Delivering HIV-positive test results via phone: Who never answered? Correlates of Undelivered Reactive HIV Results in a U.S. National Cohort of Sexual and Gender Minority Individuals

Abstract

Self-testing for HIV is an accessible, effective testing method for clinical and research contexts. From 2017–2022, we conducted annual, mail-in HIV testing with participants in our U.S. national longitudinal cohort of 6,253 sexual and gender minority (SGM) individuals. Using multivariable logistic regression, we examined factors associated with failure to deliver HIV-positive results by phone despite extensive outreach. Of 16,000 samples tested, 315 were reactive (98% cisgender sexual minority men). We successfully delivered two-thirds of reactive results (n = 209). Unsuccessful result delivery was associated with frequent (once/week or more) methamphetamine use (OR = 2.195, p = .004), one or more recent HIV-positive sexual partners (OR = 2.764, p <.001), and at least one indicator of socioeconomic vulnerability (OR = 2.413, p <.001). In the adjusted model, only socioeconomic vulnerability (AOR = 1.864, p = .034) and recent HIV-positive partner(s) (AOR = 2.220, p = .005) remained significant. Finally, we performed an exploratory cross-sectional mediation analysis, which suggested the impact of frequent methamphetamine use on failure to deliver HIV results was mediated by socioeconomic vulnerability, with significant indirect effect (ab = −0.056, p = .002) and total effect (c = −0.184, p = .008). Although remote self-testing can help SGM access HIV testing, difficulties remain in successfully delivering stigmatizing and stressful information like HIV results—particularly to those using methamphetamine or experiencing socioeconomic vulnerabilities, and those with HIV-positive sex partner(s) who may suspect a positive result. Tailored strategies are needed to better connect these populations into the status-neutral HIV care continuum.

INTRODUCTION

HIV testing is a crucial way for individuals to learn their HIV status and connect to appropriate care. Those who test HIV-negative can be connected to prevention resources, such as Pre-Exposure Prophylaxis (PrEP), and those who test HIV-positive can begin antiretroviral therapy (ART) and access supportive services, such as those provided by the Ryan White program or Housing Opportunities for Persons with AIDS (HOPWA). Thus, HIV testing is required to enter the status-neutral HIV care continuum [1]. Regardless of the outcome, an HIV test “spurs action… by recognizing the opportunity created by a negative or positive result for an individual to achieve better health” [2]. Notably, being unaware of one’s HIV-status is a known contributor towards HIV spread, with undiagnosed individuals having a transmission rate about three to seven times higher than those aware of their HIV-positive status [3, 4].

Additionally, methamphetamine (meth) use plays a significant and growing role in HIV seroconversions among sexual and gender minority individuals (SGM) [5–11]. Cohort studies with HIV-vulnerable SGM have found high rates of substance use among their participants, such as poly-substance use and methamphetamine use [5, 12, 13]. Among our cohort in particular, we have found methamphetamine use to be associated with a substantially higher risk of HIV seroconversion. Indeed, one in three seroconversions in our cohort were among participants who reported using methamphetamine persistently [5]. Further, structural inequities amplify HIV vulnerability, and people who could benefit most from increased access to HIV testing are also among those most often separated from HIV prevention and care resources. Among SGM, the intersections of sexual minority stress, economic inequities, and substance use can amplify HIV vulnerability [14, 15], with those experiencing economic insecurity and intersectional stigma also experiencing higher HIV risk than their less-marginalized peers [16–20]. Not only are these factors associated with higher risk for HIV seroconversion but they are also associated with delayed entry into care [21–24]. As such, there remains missed opportunities to connect individuals from historically disinvested communities to HIV testing and entry into the status-neutral HIV care continuum.

Self-testing for HIV allows for biospecimen collection at home or in a private space, with rapid self-tests providing results within 20 minutes and mail-in self-tests requiring result delivery later after the test is processed by a laboratory [25]. One meta-analysis found that, among men who have sex with men and transgender (trans) individuals, self-testing increased the frequency of HIV testing compared to standard HIV testing, without significant adverse effects nor any decrease in linkage to HIV care [26]. Self-testing has been widely adopted in research and healthcare settings for its convenience and ease of access [25, 27–29], particularly following the social distancing efforts of the COVID-19 pandemic [30]. However, despite extensive public health literature highlighting the importance of HIV testing, fewer studies have focused on the communication of HIV-positive test results [31]. In particular, there is limited documentation of instances in which mail-in HIV test results were not successfully delivered to the tester, especially when the result was positive, and no studies to our knowledge that characterize differences between SGM who did and did not successfully receive their HIV-positive test result via phone. This reflects an important gap in the literature, as individuals who do not receive their positive mail-in results, particularly among vulnerable and marginalized populations, may remain disconnected from HIV care and the supportive services they need.

To address this gap, we leveraged data from the Together 5000 study, an internet-based, U.S. national 4-year cohort of sexual and gender minority people (2017–22), to examine who had their HIV-positive test result successfully delivered and who did not. We used at-home, mail-in HIV tests to facilitate accessibility and ease of use for participants, and to help connect participants to care in the case of positive test results. In the present analysis, we aimed to identify factors associated with the unsuccessful delivery of HIV-positive test results via phone by examining demographic, socioeconomic, and behavioral factors. Given the heightened role of methamphetamine in HIV seroconversions among SGM, both broadly and in our specific cohort, we anticipated that methamphetamine use may have impacted our ability to successfully deliver results. Additionally, we hypothesized that mental health factors might play a role in unsuccessful result delivery, given the potential for stress and avoidance in the event that one worried about, or anticipated, a positive result [32, 33]. Finally, we postulated that demographics (e.g., age, race and ethnicity, region) and socioeconomic factors could affect delivery of results through economic, cultural, or generational differences in answering phone calls from an unknown, institution-affiliated phone number.

METHODS

Study Population & Procedures

This study uses data collected as part of the Together 5,000 study, a U.S. national cohort of cisgender (cis) men, transgender (trans) men, and trans women who have sex with men, at increased risk for HIV (see Nash et al., 2019 [34] & Grov et al., 2019 [35] for full details). Baseline enrollments occurred October 2017 through June 2018 with participants recruited via geosocial sexual networking applications. Participants were followed annually for 48-months, through 2022. Study eligibility criteria included identifying as any gender other than a cis woman, being aged 16–49 years old, self-reporting as HIV negative or unaware of their HIV status, reporting sex with >2 male partners in the 90 days prior to enrollment, not currently taking PrEP, and having at least one clinical indication for PrEP based on the 2017 CDC recommendations [36]—such as engaging in condomless sex, injection drug use, methamphetamine use, or recent STI diagnosis (see Nash et al., 2019 [34]). A total of 6,253 participants were enrolled and followed over four years (2017–2022), with annual incentivized survey assessments at baseline, 12-, 24-, 36-, and 48-months.

At each annual assessment, study staff mailed participants an at-home specimen collection HIV test. Samples were returned to a laboratory for analysis in a prepaid envelope (see Nash et al., 2019 [34] for full details). Between baseline (2018) through the middle of the 24-month assessment period (Spring 2020), we used OraSure HIV-1 oral fluid biospecimen collection devices which were analyzed by the New York State Department of Health (Avioq HIV-1 Microelisa System). Due to specimen processing disruptions related to the COVID-19 pandemic, we switched to dried blood spot testing for the remainder of the 24-month assessment period (March-July 2020), analyzed by Molecular Testing Labs (HIV immunoassays). We returned to using OraSure HIV-1 oral fluid biospecimens for the 36-month and 48-month assessment periods, which were analyzed by the Alameda County Public Health Laboratory (HIV-1EIA), who also ran a supplemental assay (GEENIUS HIV-1/2) if the initial test was positive. It is important to note that, at follow-up assessments, participants who indicated that they had tested HIV positive between study visits (i.e., outside of the study) were not asked to participate in further testing and instead were asked to provide proof of diagnosis (i.e., digital photo of a document confirming their HIV status or photo of an HIV medication bottle with their name on it). Over the course of the study, about 84% of participants completed HIV testing each year, and over 16,000 samples were tested by the laboratory with 315 HIV positive results identified through laboratory analysis: 185 at baseline, 49 at 12-months, 34 at 24-months, 30 at 36-months, and 17 at 48-months (see Table 1).

Table 1.

Number of Delivered and Undelivered HIV-Positive Test Results Across Study Timepoints

Timepoint of HIV-Positive Result	Delivery Status
	Delivered	Not Delivered	Total n at timepoint
Baseline	121	64	185
12-Months	31	18	49
24-Months	20	14	34
36-Months	23	7	30
48-Months	14	3	17
Total n of delivery status	209	106	N = 315

Study staff emailed HIV-negative results to participants, and trained research staff delivered preliminary HIV-positive results over the phone. Participants were not informed beforehand how they would receive their HIV results (i.e., negative results via email, positive results via phone). Calls followed a study-created protocol (see Figure 1) designed to answer questions, provide emotional support, offer HIV education, and connect participants with local health and social services resources to facilitate confirmatory testing and linkage to care (see D’Angelo et al., 2021 [37] for more details). Given the highly confidential nature of the information and the potential for stress and clinical emergency, trained staff were prepared with local crisis resources (e.g., behavioral healthcare hotline, study psychologist), and opted to deliver results only via phone calls to ensure that participants were provided immediate support and linkage to care resources (i.e., as “warm a hand-off” as possible to ensure participants were connected to care). However, study staff did not disclose HIV-positive participant contact information directly to healthcare facilities nor to health departments, as study staff were exempt from mandatory reporting and our IRB prohibited the disclosure of participant contact information to protect confidentiality. We referred participants who tested preliminarily positive to follow-up with a provider for confirmatory testing and diagnosis, who was responsible for reporting the result following the participants’ local and state regulations.

Figure 1:

Study HIV-Positive Test Result Delivery Process

In total, we successfully delivered approximately two-thirds of the positive test results (see Table 1). In cases when we could not reach a participant, we made multiple follow-up attempts via email, text, and voicemail, indicating an urgent need to speak with the participant without specifying a reason. If a participant re-engaged with the study at a subsequent assessment timepoint, staff resumed efforts to deliver the results.

For the present analyses, we examined characteristic differences between participants for whom HIV-positive results were successfully delivered (n = 209) and for whom results could not be delivered (n = 106).

Study Measures

We used variables taken from participants’ survey response at the corresponding timepoint in which they tested positive for HIV (i.e., if they tested positive at 24-months, we used their 24-month survey responses), with the exception of race and ethnicity, which were only asked at the baseline survey.

Result Delivery:

Successful HIV-positive result delivery (yes/no) was our outcome of interest and was determined by study documentation indicating whether study staff had successfully reached the participant via phone and delivered the positive result to the participant.

Demographics:

Participants were asked a series of demographic questions on their annual survey assessments. Race and ethnicity were determined by participants’ baseline survey responses, and was categorized as “White,” “Black,” “Latine,” and “all other races and/or ethnicities.” Age in years was calculated using participants’ date of birth. Gender was taken from participants’ survey response and included the options cisgender male, cisgender female, transgender male, transgender female, and write-in something else. Sexual orientation was also taken from survey response options, which were: gay/queer/homosexual, bisexual, straight/heterosexual, and write-in something else. Region was created with participants’ mailing address, categorized into regions as defined by the U.S. Census Bureau. Income was determined by survey responses and combined into three commonly used income tiers [38] for simplicity of analysis: less than $29K, $30–74K, and over $75K annually.

Methamphetamine:

Participants were asked how frequently they used methamphetamine in the prior 3 months, with answers including: “never,” “once or twice,” “monthly,” “weekly,” and “daily or almost daily.” We combined responses to create a 3-category variable to better capture categorical differences between patterns of meth use: frequent use (daily or almost daily; weekly), infrequent use (monthly; once or twice), and no use.

Alcohol:

Hazardous drinking was assessed using the standard 3-item AUDIT-C [39] (reliability of measure at each survey timepoint ranged from α= 0.69 – 0.80). We coded scores as “Hazardous drinking” and “Non-hazardous drinking.”

Anxiety/Depression:

To assess anxiety and depression, we used the validated 4-question ultra-brief screening tool, the Patient Health Questionnaire (PHQ-4) [40] (reliability of measure at each survey timepoints ranged from α = 0.88 – 0.94). Scores were coded as “no anxiety/depression,” “mild,” “moderate,” and “severe,” which we then combined into: “moderate or severe anxiety/depression,” and “none or mild anxiety/depression” to observe potential differences among those most severely affected by anxiety and depression.

HIV-positive male partners:

Participants were asked to write in a number in response to the question, “In the last 3 months, how many of your male sex partners were HIV-positive?” Responses were organized into two categories: “None” and “1 or more”.

Socioeconomic (SES) vulnerability composite score:

Food insecurity was measured using a standardized 6-item scale, the U.S. Department of Agriculture’s Household Food Security Survey Module [41], coded “food secure,” “low food security,” and “very low food security.” For our analysis, we combined low and very low food security into a single category to simplify the analysis and capture all levels of food insecurity (1 = yes, 0 = no). We measured housing instability by asking whether a participant had experienced periods of unstable housing in the prior year (such as living in a car or shelter, couch surfing or being homeless) (1 = yes, 0 = no). Education was measured by asking participants’ highest education earned, which we coded as below high school education vs. high school diploma/GED or higher (1 = below a high-school diploma/GED, 0 = above a high-school diploma/GED). Given education level is broadly correlated with economic stability, overall health outcomes, and other social determinants of health [42], we wanted to assess the most vulnerable and thus the lower levels of education. We measured unemployment by asking participants about their current employment status. For this analysis, responses were categorized as “unemployed/ underemployed” (combining survey responses that capture lower financial stability: not currently working; permanently/temporarily disabled and not working; permanently/temporarily disabled and working ‘under the table’), and “employed and/or student” (combining responses that thematically capture relatively higher financial stability: working full-time; working part-time; going to school while working; going to school and not working). (1 = unemployed/underemployed, 0 = employed and/or student). Because the SES variables were weakly correlated with each other (except between food insecurity and lower education, r values ranged from 0.11 to 0.25, and p values ranged from p =.044 to p = <.001), we created a single composite variable of SES vulnerability markers to reduce unreliability of regression models. The composite score captured food insecurity, housing instability, lower education, and unemployment (scored 0 through 4).

Analysis Plan

Our primary objective was to investigate potential differences between participants who study staff could and could not reach by phone to inform them of their HIV-positive test result. We used frequencies and percentages to summarize descriptive characteristics of the study sample (see Table 2). We then used chi-squared tests of independence to compare participant characteristics between those we had and had not successfully delivered their HIV positive test result (see Table 3). Using the variables that were significant in bivariate analysis, SES vulnerability composite score (0 = no markers of SES vulnerability, 1 = 1 or more markers of SES vulnerability), recent sexual activity with an HIV-positive partner (0 = none, 1 = 1 or more recent HIV-positive partners), and frequent methamphetamine use (0 = infrequent or no use, 1 = frequent use), along with variables we believed could be confounders (age, region, race and ethnicity, anxiety/depression), we conducted a multivariable analysis using logistic regression, with adjusted odds ratio (AORs) and confidence intervals (CIs) reported (see Table 4).

Table 2.

Self-identified demographic characteristics of participants, N=315

	M ± SD or n	range or %
Age	32.23 ± 7.56	18–52
Recent HIV-Positive Partners	1.05 ± 2.65	0–25
Race and ethnicity*
White	134	42.5%
Black	59	18.7%
Latine	84	26.7%
Multiracial / something else	38	12.1%
Gender
Cis male	310	98.4%
Trans female	2	0.6%
Trans male	1	0.3%
Something Else (agender, nonbinary)	2	0.6%
Sexual Orientation
Gay, Queer, homosexual	274	87.0%
Bisexual	39	12.4%
Straight, heterosexual	1	0.3%
Something Else (pansexual)	1	0.3%
Region
Northeast	29	9.2%
South	174	55.2%
Midwest	38	12.1%
West	72	22.9%
Puerto Rico	2	0.6%
Housing Instability
Stable housing	202	64.1%
Housing instability	113	35.9%
Food Insecurity
Secure	146	46.3%
Food insecure	121	38.4%
Missing	48	15.2%
Income
≤29K	192	61.0%
30–74K	108	34.3%
≥75K	15	4.8%
Employment
Employed and/or student	248	78.7%
Unemployed / underemployed	67	21.3%
Education
Less than High School	14	4.4%
High School Diploma/ GED	222	70.5%
College Degree or more	79	25.1%
Recent Meth Use
None	189	60.0%
Once or twice	55	22.5%
Daily to weekly	71	17.5%
Hazardous Alcohol Use
Hazardous drinking	134	42.5%
Non-hazardous drinking	156	49.5%
Missing	25	7.9%
Anxiety / Depression
None to Mild	224	71.1%
Moderate to Severe	91	28.9%

^*Variable taken from baseline survey. All other variables are from survey assessment corresponding to the timepoint when the participant tested positive for HIV.

Table 3.

Bivariate comparisons those who had their HIV-positive status delivered (N=315)

	Delivered		Not delivered
	N = 209		N = 106
	n	%	n	%	chi-sq	p
Age					1.193	0.551
18–29	94	69.6	41	30.4
30–39	78	64.5	43	35.5
40+	37	62.7	22	37.3
Region					2.560	0.278
Northeast	23	79.3	6	20.7
South & Puerto Rico	116	65.9	60	34.1
West & Midwest	70	63.6	40	36.4
Race and/or Ethnicity*					4.101	0.251
Black	44	74.6	15	25.4
Latine	59	70.2	25	29.8
White	83	61.9	51	38.1
Multiracial / something else	23	60.5	15	39.5
Food Security					3.442	0.064
Food Secure	106	72.6	40	27.4
Food Insecure	74	61.2	47	38.8
Housing Stability					5.549	0.019
Stable Housing	114	66.3	58	33.7
Unstable Housing	65	57.5	48	42.5
Education Status					4.806	0.028
Less than HS	5	35.7	9	64.3
HS or more	204	67.8	97	32.2
Employment					10.188	0.001
Employed and/or Student	176	71.0	72	29.0
Unemployed / Underemployed	33	49.3	34	50.7
Recent Frequent Meth Use					8.326	0.016
Frequent meth use	37	52.1	34	47.9
Infrequent meth use	39	70.9	16	29.1
No meth use	133	70.4	56	29.6
Hazardous Alcohol Use					1.418	0.234
Hazardous drinking	94	70.1	40	29.9
Non-hazardous drinking	98	62.8	58	37.2
Anxiety/Depression					1.041	0.308
None to Mild	153	68.3	71	31.7
Moderate to Severe	56	61.5	35	38.5
Recent HIV-Positive Partner					15.969	<.001
None	155	74.2	54	25.8
1 or more	54	50.9	52	49.1

^*Variable taken from baseline survey. All other variables are from survey assessment corresponding to the timepoint when the participant tested positive for HIV.

Table 4.

Logistic Regression of factors associated with HIV-positive result NOT delivered

	OR	2.5 -- 97.5% CI	p	AOR	2.5 -- 97.5% CI	p
Age	1.02	0.989 – 1.051	0.215	1.008	0.973 – 1.044	0.653
Race
White	REF			REF
Black	0.555	0.274 – 1.080	0.090	0.671	0.314 – 1.390	0.290
Latino/x	0.690	0.381 – 1.228	0.212	1.003	0.525 – 1.906	0.993
Multiracial / Something else	1.061	0.500 – 2.206	0.874	1.287	0.580 – 2.813	0.529
Location
West/Midwest	REF			REF
Northeast	0.457	0.158 – 1.153	0.116	0.445	0.148 – 1.176	0.120
South / Puerto Rico	0.905	0.551 – 1.494	0.695	0.979	0.578 – 1.668	0.936
Anxiety/Depression
None/Mild	REF			REF
Moderate/Severe	1.347	0.807 – 2.232	0.250	0.963	0.545 – 1.677	0.895
Meth Use
None/Infrequent	REF			REF
Frequent	2.195	1.277 – 3.776	0.004	1.266	0.678 – 2.345	0.456
Recent HIV Positive Partners
None	REF			REF
1 or more	2.764	1.695 – 4.532	<.001	2.220	1.280 – 3.866	0.005
SES Vulnerability*
None	REF			REF
1 or more	2.413	1.461 – 4.066	<.001	1.864	1.052 – 3.348	0.034

OR: Odds ratio

AOR: Adjusted odds ratio

CI: Confidence interval

^*Composite variable made with the following markers of SES vulnerability: food insecurity (yes=1); housing insecurity (yes=1); unemployed (yes=1); education less than high school degree/GED (yes=1)

Finally, given the bivariate and multivariable results, along with our a priori hypothesis about methamphetamine’s potential importance in our analyses, we ran a simple exploratory, cross-sectional mediation analysis to investigate associations between frequent methamphetamine use (predictor), SES vulnerability (mediator), and result delivery (outcome) (see Figure 2). We did not explore other potential mediators. The analysis was performed with bootstrapping using the Mediation R package [43]. We report coefficients and p-values along with calculations of the direct effect for the predictor and outcome variables, indirect effect of the predictor and mediator variables, and total effect (direct + indirect effects). All analyses were completed in R version 4.3.1 [44].

Figure 2.

Socioeconomic Vulnerability Mediating the Relationship between Frequent Methamphetamine Use and Successful HIV-positive Result Delivery

RESULTS

Table 2 shows descriptive statistics of all participants who tested positive for HIV (n = 315) over the four years of cohort follow-up. The sample was an average age of 32 years old (SD = 7.6, range 18–52), and 42.5% were White (n = 134), 26.7% Latine (n = 84), 18.7% Black (n = 59), and 12.1% identified as multiracial or something else (n = 38). Most identified as cisgender men (98.4%, n = 310). There was one trans woman, one trans man, and two participants who reported their gender identity was something else. All participants endorsed having sex with men; 87.0% identified as gay/homosexual/queer (n = 274), 12.7% bisexual/pansexual (n = 40), and .03% straight/heterosexual (n = 1). Two-thirds of participants had no known HIV-positive male sexual partners (n = 209), and the rest had between 1 and 12 positive partners in the prior 3 months. Over half of the sample (55.2%) resided in the Southern U.S., 35% lived in the West or Midwest, 9.2% in the Northeast, and 0.6% in Puerto Rico. Additionally, 40% reported recent methamphetamine use in the past 3 months (17.5% indicated daily to weekly use, 22.5% indicated infrequent use), and 43.5% consumed a hazardous amount of alcohol for cisgender men. Finally, over one-third had recently experienced housing instability (35.9%) and food insecurity (38.4%), and 21.3% were unemployed.

Table 3 shows bivariate associations with test result delivery category. Study staff were more successful in delivering results to participants who reported having never used methamphetamine (70.4% delivered) or infrequently used (70.9% delivered) in the prior three months, compared with those who frequently used methamphetamine (55.8% delivered), χ2 = 8.33, p = .016. Additionally, participants who reported no known HIV-positive male partners in the prior 3 months had more successful deliveries (74.2% delivered) compared to those with 1 or more recent HIV-positive partners (50.9% delivered), χ2 = 15.97, p <.001. More participants who were employed or a student successfully received their HIV test result (78.7% delivered) compared to those who were unemployed/underemployed (49.3% delivered), χ2 = 10.19, p = .001. Housing instability (χ2 = 5.55, p = .019) and lower education (χ2 = 4.81, p = .028) were also associated with not receiving results, and participants who were food insecure trended toward not having their results delivered (χ2 = 3.44, p = .06). Age, race, geographic region, and anxiety/depression were not significantly associated with result delivery.

Next, we performed a multivariable logistic regression model to examine the factors associated with unsuccessful result delivery (Table 4). Participants with one or more markers of socioeconomic (SES) vulnerability (housing instability, food insecurity, unemployed/ underemployed, and lower education) had lower odds of receiving their HIV-positive result after considering other confounding factors (AOR = 1.864, p = .034). Additionally, those with one or more recent HIV-positive sexual partners had lower odds of receiving their result (AOR = 2.220, p = .005) than those without any recent HIV-positive partners. Although frequent methamphetamine use was statistically significant in bivariate analysis (OR = 2.195, p = .004), it was not significant in the multivariate model (AOR = 1.266, p = .456).

Finally, we used a simple, exploratory, cross-sectional mediation analysis to gauge the potential relations between frequent methamphetamine use (predictor) with result delivery (outcome) as mediated by SES vulnerability (mediator). As seen in Figure 2, frequent methamphetamine use was significantly associated with higher SES vulnerability (β = 0.367, p < .001), and was negatively associated with the ability to successfully deliver HIV test results (β = −0.184, p = .003). Higher SES vulnerability was also negatively associated with successfully delivering HIV results (β = −0.187, p < .001). The mediation (indirect) effect of SES on the relationship between frequent methamphetamine use and result delivery was statistically significant (indirect effect = −0.056, [95% C.I. −0.099, −0.020], p = .002). Without involvement of SES vulnerability, there was no direct effect of frequent methamphetamine use on delivery (direct effect = −0.128, [95% C.I. −0.263, 0.020], p = .080). However, the total effect (indirect effect + direct effect) of frequent meth use, inclusive of the impact of SES vulnerability, was significantly associated with the inability to deliver HIV-positive test results (total effect = −0.184, [95% C.I. −0.324, −0.040], p = .008).

DISCUSSION

We hypothesized that, among our cohort of sexual and gender minority participants who tested HIV-positive through the study, there would be significant demographic and behavioral characteristics distinguishing participants for whom we could and could not reach to deliver their HIV-positive test result. Indeed, we found that socioeconomic status (SES) significantly impacted whether a participant learned of their HIV-positive test result through our research study and thus could enter the HIV care continuum. Participants with at least one marker of socioeconomic vulnerability had lower odds of receiving their HIV-positive test result through the study, even when controlling for factors like age, race and ethnicity, geographic region, and moderate/severe anxiety/depression. Our results are corroborated by extant literature documenting the impact of SES on HIV care, with those experiencing socioeconomic vulnerabilities more likely to have delayed entry into care [21–23] as well as more likely to have gaps in their anti-retroviral therapy (ART) regimen [21, 45, 46].

Additionally, participants who reported having used methamphetamine at least once per week had lower odds of receiving their HIV-positive test results. However, this association was no longer significant in our adjusted model. Because of the consistent role of methamphetamine use in the HIV epidemic, particularly among SGM [5–8, 47], and its prevalence among our participants, we believed that evaluating the potential role of methamphetamine use on result delivery was especially timely. Based on our cross-sectional data and exploratory mediation analysis, our results suggest that the impact of frequent methamphetamine use on delivering HIV results was mediated through SES vulnerability. Indeed, frequent methamphetamine use can lead to increased opportunities for economic vulnerabilities through several contexts, such as engaging in survival sex [48], difficulty maintaining employment [49–52], drug-related psychosis [53, 54], dropping out of school [51, 55], reduced food security [48, 56], and difficulty maintaining stable housing [48, 57–60]. Our findings suggest that these interacting pathways—from frequent methamphetamine use to SES vulnerability—may have affected participants’ chances of receiving their HIV-positive results.

Although our undelivered results could represent undiagnosed HIV infections, participants in our “undelivered results” group may have actually been aware of their HIV status through alternate means. Participants could have tested outside the study, and even started engaging in care, but never informed research staff. Indeed, we found that those with at least one recent HIV-positive sexual partner were significantly less likely to have their HIV result delivered, suggesting participants may have assumed their status and potentially deprioritized answering the study’s delivery phone call attempts. Although participants were given gift cards for completing survey assessments and sending biospecimens for HIV testing, they received no incentive to speak with study staff. Furthermore, the stress, overwhelm, and psychosocial effects of a new HIV-diagnosis [61–63] may have impeded participants’ abilities to engage in study activities, like responding to delivery attempts. Similarly, the cognitive effects of frequent methamphetamine use [64, 65] and the impacts of SES vulnerabilities, like difficulty maintaining phone access, could have also impacted participants’ ability to receive their test results from study staff.

Undelivered results could also represent avoidance of the potentially psychologically threatening information of a suspected HIV-positive test result [32, 33]. Indeed, avoidance of HIV testing has been linked to fear of HIV/AIDS and related stigma, and it has been documented in SGM who have lower education levels and lower levels of social support [22, 32, 66–68]. Although self-perception of HIV risk may have also influenced participants’ suspicions of a positive HIV result [16, 69], we found no association between result delivery and number of recent sexual partners, recency of previous HIV test, recent condomless anal sex, or recent positive STI test (results not shown). Furthermore, in addition to the potential for stigma related to infectious diseases, the high rates of spam phone calls may also contribute to individuals ignoring calls from unknown numbers, a phenomena at play in some HIV, syphilis, and COVID-19 related contact tracing efforts [70]. Although the study’s phoneline was set up with caller-ID associated with our institution to minimize this potential, participants may have ignored calls and texts from an unknown number believing they were spam; the potential for methamphetamine-induced paranoia [71–73] could have also amplified these concerns.

Our findings fit into the well-documented pattern that those who are most vulnerable to HIV are also more likely to have delayed entry into the status-neutral HIV care continuum [7, 9–11, 20, 23, 74–78]. This is particularly concerning given that undiagnosed individuals are estimated to have a transmission rate three to seven times those aware of their HIV-positive status, and thus contribute significantly to new infections in the United States [4]. However, our research environment may not reflect real-world knowledge of HIV status. For example, use of substances other than methamphetamines and alcohol have been associated with delayed entry into the HIV care continuum [79], but we found no significant association between result delivery and frequent use of cocaine, gamma-hydroxybutyrate (GHB), inhaled nitrites (“poppers”), or misuse of prescription opioids in our sample (results not shown). Thus, additional studies should continue trying to identify factors associated with knowledge of HIV status and engagement in HIV care.

Overall, this study highlights the challenges of reaching individuals to inform them of their HIV-positive test result in the context of a research study. Simply being tested does not necessarily equate to learning one’s HIV status, especially if follow-up action is required. Even before the widespread use of at-home self-testing and remote disclosure of HIV results, “failure to return” studies found that men who have sex with men, people who use injection drugs, those with lower education, and those with anxiety about HIV were less likely to return to their testing site and receive their HIV results, regardless of whether the result was positive or negative [19, 80]. Although remote self-testing has helped facilitate connection to HIV prevention and care beyond clinical settings, especially for marginalized populations [81], barriers remain in ensuring individuals receive their HIV results and connect to care. As one of the first studies to assess characteristic differences between successful and unsuccessful HIV-positive result delivery by phone, our findings highlight the sub-populations of sexual and gender minority individuals who may be less likely to receive their HIV-positive test result within the current landscape of HIV testing.

Finally, our findings add to the body of work suggesting that reducing barriers of entry into the status-neutral HIV care continuum could support the efforts of reducing HIV infection and spread. Free, rapid at-home HIV testing [68, 82, 83] and other targeted strategies [68, 84, 85]—like using phone applications, implementing targeted outreach like bringing testing resources to individuals in their own communities, and/or including financial incentives for receiving results—could be particularly beneficial for SGM who may be avoidant of or unable to prioritize HIV testing, such as those who use methamphetamine, experience SES vulnerabilities, or have HIV-positive partners. Additionally, supportive services like substance use harm-reduction and efforts to mitigate economic vulnerabilities could be powerful strategies to reduce the risk of HIV seroconversion [86–88] and provide alternate points of entry into the status-neutral care continuum. Despite these resources often being prioritized for HIV-positive individuals, assistance with housing [89–91], food access [92, 93], household economic strengthening and cash support [94–97], education assistance [98, 99], methadone clinics [100–102], syringe exchange [87, 90, 103], safer injection sites [89, 98], and other structural changes have all shown to be effective HIV-prevention methods [104]. Thus, increasing access to testing and supportive resources, including intentional strategies to facilitate successfully delivering test results, could provide a tailored approach to increasing entry into the status-neutral HIV care continuum for these important but harder-to-reach populations, including SGM who use methamphetamine or experience SES vulnerabilities. Additional research is needed to document whether such strategies, including targeted communication, testing, and structural support, increase the success rates of HIV positive deliveries among these populations.

Limitations

It is important to acknowledge that methamphetamine use and SES vulnerabilities can be interrelated and cyclical, with methamphetamine as a coping strategy to deal with the stress of living with SES vulnerabilities [15, 48, 105, 106], and frequent methamphetamine use worsening economic vulnerabilities [48, 53, 58, 60]. Our mediation model could not account for these nuances, as the model is limited to unidirectional pathways. We used the body of work documenting the impact of SES and methamphetamine use on HIV vulnerability [7, 9–11, 20, 74–78] to help inform the model creation, and used the model for exploratory purposes. Nonetheless, there could have been other predictive pathways that we did not measure in the study, such as psychological or cognitive impacts associated with methamphetamine use [52, 64, 65]. By design, comparative differences between timepoints were not assessed given the relatively small sample sizes at each timepoint. Instead, we analyzed participant experiences at the timepoint they seroconverted, which allowed us to understand the recent experiences of participants when they tested positive. However, it cannot be discounted that contextual factors outside of the study (such as the COVID-19 pandemic in 2020) may have impacted participants’ behavior. Additionally, due to disruptions in testing resources during the early COVID-19 pandemic, the study briefly used dried blood spot testing in addition to oral fluid testing for HIV. Although each type of test has its own sensitivity and specificity, these differences likely had minimal effects in testing results [107]. The downward trend in seroconversion count at each timepoint was consistent with the cohort study design, in which as participants tested positive for HIV, there was a smaller pool of HIV-negative participants who could seroconvert at each subsequent timepoint. Although overall retention in study activities remained around 77% each year, loss to follow-up over time could have affected our results, particularly given the potential for reduced retention amongst those who use methamphetamine [108]. Some participants may have also seroconverted but never returned a biospecimen sample for testing, which this analysis couldn’t account for.

Finally, staff engaged critically with the ethical dilemma of undelivered positive results throughout the study, meeting regularly as a team and one-on-one with a clinical psychologist to debrief about delivery calls and discuss unsuccessful delivery attempts. Staff balanced the nuanced considerations of disclosing sensitive and confidential health information to participants alongside that of participant autonomy and their ability to choose whether to engage with this aspect of the study [109, 110]. The decision to stop attempting result deliveries after reaching out several times across multiple modes of communication was made with care, with the researchers prioritizing what they believed was best for participants within the multifaceted considerations of: participant consent in a research study; the mental health impacts of receiving stressful health-related information and the importance of peer support; abiding by HIV-status confidentiality; and the individual and population-level implications of knowing one’s HIV status.

Conclusion

We found that individuals who used methamphetamine at least once a week or more, mediated by the effects of socioeconomic vulnerabilities such as food and housing instability, lack of higher education, and unemployment, were less likely to respond to study staff and receive their HIV-positive lab test result in our study. Additionally, those with at least one HIV-positive male sexual partner within the prior 3 months were less likely to receive their HIV-positive result, suggesting some may have been suspicious or aware of their own HIV status. Remote testing is an important tool to connect sexual and gender minority individuals to HIV-testing through non-clinical settings, especially given inequities in accessing HIV-related resources. However, our results highlight the challenges of informing individuals of an HIV-positive test result in the context of a research study. As such, tailored approaches to entering the status-neutral HIV care continuum could be particularly important for harder-to-reach populations like those who use methamphetamine and experience socioeconomic vulnerabilities.

Figure 1 is a model of the study’s HIV-positive test result delivery process. First, the lab sent de-identified HIV results via secure communication. Using the study’s password-protected database, staff noted the participant’s contact information for each HIV-positive result. Then, study staff researched local confirmatory testing resources (e.g., local health departments, STI clinics, community health centers); supportive services that may be needed (i.e., peer support groups, LGBTQ+ community resources); as well as crisis resources (e.g., mobile behavioral health crisis unit, community crisis hotline). When possible, staff also called local testing resources to clarify relevant testing information (e.g., price of test, hours of operation, kinds of testing offered, whether treatment was offered) and to identify clinic staff that could help guide someone through the testing process. Next, staff would attempt to call the participant to deliver the results. If a participant did not respond, staff would leave a voicemail, as well as email and text the participant indicating a desire to speak. Continued efforts of calling, texting, and emailing would ensue, however, if the participant never responded this constituted a result that was not delivered. If the participant answered at any point in time, staff would follow a script and deliver the test results, constituting a successful result delivered. Elements of the result delivery call script are outlined in the box in the top right corner, and include: confirming participant identity, disclosing the HIV-positive test result, validating and providing space for emotions, assessing for clinical emergency if needed (i.e., probing suicidal and homicidal ideation), providing information about HIV, debunking misconceptions, answering questions, emphasizing the importance of follow-up confirmatory testing and linkage to care, providing local resources, encouraging reaching out to support systems, and helping create a plan of next steps.

Figure 2 is a model depicting socioeconomic status (SES) vulnerability as a mediator for the relationship between frequent methamphetamine (meth) use and ability to successfully deliver an HIV-positive test result. Arrow “a” shows frequent meth use was significantly positively associated with SES vulnerability (a = 0.367, p < .001). Arrow “b” shows SES vulnerability was significantly negatively associated with delivering an HIV-positive result (b = −0.186, p < .001); those with more markers of SES vulnerability — and thus lower socioeconomic status — were less likely to receive their HIV-positive test result. The striped arrow “c’” shows that frequent meth use’s direct effect on successful HIV-positive result delivery, without accounting for SES vulnerability, was not significant (c’ = −0.128, p = .080). The “indirect effect” box highlights that SES vulnerability’s indirect/mediating effect on the relationship between frequent meth use & successful HIV result delivery was significant (ACME = −0.056, p = .002), and the “total effect” box shows that the total effect (indirect + direct) was also significant (TE = −0.184, p = .008).