Predictors and treatment strategies of
HIV-related fatigue in the combined antiretroviral therapy era
"HIV-related fatigue has a high prevalence and is strongly associated with psychological factors such as depression and anxiety......Most studies were carried out with HIV-infected patients with a major depressive disorder or clinical hypogonadism......Treatment of HIV-related fatigue is important because of its social, psychological and behavioural consequences and requires a multidisciplinary approach.......Psychoeducation and cognitive-behavioural therapy both proved to be effective in HIV-related fatigue.....the majority of studies were conducted prior to the widespread use of cART, which limits the ability to generalize the data. Aerobic exercise training reduced depressive symptoms and decreased fatigue, weight, BMI, subcutaneous fat and abdominal fat in HIV-infected individuals"|
19 June 2010
Jong, Eefje; Oudhoff, Lisanne A; Epskamp, Cynthia; Wagener, Marlies N; van Duijn, Miranda; Fischer, Steven; van Gorp, Eric CM
aDepartment of Internal Medicine, Slotervaart Hospital, Amsterdam, The Netherlands
bDepartment of Infectious Diseases, University Medical Center, Utrecht, The Netherlands
cUniversity of Amsterdam, Amsterdam, The Netherlands
dCenter of Excellence Participation Occupation and Health, Rotterdam University of Applied Science, Hogeschool Rotterdam, Rotterdam, The Netherlands
eDepartment of Integrative Medicine, The Netherlands
fDepartment of Rehabilitation Medicine, The Netherlands
gDepartment of Psychology, Slotervaart Hospital, Amsterdam, The Netherlands
hDepartment of Virology, Erasmus Medical Center, Rotterdam, The Netherlands.
Correspondence to Eefje Jong, Department of Internal Medicine, Slotervaart Hospital, Louwesweg 6, 1066 EC Amsterdam, The Netherlands. E-mail: email@example.com
Objective: To assess predictors and reported treatment strategies of HIV-related fatigue in the combined antiretroviral (cART) era.
Five databases were searched and reference lists of pertinent articles were checked. Studies published since 1996 on predictors or therapy of HIV-related fatigue measured by a validated instrument were selected.
Results: A total of 42 studies met the inclusion criteria. The reported HIV-related fatigue prevalence in the selected studies varied from 33 to 88%. The strongest predictors for sociodemographic variables were unemployment and inadequate income. Concerning HIV-associated factors, the use of cART was the strongest predictor. Comorbidity and sleeping difficulties were important factors when assessing physiological influences. Laboratory parameters were not predictive of fatigue. The strongest and most uniform associations were observed between fatigue and psychological factors such as depression and anxiety. Reported therapeutic interventions for HIV-related fatigue include testosterone, psycho-stimulants (dextroamphetamine, methylphenidate hydrochloride, pemoline, modafinil), dehydroepiandrosterone, fluoxetine and cognitive behavioural or relaxation therapy.
HIV-related fatigue has a high prevalence and is strongly associated with psychological factors such as depression and anxiety. A validated instrument should be used to measure intensity and consequences of fatigue in HIV-infected individuals. In the case of fatigue, clinicians should not only search for physical mechanisms, but should question depression and anxiety in detail. There is a need for intervention studies comparing the effect of medication (antidepressants, anxiolytics) and behavioural interventions (cognitive-behavioural therapy, relaxation therapy, graded exercise therapy) to direct the best treatment strategy. Treatment of HIV-related fatigue is important in the care for HIV-infected patients and requires a multidisciplinary approach.
The introduction of combined antiretroviral therapy (cART) has resulted in an increase in the overall survival rate of HIV-infected patients. The prevalence of HIV-related and non-HIV-related diseases and symptoms has increased [1,2]. One of the most prevalent and troubling symptoms of HIV is fatigue . The prevalence of fatigue ranges from approximately 20 to 60% in patients with chronic HIV infection [3-5] up to 85% in patients with AIDS [6,7]. Fatigue is defined as a lessened capacity for work and reduced efficiency of accomplishment, usually accompanied by a feeling of tiredness that is not relieved by a good night's sleep .
Fatigue has important social, psychological and behavioural consequences. People with an increased fatigue level have less energy and inclination to work and for social activities, reduced motivation, difficulty concentrating and increased drowsiness. Fatigue may eventually result in unemployment, social isolation, reduced self-care activities and a lessened response to treatment and medical care. It is also associated with an impaired quality of life and is a precursor to lower survival [5,9,10]. The importance of fatigue, and its impact on a patient's life, has been increasingly recognized and studied for a number of chronic diseases, including multiple sclerosis, systemic lupus erythematodes (SLE), chronic viral and cholestatic liver disease, rheumatoid arthritis and HIV .
The cause of fatigue in HIV-infected patients is probably multifactorial. According to current literature, fatigue can be associated with several disease-related factors such as the stage of the disease, the use of cART and certain laboratory parameters, as well as with sociodemographic and psychological factors. Most studies focused on a selection of physiological and/or psychological factors.
HIV-related fatigue and its clinical, social and psychological impact are challenging for the caregiver. To improve the care for HIV-infected patients, a thorough understanding of the complex interplay between sociodemographic, physiological and psychological factors in HIV-related fatigue and available treatment options for HIV-related fatigue is important. We conducted the following narrative review to summarize the latest evidence regarding predictors and treatment strategies for HIV-related fatigue in the cART era.
In this review, we assessed the importance of sociodemographic, HIV-associated, physiological and psychological factors in the aetiology of HIV-related fatigue. Furthermore, we reviewed the literature on treatment strategies for HIV-related fatigue. The reported HIV-related fatigue prevalence in the selected studies varied from 33 to 88%. The majority of prevalence rates refer to study populations from developed parts of the world and include treated and untreated HIV-infected individuals. Only one study was performed in a developing country .
The strongest predictors for sociodemographic variables were unemployment and inadequate income. Concerning HIV-associated factors, the use of cART was the strongest predictor. It was noteworthy that both CD4 cell count and HIV viral load were not associated with fatigue in the majority of studies. Comorbidity and sleeping difficulties are important factors when assessing physiological influences. Laboratory parameters were not predictive of fatigue in most studies. The most uniform and strongest associations were observed between fatigue and psychological factors like depression and anxiety.
Only studies using a validated instrument to assess HIV-related fatigue were selected. Most of the selected studies on predictors of fatigue had a cross-sectional design and only few had a longitudinal, observational design. Only one study included a HIV-negative control group . Due to the cross-sectional design, the majority of studies are unable to determine the direction of causality between fatigue and predictors. Furthermore, different predictors may be strongly dependent on each other. Although only studies with validated instruments to assess HIV-related fatigue were selected, the heterogeneity in instruments used was substantial and thus studies are difficult to compare. A number of instruments used were not validated for use in HIV-infected patients. Patients were selected mostly by self-referral. The proportion of fatigued participants may be an overestimate of the prevalence of fatigue among HIV-infected individuals in general. Furthermore, the definition of fatigue is broad and encompasses muscle aches, weakness, painful joints and fatigue.
Studies on the treatment of HIV-related fatigue are minor in nature and focus on a selected group of patients. Most studies were carried out with HIV-infected patients with a major depressive disorder or clinical hypogonadism. Remarkably, only one study focused on the effect of fluoxetine, even though depression seems an important predictor of fatigue in HIV-infected patients . Psychoeducation and cognitive-behavioural therapy both proved to be effective in HIV-related fatigue, but were not compared to a control group without an intervention. In chronic fatigue syndrome and fibromyalgia. improvement was observed with graded exercise therapy (GET) [14,86]. Where used, GET should be implemented in a controlled manner with professional supervision [17,86]. In a review article by Ciccolo et al. , exercise has been consistently listed as one of the most popular self-care therapies. A small number of studies have been conducted to examine the impact of exercise on the most common self-reported symptoms of HIV and AIDS and the adverse effects of treatment. The majority of studies were conducted prior to the widespread use of cART, which limits the ability to generalize the data. Aerobic exercise training reduced depressive symptoms and decreased fatigue, weight, BMI, subcutaneous fat and abdominal fat in HIV-infected individuals [88,89]. These studies were excluded from this review because they lacked a validated instrument to measure fatigue. No negative effect of exercise on psychological as well as physiological factors, like immune status, has been reported.
Fatigue remains a challenging construct to measure definitively, with no generally accepted gold standard definition. Cultural differences might exist in symptom representation. The HRFS is a detailed instrument designed specifically to measure intensity and consequences of fatigue in HIV-infected individuals . The HRFS has strong internal consistency and construct validity . Especially individuals with HIV/AIDS complaining of any comorbidity or other symptoms should be evaluated for fatigue. In case of fatigue, clinicians should not search only for physical mechanisms, but should question depression and anxiety in detail. There is a need for intervention studies comparing the effect of medication (antidepressants, anxiolytics) and/or behavioural interventions to direct the best treatment strategy. Behavioural interventions (cognitive-behavioural therapy, GET) have proven efficacy in chronic fatigue syndrome and are promising therapeutic interventions for HIV-related fatigue. Currently, the evidence for interventions with medication is not strong. Behavioural interventions and GET seem more promising. Treatment of HIV-related fatigue is important because of its social, psychological and behavioural consequences and requires a multidisciplinary approach. There is a need for an appropriate evidence-based practice guideline for the management of HIV-related fatigue.
Search methods for identification of studies
For the current review, published studies were extracted from five databases (PubMed, Cochrane Library, EMBASE, PsychINFO and CINAHL). Keywords used in the initial search included HIV, AIDS, fatigue, exhaustion, tiredness, weariness, antiretroviral, treatment and therapy. The search criteria were limited to age 18 and above, human and English speaking. The search was repeated without limits so the most recent articles were also included. Only original research articles were selected. All selected articles used a validated instrument to objectively measure fatigue. Additional studies were found by examining references from relevant articles. The extraction of articles was from January 1996 to August 2008 to include only studies published in the highly effective cART era. The literature search and final selection of articles is shown in Fig. 1. The selected articles on predictors and therapeutic interventions for HIV-related fatigue are summarized in Tables 1 and 2.
Pathophysiology of fatigue in chronic diseases
Parallels have been drawn between the chronic fatigue syndrome and fibromyalgia [12-14]. There is long-standing controversy surrounding the exact cause of chronic fatigue syndrome and fibromyalgia. Dysfunction of the hypothalamic-pituitary-adrenal (HPA) axis is suggested . Frequent symptoms in both chronic fatigue syndrome and fibromyalgia are pain, fatigue, sleep disorders, irritable bowel syndrome, chronic headaches, cognitive or memory impairment, dizziness and impaired coordination. The presentation is often accompanied by issues of mental health concern, in particular anxiety and depression. In chronic fatigue syndrome, there is evidence to suggest that a major significant life event or an acute illness may precede the onset of this syndrome [15-17]. In fibromyalgia, studies suggested that patients with this syndrome are more susceptible and hypersensitive to pain [12,18].
Fatigue in patients with a chronic disease is divided into central and peripheral fatigue. Central fatigue results from alterations or abnormalities in neurotransmitter pathways within the central nervous system (CNS). Peripheral fatigue results from neuromuscular dysfunction outside the CNS and relates to impaired neurotransmission in peripheral nerves and/or defects in muscular contraction, due to energy depletion, inflammation, joint abnormalities or muscle wasting (Fig. 2) [11,19]. In general, central fatigue appears to be the most relevant in patients with chronic diseases, although the absolute contribution of peripheral and central fatigue to overall fatigue in patients may vary significantly between different diseases.s
Patients with chronic diseases are known to suffer from a high degree of stress, involving variable inputs from social (loss of social position and social support), psychological (disease labelling, depression, anxiety, coping patterns) and physical (disease activity, pain and inflammation) issues. The principal CNS components of the stress response include central corticotrophin-releasing hormone (CRH) and the sympathetic nervous system. CRH is the main central regulator of the activation of the HPA, stimulating adrenocorticotropic hormone (ACTH) from the anterior pituitary. CRH that is infused centrally causes behavioural activation and signs of a deregulated HPA, such as altered diurnal cortisol rhythm and blunted cortisol stress response causing fatigue [11,19]. CRH has been shown to modulate behavioural changes during stressful conditions . However, most studies on CRH have been done in rodents. Sullivan et al.  studied in vivo in baboon two possible CRH1 antagonist ligands that are candidates for PET imaging to identify regions of CRH hypersecretion. A pattern of distribution consistent with blood-brain barrier penetration and early accumulation in the brain was observed without evidence of detectable specific binding of any radiolabelled metabolite of these radiotracers. Significant levels of the CRH1 receptor were shown in cerebellum and multiple cortical regions in primates. Further studies in humans are required.
During the inflammatory response, the acute-phase response is coordinated through the release of a number of cytokines [interleukin (IL)-1, IL-6, tumour necrosis factor (TNF) and interferon gamma (IFN-γ)]. These cytokines have been implicated in the genesis of nonspecific symptoms of illness, including fatigue . Cytokines seem to play a direct role in the secretory activity of the HPA axis. Increases in ACTH and glucocorticoid plasma levels as well as hypothalamic CRH release were observed in laboratory animals after injection of cytokines (e.g. IL-1α, IL-1ß, IL-2, IL-6, IFN-α and TNF-α). In human studies, elevated plasma ACTH and cortisol concentrations were observed after intravenous or subcutaneous administration of different cytokines. However, cytokines also produce a number of systemic acute-phase responses that themselves could elicit HPA activation, raising the question whether the effects of cytokines on the HPA axis are direct or secondary to stress produced by other acute-phase responses. Cytokine receptors are present in the CNS and their expression is increased during a number of CNS disorders such as multiple sclerosis, Down's syndrome and Alzheimer's disease. The hypothalamus and pituitary express cytokine receptors . IL-1ß activate the HPA axis and induce decreased activity and increased sleep when injected [22,24,25]. The infusion of IL-6 into humans results in fever, anorexia and fatigue . TNF upregulates the 5-hydroxytryptamine transporter and thereby alters serotonin metabolism in the brain. IFN is a pro-inflammatory cytokine that has also been used therapeutically to treat a number of diseases, and IFN therapy is associated with a number of adverse effects, including fatigue . Moreover, high levels of cytokines decrease the amount of erythropoietin secretion, which may cause anaemia [11,19].
Both the serotonergic and noradrenaline neurotransmitter systems are associated with fatigue and both systems partly control central CRH release. Specifically, studies in patients with chronic fatigue syndrome suggest increased sensitivity to serotonin-mediated hypothalamic activation, implying the existence of defective central serotonergic neurotransmission in these patients . There is also pharmacological evidence for an interaction of CRH and noradrenaline in mediating behavioural response to stress. Defective central noradrenaline release could contribute to the origin of fatigue, possibly by the inhibition of CRH . Increased levels of another central neurotransmitter, substance P, possibly causes fatigue by inhibiting the central release of CRH as well .
Fatigue in chronic diseases correlated strongly with abnormalities in mood, most typically depression and anxiety. The close association between fatigue and mood disorder has been recognized for years in patients with chronic fatigue syndrome. This may be due, in part, to overlapping symptoms, including fatigue, sleep disturbance, psychomotor change, cognitive impairment and mood changes [30-34]. In animal models of other chronic diseases like SLE and cholestatic liver disease, the development of changes in central CRH levels parallels the development of animal correlates of depression (i.e. anhedonia or the loss of interest in pleasurable activities) and anxiety. These findings suggest a link between mood disorder and fatigue in chronic disease, based on defects in the common central pathways in these disorders . Several studies have demonstrated that fatigue is significantly associated with depression, disability status and physical limitations in HIV infection [35-37]. Comorbidities in terms of endocrine and metabolic disturbances, malignancies and coinfections occur in the course of HIV infection with or without use of cART and might contribute to HIV-related fatigue. The pathogenesis of enodocrinopathies includes direct infection of the endocrine glands by HIV or opportunistic infections, infiltration by neoplasms and side-effects of drugs . Adrenal insufficiency is the commonest HIV endocrinopathy, with cytomegalovirus adrenalitis occurring in 40-88% of the cases [39,40]. Cushing's syndrome has also been reported . The adrenal glands show evidence of both inflammation and necrosis at autopsy . Thyroid dysfunction may occur as euthyroid sick syndrome or subclinical hypothyroidism. In patients with the sick euthyroid syndrome, the T3 level is reduced, the reverse T3 level elevated and the thyroid-stimulating hormone (TSH) level is relatively normal or decreased, depending on the severity of illness. The major cause of these hormonal changes is the release of IL-6 and TNF . Among HIV-infected patients with subclinical hypothyroidism, antithyroid peroxidase antibodies are rarely identified, suggesting a cause other than autoimmune . Hypogonadotropic dysfunction accounts for 75% of HIV-associated hypogonadism, with prolonged amenorrhoea being three times more likely in the women. Low testosterone with elevated luteinizing hormone and follicular-stimulating hormone levels have also been reported, suggesting primary testicular failure . Although the incidence of AIDS-defining malignancies such as Kaposi's sarcoma and cerebral lymphoma has sharply declined since the introduction of cART, large-scale observational cohort studies have reported a two-fold to three-fold higher risk of developing selected non-AIDS malignancies than in the general population. These malignancies include Hodgkin's disease, lung, head and neck cancers, liver and anal cancers [45-47]. Liver disease from chronic hepatitis B and C infection was recognized as the most important cause of non-AIDS-related death in a large cohort of HIV-infected individuals in Europe, particularly in intravenous drug users . Furthermore, metabolic disorders (dislipidaemia, body fat abnormalities, insulin resistance), cardiovascular disease and kidney disease are important cofactors contributing to HIV-related fatigue in the cART era .
Predictors of HIV-related fatigue
The majority of studies showed no significant differences in the level of HIV-related fatigue between sexes [5,7,9,10,49-51]. In one study in a representative sample of 372 HIV-infected patients including 31.7% women, women reported significantly higher fatigue intensity [6,52]. Women were in general underrepresented. In most studies, no association between age and fatigue was observed [7,10,50,51,53-55]. In one cross-sectional study in 142 HIV-infected men and women from the early cART period, a multiple regression analysis showed that being older was positively associated with higher levels of energy/fatigue as assessed by the Medical Outcome Study (MOS)-HIV . Older patients may have had higher levels of energy because they developed effective coping strategies, or may have had more experience with and more time to adjust to their medication regimes. Ethnicity could be another factor in fatigue prevalence and experience. Persons of African-American ethnicity reported lower fatigue intensity compared with whites and Latinos. This secondary data analysis focused on the intensity of fatigue by selected sociodemographic, cultural and health/illness variables. Fatigue was measured as a subscale on the SSC-HIVrev, a 74-item checklist that assesses presence and intensity of signs and symptoms currently experienced by people with HIV and AIDS . In a longitudinal study by Harmon et al , African-Americans reported higher fatigue intensity compared with whites, but this difference was not significant. The majority of participants in this study were African-American. In a study on HIV-infected women, white women reported higher evening fatigue compared with African-American and Latino women . However, several other studies showed no differences in fatigue intensity between races [4,7,50,51]. According to a study in southern Africa, there might be a positive correlation between having children and fatigue . However, no other studies executed in other countries found a significant difference in fatigue related to having children or any other kind of home-environment factors, like having a partner and the number of adults in the home [6,49,50]. A negative association was found between fatigue and the number of years of education a person had received and whether he or she had a college degree, although this association was not statistically significant [6,9,52,57]. Higher fatigue is experienced by HIV-infected patients who are unemployed or with an inadequate income, compared with employed HIV-infected patients with an adequate income [6,10,52,57]. Lee et al.  observed higher evening fatigue scores in women who were employed, but there were no differences in morning fatigue scores according to employment status. In this study, income was not associated with either type of fatigue. The study of Vosvick et al.  showed no significant association between fatigue and household income. Participation in volunteer work was not related to fatigue . Higher fatigue scores were reported in individuals with inadequate health-insurance coverage. However, these differences were not significant [6,52]. Several studies looked at the impact of sexual orientation on fatigue but no correlation was found [6,7,10].
No association was observed between symptomatic HIV-infected patients (CDC category B and C) and fatigue scores [6,9,10]. In a randomized controlled trial comparing ritonavir/saquinavir to ritonavir/saquinavir/stavudine, a better quality of life and less fatigue was reported in asymptomatic patients (CDC category A) . More HIV-related symptoms defined by higher scores on the HIV-related Symptom Scale were correlated with higher fatigue scores [5,50]. Indicators of the severity of HIV-related illness, for example HIV-related symptoms, hospitalizations and scores on the Illness Perception Questionnaire, showed the same results [6,7,10,52,54]. In a hierarchical regression model, fever, gastrointestinal symptoms and numbness significantly predicted a higher fatigue severity . Fatigue was associated with longer duration of HIV infection [4,55]. These two studies report data derived from the same pilot study in 40 patients. In a longitudinal study describing 128 HIV-infected patients, it was concluded that fatigue might be noted more often in patients with a recent diagnosis of HIV infection. The median duration since HIV diagnosis in this study was 10 years (range 6-15 years). The observed association between longer duration of infection and less reported fatigue might reflect different perceptions of fatigue among those who have been dealing with HIV for a longer time [10,57]. Yet, other studies showed no correlation between the duration of HIV infection and fatigue [6,58]. Concerning CD4 cell count and HIV viral load, most studies showed no relationship between CD4 cell count, HIV viral load and the presence of fatigue [4-7,9,54,55,57-60]. In a study in 79 patients who were all taking cART, a correlation between fatigue and a higher HIV viral load, but not CD4 cell count, was described. This study focused on correlates of sleep in HIV infection. No absolute CD4 cell counts or HIV viral loads were mentioned. The revised Piper Fatigue Scale (RPFS) used to measure fatigue in this study was originally validated in cancer patients . Another study observed greater fatigue in HIV-infected women with lower CD4 cell counts. In this study from the early cART period, 35% of women had a CD4 cell count below 200 cells/µl. No data on HIV viral load or cART use were reported . On the contrary, Henderson et al.  described greater fatigue in individuals with higher CD4 cell counts. In the multivariate analysis, higher CD4 cell counts were significant predictors of higher fatigue scores. However, CD4 cell counts between fatigue cases and noncases were not significantly different. In this cross-sectional study, women were underrepresented.
Levels of pro-inflammatory markers (TNF-α, IL-6 and soluble urokinase receptor) were not associated with fatigue in a cross-sectional study in 95 patients using cART with a mean of 12.7 (±5.1) years since HIV diagnosis . In this study, TNF-α levels were actually lower in HIV-infected patients than in healthy controls and were lowest in HIV-infected patients with severe fatigue. IL-6 levels were elevated in HIV-infected patients compared with healthy controls. These findings need confirmation in future studies. The low TNF-α levels contradict its suggested role in the pathophysiology of HIV-related fatigue. Fatigue was reported by 69% of patients as a side-effect of cART [51,61-63]. Most patients (47%) attributed the fatigue they experience to both HIV disease and the use of cART, whereas 7% attributed it mainly to cART . However, in many other studies, no association between cART and fatigue was described [4,5,7,10,53,64,65]. No difference in fatigue intensity was observed between cART regimens, including protease inhibitors or (non)nucleoside reverse transcriptase inhibitors [5,51,56,66].
Corless et al.  studied fatigue in 1203 HIV-infected patients with comorbidities from sites in Colombia, Norway, Puerto Rico, Taiwan and the United States. Higher fatigue measured on the Revised Signs and Symptoms Checklist for Persons with HIV Disease (SSC-HIVrev) was associated with hypertension, hepatitis B and C coinfection. The reported prevalence for hypertension, hepatitis B and C coinfection was 12, 7 and 9%, respectively. Depression and hepatitis B had the most profound impact on fatigue severity. No correlation between fatigue and diabetes mellitus was found. The number of comorbidities was associated with higher fatigue. In the final regression model on predictors of fatigue, the patients were distributed in a group with comorbidity and a group without comorbidity. Harmon et al.  confirmed the association of higher fatigue scores in HIV-infected patients with any other chronic illness (e.g. hypertension, arthritis, depression). In another study, no significant differences were found in fatigue related to the presence or absence of comorbidities . According to a study by Barroso et al. , hepatitis C coinfection was not correlated with fatigue. No hepatitis C prevalence for this study population was reported.
Laboratory parameters such as haemoglobin level, red blood cell count, haematocrit and serum erythropoietin were not significantly associated with fatigue [5,9,54,59]. Cortisol was suggested as a possible biomarker for HIV-related fatigue. In a pilot study, Barroso et al.  showed that the level of cortisol in the saliva of HIV-infected patients was higher than that in a control group of HIV-negative individuals. The group of HIV-infected patients with an upward trend in the cortisol level had the highest fatigue severity index measured by the HIV-related fatigue scale (HRFS). This finding is suggestive of an association between stress, the release of cortisol and fatigue, without providing information on central regulation. In this study, samples were self-collected at home. The results give an impression of the circadian rhythm of cortisol production. To really assess the HPA axis, functional testing under general conditions is required. A negative correlation between testosterone levels and fatigue was described . TSH was negatively correlated with fatigue , but no association was found in another study at all between fatigue and TSH or thyroxine levels . Both studies by Barroso et al.  used the well validated HRFS to measure fatigue and tested a number of physiological variables as predictors of fatigue. One study applied bivariate correlation and multiple regression with psychological and physiological variables in separate models. By using correlation analysis, the direction of causality could not be established. The other study used regression models only. None of the physiological variables remained significant after controlling for demographic and clinical factors . In case of the euthyroid sick syndrome, TSH levels could be both normal or decreased . This might explain the negative association in one study and the lack of an association in the other. The association between fatigue and hepatic enzymes (aspertate aminotranspertase, alanine aminotranspertase, gamma glutamyl transpeptidase, alkaline phosphatase, total bilirubin) was studied, but none of the parameters showed a significant association with fatigue . Another study supported these results and did not find any association between alkaline phosphatase and fatigue . Albumin and fatigue were not found to be associated . Barroso et al.  found a strong positive correlation between the concentration of platelets in the blood and fatigue. However, confounding factors such as iron deficiency, cancer and other conditions have to be taken into account and were not corrected for in this study. Globulin was also found to be associated with fatigue . Glucose levels and fatigue showed no association .
Parameters of body composition and cellular integrity are measurements of nutritional status and wasting. Body composition can be measured by body cell mass, fat-free mass, weight and body mass index. None of these variables had a significant association with fatigue as measured on the HRFS . Neither cellular integrity, measured by phase angle, nor cellular injury, measured by lactic acid, was associated with fatigue [9,55]. A study of Voss et al.  did find a significant correlation between fatigue and body change, but in a hierarchical regression model, body change was not a significant predictor of fatigue. In a comparison between AIDS and non-AIDS groups, objective neurocognitive complaints measured by a set of tests on verbal learning, psychomotor speed and motor/speed dexterity did not correlate with fatigue. The AIDS group demonstrated poorer performance on verbal learning and motor/speed dexterity. Subjective neurocognitive complaints did show a significant correlation with fatigue . One study used PET scanning to assess whether the fatigue experience in HIV was associated with a functional cerebral correlate, but no such association was observed . A PET scan was done in a subgroup of nine patients with severe fatigue and seven patients with no fatigue.
Sleeping difficulties are reported often in HIV-infected individuals, compared with the general population, and this may cause fatigue. Fatigue and the quality of sleep and time it takes to fall asleep were strongly correlated [5,50,59,69]. The total sleep time was not associated with fatigue [5,59,60,69], but daytime napping and sleepiness were [5,50]. Evening fatigue was associated with the number of awakenings, whereas morning fatigue was associated with daytime napping and subjective sleep disturbances during the past week. The difference between evening and morning fatigue was correlated with sleep efficiency . Significant daytime dysfunctions existed as a result of fatigue . Pain was associated with sleep disturbances and with fatigue in several studies [10,50,56].
The scores of physical functioning were strongly correlated with fatigue [7,54]. Disability was correlated with fatigue in the study of Voss et al. , but not in the study of Barroso et al. . Self-care was not found to be correlated with fatigue .
Psychological distress, as measured by the General Health Questionnaire 12 and Impact of Stressful Events, was significantly correlated with fatigue [7,70]. However, psychological well being was not associated with fatigue in the study by Walker et al. . Perceived stress, the individual's assessment of a stressful event, was highly correlated with fatigue scores [5,50,71], as well as recent stressful events . One study investigated the impact of traumas on HIV-related fatigue. The number of childhood traumas was associated with HIV-related fatigue, but this was not the case with adulthood traumas .
Depression in people with HIV/AIDS is widespread. In a study of Barroso et al. , 37.5% of the participants were moderately or severely depressed, as measured by Beck Depression Inventory-II. Fatigue and depression have a very strong positive correlation [4-6,50,52,58-60,64,72,73]. The more depressed the participants were, the worse their fatigue was . Higher levels of fatigue in patients with HIV/AIDS were strongly associated with the presence of depression and with increasing severity of depressive symptoms . However, instruments used to measure depression commonly include somatic complaints/symptoms that are similar to symptoms of fatigue .
A significant positive association was observed between anxiety and fatigue [4,52]. Anxiety can be divided into state anxiety (the emotional state at a specific moment in time and the intensity of that emotion) and trait anxiety (the differences between the way people perceive stressful situations and their reactions to these situations). Both state and trait anxiety were predictors of fatigue [4,5,50].
Certain coping styles like avoidance, behavioural disengagement and the use of self-distraction have been found to correlate with fatigue. From these studies, the direction of causality could not be established [56,74]. Denial, venting and substance use as coping style were not correlated with fatigue .
Medication and drug use
The use of antidepressants was found to be a predictor of fatigue in a study by Harmon et al.  in the bivariate and multivariate linear regression analysis. In this study, 39.1% of the study population were currently using antidepressants. It is possible that the antidepressants were not working in the intended manner, fatigue was a side effect of the antidepressants or the antidepressants were not sufficient to lift the symptoms of depression. This contrasts with studies by Barroso et al.  and Millikin et al. , who did not find a significant association between antidepressant use and fatigue. Use of anxiolytics, benzodiazepines or any alternative therapies was not found to be associated with fatigue [4,10,64]. Fatigue was significantly related to the use of sleep medication. This finding was observed in a cross-sectional study in HIV-infected women co-infected with human papillomavirus and known with low-grade intraepithelial lesions of the cervix . Another study in a predominantly male population could not confirm that association . Participants with a history of or current intravenous drug abuse were significantly more likely to report higher fatigue intensity as measured by the SSC-HIVrev and the MOS: Short Form-36 . Several other studies, however, did not find an association between any drug use, either current or in the past, and fatigue [4,10,56]. For alcohol abuse, no significant association with fatigue was found [4,10,70].
The treatment of HIV-related fatigue
Reported interventions for fatigue in HIV-infected patients can be subdivided into medication and psychological interventions.
In an 8-week, placebo-controlled trial, HIV-infected patients with a Diagnostic and Statistical Manual of Mental Disorders (DSM), fourth edition, diagnosis of major depression were randomized to fluoxetine, testosterone cypionate at an initial dose of 200 mg biweekly, which was increased to 400 mg biweekly or placebos. The Clinical Global Impressions (CGI) Scale, a standard clinician-rated instrument to evaluate mood outcome in clinical trials, was used to measure responses. Scores of 1 ('very much improved') or 2 ('much improved') on this seven-point scale were used to define responses. One hundred and twenty-three men were enrolled in the study; 90 (73%) completed the 8-week trial, with 30 completers in each study arm. All but four patients had total serum testosterone levels within the reference range at study baseline. The response rates after 8 weeks of treatment were 54% for the fluoxetine, 44% for the placebo and 47% for the testosterone group. These differences were not significant. Using a more than 50% decline in the Hamilton Depression Rating Scale score to define clinical response, differences between treatment groups remained insignificant. In an intention-to-treat analysis, response rates were 52, 61 and 51% for fluoxetine, testosterone and the placebo, respectively. In the logistic regression analysis adjusted for other explanatory variables, the response rate of the testosterone group was significantly higher than that of the fluoxetine and placebo groups . Study limitations include the rather small sample size. Given the high placebo response rate, it would require large samples to find a difference in outcome and the effect size would remain modest. The finding in this study that testosterone but not fluoxetine alleviated fatigue suggests that, in this patient sample, the fatigue was not only secondary to depression, but rather an independent problem. Use of testosterone was also evaluated in a clinical trial including HIV-infected men with diminished libido and one additional symptom of clinical hypogonadism (e.g. low mood, low energy, involuntary weight loss). Out of a study population of 108 men, 73 had clinical fatigue as rated by the clinician. After 12 weeks of treatment with testosterone, 71% had much more energy and none of them had to stop participation in the study because of side-effects. These men had significantly better scores on the CFS and VAS energy scales. The mean dose of testosterone cypionate used by the respondents was 385 mg, injected intramuscular biweekly . Another study on testosterone supplementation included HIV-infected men with a CD4 cell count below 400 cells/µl and clinical hypogonadism. A 6-week double-blind trial followed by 12 weeks of open-label maintenance treatment with biweekly injections of testosterone cypionate was done in 70 patients. About 80% of the patients had testosterone levels within the reference range. After 6 weeks, erectile dysfunction and mood disorder improved significantly in the patients taking testosterone. Among all patients, a trend was observed to report a greater decline in CFS scores compared with placebo, and significantly more so for those with fatigue at baseline. For those randomized to testosterone, mean change in serum testosterone levels between baseline and week 5 did not differ between responders and nonresponders. Interestingly, for those randomized to placebo, the mean testosterone increase in placebo responders significantly exceeded that in placebo nonresponders. After 18 weeks, all measures of depressive symptoms, distress, quality-of-life satisfaction and enjoyment, fatigue, libido and erectile response showed a statistically significant improvement from study baseline. No difference in response rate was found between men with serum testosterone levels below versus within the laboratory reference range among the 38 men randomized to receive testosterone. Study limitations include the brief duration of the double-blind phase of treatment. The significant increase in testosterone levels between baseline and week 5 is an example of the variability in testosterone levels . In a double-blind, placebo-controlled, randomized clinical trial, Knapp et al.  found that weekly intramuscular injections of 300 mg testosterone enanthate for 16 weeks in men with involuntary weight loss was associated with a significant improvement in energy/fatigue measured by the Medical Outcomes Study-Short Form 30. The studies were all done in male patients.
A precursor of both androgenic and estrogenic steroid hormones, dehydroepiandrosterone (DHEA), was also studied for its effect on HIV-related fatigue in a sample containing 39 men and six women with depressed mood (DSM-IV criteria) or a Hamilton Rating Scale for Depression score of 8 or higher, including either depressed mood or loss of interest plus low energy. In an 8-week, open-label trial, 81% of the sample responded. A respondent was defined by a CGI rating of 1 or 2 after using DHEA for 8 weeks, whereas nonrespondents did not show significant changes in fatigue scores. After randomization of the respondents for a 4-week double-blind placebo-controlled discontinuation trial, no difference in relapse rate was observed between those receiving DHEA and those receiving the placebo . DHEA administration caused large increases in serum DHEAS levels during the study, but there was no influence on serum testosterone levels among men over time. Nevertheless, the muscle mass increased, and this raised questions about the mechanism of effect. Levels of serum testosterone did increase dramatically in the two women included. The high response rate and no relapse during the double-blind continuation phase of this open label trial are suggestive of a placebo effect. However, this does not explain the maintenance of physiological effects during the double-blind phase. The sample size and study design are not suited to come to conclusions. There were not enough female completers to analyse data by sex.
HIV-associated wasting, defined as involuntary weight loss of at least 10% over the preceding 12 months, may be treated with recombinant human growth hormone (rhGH). rhGH treatment-mediated improvements in exercise physiological and functional performance have been documented in patients with adult growth hormone deficiency. Although body mass significantly increased after rhGH treatment, the changes in fatigue scores measured by Profile of Moods Scale (POMS) between the rhGH group and the placebo group were not significant .
Multiple psychostimulants have been studied for their effects on HIV-related fatigue. Wagner et al.  included 19 HIV-infected patients with a DSM-III-R depressive disorder diagnosis, debilitating low energy and CD4 cell counts less than 200 cells/µl. Fatigue scores improved significantly among all persons completing the open-label trial of 6 weeks of dextroamphetamine, with a mean maximum daily dose of 18 mg. Of the sample, 95% were defined as CGI 1 or 2 respondents. This study was an open-label trial conducted in the early cART era. It was followed by a placebo-controlled trial in 23 patients with a DSM-IV depressive disorder diagnosis and debilitating fatigue. The intervention group taking dextroamphetamine showed a significant improvement on the Chalder Fatigue Scale (CFS) and Visual Analog Scale (VAS). The patients receiving the placebo showed a smaller, but still significant, improvement on the CFS, but no significant improvement on the VAS [80,81]. In a study on ambulatory patients with persistent fatigue (for at least 2 weeks or more) randomized to the placebo or a trial of methylphenidate hydrochloride (Ritalin) or pemoline (Cylert) had significantly lower fatigue scores on the Piper Fatigue Scale and the VAS after 6 weeks of treatment, compared with patients receiving a placebo. Patients with major depression or other psychiatric disorders were excluded . Modafinil, another studied psychostimulant had a response rate of 80%, defined by a CGI 1 or 2 rating in a 4-week, open-label trial in patients with clinically significant fatigue as defined by interference with activities of daily living, including employment. Respondents had higher baseline fatigue scores and showed both statistically and clinically significant improvement on the Fatigue Severity Scale, whereas the mean score for the nonrespondents actually increased . For all studies on psychostimulants, the patient numbers are limited and the study period is relatively short. Persistent fatigue was defined by fatigue lasting for at least 2 weeks or more. One could question this definition in the light of HIV-related fatigue. Nevertheless, most studies do not define duration of fatigue at all.
A study in women with AIDS who reported a moderate-to-low baseline quality of life compared with 10-week, group-based, cognitive-behavioural stress management and expressive supportive therapy intervention with an individual psychoeducational condition on quality of life. Quality of life was measured before and after intervention using the MOS-HIV. A significant effect was noted for the total quality-of-life score, cognitive functioning, health distress and overall health perceptions, such that women in both conditions improved over the course of the study. No interactions or time effects were observed for energy/fatigue . Another study with a sample of 19 HIV-infected patients on cART, without AIDS, divided patients into relaxation techniques, psychotherapy or nonpsychiatric treatment, according to the patient's desire for stress management. A significant improvement in POMS fatigue scores was reported after the use of relaxation techniques. A posthoc t-test indicated no difference in effectiveness between relaxation, psychotherapy and nonpsychiatric treatment .