A systematic review of the association between fatigue and cognition in chronic noncommunicable diseases

Abstract

Objectives:

Fatigue is one of the most common symptoms associated with chronic noncommunicable diseases, and it may also increase cognitive impairment. However, associations between fatigue and cognitive impairment in chronic illnesses remain unclear. Therefore, the purpose of this systematic review was to examine research that investigated associations between level of fatigue and cognitive status.

Methods:

PubMed/Medline, PsycINFO, CINAHL, and Cochrane Database were searched for articles published between 2012 and 2018 using search terms fatigue, cognition, and various iterations of these terms. Study quality was assessed by the Joanna Briggs Institute Critical Appraisal Checklist tool.

Results:

Of 1799 citations, 10 studies in samples of individuals with cancer, multiple sclerosis, neurosarcoidosis, and chronic fatigue syndrome met the inclusion criteria. Fatigue was found to be significantly correlated with cognitive impairment in one cancer-related study (r=−.480, p<.001), one multiple sclerosis study (β= −0.52, p<.0001), and two chronic fatigue syndrome studies (r=0.397, p<.001; r=0.388, p<.001).

Discussion:

There is insufficient evidence examining the relationship between fatigue and cognitive impairment in patients with chronic illnesses. As a result, more studies are needed that examine potential relationships between these two symptoms in order to develop effective treatments for individuals living with a chronic noncommunicable disease.

Introduction

As the population ages, more people across the world are living with at least one noncommunicable disease (NCD) such as cardiovascular disease, cancer, diabetes, and chronic respiratory disease.¹ To enhance the quality of life, reduction of symptom burden is a priority for individuals living with NCD and their health care providers. Fatigue is the most commonly reported “transdiagnostic” symptom across chronic diseases.² While fatigue can be among the side effects of drug treatment or the result of a psychiatric disturbance in functions such as a somatoform disorder,³ it is most commonly identified as part of medical conditions. Reported conditions have included post-chemotherapy cancer survivors,⁴ multiple sclerosis,⁵ rheumatologic conditions,^6,7 diabetes,⁸ heart disease,⁹ and Parkinson’s Disease.¹⁰ Along with fatigue, symptoms reported by individuals with NCDs often include depression, pain, and sleep disturbances and are symptoms that often co-occur and may share common biological pathways across conditions.¹¹ Although there is much less research on NCDs and cognitive dysfunction, there is an emerging literature about the high prevalence of cognitive dysfunction in individuals living with NCDs, particularly those with an inflammatory pathogenesis. The intersection of fatigue and cognition is particularly compelling, given the deleterious effects of each on work and role performance.^12-14

Fatigue is defined as a subjective symptom marked by a generalized sense of depleted energy, feelings of tiredness, or lack of energy that leads to a decreased capacity for physical or mental performance.^15-17 Although healthy people may experience fatigue, it is one of the most common complaints among individuals suffering from chronic illness.^16,18 In healthy persons, fatigue occurs as a normal and temporary phenomenon, whereas illness-related fatigue may persist despite adequate amounts of rest and sleep. From an epidemiological point of view, estimates of fatigue occurring in the US general population are reported to range from 15 to 25%.¹⁸ Prolonged fatigue, or fatigue lasting one to five months, has been reported to range from 3.7 to 7.68%^18,19 while the prevalence of chronic fatigue (fatigue lasting > 6 months) reportedly ranges from 2.72 to 10.8%.¹⁸ In the workplace, Lu et al.²⁰ identified a high weekly prevalence (57.9%) of fatigue among surveyed manufacturing workers, while Ricci et al.²¹ noted that fatigued workers reported health-related and lost productive time more than twice as often as those without fatigue. From an economic point of view, Ricci et al.²¹ estimated that both short-term and long-term fatigue outcomes cost US employers $136 billion annually.

Fatigue descriptive

Fatigue is a symptom that often accompanies NCDs.²² Because it is a subjective state that has both physical and psychological elements,¹⁷ fatigue is often characterized by the context in which it is experienced. More specifically, fatigue may occur as a result of a physical or psychological event, or fatigue may be the cause of a physical or psychological event. When seeking to define the concept of fatigue, it appears to be defined by overlapping terms of cognitive or mental fatigue; however, the descriptives for each slightly vary. For example, cognitive fatigue is said to occur when cognitive performance decreases because of engaging in tasks that require sustained activity.²³ Mental fatigue has been defined as a subjective feeling of tiredness and inertia that occurs during extended periods of demanding cognitive activity.²⁴ Fatigue has also been defined as physiological fatigue, which is described as muscle weakness that may occur due to exercise.²⁵

Fatigue and cognition in chronic NCD

For individuals living with a chronic NCD, fatigue is often a common complaint, as are complaints of cognitive dysfunction.^26-28 Cognition has been defined as “mental operations that relate to logic, awareness, intellect, memory, language, and reasoning powers” with cognitive domain defined as “a category of learning that involves knowledge and thought processes within the individual’s intellectual ability [who] must be able to synthesize information at an intellectual level before actual behaviors are performed”²⁹ (p. 807). Because fatigue is often associated with the level of cognition, and both fatigue and decreased cognitive function are known to negatively impact activities of daily living,^28,30 it is relevant to seek to understand the putative nature of this relationship for purposes of prioritizing patient care. We thus hypothesized that there is a relationship between fatigue and cognition in the chronically ill individual and that there is a predictive relationship between the two variables. Therefore, the purpose of this systematic review was to examine and synthesize literature that explored potential associations between reported levels of fatigue and self-reported and/or objectively measured cognitive status in individuals diagnosed with a chronic NCD. Study findings are discussed in relation to type of illness, measures used, presence, or absence of a reported relationship between fatigue and cognition, and, if reported present, we noted whether the relationship was found to be predictive. That is, was fatigue found to be a predictor of cognitive impairment or vice versa?

Methods

Search strategy and data sources

The systematic review was guided by the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.³¹ The electronic databases PubMed/Medline, PsycINFO, Cumulative Index of Nursing and Allied Health Literature (CINAHL), and the Cochrane Database were searched for relevant articles. The primary search terms were fatigue and cognition; however, in an effort to capture types of fatigue, descriptive search terms were meshed and presented in the following format in both PubMed and PsychInfo: (((((“Fatigue” [Majr]) OR “Mental Fatigue”[Majr])) OR ((((((“illness related fatigue”[Text Word]) OR “physical fatigue”[Text Word]) OR “illness fatigue”[Text Word]) OR “emotional fatigue”[Text Word]) OR “mental fatigue” [Text Word]) OR “fatigue”[Text Word]))) AND ((((“Cognition”[Majr]))). Because the phrase “impact of fatigue on cognition” was frequently encountered when reviewing abstracts, this phrase was used in a separate search in PubMed database with 316 additional articles added from the keyword search. Using the meshed search terms in CINAHL revealed one abstract that did not meet inclusion criteria for review. The Cochrane Database would not accept the meshed format; therefore, three keywords were used (fatigue, cognition, impact) in Cochrane and repeated in CINAHL with a total of 10 and 135 abstracts retrieved, respectively. To complete the literature search, the reference lists of qualifying articles were reviewed with no additional articles added. Two reviewers (VM and DK) independently reviewed abstracts and subsequent full-text articles and summarized the key themes identified according to the PRISMA checklist.³¹ At each step, the results were discussed and discrepancies were resolved to obtain consensus. Eligible studies were sorted based on chronic illness domains, including cancer (e.g. colorectal cancer, breast cancer (BC)), multiple sclerosis, immunological/multi-system disorders, and chronic fatigue syndrome. Data of each study were extracted by three reviewers (VM, DK, and AS) and confirmed by two other reviewers (DL and GY). This strategy is depicted in Figure 1.

Figure 1.

Flowchart of identification, screening and selection process in current review.

Inclusion and exclusion criteria

Studies were included in the review if they: (a) were published in the English language, (b) examined a relationship between fatigue and cognition, and (c) were published within six years of the publication submission process (2012–2018). The limited range of six years was selected as inclusion criteria for purposes of examining the most recently reported understandings in science in relation to potential associations between reported levels of fatigue and self-reported and/or objectively measured cognitive status in individuals diagnosed with a chronic NCD. Additionally, reference lists of all qualifying articles were carefully reviewed and analyzed for potential inclusion.³² Studies were excluded if: (a) an intervention or a drug was administered to effect a change in fatigue, cognition, or both or if they included (b) sleep studies, (c) psychometric studies, (d) laboratory studies in which fatigue was artificially induced on healthy subjects, (e) surveys, (f) case studies, and (g) dissertations.

Study quality appraisal

For the articles that met the inclusion criteria, quality assessment was performed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklists for Analytical Cross-Sectional Studies and Cohort Studies, which includes 8 and 11 questions, respectively, to assess the study quality and rigor of qualitative research. Each question is answered as “Yes,” “No,” “Unclear,” or “Not applicable.” Cut-off scores between one and ten were established for evaluating the methodological quality of the articles as previously reported; scores 1–4 represented articles with low methodological quality; scores 5–7 represented articles with moderate quality; and scores 8–10 represented articles with high quality.³³ Methodological qualities of the included studies were assessed independently by two authors (AS and GY). Lack of consensus between the authors was resolved by discussion or by a third author. All of the studies were evaluated as moderate to high quality and were included in the systematic review.

Results

Overview of selected studies

After removing the duplicates, a total of 1761 abstracts were examined in depth, with 69 articles ultimately assessed for eligibility based on selection criteria. A total of 10 articles met the inclusion criteria and are included in this review.

Reported research designs were predominantly cross-sectional (N=7) followed by two longitudinal observational studies^4,5 and one secondary analysis.³⁴ Of the 10 articles reviewed, four studies were conducted in the United States; one in Canada; one in China; and four in European countries (Germany; Germany and The Netherlands; Spain; United Kingdom), for a total of 1406 participants with chronic illnesses and 185 healthy controls. Reported study participants included individuals diagnosed with cancer (N=5), multiple sclerosis (N=2), neurosarcoidosis (N=1), and CFS (N=2), and the average age of participants in the studies ranged from 46.7 to 58.8 years. Because of the variability in how fatigue and cognition were addressed and measured among these studies, the value of our analysis lies in reporting results based on patient population, research design, and noting (a) whether there was a reported relationship between fatigue and cognition and (b) whether there was a directional relationship, if any, between variables reported. Fatigue and cognitive function outcomes were measured by validated, self-reported questionnaires, such as the Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F), Patient-Reported Outcomes Measurement Information System (PROMIS), Fatigue Impact Scale (FIS), neuropsychological test battery, Functional Assessment of Cancer Therapy-Cognitive Function (FACT-Cog), and Cognitive Failures Questionnaire (CFQ). Study characteristics, main findings, and study quality appraisal are presented in Table 1.

Table 1.

Characteristics and main findings of included studies that examined association of fatigue with cognition (N=10).

Author, year, country	Objectives	Sample characteris- tics (sample size, % women, age (mean ±SD, years))	Study design	Fatigue measure- ment (timeframe, types of fatigue, instrument)	Cognitive function measurement	Other variables included in the analysis	Results	Quality level
Cancer (N=5)
Menning et al.,35 the Netherlands	To explore the effects of cancer-related psychological or biological processes in cognitive functioning and associated aspects of brain structure and function prior to adjuvant treatment	Women with BC scheduled to receive chemotherapy (Pre-ChT+, N=32); patients with BC who do not require chemotherapy (Pre-ChT−, N=33); no-cancer controls (NC, N=38), Age=Pre-ChT+: 50.2±9.2; Pre-ChT−: 52.4±7.3; NC: 50.1±8.7	Cross-sectional three-group design	Fatigue measured with EORTC-QLQ-C30 and POMS	Cognitive performance (e.g., executive function, attention, visual memory, verbal memory, processing speed, and motor speed) assessed with a neuropsychological test battery	Health-related QOL, anxiety and depression, mood, stress, cognitive problems, and personality dimensions (EORTC QLQ-C30, HSCL-25, PSS and POMS)	Worse cognitive performance, prefrontal activation and lower white matter integrity were observed in patients with BC compared to NCs, and fatigue was an important factor that contributes to these results. Worse cognitive performance in pre-ChT+ and pre-ChT− groups was found compared to HCs; however, when fatigue, stress, anxiety or depression were included in the model, no significant difference was observed. Processing speed was modestly, but significantly, associated with fatigue (r=−.292, p=.003).	Moderate
					MRI	Premorbid verbal IQ (Dutch Adult Reading Test) Long-term stress (cortisol levels using hair samples)
Li et al.,36 China	To examine extent to which chemotherapy and psychological factors (e.g. fatigue) were related to PCI	BCSs at least one week post-surgery, N=202 (women 100%), age=45.2±8.0	Cross-sectional, observational study	Psychological fatigue measured with FACIT-F (Chinese version)	PCI measured with PCI subscale of FACT-Cog (Chinese version)	PTSD (PCL-S) Depression and anxiety (HADS)	PCI was significantly associated with fatigue, PTSD symptoms, depression, anxiety, and chemotherapy or radiotherapy. Fatigue and PTSD symptoms (ΔR2=0.26, p<.001) independently contributed to PCI.	High
Visovatti et al.,38 USA	To evaluate cognitive function in individuals with CRC and examine factors associated with cognitive effects	Individuals with a diagnosis of CRC (N=50, 26 women and 24 men) and HCs (N=50, women 50%), age=CRC: 55±12; HC: 58±7	Cross-sectional, comparative design	Fatigue in the past week assessed with POMS-BF subscale	Attention, cognitive control, and memory function assessed with neuropsychological tests including Attention Network Test, Digit Span Test, Trail Making Test, Rey Auditory Verbal Learning Test, Attentional Function Index, and Everyday Memory Questionnaire	Anxiety and depression (POMS-BF subscales)	Individuals with CRC presented worse performance on tasks of attention and cognitive control (p<.005), but not long-term memory compared with HCs. Poorer performance on tasks of attention and cognitive control was significantly associated with having CRC, older age, and less education after controlling for covariates (p<.001). Patients with CRC reported greater fatigue and more anxiety than HCs (p<0.01).	High
Vardy et al.,4 Canada and Australia	To evaluate longitudinal changes in cognitive function and underlying mechanisms in people with CRC and HCs	Patients with localized CRC (N=289, 106 women and 183 men); patients with metastatic CRC (N=73, 33 women and 40 men); HCs (N=72, 41 women and 31 men)	Longitudinal (four time points for 24 months) design	Fatigue assessed with FACT-F	Cognitive function evaluated with clinical neuropsychological tests, the computerized Cambridge Neuropsycho-logical Test Automated Battery, and the modified Six Elements Test	Anxiety and depression (General Health Questionniare-12) – QOL (FACT-General)	Patients with localized CRC had more frequent cognitive impairment than HCs, with most affected in attention and working memory, verbal learning and memory, and complex processing speed. However, no association between overall cognitive function and fatigue, QOL, anxiety/depression, or any serum biomarkers was found.	Moderate
Henneghan et al.,27 USA	To explore the effects of psychological-factors (e.g. fatigue, perceived stress, loneliness and sleep quality) on PCF in BCSs after chemotherapy completion	BCSs for up to 10 years following the end of chemotherapy in a community setting; N=90 (women 100%); age=49±8.99	Descriptive cross-sectional design	Fatigue assessed with PROMIS Item Bank	PCF assessed with FACT-Cog	Blood tests (CBC, creatinine, liver function tests, carcinoembryonic antigen, sex hormones, selected cytokines, markers of blood clotting, and apolipoprotein genotyping)	Anxiety and depression (PROMIS)	More stress, social isolation, and experiencing worse sleep quality contributed to poorer PCF in BCSs and these effects were likely to be mediate by feelings of fatigue and anxiety.
						Moderate	Stress (PSS)	Importantly, fatigue was a significant mediator in all three of the mediation models (β=−1.66, p<.00001 in PSS model; β=−1.96, p<.00001 in UCLA-R model; β=−1.83, p<.0001 in PSQI model) Suggesting that the effects of stress, loneliness and sleep quality on PCF could share a common pathway, i.e. fatigue.
							Loneliness (UCLA-R)
Sleep quality (PSQI)
Multiple sclerosis (N=2)
Beier et al.,34 USA	To investigate the association between cognition and demographic and psychological variables in persons with a self-reported diagnosis of multiple sclerosis	Community-dwelling individuals from the Pacific Northwest with multiple sclerosis, N=407 (women 83%), age=52.95±10.67	Secondary analysis of a longitudinal study design	Physical, cognitive and psychosocial functioning of fatigue assessed with modified FIS	Cognition over the past 7 days assessed from NeuroQOL-GC to measure perceived cognitive dysfunction and NeuroQOL-EF to measure perceived dysfunction in higher order cognitive tasks such as planning, organizing, working memory, and mathematics	Depression (PHQ and DSM-IV) Stress (PSS) Anxiety (PROMIS anxiety questionnaire) Pain (PROMIS pain interference Sleep-Wake disturbance (PROMIS sleep disturbance items) Disability (Expanded Disability Status Scale)	Fatigue had strong, negative correlations with general cognitive concerns (r=−.63) and perceived executive functioning (r=−.59) although it is not significant. Fatigue was the strongest predictor of general concerns (β=−.51, p<.0001) and self-reported executive functioning (β=−.41, p<.0001).	Moderate
Ukueberuwa and Arnett,5 USA	To assess coping style as a moderator of the relationship between fatigue and cognition	Persons with multiple sclerosis, N=50 (38 women and 12 men), age=47±7.7	Longitudinal design (three years)	Physical, cognitive, and psychosocial functioning over the past month assessed with FIS	Executive function and processing speed changes assessed from five tests, such as Affective Reading Span Test, Paced Auditory Serial Addition Test (3s version total correct), Oral Symbol Digit Modalities Test total correct, Tower of Hanoi total time for blocks 1 and 2 combined, and Visual Elevator from the test of everyday attention time per correct switch	Coping style (Coping Orientation to Problems Experienced Inventory)	Avoidant coping moderated the relationship between fatigue and cognitive performance. Patients who experience relatively high fatigue performed better on cognitive tests if they used less avoidant coping. Patients who reported lower fatigue had relatively good cognitive performance regardless of their coping style.	High
Neurosarcoidosis (N=1)
Beste et al. (2015), Germany	To examine the impact of fatigue on conflict processing and response monitoring functions	Neurosarcoidosis patients (N=25, 16 women and 9 men) and case-matched HCs (N=25)	Cross-sectional, two-group design	Cognitive and motor aspects of fatigue assessed with Fatigue Scale for Motor and Cognitive Functions	Basic neuropsychological performance assessed with standard neuropsychological tests to assess general processing speed and working memory capacity	Depressive symptom (BDI-II)	Both cognitive and motor fatigue scores were higher in neurosarcoidosis patients than HCs, and cognitive and motor scores in neurosarcoidosis Patients were highly correlated (r=.766, p<.001). However, no difference for the neuropsychological tests was observed between two groups, and no effect of motor fatigue was evident in either group. Neurosarcoidosis was associated with cognitive deficits related to selection between different responses, but no changes in attentional processes were evident. Fatigue seemed to strongly modulate response selection processes in neurosarcoidosis patients, suggesting that the cognitive dimension of fatigue is relevant to the modulation of response selection processes. Neuroimmunological parameters like TNF-α and sIL-2R concentrations and depressive symptom did not mediate the effects observed.	High
EEG recording				Response selection processes assessed with a classical flanker task		Pro-inflammatory cytokines (e.g. TNF-α, sIL-2R)
Chronic fatigue syndrome (N=2)
Santamarina-Perez et al.,45 Spain	To explore neuropsycholo-gical impairment associated with chronic fatigue syndrome and determine their association with related clinical factors	Patients with chronic fatigue syndrome who were diagnosed by carrying out a physical examination and clinical laboratory and attended the Chronic Fatigue Unit of the University Hospital and studies (women 100%), N=68, age=46.7±8.34	Cross-sectional design	Mental, physical, and psychosocial dimensions of fatigue assessed with FIS	Cognitive function (i.e. attention, speed of information processing, verbal memory, visual memory, executive functioning, problem solving and motor functioning) assessed with a battery of neuropsychological tests	Anxiety and depression (HADS)	Fatigue was positively correlated with cognitive deficits in attention (r=.373, p<.01), verbal memory (r=.330, p<.01), and executive functioning (r=.397, p<.001). Fatigue was significantly associated with attention and executive functioning after controlling for all variables correlated with each domain. Especially, the attention deficits could be explained by the level of fatigue, which may be a consequence of alterations in neural pathways underlying cognitive dysfunction in chronic fatigue syndrome.	High
Attree et al.,47 UK	To examine the relationship among contributing factors to cognitive performance (e.g. fatigue, depression, and general self-efficacy)	Patients diagnosed with myalgic encephalomyelitis/chronic fatigue syndrome, N=87 (85 women and 2 men), age=58.8±10.32	Cross-sectional survey design	Physical fatigue experienced in the last week assessed from Chalder Fatigue Scale	Cognitive failures assessed from CFQ Prospective and retrospective emory assessed from prospective and retrospective memory questionnaire	Depression (CES-D) Anxiety (HADS) Social support (Medical Outcome Study social support survey) Self-efficacy (General self-efficacy scale)	Fatigue was significantly associated with cognitive failures (r=0.388, p<.001), prospective memory (r=.298, p=.005), and retrospective memory (r=0.354, p=.001). Depression and general self-efficacy were also associated with cognitive failures and retrospective memory. Fatigue and depression may be a consequence of the neurobiological effects of myalgic encephalomyelitis/chronic fatigue syndrome and the neurobiological effects of the illness may impact fatigue and cognitive deficits independently.	High

BC: breast cancer; BCS: breast cancer survivor; BDI: Beck Depression Inventory; CBC: complete blood count; CES-D: Center for Epidemiologic Studies Depression scale; CFQ: Cognitive Failures Questionnaire; CRC: colorectal cancer; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders Fourth Edition; EEG: electroencephalography; EORTC-QLQ-C30: European Organization for Research and Treatment of Cancer-Quality of Life Questionnaire-Core 30; FACIT-F: Functional Assessment of Chronic Illness Therapy-Fatigue; FACT-Cog: Functional Assessment of Cancer Therapy-Cognitive Function; FACT-F = Functional Assessment of Cancer Therapy-Fatigue; FIS: Fatigue Impact Scale; HADS: Hospital Anxiety and Depression Scale; HC: healthy control; HSCL-25: Hopkins Symptom Checklist-25; IQ: intelligence quotient; MRI: magnetic resonance imaging; NC: no-cancer controls; NeuroQOL-GC: Neuro-quality of life applied cognition-General Concerns-Short Form; NeuroQOL-EF: Neuro-quality of life applied cognition-Executive Function-Short Form; PCF: perceived cognitive function; PCI: perceived cognitive impairment; PCL-S: Posttraumatic stress disorder Checklist-Specific Stressor Version; PHQ: Patient Health Questionnaire; POMS: Profile of Mood States; POMS-BF: Profile of Mood States-Brief From; Pre-ChT: Pre-chemotherapy treatment; PROMIS: Patient-Reported Outcomes Measurement Information System; PSQI: Pittsburgh Sleep Quality Index; PSS: Perceived Stress Scale; PTSD: posttraumatic stress disorder; QOL: quality of life; sIL-2R: soluble interleukin-2 receptor; TNF-α: tumor necrosis factor-α; UCLA-R: University of California at Los Angeles loneliness scale-Revised.

Literature review by illness domain

Cancer

Breast cancer.

Because both neuropsychological and imaging studies point to the potential existence of pretreatment cognitive and brain dysfunction in patients with BC, Menning et al.³⁵ examined differences in symptoms and cognition/cognitive performance and their relationship to one another in a comparative sample of women with BC. The sample included women who were scheduled to undergo chemotherapy (N=32), women with BC who did not need chemotherapy (N=33), and women without cancer who were the controls (N=38). According to the reported findings, women with BC (with and without chemotherapy) demonstrated worse cognitive performance scores (p=.021) than the no-cancer control group. Cognition was measured with a neuropsychological battery of 18 test items. While fatigue levels were reported to be higher in cancer patients than in the no-cancer controls, group differences in cognitive function and various MRI measures “did not survive… stringent statistical thresholding” (p. 553). That is, when adjusting for fatigue levels, no statistically significant differences were found between groups in cognitive function. The authors concluded that they could not justifiably make the connection between fatigue and cognition based on their study findings. In another study, Li et al.³⁶ examined the extent to which chemotherapy and psychological factors of posttraumatic stress disorder (PTSD), fatigue, anxiety, and depression were related to perceived cognitive impairment (PCI) in Chinese women with BC (N=202) who were at least one week post-surgery and who would receive or not receive chemotherapy or radiation. PCI was reported to be significantly, but moderately, associated with “PTSD symptoms (r=0.54, p<.001) and higher fatigue (r=0.48, p<.001)” (p. 678). Applying a hierarchical regression model to their data, the authors noted that chemotherapy, but not radiotherapy, was significantly associated with PCI and symptoms of hyperarousal and fatigue and that both variables “contributed to 28.8%” of the PCI variance (p. 679). Cognition status was self-reported as measured by the FACT-Cog.³⁷ No discussion indicated that fatigue had any specific impact on cognition outside of the reported correlations. One study by Henneghan et al.²⁷ determined the effects of modifiable factors such as stress, loneliness, and sleep quality on perceived cognitive fatigue (PCF) in women with BC following chemotherapy completion. In this descriptive cross-sectional study, 90 BC survivors who were on average three years post chemotherapy completion were recruited from the community setting. The authors reported moderate to largely negative correlations between PCF, sleep quality, and psychological variables, including fatigue “(r=−0.31 to −0.70, p<.0009)” (p. 226); noting that the effects of three variables, i.e. stress, loneliness, and sleep quality in PCF, was mediated by fatigue. That is, worse-perceived cognitive functioning was associated with higher stress levels, loneliness, daytime sleepiness, and poorer sleep quality. Findings indicated that interventions targeting psychological factors, such as fatigue, stress, loneliness, and sleep quality, may enhance perceived cognitive functioning in BC survivors post chemotherapy treatment.

Colorectal cancer.

Visovatti et al.³⁸ conducted a cross-sectional study to examine cognition, anxiety, depression, and fatigue in persons with CRC (N=50) compared to healthy controls (N=50). Cognition was measured objectively with neuropsychological measures and subjectively with self-report measures such as Attentional Function Index³⁹ and Everyday Memory Questionnaire.⁴⁰ Fatigue, depression, and anxiety were measured with the Profile of Mood States-Brief Form.⁴¹ The authors reported that fatigue was the only variable that significantly predicted (p<.05) cognitive complaints related to attentional capacity and cognition; data were collected via self-report in both persons diagnosed with CRC and healthy controls. Alternatively, fatigue was not found to be a significant predictor of cognitive performance when it was objectively assessed using neuropsychological measures. The authors explained these findings by suggesting that subtle changes in fatigue may not influence neuropsychological test performance; however, they did not offer suggestions on potential mechanisms that may underlie cognitive changes self-reported by persons diagnosed with CRC. In another study, which consisted of a longitudinal quasi-experimental three-group trial, Vardy et al.⁴ examined cognition, fatigue, depression, and anxiety in persons with CRC (N=289) compared to healthy controls (HCs) (N=72) at baseline, 6, 12, and 24 months. Cognitive function was assessed objectively with neuropsychological testing and subjectively with a self-report measure (FACT-Cog)³⁷ with fatigue assessed using the FACT-fatigue (F) subscale. After controlling for chemotherapy treatment, cognitive impairment was higher in CRC patients than HCs at baseline “(43% vs. 15%, respectively, p<.001) and 12 months (46% vs. 13%, respectively, p<.001)”⁴ (p. 4085). The authors reported a “moderate association between self-reported cognitive symptoms and fatigue (r=.37–.55) and anxiety/depression (r=.36–.43)” (p. 4090) at each assessment; however, none was associated with objectively measured neuropsychological performance nor was there a reported predictive relationship.⁴ The authors noted associations between cognitive symptoms and fatigue but acknowledged that mechanisms of cognitive impairment remain unknown. In summary, two of the reported cancer-related studies noted a relationship between fatigue and cognition, with fatigue reported as contributing to, or being a predictor of, cognitive impairment.^36,38 In the other three studies reported here, Menning et al.³⁵ reported finding worse cognitive performance scores and higher fatigue levels in women with BC but no association between cognition and fatigue. Vardy et al.⁴ reported higher levels of cognitive impairment in colorectal cancer (CRC) patients but no association between overall cognitive function and fatigue. Henneghan et al.²⁷ found that the potentially modifiable variables of stress, loneliness, and sleep quality could indirectly predict PCF through feelings of fatigue. In the studies of cancer, two studies were assessed as high quality,^36,38 and three studies^4,27,35 were evaluated as moderate quality because objective, standard criteria for measurement of the conditions were not used or strategies to address incomplete follow up were not utilized in the longitudinal study.

Multiple sclerosis.

In a secondary analysis of data collected as part of an ongoing longitudinal survey tracking symptoms in individuals (N=470) with MS, Beier et al.³⁴ sought to determine potential predictor variables for cognitive impairment. Cognition, fatigue, pain, depression, and anxiety were among the wide array of symptoms measured over a period of two years. Self-reported cognition was measured by the Applied Cognition-General Concerns-Short Form (NeuroQOL-GC)⁴² while fatigue was assessed using FIS.⁴³ Using univariate and multivariate statistical modeling, the authors reported “fatigue (β=−0.52, p<.0001) and anxiety (β=−0.17, p=.003)” as statistically significant predictors of general cognitive concerns. “Fatigue (β=−0.41, p<.0001) and perceived stress (β=−0.12, p=.049)” were reported to be statistically significant predictors of self-reported executive functioning. It was also noted that of all the variables measured, fatigue had the “strongest association with both cognitive measures (e.g. general cognitive concerns and executive difficulties)” (p. 258).

Ukueberuwa and Arnett⁵ conceptualized fatigue as stress that some individuals with MS must cope with over time and hypothesized that the impact of stress from fatigue over time may be related to later cognitive problems. Then they examined the possible role of coping as a moderator in the relationship between fatigue and cognitive performance in persons (N=50) diagnosed with MS. Using a longitudinal survey design, participants completed a comprehensive neuropsychological test battery composed of five tests sensitive to executive function and processing speed changes in MS at baseline and twice more over an interval of three years. Active and avoidant coping was measured using the Coping Orientation to Problems Experienced⁴⁴ with fatigue measured using the FIS.⁴³ Fatigue as measured by the FIS significantly predicted the cognitive index at baseline “(ΔF (1.48)=4.19, p<0.05),” with a statistically significant interaction effect between fatigue and avoidant coping “(ΔF (1.45)=4.16, p<.05)” (p. 753). The authors suggested that high levels of fatigue combined with avoidant coping strategies (i.e. coping skills are inadequate) are predictive of later cognitive problems in persons with MS. In summary, findings in one of the MS studies suggest the predictive nature of fatigue in relation to cognition; however, it was not fatigue alone but also anxiety and perceived stress that predicted cognition.³⁴ In the other study reported here, fatigue was assumed a priori to be a predictor of diminished cognitive performance, and an individual’s coping style seemed to moderate the impact of fatigue on cognition. One study was evaluated as moderate quality,³⁴ and the other one was assessed as high quality.⁵

Immunological/multi-system disorders

Chronic fatigue syndrome.

Seeking to identify predictors of cognitive deficit in a sample of female CFS patients (N=68), Santamarina-Perez et al.⁴⁵ examined neuropsychological impairments associated with clinical factors of fatigue, anxiety, and depression. This was a cross-sectional study in which fatigue was measured using FIS;⁴³ anxiety and depression were measured by the Hospital Anxiety and Depression Scale (HADS).⁴⁶ Cognition was measured with a battery of objective neuropsychological tests. The authors reported high levels of anxiety, depression, and fatigue in three-quarters of the study participants with significant correlations reported between “attention and the FIS scores (r=.373, p<.001) and the HADS score (r=.309, p<.05), verbal memory (FIS: r=.330, p<.01; HADS: r=.407, p<.001), and executive functioning (FIS: r=.397, p<.001; HADS: r=.236, p=.056)”⁴⁵ (p. 124). There were no reported associations among age, education, and length of illness with cognitive deficit. Conducting linear stepwise forward regressions and controlling for all variables associated with each cognitive domain, deficits in attention and executive function were reported to be associated with fatigue while deficits in verbal memory were associated with depression. The authors concluded that an individual’s level of fatigue appeared to be one of the primary reasons for attention deficits in CFS and suggested that cognitive dysfunction may be attributed to alterations in neural pathways; however, they did not elaborate on what these pathways would be. Using a cross-sectional survey design, Attree et al.⁴⁷ examined the relationships between a set of variables, including fatigue, that purported to contribute to cognitive performance in a sample of individuals diagnosed with CFS (N=85 women; N=2 men). Self-reported prospective memory (r=.298, p=.005), retrospective memory (r=0.354, p=.001), measured using the Memory Questionnaire⁴⁸ and cognitive failures (r=.388, p<.001) using the Cognitive Failures Questionnaire⁴⁹ each moderately but significantly correlated with fatigue as measured by the Chalder Fatigue Scale.⁵⁰ Using a regression model to examine which variables would contribute independently and significantly to cognitive performance, fatigue was reported to be a predictor of cognitive failures and retrospective, but not prospective, memory. Noting that the study findings reported an association between fatigue and cognition, Attree et al.⁴⁷ acknowledged that the mechanism of action between and among these variables remains unknown. In summary, both studies in CFS patients listed fatigue as part of a larger phenotype. Adding to this that CFS is a disease process that includes fatigue as part of the diagnosis, in these studies, there was an a priori assumption that fatigue and cognition are intertwined. However, because CFS symptoms cluster with other psychophysiological or psychoneuroimmunological symptoms, it is difficult to discern whether a predictive relationship exists between fatigue and cognition. Thus, to date, it is not possible to determine if a potential directional relationship between fatigue and cognition in CFS exists, as both studies reviewed here have indicated. The methodological qualities of these studies were high.

Neurosarcoidosis.

Using a cross-sectional, two-group design, Beste et al.²⁶ investigated conflict monitoring and cognitive response selection processes in relation to cognitive fatigue and motor fatigue in a sample of adult neurosarcoidosis patients (N=25) compared to healthy controls (N=25). The authors tested the hypothesis that interindividual response selection and conflict monitoring would be modulated on the basis of interindividual degree of fatigue and that higher levels of fatigue would be related to stronger deficits in responding to conflict monitoring. Measuring cognition with a battery of neuropsychological tests and self-reported fatigue using the Fatigue Scale for Motor and Cognitive Functions (FSMC),⁵¹ the authors reported that neuroscarcoidosis patients, compared to healthy controls, had higher motor fatigue (p<.001) and cognitive fatigue (p<.001) scores on the FSMC with increases in cognitive fatigue having a stronger negative effect on response selection processes. The authors concluded that increases in cognitive, but not motor, fatigue lead to greater deficits in cognitive control processes in persons suffering from the neuroimmunological disease of neurosarcoidosis. Beste et al.²⁶ acknowledged that it is unknown how far fatigue modulates overall cognitive functions and suggested that fatigue in this patient population may arise as a consequence of immunological alterations. The methodological quality of this study was high.

Discussion

As reported, most studies examined non-directional associations among several covariates of fatigue and cognition, thus contributing to an inability to determine whether a relationship between fatigue and cognitive function existed in the chronically ill patient population. This is evidenced by the fact that several authors began their studies with the a priori assumption that this relationship existed and asked how the constructs of coping or self-efficacy⁴⁷ moderated the relationship between the variables of fatigue and cognition. In several studies, both depression and anxiety were included among the variables measured along with cognitive status. Of the eight studies in which depression and anxiety were among the variables measured as potential predictors of cognitive impairment in a chronically ill patient population, fatigue was often one of the significant predictors of cognitive concerns. For example, Beier et al.³⁴ found that only fatigue (β=−0.52, p<.0001) and anxiety (β=−0.17, p=.003) were statistically significant predictors of general cognitive concerns. In another example, Li et al.³⁶ found that PCI was significantly associated with fatigue, PTSD symptoms, depression, anxiety, and chemotherapy/radiotherapy, but it was only the combination of fatigue and PTSD symptoms (ΔR²=0.26, p<.001) that independently contributed to PCI. In other studies that examined multiple covariates of fatigue, the opposite was true. That is, authors reported finding no relationship between cognitive dysfunction and depression, anxiety, or other covariates.^4,26,27 In a systematic review to examine the relationship among fatigue in BC survivors, quality of life, and potentially modifiable factors for psychological interventions, Abrahams et al.⁵² reported study findings that supported our own. Indeed, they noted a myriad of biopsychosocial factors that they attributed to fatigue. These included depression, anxiety, and sleep disturbances, and dysfunctional cognition defined as catastrophizing about symptoms; however, they did not report cognitive impairment as a symptom that emerged from their systematic review of the literature. In summary, this review indicates that, to date, efforts to determine if, in chronic NCDs, fatigue is a major variable that contributes to cognitive impairment, or whether cognitive impairment contributes to fatigue has been challenged not only by the presence of multiple co-factors, for example depression and anxiety, but also by variations in study designs and data collection instruments. Nonetheless, there is some evidence to support the a priori assumption that some authors have made—that there is a relationship between fatigue and cognition; however, precisely what that relationship might remains to be determined.

Another concern that contributes to our conclusion that further, more refined or targeted research is warranted in the field of fatigue and cognition is the wide variation in how fatigue and cognition were measured among the reported studies (see Table 1). Among the 10 studies reviewed, there were six fatigue measures used. These included the PROMIS Fatigue Short-Form 8a,⁵³ the Chinese version of the (FACIT-F),⁵⁴ the subscale of the European Organization for Research and Treatment of Cancer Quality of Life-C30 (EORTC-QOL C30),⁵⁵ the FIS,⁴³ the FSMC,⁵¹ and the Chalder Fatigue Scale.⁵⁰ For cognition, four studies used objective neuropsychological testing only, five studies used subjective, self-report measures only, and one study combined objective neuropsychological testing with subjective cognitive measures. Subjective cognitive measures, along with the fatigue measures, also varied among the studies. These included the FACT-Cog,³⁷ FACIT-F Version 4 Chinese Version,⁵⁴ Attentional Function Index,³⁹ Everyday memory Questionnaire,⁴⁰ Applied Cognition-General Concerns-Short Form,⁴² Memory Questionnaire,⁴⁸ and Cognitive Failures Questionnaire.⁴⁹ Notably, objective measures of cognition were frequently reported as not having a significant relationship to self-reported fatigue scores, which suggests that the use of such measures, which are both expensive as well as time consuming, may have possibly confounded fatigue and cognition findings in these studies. Other issues that may have influenced study outcomes involve the terms, feeling fatigued/feeling cognitively down. That is, when considering that the correlations between fatigue and self-reported cognitive dysfunction were more prevalent than with objectively measured cognitive dysfunction, the close overlap in what these terms mean may have also confounded study findings. When designing future research studies and directions, this “cognitive fatigue” phenomenon should be taken into account. Overall, study findings were mixed and inconsistent, thus contributing to the challenge of capturing information as it relates to fatigue and cognition in the chronic NCD patient population.

Limitations

Study designs reported in the literature from 2012 to 2018 were predominantly cross-sectional, which suggests that efforts are commencing to determine if a priori assumptions can actually be made about a specific relationship between fatigue and cognition. Drawing concise conclusions regarding a relationship between the variables of fatigue and cognition was not possible due to the wide variations of confounding variables associated with chronic illness conditions such as depression, stress, and anxiety. While a few studies did seek to ask, through multivariate modeling, whether there might be a directional relationship between fatigue and cognition, and asked whether fatigue impacted cognition and not vice versa, results of this literature review suggest that there is minimal literature reporting studies in which researchers addressed potential underlying mechanisms that may confirm directional linkages biologically. For example, Matura et al.⁵⁶ addressed the question of biological mechanisms of fatigue in chronic illness through their systematic review of the literature. They reported not just cancer-related fatigue, but also fatigue as it is found in chronic NCDs such as heart failure, MS, and rheumatoid arthritis. Matura et al.⁵⁶ reported inflammation, hypothalamic-pituitary axis (HPA), and autonomic nervous system (ANS) dysfunction as integral to the mechanisms of fatigue. From a clinical perspective, understanding the relationships between biological and non-biological linkages may enable interventionist researchers to improve the care of the chronically ill.

Conclusions and implications for future research

The putative relationship between fatigue and cognition has been established in the literature over many chronic NCDs such as fibromyalgia, multiple sclerosis, and cancer. Other research has demonstrated that such illnesses may share an underlying inflammatory process which may play a role in the genesis, severity, or persistence of fatigue and cognition. Thus, this review examined fatigue and cognition across chronic illness. As these two symptoms may be inextricably linked, we endeavored to conduct this systematic review to elucidate the ways in which other researchers examined and measured these two phenomena to provide direction for future research. Further research is necessary to determine the interconnectedness of these two most distressing symptoms for patients living with chronic NCDs so as to design and test interventions that could assuage the effects of fatigue and cognitive dysfunction. It remains to be known if mitigation of one of these connected symptoms has influence on the mitigation of the other. Knowing this would indeed provide evidence to direct future-targeted interventions that may greatly improve the quality of life for patients suffering with fatigue and cognitive dysfunction while living with a chronic NCD.