Anxiety symptom ratings before and during the interruption of antiretroviral therapy as part of HIV cure-related studies

Abstract

Background:

An analytical treatment interruption (ATI) is a critical component of HIV cure research, allowing researchers to assess potential strategies aimed at durable antiretroviral treatment (ART)-free viral control. However, pausing ART during an ATI may lead to uncertainty and psychological distress. Limited empirical data exist on the mental health impact of an ATI, specifically anxiety symptoms.

Methods:

We conducted a longitudinal study to assess self-reported anxiety levels before and during an ATI in two HIV cure-related research studies at the University of California, San Francisco. Study 1 was an observational ATI study (NCT04359186), while Study 2 was an interventional HIV cure trial (NCT04357821). Anxiety was measured at multiple time points using the Generalized Anxiety Disorder-7 (GAD-7) and the State-Trait Anxiety Inventory. We analyzed changes in anxiety scores over time using interrupted time-series models.

Results:

In Study 1, anxiety was elevated at baseline and gradually declined over time but remained within the mild severity range. In Study 2, anxiety increased significantly from before the ATI and during ATI, with higher peak anxiety levels compared to Study 1. Anxiety was highly prevalent in both studies and increased during the ATI in Study 2.

Conclusions:

The uncertainty with an ATI may contribute to increased mild anxiety. However, this uncertainty can be measured and with resources, easily addressed to reduce risks for psychological harm.

Introduction

An analytical treatment interruption (ATI) may intersect with the mental health of a person with HIV (PWH) participating in HIV cure-related research [1]. An ATI is one specific method in HIV cure research that is a monitored pause (of various lengths of time) of antiretroviral therapy (ART) [2]. The uncertainty surrounding how an ATI is experienced and associated risks, including transmission to sex partners during an ATI with detectable viral loads, may either trigger or exacerbate mental health problems, particularly anxiety [3–5].

Despite this uncertainty, an ATI is an important part of HIV cure research. As observed in PWH who naturally suppress HIV without ART (called elite controllers), high viral loads may need to be tolerated before control is ultimately achieved [6]. A short-term ATI (usually 2 – 12 weeks) is relatively safe; however, clinical risks are elevated with a longer ATI (e.g. lasting 3 months or longer), which is where new knowledge may be gained [7]. Further, the immunological, virological and clinical monitoring strategies and criteria for re-starting ART after an ATI are heterogeneous, which makes it nearly impossible to predict how a person’s immune system will respond during an ATI and what a person’s experience with an ATI will be like.

The 2019 Lancet HIV statement on the use of ATI co-authored by around 50 HIV cure research experts established that: “Monitoring of participants’ psychosocial experiences during ATIs is crucial. (…) [R]esearchers should also examine participants’ psychosocial tolerance for longer ATIs.” [2] Further, community advocates have issued recommendations to monitor psychosocial risks of a prolonged ATI, including unintended social harms like transmission to sex partners during an ATI when they are no longer virally suppressed [8].

Thus, there is a clear scientific need to investigate the potential risk for psychosocial harms with an ATI. PWH feel a tension between the altruistic desire to advance science and their own safety and that of sex partners [3, 5, 9–12]. There may also be risks that sustaining an undetectable viral load during an ATI triggers an erroneous belief of becoming HIV-free and create HIV transmission risks to sex partners following viral rebounds [11, 13]. Boston Patient B (also publicly known as Gary Steinkohl) noted experiencing emotional and psychological distress after finding out his HIV returned following a stem cell transplant which he believed had cured him [14]. Therefore, it is plausible that participants may experience heightened anxiety if they continuously worry about viral rebound [3, 5, 9–11]. Given the higher rates of mood disorders among PWH compared with the general population [15], and the established impact of HIV cure interventions on the central nervous system (e.g., neurological changes caused by experimental interventions crossing the blood-brain barrier) [16, 17], we sought to test for changes in anxiety symptoms in PWH before and during an ATI in HIV cure-related research.

Hypothesis

Surveying participants throughout two unique HIV cure-related studies, we hypothesized that mean levels on anxiety-related outcomes would fluctuate from before the ATI to over time during the ATI.

Methods

Overview

We present survey data from two HIV cure-related studies conducted by the same clinical research team at the University of California, San Francisco (UCSF). For ease of interpretation, we are labeling these two studies as: Study 1 – The Analytic Treatment Interruption (ATI) in HIV Infection (an observational HIV cure-related study) (NCT04359186) and Study 2 - The UCSF-amfAR trial, Combinatorial Therapy with a Conserved Element DNA Vaccine, MVA Boost, TLR-9 Agonist and Broadly Neutralizing Antibodies (NCT04357821) (an interventional HIV cure-related trial). All participants enrolled in Study 1 and 2 were eligible to participate in this longitudinal survey study. We selected these two ongoing studies being conducted at UCSF because they represented complementary ATI designs that allowed us to study anxiety across distinct yet related contexts. Embedding one harmonized survey across both studies at a single site with the same clinical research team and one clinical research coordinator who collected data for this research component aligned survey timing. This design allowed us to describe anxiety from before to during theATI and to explore whether these patterns differed by ATI context.

Study 1: ATI in HIV Infection (Observational Study)

Participants and Eligibility Criteria

Study 1 was a single-arm, uncontrolled, prospective observational study designed to characterize the interaction between the host and virus at the earliest stages of rebound and to identify biomarkers of viral rebound. Participants were age ≥ 18 years, on effective ART with viral loads below the level of quantification (generally <40 copies/mL) for at least 12 months, and screening CD4+ T cell count >350 cells/uL. Participants could not be taking a non-nucleoside reverse transcriptase inhibitor or a long-acting injectable ART. Individuals with chronic health conditions such as active viral hepatitis, significant cardiovascular, renal or hepatic impairment, or serious psychiatric or substance dependance were excluded from participation. Participants underwent four baseline on-ART measurements, plus optional procedures (i.e., leukapheresis, lymph node fine needle aspiration, and/or gut biopsy) prior to ART interruption. They then entered an ATI lasting up to 28 days with laboratory monitoring thrice weekly. The follow-up plan and restart criteria differed by whether a participant was previously known to be an HIV controller or non-controller. Controllers were PWH who had HIV RNA levels <2,000 copies/mL for a minimum of a year without ART, thus demonstrating natural viral ‘control’, whereas non-controllers did not meet this criterion. For prior controllers, the specific virologic criteria for ART restart were plasma HIV RNA level >50,000 copies/mL for 4 weeks, >10,000 copies/mL for 6 weeks, >2,000 copies/mL for 12 weeks, or >400 copies/mL for 24 weeks. Non-controllers were instructed to restart ART once a confirmed viral load test showed ≥ 200 copies/mL. As a result, most individuals in this protocol would resume ART prior to prolonged high-level viremia.

Study 2: UCSF-amfAR Trial (Interventional Trial)

Participants and Eligibility Criteria

Study 2 was a single-arm, open-label, non-randomized interventional trial that included multiple cure-related interventions. The goal was to induce durable ART-free viral suppression in PWH. The trial tested a combination of five interventions, including a HIV DNA vaccine, a MVA boost, a TLR-9 agonist, and a combination of 2 broadly neutralizing antibodies (bNAbs), administered over 34 weeks that showed safety and potential efficacy in pre-clinical research models, followed by an ATI lasting up to 52 weeks. Eligibility criteria included age 18 – 65 years old at trial screening, continuous ART for ≥1 year prior to entry, CD4+ T cell counts ≥ 500 cells/mL, and susceptibility to the trial’s bNAbs as determined by the PhenoSense Assay (Monogram Biosciences, Inc.). None of the participants were known to be HIV controllers prior to enrollment. Participants agreed to undergo weekly viral load monitoring. ART restart criteria included plasma HIV RNA level >50,000 copies/mL for 4 weeks, >10,000 copies/mL for 6 weeks, >2,000 copies/mL for 12 weeks, or >400 copies/mL for 24 weeks. Additionally, participants in both Study 1 and Study 2 were advised to restart HIV medications if they developed symptoms of acute retroviral syndrome, their CD4+ T cell count dropped to low levels (<350 cells/uL), or if at any point the participant or their primary care provider requested ART to be resumed.

Recruitment

During screening visits for Study 1 and 2, research staff asked participants if they were interested in enrolling in this longitudinal survey study. These discussions occurred after participants had already completed the informed consent processes. Those interested were given links for the baseline survey that included an Institutional Review Board (IRB)-approved electronic informed consent form. Participants received $30 as compensation for each survey and $100 bonus for completing all surveys.

Procedures and Survey Schedule

Given the intensity of Study 1 and 2 clinical protocols, both in terms of frequency of study visits and procedures, we designed our survey so that it could be quickly administered and completed. First, we reviewed the clinical research protocols and identified strategic and milestone timepoints in Study 1 and 2 relevant to our study (e.g., visit immediately before the start of the ATI). Survey data analyzed are from (1) baseline visit, (2) at the study visit immediately prior to the start of the ATI, and (3) during the ATI (weekly for 28 days and then monthly). During the ATI (3), participants’ anxiety was measured as long as the ATI lasted; thus, they contributed multiple data points during this time period. As noted in the Introduction and in description of Study 1 and 2 above, ATI lengths were approximate and could be shorter or longer than expected. Our survey study was developed independently from the clinical research team as the goal was to collect participants’ self-reported anxiety symptoms during each study. The clinical research team was responsible for recruiting participants into our survey study, sending reminders, and tracking completion as they had weekly contact with all participants going through these HIV cure-related studies.

From 2020 – 2023, online surveys could be completed remotely or during an in-person clinical research visit. Each survey took approximately 10–15 minutes and could be completed on a tablet in-person during their weekly study visit per Study 1 and 2 protocols, or via an Qualtrics^XM link the study staff e-mailed directly to study participants. The UCSF IRB approved the survey study.

Survey Measures

Demographics and Clinical Data

Baseline Anxiety

We used the 7-item Generalized Anxiety Disorder scale (GAD-7)[18,19] to assess baseline self-reported anxiety symptoms over the past two weeks, corresponding to the 14 days preceding the start of Studies 1 and 2. We calculated total scores by summing responses to all seven items. For instance, at baseline, we asked participants, “In the past two weeks, how often have you been bothered by the following problems?” One example item was “Feeling nervous, anxious, or on edge.” Responses ranged from 0 (Not at all) to 3 (Nearly every day), with total scores of 5 or higher indicating mild anxiety severity.

Follow-Up Anxiety.

For all subsequent survey assessments, we measured self-reported “state” anxiety using the abbreviated 6-item State-Trait Anxiety Inventory (STAI) [20–22], a measure of self-reported anxiety symptoms at the moment of assessment (rather than past 2-week symptoms with the GAD-7). We selected this instrument because it seeks to capture anxiety at the time of measurement at follow-up visits, rather than relying on the GAD-7 measure that would reflect symptoms from the prior two weeks. For example, participants read “indicate how you feel right now” (e.g., I feel tense) and rated each item from 1 (Not at all) to 4 (Very much). Scores of 6 or higher were considered mild severity.

Trait anxiety.

We measured trait – or a person’s disposition for experiencing anxiety – using the Big Five Inventory Neuroticism 8-item sub-scale [23]. For example, participants rated from 1 (Disagree strongly) to 5 (Agree strongly) items such as “I see myself as someone who gets nervous easily.” This was done to account for each participant’s propensity to self-report generalized and state anxiety symptoms. Total scores were the summation of all items.

Statistical Analysis

We chose an interrupted time series analysis for its ability to show patterns of predicted anxiety over time within each measure of self-reported anxiety, emphasizing relative change and associations with ATI-related factors, rather than absolute cross-measure differences. A priori, we had planned to pool data across Study 1 and Study 2 as early phase clinical studies have small sample sizes (N<30) and robust inferential analyses require larger Ns. Accordingly, we performed both a pooled analysis and separate analyses for each study, the findings of which are presented below. We conducted a descriptive analysis of the data, reporting means and standard deviations for continuous variables and frequencies for categorical variables. Then, we examined the correlation matrix of generalized and state anxiety scores and trait anxiety scores.

Next, we used an interrupted time series (ITS) model, which provides a straightforward and flexible method for longitudinal data. We used a piecewise random coefficient model, implemented using SAS’s PROC MIXED, as it is a robust method for modeling distinct changes over time, specifically before an interrupting event (i.e., “pre-ATI) and post the event (i.e., the start of the ATI). The “pre-post” term is used as a statistical term to designate scores before (“pre”) and during (“post”) the ATI. Mixed models permit tests of fixed effects through either maximum likelihood or restricted maximum likelihood estimation. These methods are superior to traditional repeated measures analysis and allow us to specify covariance (correlation) structures [22]. Thus, more appropriate covariance data structures can be analyzed. Specifically, we used ITS to test whether the slope of anxiety increased significantly during the ATI period. In these models, we could examine time (e.g., Does anxiety overall increase with time?), pre-post ATI (e.g., Is anxiety higher in the pre (before)- vs. post-ATI (during) period?) and interaction effects (e.g., Is there a difference in slope in the pre- vs post-ATI periods?). Specifically, time was modeled as the sequential order of observations (rather than calendar time), given that follow-up intervals varied across participants. Models included random intercepts to account for individual differences in baseline levels, and an ARH (1) covariance structure was specified to account for within-person correlation and unequal spacing between observations. Because some participants contributed a large number of observations, we capped the number of observations at ten per person to reduce extreme imbalance; a sensitivity analysis including all available observations produced comparable results.

We ran three models: 1) pooled data from Study 1 and 2, 2) Study 1 separately, and 3) Study 2 separately; all controlled for trait anxiety and study type. Because the number of observations during the ATI period varied by study, and to ensure equal weighting of participants, we capped the total number of observations at 10 per participant. We conducted all statistical analyses using SAS Version 9.4 of the SAS System for Windows, and we used the SAS PROC MIXED procedure to model the data and produce the effect estimates.

Results

Study 1 Participants: ATI in HIV Infection (Observational Study)

Of the 21 participants, fifteen completed the baseline survey, seventeen completed all three surveys, and one completed only two surveys. This resulted in a total of 107 participant observations over the entire study period. Participants were an average of 54.85 years old (SD = 11.9, min = 32, max = 75), predominantly male (3 female), and represented a range of incomes and educational backgrounds. Most participants were non-Latino (N = 15, 88.24%) (Table 1).

Table 1.

Demographic Characteristics of SCOPE-ATI and amfAR-UCSF Study Participants (2000 – 2023, San Francisco, United States)

Variables	SCOPE ATI (Study 1)	amfAR-UCSF (Study 2)
	M (SD) or N (%)	M (SD) or N (%)
Age	54.85 (11.9)	41.30 (8.99)
Gender ID man
Cis-gender Male	16 (80.00%)	9 (90.00%)
Cis-gender Female	3 (15.00%)	0
Transgender female	1 (5.00%)	1 (6.67%)
Income
< $25,000	8 (40.00%)	1 (11.11%)
$25,000 to $50,000	5 (25.00%)	3 (33.33%)
$50,001 to $75,000	1 (5.00%)	1 (11.11%)
$75,001 to $100,000	4 (20.00%)	1 (11.11%)
> $100,000	2 (10.00%)	3 (33.33%)
Latino	2 (11.76%)	5 (50.00%)
Race
White	14 (77.78%)	8 (100.00%)
Black	3 (16.67%)	0
More than one race	1 (5.56%)	0
Male sex at birth	17 (85.00%)	10 (100.00%)
Education
Some high school	1 (5.26%)	0
High school	4 (21.05%)	0
Some college	6 (31.58%)	2 (22.22%)
College degree	3 (15.79%)	3 (33.33%)
Graduate degree	5 (26.32%)	4 (44.44%)
Baseline Anxiety (GAD-7)	9.50 (3.6)	2.30 (2.4)

Note: In the SCOPE ATI Study, 1 participant did not respond to questions of age, gender, income, sex; 4 participants did not respond to the Latino ethnicity question; 3 participants did not respond to the race question; and 2 participants did not respond to the education question. In the amfAR-UCSF Study, 2 participants did not respond to the race question and 1 participant did not respond to the education question.

Participants mean anxiety scores were the following: baseline (GAD-7 measure: μ = 9.50, SD=3.6), Day 0 (prior to ATI, STAI measure: μ = 6.61, SD=4.2), and during the ATI period (STAI measure: M=6.26, SD=3.6). Over the course of the study, all participants reported at least mild anxiety symptoms at one point in the study.

Study 2 Participants: UCSF-amfAR Trial (Interventional Trial)

Ten UCSF-amfAR trial participants completed the baseline survey, and 8 completed all three surveys. This resulted in a total of 80 participant observations over the entire trial period. Participants were a mean age of 41.30 (SD = 8.9, min = 32, max = 55), primarily male, including one transgender woman, and represented a range of incomes and educational achievement. 50% of participants were Latino, and 8 participants were White. All participants were assigned male sex at birth (Table 1).

Participants’ mean anxiety scores were the following: baseline (GAD-7 measure: μ = 2.30, SD = 2.5), pre-ATI period (STAI measure: μ = 5.44, SD = 2.5), and during the ATI period (STAI measure: μ = 7.94, SD = 3.5). Over the course of the study, 7 out of 10 reported at least mild anxiety symptoms at one point.

Correlations

When data were pooled and separated by study, bivariate correlations among generalized and state anxiety scores, and trait anxiety scores were positively associated with one another. The correlations between trait and generalized anxiety between studies ranged from .65 (p=.001) and .86 (p=.001), and .36 (p=.0003) and .72 (p=.0001) with state anxiety.

ITS Models

Pooled ITS Model

Pooling all data, there was not a statistically significance effect of time, pre-post effects, and the interactions of time and pre-post effect were not significant (Table 2).

Table 2.

Interrupted Time Series Model Results

Model	Time	Pre-Post Effect	Interaction (Time X Pre-Post effect)
Pooled Analysis	β = −0.76	β = 0.96	β = 0.66
	SE = 0.77	SE = 1.05	SE = 0.80
Study 1 – SCOPE ATI Observational Study	β = −2.99	β = 2.77	β = 2.68
	SE = 0.86	SE = 1.42	SE = 0.92
Study 2 – amfAR ATI Combination Trial	β = 3.11	β = −2.54	β = −3.13
	SE = 0.56	SE = 0.92	SE = 0.59

Note. The term “pre” refers to anxiety scores prior to the start of the ATI. The start of the ATI is defined as the interrupting event in the interrupted time series model. Thus, the “post-effect” are anxiety scores during the ATI or after the start of the ATI.

Study 1 ITS Model

There was a statistically significant effect of time (β = −2.98, SE = 0.85, p=0.0008), as well as an interaction between time and pre-post (β = 2.76, SE = 1.4, p=0.06). Specifically, the pre-ATI slope (β = −0.30, SE = 0.1, p=0.04) was less steep than the post-ATI slope (β = −2.99, SE = 0.85, p=0.0008) (Table 2 and Figure 1). The significant interaction between time and the pre-post ATI effect indicated that while anxiety decreased over time, the post-ATI slope of decline was significantly steeper than the pre-ATI slope.

Fig 1.

Interrupted time series for Study 1 showing baseline anxiety scores elevated (mild level) prior to the start of the analytical treatment interruption (ATI) (solid line), declining, but remaining stably mild during the ATI period (dotted line). Time is coded to represent sequential follow-up visits with higher numbers indicated later visits. Baseline anxiety (Time = −2) was measured with Generalized Anxiety Disorder – 7 Scale, and follow-up anxiety was measured with State Trait Anxiety Inventory (Time = 2, 4, 6 and 8).

Study 2 ITS Model

Anxiety significantly increased over time (β = 3.11, SE = 0.6, p < .0001), and mean anxiety was significantly lower pre-ATI (μ = 3.79, SD = 2.9) than post-ATI (μ = 7.03, SD = 3.3) (β = −2.54, SE = 0.9, p = 0.0247). The interaction between time and pre-post ATI period was statistically significant (β = −3.12, SE = 0.6, p < .0001). Specifically, the anxiety slope was relatively stable pre-ATI, and significantly increasing post-ATI (β = 3.11, SE = 0.6, p=<.0001) (Table 2 and Figure 2).

Fig 2.

Interrupted time series for Study 2 showing state anxiety scores below mild threshold before the analytical treatment interruption (ATI) (solid line) and increasing during the period of the ATI period (dotted line). Time is coded to represent sequential follow-up visits with higher numbers indicated later visits. Baseline anxiety (Time = −2) was measured with Generalized Anxiety Disorder – 7 Scale, and follow-up anxiety was measured with State Trait Anxiety Inventory (Time = 2, 4, 6 and 8).

Discussion

Our findings indicate that self-reported anxiety was highly prevalent among PWH who went through an observational ATI and an interventional ATI in two HIV cure-related studies. In Study 1, anxiety levels were elevated at baseline but gradually decreased over the course of the study, though all participants at some point reached at least mild severity. In contrast, in Study 2, anxiety levels were initially below the mild threshold but increased significantly during the ATI, even when accounting for baseline trait anxiety. These novel findings provide preliminary empirical evidence that an ATI associates with heightened anxiety symptoms, both at baseline and as the ATI progressed.

It is well documented that mental health problems are highly comorbid with HIV [15]. Given the risks involved with ATIs (e.g., forward transmission during viral rebound) and background base rate of mental health problems in PWH, especially anxiety and depression symptoms, this unique feature of HIV cure-related studies may contribute to increased anxiety even under the close clinical monitoring and frequent viral load testing.. Despite regular assessments, the psychological burden of ATI persists, likely due to multiple contributing factors. First, for participants who have been on suppressive ART for years or decades, experiencing viremia—even transiently—represents a fundamental disruption to their understanding of U = U (Undetectable = Untransmittable). This shift may provoke emotional distress and uncertainty about their personal health and transmission risk. Second, the concerns and uncertainty of viral rebound for prolong periods off ART, though minimized through strict monitoring, remains a background concern [11]. Lastly, as noted earlier, PWH have high rates of anxiety and other mood disorders, with nearly one-third globally reporting such symptoms [24]. However, it is reassuring that in most cases, anxiety severity remained mild, for which clinical guidelines recommend simple monitoring and follow-up rather than immediate clinical intervention. Notably, methodological differences between the two studies preclude direct comparisons, and as discussed in the limitations, the heterogeneity of study designs and participant populations must be considered in interpreting these findings.

The observed differences in anxiety trajectories between the two studies likely reflect an interplay of demographic, experiential, and structural factors. Participants in Study 1 at baseline had elevated anxiety, which declined over time but remained in the mild category during the short ATI. Participants were predominantly older male PWH and from a long-standing, treatment experienced cohort in San Francisco. Participants in Study 2 at baseline had below mild anxiety, which increased during the extended ATI. Given the nature and intensity of the interventional clinical trial, participants were recruited throughout San Francisco and beyond and may have had to be generally slightly younger and healthier given the high number of inclusion criteria and screening steps for this unique trial. Indeed, screening criteria were more restrictive for Study 2 (e.g., specific requirements of viral suppression, CD4+ cell counts, and regimen restrictions) than Study 2, which is why we speculate the sample in Study 1 entered with highly prevalent mild anxiety as this reflects the broader base rate of anxiety in PWH. Given the extended ATI in Study 2, we believe this associated with the increase in mild anxiety we observed.

Beyond biomedical risks, an ATI affects participants’ lived experiences and perceptions of control.[25] The return of detectable viral load may provoke a sense of vulnerability, particularly in individuals who associate suppression with social acceptance and reduced stigma.[26] Additionally, the tension between the hope for a cure and the stress of interventional trial protocols has been previously documented in HIV cure research.[27] These findings reinforce the need for understanding secondary impacts of requiring PWH to pause effective ART.

Implications for Research and Ethics

Our findings emphasize the importance of embedding simple and cost-effective behavioral health protocols around HIV cure-related research. First, there could be brief pre-trial counseling that not only outlines the biomedical risks but also prepare participants for the psychosocial challenges of an ATI, equipping them with strategies to manage discussion with close networks and sex partners. Our team has developed a local protocol for such a discussion. Second, regular brief, 2 – 3 item mental health assessments during and after ATI can help detect and address participant distress in real time. Even among older, more experienced PWH, clear and transparent communication remains critical to maintaining trust and alleviating concerns.

From an ethical standpoint, these findings underscore researchers’ obligation to minimize psychological harm while pursuing scientific progress. Even mild anxiety can influence participants’ quality of life, decision-making, and retention in research. Implementing proactive support strategies, structured debriefing sessions, and enhanced monitoring, can ensure that an ATI adheres to ethical principles of respect, beneficence, and equity.

Limitations

While this study provides valuable insights, several limitations should be acknowledged. The relatively small sample sizes and reliance on self-reported anxiety measures may limit the generalizability of findings. Additionally, variations in study designs and participant demographics introduce potential confounding factors. As noted elsewhere, HIV cure-related studies are typically small in size and male-dominated with sub-optimal gender and racial/ethnic diversity in terms of the representativeness of populations who carry the greatest burden of HIV, which we also observed in these two studies and which limit generalizability. We also chose to use two different measures of self-reported subjective anxiety symptoms given our focus was on change over time rather than cross-measure differences, which makes it difficult to explore within-measure differences and does not allow for the easy interpretation of standardized scores. Thus, results should be interpreted as showing patterns of increases/decreases rather than within-person change. Lastly, future research should prioritize larger, more diverse cohorts and explore objective markers of psychological distress, such as stress-related biomarkers or neurocognitive assessments, to provide a more comprehensive understanding of the psychological impact of an ATI.

Conclusion

Our data suggest that an ATI can heighten anxiety symptoms in PWH in clinical research studies, especially given the high base rate prevalence of mental health challenges in this population. Future studies can easily integrate cost-effective repeat testing as an ethical safeguard to mitigate distress and enhance participant well-being in the context of cure-related clinical trials.