Keywords

1 Screening for HAND: Needs Versus Practice

The neurocognitive impairment (NCI) related to HIV-associated neurocognitive disorder (HAND) is one of the most common sequelae and comorbid conditions of HIV infection and has significant medical, functional, and public health consequences (Hinkin et al. 2002, 2004; Heaton et al. 2004; Marcotte et al. 2004; Van Gorp et al. 2007; Gorman et al. 2009; Ettenhofer et al. 2009, 2010; Vivithanaporn et al. 2010; Schouten et al. 2011; Umaki et al. 2013). NCI can be caused by HIV infection of the central nervous system (CNS; resulting in neuropathology of the basal ganglia and white matter) or by other comorbid or preexisting conditions (e.g., chronic substance abuse, head trauma, CNS opportunistic infections) and/or a combination of both (Grant 2008; Heaton et al. 2010, 2011; Grant and Sacktor 2012). NCI in HIV typically causes impairments in mental processing speed, learning, memory, attention and concentration, higher-order executive functions, and motor speed (Grant 2008; Heaton et al. 2011; Grant and Sacktor 2012). Collectively known as HAND, HIV-related NCI ranges in severity from its mild forms, asymptomatic neurocognitive impairment (ANI) and mild neurocognitive disorder (MND), to its most severe form: HIV-associated dementia (HAD) (Antinori et al. 2007).

Diagnosing HAND requires neuropsychological testing across the domains of language, attention/working memory, abstraction/executive, memory [learning and recall], speed of information processing, and sensory-perceptual, motor skills (Antinori et al. 2007). Specifically, an ANI diagnosis requires acquired NCI in at least two domains at least one standard deviation below the mean for age- and education-adjusted norms on the neuropsychological tests. Furthermore, the NCI does not interfere with everyday functioning. A diagnosis of MND requires the same acquired NCI in at least two domains at least one standard deviation below the normative mean as ANI but causes mild interference in daily functioning by self-report and/or observation by knowledgeable others. Diagnosis of HAD requires marked acquired NCI in at least two domains at least two standard deviations below the normative mean with the NCI causing marked interference with everyday functioning. For ANI, MND, and HAD, the NCI does not meet criteria for delirium (or dementia for ANI and MND) and cannot be accounted for by a preexisting condition.

The mild forms (ANI and MND) are by far the most common and even occur in people living with HIV (PLWH) on antiretroviral therapy (ART) with well-controlled viremia; prevalence ranges from 22 to 70% (Becker et al. 2004; Simioni et al. 2010; Heaton et al. 2010; Grant and Sacktor 2012; Bonnet et al. 2013; NNTC Public Reports 2014) depending on the study, participant characteristics, and method of assessing neurocognitive and functional status. One study, with over 1,500 PLWH, estimated rates of HAND at about 50% – excluding those with other comorbidities that could better explain the NCI present (Heaton et al. 2010). Prevalence of HAD is much lower, from 2 to 9% (Becker et al. 2004; Simioni et al. 2010; Heaton et al. 2010, 2011; Grant and Sacktor 2012; Bonnet et al. 2013). Compared to younger PLWH, those 50 years of age and over have been shown to have higher rates of NCI and are at greater risk for developing it (Becker et al. 2004; Valcour et al. 2004; High et al. 2006; Ettenhofer et al. 2009; Mateen and Mills 2012). Prevalence of NCI among PLWH in the presence of multiple confounders ranges from 7% to as high as 33%, depending on the sample (Heaton et al. 2010; NNTC Public Reports 2014). Less is known about the prevalence of HAND in children, adolescents, and adults who acquired HIV at birth (Hoare et al. 2016). Having even mild (or asymptomatic) HAND has been associated with increased risk for developing more severe HAND and mortality (Vivithanaporn et al. 2010). Research has also established a strong relationship between NCI and worse ART adherence (Hinkin et al. 2002, 2004; Ettenhofer et al. 2009, 2010), thus jeopardizing positive health outcomes. NCI in HIV is also associated with work difficulties, impaired activities of daily living, (e.g., planning, driving, finance management), worse overall quality of life, and need for more social services (Heaton et al. 2004; Marcotte et al. 2004; Van Gorp et al. 2007; Gorman et al. 2009; Umaki et al. 2013). Finally, neurocognition deficits are associated with poor decision-making and greater HIV transmission risk behaviors (e.g., unprotected sex; Wardle et al. 2010; Thames et al. 2012; Iudicello et al. 2013).

Routine screening for HAND – especially those newly diagnosed with HIV – has been recommended as good clinical practice (Valcour et al. 2011; Cysique et al. 2012; The Mind Exchange Working Group et al. 2012; Haddow et al. 2013; Morley et al. 2013; Zipursky et al. 2013; Barber et al. 2014; Kim et al. 2014). As a first step in the clinical decision-making process, screening can enable providers to determine who is most likely to have HAND, detect early signs of HAND, allocate limited resources more effectively, track and monitor neurocognitive function, and educate patients about the impact of HAND and ways to minimize it – all of which can improve health outcomes (Cysique et al. 2012; The Mind Exchange Working Group et al. 2012). Once adjuvant behavioral therapy and/or pharmacotherapy become available, screening for HAND will assist providers in appropriate treatment referrals. However, it rarely occurs (Valcour et al. 2011; Haddow et al. 2013; Morley et al. 2013; Kim et al. 2014).

Globally, there are 36.7 million people living with HIV (Joint United Nations Programme on HIV/AIDS 2016), and millions have and are at risk for developing HAND. Hence, screening for HAND is critical. Even as greater numbers of PLWH gain access to effective ART and live longer, healthier lives, HAND persists (Becker et al. 2004; Simioni et al. 2010; Heaton et al. 2010; Grant and Sacktor 2012; Bonnet et al. 2013). Furthermore, as PLWH grow older, they are at a higher risk than the general population of developing dementia. Screening for and detecting HAND early can play a critical role in the lives of PLWH through understanding its impact on health and health outcomes and developing strategies to optimally manage it.

There are numerous reasons why screening does not routinely occur in clinical practice. Most screening tests developed specifically for HAND were either designed to detect only the most severe form (HAD) or lack accuracy to detect the more common milder forms to be clinically useful (i.e., maximizing true positives and minimizing false positives; Power et al. 1995; Smith et al. 2003; Sacktor et al. 2005; Bottiggi et al. 2007; Valcour et al. 2011; Muñoz-Moreno et al. 2013; Zipursky et al. 2013). Many of the currently available, conventional, paper-and-pencil screening tests and short batteries for HAND require specialized skills and knowledge to properly administer, score, and interpret them (Valcour et al. 2011; Zipursky et al. 2013). Some screeners are cumbersome, requiring additional proprietary test forms and expensive equipment (Valcour et al. 2011; Zipursky et al. 2013). Furthermore, conventional, paper-and-pencil screening instruments are prone to human error in administration and scoring, nor are they well-suited for easy integration with electronic medical records (test data must be manually entered into electronic systems – a task also prone to error). Two studies have demonstrated that computer-based neurocognitive testing may be useful to detect mild HAND; however, it requires expensive laptop or desktop computers plus proprietary software, and more research is needed to determine its accuracy in detecting HAND, whether it is feasible to use in the clinical setting and the facilitators and barriers to its incorporation into clinical practice (Becker et al. 2011; Overton et al. 2011).

2 Screening Tests for HAND

Numerous screening tests have been developed for and/or evaluated to detect HAND. Two recent review articles by Kamminga et al. (2013) and Zipursky et al. (2013) summarized results across a total of 36 different studies that examined 40 different screening tests, subtests, or short batteries of individual neuropsychological tests as screening instruments for HAND. We have presented the tables with permission from Kamminga and Zipursky (see Tables in the Appendix to this chapter). Most of the studies (N = 20) were conducted in the United States, with three from South Africa, three from Australia, one from an AIDS Clinical Trials Group study from multiple countries, and the rest from Europe, Asia, and sub-Saharan Africa. The most widely reviewed screening tests were the HIV Dementia Scale (HDS) (Power et al. 1995) and the International HIV Dementia Scale (IHDS) (Sacktor et al. 2005). Results greatly varied between studies with some evidence to suggest some individual screening tests, such as the HDS and IHDS, may be adequate to detect only HAD in some settings (Haddow et al. 2013), and combinations of individual tests may be useful in the detection of more mild forms of HAND (Kamminga et al. 2013, 2017; Monteiro de Almeida et al. 2017). For example, one study found that different pairs of six different neuropsychological measures (i.e., WAIS-III Digit Symbol, HVLT-R Total Recall, PASAT-50, and Grooved Pegboard, and Trail Making Test Part B) were more accurate than the HDS in classifying NCI in HIV+ individuals (Carey et al. 2004). One additional scoping review of screening tests for HAND in sub-Saharan Africa is not reviewed or presented here due to it being an open-access and open peer review study with only two reviews, one of which requires revisions (the study can be viewed at https://aasopenresearch.org/articles/1-28/v1).

Since the Kamminga et al. (2013) and Zipursky et al. (2013) articles, 23 additional studies from the past 5 years presented data on 18 different screening tests for HAND and/or the NCI associated with it, several of which were novel computer or mobile device-based screeners (see Table 1). One study published before the Kamminga et al. and Zipursky et al. reviews but not mentioned in those reviews is also included here (Kvalsund et al. 2009). Nine of the studies were conducted solely in the United States. One study was conducted in both the United States and South Africa. Other countries represented across the studies are Australia, Brazil, Malaysia, the Netherlands, Nigeria, South Korea, Uganda, the United Kingdom, and Zambia.

Table 1 HAND screening tests
Full size table

All but one of the studies were of adults living with HIV with a mean age across adult studies of about 46 years. Three studies had participants with mean ages of 50 years or older. One study had perinatally HIV-infected children and adolescents between 9 and 12 years of age (Phillips et al. 2019).

The Montreal Cognitive Assessment (MoCA; Nasreddine et al. 2005) was the most widely examined screening tool (included in 11 of the 23 studies), and it was used in several countries (e.g., South Africa, Zambia, the Netherlands, and the United States). The IHDS was examined in six of the studies and the HDS in three of the studies. One study included combinations of tests from a comprehensive neuropsychological test battery (Monteiro de Almeida et al. 2017).

Nineteen of the studies compared screening test results to a comprehensive neuropsychological test battery. About half of the studies (n = 13) did not include an assessment of functional ability and hence did not provide HAND diagnostics. Seven of the studies included HIV-uninfected comparison groups. Sample sizes ranged from a minimum of N = 39 to a maximum of N = 342. The mean sample size was 126 (SD = 83). Base rates for NCI and HAND (based on a comprehensive neuropsychological test battery [for NCI] and a functional assessment [for HAND]) ranged widely from 26 to 75%, depending on how NCI was classified (e.g., global domain scores vs. clinical ratings).

Most of the screening tests examined were reported to take between 10 and 20 min to complete. Ten studies did not report on who administered the screening tests. Of those that did, screening test administrators ranged from lay counselors (e.g., Robbins et al. 2018), to non-physician healthcare workers (e.g., Kvalsund et al. 2009), to trained research assistants, to neuropsychologists (e.g., Janssen et al. 2015), and to physicians (e.g., Hasbun et al. 2013).

Joska et al. (2016), comparing adults living with HIV in the United States and South Africa, found that assessing for HAD the cognitive assessment tool – Rapid (CAT-Rapid) demonstrated good sensitivity and weak specificity (94% and 52%; cutoff score ≤ 10), the IHDS showed fair sensitivity and good specificity (68% and 86%; cutoff score ≤ 10), and the MoCA showed excellent sensitivity but poor specificity (100% and 22%; cutoff score ≤ 26. The Mini-Mental Status Exam (Folstein et al. 1983) and Simioni symptom questionnaire (Simioni et al. 2010) did not demonstrate sufficient psychometric properties to detect HAND in the sample. None of the five tools were sufficient to adequately detect less severe forms of HAND.

Several international, single-country studies explored the ability of adapted versions of the IHDS and MoCA in Brazil, Nigeria, South Africa, and Korea to detect HAND (Royal et al. 2012; Ku et al. 2014; Monteiro de Almeida et al. 2017). Two of the studies concluded that the IHDS had enough sensitivity to be effective in detecting ANI and MND (Ku et al. 2014) and more generally HAND (Royal et al. 2012) in these populations, but it had poor specificity. In Brazil, Monteiro de Almeida et al. (2017) reported that a combination of several gold standard neuropsychological tests (see Table 1 for list of test combinations) was more accurate in detecting HAND than the IHDS. Ku et al. (2014) and Mukherjee et al. (2018) also evaluated the utility of the MoCA in detecting HAND in Korea and Malaysia, respectively. Both studies reported that the MoCA was effective at detecting HAND, with the Mukherjee et al. study concluding so only after accounting for demographic factors.

The computerized CogState Brief (Cysique et al. 2006; Maruff et al. 2009) was examined in Uganda and Australia (Bloch et al. 2016; Yechoor et al. 2017). A Ugandan study evaluated the CogState Brief in detecting HAND (Yechoor et al. 2017). In Uganda, the CogState had adequate specificity to detect HAND but poor sensitivity. Yechoor et al. concluded that additional research is required to identify tools with high sensitivity in detecting HAND in resource-limited settings. In Australia, Bloch et al. (2016) evaluated an updated CogState screener that specifically targets the NCI observed in HAND. The updated CogState had 76% sensitivity and 71% specificity to detect any HAND defined by a gold standard neuropsychological evaluation. CogState’s classification accuracy was increased to 100% sensitivity and 98% specificity when only MND and HAD were considered.

Two studies examined a mobile device-based screening tool, NeuroScreen (Robbins et al. 2014, 2018). The first study examined a large format smartphone version of NeuroScreen to detect NCI among adults living with HIV in the United States (Robbins et al. 2014) and found that sensitivity ranged from 89 to 94% and specificity ranged from 63 to 64%, depending on the method of NCI classification used. The second study examined the same screening tool using a tablet-based device administered by lay counselors to detect NCI among adults living with HIV in South Africa (Robbins et al. 2018). Sensitivity ranged from 82 to 93% and specificity from 75 to 81% depending on the combination of NeuroScreen subtests used.

A recent American study investigated whether a functional capacity measure, the UCSD Performance-Based Skills Assessment – Brief (UPSA-B), was able to detect functional dependence and neurocognitive impairment in adults in San Diego, California (Moore et al. 2017). This study found that the UPSA-B had sensitivity, specificity, and accuracy rates in 70%, but scores on this measure were unrelated to self-reported functional dependence.

Only one study examined a HAND screening tool for youth living with HIV. Phillips et al. (2019) examined an adapted for youth IHDS (y-IHDS) among perinatally HIV-infected children and adolescents between 9 and 12 years of age in South Africa. The y-IHDS had good sensitivity but poor specificity (94% and 24%, ≤10) for assessing HAND in children and adolescents. The authors concluded that while the y-IHDS has both clinical practice and research value in low-resource settings, further research to optimize the tool would be beneficial.

Finally, a recent meta-analysis of the MoCA’s accuracy to detect HAND (not included in Table 1) that includes several of the studies cited herein examined sensitivity and specificity across multiple cutoff scores (Rosca et al. 2019). The authors recommend using a modified cutoff score of ≤23 (versus the standard ≤25) to define impairment. Using this cutoff best balances true and false positives (see Sect. 2.2) and yields sensitivity of 44% and specificity of 79%.

2.1 Need for Functional Assessment

Screening tests for HAND focus on neurocognitive impairment. However, to fully diagnose HAND, a functional assessment is required. Only one of the reviewed screening tools for HAND included a performance-based functional assessment (see UPSA-B above), and none of the others included a functional assessment within the screening tool or procedure. Of the 11 studies with a functional assessment, the functional assessment was most often used in conjunction with the comprehensive neuropsychological test battery to define the base rate of HAND. In fact, there is a dearth of research on functional assessment tools for HAND across countries and contexts, and few tools to assess functional ability are available for those regions and countries most affected by HIV. Levels of everyday functioning vary greatly within and across countries depending on socioeconomic and cultural factors, for example, differences in gender roles across societies. These factors need to be accounted for when doing functional assessments in local and international settings. Furthermore, most of the commonly used assessment tools for everyday functioning, like Lawton and Brody’s Instrumental Activities of Daily Living (Lawton and Brody 1969), are self-report and often contain items that are not appropriate for certain contexts, like ability to manage finances via check writing or managing bank accounts. Many individuals in the most affected communities of sub-Saharan Africa do not have bank accounts, let alone use checks. Hence, this activity would need to be appropriately adapted to the context in which the measure is being used. Though self-report assessments do not provide an objective assessment of functional abilities, they can be useful to screen individuals, as many are fairly quick.

2.2 Screening Test Psychometrics

There are several important factors that need to be considered when choosing a screening test: (1) sensitivity and specificity; (2) positive and negative predictive values, and (3) base rates of the disease the screening test is trying to identify. Sensitivity refers to a test’s ability to correctly classify a patient as having the disease or disorder, whereas specificity refers to a test’s ability to correctly identify a patient as being disease or disorder free (Rosenfeld et al. 2000). A test with high sensitivity and low specificity will accurately detect those who have the disease (true positives) but also produce more false positives (those without the disease but who screen positive for it). Similarly, if the sensitivity is lower than the specificity, the test will be better at detecting those who do not have the disease (true negatives) than those who do. Positive predictive value (PPV) is the percentage of patients who test positive and who do in fact have the disorder. Higher PPV indicates that the screening test is more able to detect those patients who have the disorder. PPV is related to the prevalence of the disorder in the population and will increase as prevalence increases. Negative predictive value (NPV) is the percentage of patients with a negative test who do not have the disease. Higher NPV indicates that the screening tests are more accurately classifying those people who do not have the disorder as disorder free.

Base rates of the disease or disorder further complicate the issue of a test’s predictive ability. A screening test with 80% sensitivity and 70% specificity used in a population where the known base rate of HAND is 50% would result in a PPV of 73% and a NPV of 78%, which would result in a 73% probability that a person who screens positive actually has HAND. However, if the base rate of HAND is actually 30% in the population but the screening test was calibrated on a base rate of 50%, then the psychometric properties change dramatically such that PPV would be 53%, NPV would be 89%, and the probability that a positive screen is truly positive would be 53%.

Choosing and interpreting a screening test requires a thorough understanding of these psychometric properties. Few, if any, screening tests are perfect. Hence, clinicians and clinics must consider the ethics and costs of a screening test’s limitations. What are the consequences of missing a patient with the disorder (lower sensitivity/higher false-negative rate) and providing additional resources for someone who screens positive but does not truly have NCI (lower specificity/high false-positive rate)?

3 Global Perspective

The burden of HIV is in low- and middle-income countries (LMICs), particularly those in sub-Saharan Africa where, in some countries, the national prevalence of HIV is ~12% (Joint United Nations Programme on HIV/AIDS 2016). To date, most of the research on HAND screening has been focused on adult populations with HIV and populations in the United States. Given that the burden of the disease is disproportionally represented in sub-Saharan Africa (~27 million PLWH; Joint United Nations Programme on HIV/AIDS 2016), more studies are needed to evaluate properly adapted or locally developed screening tools in the most affected countries. Furthermore, there are approximately 2.1 million children under 15 who are living with HIV (Joint United Nations Programme on HIV/AIDS 2016), yet little attention has been paid to screening tools for them.

LMICs, such as South Africa, are facing massive resource shortages of healthcare professionals to test for and manage HIV, let alone screen and assess for HAND. For example, in the United States and Australia, there are 25.95 and 35.88 physicians per 10,000 population, whereas in South Africa there are only 9.01 physicians per 10,000 population, and in Zimbabwe there are only 0.76 physicians per 10,000 population (World Health Organization 2019). To address these shortages, many LMICs have been practicing task shifting where certain aspects of the HIV care continuum have been shifted to lay health professionals, such as community healthcare workers (Callaghan et al. 2010). Task shifting for HAND screening has been considered in South Africa. Unfortunately, the research has shown that certain HAND screening tools (i.e., IHDS) while developed to be used in any culture/country, when used by lay professionals, may highly over- or underestimate rates of HAD due to administration and interpretation errors (Robbins et al. 2011; Breuer et al. 2012).

Given the global distribution of HIV, considerations for the use of any screening test for HAND must take into account the psychometric validity of the test in the new population, the test’s cultural and demographic appropriateness, and who will be administrating the test. For example, the MoCA (Nasreddine et al. 2005) has test items (e.g., cube drawing) that are not well designed for some populations in South Africa and could erroneously indicate NCI (see Robbins et al. 2013).

Another challenge in LMICs is the issue of normative performance on cognitive tests. Many countries with the highest burden of HIV do not have formally validated and normed cognitive tests, let alone screening tests, to detect the NCI associated with HAND. This creates problems in understanding how individual performance compares to the general population. For example, using the MoCA’s North American norms in a South African population suggested very high rates of dementia among normal adults (see Robbins et al. 2013).

4 Future Directions

Computer- and tablet-based screening tests may provide a platform that greatly increases the feasibility and accuracy of NCI screening. The Robbins et al. (2018) study demonstrated that a lay counselor administered tablet-based screening test had robust sensitivity and specificity to detect NCI among South African PLWH. This creates an opportunity to make screening more feasible and widely available in resource-limited settings. With their ease of use, low unit cost, reliability, and accuracy, tablet-based cognitive tests may also make the collection of normative performance data more feasible and less expensive across larger segments of societies, as it would not require highly trained neuropsychologists or psychometrists to administer a lengthy paper-and-pencil battery. Furthermore, because data from tablets could be easily linked to electronic databases, new opportunities to use big data science approaches to examine prevalence and predictors of NCI among and across populations will arise.

Regardless of using a tablet- or paper-based screening test, a first step that is critical to developing and implementing screening and referral programs is research to evaluate any potential program’s acceptability and feasibility from patient, provider, and clinic system perspectives. Understanding what the implementation challenges will be to making screening in clinical settings routine, such as time and space constraints, training requirements, and insurance reimbursement, is an essential step to developing a scalable and sustainable program. Finally, as screening programs are developed and implemented, they will also have to demonstrate positive impact on clinical care and patient outcomes.

5 Summary

HAND remains prevalent among PLWH. There are numerous tests available to screen for HAND, though many have a variety of limitations that make them less appealing for routine use. Common limitations among most available tests include who can administer it, how well it can detect the range of HAND severity (most lack accuracy to detect mild forms), and lack of norms for specific populations. Short neuropsychological test combination screeners often have better performance criterion validity than individual screening tests, but they too have limitations in who can administer, score, and interpret them, as well as costs for testing materials and equipment (Kamminga et al. 2013). Neuropsychological tests must be interpreted by a qualified neuropsychologist. In most LMIC settings, where the burden of HIV is the greatest, qualified staff are rarely available and not easily accessible.

Computerized screeners, such as CogState (Cysique et al. 2006) and NeuroScreen (Robbins et al. 2014, 2018), offer new possibilities to screen for HAND. While the requirements of staff administrators for these tools are greatly reduced (and may be appropriate for community health workers and lay professionals) and the hardware for using them (i.e., tablets and computers) are becoming more affordable and ubiquitous, there are challenges to using these types of tools in many settings. Software and technical support costs may be prohibitive. Moreover apps for mobile devices require ongoing programming maintenance as mobile devices constantly receive operating system updates and changes in form factors which require developers to provide timely updates. Moreover, computers and mobile devices are also dependent on infrastructure, such as electricity to charge devices and Wi-Fi for software updates. These may be limiting factors in some settings. Paper-and-pencil tests do not suffer from these limitations.

Despite these limitations, newer screening tools may help to make screening more widely available to larger segments of society. Routine screening for NCI among PLWH – especially those newly diagnosed – constitutes good clinical practice. It can enable providers to detect early signs of NCI, determine if and when to adjust ART regimens, track and monitor neurocognitive function, and educate patients about the impact of NCI and ways to minimize it – all of which can improve health outcomes. The impact of NCI on ART adherence can be minimized through behavioral planning, and detecting NCI in highly infectious PLWH (i.e., those with detectable viral load) may help in tailoring transmission prevention strategies. Furthermore, if/when pharmacotherapies or emerging behavioral interventions become widely available for HIV-related NCI, screening will be essential to link patients to appropriate services.