With the advent of combination antiretroviral therapies (cARTs), people living with HIV (PLWH) are healthier and living longer [1,2]. Similar to other ageing populations, PLWH are experiencing multiple comorbidities and age-associated morbidity [3,4]. In addition, PLWH have added complexities of chronic infection and the inherent inflammation as well as long-term exposure to antiretroviral medications that have been linked to important adverse actions, all of which may be associated with an increased risk of poor health outcomes [5–7] (Fig. 1).
The impact of chronic HIV and cART on brain health is complex; HIV enters the central nervous system (CNS) early after initial exposure and is associated with mild cognitive changes, which persist despite successful viral suppression with cART [6–9]. When untreated, HIV infection causes a progressive subcortical dementia (HAD) and immunodeficiency from HIV can contribute to opportunistic infections, including progressive multifocal leukoencephalopathy and lymphoma [9,10]. Rarely, among people on cART with undetectable plasma viral loads, cerebrospinal fluid (CSF)-viral escape occurs and is associated with cognitive impairment and other neurological symptoms . Observational neuroimaging studies link chronic HIV to progressive atrophy, even among study participants with persistent viral suppression [11,12]. It has also been established that PLWH on cART are at an increased risk of cerebrovascular, cardiovascular and metabolic diseases compared with people with similar risk profiles without HIV, all of which are known to affect brain health and dementia risk .
In the context of an ageing population of PLWH, it is unclear whether PLWH are at an increased risk of developing dementia possibly from HAD progression or greater risk factor burden or some other mechanism [14–16]. The objective of our study was to determine whether HIV-infection was associated with incident dementia diagnosis in a prospective cohort of older U.S. veterans, primarily men, after accounting for important potential confounders and competing risk of death. We also examined whether exposure to cART modified this risk.
Materials and methods
Data and study participants
We studied a random sample of veterans ages 55 years and older enrolled in the Veterans Health Administration (VHA) system. Veterans were included in the study from their first encounter during the study period between 1 October 2004 through 30 September 2015. Data were extracted from the National Patient Care Database, an electronic database that captures information on all inpatient and outpatient encounters that occur at VHA healthcare facilities nationwide.
From this sample, we identified all veterans with a HIV diagnosis (n = 1114) at baseline (defined as within 5 years before random selection date) and performed 1 : 1 propensity score matching to match veterans with HIV to with those without (n = 1114.) Participants were matched on age, sex, race and substance use (drug abuse, alcohol abuse, tobacco use). Veterans with prevalent dementia during the baseline period and those who did not have at least one visit during the follow-up period were excluded. We evaluated veterans for incident dementia during an outcome period that lasted until a diagnosis of dementia, death or the end of the study (September 2015).
HIV, CD4+ and combination antiretroviral therapy exposure
HIV diagnosis was coded using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes diagnosis codes for AIDS (042), asymptomatic HIV (V08) and related Diagnostic-Related Group codes (488–490).
CD4+ test data were obtained from electronic laboratory data starting from 1 October 2004 to study end. We used the mean CD4+ cell count (cells/μl) from the test closest to dementia diagnosis, death or study end.
We categorized participants by cART exposure (ever, never) and by regimen type among those cART-exposed using electronic prescription records starting from 1 October 2001. All cART medications that were on Veterans Affairs (VA) formulary during the study period were included. It has been previously shown in a nested sample that 98% of HIV-infected veterans obtain their cART medications from the VA .
Veterans with dementia were identified using ICD-9-CM codes recommended by the VA Dementia Steering Committee (2016 version) during the baseline period to exclude prevalent cases. Incident dementia during the follow-up period was classified using a slightly modified version of the VA Steering Committee ICD-9 codes in which we excluded prion disease (ICD-9 codes 046.11, 046.19, 046.3 and 046.79) and alcohol-induced or drug-induced dementia (ICD-9 codes 291.2 and 292.82) . We defined dementia subtype using ICD-9-CM codes as follows: HIV-associated dementia (HAD) 042 in conjunction with 294.10 or 294.11; Alzheimer's type: 331.0, 290.0–290.3; vascular dementia: 290.4–290.43; or dementia not otherwise specified (NOS): 294.8.
Evaluation of incident dementia diagnosis occurred between 1 October 2004 through 30 September 2015. If dementia was not diagnosed before the patient's death or the end of the follow-up period, the patient's data were censored at whichever event occurred first. Date of death was determined using the VHA Vital Status File, which combines information from the VHA, the Center for Medicare & Medicaid Services and the Social Security Administration to determine date of death. Prior studies have found that the VHA Vital Status File is comparable to the National Death Index in accuracy and completeness .
Demographic information, medical comorbidities, psychiatric conditions and number of follow-up visits were obtained from the VHA inpatient and outpatient files. Demographic data were based on self-report and included baseline age calculated at 1 October 2004; sex and race/ethnicity (categorized as non-Hispanic white individuals, non-Hispanic black individuals, Hispanic individuals or individuals of other or unknown races/ethnicities). We classified veterans as living in broad educational and income strata according to zip code tabulation areas (ZCTA) by linking our data to 2016 American Community Survey (ACS) data. For educational level, veterans were categorized according to whether they were living in a ZCTA where 25% or less versus more than 25% of the adult population had completed a college education (bachelor's degree or higher). For income, veterans were categorized in tertile of median ZCTA income for adults less than 75 or at least 75 years old. We calculated the number of follow-up visits by summing the number of inpatient episodes or outpatient visits per person per year. Indicators of prevalent medical and psychiatric comorbidity during the baseline period were determined using ICD-9-CM codes.
This study was approved by the Committee on Human Research at the University of California, San Francisco; the Research & Development Committee at the San Francisco VA Medical Center; and the Human Research Protection Office of the US Army Medical Research and Materiel Command and has been granted a waiver of informed consent. The study is HIPAA compliant.
Matching covariates were age, sex, race and substance use (drug abuse, alcohol abuse, tobacco use). Baseline characteristics were summarized with means and standard deviations or percentages for the veterans with and without HIV. The characteristics of the two groups were compared using t-tests for continuous variables and chi-square tests for categorical variables.
Cumulative incidence of dementia was plotted by patient age in years for veterans with and without HIV accounting for the competing risk of death. To account for the competing risk of death, Fine-Gray proportional hazards models were used to determine whether HIV status was associated with a greater risk of incident dementia with censoring at the date of the last medical encounter and age as the time scale.
Models were unadjusted and adjusted for age, sex and race, measures of education and median income level, substance use. Effects are reported as hazard ratios with 95% confidence intervals (95% CIs). We tested for interactions by cART exposure, and we performed sensitivity analyses with additional adjustment for mean CD4+ cell count and for the number of follow-up visits per person per year. Proportional hazards model assumptions were checked for all final models. P values were two-sided with statistical significance defined as P value less than 0.05. Analyses were performed using SAS, version 9.4 (SAS Institute, Cary, North Carolina, USA).
Our final cohort included 2228 veterans with HIV (n = 1114) and without (n = 1114). Study participants had a mean age of 62.5 years (SD 6.33) at baseline. Over 98% of the sample were men (n = 2202) and 52% were non-Hispanic white (n = 1167), 38% non-Hispanic black (n = 844) and 10% other (n = 217.) Health risk behaviours and chronic noncommunicable conditions were common. Twenty-two percent had a history of tobacco use, and about 18% had an alcohol/drug abuse diagnosis. The prevalence of diabetes at baseline was 17%, high blood pressure 34% and 7% a history of stroke or transient ischemic attack. The comparison groups were generally well matched on individual-level variables (see Table 1). Participants with HIV were more likely to live in zip codes with higher education levels but lower median incomes.
During the follow-up period of up to 11 years (mean 5 years, SD 3), 57 cases of incident dementia (5.1%) were diagnosed in people with HIV compared with 33 cases in those without HIV (3%), P = 0.01. Most had a diagnosis of dementia NOS. Specific dementia subtype was only recorded in one-third of participants; in veterans with HIV, 11 (19%) developed a diagnosis of HAD during follow-up. Excluding HAD, the patterns of dementia subtype diagnoses where recorded were similar between the groups; four Alzheimer's disease (7%) and six vascular dementia (10%) in people with HIV, compared with 4 with Alzheimer's disease (10%), six vascular dementia (18%) and two frontotemporal dementia (FTD) in those without HIV.
Accounting for the competing risk of death and adjusting for age, sex, race/ethnicity, substance use, education and income, veterans with HIV were still 50% more likely to receive a diagnosis of dementia (hazard ratio 1.50, 95% CI 0.96–2.35). The cumulative incidence of dementia in veterans with and without HIV adjusted for demographic and comorbidities is shown in Fig. 2. A total of 467 (21%) veterans died during the follow-up period. Veterans with HIV were more likely to die during follow-up (unadjusted relative risk 1.67, 95% CI 1.41–1.97) than those without.
On the basis of electronic prescribing data, 61% of people with HIV were prescribed cART; of these, 198 (29%) were prescribed regimens containing a single cART class [75 prescribed a nucleoside reverse transcriptase inhibitor (NRTI), 72 a non-nucleoside reverse transcriptase inhibitor (NNRTI), and 49 a protease inhibitor (PI)], 369 (54%) were prescribed two cART class containing regimens and the reminder were prescribed regimens containing three cART classes. Among people without HIV, two were taking NRTI agents for viral hepatitis. Comparing veterans with HIV who were and were not prescribed cART, we found the groups had similar sociodemographic profiles. There were differences in some health risk factors (hypertension, arthritis, mental health diagnoses and substance use), measures of HIV control (CD4+ cell count) and time in the study with people taking cART having worse risk factor profiles and also fewer visits per-person per-year though longer follow-up time (see Table 2.) Veterans with HIV and cART exposure were significantly more likely to be diagnosed with dementia during follow-up (aHR 1.78, 95% CI 1.11–2.85) compared with those without cART exposure and those without HIV (See Table 3). However, this difference was no longer significant after adjustment for CD4+ cell count (aHR 1.20, 95% CI 0.50–2.86). In additional sensitivity analyses comparing type of cART regimes (NNRTI to non-NNRTI and NRTI to non-NRTI based regimens), we did not find convincing evidence that one regimen type had more or less of an effect on risk estimates than the other; aHR 0.73 (95% CI 0.39–1.35) and aHR 1.78 (95% CI 0.99–3.21), respectively, and formal tests of difference were not significant.
In a cohort of 2228 mostly male veterans with and without HIV, we found that HIV infection was associated with an increase in risk of dementia diagnosis during follow-up. We also found that the increase associated with cART exposure was driven by differences in measures of virologic control and other health differences. If causal, the effect size estimates we found are comparable to other established potentially modifiable risk factors for dementia . The mechanisms by which HIV may increase the risk of dementia diagnosis remain unclear. Clarifying the risks whether biological or socially patterned is key to developing appropriate prevention and mitigation strategies for this population group.
Few prior studies have assessed the risk of dementia associated with HIV-infection comparing older people with HIV to those without. To our knowledge, only one matched-cohort study from Taiwan prospectively assessed dementia risk in adults over the age of 50 with and without HIV. The study included 1261 people with HIV and 3783 without but only 25 and 227 incident dementia diagnoses during the follow-up period. The results showed that HIV was not associated with an increase in risk of dementia diagnosis (aHR 0.852, 95% CI 0.19–2.9, P = 0.4). Analysis of the risk by cART (no association) was limited by the fact that only one participant taking cART developed dementia during the follow-up period . Results from a cross-sectional study screening older adults in South Africa for dementia which included HIV-testing showed higher, but not significantly different, dementia point prevalence estimates in people with HIV compared with without. However, the sample only included 55 people with HIV and only 10 screen-positive dementia cases making interpretation of these results difficult .
Much of the published research to date exploring the risk of dementia in people with HIV has focused on the risks of HAD [23–25]. HAD is less common in the era of widely available cART. As a dementia subtype, it was recorded in in nearly one-fifth of our study participants who developed a dementia diagnosis during follow-up (11 out of 57), and is probably some of the dementia NOS . Ongoing HAD risk is one of several potential reasons for an increase in dementia risk in PLWH and in this age cohort may be due to legacy effects of early treatment strategies for HIV (regimens containing only one or two agents and waiting to start treatment until symptomatic or a CD4+ cell count threshold reached) and newly recognized CNS viral escape syndromes, that is ongoing viral replication in the CNS despite suppressed peripheral viral loads often associated with previous episodes of poor treatment compliance .
Chronic inflammation and immune activation associated with HIV are known to increase the risk of cardiovascular disease, which includes cerebrovascular disease and stroke and may increase the risk of vascular dementias . Although dementia subtype was only available for about one-third of our study participants, we found that the patterns of non-HAD dementia subtypes were similar in those HIV and without, and that vascular dementia was the most frequently recorded subtype. We did not find convincing evidence for an increase in Alzheimer's disease diagnosis in this cohort. Other socially patterned and modifiable established dementia risk factors for dementia may also be more prevalent in PLWH than those without (e.g. substance use), which combined with increasing longevity in this patient group would lead to increases in dementia. Although we adjusted for these, we cannot rule out residual confounding due to imperfect measurement of these factors.
Veterans prescribed cART were different from those not across a range of measured health risk factors and HIV control. Our finding that veterans with HIV and cART exposure were at a higher risk of an incident dementia diagnosis than those not taking cART was driven by differences in measures of virologic control (CD4+ cell count) and other health differences. Although we would expect that veterans prescribed effective treatments would have better outcomes, CD4+ cell count data indicate the group prescribed cART had lower CD4+ nadirs and lower CD4+ cell counts at the time of dementia diagnosis. We think this is indication bias, veterans who were sicker, with lower starting CD4+ cell counts, were more likely to be prescribed cART. In addition, in the USA in the years covered by this study, cART management strategies (e.g. ‘drug holidays’) changed and new cART regimens were established, and we cannot rule out that our results reflect a cohort effect.
Our study has several important strengths, including its longitudinal design and conservative approach to excluding people with prevalent dementia (5 years before baseline). We matched people with HIV to those without on important covariates and we adjusted models using a-priori specified key potential confounders. We also treated death as a competing risk, which is a more conservative analytic approach. However, there are several limitations to consider when interpreting these results. The study used medical record databases, laboratory tests and electronic prescribing records to ascertain exposures and outcomes. Clinician ICD-9 codes and dementia medication prescriptions to establish dementia diagnoses can be less sensitive than structured diagnostic interviews, particularly for dementia subtype classification. In recent years, new diagnostic tools and protocols have been developed that allow for much improved dementia subtype classification; for PLWH understanding, the type and distribution of dementia subtypes will help clarify potential mechanisms and interventions. We did not have information on HIV viral load. We also cannot rule out that clinicians managing older PLWH have a lower threshold for asking about subjective cognitive complaints leading to increases in diagnostic rates in this patient population; however, if this were the case, we might expect to see differential dementia subtype coding rates, which was not the case in this study. Two-thirds of our sample were exposed to cART during the study that likely reflects some combination of access to and engagement in care and historical prescribing practices and is different to current best practice in HIV management in the USA. However, it is not very different from the proportion of people accessing cART globally (62% in 2018) and available cART regimens . In addition, sex-based differences are known for several dementia risk factors and outcomes and we were not able to explore for differences in risk by sex because our sample included only 1% women. Finally, as our sample included only veterans using VHA facilities, it is not clear whether results are generalizable to veterans not using this health service and to nonveterans.
Our study is the first to examine whether HIV-infection was associated with incident dementia in prospective cohort of older U.S. veterans. We found evidence that HIV infection is associated with an increased risk of dementia diagnosis and we found that differences in risk associated with cART exposure were driven by differences in disease severity. It is critical to understand the mechanisms by which risk is increased (particularly dementia subtypes) whether biological or socially patterned. As the population of PLWH in the US ages, understanding general and unique dementia risk factors for this group is needed to develop appropriate prevention and mitigation strategies.
V.V. serves as a faculty at the IAS-USA and has been a consultant for Merck and ViiV Healthcare on matters of cognition, ageing and HIV. We have no known conflict of interest to disclose.
K.Y. and K.B. conceived and planned the project. F.X. analysed the data. K.B., K.Y., F.X. and T.H. drafted the manuscript and designed the tables and figures. K.B., K.Y., F.X., T.H. and V.V. interpreted results and revised the manuscript.
Conflicts of interest
There are no conflicts of interest.
1. Bekker L-G, Alleyne G, Baral S, Cepeda J, Daskalakis D, Dowdy D, et al. Advancing global health and strengthening the HIV response in the era of the Sustainable Development Goals: the International AIDS Society—Lancet Commission
2. Wing EJ. HIV and aging
. Int J Infect Dis
3. Guaraldi G, Palella FJ. Clinical implications of aging with HIV infection: perspectives and the future medical care agenda
2017; 31: (Suppl 2): S129–S135.
4. Blanco J-R, Barrio I, Ramalle-Gómara E, Beltran MI, Ibarra V, Metola L, et al. Gender differences for frailty in HIV-infected patients on stable antiretroviral therapy and with an undetectable viral load
. PLoS One
5. Gali B, Eyawo O, Hull MW, Samji H, Zhang W, Sereda P, et al. Incidence of select chronic co-morbidities among a population-based cohort of HIV-positive individuals receiving highly active antiretroviral therapy
. Curr Med Res Opin
6. Grant I. Evidence for early central nervous system involvement in the acquired immunodeficiency syndrome (AIDS) and other human immunodeficiency virus (HIV) infections: studies with neuropsychologic testing and magnetic resonance imaging
. Ann Intern Med
7. Dore GJ, McDonald A, Li Y, Kaldor JM, Brew BJ. National HIV Surveillance Committee. Marked improvement in survival following AIDS dementia complex in the era of highly active antiretroviral therapy
. AIDS Lond Engl
8. Ellis RJ, Badiee J, Vaida F, Letendre S, Heaton RK, Clifford D, et al. CD4 nadir is a predictor of HIV neurocognitive impairment in the era of combination antiretroviral therapy
. AIDS Lond Engl
9. Del Valle L, Piña-Oviedo S. HIV disorders of the brain: pathology and pathogenesis
. Front Biosci J Virtual Libr
10. Valcour V, Sithinamsuwan P, Letendre S, Ances B. Pathogenesis of HIV in the central nervous system
. Curr HIV/AIDS Rep
11. Mukerji SS, Misra V, Lorenz D, Cervantes-Arslanian AM, Lyons J, Chalkias S, et al. Temporal patterns and drug resistance in CSF viral escape among ART-experienced HIV-1 infected adults
. J Acquir Immune Defic Syndr
12. Nir TM, Jahanshad N, Ching CRK, Cohen RA, Harezlak J, Schifitto G, et al. Progressive brain atrophy in chronically infected and treated HIV+ individuals
. J Neurovirol
13. Shah ASV, Stelzle D, Lee KK, Beck EJ, Alam S, Sarah C, et al. Global burden of atherosclerotic cardiovascular disease in people living with HIV
14. High KP, Valcour V, Paul R. HIV infection and dementia in older adults
. Clin Infect Dis
15. Hong S, Banks WA. Role of the immune system in HIV-associated neuroinflammation and neurocognitive implications
. Brain Behav Immun
16. Milanini B, Valcour V. Differentiating HIV-associated neurocognitive disorders from Alzheimer's disease: an emerging issue in geriatric NeuroHIV
. Curr HIV/AIDS Rep
17. Justice AC, Dombrowski E, Conigliaro J, Fultz SL, Gibson D, Madenwald T, et al. Veterans Aging Cohort Study (VACS)
. Med Care
2006; 44: (8 Suppl 2): S13–S24.
18. VHA Dementia Steering Committee. VHA dementia steering committee recommendations for dementia care in the VHA health care system 2016
. Washington, DC: Department of Veterans
19. Fisher SG, Weber L, Goldberg J, Davis F. Mortality ascertainment in the veteran population: alternatives to the National Death Index
. Am J Epidemiol
20. Barnes DE, Yaffe K. The projected effect of risk factor reduction on Alzheimer's disease prevalence
. Lancet Neurol
21. Yang C-C, Chien W-C, Chung C-H, Liu Y-P, Yeh C-B, Chen K-H, et al. No association between human immunodeficiency virus infections and dementia: a nationwide cohort study in Taiwan
. Neuropsychiatr Dis Treat
22. Joska JA, Dreyer AJ, Nightingale S, Combrinck MI, De Jager CA. Prevalence of HIV-1 Infection in an elderly rural population and associations with neurocognitive impairment
. AIDS Lond Engl
23. Joska JA, Gouse H, Paul RH, Stein DJ, Flisher AJ. Does highly active antiretroviral therapy improve neurocognitive function? A systematic review
. J Neurovirol
24. Gannon P, Khan MZ, Kolson DL. Current understanding of HIV-associated neurocognitive disorders pathogenesis
. Curr Opin Neurol
25. Dastgheyb RM, Sacktor N, Franklin D, Letendre S, Marcotte T, Heaton R, et al. Cognitive trajectory phenotypes in human immunodeficiency virus infected patients
. J Acquir Immune Defic Syndr
26. Heaton RK, Franklin DR, Ellis RJ, McCutchan JA, Letendre SL, Leblanc S, et al. HIV-associated neurocognitive disorders before and during the era of combination antiretroviral therapy: differences in rates, nature, and predictors
. J Neurovirol
27. Gutierrez J, Albuquerque ALA, Falzon L. HIV infection as vascular risk: a systematic review of the literature and meta-analysis
. PLoS One]
28. HIV.gov. Global Statistics (2019). HIV.gov. [Internet].[cited 22 January 2020]. https://www.hiv.gov/hiv-basics/overview/data-and-trends/global-statistics