Symptoms of common mental disorder and cognitive associations with seropositivity among a cohort of people coming for testing for HIV/AIDS in Goa, India: a cross-sectional survey

High levels of common mental disorder have been identified among people living with HIV/AIDS in communities around the world [1–5], including in India [6]. There is evidence from low and middle income and high income settings that Common Mental Disorders (CMD) such as depression have an adverse impact upon adherence [7–9], as well as additional negative effects upon HIV-related clinical outcomes such as disease progression and survival, that are independent of the effects of common mental disorder upon adherence [10, 11]. Neurocognitive impairment usually thought to be due to HIV-related damage to the frontostriatal region of the brain is generally found to occur among 20–37 percent of people living with HIV/AIDS [12] and is associated with HIV-associated dementia, disease severity [13], mortality [14], adherence to antiretroviral therapy [15] and employment [16]. The precise mechanisms for the high prevalence of comorbid neurocognitive impairment with HIV/AIDS are not well understood but recent studies have found reduced subcortical volumes and metabolite abnormalities among HIV-positive participants on cART [17].

Although diagnosis and life events related to diagnosis and disclosure undoubtedly contribute to the poor mental health of people living with HIV/AIDS [18, 19], there is evidence to suggest that the presence of HIV in the brain may also play a role in producing depression-like symptoms [20]. The majority of research on HIV/AIDS and mental health has been carried out among clinical populations: the time of onset of comorbid depression and the mechanisms for this are therefore unclear. Although there is evidence to suggest that asymptomatic people living with HIV/AIDS exhibit some cognitive deficits [21], the prevalence of poor cognitive functioning among people in low income settings at an early, pre-clinical stage has not yet been investigated.

The few researchers who have measured depression among people coming for testing in low income settings have generally found a high prevalence (45 percent in an Indian sample [6]; 41 percent among pregnant women in South Africa [22]). Depression was not found to be associated with testing positive for HIV/AIDS in either these samples.

The limited data available on HIV-associated cognitive impairment in low income settings generally indicates lower cognitive functioning among seropositive participants as compared to seronegative controls [12]. Prevalence estimates of impairment vary- a multicentre WHO survey found impairment among symptomatic individuals of between 13 and 19 percent [21], whereas recent studies carried out in India, China, Uganda and Botswana comparing HIV-positive with HIV-negative controls reported prevalence of between 31 and 56 percent [12, 23].

The aim of the Umeed study was to investigate the relationship between CMD symptoms cognitive functioning and HIV/AIDS among a cohort of people attending a public clinic for HIV-testing. At the time of interview, study participants were not aware of their HIV status. Given the evidence to support a biological pathway between HIV infection and common mental disorder and cognitive impairment, we hypothesised that CMD symptoms and cognitive impairment would predict testing positive for HIV/AIDS among a sample that were not yet aware of the outcome of their test.

The Umeed cohort is the largest study to describe the mental health of people coming for testing for HIV/AIDS in a low income setting. The inclusion of seropositive and seronegative attendees of HIV testing provided us with a unique opportunity to test hypotheses about mental health and HIV status in the absence of the impact of diagnosis and consequent experiences of living with HIV/AIDS upon participant’s mental health. We hypothesised that CMD symptoms and cognitive impairment would be associated with testing positive. Our findings were inconclusive. Although CMD symptoms was strongly associated with testing positive in bivariate analyses, after adjustment for confounders, the effect size of this association was reduced and no longer statistically significant. We found some evidence to support the hypothesised association between low cognitive functioning and seropositivity. Although there was only a bivariate trend towards an association between delayed recall and seropositivity, scoring below education-adjusted norms on the test of verbal fluency (animal naming) was associated with testing positive for HIV/AIDS (AOR = 1.77, 1.27-2.48).

Almost all of the large body of research about HIV/AIDS and CMD in low and middle income settings has been carried out among single groups, with a HIV-positive diagnosis, without comparison with seronegative controls (for example, [5, 38, 39]). A few studies have compared prevalence of depression between seropositive and seronegative groups (ie. [4, 40]). In one of the few studies similar in design to Umeed (a sample of test-seekers in Pune where 45.5 percent subsequently received a positive diagnosis), univariate analyses revealed no evidence of an association between common mental disorder and seropositivity [6]. In the Umeed sample, bivariate analysis revealed a strong association between CMD and testing HIV-positive (OR = 2.27, 1.31-3.92). However, after the gradual introduction of covariates into the model, the effect size was reduced and no longer statistically significant (AOR = 1.56, 0.86-2.85).

Although the multivariate analysis presented in Table 3 demonstrates that the association between CMD symptoms and seropositivity was partially confounded by sociodemographic and HIV-related covariates, we should be cautious before interpreting these findings as a demonstration that there was no true association between CMD symptoms and seropositivity. As indicated by the relatively wide confidence intervals presented in Table 3, the low prevalence of both CMD and seropositivity among Umeed participants (and particularly in the sub-sample used for this analysis) resulted in reduced power to detect an association, therefore we cannot rule out the existence of a true association between CMD and seropositivity.

As described, the prevalence of CMD symptoms identified among the Umeed sample was much lower than that reported among the two other samples of people coming for testing in low and middle income settings described in the literature. The relatively low prevalence of HIV/AIDS measured in Umeed as compared to these other samples (45.5% in India; 40.9% among pregnant women in South Africa) may help to explain the difference in prevalence of CMD symptoms. The Umeed sample is heterogeneous and includes participants at high and low-risk of testing positive. Therefore, it is plausible that the overall prevalence in the Umeed sample may be more similar to that found in the local community [41], rather than that identified in primary care [42]. Without further research among people coming for testing in similar settings, it is difficult to contextualise our findings further. Of course, the possibility that our mode of measurement of CMD symptoms may have contributed to the low reported prevalence of CMD symptoms cannot be ruled out. Although the PHQ-9 is commonly used in Western settings, the lack of validation of cut-points for use among HIV-affected populations in low income settings may have had an impact upon the low prevalence of common mental disorder identified in the Umeed cohort. As described, all interviews were carried out in temporary, private cabins. All participants were given detailed information about CMD symptoms and the aims and objectives of the study; the PHQ has been found to be valid among a Goan sample of people attending primary care. However, the possibility remains that using the PHQ-9 immediately prior to testing in the busy clinic setting may have inhibited participants willingness to disclose potentially sensitive information about symptoms of anxiety and depression. Our analysis demonstrated a high degree of co-morbidity of symptoms of anxiety and depression within our sample (of those scoring >10 on the GAD-7, 75 percent scored >10 on the PHQ-9; of those scoring >10 on the PHQ-9, 20 percent scored >10 on the GAD-7 and 47.4 percent scored between 5 and 10 on the GAD-7), thus supporting the dichotomous “symptoms of CMD” variable used in our analysis.

Given that HIV/AIDS is associated with morbidity, stigma and discrimination [43, 44], poverty [45] and mortality [46, 47], it is reasonable to expect that living with the disease contributes greatly to the prevalence of CMD among people with HIV/AIDS. However, there is evidence from neurological research to suggest that the inflammatory response that contributes to the neuronal loss that causes HIV-associated cognitive impairment may also trigger cytokine-induced depression [20]. Recent research suggests that even in the absence of other symptoms of depression and cognitive impairment, HIV-related damage to white matter may be associated with symptoms of apathy (“reduced self-initiated cognitive, emotional and behavioural activity”) [48]. In addition to social and environmental risk factors associated with their disease status, people living with HIV/AIDS (and other chronic conditions) may have a biological vulnerability to the development of depressive symptoms [46]. Although evidence suggesting that CMD may be associated with risky behaviour [47, 49] means that a reverse relationship (pre-existing depression promoting sexual risk-taking and acquisition of HIV) cannot be ruled out, presence of genital warts and perception of HIV-risk were the only sexual risk-related variables, that confounded the association between CMD symptoms and seropositivity- therefore this pathway seems unlikely.

The Umeed finding that low scores on the test of verbal fluency were associated with seropositivity corresponds with the results of a recent meta-analysis that revealed a statistically significant deficit in mean scores on tests of verbal fluency among people living with HIV/AIDS, as compared to HIV-negative controls (−0.31, −0.44 to −0.18) [50]. In the only study to compare verbal fluency between seropositive individuals and seronegative controls in India, although HIV-positive individuals differed in other language-related domains (phonemic fluency, verbal working memory and verbal learning and memory); differences in animal naming fluency were not statistically significant [51]. Similarly, statistically significant deficits in verbal learning and memory were identified between seropositive individuals and seronegative controls in a previous south Indian study [52].

Neurological evidence supports the presence of cognitive deficits early in the course of the disease [12]. Both cognitive domains measured in Umeed (category fluency and verbal learning and memory) are consistent with diffuse or frontostriatal pathology, both of which may be characteristic of the effects of HIV-1 on the brain [50]. There is now a growing body of evidence on the cognitive profile of people living with HIV/AIDS in low income settings, including India [52]. Whereas small studies with detailed neuropsychological assessments tend to have stringent exclusion criteria which may limit the generalisability of findings, in Umeed, exclusion criteria were limited, meaning that our sample was likely to be more representative of the base population. The aim of Umeed was to measure cognitive functioning in the selected domains most likely to have an impact upon linkage to care and adherence and to examine whether these were associated with testing positive. The animal naming test of category fluency and the word list learning test of verbal learning and memory were selected as appropriate measures that would be relatively quick and simple for trained research assistants to administer among a large cohort of participants in a busy clinic setting.

The Umeed study design had several other methodological strengths. Unlike other studies that have investigated mental health and HIV/AIDS in a low income setting, the Umeed study was carried out among a large sample that included both seropositive and seronegative individuals recruited prior to diagnosis. Thus we were able to test hypotheses that CMD symptoms, cognitive functioning would be associated with seropositivity independently of the effects of diagnosis. Blind recruitment of seropositive and seronegative participants from the same clinic helped to ensure comparability of the two groups and reduced risk of selection bias. There was evidence from previous research to support the validity of measures of CMD and cognitive functioning among Goan populations.

However, there were also limitations. Due to ethical considerations, those who were potentially most unwell: (ie. unable to participate in 30 minute interview without experiencing undue distress or discomfort or lacking capacity to understand aims, participation demands and give informed, voluntary consent) were excluded from our sample (n = 102). Therefore, estimates for CMD symptoms, low cognitive functioning and seropositivity are likely to be lower than the true prevalence among people coming for testing for HIV/AIDS in Goa Medical College. Subsequently, this may have contributed to smaller effect sizes for associations between mental health exposures and HIV status.

It is possible that social desirability bias may have had an impact upon our measures of risk-taking behaviours. Reported risk-taking behaviours were fairly low relative to prevalence of HIV: 0.1% reported injecting drugs, 0.8% of men reported having had sex with other men, 1.9% reported having received money or gifts in exchange for sex; 5% reported having sex with a sex worker and 12.6% reported having sex with someone other than a regular partner in the last year. This may have limited our ability to control for the confounding effects of risk-taking behaviours upon observed associations between common mental disorder/low cognitive functioning and HIV status.

There were other limitations to our approach to cognitive testing. The cognitive screening instruments used in Umeed were chosen because they measured the domains most likely to have an impact upon access to care. The education-adjusted norms for cognitive functioning have not been validated among the younger Goan population or in the context of HIV/AIDS. We were unable to discriminate between sub-cortical deficits commonly caused by HIV and cognitive deficits with other causes (educational deprivation, mental sub-normality, language issues etc.), Unmeasured confounding is a possibility. Reverse causality cannot be ruled out. Low cognitive test scores may have been caused by a factor other than HIV that was associated with HIV-risk and acquisition. There is evidence to support a link between heavy alcohol use and poor cognitive functioning [53], as well as an association between alcohol-related impairment and reduced perception of sexual-risk [54]. However, given that we found little evidence of any association between alcohol use and HIV status, it seems unlikely that this pathway explains the observed relationship between low cognitive functioning and testing positive for HIV/AIDS.

Lack of detection of an association between delayed recall and seropositivity may be due to error in the measurement of delayed recall in Umeed. Although every effort was made to carry out interviews in an environment conducive to cognitive testing, it is possible that noise and interruptions to interviews had an impact, particularly on measurement of delayed recall, the more complex of the two cognitive tests. Measurement error may have led to some random misclassification of delayed recall test results, which may have contributed to masking a true association between delayed recall test results and testing positive.

Umeed study findings provide tentative evidence that prior to knowledge of HIV status, low cognitive test scores (and perhaps CMD), may be associated with seropositivity among people coming for HIV-testing. Our findings suggest that a biological mechanism may contribute to the high prevalence of CMD and poor cognitive functioning commonly found among people living with HIV/AIDS. Impaired cognitive functioning and depression-like symptoms may be the result of the same underlying neurological damage caused by the presence of HIV in the brain. CMD and cognitive impairment may overlap to a greater extent than previously assumed. In order to clarify the relationship between CMD, cognitive impairment and HIV/AIDS, further research among pre-diagnosis samples is necessary. It would be helpful to include detailed assessments of key cognitive domains. If a closer interrelationship between cognitive functioning and CMD among people living with HIV/AIDS is demonstrated, this may have implications for the way in which we measure and treat CMD and cognitive functioning among people living with HIV/AIDS. It will be important to explore the possibility of early behavioural markers (such as apathy) as predictors of later cognitive decline. Depending on the outcome of future research, routine integration of screening for both CMD and cognitive deficits within Voluntary Counselling and Testing should be considered.