HIV Testing Among Adolescents with Acute Sexually Transmitted Infections

Abstract

Background and Objective:

Rates of sexually transmitted infections (STI) have increased over the decade. Guidelines recommend HIV testing with incident STIs. Prevalence and factors associated with HIV testing in acute STIs are unknown in adolescents. Our objective was to determine the prevalence of completed HIV testing among adolescents with incident STIs and identify patient and healthcare factors associated with HIV testing.

Methods:

Retrospective study of STI episodes (gonorrhea, chlamydia, trichomoniasis, or syphilis) of adolescents between 13–24 years-old from July 2014 to December 2017 in two urban primary care clinics. We performed mixed effects logistic regression modeling to identify patient and healthcare factors associated with HIV testing within 90 days of STI diagnosis.

Results:

The 1,313 participants contributed 1,816 acute STI episodes. Mean age at STI diagnosis was 17.2 years (SD=1.7), 75% of episodes occurred in females, and 97% occurred in African-Americans. Only half (55%) of acute STI episodes had a completed HIV test. In the adjusted model, female sex, prior STIs, uninsured, and a confidential sexual health encounter were associated with decreased odds of HIV testing. Patients enrolled in primary care at the clinics, compared to those receiving sexual healthcare alone, and those with multi-pathogen STI diagnoses were more likely to have HIV testing.

Conclusion:

HIV testing rates among adolescents with acute STIs are suboptimal. Patient and healthcare factors were found to be associated with receipt of testing and should be considered in clinical practice.

Table of Contents Summary

This study demonstrates that only 55% of adolescent STI episodes are tested for HIV and identifies significant patient and healthcare factors associated with HIV testing.

Introduction

The past decade has seen dramatic increases in sexually transmitted infections (STIs), specifically chlamydia, gonorrhea, syphilis, and trichomoniasis--among the highest in recorded United States (U.S.) history.¹ From 2013–2017, the STI rates increased by 31%, with gonorrhea and syphilis incidence climbing by 67% and 76%, respectively.² Much of the STI burden falls among adolescents and young adults (AYA). While youth 15–24 years-old represent only one-quarter of sexually active individuals, they account for half of incident STIs.³ Youth with acute STIs are at increased risk of HIV due to both non-condom-protected sexual behavior and genital tract inflammation.^4,5 Over the lifespan, each STI episode increases one’s susceptibility to HIV infection.^6,7

In recognition of this increased risk, in addition to routine HIV testing, the Centers for Disease Control and Prevention (CDC) and U.S. Preventative Services Task Force recommend risk-based HIV testing for individuals who test positive for an STI.^8,9 However, these guidelines have not been adequately followed.^9–12 HIV screening rates are as low as 36% among adults, and lower among AYA,^10,13,14 which has contributed to 44% of 13–24 year-olds living with HIV being undiagnosed.¹⁵ Undiagnosed individuals living with HIV are at risk for HIV progression and contribute to one-third of HIV transmissions due to uncontrolled HIV viremia.^16–18 Unfortunately, AYA 13–24 years-old have the highest transmission rate amongst all people living with HIV.¹⁹

In order to make strides toward the Presidential Plan to End the HIV Epidemic,²⁰ we must identify missed opportunities for HIV testing, and use these as gateways to treatment and prevention, especially in vulnerable AYA.²¹ Our primary research objective was to examine rates of HIV testing after an incident STI among AYA in primary care settings and identify patient and healthcare factors related to receipt of HIV testing, which may serve as targets for future interventions to improve HIV testing.

Participants and Methods

Study Design and Data Source

This retrospective study utilized electronic health record (EHR) data from two urban, pediatric/adolescent, primary care clinics, with high rates of incident STIs, affiliated with a large, academic health system in Philadelphia.²² During the observation period, HIV screening was performed only by laboratory-based, fourth-generation antigen/antibody testing. We intentionally limited the observation period to when rapid HIV testing was unavailable because rapid test results were not routinely documented in the EHR, and the barriers to laboratory-based versus rapid HIV testing likely differ. Therefore, the observation period differed between the two sites, to reflect when funding for rapid testing was eliminated at these clinics. For Site 1, the observation period was July 1, 2014 – December 31, 2017. For Site 2 the observation period was January 1, 2015 – December 31, 2017. Data were automatically abstracted from the EHR, initially for quality improvement measurement, using a commercial business intelligence platform (Qlik, Radnor, PA), which generated a dataset of all STI and HIV screening episodes. The final analytic dataset contained all visits with an incident STI episode, and all follow-up visits within 90 days of that episode. If a patient tested positive for HIV, they were removed from the cohort after diagnosis. In cases of missing data, we performed EHR abstraction. This project was approved by the Institutional Review Boards of Children’s Hospital of Philadelphia and Access Matters.

Study Participants and Setting

Participants were included if they were 13–24 years-old, and had an incident STI episode in the study period. Incident STI episodes were defined as new diagnosis of chlamydia (Chlamydia trachomatis) or gonorrhea (Neisseria gonorrhea) via urine, rectal, oral, or cervical nucleic acid amplification tests (NAATs), trichomoniasis (Trichomonas vaginalis) via NAAT or wet mount microscopy, and/or syphilis (Treponema pallidum) by serum rapid plasmin reagin testing.

Patients at both sites are mostly African-American (87%) and Medicaid-insured (79%).²² Each clinical site has a co-located primary care practice and Title X-funded adolescent program that provides confidential sexual health (family planning) services irrespective of insurance coverage. Primary care patients may receive sexual health services through either primary care or family planning. Uninsured patients or those not enrolled in primary care at the sites receive sexual health services solely in the Title X family planning program. The EHR for family planning visits has additional confidentiality protections and no billing statements are sent to parents/guardians for these visits. Both sites have similar patient demographics; however, site #1 has larger patient volume and has more adolescent trained providers.

Outcome Measure

The primary outcome measure was receipt of an HIV test within 90 days of STI diagnosis. This 90-day limit was chosen as this timeframe is sufficient for providers to obtain the STI result, communicate it to the patient, invite the patient to return for HIV testing, and have patients complete HIV testing. HIV testing was defined as having a laboratory-based fourth generation HIV antibody/antigen test. Visits were excluded if an encounter was not conducive to an HIV test being performed (e.g. clinic visits for behavioral health only, immunization visits). For visits where HIV testing was ordered and not completed, we manually reviewed the EHR to ascertain reason for test non-completion.

Covariates

Patient- and healthcare-level factors were collected. Patient factors included age, sex, multi-pathogen STI diagnosis, and prior STIs. Age was categorized as younger (13–17 years-old) and older adolescents (18–24 years-old), as we hypothesized older adolescents could be more likely to receive testing. Healthcare factors included adolescent medicine (AM) training of provider (yes or no), insurance status (insured or uninsured), visit type (family planning or primary care), and patient enrollment in primary care at the sites during the study time. Providers were considered AM trained if they were a physician who was enrolled in or had completed AM fellowship, or if they were a nurse practitioner assigned to specifically see adolescent family planning patients.

Statistical analysis

Descriptive statistics summarized characteristics of participants and STI episodes across HIV test outcome. We determined prevalence of HIV testing by calculating the proportion of completed HIV tests within 90 days of the STI episode among all STI episodes. We conducted bivariate analyses using t-tests and X² to test associations between characteristics of STI episodes and HIV test completion.

To examine adjusted associations between patient and healthcare factors on HIV test completion, we used mixed effects logistic regression models. Odds ratios were estimated accounting for random effects of subjects and clinic sites.

We first conducted models assessing associations between patient factors, and separately, healthcare factors on HIV testing. The final multivariable model contained both patient and healthcare factors, as prior literature suggests HIV testing may be influenced by all of these variables. We included an interaction term between insurance and family planning visit because we hypothesized that family planning visits could function differently for insured and uninsured patients. Insured patients may opt for family planning visits for confidentiality reasons.

In sensitivity analysis, we examined if time since last HIV testing had an effect on HIV test completion with incident STI diagnosis. This analysis was performed on a subset of the episodes, from January 2016-December 2017, to allow for record review of at least one full year prior to each patient’s STI episode. We categorized time from last HIV test into three groups: those who had an HIV test in the past year, those who had an HIV test more than a year ago, or those with no prior documented HIV test. We performed similar multivariable models as defined above and added the time since last HIV testing variable in two ways: categorical, and continuous, denoting time in months.

All statistical analyses were done using STATA 15 (StataCorp LLC., College Station, TX) and SAS 9.4 (SAS Institute Inc., Cary, NC).

Results

Out of the 16,392 STI tests performed during our study period, 1,816 (11.1%) were positive and included in our study. These 1,816 episodes occurred in 1,313 unique individuals. The majority of patients (97.4%) were African-American, reflecting the demographics of the clinics. The patients were 70.8% female, 1.4% Hispanic, with a median age of 17 (Interquartile range [IQR]: 16–18). Twenty-seven percent of individuals had multiple STI episodes during the study period. Table 1 shows characteristics of the STI episodes by HIV test completion. The most prevalent STI was chlamydia, followed by gonorrhea, trichomoniasis, and syphilis (Table 1).

Table 1:

Patient and healthcare characteristics of STI episodes by HIV test completion, 2014–2017

	All (n=1,816) n(%)	Completed HIV test (n=1,001) n(%)	No HIV test (n=815) n(%)	p-value
Age – Mean (SD)	17.2 (1.7)	17.1 (1.6)	17.3 (1.7)	0.04
Sex
Female	1,356 (74.7%)	703 (51.8%)	653 (48.7%)	<0.001
Male	460 (25.3%)	298 (64.8%)	162 (35.2%)
Race
African-American	1,766 (97.2%)	979 (55.1%)	787 (44.6%)	0.11
Other	50 (2.8%)	22 (44.0%)	28 (56.0%)
Hispanic	25 (1.4%)	5 (20.0%)	20 (80.0%)	<0.001
Insurance
Government	1,101 (60.6%)	620 (56.3%)	481 (43.7%)	0.06
Private	422 (23.2%)	238 (56.4%)	184 (43.6%)
None	293 (16.1%)	143 (48.8%)	150 (51.2%)
STI1 diagnosis
Chlamydia	1,533 (84.4%)	857 (55.9%)	676 (44.1%)	0.28
Gonorrhea	307 (16.9%)	178 (58.0%)	129 (42.0%)	0.30
Trichomoniasis	134 (7.4%)	57 (42.5%)	77 (57.5%)	0.08
Syphilis	8 (0.4%)	8 (100.0%)	0 (0.0%)	0.91
Multi-pathogen STI	162 (8.9%)	99 (61.1%)	63 (38.9%)	0.11
STI History
No prior STIs	1,239 (68.2%)	735 (59.3%)	504 (40.7%)	<0.001
Prior STIs	577 (31.8%)	266 (46.1%)	311 (53.9%)
Mean STI number	1.5 (0.86)	1.4 (0.80)	1.6 (0.93)	<0.001
AM3 Trained
Yes	1,022 (56.3%)	544 (53.2%)	478 (46.8%)	0.07
No	794 (43.7%)	457 (57.6%)	337 (42.4%)
Family Planning visit
Yes	1,489 (82.0%)	770 (51.7%)	719 (48.3%)	<0.001
No	327 (18.0%)	231 (70.6%)	96 (29.4%)
Visit Site4
Site #1	747 (41.3%)	350 (46.9%)	397 (53.1%)	<0.001
Site #2	1,060 (58.7%)	636 (60.0%)	424 (40.0%)

Statistical Significance: p=0.05

¹Sexually transmitted infection

²Certified Registered Nurse Practitioner

³Adolescent Medicine

⁴Denominator is 1,807 due to missing data

All analysis was done in episode level using t-test and X² test and presented by frequency (%) or mean (standard deviation).

Table 1 also shows bivariate associations between patient and healthcare factors and HIV test completion within 90 days of an STI diagnosis. The proportion of episodes where HIV testing was completed within 90 days of STI diagnosis was 55.1% (n=1,001/1,816 episodes). There was n=1 confirmed positive HIV result (0.1%) among completed tests. Figure 1 shows STI episodes and whether HIV testing was ordered and/or performed at baseline and/or follow-up visits. Of those episodes where HIV testing did not happen concurrent with STI testing, about three-quarters had a follow-up visit, of which the majority did not have completed HIV testing. The majority of episodes with completed HIV testing had HIV testing concurrent with STI testing. Of the n=815 STI episodes where HIV testing was not completed, n=219 (26.9%) had a test ordered by the provider and not completed by the patient. The main reasons for HIV test non-completion included patients leaving the lab before the test could be performed (n=146, 66.7%), followed by not showing up at all (n=40, 18.3%), errors in the medical record or laboratory (n=10, 4.6%), patients declining HIV testing and closed laboratory (n=8, 3.6%). Fifteen (6.8%) had no reason listed for HIV test non-completion.

Figure 1.

Caption: Receipt of HIV testing in the study sample by STI episode (n=1,816)

Black=HIV test completed; Gray= HIV test ordered and not completed

The results of the three mixed effects logistic regression models examining associations between patient and healthcare characteristics and adjusted odds of HIV test completion are displayed in Table 2. With respect to patient characteristics, females and participants with a prior history of STIs had significantly lower adjusted odds of HIV test completion compared to males and those with no prior history of STIs. For the healthcare factors model (Table 2), we found not having insurance and having a family planning visit were both associated with decreased odds of HIV testing compared to their counterparts. There was no significant association between AM training, enrollment in primary care, or the interaction between insurance and family planning and the odds of HIV test completion.

Table 2:

Mixed Effects Logistic Regression model assessing patient, healthcare, and combined factors and adjusted odds of HIV test completion

	Patient Model		Healthcare Model		Final Combined Model
Covariates	aOR (95% CI)	p-value	aOR (95% CI)	p-value	aOR (95% CI)	p-value
STIs at diagnosis			---	---
Having 1 STI at diagnosis	Reference				Reference
Multiple STIs at diagnosis	1.40 (0.99 – 2.00)	0.06			1.43 (1.00 – 2.04)	0.05
STI History			---	---
No prior history of STIs	Reference				Reference
Prior STI history	0.59 (0.47 – 0.73)	<0.001			0.58 (0.47 – 0.73)	<0.001
Younger vs. Older adolescents			---	---
13–17 years-old	Reference				Reference
18–24 years-old	0.88 (0.72 – 1.09)	0.24			0.91 (0.74 – 1.12)	0.58
Patient Sex			---	---
Male	Reference				Reference
Female	0.63 (0.49 – 0.80)	<0.001			0.63 (0.49 – 0.81)	<0.001
Adolescent trained providers	---	---
No			Reference		Reference
Yes			1.00 (0.81 – 1.24)	0.99	1.11 (0.89 – 1.37)	0.36
Insurance status	---	---
Private or Government			Reference		Reference
No insurance			0.44 (0.21 – 0.92)	0.03	0.48 (0.23 – 0.99)	0.05
Family planning visit	---	---
No			Reference		Reference
Yes			0.48 (0.34 – 0.67)	<0.001	0.55 (0.39 – 0.76)	<0.001
Interaction (insurance and family planning visit)	---	---	1.87 (0.85 – 4.14)	0.121	1.72 (0.78 – 3.77)	0.18
Primary care enrollment	---	---
No			Reference		Reference
Yes			1.24 (0.90 – 1.70)	0.185	1.46 (1.06 – 2.01)	0.02

aOR= Adjusted Odds Ratios; CI=Confidence Intervals; STI=Sexually transmitted infection

A Mixed Effects Logistic Regression model was used including all variables above after considering hierarchical structure of STI related visits by clinic and individual.

In our final model (Table 2), factors associated with increased odds of completed HIV testing included multi-pathogen infection at STI diagnosis and enrollment in primary care. Factors associated with decreased odds of HIV testing included having history of prior STIs, being female, uninsured, and having the incident STI diagnosed at a family planning visit.

In the sensitivity analysis (Table 3) examining time from last HIV test with receipt of HIV testing within 90 days of incident STI, participants who had a previous completed HIV test over one year ago or never had a documented HIV test had higher adjusted odds of receipt of HIV testing at incident STI, compared to those with HIV testing within the year prior to incident STI. Among those with a prior HIV test, there was an 8% (OR=1.08, 95% CI: 1.04 – 1.12) increased odds of being appropriately HIV tested within 90 days of STI diagnosis with each additional month since the last documented HIV test.

Table 3:

Sensitivity Analysis: Mixed Effects Logistic Regression model assessing time since last HIV testing adjusting for patient, healthcare, and combined factors and the odds of HIV test completion, 2016–2017

Covariates	aOR (95% CI)	p-value
Last HIV test
Tested less than 1 year ago	Reference
More than 1 year ago	2.47 (1.50 – 4.07)	<0.001
No documented HIV test	2.10 (1.38 – 3.20)	0.001
STIs at diagnosis
Having 1 STI at diagnosis	Reference
Multiple STIs at diagnosis	1.65 (0.96 – 2.83)	0.07
STI history
No prior history of STIs	Reference
Prior STI history	0.86 (0.61 – 1.22)	0.41
Younger vs. older adolescents
13–17 years-old	Reference
18–24 years-old	0.87 (0.64 – 1.19)	0.39
Patient Sex
Male	Reference
Female	0.80 (0.56 – 1.13)	0.21
Adolescent trained providers
No	Reference
Yes	1.20 (0.87 – 1.65)	0.26
Insurance status
Private or Government	Reference
No insurance	0.60 (0.19 – 1.87)	0.38
Family planning visit
No	Reference
Yes	0.47 (0.30 – 0.75)	0.001
Interaction (insurance and family planning visit)	1.45 (0.42 – 4.92)	0.55
Primary care enrollment
No	Reference
Yes	1.76 (1.06 – 2.92)	0.03

aOR= Adjusted Odds Ratios; CI=Confidence Intervals; STI=Sexually transmitted infection

A Mixed Effects Logistic Regression model was used including all variables above after considering hierarchical structure of STI related visits by clinic and individual.

Discussion

In our sample of STI-positive adolescents receiving care at large, urban, primary care clinics, we found that only 55% of incident STI episodes had HIV testing completed within 90 days following STI diagnosis. Previous studies have found adolescents to be under-screened for HIV, especially in primary care.^13,23 We have expanded on previous work by showing when adolescents are at highest risk for HIV (i.e. diagnosed with an incident STI), only half of STI episodes were appropriately HIV tested. Most completed HIV tests were done at the time of STI testing as a part of sexual health screenings. Promoting the inclusion of HIV testing in comprehensive sexual health screening for adolescents may improve the rates of HIV testing overall. Leaving or not showing up to the laboratory for blood draw were the main reasons for cancelled HIV tests. This suggests that even when a provider orders a laboratory-based HIV test, additional barriers prevent adolescent test completion. Focusing on rapid HIV testing in primary care for adolescents would likely reduce barriers to patients and increase HIV test completion rates.

Clinicians appropriately identified increased HIV risk in cases of multi-pathogen STIs, as these episodes had higher odds of HIV testing. However, they did not appropriately identify increased lifetime HIV risk in patients with prior STIs, who had lower odds of HIV testing than those without a history of STIs. This suggests a significant missed opportunity for targeted HIV screening, early HIV diagnosis, and linkage to HIV care if individuals were HIV-positive or initiation of pre-exposure prophylaxis (PrEP) if HIV-negative. This finding adds significantly to the literature, as little is known about how prior STIs may influence HIV testing in adolescents with acute STIs.^6,7 Youth with prior STIs may paradoxically have a lower perceived HIV risk, given that they have had STIs previously and ostensibly remained HIV negative. Conversely, youth with no prior STI may experience a heightened fear of HIV infection. Finally, patients with prior STIs may have undergone HIV testing concurrent with previous STI diagnoses, thus biasing providers against the need for repeat HIV testing. In this clinical cohort, a single HIV infection was detected within three months of STI diagnosis. As many of the highest risk youth did not receive testing, the number of missed HIV infections and missed opportunities for linkage to PrEP among high-risk HIV-negative youth in a high HIV-incidence urban area, remain unknown.

Our sensitivity analysis demonstrated that youth who had an HIV test over one year ago or those never tested at these clinics had higher odds of receipt of HIV testing with incident STI. This finding demonstrates that providers may be biased toward testing on a yearly basis, rather than recognizing incident STIs as a biomarker of risk and testing appropriately with each new episode. While testing guidelines for HIV state those who are sexually active or those with an STI should be tested for HIV, they do not state that HIV testing should be done solely yearly. To increase HIV testing in those with STIs, health systems should examine strategies for identifying patients with STIs through the EHR and alerting providers on the need for intervention in order to close this gap in HIV testing. Furthermore, our findings suggest a need for further provider education and guideline clarification since each new STI diagnosis represents an increased temporal risk for HIV and a unique opportunity to link HIV-negative youth to PrEP when motivation for behavior change may be highest.

Our finding of lower odds of HIV testing in those without insurance is in accordance with previous literature showing reduced HIV testing rates in uninsured individuals.^24–27 Because our patients could receive HIV and STI testing and treatment confidentially, without insurance, and without being capitated to primary care at the site, many of the traditionally identified barriers to HIV testing were mitigated at the study sites. Unexpectedly, having a family planning visit was associated with lower odds of HIV testing and being enrolled in primary care at the clinics was associated with higher odds of HIV testing. These findings emphasize the importance of receiving primary preventative health care and having an ongoing relationship to clinical sites in receipt of adolescent sexual health care. The literature has shown primary care to be an effective delivery site for adolescent sexual and mental health services.^28–32 Clinicians in sexual health clinics should consider how to connect their patients to primary care to improve ongoing sexual health and HIV testing outcomes.

This study is subject to limitations. Due to the nature of our secondary data, there is the possibility of unmeasured confounders which would have further enriched our investigation (e.g. patient sexual orientation, sexual behavior, and additional provider characteristics). Despite this, our sample size was robust and missing data was marginal or nonexistent for measured variables. Second, patients could have received care from other facilities and health systems outside these sites, making it possible that we did not capture the patients’ complete HIV testing histories. However, we believe this is unlikely as both of study clinics offered free HIV screening and sexual health services. Therefore, this would limit bias based on insurance status or cost of care. Third, our analysis was limited to a specific section of Philadelphia, and may not be generalizable outside of that area. However, Philadelphia STI rates are almost three-times the national average, and the urban area where the research was conducted has some of the highest rates in Philadelphia.^34,35 Thus, our study represents a sample at particularly high risk for contracting HIV in which testing rates would be expected to be higher than in low HIV and STI prevalence regions.

Conclusions

Our findings identified suboptimal rates of HIV testing in adolescents with STIs, with nearly half of episodes not receiving testing within 90 days of STI diagnosis. In addition, we identified that rates of testing were actually lower in participants who were at highest risk of HIV infection, including those with prior history of STIs. This study focused its analyses to adolescents in a primary care setting where HIV and STI testing may occur confidentially and without insurance, ameliorating most barriers to testing and thus representing a unique addition to previous HIV testing research. However, we also identified that some adolescents did not receive HIV testing even when ordered by providers, suggesting that rapid HIV testing at the time of encounter may be beneficial to test completion for adolescents and easier to complete for providers. As we enter the fourth decade of the HIV epidemic, it remains clear that missed opportunities for diagnosis have the potential to delay HIV diagnosis and linkage to antiretroviral therapy or PrEP and prevention services, thus increasing the population risk of HIV transmission.^18,36 Our data underscore the need for improved HIV testing education for providers of all levels of training, and the need for public health agencies to clearly communicate the need for testing at the time of STI infection to reduce the number of missed opportunities for testing. It is our hope that our findings can be utilized to target HIV testing interventions in primary care settings.

What’s Known on This Subject

Prior literature states adolescents have suboptimal HIV testing rates and that pediatricians are often unaware of clinical guidelines for HIV testing. Adolescents with STIs are at higher risk for HIV and should be considered for initiation of HIV pre-exposure prophylaxis.

What This Study Adds

This study finds that HIV testing rates among adolescents with STIs are suboptimal. The study also identifies that patients enrolled in primary care at the clinic sites had increased odds of receipt of HIV testing after STI diagnosis.

Abbreviations:

STI

Sexually transmitted infection

HIV

Human Immunodeficiency Virus

US

United States

AYA

Adolescents and young adults

CDC

Centers for Disease Control and Prevention

USPSTF

United States Preventative Services Task Force

EHR

Electronic health record

NAAT

Nucleic acid amplification test

CRNP

Certified Registered Nurse Practitioner

SD

Standard deviation

OR

Odds ratio