Prevalence and predictors of HIV-associated Neurocognitive Disorder in Ethiopians

Two hundred and forty-four people living with HIV were interviewed in a study using a cross-sectional design and International HIV Dementia scale (IHDS). Additionally, the sociodemographic, clinical and psychosocial characteristics of the patients were recorded. Chi-square and binary logistic regression analysis were used to determine the level of signicance among the independent risk factors and probable HAND. This study showed the point prevalence of probable HIV associated neurocognitive disorder in an Ethiopian population living with HIV as 39.3%. The use of social/recreational drugs, poor medication adherence, and 40 years of age or older were signicant predictors of cognitive impairment in people living with HIV. Additional strategies to curb this epidemic would emphasize early screening for the diagnosis of dementia in the resource-poor sections of the Ethiopian community. This would further help to direct the targeted patients for further follow-up in higher psychiatric referral hospitals for assessment and management, given the positive association between people living with HIV and HAND. Furthermore, awareness campaigns about the deleterious effects of khat chewing and drinking excess alcohol on the functional effects of the brain need to be investigated.


Background
In recent decades, the successful development and widespread implementation of anti-retroviral therapies have seen an associated increase in the longevity of HIV-infected patients. However, this important public health achievement has presented new challenges of clinically maintaining the health and quality of life of these patients. One of the major health de cits a icting HIV-infected individuals is the development of neurocognitive disorders, including progressive symptoms of dementia. The successful implementation of cART regimens has considerably reduced the prevalence of the most severe form of dementia, i.e., HIV-associated dementia (HAD) (1,2). However, milder forms of this condition, namely, mild neurocognitive disorder (MND) and asymptomatic neurocognitive impairment (ANI), are commonly prevalent and may reduce the quality of life of HIV-infected patients. These clinical manifestations have been an important concern, especially for patients belonging to special clinical categories, including those with a late diagnosis of HIV, untreated adults, pregnant women, patients with low medication adherence, and pediatric patients, associated diagnoses which make affected patients particularly vulnerable to neurocognitive impairment could potentially interfere with daily life functioning such as work ineptitude, increased driving errors, and poorer adherence to treatment, which may indirectly in uence the viral load and CD4 lymphocyte status. A lower CD4 count is the hallmark of a weaker immune system and compromises the ability to ght infections, thus increasing the degree of neurodegenerative insult by attacking the microglia and macrophages in the central nervous system. Additionally, HIV-infected persons frequently present with comorbidities such as Hepatitis-C co-infection, drug abuse, or prior head injuries, which can further exacerbate HIV-related effects on the brain. Thus, early diagnosis of the neuroHIV symptoms is important for improved CD4 count and reduced viral load. In support of this public health objective, the UNAIDS, an HIV education, and prevention program of the United Nations, has strategically supported to improve the quality of life of patients with existing sexual and reproductive health services (10).
Post cART era there has been decrease in the prevalence of HAD but up to 40% still suffer from HAND (11,12).
Additional study by Heaton et al group in 2011 revealed greater than 50% of HIV + patients showed HAND symptomology (13). Nevertheless it is prudent to mention that unavailability of sophisticated tools to parse the asymptomatic neurocognitive disorder (ANI) from mild neurocognitive disorder (MND) in the present population, IHDS was used as a screening tool for those individuals who are at high risk of dementia only. However, no studies have investigated the prevalence of HAND and its associated conditions in a population of Ethiopians living with HIV in the present facility setting. Therefore, the present study sought to determine the prevalence and predictors of HIV-associated neurocognitive disorder (HAND) using International HIV dementia scale (IHDS).

Participants and procedures
The target population was people living with HIV, who were residents of Mizan-Aman, Ethiopia. The accessible population consisted of people living with HIV who were attending the HIV/ART clinic of Mizan-Tepi University Teaching Hospital (MTUTH), Aman, Ethiopia. The study was carried out for a period over two months from February 2018 to April 2018. Out of the total of 384 patients attending the HIV/ART clinic of MTUTH during these months, 300 eligible patients initially agreed and signed an informed consent form. Of the 300 eligible patients, 250 were nally selected using a simple random sampling method. Inclusion criteria were (i) age greater than 18 years and (ii) mentally stable, as determined by the attending clinicians. Finally, after removing the construct-level missing data, 244 samples were nally used for quantitative analysis (See Supplementary File attached). Three trained psychiatric nurses from the HIV/ART clinic of MTUTH performed the structured interview. The four words used as a part of recall memory test were translated into Amharic by a native Amharic language expert. These four words were native to local Ethiopian community and are used commonly in the Ethiopian cultural context. The dementia task comprised 3 tasks which assessed memory recall, motor speed, and psychomotor speed. The rst task involved a short-term memory task in which the participants were given four words to recall (dog, hat, bean, red) (translated into Amharic as wusha, ko ya, bakele, keyi) and were provided one second to say each word. Amharic translations of these four words are native to the local Ethiopian community and are used commonly (i.e. High-frequency words) in the Ethiopian cultural context. Though, it is plausible to think that the word length and number of syllables might play some role in recall memory, because the Amharic translation of these words are slightly longer and have slightly higher number of syllables. However, the associative retrieval mechanism facilitated by the high-frequency nature of these four words in both English and Amharic versions would have somewhat compensated the effect caused by the variation in the word lengths and number of syllables. The participants were then asked to remember the 4 words and told that they would be asked to recall the words again a bit later. This was followed by a motor speed task in which the patient was asked to tap the rst two ngers of the non-dominant hand as quickly and as rapidly as possible. The maximum score for the motor task was 4 points, with speci c performance levels being scored as follows: 4 = ≥ 15 taps in 5 seconds; 3 = 11-14 taps in 5 seconds; 2 = 7-10 taps in 5 seconds, 1 = 3-6 taps in 5 seconds; and 0 = 0-2 taps in 5 seconds with the maximum 4 points for motor speed task. Psychomotor speed was further assessed by asking the patient to perform several movements with the non-dominant hand as quickly as possible. Primarily the patients needed to clench their hand into a st on a at surface. They were then asked to put their hand at on the surface with their palm down. Finally, they were asked to place their hand perpendicular to the at surface while displaying the 5th digit. The whole task was demonstrated once to the patients who were then allowed to practice twice before starting the test. A maximum of 4 points was possible for the psychomotor task, with patient performance being scored as follows: 4 = 4 sequences in 10 seconds; 3 = 3 sequences in 10 seconds; 2 = 2 sequences in 10 seconds; 1 = 1 sequence in 10 seconds; and 0 = unable to perform the task. Finally, the patients were given the follow-up to the memory recall in which they were asked to recall the four words. For the words that were not recalled correctly the patients were prompted with a semantic clue as follows: animal (dog); a piece of clothing (hat); vegetable (bean); color (red). A maximum 4 points was possible for the memory recall task, with the scoring as follows: one point for each word spontaneously recalled, and 0.5 points for each correct answer after prompting. The nal score was a sum of the three tasks, the maximum being 12. Patients who scored 10 points or less were referred for further dementia testing.

Sociodemographic measures
A questionnaire was used to gather sociodemographic information regarding the participants' age, gender, marital status, religion, ethnicity, and occupation. The questionnaire also recorded information related to substance use: this included information about participants' habitual use of commercial and indigenous alcoholic drinks such as tej, tella, areki, shamita, borde, and korefe, as well as about habits such as tobacco smoking, consumption of caffeinated drinks, and khat chewing (14) Clinical measures Data regarding participants' clinical symptoms were also recorded. These included the patients' current CD4 count, baseline CD4 count, viral load, duration on combination antiretroviral therapy regimen (cART), side effects from cART, opportunistic infections, duration since HIV diagnosis, other neuropsychiatric diagnosis, and duration of hospital stay.

Psychosocial measures
Information related to psychosocial factors such as support from family, perceived stigma accruing from HIV status, and discrimination from society was collected. Further, questions about perceived memory de cits in the past month which might have interfered with daily functioning were assessed by questions such as "Do you experience frequent memory loss?", "Do you feel you are slower when reasoning or solving problems?", and "Do you have di culties in paying attention?" (15).

The International HIV dementia scale (IHDS)
The International HIV dementia scale (IHDS) (16) was used to screen for HIV-associated neurocognitive disorder or HIV dementia. This tool has been validated in different African and Caucasian populations and has been shown to have a sensitivity of 88% and 80% and speci city scores of 50% and 55% respectively (16)(17). The tool measures three essential components of neurological impairments: these include cognition, motor, and psychomotor de cits. Each component has a maximum score of 4, with a total score of three components summing to 12 (16). Any value of less than 10 is indicative of neurocognitive de cits, and patients receiving such a score are referred for further psychiatric follow-up by higher referral hospital (16). The IHDS does not require pro ciency in the English language and is ideal for measuring probable HAND in people living with HIV.

Statistical analysis
Data were analyzed using the Statistical Package for the Social Sciences (SPSS), version 21 (SPSS Inc., Chicago, IL). Bivariate correlation and Chi-square tests were used to determine the correlation between the independent variables and probable HAND status; those sociodemographic, clinical and psychosocial risk factors with a p-value of less than 0.25 were selected for the multivariate association analysis with the outcome measure. Binary logistic regression was used to assess the multivariate association between the dependent variable, i.e., HAND status and independent variables. Models were adjusted for age, gender, education, marital status, severity of illness (hospital stay) and stigma (indirectly related to increased anxiety and depression) towards people living with HIV. A Mann Whitney test was performed to assess the difference between the mean scores of participants in the probable HIV dementia group and the non-demented group.
Logistic regression was applied after verifying all its assumptions in the study data. In general, the dependent variable, namely, HAND status was measured as a dichotomous variable. Second, there was independence of observations as assessed by the Durbin Watson test. No outlier was found, as assessed by Mahalanobis Distances for multivariate outliers and box plot analysis for univariate outliers. Independent variables (continuous) were linearly related to log odds as determined by the absence of signi cance for the interaction effect. Additionally, the study data also satis ed the minimum sample size requirement. Based on the present prevalence of 41% for HAND, the sample size calculation for 9 independent variables included in the model and the expected probability of the least frequent outcome being .10, it was determined that a minimum sample size of (10*9/.4) =90/.4=225 was needed.

Results
Sociodemographic characteristics of the study population Table 1 presents the patients' characteristics. Most of the patients were under 40 years of age (85%). The majority of the patients (64%) were female and approximately three-fourths (69%) were married. Most of the patients (70%) reported having completed less than eight years of primary school or had no formal education. One-fourth of the patients admitted that they used recreational drugs (25%). Nearly 65% of the patients belonged to the low and very low-income groups. Most of the patients lived in urban locations (86%). No signi cant association between education level and gender groups was found; χ2(3) = 4.985, p = .173.
Clinical characteristics of the study population Table 2 shows the clinical characteristics of the study participants. Approximately 96% (235, 96.4%) of patients had been diagnosed with HIV for greater than 6 months of which 94 (40%) showed symptoms of HAND. The current CD4 count is a good marker for dementia as per the WHO recommended criteria; patients with current CD4 count < 500 cells/mm 3 could potentially show signs of cognitive impairment. In the present study, approximately 44% of the HIV diagnosed patients had a current CD4 count <500 cells/mm 3 . Most of the patients reported that they did not adhere to their prescribed medication regimen (88%). About one-third of the patients were diagnosed with opportunistic infections at the time of participation in the study. None of the clinical characteristics showed any signi cant statistical association with HAND in the bivariate analysis.
Psychosocial characteristics of the study population Table 3 shows the psychosocial characteristics of the patients enrolled in the study. Most of the patients (87%) reported no social support for medical adherence from their family or friends. Most (about three fourths) of the participants reported that they did not experience any stigma and discrimination from society.

Differences in HAND component scores between demented and non-demented groups
The prevalence of HAND identi ed by IHDS was 39.34% with the IHDS total score of 9.57±1.57. Independent mean IHDS recall score was revealed as 3.36±0.80, motor score as 3.32±0.66 and psychomotor score as 2.90±0.72 (Table 4). The difference between those who had HAND and those who didn't was evident in all the three components of IHDS, i.e., recall [U = 2732, p = p<.001], motor performance [U = 2704.500, p<.001] and psychomotor abilities [U = 1930.500, p<.001; see Figure 1].
Factors associated with the neurocognitive de cit: a multivariate analysis Binary logistic regression was performed to measure the effects of age, gender, current CD4 count, medication adherence status, stigma, marital status, hospital stay, social drug use and education on the HAND status. Gender, education, marital status, current CD4 count and hospital stay were included in the model based on previous studies showing them to be consistent predictors for HAND (18)(19)(20)(21)(22)(23)(24)(25). The logistic regression model was statistically signi cant, χ2(9) = 72.91. The model explained 35.0% (Nagelkerke R 2 ) of the variance in dementia status and correctly classi ed 75.8% of cases. Participants who were older than 40 years of age were 2.8 times more likely than younger subjects to exhibit HAND (p<.05, Table 5). Those who admitted to that they did not adhere to their medication regimens were 2.9 times more likely to have symptoms of HAND (p<.05, Table 5). Study subjects who stated that they used recreational drugs were 13.6 more likely than non-recreational drug users to have symptoms of HAND (p<.001, Table 5 Discussion The present study determined the prevalence, and identi ed predictors for HIV dementia (HAND) in Ethiopians living with HIV. People living with HIV were found to have a high prevalence of neurocognitive, motor and psychomotor de cits. The present ndings con rmed the hypothesis that some of the patient-associated socio-demographics and clinical factors can serve as predictors of HAND. The identi cation of predictors, some of which may be more closely related to the Ethiopians living with HIV, may help in targeted screening of vulnerable groups during cART follow-up visits. This may greatly help in strategizing and implementation of the prevention program, more so, because: (i) HAND is an asymptomatic condition for considerable durations, and (ii) clinical trials of HAND therapies have been unsuccessful (9).
The majority of the present study sample (96%) had been diagnosed as having had HIV for more than 6 months and thus most of the patients (99%) were already on the cART regimen. Despite being treated with cART medication, some of the patients still showed symptoms of cognitive impairment. The possible explanation for this would be that many patients reported having poor medication adherence (88%) or inability of cART to achieve optimum therapeutic concentration in the CNS due to its poor penetration beyond the blood-brain barrier (34).
Similar to previous studies, being older than 40 years was found to have a positive association with HAND status in the people living with HIV (35)(36)(37)(38)(39). Generally, cognitive performance deteriorates with increasing age, and some studies have shown that being in the 40 plus age group is a signi cant predictor of HAND. One study in Uganda did not show any association between advanced age and HAND (33). This might be because of some moderator variables in their study (33), because consistency in association is suggested by multiple reports (35)(36)(37)(38)(39).
Recreational drug use has been shown to be a strong predictor of HAND in the present study. The present study assessed selfreported use of social drugs such as khat, alcohol, cigarette, coffee and other caffeinated drinks in people living with HIV. Drinking indigenous varieties of alcohol and khat chewing is among the most prevalent social habits in Ethiopia (40). There has been a growing body of evidence which points towards cognitive decline due to drug misuse (41)(42)(43). Social drug misuse frequently involves drug takers in various risk-taking behaviors (44). Some studies have shown that low to moderate alcohol consumption may actually facilitate cognition (45)(46); others have shown that it impairs cognition (47) or produces no change (48). Another recreational drug called Khat is abused widely in Ethiopia and has been associated with short-term memory de cits and loss of cognitive exibility in adults (49)(50)(51). Additionally, the deleterious effects of khat on cognitive performance have been similarly demonstrated in people living with HIV (52). Nevertheless, the association between khat and cognition is complex. Khat has a biphasic response on cognition, wherein low doses may enhance cognition until reaching a peak, at which progressively higher doses begin to produce adverse cognitive effects, and, additionally, are associated with classic signs of addiction similar to psychostimulants (53). Other lines of evidence have tended to suggest the presence of an adverse additive effect, in as much as, drug users without HIV are more vulnerable to cognitive impairment than patients who are HIV positive (54)(55)(56)(57). It's worth mentioning that since Khat is a social drug being extensively used in the present Ethiopian community setting and clinical and preclinical evidences show that it produces spatial memory, working memory and cognitive exibility de cits (49). Delineating khat neurocognitive effects from other social drugs like coffee, tobacco, alcohol and soft beverages may need a longitudinal follow-up of the HIV+ patients which was out of the scope of the present study. Furthermore. social drug abuse has led to a synergistic interaction with HIV which may need future longitudinal follow up trials and experiments for its con rmation. Nevertheless, this is consistent to a study in the preclinical population which demonstrated the combined exposure of HIV-1 Tat protein and clinical opioid namely Oxycodone to promote the psychomotor behavioral effects, thus making the drug more rewarding, with the initial perceived potency of the drug being an important predictor of whether the patient would enter into the addiction cycle (58,59) The study sought to identify which clinical factors might be most important for predicting HAND. The candidate predictors included the current CD4 count, duration of time since HIV diagnosed, presence of opportunistic infections, and WHO clinical staging. These factors, however, did not show any signi cant association with HAND status in our study. Inconsistent results have been reported between different clinical correlates of HAND (60,61). One of the possible associations, i.e., between selfreported lapses in medication adherence and HAND, was of interest inasmuch as patients frequently reported that being on multiple drug regimens made it particularly burdensome for them to reliably take their medicines (62). However, this potential association failed to reach statistical signi cance. The study subjects positive to have HAND symptomology should further be con rmed by more robust cognitive neuropsychological test batteries, computerized testing, Mini-Mental State Examination (MMSE), and grooved pegboard action uency (63).
There was an apparent contradiction in the ndings of bivariate and multivariate analysis with regards to the association of medication adherence with HAND. Poor medication adherence did not show a signi cant relationship with HAND in the bivariate analysis, but it did have a signi cant positive association in the multivariate analysis. This is most plausibly explained by statistical consideration that some of the covariates in the multivariate analysis in this study might have mediated the relationship (64).
The limitations of assessing the cellular and molecular changes in the brain morphology in clinical population makes it pressing need for identifying these aspects in preclinical population for further understanding the repercussions of combined expression of the HIV+ proteins and other factors which predict cognitive impairment.

Limitations of the study
The study failed to include any psychosocial correlates such as depression, stress, and anxiety, which could potentially in uence the outcome variable. The study also could not include patients belonging to special clinical categories, those with a late diagnosis of HIV, untreated adults, and pediatric patients, associated diagnoses which make these HIV patients particularly vulnerable to severe forms of dementia (HAD) (3)(4).
Moreover IHDS tool comes with limitations. Previous studies signify the use of IHDS to be suboptimal in screening cognitive impairment among HIV + individuals (65). However, in present limited resource setting where full neuropsychological testing was not possible, IHDS was used as a screening tool for those individuals who are at high risk of dementia only. Any patients showing probable neurocognitive symptoms would then be referred to a neurological facility for further diagnosis.. However, IHDS is not the gold standard and further follow-up of patients by robust neuropsychological test batteries who report memory issues is needed. Moreover, future studies may need to explore the psychometric properties of the IHDS scale to determine the sensitivity and speci city scores in the Ethiopian demographics. Moreover, future assessments of the sensitivity and speci city for recall, motor and psychomotor which are the independent components of the IHDS tool is the need of the hour.

Conclusion
This study showed the point prevalence of probable HIV associated neurocognitive disorder in an Ethiopian population living with HIV as 39.3%. The use of social/recreational drugs, poor medication adherence, and 40 years of age or older were signi cant predictors of cognitive impairment in people living with HIV. Additional strategies to curb this epidemic would emphasize early screening for the diagnosis of dementia in the resource-poor sections of the Ethiopian community. This would further help to direct the targeted patients for further follow-up in higher psychiatric referral hospitals for assessment and management, given the positive association between people living with HIV and HAND. Furthermore, awareness campaigns about the deleterious effects of khat chewing and drinking excess alcohol on the functional effects of the brain need to be investigated.

Declarations
Ethics approval and consent to participate This cross-sectional study was approved by the Institutional review board of the College of Medicine and Health Sciences, Mizan-Tepi University, Ethiopia. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed written consent was obtained from all participants prior to the commencement of the study.

Consent for publication
Not applicable.

Availability of data and material
The de-identi ed dataset used and/or analyzed during the current study is available as supplementary le.

Competing interests
The authors have read the journal's policy and have the following potential con icts: This study was not an industry-supported study. S.R. Pandi-Perumal is a stockholder and the President and Chief Executive O cer of Somnogen Canada Inc., a Canadian Corporation. This does not alter his adherence to all of the journal policies. Pandi-Perumal has edited several academic volumes for which he receives occasional annual royalties. He declares that he has no competing interests that might be perceived to in uence the content of this article. Other remaining authors declare that they have no proprietary, nancial, professional, nor any other personal interest of any nature or kind in any product or services and/or company that could be construed or considered to be a potential con ict of interest that might have in uenced the views expressed in this      Figure 1 Independent components of IHDS scores like recall, motor and psychomotor were calculated. * indicates a main effect wherein probable HIV dementia patients differ from respective non-dementia patients. p<0.05.