Loss to Follow-Up and Re-Engagement in HIV Pre-Exposure Prophylaxis Care in the United States, 2013–2019

Abstract

Retention in HIV pre-exposure prophylaxis (PrEP) care is critical for effective PrEP implementation. Few studies have reported long-term lost to follow-up (LTFU) and re-engagement in PrEP care in the United States. Medical record data for all cisgender patients presenting to the major Rhode Island PrEP clinic from 2013 to 2019 were included. LTFU was defined as no PrEP follow-up appointment within 98 days. Re-engagement in care was defined as individuals who were ever LTFU and later attended a follow-up appointment. Recurrent event survival analysis was performed to explore factors associated with PrEP retention over time. Of 654 PrEP patients, the median age was 31 years old [interquartile range (IQR): 25, 43]. The majority were male (96%), White (64%), non-Hispanic (82%), and insured (97%). Overall, 72% patients were ever LTFU and 27% of those ever LTFU re-engaged in care. Female patients were 1.37 times [crude hazard ratio (cHR): 1.37; 95% confidence interval (CI): 0.86–2.18] more likely to be LTFU than male patients, and a 1-year increase in age was associated with a 1% lower hazard of being LTFU (cHR: 0.99; CI: 0.98–0.99). Being either heterosexual (aHR: 2.25, 95% (CI): 1.70–2.99] or bisexual (aHR: 2.35, 95% CI: 1.15–4.82) was associated with a higher hazard of loss to follow-up compared with having same-sex partners only. The majority of PrEP users were LTFU, especially at the first 6 months of PrEP initiation. Although a significant number were re-engaged in care, targeted interventions are needed to improve retention in PrEP care. This study characterized the natural projection of loss to follow-up and re-engagement in HIV PrEP care using a longitudinal clinic cohort data and explored associated factors for guiding future interventions to improve retention in PrEP care.

Introduction

HIV pre-exposure prophylaxis (PrEP) is a highly efficacious biobehavioral intervention to prevent HIV acquisition. The efficacy of PrEP in preventing HIV acquisition has been demonstrated in numerous randomized controlled trials and open-label studies among populations at risk.¹ Populations at high risk of HIV acquisition include men who have sex with men (MSM), and heterosexuals who engage in high-risk sexual activities, as well as people who inject drugs.² Modeling studies have shown that PrEP has the potential to dramatically reduce HIV incidence in the United States, particularly among MSM.^3,4 Engagement and retention in PrEP care are critical to achieve the optimal preventive effect of PrEP.

The PrEP care continuum includes PrEP awareness, uptake, and adherence.⁵ Studies show increasing PrEP awareness in the United States among MSM^6,7 and heterosexual women.^8–10 In the fourth quarter of 2017, the overall prevalence of PrEP use in the United States was 26 per 100,000, with rates varying across gender, age, race, ethnicity, and region.¹¹ The number of PrEP users increased by 39% from 2017 to 2018 in the United States.¹² Although a consistent increasing trend in PrEP use has been observed since 2012, rates of PrEP uptake remain low in the United States.¹² Based on this analysis, fewer than 10% of the nearly 1.2 million people indicated for PrEP by CDC guidelines use PrEP.¹¹ Notably, rates of PrEP awareness and uptake vary significantly across subpopulations, highlighting disparities among Black/African Americans (B/AA) and Hispanic/Latino (H/L) communities, those with lower income, and those in other marginalized socioeconomic positions.^13–15 Factors such as medical mistrust and reduced accessibility of health care rooted in systemic racism likely contribute to lower PrEP awareness and uptake among B/AA and H/L MSM.^13,14,16 In one study, B/AA and H/L MSM in serodiscordant partnerships were significantly less likely than White MSM to use PrEP.¹⁷ On a state level, PrEP uptake relative to epidemiological need (HIV incidence) is lowest in states with high poverty, low health insurance coverage, and high concentrations of B/AA and H/L populations.¹¹ These racial and ethnic disparities in PrEP awareness and uptake may carry forward into retention in PrEP care and impact loss to follow-up and re-engagement in PrEP care.

Retention in PrEP care remains the least clearly understood level of the PrEP continuum. Retention in care is essential for ensuring optimal PrEP effectiveness at both the individual and population levels.^18,19 PrEP patients may discontinue PrEP and be lost to follow-up (LTFU) in PrEP care due to decreased perception of risk with changes in sexual activity, reluctance or inability to take daily medications, or barriers related to cost, among others^20–23; however, loss to follow-up and re-engagement in PrEP care are not yet well described in the literature.²⁴ Understanding loss to follow-up and re-engagement in PrEP care over time is necessary to characterize PrEP patients who are LTFU but remain PrEP eligible. Understanding these patterns can also inform the development of appropriate interventions to identify and target those individuals in ongoing PrEP care practice. This is especially important in the interest of identifying and mitigating any racial, ethnic, or other socioeconomic disparities in retention in PrEP care.

Measurement of retention in PrEP care has been limited due to the lack of large-scale longitudinal data. A prior study that evaluated short-term retention among patients taking PrEP using pharmacy fill data showed that ∼41% of patients persisted on PrEP over 2 years.²⁵ However, this study was limited by the inability to track patients who switch pharmacies, as well as the selection bias from sampling a single retail pharmacy chain. Another study using insurance claims data found a median retention time of 13.7 months among commercially insured persons and 6.8 months among Medicaid insured persons, with higher retention among men than among women and White persons than persons of color, and lower retention among young persons.²⁶ None of these studies accounted for re-engagement in PrEP care, which could introduce bias to study findings.

In this study, we expand on the existing literature by evaluating retention in PrEP care—including loss to follow-up and re-engagement—over a 7-year time period and investigating potential disparities using medical record data from a centralized clinic that serves the majority of PrEP patients in the Providence, Rhode Island area. By tracking PrEP follow-up visits, we construct a comprehensive and long-term PrEP retention profile among patients at high risk of HIV infection. This study addresses the knowledge gap and sheds light on future strategies to improve retention in PrEP care.

Methods

We reviewed medical records of cisgender patients attending a hospital-based PrEP clinic between 2013 and 2019. During the study period, all PrEP patients were prescribed daily oral treatment with tenofovir disoproxil fumarate/emtricitabine (TDF/FTC) and tenofovir alafenamide (TAF)/FTC. Data reviewed include basic demographic (race, ethnicity, gender) and socioeconomic (insurance status, education level, income) characteristics, sexual HIV risk behaviors in the past 6 months, lifetime drug use, and gonorrhea and chlamydia diagnosis. These data were collected at patients' initial PrEP clinic visits. We reviewed clinic visit attendance and classified patients as LTFU if they did not attend a PrEP follow-up appointment within 98 days. This duration was determined by the median number of days between follow-up visits in this data set. Re-entry to care was defined as individuals who missed at least two consecutive follow-up periods (196 days, 6.54 months) and later attended a follow-up appointment. Review of clinic data was approved by the Miriam Hospital Institutional Review Board.

Bivariate analyses were conducted to compare the demographic and behavioral characteristics between individuals retained in care and individuals LTFU at initial engagement in PrEP care (before any re-entry). Chi-square tests were used for categorical variables and Kruskal–Wallis tests were used for continuous variables. Recurrent event survival analyses were performed to explore factors associated with LTFU in PrEP care. We used the Prentice, Williams, and Peterson (PWP) model analyses,²⁷ which orders multiple events by stratification based on the number of prior events during the entire follow-up period. All participants are at risk of LTFU at the initial engagement, but only those who were LTFU in the previous stratum are at risk for the successive stratum. Retention in PrEP care was a time-to-event variable. The start time point was the date of the first PrEP visit, or the date of re-engagement in PrEP care. The end time point was either the date of loss to follow-up or the end of the study. Confounding variables were determined by direct acyclic graphs and a priori. An alpha of 0.05 was set as the significance level. All analyses were conducted in Stata 15.0 (StataCorp LLC, College Station, TX). Participant consent was waived for this study as data were collected through a retrospective review of medical records. The study was approved by The Miriam Hospital Institutional Review Board.

Results

Demographic characteristics

Of 654 cisgender PrEP patients, the median age was 31 years old [interquartile range (IQR): 25, 43]. The majority were White (64%), non-Hispanic (81%), and male (96%), had a college education or above (62%), and had health insurance (97%; Table 1). Those who were young adults and had a high school or less education were less likely to be retained in PrEP care than their counterparts. Individuals with higher income were more likely to be retained in PrEP care.

Table 1.

Demographics and Behavioral Characteristics Among Pre-Exposure Prophylaxis Patients in a Public Clinic

	PrEP persistence at the initial engagement
	Retained in care		Loss to follow-up		p
	N = 274	%	N = 380	%
Age (median, IQR)	34 (27, 47)		30 (24, 39)		<0.001
Race					0.1
White	187	70.3	233	62.3
AA/Black	20	7.5	39	10.4
Other	59	22.2	102	27.3
Ethnicity					0.45
Non-Hispanic	217	82.5	298	80.1
Hispanic/Latino	46	21.2	74	24.8
Gender					0.07
Cis-male	268	97.8	360	94.7
Cis-female	6	2.2	20	5.3
Sexual orientation					0.003
Same-sex only	243	89.0	299	78.9
Bisexuality	17	6.2	51	13.5
Heterosexuality	13	4.8	29	7.7
Insurance status					0.08
Insured	263	96.0	374	98.4
Uninsured	11	4.0	6	1.6
Education					<0.001
high school or less	68	26.2	150	41.0
college or above	192	70.1	216	56.8
Income (median, IQR)	50K (27K, 80K)		30K (10K, 55K)		<0.001
HIV-positive male partner					0.02
No	204	86.1	249	78.1
Yes	33	13.9	70	21.9
>5 Male sex partners					0.72
No	73	26.6	106	27.9
Yes	201	73.4	274	72.1
Condomless anal sex					0.29
No	46	19.4	51	16.0
Yes	191	80.6	268	84.0
Drug use, ever					0.48
No	145	53.1	190	50.1
Yes	128	46.9	189	49.9
Chlamydia positive any site (urine, rectal, and pharyngeal)					0.65
Negative	245	93.2	306	92.2
Positive	18	6.6	26	6.9
Gonorrhea positive any site (urine, rectal, and pharyngeal)					0.69
Negative	237	90.5	297	89.5
Positive	25	9.2	35	9.2
Observation time (median, IQR)	16.1 (5.8, 35.8)		6.4 (3.2, 15.3)		<0.001

IQR, interquartile range; PrEP, pre-exposure prophylaxis.

LTFU and re-engagement at each period of engagement in PrEP care

By the end of this study period, 185 (28%) participants were retained in care at their initial engagement, before any loss to follow-up and subsequent re-engagement, while 469 (72%) participants were LTFU. The median observation time among those LTFU was 6.2 months (IQR: 3.2, 12.0), compared with 10.1 among those retained (IQR: 4.1, 20.9; p < 0.001) during their initial engagement. Of 469 PrEP patients who were ever LTFU, 27% (127) had at least one episode of re-engagement in care. During the 6-year study period, we observed a maximum of six engagements in PrEP care by our definition. The retention and LTFU status for each period of engagement in care are listed in Table 2.

Table 2.

Retention in Pre-Exposure Prophylaxis Care by Number of Engagements in Pre-exposure Prophylaxis

Order of events	No. of patients eligible to engage in care	Total number engaged in care	Retained			LTFU
			n	%	Time retained/month (median, IQR)	n	%	Time retained/month (median, IQR)
First engagement	654	654	185	28.3	10.1 (4.1, 20.9)	469	71.7	6.2 (3.2, 12.0)
Second engagement	469	127	42	33.1	6.6 (3.1, 12.8)	85	66.9	3.2 (3.2, 9.4)
Third engagement	127	35	10	28.6	4.3 (2.0, 5.8)	25	71.4	3.2 (3.2, 9.9)
Fourth engagement	35	9	4	44.4	6.5 (4.5, 10.1)	5	55.6	3.2 (3.2, 6.5)
Fifth engagement	9	3	0	0.0		3	100.0	3.2 (3.2, 8.5)
Sixth engagement	3	1	1	100.0	2.5 (2.5, 2.5)	0	0.0

IQR, interquartile range; LTFU, lost to follow-up.

Behavioral characteristics and sexually transmitted infections

In this study, 83% reported having exclusively same-sex partners, 92% reported sex with more than five male partners in the past 6 months, 83% reported condomless anal sex in the past 6 months, and 49% reported any lifetime drug use. The prevalence of chlamydia and gonorrhea at first visit was 7% and 9%, respectively. Individuals who reported having exclusively same-sex partners were less likely to be LTFU compared with bisexual and heterosexual patients. Those reporting sex with an HIV-positive male partner in the past 6 months were more likely to be LTFU. Patients with more than five sex partners in the past 6 months were more likely to be retained in PrEP care. A higher rate of chlamydia diagnoses was observed in the LTFU group.

Multivariate Cox proportional hazard analysis

In the crude Cox hazard regression models, age, gender, and race were significantly associated with loss to follow-up. A 1-year increase in age was associated with a 1% lower hazard of loss to follow-up [crude hazard ratio (cHR): 0.99; 95% confidence interval (CI): 0.98–0.99; Table 3]. Women were more likely to be LTFU than men (cHR: 1.37; 95% CI: 0.86–2.18). Black patients had 1.38 times the hazard of being LTFU than White patients (cHR: 1.38, 95% CI: 1.02–1.89). In a multivariate model adjusted for age, gender, race, and ethnicity, those with a college education or above were 31% less likely to be LTFU compared with those with a high school education or lower [adjusted HR (aHR): 0.69; 95% CI: 0.57–0.82]. In a separate model adjusted for age, race, and ethnicity, both bisexual (aHR: 2.25; 95% CI: 1.70–2.99) and heterosexual (aHR: 2.35; 95% CI: 1.15–4.82) patients were more likely to be LTFU than those with exclusively same-sex partners.

Table 3.

Multivariate Proportional Hazard Regression of Persistence in Pre-Exposure Prophylaxis Care, 2013–2019

Variables	Hazard of PrEP nonpersistence
	cHR/95% CI	Adjusted hazard ratio (aHR)/95% CI
Age	0.99 (0.98–0.99)
Gender
Male	1.00
Female	1.37 (0.86–2.18)
Race
White	1.00
Black	1.38 (1.02–1.89)
Other	1.01 (0.84–1.22)
Ethnicity
Non-Hispanic	1.00
Hispanic	1.15 (0.94–1.41)
Educationa
High school or below	1.00	1.00
College or above	0.65 (0.55–0.77)	0.69 (0.57–0.82)
Sexual orientation
Same-sex only	1.00	1.00
Bisexuality	2.25 (1.69–2.99)	2.25 (1.70–2.99)
Heterosexuality	1.74 (1.15–2.64)	2.35 (1.15–4.82)

^aAdjusted for age, gender, race, and ethnicity.

CI, confidence interval; PrEP, pre-exposure prophylaxis.

Kaplan–Meier survival curves

Figure 1 shows the Kaplan–Meier survival curves of retention in PrEP care from 2013 to 2019 by the order of loss to follow-up. We observed a steep decrease in the first 3 months and a gradual decrease afterward. Of 654 PrEP patients during their initial engagement in PrEP care, 67% of patients retained in care at 3 months, 52% at 6 months, 33% at 1 year, 20% at 2 years, 13% at 3 years, 7% at 4 years, 5% at 5 years, and 5% at 6 years. During the second and third periods of engagement in care, a slowing decrease in retention in care was observed, which was different from the initial period of engagement. As we only had a limited number of patients who had the fourth, fifth, and sixth periods of engagement, the Kaplan–Meier survival curves were less likely to reflect the overall retention in care.

FIG. 1.

Probability of retention in PrEP care among patients in Rhode Island. PrEP, pre-exposure prophylaxis.

Discussion

This study is among the first to evaluate loss to follow-up and re-engagement over a long time period and presents several critical findings about long-term retention in PrEP care using longitudinal clinical data. The median length of time retained in PrEP care during the initial period of engagement was 1.6 times longer among patients who were still retained in PrEP care by the end of the study than those who were LTFU. Around 52% of PrEP patients were retained at the first 6 months, and 33% at 1 year, which is comparable with findings in other studies.^{21,23,28–31} A study in New York reported 42% retention in PrEP care at 6 months.²⁸ In another study, 39% of patients were retained 6 months after initiating PrEP.³⁰ Our retention rate at 12 months was lower than that in studies in San Francisco and Chicago, which reported 47% and 43% retention rates at 12 months, respectively.^23,31 Given our findings and evidence in the literature, PrEP patients are at a high risk of loss to follow-up at the early stage of PrEP initiation.

Several factors could contribute to loss to follow-up at the early stage of PrEP initiation, including side effects, financial or insurance issues, perceived low risk of HIV infection, and PrEP-related stigma.^22,32,33 At least 2% of PrEP patients have short-term side effects, including nausea, abdominal cramping, vomiting, dizziness, headache, and fatigue.^34–36 Patients may not tolerate these side effects and choose to discontinue PrEP. Awareness of side effects and a follow-up visit 2–4 weeks after PrEP initiation may reduce loss to follow-up at the early stage. We also found that income and insurance were associated with retention in PrEP care in our study. Support in applying for financial assistance programs to cover the cost of clinical visits and PrEP may also reduce loss to follow-up. A few PrEP patients discontinued PrEP after initiation and could re-engage in PrEP care later.³³ We also observed that 27% of patients who were LTFU during their initial period of engagement had at least one period of re-engagement in PrEP care. PrEP-related stigma is also a contributor to loss to follow-up. Interventions to reduce PrEP-related stigma may improve retention in PrEP care.³⁷ Providers' suggestion and support of PrEP use have also been suggested as effective solutions to improve retention in PrEP care in a variety of diverse situations.^38,39

Racial disparities are also pertinent to retention in PrEP care. Prior research shows that B/AA MSM are less likely to be aware of and initiate PrEP^13,40; in this study, they were also less likely to be retained in PrEP care. We found that being B/AA was associated with a 38% increase in the hazard of loss to follow-up compared with being White. H/L PrEP patients had poorer retention in PrEP care, but this was not statistically significant in this study. This null finding may be due to the limited number of H/L PrEP patients in our study population. As B/AA persons experience a disproportionate burden of HIV compared with other populations, innovative interventions are urgently needed to promote PrEP uptake and retention in PrEP care in this population.

A college education or above was associated with a 31% decrease in hazard of loss to follow-up in our study, which is comparable with prior research of PrEP retention.³⁰ This finding suggests that interventions should target less educated PrEP patients to improve retention in PrEP care. Individuals who had both male and female partners were more likely to be LTFU than their same-sex partnering counterparts. These individuals could still be at high risk of HIV infection, and may be less likely to engage in care because of messaging around the use of PrEP, which has disproportionately been aimed at sexual minority men. Future research should explore ways to more adequately engage this population as there is still potential to acquire and transmit HIV. Tailored interventions should focus on both promoting PrEP uptake and retaining in care among these populations.

Expanding our understanding of how PrEP patients interact with care may enable practitioners to redesign social support and patient safety nets to reach our most vulnerable populations: MSM of color and of low socioeconomic status. Engaging and maintaining PrEP care with these MSM populations by increasing long-term access, adherence, and personal agency will contribute to a dramatic reduction in HIV incidence.

We used a longitudinal clinical data set to reflect the natural trajectory of retention in PrEP care over 7 years and provided valuable insights to develop targeted and tailored interventions to improve retention in PrEP care. We conducted a recurrent event survival analysis, which allowed us to model patients falling out of care and re-engaging in care over time, which is arguably more representative of the PrEP clinical care continuum. This approach allowed us to best estimate scenarios of engagement in PrEP care and to improve power for detecting true differences if they exist. Although our study focused on an academic PrEP clinic, our study findings are comparable with studies in other settings with either a small or large patient population.

This longitudinal study shows that PrEP patients are at high risk of loss to follow-up at the early stage of PrEP initiation. Notably, however, 27% of patients who were LTFU after PrEP initiation were later reengaged in PrEP care. We observed racial and socioeconomic disparities in retention in PrEP care, as patients who were persons of color, with lower income, and with less education had increased hazard of being LTFU. Intensive interventions to address clinic engagement at the early stage of PrEP initiation could be promising to prevent loss to follow-up among those likely to discontinue. Targeted interventions are needed to focus on populations currently underserved by existing services to improve retention in PrEP care.

Authors' Contributions

P.A.C., A.S.N., and J.T. designed the study. M.C.M., M.Ma., S.C.N., C.C., and A.A. performed data collection. J.T. conducted the analysis. J.T. and M.C.M. wrote the article. R.W., B.G.R., C.S., M.Mu., M.Ma., S.C.N., C.C., A.A., A.S.N., and P.A.C. assisted in revisions and reviewed and approved the final version of the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This study was supported by the National Institute of Mental Health (K01MH19660 and R01MH114657) and the National Institute of Allergy and Infectious Diseases (P30AI042853) at the National Institutes of Health.