Experience of fatigue and related factors among HIV-infected adults attending ART clinic in Ethiopia; a cross-sectional study

Background: Fatigue is one of the most common bothersome HIV-related morbidity. The HIV prevalence in Ethiopia is heterogeneous by sex, geographic areas, and population groups. In Ethiopia, there is a need to estimate the burden of fatigue among HIV/Acquired Immune Deciency Syndrome (AIDS) adults to gain regional insight into this disabling symptom. Method: An institutional-based cross-sectional study was conducted among 392 HIV/AIDS patients attending an antiretroviral therapy clinic at the University of Gondar Hospital, Ethiopia using a systematic random sampling technique. Data were collected using a structured questionnaire, nine-item version Fatigue Severity Scale (FSS), and PHQ-9 (Patients Health Questionnaire 9). Logistic regression model was used to identify factors associated with the reported presence of fatigue. Result: The mean age of the participants was 40.5 ± 8.5 years. The prevalence of HIV-related fatigue was 53.3% and about 66% of HIV-infected women experienced fatigue. The factors associated with fatigue experience were; female gender (AOR: 0.196, 95%CI; 0.05, 0.92), being married (AOR: 0.13, 95% CI 0.23, 0.7), low income (AOR: 12.3, 95% CI 2.5, 60.15), unemployed (AOR: 3.9, 95%CI (1.02, 14.739), parity (AOR: 7.99, 95% CI 1.66, 38.41), being anemia (AOR: 13.34, 95% CI 2.74, 65.01), mild weight loss (AOR: 4.9 95% CI 4.33, 19.5) and moderate weight loss (AOR: 5.5 95% CI 3.11, 21.3), respectively. Conclusion: The ndings of this study revealed that experiencing fatigue is quite high among adults living with HIV. It is important for health care professionals and people living with HIV to understand; the possible causes of fatigue, remedies, and ways to reclaim energy. The predisposing factors and complications that cause fatigue should be aggressively diagnosed and treated by the clinicians.

related symptoms. Several studies report an estimated prevalence of fatigue ranging from 33 to 88% [8,[13][14][15] and the symptom is highest reported among the other entire HIV-related and ART-related symptoms in HIV positive patients [16].
A previous study from a different region in Ethiopia reported the prevalence of fatigue among HIV/AIDS adults to be 51.7% [17]. The prevalence of fatigue among HIV/AIDS patients and predictors vary widely across countries and regions in the country. Several studies have reported that fatigue can negatively impact the patient's activities of daily living, quality of life, sociability, job desire, productivity, level of physical activity, psychological wellbeing, health-seeking behavior, and adherence to the HAART regimen [15,16,[18][19][20][21][22]. The cause of fatigue in HIV-infected patients is probably multifactorial. Most of the studies focused on the physiological and/or psychological factors [21,23,24]. Further, evidence suggests that fatigue is associated with age, gender, malnutrition, insomnia, unemployment, poor income, family burden, depression, social support system, several disease-related factors such as the stage of the disease, anemia, the use of ART and certain laboratory parameters, as well as with socio-demographic and psychological factors [8,9,13,14].
The attention in HIV care is shifting towards symptom control and improving the quality of life of HIV infected patients elsewhere [25]. Sub-Saharan Africa is the hotspot of the HIV/AIDS pandemic and a large number of people living with HIV/AIDS reside in Ethiopia [1,2]. Yet, there is a dearth of research in Ethiopia focusing on HIV-related fatigue as a primary outcome of interest and the study area is a challenging geographical terrain where fatigue can be very disabling. The objective of this study was to determine the prevalence of fatigue de ned by the Fatigue Severity Scale (FSS) and the factors associated among adults living with HIV/AIDS attending antiretroviral therapy at public health facility of Gondar city, Amhara, Northwest Ethiopia.

Study design, setting, and population
An institutional-based cross-sectional design study was conducted from March to May 2019 at the antiretroviral therapy (ART) clinic, University of Gondar Specialized comprehensive hospital (UOGSCH).
The hospital is found in Gondar town at 748 km far from Addis Ababa, a capital city of Ethiopia, to the northwest, at an altitude of 2,706 meters above sea level. UOGSCH is a 550 bedded multidisciplinary specialized governmental teaching hospital. Presently, it provides health care services to more than 5 million urban and rural inhabitants in its catchment area (19,20). The HIV care unit and ART clinic started in 2003. As of 2018, this institution has served about 11277 HIV infected patients, on average 90 to 130 patients every working day. This hospital provides cost-free HIV testing, CD4 count monitoring regularly, medical consultations, counseling, and ART medications. Ethical approval was obtained from the College of Medicine and Health Sciences, ethical review committee, University of Gondar. Each participant signed written consent before participation. Adult HIV positive out-patients diagnosed by infectious disease physicians, both genders aged 18 and above, conscious and able to speak Amharic (local language) attending UOGSCH, ART clinic were eligible for inclusion. Pregnant and hospitalized HIV-positive patients were excluded from participation.

Sampling and data collection
The sample required for this study was determined using a single population proportion formula by assuming the prevalence of fatigue among HIV infected adults to be 50%, with a 95% con dence interval, and marginal error 5%. The derived power calculated sample was n = 371. Accordingly, the nal sample size with added 10% contingency was found to be 408. From the registered list of HIV infected patients attending the HIV care/ART clinic each day during the study period, K th patient was selected and the 1 st participant between 1 and K th was randomly selected, then taking every K th number thereafter. Data were collected by two trained physiotherapists who were randomly recruited from the registered list and were paid per diem. Interview method, measurements, and a structured questionnaire were used for data collection. Socio-demographic and clinical information (comorbid, duration of illness, stage, CD4 count, viral loads, prophylaxis history, hematological values) were extracted from the patients' medical records, ART logbook, and follow up cards. Fatigue Severity Scale for measuring fatigue, Patient's questionnaire-9 for measuring depression, and Insomnia severity scale were also used.

Variable de nitions
Fatigue is de ned using the 9-item Fatigue severity scale (FSS) [28]. Each item in FSS is scored on a 7point Likert scale ranging from 1 ("strongly disagree") to 7 (strongly agree) on how much fatigue affects the activities and lifestyle of a person [29,30]. The minimum score is "1" and the maximum score being "63", another way of scoring is by calculating the mean of all the scores with "1" and "7" being minimum and maximum scores respectively. A cut-off score of 4 or more was considered indicative of problematic fatigue [29]. A patient was considered physically active if the participant reported activity ≥ 150 min per week [31]. The Patient Health Questionnaire-9 (PHQ-9) was used to measure the depression level (case cut off ≥ 5) of the participants [32]. A participant who scored > 7 on the Insomnia Severity Scale was de ned as an insomnia case. Anemia was de ned according to WHO criteria. For men, anemia was de ned as hemoglobin concentration (Hb) less than 13g/dl, while for women; the cut-off is less than 12 g/dl [33]. HIV infected patients with weight loss or wasting were recognized using a set of rules/criteria developed by the expert physicians in the HIV care clinic, UOGH, and investigators of this study (Additional le 1).

Statistics
Data were coded and entered into EPI Info version 7.0 and exported to IBM Statistical Package for Social Sciences (SPSS) version 24 for Windows. The representativeness of the study sample to the research setting population during the study period was examined using formulas proposed by Cochrane to calculate the normal approximation frequency [34]. Descriptive statistics were used for all participant characteristics and factors associated with fatigue de ned by FSS. Co-linearity diagnostics were performed for each of the variables in the full model. All variance in ation factors were <10, and the condition in ation factors were <30, indicating that multi-colinearity was not a problem for this model. With fatigue (categorized: yes versus no) as the dependent variable, univariate and multivariate binary logistic regression analyses were carried out to examine the association with different independent variables. Variables were entered into the model using forced entry and categories were used as covariates for detailed analyses. Model t was assessed by the described method [35] and results were considered statistically signi cant when 95% con dence intervals not containing unity (equal to p-value < 0.05). Chi-square test or Fisher's exact test was used to determine the prevalence distribution of fatigue and estimate its association with different predictor variables. Predictor variables that were found to be associated with HIV-related fatigue (categorized; yes or no) in univariate model were gender, marital status, income, educational status, occupation, having children, anemia, weight loss, total duration since HIV con rmed, co-morbid condition, WHO classi cation clinical stage of HIV/AIDS and CD4 counts and the same were included in multivariate analysis. Interaction terms were used to examine the potential association between predictor variables and fatigue. When a clear sub-group seemed present in the data set, signi cance testing (Pearson X 2 ) and, if appropriately sized subgroups per category remained, the same was exported to the logistic model. This study is reported in accordance with the STROBE reporting guidelines (Additional le 2).

Socio-demographic characteristics
A total of 408 HIV positive adults were approached for consent, among which 392 participants consented to participate in this study, with a response rate of 96.1% and this is more than 100% of the power calculated sample size (n = 371). Those 16 patients who did not consent or agree to participate in the study reasoned lack of interest and time constrain.
The mean age of the participants was 40.5 (± 8.5) years and the majority of them 259 (66.1%) were females. Only 7.9 % of them reported to be from rural, majority of the participants were underweight (52.8%), above one-fourth did not have formal education. About 47% reported to be jobless and 43% answered to have low income (< 1500 Ethiopian birr). The majority of the participants 77.3 % reported lack of family support. Most of the subjects (94.1% and 86.9%) self-reported no previous or current smoking and alcohol habits respectively Table 1.  The frequency distribution of fatigue symptoms for the 9-item fatigue severity scale (FSS) reported by the respondents was almost even and majority of them reported that "exercise brings on my fatigue" 57.4% (225/392), and "I am easily fatigued" 55.9% (219/392) Figure 1.

Regression analysis
Prior to analysis, fourteen variables potentially related to the experience of fatigue were identi ed for regression analysis: gender, age, marital status, education level, employment status, income index, parity, duration of HIV, HIV/AIDS clinical-stage, weight loss, anemia, comorbid conditions, depression, and physical activity. Of those, age, duration of HIV, and physical activity were found not signi cantly associated in univariate analyses. The remaining eleven variables were entered into the multivariate model. In multivariate analyses when adjusted for other independent variables; gender, marital status, income level, employment status, parity, anemia, depression, and weight loss were found to be signi cant predictors.
HIV-infected adult women patients were 3 times more likely to experience fatigue than their counterparts [AOR 3.19, 955 CI: 1.05, 6.92], patients who were married and living with their spouse were 87% less likely to experience fatigue than those who were not with their spouse. Those who were unemployed and had low income were four times (AOR 3.9, 95% CI:

Discussion
The ndings of this study showed that the overall prevalence of fatigue using the 9-item Fatigue Severity Scale (FSS) among HIV/AIDS patients attending the ART clinic at University of Gondar specialized comprehensive hospital in Gondar city, Ethiopia was 53.3% (95% CI: 48.5, 58.4). This is to some extent at the higher end of the prevalence range (33 to 88%) reported previously in the literature [8,[13][14][15]. This nding is harmonized with the regional study conducted in Tigray, Ethiopia 51.7%, studies in USA 54%, South Africa 55%, and Canada 54% [17,29,36,37]. Larger sample size, similar patient characteristics, and inclusion criteria in these studies could explain the consistent ndings.
Surprisingly, in this study, there is no association between advanced HIV disease marked by CD4 count, duration of HIV, ART regimen, clinical stage of HIV, and the experience of fatigue when adjusted for the other independent variables in the regression model. Although drugs like dovudine and didanosine are frequently reported to have fatigue as side effects. Antiretroviral agents might help the control of viral replication and slower disease progression, which could mitigate the risk of fatigue associated with ART.
Further, the distinct difference between the perception of fatigue and performance fatigue, level of fatigue, and energy demand of the respondents could have in uenced the ndings [38]. Then again, the nding of this result was lower than the ndings reported in other studies done in a rural district in Southwest Uganda 61% [39], South Africa 66.7% [40], Rochester, USA 64% [41], China 86.8% [42], and the United Kingdom 65.1% [43]. This discrepancy might be due to the reasons that the Ugandan and the USA included lesser samples 212 and 128 respectively. Moreover, the Ugandan study used Memorial Symptom Assessment Scale-Short Form (MSAS-SF), the UK used a self-administered questionnaire with Chalder Fatigue Scale to measure fatigue, though the study from USA used similar outcome measure (FSS) the cutoff was lower, the study in South Africa used convenient sampling to report multiple selfreported symptoms with fatigue one among those. The Chinese study was based secondary source data (case-report form) using a retrospective design. Overall, the variations in the sampling method, study design, outcome tools, data collection method, and clinical characteristics of the patients could explain these discrepancies.
In this study, the key predictors that were signi cantly associated with fatigue among HIV/AIDS adults were clinically diagnosed anemia, low income, parity, and living without a spouse. Likewise, studies done in the USA and China reported an association between fatigue and anemia [42,44,45]. Fatigue is the cardinal symptom of anemia, a prognostic marker of disease progression, and frequently reported as a predictor of morbidity and mortality among HIV/AIDS patients. The negative effects of HIV infection and cART on the hematopoiesis contribute to low hemoglobin concentration leading to impaired oxygen transport to the vital organs and musculoskeletal system resulting in fatigue. The resolution of HIVrelated anemia has been shown to improve fatigue individuals with HIV [46].
We are unsurprised that depression is signi cantly associated with fatigue, low energy or tiredness itself is a depressive symptom. Similar to this study, many studies and a systematic review of 42 studies reported that inadequate income, unemployment, and depression are stronger predictors and uniformly associated with HIV related fatigue [36,[47][48][49]. Less energy, lack of work capacity, an inclination to work and reduced motivation may eventually result in unemployment and inadequate income. Importantly, fatigue that precedes depression or results from depression is a distinct health outcome among HIV/AIDS patients and the interaction can sometimes become a vicious cycle. The present study also found that female patients were more likely to perceive fatigue than men. This nding is similar to the studies conducted in South Africa [50], Canada [51], and a systematic review [49]. The mean age of women in this was lower than men, majority of those who diagnosed with anemia and weight loss were women.
Besides, a higher representation of women in this study could have favored this association.
In this study, having children was associated with the experience of fatigue. A similar association was reported by studies in the USA, France, and South Africa [36,48,52]. The ndings might be attributed to additional responsibility and workload in terms of the physical, social, and nancial burden that are associated with having children. However, participants who were married and living with their spouse was a protective factor for fatigue than those who lived without a spouse. Emotional stress, grief, loneliness, and depression attributed to the loss of spouse, separation, and lack of support could be the possible explanation for likely risk of fatigue among those living alone.
The current study also found the mild and moderate weight loss was signi cantly associated with fatigue. This was similar to the study done in Western Cape, Bellville, South Africa [53], and Southwest Uganda [39]. Although the association between fatigue and severe weight loss is non-signi cant, the coe cient above 1 in the regression model explains the risk of fatigue. Malnourishment, wasting, loss of appetite, side-effects of drugs, illness related to co-infections, and HIV related complications are blamed for unexplained weight loss among HIV/AIDS patients. Each of these factors can cause loss of energy and/or tiredness and are predictor or risk factor for developing fatigue.
Limitations of this study include, unlike longitudinal design, which explains the temporal relationship, the cross-sectional nature of the data reported in this study might reduce generalizability. We were not able to distinguish the role of the HAART regiment due to the limited enrolment of pre-ART patients. Further, we could not draw any conclusions related to the direction of causality, and cause of fatigue. However, as a description of HIV-infected adult patients in Ethiopia in clinical practice, we believe our power calculated sample size, representative sample of clinical population and use of a valid outcome tool should enhance the external validity of the data presented and provide a powered preliminary insight of the burden and factors related to HIV-related fatigue in Ethiopia. Future studies should explore the effect of fatigue on the adherence to ART regimen, disease status, and other relevant clinical outcomes among adults with HIV/AIDS in Ethiopia.

Conclusion
The prevalence of fatigue was found to be high in people living with HIV/AIDS. From all the sociodemographic factors and HIV/AIDS related medical factors that were studied, being female, being married, low income, unemployed, having children, severe weight loss, and anemia were found statistically associated with fatigue. Hence, early detection of the HIV-related fatigue and its underlying treatable causes permits patient education, controlling risk factors, and treat these conditions when they are present or predisposed. The study was conducted after obtaining ethical approval from the Institutional Review Board (IRB), CMHS, University of Gondar (Ref no; SOM/087/9/2017). Permissions were obtained from the regional public health institute and the authorities of the study site before the study. Written consent was obtained from each participant after explaining the study, its objective, bene ts, and its importance. Information's were recorded anonymously and con dentiality and bene cence were assured throughout the study period.

Consent to publish
Not applicable

Competing interests
The authors declare that they have no con ict of interest.

Funding
This study was funded by the University of Gondar (Grant ref no: 08792017). The views presented in the article are the authors and not necessarily express the views of the funding organization. University of Gondar did not involve in the design of the study, data collection, analysis, and interpretation.
Availability of data and materials: All data relevant to our ndings are contained within the manuscript. Requests for further details on the dataset and queries concerning data sharing shall be arranged based on a reasonable request to the corresponding author.
Authors' contributions MB brought the original idea and was involved in the proposal writing, designed the study, and participated in all the implementation stages of the project. MB and BJ analyzed the data and wrote the manuscript. BF, SF, SM, and MB participated in the conception of the original idea and were involved in the proposal writing, nalized the write-up of the manuscript, and critically revised the manuscript for important intellectual content. BJ, MB, SM, SF, and BF were responsible for critically revising the research proposal and the manuscript, and participated in its design and interpretation. BJ and MB were involved in the design of the work and approved the version to be published. All the authors read and approved the nal version of the manuscript.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.