Frequent Missed Opportunities for Earlier HIV Diagnosis in a Routine Opt-out Testing Environment in Atlanta

Abstract

Background

Routine, opt-out HIV testing strategies aim to diagnose HIV earlier and decrease ongoing transmission. We report missed opportunities (MO) and missed testing opportunities (MTO) for earlier HIV diagnosis in a safety-net healthcare system using routine opt-out testing.

Methods

This retrospective study analyzed adults diagnosed with HIV between 2015 and 2022 with eligible encounters 30–365 days before diagnosis. MO is defined as no HIV testing in the year before diagnosis. MTOs are encounters where testing was indicated but not performed. Logistic regression identified factors associated with MO and MTO. To evaluate the opt-out testing program, we measured the number of individuals eligible for testing, tests completed, reactive results, and new HIV diagnoses.

Results

Of 713 newly diagnosed individuals, 499 (70%) had MO. Among 1845 encounters, 1235 (67%) were MTO. Sexual health–related encounters and STI testing had lower MO odds (adjusted odds ratio [aOR] 0.62; 95% confidence interval [CI], 0.42–0.90; P = .013) and (aOR 0.36; 95% CI, 0.28–0.48; P < .0001), respectively. MO was associated with higher odds of CD4 < 350 cells/mm³ at diagnosis (aOR 1.8; 95% CI, 1.2–2.9; P = .011). Outpatient encounters, particularly primary care, had higher MTO odds than emergency department (OR 0.67; 95% CI, 0.54–0.82; P < .0001). Among 531,848 eligible individuals, 357,771 tests were conducted in 199,004 individuals (37.4%); 4719 (1.3%) tests were reactive, resulting in 861 (0.4%) new HIV diagnoses.

Conclusions

Despite routine opt-out HIV testing, most individuals newly diagnosed with HIV had no testing in the prior year, highlighting the need to optimize opt-out testing procedures, particularly in primary care and nonsexual health visits.

Despite routine opt-out HIV testing, missed opportunities for earlier diagnosis remained prevalent among newly diagnosed individuals and were associated with advanced HIV. Optimized HIV testing, particularly in primary care and nonsexual health visits, is imperative to facilitate diagnosis and improve outcomes.

BACKGROUND

In the United States, individuals with undiagnosed HIV account for 13% of all people with HIV (PWH) [1] but contribute to more than 40% of all new transmissions [2, 3], highlighting the need to improve HIV testing interventions.

Earlier HIV diagnosis via expanded testing is a primary pillar of the Ending the HIV Epidemic Initiative [2]. The US Centers for Disease Control and Prevention recommends 1-time opt-out HIV testing for all individuals aged 13–64 years without assessment of risk (i.e., routine opt-out HIV screening) and annual screening for persons with a higher likelihood of HIV acquisition [4–6]. Healthcare systems in the United States, including Grady Health System (GHS) in Atlanta, have adopted routine opt-out HIV screening. This method integrates HIV testing into standard care, offering it to patients by default unless they decline, unlike traditional opt-in testing, which requires patients to actively request the test.

Despite these efforts, recent studies have identified missed opportunities (MO) for earlier HIV testing among newly diagnosed PWH [2, 7–15]. Studies conducted in urban academic medical centers with similarities to GHS have demonstrated that persons perceived at a lower HIV risk are at higher risk for MO and that missed testing opportunities are most common in the emergency department (ED) [2, 8, 9 , 14]. Prior analyses of MO have been limited by small sample sizes [12] and varying definitions of MO that often do not consider testing indications [7–9, 11–14], and may be impacted by confounding factors related to healthcare access and overutilization [2, 7–11, 14].

We aimed to identify missed opportunities for earlier HIV diagnosis within our high-volume safety-net healthcare system in Atlanta, Georgia. Using a novel definition of missed opportunities that considers healthcare utilization and testing indications, we sought to quantify and characterize missed opportunities, identify associated risk factors, and evaluate the impact of MO on people diagnosed with HIV. To contextualize our findings, we analyzed the testing rate and diagnostic yield of our routine opt-out HIV screening program.

METHODS

Study Setting and Data Source

This retrospective cohort study was conducted at GHS, an urban safety-net hospital in Atlanta, Georgia. The center features a Level 1 trauma unit, handling 250,000 ED encounters and 120,000 primary care visits annually. It serves a catchment area of >6 million people living in Fulton and Dekalb counties [16], which are 2 of the 42 Ending the HIV Epidemic Initiative counties with the highest incidence of HIV in the United States [17].

In 2013, GHS implemented routine opt-out HIV testing in the ED and urgent care, and by 2017, this initiative was expanded to outpatient and inpatient settings. In the ED, an automatic electronic health record (EHR)-based best practice alert (BPA) was activated if an individual had not completed HIV testing in the prior 6 months. This alert prompted triage staff to offer HIV screening using normalizing language—emphasizing that testing is a routine and expected part of medical care to reduce potential discomfort or stigma—and to obtain verbal consent, in compliance with Georgia statute [18]. If the HIV test was not ordered, a second BPA was triggered when the provider accessed the individual's chart. In outpatient and inpatient settings, the BPA would activate if the last HIV screening occurred more than 1 year prior. Between 2020 and 2021, the GHS ED prioritized COVID-19–related BPAs, which resulted in the HIV screening BPA being moved from primary to secondary screening.

Data were extracted from the GHS Epic EHR database. Extracted data included patient demographics, encounter details, laboratory tests and results, and diagnosis codes.

Patient Consent Statement

This research was approved by the Emory University institutional review board and the GHS Research Oversight Committee with a waiver for informed consent. No patient images or otherwise patient-identifying information are included in this manuscript.

Study Population

All adult individuals with at least 1 encounter at GHS between 01 January 2014 and 31 December 2022 were included. PWH were defined as anyone in the cohort with a diagnosis of HIV per the Georgia Department of Public Health [19]. Eligible testing encounters were defined as provider-based encounters within departments where opt-out testing was implemented (e.g., inpatient, emergency, urgent care, selected outpatient settings, including primary care, women's health, dermatology, medical subspecialty clinics). Visits for nursing encounters, procedures, imaging, therapy, and surgical specialties were excluded since opt-out testing was not implemented in these settings. PWH with an HIV diagnosis date between 2015 and 2022 and at least 1 eligible GHS healthcare encounter 30–365 days before their diagnosis were included in the analysis for missed opportunities (Figure 1). Encounters within 30 days of the HIV diagnosis date were excluded to avoid miscategorizing the GHS encounter where HIV diagnosis occurred as an MO. To prevent overcounting of linked encounters (e.g., an ED encounter that led to an admission), encounters within 3 days of each other were named “encounter groups” and analyzed as a single encounter.

Figure 1.

Cohort definitions for individual-level and encounter-level analyses and analysis of opt-out testing system. Abbreviations: ED, emergency department; GDPH, Georgia Department of Public Health; GHS, Grady Health System; IP, inpatient; OP, outpatient; UC, urgent care.

To evaluate our opt-out testing program, we included all people eligible for HIV testing at GHS, defined as persons not living with HIV before 2014 who had at least 1 eligible testing encounter during the study period.

Primary Outcome: Missed Opportunity at the Individual Level

The primary study outcome was MO, defined as people who did not receive HIV testing at any eligible encounter group 30–365 days before their HIV diagnosis date.

Secondary Outcome: Missed Testing Opportunity at the Encounter Level

An encounter was considered a missed opportunity for HIV testing (MTO) if it occurred 30–365 before an HIV diagnosis and HIV testing was indicated but not performed. Testing was considered indicated if at least 1 of the following conditions were met: the encounter (1) occurred more than 6 months after the last HIV test or (2) included a documented sexual health–related diagnosis or chief complaint. These conditions were chosen based on GHS BPA firing parameters (criteria 1) and US Centers for Disease Control and Prevention testing guidelines for testing and prior studies (criteria 2) [14, 20].

Variable Selection

We cleaned the extracted EHR data using R version 4 statistical software [21] and created variables encompassing sociodemographic, healthcare utilization, and clinical information. The selection of variables was guided by an extensive literature review and collaboration between clinicians and biostatisticians to ensure their relevance. A complete list of variables and the data dictionary can be found in the Supplementary material. We generated a correlation heatmap to address multicollinearity, identifying pairs of variables with high correlation values. Only those variables with sufficient data and without multicollinearity were retained for further analysis.

Statistical Analysis—Primary Outcome, MO at the Individual Level

All statistical analyses were performed in R version 4 statistical software [22]. Descriptive statistics were computed for the overall study population and stratified by the presence of MO. Categorical variables were summarized using frequencies and column percentages, wheres continuous variables were expressed as means with standard deviations. Using univariable and multivariable logistic regression, we estimated odds ratios (OR), 95% confidence intervals (CI), and P values for the association between each descriptive variable and MO. Statistical significance was set at P < .05 for all analyses. Age, race, sex, number of encounters in the 30–365 days before diagnosis, and HIV testing in the 2 years before diagnosis were identified as potential confounders and controlled for in the multivariable analysis.

Statistical Analysis—Secondary Outcome, Encounter-level MTO

We stratified all relevant encounters by the presence of an MTO. Univariable logistic regression was again employed to analyze the association between encounter-level descriptive variables and MTO (secondary outcome), presented as OR (95% CI). We used Firth's logistic regression for more reliable parameter estimation to address potential bias caused by predictors that perfectly separated groups (e.g., when 1 group had either 0% or 100% of the outcome). Statistical significance was set at P < .05 for all analyses.

Analysis of Routine Opt-out HIV Testing Program

In a secondary analysis, we identified the number of individuals who were eligible for HIV testing and how many of those received HIV testing at GHS during our study period. To determine the prevalence of positive test results and the number of people newly diagnosed with HIV, we identified all accurate positive tests and individuals with at least 1 positive HIV test. A positive HIV test was defined as a positive Architect HIV Antigen/Antibody Combination Assay. New diagnoses of HIV were defined as individuals with positive HIV testing at GHS occurring within 30 days of the Georgia Department of Public Health diagnosis date.

RESULTS

Primary Outcome—Individual-level Missed Opportunities (MO)

Among PWH diagnosed during the study period per the GDPH database, 713 individuals had at least 1 eligible encounter at GHS in the 30–365 days before diagnosis and were included for analysis of missed opportunities. Of the 713 individuals, 54% were aged 20–34 years (median 32 years; interquartile range [IQR] 24 years), 66% (471/713) were male, and 91% (629/713) were Black. Of these individuals, 70% (499/713) met the MO definition (i.e., they had no HIV testing at any encounter in the year preceding HIV diagnosis, which consisted of 1098 encounters, including 501 outpatient, 551 ED/urgent care, and 46 inpatient encounters). Individuals with MO had a median of 1.0 encounters (IQR 1–3) the year before diagnosis (Table 1).

Table 1.

Characteristics of Individuals Diagnosed With HIV in Health System 2015–2022

Demographics, n (%)	Cohort (n = 713)	Missed Opportunitya		Univariable		Multivariable
		No (n = 214, 30%)	Yes (n = 499, 70%)	OR (95% CI)	P Value	OR (95% CI)	P Value
Age at diagnosis, y
20–34	383 (54%)	142 (66%)	241 (48%)	Ref		Ref
35–49	156 (22%)	41 (19%)	115 (23%)	1.7 (1.1–2.5)	.017	2.1 (1.3–3.4)	.002
50+	174 (24%)	31 (15%)	143 (29%)	2.7 (1.8–4.3)	<.0001	3.4 (2.2, 5.7)	<.0001
Legal sex
Male	471 (66%)	146 (68%)	325 (65%)	Ref		Ref	…
Female	242 (34%)	68 (32%)	174 (35%)	1.1 (0.82–1.6)	.43	1.4 (0.93–2.0)	.12
Raceb
White	45 (6.5%)	11 (5.1%)	34 (6.8%)	Ref		Ref
Black	629 (91%)	189 (88%)	440 (88%)	0.75 (0.36–1.5)	.43	0.92 (0.42–1.9)	.83
Other	14 (2.0%)	6 (2.8%)	8 (1.6%)	0.43 (0.12–1.6)	.19	0.75 (0.19–3.0)	.68
Ethnicityb
Hispanic	25 (3.5%)	8 (3.7%)	17 (3.4%)	Ref		Ref
Non-Hispanic	677 (95%)	201 (94%)	476 (95%)	1.1 (0.45–2.5)	.80	1.2 (0.20–6.0)	.85
TGWb	12 (1.7%)	6 (2.8%)	6 (1.2%)	0.42 (0.13–1.4)	.14	0.99 (0.28–3.6)	.99
MSMb	51 (7.2%)	17 (7.9%)	34 (6.8%)	0.85 (0.47–1.6)	.59	1.2 (0.60–2.4)	.66
Black MSMb	40 (5.6%)	12 (5.6%)	28 (5.6%)	1.0 (0.51–2.1)	.99	1.3 (0.63–2.8)	.47
Black Femaleb	216 (30%)	61 (29%)	155 (31%)	1.1 (0.80–1.6)	.50	0.85 (0.21–3.2)	.89

Abbreviations: CI, confidence interval; MSM, men who have sex with men; OR, odds ratio; TGW, transgender woman.

^aIn this context, a missed opportunity is defined as an individual who was not screened for HIV in all relevant encounters 30–365 days before HIV diagnosis or screened 30 days prior. This outcome is stratified by those with all missed opportunities in the year prior to diagnosis vs. those who had no or some missed opportunities for HIV screening.

^bData are incomplete due to missing data.

Individuals aged 35–49 years and older than 50 years had higher odds of MO compared to those aged 20–34 (adjusted OR [aOR] 2.1 [95% CI, 1.3–3.4] and aOR 3.4 [2.2–5.7], respectively, P < .0001). There was a trend toward women having higher odds of MO than men (aOR 1.4 [0.93–2.0], P = .12). There were no associations between race, ethnicity, gender identity, or sexual orientation and the odds of MO (Table 1).

For individuals with MO, the median number of days between HIV diagnosis date and last negative HIV screening was 715 days (IQR 634), compared to 194 days (IQR 160) for those without MO. Individuals who received screening in the 1–2 years before HIV diagnosis had higher odds of receiving testing the following year (aOR for MO 0.41 [0.27–0.63], P < .0001) (Table 2). Individuals with MO had more than double the odds of having CD4 < 350 cells/mm³ at diagnosis (aOR 1.8 [1.2–2.9], P = .011).

Table 2.

Factors Predictive of Having a Missed Opportunity for HIV Screening Among Individuals Diagnosed With HIV 2012–2022^a

n (%)	Missed Opportunityb		Univariable		Multivariable
	No (n = 214, 30%)	Yes (n = 499, 70%)	OR (95% CI)	P value	Adjusted OR (95% CI)	P value
HIV-related factors
HIV screened 366–730 days Before Dx	65 (30%)	68 (14%)	0.36 (0.25–0.53)	<.0001	0.41 (0.27–0.63)	<.0001
CD4 < 200	13 (6.1%)	63 (13%)	2.2 (1.2–4.2)	.014	1.7 (0.92–3.4)	.11
CD4 < 350	29 (14%)	133 (27%)	2.3 (1.5–3.6)	.0002	1.8 (1.2–2.9)	.011
Healthcare utilization
No. of encounters 30–365 days before Dxc	3.5 (3.5)	2.2 (2.3)	0.48 (0.30–0.75)	<.0001	0.83 (0.77–0.89)	<.0001
No. of encounters evera	92 (146)	76 (121)	1.0 (0.99–1.0)	.13	1.0 (0.99–1.0)	.027
>2 visits to ED/UC 2 y ≤ Dx	103 (48%)	142 (28%)	0.43 (0.31–0.60)	<.0001	0.68 (0.46, 1.0)	.054
ED encounter without labs 2y ≤ Dx	113 (53%)	332 (67%)	1.8 (1.3–2.5)	.0007	2.6 (1.8–3.7)	<.0001
STI exposure complaint 2 y ≤ Dx	25 (12%)	15 (3.0%)	0.23 (0.12–0.45)	<.0001	0.43 (0.21–0.88)	.022
SH-related complaint 2 y ≤ Dx	99 (46%)	138 (28%)	0.44 (0.32–0.62)	<.0001	0.62 (0.42–0.90)	.013
No SH-related visit	107 (50%)	332 (67%)	2.0 (1.4–2.8)	<.0001	1.4 (0.98–2.1)	.064
Behavioral/mental health
STI 2 y ≤ Dx	35 (16%)	26 (5.2%)	0.28 (0.16–0.48)	<.0001	0.47 (0.26–0.83)	.009

Abbreviations: CI, confidence interval; ED, emergency department; IV, intravenous; Meth, methamphetamine; OR, odds ratio; PC, primary care; PID, pelvic inflammatory disease; SDOH, social determinants of health; SH, sexual health; STI, sexually transmitted infection; UC, urgent care; WH, women's health.

^aThis table has been condensed to display only factors which reached statistical significance. A full report of factors evaluated can be found in Supplementary Table 1.

^bIn this context, a missed opportunity is defined as an individual who was not screened for HIV in all relevant encounters 30–365 days before HIV diagnosis or screened 30 days prior. This outcome is stratified by those with all missed opportunities in the year prior to diagnosis vs. those who had no or some missed opportunities for HIV screening.

^cRepresented as μ (σ).

Several factors related to healthcare utilization were significantly associated with the odds of MO. Individuals without MO had more healthcare visits in the year before diagnosis compared to those with MO (3.5 encounters versus 2.3; aOR 0.83 [0.76–0.89], P < .0001). In univariate analysis, individuals with at least 2 encounters in the ED or urgent care in the 2 years before diagnosis had lower odds of MO (OR 0.43 [0.31–0.60], P < .0001). This finding was maintained with multivariate analysis, although the association only trended toward statistical significance (aOR 0.68 [0.46–1.0], P = .054). In contrast, individuals with at least 1 ED encounter without a blood draw for other testing in the 2 years before HIV diagnosis had nearly double the odds of MO (aOR 2.6 [1.8–3.7], P < .0001). The number of primary care encounters in the 2 years before HIV diagnosis was similar between individuals with and without MO (aOR 1.0 [0.99–1.0], P = .25) (Table 2).

Individuals who sought care for sexually transmitted infection (STI) exposure or other sexual health–related complaints in the 2 years before diagnosis had lower odds of MO (aOR 0.43 [0.21–0.88], P = .022 and aOR 0.62 [0.42–0.90], P = .013, respectively). Similarly, individuals without a sexual health–related encounter had double the odds of MO in unadjusted models (OR 2.0 [1.4–2.8], P < .0001), although the statistical significance of this association was not maintained after controlling for confounding variables (aOR 1.4 [0.98–2.1], P = .064). Social determinants of health (SDOH) were not found to be associated with MO (Table 2).

Secondary Outcome—Encounter Level Missed Testing Opportunities

Of the 1845 testing-indicated encounters that occurred 30–365 days before an HIV diagnosis, 1235 (67%) encounters were MTO (Table 3). Within clinical settings, with outpatient as the reference group, MTO were less common in the ED/urgent care (OR 0.67 [0.54–0.82], P < .0001). Within outpatient departments, compared to primary care, women's health departments (i.e., obstetrics and gynecology) had lower odds of MTO (OR 0.37 [0.24–0.57], P < .0001). Encounters with any bacterial STI testing had lower odds of being an MTO (OR 0.36 [0.28–0.48], P < .0001), and this was true for each bacterial STI independently (gonorrhea: OR 0.33 [0.28–0.48]; chlamydia: OR 0.34 [0.25–0.46]; and syphilis: OR 0.25 [0.17–0.36]; P < .0001). The presence of SDOH, mental health, and substance use diagnoses were not associated with the odds of MTO.

Table 3.

Encounter Characteristics:^a Labs, Diagnoses, and Encounter Locations on Relevant Encounters^b in the Year Prior to HIV Diagnosis

n (%)	Encounters (n = 1845)	No Missed HIV Testing Opportunity (n = 610; 33%)	Missed HIV Testing Opportunityc (n = 1235; 67%)	Odds Ratio (95% CI)	P Value
Encounter types
Outpatient	762 (41%)	213 (35%)	549 (44%)	Ref
ED/UC	1003 (54%)	369 (60%)	634 (51%)	0.67 (0.54–0.82)	<.0001
Inpatient	80 (4.3%)	28 (4.6%)	52 (4.2%)	0.72 (0.45–1.2)	.19
Outpatient departmentsd
PC	470 (62%)	116 (55%)	354 (65%)	Ref
OB/GYN	109 (14%)	51 (24%)	58 (11%)	0.37 (0.24–0.57)	<.0001
Specialty	183 (24%)	46 (22%)	137 (25%)	0.98 (0.66–1.5)	.90
Lab tests taken
Bacterial STI	257 (14%)	138 (23%)	119 (9.6%)	0.36 (0.28–0.48)	<.0001
Gonorrhea	193 (11%)	110 (18%)	83 (6.7%)	0.33 (0.24–0.44)	<.0001
Syphilis	123 (6.7%)	79 (13%)	44 (3.6%)	0.25 (0.17–0.36)	<.0001
Chlamydia	196 (11%)	110 (18%)	86 (7.0%)	0.34 (0.25–0.46)	<.0001
Behavioral/sexual health
STD exposure complaint	36 (2.0%)	18 (3.0%)	18 (1.5%)	0.49 (0.25–0.94)	.045
SDOH Dx	86 (4.7%)	37 (6.1%)	49 (4.0%)	0.64 (0.41–0.99)	.058

Abbreviations: Dx, diagnosis; ED, emergency department; Meth, methamphetamines; OB/GYN, obstetrician and gynecologist; PC, primary care; SDOH, social determinants of health; SH, sexual health; Specialty, specialty clinics (e.g., dermatology; cardiology); STD, sexually transmitted disease; STI, sexually transmitted disease (encompasses gonorrhea, syphilis, or chlamydia); UC, urgent care.

^aThis table has been condensed to display only encounter characteristics that reached statistical significance. A full report of encounter characteristics analyzed can be found in Supplementary Table 2.

^bRelevant encounters are 30–365 days before HIV diagnosis that were in primary care, ED, UC, or inpatient.

^cIn this context, a missed HIV testing opportunity is defined as an encounter 30–365 days before HIV diagnosis in which they were not screened within 6 months or not screened upon an encounter in which there was a sexual health related chief complaint or diagnosis.

^dPC, OB/GYN, and Specialty are departments from outpatient encounters. Therefore, their column % only considers outpatient encounters.

GHS Routine Opt-out HIV Testing Program Analysis

Between 2014 and 2022, 531,848 individuals were eligible for opt-out routine HIV screening at GHS. 357,771 tests were performed on 199,004 unique individuals (37.4%). Of these, 4719 tests (1.3% of tests) were reactive, producing positive results in 4211 individuals (2.1% of individuals) and resulting in 861 (0.4% of individuals) new diagnoses of HIV (Figure 1).

DISCUSSION

In this 7-year retrospective study, we characterized MO for HIV testing among individuals newly diagnosed with HIV in a large, safety-net hospital system using a routine opt-out HIV testing approach. In our study population, 70% of individuals newly diagnosed with HIV were not tested for HIV at any encounter in our health system in the year preceding diagnosis. This proportion of MO is double what has been previously reported [2, 9]. Factors associated with higher odds of MO included being older than 35 years of age, lower healthcare utilization, ED visits without blood drawn for other testing, and visits unrelated to sexual health. Outpatient encounters, particularly in primary care, were more likely to represent MTO than ED visits. Notably, individuals with MO experienced longer periods of undiagnosed HIV and, in contrast to prior studies, had nearly twice the odds of presenting with advanced HIV, defined as a CD4 count less than 350 cells/mm³ at diagnosis, compared to those without MO [2, 9]. Delayed diagnosis and advanced disease at presentation are associated with more significant morbidity and mortality for PWH [22–25]. Further, a recent modeling study showed that individuals with chronic HIV infection who were unaware of their status had more than double the transmission rate compared to PWH in care, regardless of viral suppression [3]. These individual and public health impacts underscore the urgent need to address missed opportunities to facilitate earlier linkage to HIV care. We analyzed our opt-out testing program to contextualize our results and found that despite an adequate overall positive test prevalence of 1.3% [11, 26], only about one third of all eligible individuals were tested for HIV. These results, particularly in the context of clinically significant missed opportunities, highlight the need to optimize our routine opt-out HIV testing strategy.

Contrary to prior studies, we found that outpatient encounters, particularly in primary care, were more likely to be MTO [2, 15]. Although primary care providers are well positioned to incorporate HIV testing into routine care due to their long-term relationships with patients [27], they face significant barriers. These include persistent HIV-related stigma, particularly in Black communities in the Southern United States [28], and associated concerns about straining the physician-patient relationship by offering testing [27]. Furthermore, primary care providers are uniquely vulnerable to systemic limitations, including time-constrained visits and competing priorities [29–31], with a simulation study demonstrating they would require 26.7 hours daily to deliver all recommended care [31]. Digital strategies for direct-to-patient education outside of clinical visits could help reframe HIV testing as routine, reduce HIV-related stigma, encourage patient-initiated testing, and save valuable time during encounters [32–34]. Further studies examining MO in the primary care setting are needed.

Unsurprisingly, seeking sexual health–related care was protective against MO and MTO. In line with prior studies [35, 36], we hypothesize that this finding reflects HIV testing based on risk perception by clinicians. However, nearly half of newly diagnosed PWH did not seek sexual health-related care during the study period, and lack of sexual health–related care nearly doubled the odds of MO, indicating that this risk perception is inaccurate and/or incomplete. While testing based on inaccurate risk perception is problematic, accurate risk assessment that reflects epidemiologic and behavioral factors could optimize HIV testing [37]. Emerging studies have developed HIV risk scores using machine learning algorithms [11, 38–40]. A hybrid strategy combining routine testing with validated HIV risk assessment tools may enhance screening, support early diagnosis, and broaden prevention through provider- or patient-directed prompts.

Our study has several limitations. First, this is a single-site study, so our results may not be generalizable. Second, we were unable to determine why HIV testing did not occur during specific encounters, whether because of the provider's oversight, patient decision to opt out, or other factors. Additionally, PWH in our database that did not seek care at GHS the year before diagnosis were excluded from our study. There is a possibility that some of these individuals were tested for HIV outside of our system, but we were unable to capture it. Although we did not find an association between race and MO, bias may still play a role in other settings and warrants further investigation. Finally, individuals with at least 1 HIV test in the year before diagnosis were categorized as not having MO, which could have led to an underestimation of missed opportunities in cases where more frequent HIV testing was clinically indicated.

This study provides a comprehensive evaluation of a real-world HIV opt-out testing strategy within 1 of the largest safety-net healthcare systems in the Southern United States, serving a population with 1 of the highest HIV incidence rates in the nation. In a novel approach, we used a statewide HIV diagnosis database to identify newly diagnosed PWH and combined these data with our internal database, enabling us to include a broader study population who had interacted with our healthcare system before their HIV diagnosis. Additionally, we employed a more specific definition of MO than previous studies [2, 8, 9, 13, 14], requiring no HIV testing during any encounter in the year before diagnosis and reducing the likelihood that overutilization of healthcare confounds our findings. These innovations allowed us to quantify the prevalence of clinically and temporally significant missed opportunities and offer clinical and public health insight into the impact of MO on efforts to end the HIV epidemic. Our findings prompt further exploration of the barriers and facilitators of our opt-out testing program and to develop, implement, and study interventions to address them.

CONCLUSIONS

Despite a systemwide opt-out HIV testing program, most individuals newly diagnosed with HIV in a high-volume safety-net healthcare system had missed opportunities for earlier HIV diagnosis. These missed opportunities have important clinical and public health implications and underscore the need to optimize HIV testing strategies to facilitate earlier HIV diagnosis.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.