Health Questionnaire (PHQ-9; a four-point Likert scale ranging from 'not at all' to 'nearly every day') which is designed to correspond to the diagnosis of depression. A higher score indicates lower mood/ depression (range 0-27), and scores !10 represent clinically depressive symptoms. 35,36 Physical activity. Physical activity over the preceding seven days was measured with the four-item International Physical Activity Questionnaire -short form (IPAQ). The metabolic equivalent of task was calculated over seven days (METS/minutes/week -IPAQMETS), and scores !600 MET/minutes/week were considered physically active. 37 Sleep quality. The quality of sleep over the past month was assessed using the seven-component 19-item Pittsburgh Sleep Quality Index (PSQI; four-point Likert scale ranging from 'not during the past month' to 'three or more times a week'). A higher total PSQI score (range 0-21) represents lower sleep quality. A total PSQI score of > 5 represents severe difficulties in at least two components. 38,39 Perceived social support. Perceived social support from family, friends and significant others was evaluated via the 12-item Multi-Dimensional Perceived Social Support Scale (MSPSS; seven-point Likert scale ranging from 'strongly disagree' to 'strongly agree'). Higher scores indicate higher perceived social support (1-2.9, low; 3-5, moderate; 5.1-7, high). 40,41 Qualitative phase A purposive subsample of participants enrolled in the cross-sectional survey was interviewed by one of two researchers (J.A. or S.G.). 42 The interviewees varied in relation to most sociodemographic characteristics, APS type and lived experiences.
Interviewers were not directly involved in participants' health care and were supervised by an experienced qualitative researcher (H.L.). The audio-recorded interviews were conducted either face to face or by telephone, subject to each participant's preference. A topic guide was developed a priori derived from other RMD fatigue literature 8,9,11 and refined with researchers' and patients' feedback to include questions aligned to the key quantitative variables. The semi-structured interview schedule allowed the interviewer to ask open-ended questions so that the interviewee could diverge or expand the experiences or situations in more detail. 43 A pilot study was completed with three participants to assess comprehension and relevance of questions, timing and subjectively perceived research burden. No further changes were necessary to the interview guide (Table 1). Subtle judgement between interviewers and the supervisor led to recruitment termination when data saturation of themes had been reached (i.e. no new information was reported by the participants). 44 Data analysis Statistical analysis. Analyses were completed using IBM SPSS Statistics for Windows v25.0 (IBM Corp., Armonk, NY). Statistical significance was set at p < 0.05. Frequencies and mean (standard deviation (SD)) values were reported for categorical and continuous descriptive variables, respectively. Relationships with FS (criterion variable) were explored using twotailed Spearman's rho correlation coefficients. A backward stepwise regression was performed including all measures associated with FS at p < 0.05 in the bivariate analyses to determine whether they predicted the FS score. The model with the largest adjusted R 2 model was used to perform a multiple linear regression analysis. Missing values were excluded pairwise. Univariate (Studentised residual valuesAE3SD) and multivariate outliers (Mahalanobis distance p < 0.01) were excluded, and models were evaluated for multicollinearity, normal and independent errors and homoscedasticity.
Qualitative data analysis. The interviews were anonymised and transcribed verbatim by one researcher and one professional transcribing agency, and the text was analysed by two researchers (J.A. and H.L.) using the principles of Framework Analysis 45 via a five-stage structured approach 46 that included (a) familiarisation with text, (b) coding within the qualitative computer package NVIVO v10 (QSR International Pty Ltd), 47 (c) building categories and themes, (d) identification of a thematic framework and (e) linking findings with theoretical concepts. Validation of data included (a) checking initial codes with one external researcher from one transcript, (b) presentation of initial findings from the pilot study to the supervisor, (c) discussing emerging themes from two additional transcripts and (d) single counting. Transcript accuracy was checked by the participants against original recordings. A balanced reporting of data is a recognized validation strategy. 47 Therefore, a range of accounts are presented when available.

Descriptive statistics
A total of 105 potentially eligible patients were approached between July 2014 and June 2017. Two people with communication difficulties (one with a hearing impairment and one with insufficient English) were excluded. A total of 103 participants were therefore enrolled onto the study (18 males, M age ¼ 50.3 years, SD ¼ 10.1 years). Data from 16 participants (10 male, M age ¼ 49.7 years, SD ¼ 10.1 years) were omitted due to incomplete data. Therefore, 87 participants (8 male, M age ¼ 50.3 years, SD ¼ 10.4 years) were included in the analysis (Table 2). A total of 62% of participants reported severe fatigue, 40% reported depression, 24% were classified as physically inactive, 69% revealed severe sleep difficulties and 35% reported low-moderate levels of perceived social support.  Table 3).

Multivariate analyses
The backward stepwise regression included all variables ( Table 4). The final best-fit model explained 56% of the variance in FS (adjusted R 2 ¼ 0.560, F(3, 74) ¼ 33.65, p < 0.001) and included PHQ9 and IPAQMETS as significant predictors and PSQI as a non-significant predictor but not MSPSS (Table 5).

Results: qualitative phase
Twenty consenting participants were interviewed (duration: 20-60 minutes). The participants were all female and predominantly white ( Table 2). Three key themes were identified: (a) characteristics of fatigue, (b) impact on life and (c) coping strategies, with two to four linked subthemes. Unpredictability and fluctuation of fatigue. Our participants (14/20) described that fatigue was unpredictable and variable and therefore interfered with their daily activities of living. They were uncertain about the causes or triggers for their fatigue. For example: 'You know, the fatigue like I said is, it can happen any day and come out of the blue' (P15) and 'So it's very variable [fatigue], and it's not even the fact that I've overdone it the day before, which sometimes can help bring on the fatigue even more, but sometimes, umm, I've been   Impact of mood. Some participants (8/20) described how mood and negative thoughts affected their fatigue. These thoughts could be intrusive and unhelpful, and this influenced their ability to accomplish their daily activities. For example: . Impact on work. There was no doubt that fatigue impacted upon participants' salaried or daily work, which meant they had to make adaptations and were feeling unable to carry out complex tasks at times. For example: '[I] work flexi-time, so on a day where I just think I cannot do this anymore, I just need to get myself home, I can leave within reason' (P10) and '. . .I only work part time, whether I would be able to work full-time? I don't know the answer to that. Maybe I would get too fatigued if I worked full time' (P6).

Theme 3: coping strategies
All participants (20/20) employed a range of coping strategies to manage their fatigue. The strategies selected depended on the severity of the fatigue and were often informed by past successful experiences of dealing with their fatigue.
Individual coping strategies. Most participants (16/20) described how they applied individual coping strategies in their personal life, and pacing was a popular strategy.

Discussion
This mixed-methods study is one of the first to explore the experience and impact of fatigue in adults with pAPS. The findings showed that fatigue was common, unpredictable and sometimes overwhelming in people with pAPS. It was influenced by factors such as low mood and physical inactivity. Participants had developed their own coping strategies to manage their fatigue and generally reported receiving high levels of social support. They reported minimal or variable understanding from health-care professionals. Almost two thirds of our participants reported experiencing severe fatigue at times. Participants were vigilant about monitoring their fatigue so they could adopt strategies to mitigate the symptom. However, they found fatigue challenging to manage and difficult to explain to others, including health-care professionals. Consequently, participants used metaphors to express the character of fatigue and convey their increased effort and loss of energy, similar to people with long-term conditions such as SLE 9 or RA. 11 More than a third of our participants reported low mood, and this was associated with higher fatigue levels. The relationship between mood and fatigue is well described in people with other long-term conditions, 16,48 and mood predicts fatigue, even after adjustment for pain and other potentially confounding variables in some RMDs. 20 Low mood can affect sleep, and many of our participants reported sleep difficulties like people with other RMDs. 15,17,20 Sleep disturbances could be due to pain, stress or influenced by medication. 17 Whilst many interviewees described a clear link between fatigue and sleep, sleep quality did not independently predict fatigue in our quantitative data. This may be because sleep quality has an indirect effect on fatigue and mediates the direct influence of depression or physical inactivity on fatigue. 15,17 More than three quarters of our respondents reported meeting physical activity recommendations, which is higher than people with other RMDs. 22,23,49 This may be due to over-reporting because these data were collected using self-completed questionnaires, which is subject to recall bias. Alternatively, since many participants were employed or had active caring responsibilities, this finding may reflect activity undertaken for transportation (i.e. walking) related to their daily work. This study showed that greater physical activity was directly related to lower fatigue, and some interviewees stated that physical activity/exercise was a helpful management strategy for fatigue. Brief home-based physical activity and exercise interventions improve fatigue and sleep quality in people with RA. So, this may be a promising intervention approach for people with pAPS. 50,51 However, some interviewees found the physical activity hard to complete due to the unpredictability of fatigue. So, tailored strategies or interventions may be needed to accommodate fluctuations in fatigue.
Participants developed their own coping strategies, particularly pacing, to manage their fatigue. They mostly received good social support from family, friends and significant others (i.e. perceived emotional support), concurring with other research, 52 and it is proposed that social support influences self-efficacy and has a buffering effect on fatigue. 20,26 Whilst social support correlated with fatigue in our study, it did not independently explain the variance in fatigue in our multivariate analyses and exploration of the other types of perceived social support (i.e. instrumental (practical support such as housework) or informational (education) support) may clarify this relationship. 52 Informational social support could be provided by clinicians, although our participants commented that fatigue was not regularly discussed in consultations. This could be because clinicians do not recognise the presence or impact of fatigue in people with pAPS or because it is not acknowledged in management recommendations. 27 Alternatively, the lack of evidence-based management strategies for fatigue in pAPS may mean that it is rarely a treatment focus. A multidisciplinary team approach may optimise the management of pAPS, similar to other RMDs such as RA. [53][54][55] The conceptual model of fatigue for patients with RA by Hewlett et al. aligns with our findings. 14 Factors such as pAPS itself, cognitive and behavioural dimensions (e.g. mood, physical activity) and personal aspects (e.g. social support) play a dynamic role in the experiences of fatigue and how people cope with it. This model may help researchers, clinicians and patients clarify the focus for future interventions and measurements to inform treatment and selfmanagement.
This study has several strengths. The mixedmethodology approach facilitated the integration of the findings from both phases within a single study. The findings are complementary and provide a deeper insight into the experience of fatigue in adults with pAPS. 28,29 Our eligibility criteria were robust, and adults with co-morbidities known to influence fatigue (e.g. obesity) were excluded to minimise confounding factors. The number of medical records reviewed and reasons for patients' ineligibility following review of medical notes were not documented. However, the team enrolled a range of participants with pAPS and only two potentially suitable patients, who were identified and approached by the clinic team, were excluded. Our cohort reflected the female predominance which exists in pAPS. However, patients were recruited from a single centre, there was relatively limited ethnic and sex diversity in our study population and the educational status of the participants was not recorded.
Our survey explored key factors known to influence fatigue in other RMDs, but other aspects may contribute to fatigue (e.g. physical capacity, end organ damage), and these require investigation. As data from our quantitative phase was cross-sectional, no causality can be inferred from the findings, and the reciprocal effects of fatigue on other features cannot be observed (e.g. greater fatigue may lead to lower mood or less activity).
Fatigue is common and impacts the lives of adults with pAPS. Enhancing communication between patients and clinicians so that symptoms which are important to patients, such as fatigue, are identified and addressed appropriately is crucial to optimise the management of pAPS. Developing effective multidisciplinary interventions to support self-management of fatigue with patients, clinicians and researchers is vital.