Worldwide Occurrence of HIV-Associated Neurocognitive Disorders and Its Associated Factors: A Systematic Review and Meta-Analysis

Background HIV-associated neurocognitive disorders are common in people living with HIV/AIDS and affect the adherence of patients to prescriptions, activities of daily living, and quality of life of patients. However, there is a lack of summative evidence in the area. The present meta-analysis was therefore addressing this gap. Methods We did our electronic search in Psych-Info, EMBASE, Scopus, and PubMed. The retrieved articles were stored with the endnote reference manager and data was extracted using Meta-XL version 5.3. The quality of studies was evaluated with the modified Newcastle–Ottawa Scale (NOS). A random-effect model and STATA-16 were used to compute the average estimate of HAND. Heterogeneity was weighed with I2 statistics. A sensitivity analysis and subgroup analysis were employed. The existence/nonexistence of a publication bias was checked with the Eggers test of publication bias. Results The average prevalence of HAND was 50.41% (95% CI: 45.56, 55.26). The average estimate of HAND in Europe was found to be 50.015% whereas in Africa, Asia, and the United States of America (USA) it was 49.566, 52.032, and 50.407% respectively. The prevalence of HAND in studies that used the HIV Dementia Scale (IHDS) was 36.883% and 59.956% at cutoff points of IHDS <9.5 and IHDS <10 respectively. Besides, the estimated average of HAND with the global dementia scale (GDS) was 40.766%. The prevalence of HAND in cross-sectional, cohort, and case-control studies was 49.52, 54.087, and 44.45% in that order. Socio-demographic variables; low level of education and older age, clinical and HIV related variables; the advanced stage of the illness and CD4 count of 500 cells/dl or less and psychological variables such as comorbidity of depression increases the risk of HAND. Conclusion The prevalence of HIV-associated neurocognitive disorders was about 50.41%. Low level of education and older age, clinical and HIV related variables such as the advanced stage of the illness and CD4 count of 500 cells/dl or less, and comorbidity of depression were associated with HIV associated neurocognitive disorders. Public health interventions for HIV patients should target these essential problems.


INTRODUCTION
HIV/AIDS is a global public health issue with more than 34 million people living with HIV/AIDS(PLHIV) currently (1). Mental, neurological, and substance (MNS) related disorders are very common in People living with HIV/AIDS (2). The latest systematic review and Meta-analysis studies by Necho et al. (3) revealed that 35.8% of HIV/AIDS patients had depressive symptoms (3). Another systematic review and meta-analysis studies reported that the prevalence of post-traumatic stress disorder (PTSD), alcohol use disorder (AUD), and suicidal ideation in individuals living with HIV/AIDS were 32.67% (4), 22.02% (5), and 21.7% (6), respectively.
Since HIV is a neurotropic virus, it affects the cortical and subcortical parts of the brain resulting in cognitive impairment (7). This impact of HIV on the cognitive domain of patients is known as HIV-associated neurocognitive disorder (HAND) (8,9). The level of HAND ranges from asymptomatic impairment to minor neurocognitive disorder and full-blown dementia (10)(11)(12)(13). The HIV-associated neurocognitive disorder affects memory, attention, problem-solving ability, language, higher executive function, and independent activities of daily living (14).
Different studies reported varieties of socio-demographic and clinical factors associated with HIV-associated neurocognitive disorders in people living with HIV/AIDS. For example studies from, Cameroon, Nigeria, Botswana, Singapore, Malawi, and Dessie Ethiopia reported that socio-demographic variables such as older age, female sex, and lower educational level were risk factors for HIV associated neurocognitive disorder (13,14,46,50,52,53). Besides, from Clinical variables CD4 count of < 500 cells/mm3 was related to HIV-associated neurocognitive disorder based on reports of studies from Brazil, Singapore, and Northern Nigeria (14,50,54). Moreover, advanced stages of AIDS and not being on highly active antiretroviral treatment (HAART) were associated with HIV-associated neurocognitive disorder in South Africa (51-53). In Uganda, behavioral and psychological variables such as depression, Body mass index, and alcohol abuse were associated with HIV-associated neurocognitive disorder (10). Moreover, medication non-adherence and opportunistic infections were associated with HIV-associated neurocognitive disorder (46,55).
The presence of HIV-associated neurocognitive disorder predisposes people living with HIV/AIDS to substance abuse, poor medication adherence, and unsafe sex so the poor quality of life and loss of follow-up from treatment are outcomes. These conditions speed up the progression of the virus to its advanced stages and the development of severe opportunistic infections and death (11,12).
Even though a high proportion of the world population has been living with HIV/AIDS and a high prevalence of mental, neurological, and substance use disorders in this population, these problems, especially neurocognitive disorders, are not investigated well. Despite the presence of some studies in the area, they are mostly confined to a small population and a narrow geographical area (8,10,(12)(13)(14). Consequently, there arises a need to have aggregate data regarding HIV-associated neurocognitive disorder and its associated factors from the global context. Therefore, this systematic review and meta-analysis aimed to estimate the prevalence of HIV-associated neurocognitive disorder in people living with HIV AIDS and to analyze the associated factors for HIV-associated neurocognitive disorder in people with HIV AIDS.

Search Strategy
We have performed our search strategy for this review in different ways. Initially, we did an electronic exploration for eligible articles regarding HIV-associated neurocognitive disorders in people living with HIV AIDS in the databases of Psych-Info, EMBASE, Scopus, and PubMed. As a sample of our search strategy with the PubMed database, we have used the following key terms: [(neurocognitive disorder or HIVassociated neurocognitive disorder or HAND) and (adults)] and (Human immunodeficiency virus or HIV or acquired immunodeficiency syndrome or AIDS). Moreover, Psych-Info, EMBASE, and Scopus databases were investigated in line with the searching guidelines of each database. Besides, the reference lists of included studies were searched manually for additional eligible articles. There was no time restriction to the publication year of the articles during the searching process.

Eligibility Criteria
During our study of a systematic review and meta-analysis on HIV-associated neurocognitive disorder in people living with HIV AIDS, we have set the following inclusion and exclusion criteria based on the: 1) the primary inclusion criteria to analyze where the study should assess prevalence OR associated factors of HIV-associated neurocognitive disorder in people living with HIV AIDS. 2) The HIV-associated neurocognitive disorder had also to be investigated using the International HIV Dementia Scale (IHDS), Frascati criteria, Mini-mental state exam(MMSE), global dementia scale(GDS), Brief Neurocognitive Screen, Neuropsychological battery, Montreal Cognitive Assessment (MoCA), In-depth neuropsychological assessment, Wechsler Adult Intelligence Scale and ADC.
We excluded studies (1) that assessed neurocognitive disorder in samples other than people living with HIV/AIDS. (2) That assessed neurocognitive disorder in individuals taking psychotropic medication. (3) Studies that are letters to the editor with non-original data content, earlier reviews, case studies, studies involving non-human subjects, and articles published in a language other than the English language were also excluded from the analysis. After all relevant articles were searched in the mentioned databases; they were stored in an endnote reference manager. Two of the authors (MN and YZ) individually screened the titles and abstracts of articles stored in an endnote reference manager using the eligibility criteria. Next to that, the above two authors carefully read the full length of articles that passed the initial screening and decided independently articles suitable for inclusion in the final meta-analysis. Any disagreement between them regarding eligibility criteria was resolved by agreement and with a third reviewer (WY).

Data Extraction and Quality Assessment Techniques
The extraction of relevant data from the 40 final included articles was separately done by two authors (MN and YZ) using an identical data extraction form as suggested by PRISMA guidelines (56), using Meta-XL version 5.3 (57) and the result was summarized in a table. Disagreements among these two authors were settled with a discussion. The contents of the data extraction template were author name, year of publication, the country where the study was done, study design, study sample population, an assessment tool for HIV associated neurocognitive disorders, number of cases with HIV associated neurocognitive disorders, prevalence of HIV associated neurocognitive disorders, sampling technique employed to recruit participants, and response rate of the study.
The quality of 40 included studies (8,10,(12)(13)(14) had been evaluated using the modified Newcastle-Ottawa Scale (NOS) (58) as the gold standard. Representativeness of sample and sample size, statistical quality, comparability among participants, and ascertainment of cases were the components of this quality assessment scale. Based on this scale studies with a quality score of 7 to 10 were categorized as very good/good, a score of 5 to 6 was categorized as having satisfactory quality, and a score less than 5 was taken as unsatisfactory quality.

Data Analysis and Synthesis
The random-effect model was used to compute the average estimate of HIV-associated neurocognitive disorders and their associated factors with 95% CIs (59). The STATA-16 Meta-prop package (60) was employed to find the average estimate of HIVassociated neurocognitive disorders. Heterogeneity among the 40 involved studies (8,10,(12)(13)(14) was weighed with Q and I 2 statistics (61). An I 2 numerical value of more than 50% implies a significant degree of heterogeneity among 40 studies (61). As there existed a potential heterogeneity during analysis, we further conducted a sensitivity analysis to identify an influential study outweighing the study found. Additionally, we did a subgroup analysis regarding the country of the study, study design, and the assessment tools used to screen HIV-associated neurocognitive disorders. The presence/absence of a publication bias was done with the funnel plot test (62) and eggers test of publication bias.

Identification of Studies
Our electronic search gave a total of 10231 articles; Psych-Info (386), EMBASE (2,452), Scopus (1,594), and PubMed (5,799). Additionally, 12 articles were retrieved by looking for a reference list of earlier articles. Thus, a total of 10,243 articles were retrieved during the overall searching process, of which 39 were removed as they were duplicates. During the initial stage of screening, most of the articles (10,118) were excluded merely by looking at their title or abstract. The remaining 86 articles were completely inspected for suitability for inclusion in the study but only 40 articles were suited for the final meta-analysis as the remaining 46 studies were excluded; 10 reported an incomplete prevalence rate for HAND, 32 had a poor methodological evaluation of HAND, and 4 were letters to the editor (Figure 1).

Characteristics of Included Studies
A total of 40 studies (8, 10, 12-14, 17-50) that surveyed HIVassociated neurocognitive disorders in 14,107 HIV/AIDS patients were integrated into the current systematic review and metaanalysis study. Of the 40 included studies; 11 were from Europe, 21 were from Africa, six were from Asia, and two were from the United States of America (USA). Most of the included studies were cross-sectional in design whereas the remaining 10 and 2 were cohort and case-control, respectively. Regarding tools used for the assessment of HIV-associated neurocognitive disorders, half of the included studies (twenty) used the International HIV Dementia Scale (IHDS). Frascati criteria, global dementia scales (GDS), and Montreal Cognitive Assessment (MoCA) were also used to assess HIV-associated neurocognitive disorders in three, three, and three studies, respectively. The reported prevalence of HIV-associated neurocognitive disorders included in the metaanalysis differs from 7.3% in the United Kingdom (28) to 88% in Kenya (35) ( Table 1).

Quality of Included Studies
Among the 40 included studies; the majority (twenty-nine) had scored from 7 to 10 so good quality scores on the scale. Of the remaining 11 studies, seven had a satisfactory quality, and the remaining four studies had unsatisfactory quality (Additional File 1).

The Prevalence of HIV-Associated Neurocognitive Disorders Among People Living With HIV/AIDS
Forty studies that evaluated HIV-associated neurocognitive disorders in HIV/AIDS had been included to determine the average prevalence of HIV-associated neurocognitive disorders.

Subgroup Analysis of the Prevalence of HIV Associated Neurocognitive Disorders Among People Living With HIV/AIDS
Since the average estimate of HIV-associated neurocognitive disorders was predisposed to considerable heterogeneity, we employed a subgroup analysis based on the country where   (Figure 3 and Table 2

Sensitivity Analysis
In addition to subgroup analysis, we did a sensitivity analysis to know whether one or more of the individual  that there was no single influential study outweighing the average estimate.

Publication Bias
Although a graphical inspection from a funnel plot for a Logit event rate of occurrence of HIV-associated neurocognitive disorders in people living with HIV/AIDS alongside its standard error suggests asymmetrical distribution, the quantitative Eggers test of publication bias had been run and its p-value was not significant; (P = 0.55). This suggests there was no publication bias for the prevalence HIV associated neurocognitive disorders (Figure 6).  (49), having a comorbid opportunistic infection (20) and psychological variables like negative life events, high-stress score index (score>10) (10) were related to HIV associated neurocognitive disorders ( Table 3).

Association Between Old Age and HIV Associated Neurocognitive Disorders Among People Living With HIV/AIDS
Older age was reported as the risk factor for HIV-associated neurocognitive disorders by five studies (8,14,20,46,49). The pooled odds ratio for the association between old age and HIVassociated neurocognitive disorder among these five studies was found to be 3.68 (95% CI: 2.95, 4.11) (I 2 = 98.2%%, P = 0.000; Figure 7).

Association Between Depression and HIV Associated Neurocognitive Disorders Among People Living With HIV/AIDS
As reported in three studies (14,21,35) that assessed HIVassociated neurocognitive disorders, depression increases the risk of HIV-associated neurocognitive disorders. The pooled odds ratio for the association between depression and HIV-associated neurocognitive disorder among these three studies was found to be 2.87 (95% CI: 0.87, 4.87) (I 2 = 99.6%%, P = 0.000).

Association Between Advanced Stages of AIDS and HIV Associated Neurocognitive Disorders Among People Living With HIV/AIDS
Advanced clinical stages of the illness (stage III and stage IV AIDS) (20,21,25,49) were also associated factors for HIV-associated neurocognitive disorders. The pooled  odds ratio for the association between advanced stage of AIDs(stage III& IV) and HIV-associated neurocognitive disorder among the four included studies was found to be 5.68 (95% CI: 3.06, 8.29) (I 2 = 99.9%%, P< 0.001; Figure 8).

Meta-Regression
Firstly, we performed a univariate regression analysis to select the independent variables to incorporate in the final metaregression model. Then, all variables with P < 0.8 were included in the final regression analysis as recommended by Ferrari et al. 2013 (67). The impacts of the country of origin of the study (Europe, Asia, and Africa), Assessment Tool (IHDS < 9.5, IHDS < 10, Frascati criteria, ADC, MMSE, GDS > 0.5, MoCA, and Other Tools), and study design (case-control, cohort and cross-sectional) were quantified in the meta-regression model. The analysis was conducted for the overall effect on the burden of HIV-associated neurocognitive disorders in HIV/AIDS patients. The overall proportion of variance explained by these covariates in the final model was 11% (R 2 = 11%; P = 0.302). All the three covariates such as country of origin of the study, Assessment Tool, and study design were not statistically significant determinants for the observed variation in the association between HIV-associated neurocognitive disorders and HIV/AIDS.

DISCUSSION
To our knowledge, this is the first systematic review and meta-analysis that assessed the global burden of HIVassociated neurocognitive disorders in HIV/AIDS patients. So, the data synthesized will be an important suggestion to varied stakeholders. Overall, 40 studies (8,10,(12)(13)(14) that measured the prevalence of HIV associated neurocognitive disorders in 14,107 participants from over 28 different countries and 15 studies that described the factors related HIV associated neurocognitive disorders (8,10,12,14,18,20,21,25,26,34,35,(46)(47)(48)(49) were included. The result of the present study advocated that more than half of people living with HIV/AIDS were affected with HIVassociated neurocognitive disorders. This suggests that HIVassociated neurocognitive disorders are an important public health problem in people living with HIV/AIDS. The result of HIV-associated neurocognitive disorders is consistent with the prevalence of neurocognitive disorders in patients with chronic kidney disease and diabetes; 48% In the UK (68). However, it was higher than the magnitude of neurocognitive disorders in the general urban population in India (1.5%) (69), and the prevalence of dementia in an urban Turkish population (20%) (70). It was also higher than the prevalence of dementia in Norwegian elders of age 75 and above where dementia was prevalent in 16.3% of the population (71). Besides, the present finding was higher than the prevalence rate of dementia among japans community aged 65 or older; 6.7% (72). The added presence of opportunistic central nervous disease (73)(74)(75) in HIV-positive people could be attributing factor to this.
The current study result was higher than the result of a metaanalysis that assessed 16 studies in sub-Sahara Africa where the prevalence of HAND was 30.39% (16). The difference in the study population and significant variance in the number of included studies might cause the variation. Moreover, the risk factors for HIV-associated neurocognitive disorders might be different in Sub-Saharan Africa and the world in general, which may cause the difference in HIV-associated neurocognitive disorders. On the other hand, the current study was higher than the result of the prevalence of Frascati-Criteria-Based HIV-Associated Neurocognitive Disorder; 43.9% (76). The possible reason for this could be due to the difference in sensitivity and specificity between Frascati-Criteria and other tools to screen HIV-Associated Neurocognitive Disorder.
In a subgroup analysis by type of measurement instrument used for HAND, we found that the prevalence was higher as measured with IHDS < 10 (8, 10, 12-14, 17, 20, 25, 35-37, 39, 43, 44, 46, 48-50, 66) (59.956%) than the result measured with Frascati criteria (53.5%), ADC (27.65%), MMSE (41.349%), and GDS (40.766%) with all these differences at statistically significant level(P < 0.001). The possible differences in the estimated magnitude of HIV associated neurocognitive disorders among measurement instruments will be attributed to the difference in the ability of measurement tools to identify truly those HIV patients who have HIV associated neurocognitive disorders (sensitivity) and to exclude HIV patients who do not have HIV associated neurocognitive disorders (specificity) (33,(77)(78)(79)(80)(81)(82). In addition, the lack of country-specific NP norms for the assessment of HAND would have contributed to this variation.
Even though we conducted a subgroup analysis with a country of origin of the study (categorized as Africa, Europe, Asia, and the USA) as a moderator variable, we found a more or less consistent average estimate of HAND; Europe (50.015%), Africa (49.566%), Asia (52.032%), and United States of America (USA)   Frontiers in Psychiatry | www.frontiersin.org (50.407%) with the difference of prevalence estimate of HAND being insignificant. Concerning the associated factors for HIV-associated neurocognitive disorders, our systematic review suggested that older age was among the commonly reported risk factor for HIV-associated neurocognitive disorders (8,14,20,46,49). Older age group people were 3.68 times more likely to develop HIV-associated neurocognitive disorder than younger age people. This is supported by multiple earlier studies (83). The accumulation of excessive amyloid deposits and various chronic diseases in old age could account for this (84).
Besides, depression also increases the risk of HIV-associated neurocognitive disorders as reported in three studies (14,21,35). The HIV-associated neurocognitive disorder was 2.87 times higher among individuals having comorbidity of depression than those not having comorbidity of depression. Multiple supportive findings for this association are reported (85)(86)(87). Pseudodementia associated with depression could mimic neurocognitive disorders and amplify the prevalence of neurocognitive disorders.
Moreover, advanced clinical stages of the illness (stage III and stage IV AIDS) (20,21,25,49) were also associated factors for HIV-associated neurocognitive disorders. Advanced clinical stages of the illness increase the risk of HIV-associated neurocognitive disorder by 5.68.

Difference Between Studies Incorporated in the Current Review and the Meta-Analysis Study
This meta-analysis study on HIV-associated neurocognitive disorders was subjected to a high degree of heterogeneity from the variation between 40 of the studies integrated with the final analysis. Therefore, the need for a subgroup and sensitivity analysis was mandatory. The country of origin of the study (Africa, Europe, Asia, and the USA), design of the study, and measurement tool were the parameters for subgroup analysis. However, no major difference in the pooled prevalence of HIV-associated neurocognitive disorders was observed between high-income/(Europe, Asia, and the USA) and low-to-middle income/African countries. These should be a focus of future researchers. Results of the subgroup however revealed the design of the study and measurement tool used to screen HIVassociated neurocognitive disorders were contributors to the high heterogeneity between the 40 studies integrated with the analysis. Furthermore, we did a sensitivity analysis to search for additional sources for the high degree of heterogeneity by exploring a particular study outweighing the overall estimate. However, the result showed that none of the 40 studies had outweighed the average estimated prevalence of HIV-associated neurocognitive disorders in HIV patients.

STRENGTH AND LIMITATIONS OF THE STUDY
This systematic review and meta-analysis on HIV-associated neurocognitive disorders have several strengths. The primary strength of the study was the large number of studies included in the review, representing the worldwide spread of HAND in people with HIV. The application of subgroup analysis and sensitivity analysis to identify the source of heterogeneity was another quality of the present study. The study also used a well-designed and pre-determined search approach to lessen the reviewer's bias. The third for this study was the extraction of relevant data and assessment of the quality of 40 studies by self-determining assessors which decrease the reviewer's bias. Quality analysis of all the included studies also increases the study validity. On the other hand, the study has many limitations. First, the definition of HIV-associated neurocognitive disorder could be more explicit. In addition, the reasons for exclusions of 46 identified full texts are not explicitly mentioned in the PRISMA flowchart. The presence of a high degree of heterogeneity that influences the inference of the study outcomes was also another limitation. Moreover, biases and lack of methodological rigor could affect the validity of this finding.

CONCLUSION AND RECOMMENDATION
This systematic review and meta-analysis aimed to estimate the prevalence of HIV-associated neurocognitive disorder in people living with HIV AIDS and to analyze the associated factors for HIV-associated neurocognitive disorder in people with HIV AIDS. The prevalence of HIV-associated neurocognitive disorders was about 50.41%. Low level of education and older age, clinical and HIV related variables such as the advanced stage of the illness and CD4 count of 500 cells/dl or less and psychological variables such as comorbidity of depression were associated with HIV associated neurocognitive disorders. Therefore, to increase independent functioning and improve the quality of life of people with HIV/AIDS, much attention has to be given to lessening these neurocognitive disorders and adjusting the allied factors essentially through routine screening and timely intervention of HAND. Moreover, policies and procedures that integrate routine screening and timely intervention of HAND into the routine anti-retroviral therapy should be designed and implemented. Further experimental and follow-up studies with a greater sample population should be done.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
YZ conceived the idea for the study. YZ and MN established the search approach, extracted the relevant data, accomplished the analysis, and wrote the manuscript. MN, YZ, BA, and WY did the quality assessment studies. All authors confirmed the final draft of the manuscript.