High burden of fatigue amongst socioeconomically disadvantaged South Africans with rheumatoid arthritis

Introduction

Fatigue is a common and incapacitating symptom among rheumatoid arthritis (RA) patients, often causing a profound negative impact on health-related quality of life (HRQoL). Acknowledging its significance, the Outcome Measures in Rheumatology Clinical Trials (OMERACT) group has emphasized the importance of evaluating fatigue in all RA-related research (1).

Fatigue is a subjective experience characterized by feelings of weakness, diminished energy, or exhaustion that occur independent of physical activity and do not subside with rest (2). This multidimensional symptom has been described by patients as overwhelming, unearned, and unpredictable (3, 4). The prevalence of fatigue in RA varies, with studies reporting that up to 70% of patients experience some fatigue, with a significant proportion experiencing severe fatigue (5–8).

Fatigue has a significant impact on HRQoL, affecting physical and social functioning, physical and emotional roles, and mental health (9). Patients have also identified fatigue as a significant barrier to employment, and high levels of fatigue are associated with increased absenteeism, activity impairment, and poor work productivity (10, 11).

Despite the prevalence and impact of fatigue in RA, the causes and associations of this problem are complex and not fully understood. Of note, fatigue is only weakly correlated with disease activity, and medications that improve disease activity, including biologics, only partially improve fatigue (12–14). Many patients with well-controlled inflammation assessed as being in remission experience persistent fatigue that prevents them from perceiving favorable clinical outcomes (15, 16). Several models have been developed highlighting the characteristics linked with fatigue that are unrelated to RA. These factors include demographic characteristics such as age, level of education, gender, employment status, socioeconomic status, obesity, and comorbidities. In addition, RA-related symptoms, medications used to treat RA, particularly prednisone, and patient-related aspects, particularly mental health, have an impact (17–19). The magnitude of the influence of these factors may differ across populations.

These models are consistent with the conceptual model of RA fatigue proposed by Hewlett et al. (20), which identifies three interacting domains: RA-related factors (including inflammation, medication, pain, sleep, anemia, deconditioning, disability, and joint damage), personal factors (health, environment, work, caring responsibilities, and social support), and cognitive/behavioral factors (activity patterns, illness beliefs, stress, anxiety, and depression) (20, 21). This framework guided the selection of variables for the present study: disease activity, prednisone use, pain, and functional disability represent the RA-related domain; socioeconomic status, employment, and multimorbidity represent the personal domain; and anxiety and depression represent the cognitive/behavioral domain.

Fatigue and factors associated with fatigue have not been specifically explored in sub-Saharan Africa. This study is undertaken to explore fatigue in a cohort of socioeconomically disadvantaged RA patients in South Africa (SA) and to provide valuable insight into the challenges faced by people living with RA in this region, hopefully leading to customized interventions that address their needs effectively.

Patients and methods

A cross-sectional analysis was conducted using data from an ongoing rheumatoid arthritis registry at a tertiary state-sector academic hospital in the Western Cape, South Africa. The registry was established under ethics approval HREC R045/2020, and the present fatigue analysis was approved separately under HREC 629/2023. Consecutive adults aged ≥18 years who fulfilled the 2010 ACR/EULAR classification criteria for RA (22) and attended the registry clinics between 14 January 2021 and 12 December 2023 were eligible. All participants provided written informed consent before enrolment.

No a priori sample-size calculation was performed; all consecutive eligible registry participants during the study period were included, yielding a pragmatic sample of 550.

Data collection

Socio-demographic details, including age, gender, self-expressed ethnicity, level of education, employment status, smoking status, and living situation, were collected. Socioeconomic status was assessed using a composite multidimensional deprivation index comprising four equally-weighted binary indicators: receipt of a disability grant or unemployment, low educational attainment (primary school or less), household crowding (>7 household members), and lack of access to domestic technology, defined as washing clothes by hand. Education alone was considered an inadequate proxy for current socioeconomic circumstances because educational attainment among older South African adults remains strongly influenced by historical inequalities under apartheid-era policies. Participants were classified as socioeconomically disadvantaged using the union approach within the Alkire–Foster multidimensional poverty framework, whereby each indicator was weighted equally and deprivation in at least one indicator was sufficient for classification (poverty cut-off k = 1) (23). This study-specific index has not undergone separate external validation as a formal SES instrument.

Disease-specific information, including disease duration, medication, extra-articular manifestations, the presence of rheumatoid factor (RF) and/or anti-cyclic citrullinated protein antibodies (ACPA), and comorbidities, was collected. Disease activity was evaluated using the Clinical Disease Activity Index (CDAI), which combines physician and patient global assessments with swollen and tender joint counts (24). Patients were classified by CDAI into remission/low disease activity (LDA) (< 10.0), moderate (MDA) (≥ 10.0 to < 22.0), and high (HDA) (≥ 22.0) states. Remission and LDA were combined into a single category (“LDA”) for descriptive and analytic purposes, given the small remission subgroup. Multimorbidity was defined as the presence of ≥2 comorbidities.

Assessment of fatigue

Fatigue was evaluated using the FACIT fatigue (FACIT-F) scale, a 13-item validated tool measuring self-reported fatigue. Participants provided responses to each item on a 0–4 scale, allowing the calculation of the composite FACIT-F score, ranging from 0 to 52, and two sub-scores: fatigue experience and fatigue impact scores (25–27). Higher scores indicate less fatigue. Participants scoring 39 or lower on the composite FACIT-F scale were categorized as experiencing fatigue, with four grades of severity: no or little fatigue (score 40–52), some fatigue (score 27–39), quite a lot of fatigue (score 14–26), and extreme fatigue (score between 0 and 13).

For binary analyses of severe fatigue, patients were dichotomized at the boundary between the “some” and “quite a lot” categories (FACIT-F < 27), combining the two most severe categories to define “severe fatigue”. This is an internally derived, not externally validated, cut-point based on the pre-specified four-category severity classification above.

Other patient-reported outcome measures (PROMs) included the Health Assessment Questionnaire-Disability Index (HAQ-DI) to evaluate functional disability, EuroQol (EQ-5D) to measure health-related quality of life, the Hospital Anxiety and Depression Scale (HADS), and the Brief Pain Inventory, from which the pain severity score (PSS) and pain interference score (PIS) were calculated (28–31).

Anxiety and depression were assessed using the Hospital Anxiety and Depression Scale (HADS). Each seven-item subscale ranges from 0 to 21, with higher scores indicating greater symptom burden. A subscale score ≥11 was used to define anxiety or depression caseness. Incomplete subscale scores were treated as missing.

All questionnaires (FACIT-F, HADS, HAQ-DI, EQ-5D, and BPI) were administered in English, with assistance from the study nurse provided if needed.

Statistical analysis

Continuous variables were summarised as mean (SD) or median (IQR), according to their distribution, and categorical variables as frequencies and percentages. Distributional assumptions were assessed using skewness statistics and inspection of the distributions. Variables with an absolute skewness value <1 were treated as approximately symmetric. Associations with the continuous FACIT-F score were assessed using Pearson's correlation coefficient for approximately symmetric variables and Spearman's rank correlation coefficient for skewed variables.

For the binary severe-fatigue analysis, participants with FACIT-F scores ≥27 and <27 were compared using independent-samples t-tests for approximately symmetric continuous variables, Mann–Whitney U tests for skewed continuous variables, and chi-square or Fisher’s exact tests for categorical variables, as appropriate.

Variables associated with the relevant outcome at p < 0.05 in univariable analyses were considered for inclusion in forced-entry multivariable linear regression, with the continuous FACIT-F score as the dependent variable, and binary logistic regression, with severe fatigue as the dependent variable. Where variables were mathematically or conceptually overlapping, the more specific measure was prioritized. The EQ-5D anxiety/depression and pain/discomfort domains were excluded in favour of HADS and Brief Pain Inventory scores, respectively. The EQ-5D-3L Level Sum Score was excluded because it is derived from the five EQ-5D domains and overlapped with the individual functional domains, pain measures and HADS scores. CDAI was excluded in favour of its individual components: TJC, SJC, PGA and PhGA. Multicollinearity was assessed using variance inflation factors.

Univariable analyses used available observations for each variable, whereas multivariable analyses used complete cases. All tests were two-sided, and p < 0.05 was considered statistically significant. Analyses were performed using IBM SPSS Statistics version 25.

Results

Of 550 patients, 467 (84.9%) were female, and the mean (SD) age was 55.8 (12.9) years and disease duration 11.5 (9.7) years (Table 1). Most patients (67.1%) had poor SES, with only a fifth employed. The vast majority (91.6%) were seropositive for either RF or ACPA (among the 440 patients with available serology results; 110 patients, 20.0%, had no recorded serology).

Table 1. Baseline characteristics of 550 RA patients.

ACPA, anti-citrullinated protein antibody; bDMARDs, biologic disease-modifying antirheumatic drugs; csDMARDs, conventional synthetic disease-modifying antirheumatic drugs; HAQ-DI, health assessment questionnaire–disability index; LDA, low disease activity; ns, not significant; RA, rheumatoid arthritis; RF, rheumatoid factor.

Percentages are based on participants with available RF/ACPA results (n = 440).

Remission (n = 65) and LDA (n = 198) were combined for descriptive analysis.

Medication-category data were available for 516 participants. Categories were mutually exclusive, and patients receiving biologic therapy were classified in the bDMARD category regardless of concomitant csDMARD use.

Respiratory disease included interstitial lung disease, asthma, chronic obstructive pulmonary disease and pulmonary tuberculosis.

Cardiovascular disease included ischaemic heart disease, valvular heart disease, arrhythmia and congenital heart disease. Unless otherwise stated, percentages were calculated using valid, non-missing responses; denominators therefore vary slightly between variables.

In terms of disease activity, 47.8% were in LDA, 30.2% in MDA, and 22.0% in HDA. Regarding treatment, 39.1% were using a single conventional synthetic disease-modifying antirheumatic drug (csDMARD), 58.3% used two or more csDMARDs, and 2.5% were prescribed biologic DMARD therapy.

Overall, 489 patients (88.9%) were receiving prednisone: 349 (63.5%) received ≤7.5 mg/day and 140 (25.5%) received >7.5 mg/day; 61 patients (11.1%) were not receiving prednisone.

Comorbidities were present in 75.1% of patients, and 225 (40.9%) had multimorbidity (≥2 comorbidities).

Patients experienced significant functional disability, with a mean (SD) HAQ-DI of 1.5 (0.7) and two-thirds reporting problems with mobility and self-care (Table 2). Pain scores were high on the EQ-5D and BPI-SF, and a third of patients had significant anxiety or depression.

Table 2. Patient-reported outcome measures among 550 patients with rheumatoid arthritis.

Values are presented as mean (SD) or n/N (%). Percentages are based on participants with valid data. The EQ-5D-3L pain/discomfort and anxiety/depression variables represent participants reporting any problem in the respective domain. HAQ-DI, health assessment questionnaire—disability index; EQ5D, European quality of life-5 dimension.

Fatigue

The composite FACIT-F score was 28.7 (13.4), with 73.3% experiencing fatigue. In terms of severity, 30.5% reported “some fatigue,” 26.9% experienced “a lot of fatigue,” and 15.8% grappled with “extreme fatigue” (Figure 1). There was a robust linear correlation between the total FACIT-F score and its two sub-components (experience r = 0.91 and impact r = 0.96).

Individual FACIT-F item responses varied by domain. Notably, 36.1% and 37.1% felt their fatigue made starting or finishing tasks difficult or very difficult; 34.4% experienced “quite a bit” or “very much” limitation in their social activities; 31.0% needed to sleep during the day; and 37.0% felt frustrated at being too tired to do the things they want to do.

There were significant differences in the level of fatigue scores between disease activity states, with patients in LDA showing less severe fatigue levels and patients in HDA being more likely to have severe or extreme fatigue (Figure 2).

The correlation analysis revealed significant associations between FACIT-F scores and various factors (Table 3). Age, but not disease duration, was correlated with fatigue (r = 0.131, p = 0.002; with younger individuals reporting higher levels of fatigue). Smoking (r = −0.136, p = 0.001), low SES (r = −0.106, p = 0.013), fibromyalgia syndrome, and the presence of multimorbidity (r = −0.132, p = 0.002) were associated with increased fatigue, but there was no association with gender, body mass index, or education level. Disease activity and its components also displayed negative correlations with the composite FACIT-F total score and sub-scores. Significant negative correlations were found with all PROMs, including general quality of life, pain (PSS r = −0.415; PIS r = −0.434, both p < 0.001), functional disability, anxiety (r = −0.477, p < 0.001), and depression (r = −0.472, p < 0.001). In terms of RA therapy, prednisone use showed a very weak correlation with fatigue.

Table 3. Factors associated with FACIT-F score: univariable correlations and multivariable linear regression.

Higher FACIT-F scores indicate less fatigue; therefore, negative adjusted coefficients indicate greater fatigue.

RA, rheumatoid arthritis; BMI, body mass index; TJC, tender joint count; SJC, swollen joint count; PGA, patient global assessment; PhGA, physician global assessment; CDAI, Clinical Disease Activity Index; csDMARDs, conventional synthetic disease-modifying antirheumatic drugs; HAQ-DI, Health Assessment Questionnaire–Disability Index; EQ-5D-3L, EuroQol 5-Dimension 3-Level questionnaire; HADS, Hospital Anxiety and Depression Scale; PSS, Pain Severity Score; PIS, Pain Interference Score; DM, diabetes mellitus; CVD, cardiovascular disease.

Univariable analyses used continuous HADS scores; multivariable analyses used anxiety or depression caseness, defined as a subscale score ≥11.

Spearman's ρ; all unmarked univariable coefficients are Pearson's r.

Variables associated with FACIT-F scores at p < 0.05 in univariable analyses were considered for simultaneous forced-entry regression after resolving overlapping measures. Multivariable model: n = 516.

In the multivariable linear regression model involving 516 patients, younger age, current smoking, multimorbidity, problems with usual activities, greater pain severity and interference, and HADS anxiety and depression were independently associated with lower FACIT-F scores (i.e., greater fatigue). Current prednisone use showed a borderline adjusted association with lower FACIT-F scores.

Factors associated with severe fatigue

Patients with no/little or some fatigue (FACIT-F ≥ 27) were compared with those with severe fatigue (FACIT-F < 27); univariable comparisons are presented in Table 4. In the multivariable logistic regression model involving 516 participants, severe fatigue was independently associated with current smoking [adjusted odds ratio (aOR) 1.78, 95% CI 1.14–2.79], multimorbidity (aOR 2.22, 95% CI 1.40–3.53), problems with usual activities (aOR 1.91, 95% CI 1.02–3.60), greater pain severity (aOR 1.23 per one-point increase, 95% CI 1.06–1.42), greater pain interference (aOR 1.18 per one-point increase, 95% CI 1.04–1.35), HADS anxiety (aOR 3.13, 95% CI 1.94–5.06), and HADS depression (aOR 1.96, 95% CI 1.22–3.16).

Table 4. Factors associated with severe fatigue in patients with rheumatoid arthritis: univariable comparisons and multivariable logistic regression.

Data are presented as mean (SD), median (Q1–Q3), or n (%), as appropriate. Severe fatigue was defined as a FACIT-F score <27. Univariable analyses used available cases; the multivariable logistic regression included 516 complete cases. Adjusted odds ratios for continuous predictors represent the change in odds per one-unit increase. Denominators vary for variables with missing data (HADS Anxiety: 311 minimal + 231 severe = 542; HADS Depression: 300 minimal + 217 severe = 517).

RA, rheumatoid arthritis; BMI, body mass index; RF, rheumatoid factor; ACPA, anti-citrullinated protein Antibody; TJC, tender joint count; SJC, swollen joint count; PGA, patient global assessment; PhGA, physician global assessment; CDAI, Clinical Disease Activity Index; LDA, low disease activity; MDA, moderate disease activity; HDA, high disease activity; ns, not significant; csDMARDs, conventional synthetic disease-modifying antirheumatic drugs; bDMARDs, biologic disease-modifying antirheumatic drugs.

Overall p-value for the three-category disease-activity comparison using the Pearson chi-square test.

Discussion

Fatigue is a pervasive and debilitating yet invisible symptom that significantly impacts the HRQoL of RA patients. Our study, the first in SA, shows the significant burden of fatigue, with almost three-quarters of patients reporting fatigue to varying extents. This aligns with the worldwide literature, highlighting the need for healthcare providers to acknowledge fatigue as an important symptom of RA to be assessed and managed in conjunction with conventional joint-related symptoms and signs of RA (5–8, 32). Our study revealed that both RA-related factors and non-RA-related patient factors contribute to fatigue: pain, functional disability, younger age, smoking, multimorbidity, anxiety, and depression were significantly associated with fatigue.

Higher CDAI scores were associated with greater fatigue in univariable analysis. CDAI was not entered into the multivariable models because it is a composite of TJC, SJC, PGA and PhGA, which were included separately to avoid mathematical redundancy; none of these individual components remained independently associated with fatigue after adjustment. Consistent with the univariable findings, high disease activity was more frequent among patients with severe fatigue than among those with FACIT-F scores ≥27. Although several studies have reported an association between fatigue and disease activity, others have found little or no relationship, supporting the view that fatigue in RA is influenced by both inflammatory and non-inflammatory factors (33–36).

Pain, a hallmark of RA, emerged as a factor strongly associated with fatigue in our study. These results align with previous research in the field, which consistently highlights the close relationship between pain and fatigue (8, 33, 35, 37, 38). In addition, our observation that psychological distress, particularly anxiety, was strongly associated with fatigue has been shown in multiple previous studies (17, 39, 40). Functional disability, resulting in constraints in performing daily activities like walking, dressing, and grooming, was also significantly associated with fatigue in this study and elsewhere (33, 41). Strategies to manage pain, mental health, and disability could potentially alleviate fatigue.

Consistent with Hewlett et al.’s dynamic conceptual model of RA fatigue (20), in which RA-related, personal, and cognitive/behavioural factors interact bi-directionally both within and between domains, the associations we observed between pain, disability, anxiety and depression, and fatigue likely reflect mutually reinforcing relationships rather than a simple unidirectional causal pathway.

We observed that younger patients experienced significantly higher levels of fatigue compared to their older counterparts, which is consistent with similar research conducted elsewhere (41). A qualitative study from Sweden suggested that younger individuals, especially those in the working age group, often contend with elevated levels of fatigue due to their numerous daily responsibilities and commitments (42).

Additional demographic and lifestyle factors, together with comorbidities, including smoking habits, lower SES, a higher burden of comorbid conditions, and the presence of fibromyalgia, were linked to heightened fatigue levels, albeit with modest correlations. Davis et al. also showed that multimorbidity was linked to higher levels of fatigue in RA, with an impact on physical, living, cognitive, and emotional aspects of life (43).

This study has several limitations. Participants were recruited from a single tertiary academic hospital and were limited to those who consented to participate, which may reduce the generalizability of the findings to the broader RA population. The cross-sectional design precludes causal inference and cannot exclude reverse causation; for example, fatigue may influence pain perception, coping and activity levels, rather than only resulting from these factors. Several patient-reported measures retained in the multivariable models, including HAQ-DI, EQ-5D functional domains and pain interference, assessed related functional constructs. Although multicollinearity was acceptable (maximum variance inflation factor 2.35), this overlap may have attenuated the estimated independent contribution of individual measures. Objective inflammatory markers, including CRP and ESR, hemoglobin, sleep quality and physical activity were not assessed and may represent unmeasured confounders. Questionnaires were administered only in English, with the assistance from study-nurse as needed, which may have introduced measurement bias among participants with limited English proficiency, particularly among the 20.5% of the cohort with primary-level education or less.

In conclusion, our study provides valuable insights into the prevalence, severity, and factors associated with fatigue in RA patients in SA. It confirms the importance of fatigue in RA and highlights the multifaceted nature of this symptom. Integrating psychological support, pain management, comorbidity care, and patient awareness into RA management will be essential to address this complex interplay of factors associated with fatigue. Ultimately, our findings emphasize the need for a patient-centered approach that recognizes the individualized symptoms of RA. Future qualitative research will explore cultural, social, and healthcare access factors unique to our socioeconomically disadvantaged population. Interventional and longitudinal studies assessing the impact of interventions are planned.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Maria Chiara Maccarone, University of Padua, Italy

Reviewed by: Manali Shah, Icahn School of Medicine at Mount Sinai, United States

Samer Imad Mohammed, University of Baghdad, Iraq