This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on http://formative.jmir.org, as well as this copyright and license information must be included.
The development of neurocognitive deficits in people infected with HIV is a significant public health problem. Previous cross-sectional studies have shown that performance on central auditory tests (CATs) correlates with cognitive test results in those with HIV, but no longitudinal data exist for confirmation. We have been performing longitudinal assessments of central auditory and cognitive function on a cohort of HIV-positive and HIV-negative individuals in Dar es Salaam, Tanzania to understand how the central auditory system could be used to study and track the progress of central nervous system dysfunction.
The goal of the project was to determine if CATs can track the trajectory of cognitive function over time in people diagnosed with HIV.
Tests of peripheral and central auditory function as well as cognitive performance were performed on 382 individuals over the course of 3.5 years. Visits were scheduled every 6 months. CATs included tests of auditory temporal processing (gap detection) and speech perception in noise (Hearing in Noise Test and Triple Digit Test). Cognitive tests included the Montreal Cognitive Assessment (MoCA), Test of Variables of Attention (TOVA), and subtests from the Cogstate battery. HIV-positive subjects were divided into groups based on their CAT results at their final visit (bottom 20%, top 20%, middle 60%). Primary analyses focused on the comparison between HIV-positive individuals that performed worse on CATs (bottom 20%) and the overall HIV-positive group (middle 60%). Data were analyzed using linear mixed-effect models with time as the main fixed effect.
The group with the worst (bottom 20%) CAT performance showed a difference in trajectory for the MoCA (
The results of this study support the ability for CATs to track cognitive function over time, suggesting that central auditory processing can provide a window into central nervous system performance. CATs can be simple to perform, and are relatively insensitive to education and socioeconomic status because they only require repeating sentences, numbers, or detecting gaps in noise. These tests could potentially provide a time-efficient, low-cost method to screen for and monitor cognitive decline in patients with HIV, making them a useful surveillance tool for this major public health problem.
Even with advanced antiretroviral therapy, people infected with HIV can develop neurocognitive deficits [
We have been examining the use of central auditory tests (CATs) as an approach for tracking cognitive function in HIV-positive individuals. Our earlier work established that performance on CATs in HIV-positive individuals is strongly related to cognitive test results [
The central auditory system provides a window into brain function because processing complex auditory information is a neurologically demanding task [
Problems with central auditory processing often manifest as difficulty understanding speech, particularly in the context of background noise. Accurate speech perception requires complex processing in the auditory midbrain and cortex [
To test central auditory processing, we have assembled a battery of behavioral CATs that measure the system in two critical ways: temporal auditory processing and speech perception in noise. Temporal processing refers to the precise perception of time alterations on audible acoustic events [
CATs have several practical advantages over cognitive assessments: they do not require literacy or a high level of education to complete; they are short and easy to explain; and some tests can even be administered remotely, by phone or internet. Thus, these tests could be a major advance for following HIV-positive patients, particularly in the developing world. If performance on CATs can track or provide an early marker of CNS dysfunction in HIV infection, detecting these changes in clinical practice could lead to appropriate adjustments in HIV treatment. CATs could be used to identify CNS comorbidities or to track treatment effects. Antiretroviral drugs differ in their ability to penetrate the CNS to treat HIV, and resistance to particular antiretrovirals can develop over time [
This study was designed to ascertain how longitudinal performance on CATs relates to neurocognitive performance in a cohort of HIV-negative and HIV-positive individuals. We hypothesized that the diffuse white matter disease associated with HIV infection would affect central auditory processing progressively [
We recruited participants in this study from a unique cohort of approximately 670 HIV-positive and HIV-negative individuals in Dar es Salaam, Tanzania, who have been performing central auditory, peripheral auditory, and cognitive testing at approximate 6-month intervals for the last 4 years. The research protocol was approved by the Committee for the Protection of Human Subjects of Dartmouth College and the Research Ethics Committee of Muhimbili University of Health and Allied Sciences. All participants provided written informed consent.
Subjects completed a series of questionnaires, and performed cognitive and auditory tests at the Infectious Disease Center in Dar es Salaam, Tanzania. The questionnaires gathered data on the participants’ self-reported hearing ability (hearing status questionnaire) and general health (health history questionnaire). The questions covered noise exposure, tinnitus, ear drainage, ear infections, chemical exposure, and balance problems. The questionnaire also asked about past or current tuberculosis treatment; HIV treatment; gentamicin exposure; and the use of antimalarials, aspirin, and diuretics. All participants completed testing at approximately 6-month intervals; not all participants adhered to the schedule and some dropped out of the study during this time.
To ensure accuracy of longitudinal analysis, and control for variables that could affect central auditory and cognitive function tests, we used a series of data selection techniques. First, individuals were excluded if they only completed 3 or less visits. Second, data from visits beyond 3.5 years were excluded to limit bias from the subset of subjects with longer follow up (ie, a few subjects with long follow-up times could have greater leverage in the model). Third, individuals were excluded if they had abnormal hearing sensitivity (>25 dB HL from 0.5 to 4 kHz) or abnormal middle ear function. Fourth, individuals were also excluded if they had a positive history of ear drainage, concussion, significant noise or chemical exposure, neurological disease, mental illness, ototoxic antibiotics (eg, gentamycin), or chemotherapy. This selection technique resulted in a final sample of 382 individuals.
Peripheral auditory tests included tympanometry and audiometry after otoscopy with cerumen removal as needed to ensure a clear ear canal. A Madsen Otoflex 100 system (GN Otometrics, Denmark) was used to perform tympanometry at 226 Hz. Measurements of ear canal volume, static admittance, tympanometric peak pressure, tympanometric width, and tympanogram type (A, As, Ad, B, C) were collected. Type A tympanograms (including As and Ad) were required for inclusion in this study, with pressure limits from –100 to +50 daPa and static admittance limits from 0.3 to 1.7 milimho.
Pure-tone air conduction thresholds were measured at frequencies of 0.5, 1.0, 2.0, and 4.0 kHz using a Békésy-like tracking procedure as previously described [
Audiometry and all behavioral audiometric testing were completed using a Creare LLC wireless automated hearing test system (WAHTS) controlled through a laptop. The WAHTS allowed for testing in rooms with minimal background noise, as the device speakers are mounted in the ear cups. The attenuation provided by this headset is on par with a portable sound booth as measured by an independent laboratory according to the relevant American National Standards Institute standards [
CATs included the Hearing In Noise Test (HINT), Triple Digit Test (TDT), and gap detection test (GAP). The HINT was administered in four test conditions: noise front, noise right, noise left, and quiet. In each HINT, a different list of 20 sentences was presented in random order in the presence of the masking noise spectrally matched to the long-term average of the target material. The presentation level of the noise remained fixed at 65 dB (A-weighting), and the test instrument adjusted the level of each sentence adaptively depending on whether the test administrator indicated that the previous sentence was repeated correctly. The presentation level of the sentence was reduced if the previous sentence was repeated correctly and was increased if the previous sentence was repeated incorrectly. This adaptive procedure was used to determine the presentation level of each sentence in the list. The average presentation level of all sentences after the first four sentences defined the speech reception threshold for the test condition expressed as a signal-to-noise ratio (SNR). The WHATS displayed and recorded the SNR for each test condition. A composite SNR of all three noise conditions was calculated and used as the primary variable of interest for the HINT.
In the TDT, recordings of natural productions of three-digit triplets such as 3-5-9 (spoken as “tatu-tano-tisa” in Kiswahili) were used as target stimuli (Kiswahili numbers below 10 have the same number of syllables). All digit triplets were produced and recorded by a male speaker in a soundproof booth. Triplet digit recognition was tested in the presence of competing Schroeder-phase masking noise. The test included 30 total presentations of pseudorandom triplet digits with six practice presentations. Presentations were delivered in pairs of positive- and negative-phase maskers. Each pair was presented at the same SNR, and the order of the masker was randomized for each pair. The test started at a 0 dB initial SNR with the masker fixed at a 75 dB sound pressure level. SNRs were then adjusted after each presentation or pair of presentations by varying the target level; a 1.5 dB sound pressure level was added to the target level for each incorrect digit and a 1.5 dB sound pressure level was subtracted for each correct digit from the previous positive-phase presentation. The speech reception threshold was calculated as the SNR of the last 14 positive-phase presentations, which was used as the primary variable of interest.
We also implemented an adaptive GAP test to evaluate temporal auditory processing. The adaptive gap detection algorithm applies a single staircase and has been used extensively in our previous studies [
We used three cognitive tests: the Montreal Cognitive Assessment (MoCA), the Tests of Variables of Attention (TOVA), and selected subtests from the Cogstate battery. The MoCA was used to assess the participants’ general cognitive abilities and screen for potential cognitive impairment [
The TOVA (TOVA Company, Los Alamitos, CA, USA [
The final cognitive test battery, the Cogstate battery [
To test the applicability of CATs to track cognitive function over time, we created four experimental groups. The first group consisted of HIV-negative individuals. We used this group to create normative values for each central auditory and cognitive test. To divide the HIV-positive group on the basis of CAT performance, we used a combination of transformed
Three groups were created based on the combination CAT score. One group included HIV-positive individuals whose performance on the GAP, HINT, and TDT combination
Analyses were conducted using linear mixed-effects models with MATLAB 2020a (Mathworks, Natick, MA). Response variables included measures from the TOVA, Cogstate, and total score on the MoCA. Fixed effects included group (HIV-negative, HIV-positive, TopCATs, BottomCATs), age at last visit, and time between tests. Random effects included individual subject result variation over time. Using age and time as fixed effects allowed for analyses of cross-sectional age differences between subjects and longitudinal changes within subjects across time. This approach was developed by Laird and Ware [
We calculated
Demographic information.
Characteristic | Overall Cohort (N=382) | HIV-negative (n=90) | HIV-positive (n=164) | TopCATsa (n=53) | BottomCATsb (n=75) | ||||
|
HIV-negative vs HIV-positive | TopCATs vs BottomCATs | HIV-positive vs BottomCATs | ||||||
|
|||||||||
|
Male | 130 (34.0) | 44 (49) | 44 (26.8) | 32 (59) | 17 (23) | N/Ad | N/A | N/A |
|
Female | 239 (62.7) | 46 (51) | 119 (72.6) | 21 (41) | 58 (77) | N/A | N/A | N/A |
Age (years), mean (SD) | 37.8 (14.8) | 25.9 (11.8) | 40.8 (13.6) | 38.1 (12.8) | 42.1 (8.4) | <.001 | .001 | .001 | |
|
|||||||||
|
Right ear | 7.62 (6.1) | 3.93 (6.8) | 7.25 (5.2) | 5.06 (5.2) | 8.42 (8.2) | <.001 | .001 | .02 |
|
Left ear | 6.43 (6.6) | 3.53 (7.2) | 6.87 (5.9) | 5.88 (5.0) | 7.29 (6.1) | <.001 | .001 | .03 |
Education (years), mean (SD) | 9.01 (2.7) | 10.23 (2.6) | 8.83 (2.6) | 9.7 (2.3) | 8.62 (2.8) | <.001 | .001 | .34 | |
MoCAf, mean (SD) | 27.6 (3.0) | 28.2 (3.4) | 27.2 (3.1) | 27.9 (2.8) | 26.7 (3.3) | .009 | .01 | .04 |
aTopCATs: in the top 20% of central auditory test results for HIV-positive individuals.
bBottomCATs: in the bottom 20% of central auditory test results for HIV-positive individuals.
cBased on two-sample
dN/A: not applicable.
ePTA: pure tone average (0.5, 1.0, 2.0, 4.0 kHz).
fMoCA: Montreal Cognitive Assessment.
Results of linear mixed effect models.
Variable | Time estimate (slope, |
Age×group |
Time×group |
|||||||
|
|
HIV+ | Topa | Bottomb | HIV+ | Top | Bottom | HIV+ | Top | Bottom |
|
||||||||||
|
Gap detection threshold (ms) | –4.22 ×10–4 | –7.01 ×10–4 | –5.33×10–4 | Refd | .54 | .45 | Ref | .03 | .43 |
|
HINTe (SRTf) | –4.68×10–4 | –5.61 ×10–4 | –4.10×10–4 | Ref | .53 | .16 | Ref | .17 | .89 |
|
TDTg (mean SNRh) | –8.93×10–4 | –9.01×10–4 | 2.16×10–4 | Ref | .32 | .92 | Ref | .99 | <.001 |
MoCAi (total score) | 15.5×10–4 | 12.5×10–4 | –1.96×10–4 | Ref | .19 | .99 | Ref | .56 | .003 | |
|
||||||||||
|
Response time (mean) | 5.70×10–4 | 4.91×10–4 | 2.19×10–4 | Ref | .63 | .06 | Ref | .47 | .007 |
|
ExGaussiank |
3.55×10–4 | 1.01×10–4 | 0.942×10–4 | Ref | .47 | .23 | Ref | .08 | .048 |
|
Attention comparison score | 4.50×10–4 | 3.42×10–4 | 4.08×10–4 | Ref | .72 | .29 | Ref | .09 | .13 |
|
||||||||||
|
Groton maze learning (moves per second) | 1.05×10–4 | 0.721×10–4 | 0.511×10–4 | Ref | .84 | .69 | Ref | .18 | .046 |
|
Groton maze learning (total errors) | 13.8×10–4 | 11.3×10–4 | 6.15×10–4 | Ref | .24 | .16 | Ref | .45 | .03 |
|
One Card Learning (accuracy) | 3.86×10-4 | 2.95×10-4 | 0.761×10-4 | Ref | >.99 | .71 | Ref | .78 | .08 |
|
One Back Test (reaction time) | 13.2×10-4 | 9.12×10-4 | 10.9×10-4 | Ref | .62 | .18 | Ref | .06 | .16 |
|
Continuous paired associate learning (accuracy) | 4.44×10-4 | 2.98×10-4 | 4.05×10-4 | Ref | .28 | .08 | Ref | .16 | .07 |
Global executive scorel | 9.61×10-4 | 8.62×10-4 | 7.31×10-4 | Ref | .47 | .21 | Ref | .30 | .02 | |
Global speed scorem | 3.04×10-4 | 3.00×10-4 | 0.511×10-4 | Ref | .35 | .07 | Ref | .89 | .006 | |
Global CAT score | 3.59×10-4 | 4.21×10-4 | 0.540×10-4 | Ref | .45 | .23 | Ref | .58 | .01 |
aTop: HIV-positive individuals in the top 20% of combined central auditory test scores.
bBottom: HIV-positive individuals in the bottom 20% of combined central auditory test scores.
cCAT: central auditory test.
dReference for comparison.
eHINT: Hearing In Noise Test.
fSRT: speech reception threshold.
gTDT: Triple Digit Test.
hSNR: signal-to-noise ratio.
iMoCA: Montreal Cognitive Assessment.
jTOVA: Tests of Variables of Attention.
kExGaussian: exponentially modified Gaussian distribution.
lCombination of speed-of-processing subtests.
mCombination of executive functioning subtests.
Trajectory of central auditory tests over time. Time 0 is the first visit in the study. Subjects were tested at roughly 6-month intervals thereafter. The blue dashed line shows the slope of the HIV-negative group, the red line shows the slope of the HIV-positive group, the magenta line shows the slope of the TopCATs group, and the black line shows the slope of the BottomCATs group. GAP, HINT, and TDT scores were used in combination to create the TopCATs and BottomCATs groups. The lines were fit to the data using linear fitting procedure in MATLAB. CAT: central auditory test; TopCATs: HIV-positive individuals in the top 20% of CAT scores; BottomCATs: HIV-positive individuals in the bottom 20% of CAT scores.
Trajectory of MoCA, TOVA, and Cogstate measures over time. The color scheme of lines for each group is the same as that in
Trajectory of derived global scores over time. The color scheme of lines for each group is the same as that in
Among HIV-positive individuals, the trajectory of cognitive performance over time differed as a function of CAT performance. Over the time period studied, those individuals with HIV that fell in the bottom 20% on a combination of CATs either worsened or improved more slowly than the other groups. In contrast, both HIV-negative patients and HIV-positive patients with good CAT performance improved in their cognitive performance, potentially reflecting learning effects on the tasks. This suggests that CAT scores in those diagnosed with HIV are correlated with a worsening trajectory or failure to improve on measures of cognitive function.
Results from this study suggest that CATs may be a useful way to provide surveillance for the development of neurocognitive problems in people with HIV. Previous studies have suggested that HIV-positive individuals develop signs of central auditory processing deficits [
Results of this study also suggest that the main problems of CATs are in the domains of executive function and processing speed. This is evidenced by significant results on the MoCA, TOVA, Cogstate, and global scores. The MoCA displayed the strongest interaction (
HIV-positive individuals with poor central auditory function also showed degradations in processing speed on a variety of cognitive subtests over time. For example, the mean response time and ExGaussian
One interpretation of our results could be that some individuals with HIV experience an accelerated aging process. That is, HIV could be associated with accelerated cognitive aging such that a subset of people with HIV in their 40s and 50s are functioning with a cognitive processing speed typical of that found in people in their 60s and 70s. Cognitive and central auditory deficits in adults with HIV may result from additive effects of the pathophysiological mechanisms of aging (ie, the “common cause” hypothesis [
Future studies should seek to examine executive functioning and speed of processing in the central auditory pathway in HIV-positive individuals to improve upon surveillance of this public health concern. Declines in general cognitive processing speed have been considered a hallmark of the aging process, beginning in young adulthood and continuing nearly linearly across the lifespan [
This study has limitations. The main limitation in interpreting the cognitive variables accurately over time was an overall learning effect. Although previous results have shown minor learning effects on these cognitive variables [
Another limitation is that this study was conducted over a 3.5-year period, which is not an exceptionally long time to develop cognitive decline due to HIV. These data were from an ongoing project in Dar es Salaam, Tanzania. More time is needed for more individuals to complete multiple visits and for deterioration in neurocognitive performance to develop to fully answer the question of whether CATs can predict future cognitive decline. This study was only able to show the association.
Differences in PTA may have affected the results. Although normal hearing sensitivity from 0.5 to 4.0 kHz was required for inclusion, the difference in hearing thresholds (ie, the difference in PTA between the BottomCATs and HIV-positive groups) might have affected the CAT results. Although it is unlikely that an averaged difference for both ears of 1.2 dB in PTA affected the results, it is possible that peripheral hearing sensitivity also factored into the trajectory of cognitive variables over time. This could, however, also be related to damage not reflected in hearing thresholds, such as damage to the synapses between hair cells and the cochlear nerve or further along the auditory pathway. Studies have suggested that peripheral hearing sensitivity is not a comprehensive picture of auditory function [
The overall results from this study suggest that CATs may be useful to track cognitive function over time in people with HIV. This could provide an easy-to-use, quick method of surveillance for this important public health problem. Subjects that performed in the bottom 20% of a battery of three CATs had a significantly different trajectory of cognitive variables over time, suggestive of cognitive dysfunction. The cognitive dysfunction seen was consistent with a failure to improve or decrease in executive functioning and speed of processing in those with poor central auditory function over time. This study supports the ideal that CATs should be studied further to track cognitive dysfunction in those with HIV-related cognitive deficits.
central auditory test
central nervous system
exponentially modified Gaussian distribution
gap detection threshold
Hearing In Noise Test
Montreal Cognitive Assessment
pure tone average
signal-to-noise ratio
Triple Digit Test
Tests of Variables of Attention
wireless automated hearing system
We thank the team at the DarDar clinic in Dar es Salaam, Tanzania who collected these data (Esther Kayichile, Joyce Ghatty, Claudia Gasana, Filmon Sulle, Pascal Donard, Godfrey Njau, Matilda Kabeho, and Betty Mchaki). We thank the team at Creare, LLC that assembled and tested the hearing testing systems. We appreciate the support of Erika Kafwimi and Sabrina Yegela who helped with building the video questionnaire and translating the questions. This work was supported by the National Institute on Deafness and Other Communication Disorders (grant R01DC009972) and by the National Institutes of Health (grant number 5R01DC009972-10; JB principal investigator). The content of this report is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
CN performed the statistical analysis and data interpretation, and was primarily responsible for writing the manuscript. AF was primarily responsible for training and study management; she assisted with study design, data analysis, and data interpretation. JL and TS assisted with data analysis and data interpretation. AM was primarily responsible for study management in Tanzania, enrolling patients, and data review. JG assisted with statistical analysis. OC assisted with study design, test development, and data analysis. EM assisted with study management, data analysis, and interpretation. NM assisted with study design and was primarily responsible for study oversight in Tanzania. MB assisted with cognitive data interpretation. NK assisted with study design, data analysis, and data interpretation. JB was the principal investigator, and was involved in study design, training, data analysis, and data interpretation. All authors assisted with revising the final work and approved the final version to be published. All authors agree to be accountable for all aspects of the work and ensuring that questions about the accuracy or integrity of any part of the work are appropriately investigated and resolved.
None declared.