HIV pre-exposure prophylaxis in the time of COVID-19: How a robust and responsive HIV PrEP intervention can avert loss of HIV prevention coverage during a global pandemic

Introduction

Severe Acute Respiratory Syndrome coronavirus-2 (SARS-CoV-2), the virus that causes the coronavirus disease (COVID-19), can be transmitted through close proximity and physical contact with an infected individual. Mitigating behaviors to minimize the risk of transmission include physical distancing, which also restricts sexual contact, specifically casual sex. Recent data show a reduction in sex with casual and non-primary partners among gay and bisexual men^1–4. As a result, many cities and countries have seen a sharp reduction in the initiation and continuation of HIV pre-exposure prophylaxis (PrEP)¹. For example, a high-volume community clinic in Boston saw a 72% decline in new PrEP initiations and a nearly 3-fold increase in lapses in PrEP refills among existing patients since the beginning of the SARS-CoV-2 pandemic in the US⁵. PrEP programs in various other settings have observed similar trends^6,7.

However, other reports have noted that declines in PrEP use during the pandemic may not necessarily reflect changes in HIV risk^3,8. Indeed, respondents in one study reported an increase in the number of sex partners during the pandemic and many felt that it was not important to change their sexual behaviors during the pandemic⁸. Thus, disruptions in PrEP care due to pandemic restrictions have the potential to exacerbate HIV disparities in key populations and hamper national HIV prevention goals⁹.

Recently, studies found that young adults were more likely to report experiencing barriers to accessing sexual health services during the pandemic and that non-whites, Latinx, and those on public health insurance were more likely to have lapses in PrEP refills^3,5. To inform the development of strategies that can alleviate disruptions in PrEP care due to the pandemic, there is an ongoing need to evaluate the pandemic’s impact on PrEP uptake and persistence. In this study, we investigated the impact of the pandemic on referral for PrEP initiation, PrEP continuation, and loss to follow up from PrEP services among 12 San Francisco Department of Public Health (SFDPH) clinics.

Methods

The PrEP Optimization Intervention (PrEP-OI) study examined the impact of a PrEP panel management strategy involving a PrEP Coordinator and a web-based panel management tool on PrEP initiation in 12 SFDPH primary care clinics (three clinics under one administration with overlapping providers and drop-in services for adolescents and young adults were combined for a total of 10 clinical sites). The study began in August 2018 and included a three-month observational phase (August–October 2018), followed by a 10-month stepped wedge randomization phase (November 2018–September 2019), and a 12-month follow-up phase (October 2019–September 2020). The PrEP-OI study protocol has previously been published¹⁰. On March 16, 2020, the San Francisco shelter-in-place orders from the SARS-CoV-2 pandemic went into effect; therefore, over half of the follow-up phase occurred during the pandemic. Here, we report the impact of these public health orders on PrEP use among patients in study clinics.

The PrEP-OI intervention included two components: (1) a centralized PrEP coordination overseen by a PrEP Coordinator (a nonclinical support staff member) who outreached to patients who may benefit from PrEP, assisted in PrEP initiations, supported continuity of care for existing PrEP patients; and (2) an integrated web-based panel management tool called PrEP-Rx, which included an HIV risk assessment survey, a patient PrEP timeline, questions for PrEP Coordinators to ask patients at PrEP initiation and continuation visits, and automated reminders for PrEP Coordinators to order quarterly laboratory tests and/or follow-up with patients regarding PrEP adherence and side effects¹¹. PrEP Coordinators spoke to patients at baseline to assist in PrEP initiation and then at one week, one month, and quarterly thereafter.

During the shelter-in-place orders, we instituted several measures to maintain connection with patients and tailor PrEP care to their changing needs: (1) Using our already robust telehealth infrastructure, the PrEP Coordinators ensured continued connectivity with patients through telephone encounters, text messages, or emails. (2) Due to limited laboratory availability and many patients’ request to minimize clinic visits, we developed internal PrEP guidelines. For those due for their quarterly laboratory visit and refill, we provided an additional three-month PrEP supply and deferred laboratory tests for three months if patients were otherwise adherent and were not experiencing any PrEP side effects; provided an additional one-month PrEP supply and deferred laboratory tests for one month if patients had experienced challenges with adherence but were not experiencing any side effects; and communicated with their provider about next steps for patients experiencing any side effects, symptoms of sexually transmitted infections, or symptoms of acute HIV infection. (3) With all who self-identified as men who have sex with men or transgender women, the PrEP Coordinators communicated the nuances of 2-1-1 PrEP¹². These individuals were notified that this dosing strategy was an alternative to consider if they wanted to discontinue daily PrEP yet wanted to have an HIV prevention option in case of future sexual encounters. Other available measures included assisting patients in receiving PrEP by mail, providing patients with ride-sharing vouchers in case of need for in-person clinic visits, and communicating non-PrEP health-related questions/concerns to patients’ provider.

From January 1, 2020 through October 31, 2020, we captured PrEP use on a monthly basis from the PrEP-Rx database reports (due to technical issues, PrEP-Rx backup reports for June and July were missing). Based on the PrEP Coordinators’ assessment of patients, the PrEP status of each patient was categorized as active, referred/outreached, needs support, or lost to follow-up (LTFU). “Active” patients were those who reported taking PrEP (daily or 2-1-1 dosing) or starting PrEP within a week (e.g., awaiting baseline laboratory results). The “referred/outreached” status was used for patients who were newly referred to PrEP-OI and were awaiting communication with the PrEP Coordinator. “Needs support” was used for patients whose PrEP care was on hold and/or needed additional support prior to PrEP re-initiation. The PrEP Coordinators contacted these individuals on a weekly basis for one month before categorizing them as “LTFU”. The “LTFU” status was used for patients who did not respond to any telephone calls, text messages, or other forms of communication. The PrEP Coordinators contacted LTFU patients on a monthly basis for six months. We combined patients who were in the “needs support” and “LTFU” categories for ease of analysis and because these two groups required more frequent follow-up from PrEP Coordinators.

Data reports summarizing the status of patients were generated at the end of each month for all clinics. For the month of October, a demographic data report (age, race/ethnicity, and gender) was also generated and univariate analyses summarized data. We tested for trends over time in the counts of active, referred, and LTFU/needs support patients per month. Since outcomes were counts, we used negative binomial regression to assess the association of month with each outcome¹³. A mixed effects modeling approach with random intercepts was used to account for the correlation among repeated measures within participants. Linearity of the time trend for each outcome was assessed by comparing a model with month treated as a straight line (linear). Cubic splines (CS) were included to allow for curvature in the outcome-to-months association to a model containing only the linear effect of month which allows only a straight line association between the outcome and months. The fit of these two competing models were compared using Schwarz’s Information Criterion (SC)¹⁴; the model with the lowest SC had the best fit to the data and was selected as final. A follow-up analysis was conducted to explore whether counts from January through March 2020 (reference group) differed from those from April through October 2020. Rate ratios (RR) and p-values are reported for effects from the final models. All data analyses were conducted with Stata 16.1 (College Station, TX).

Results

In January 2020, the study included 267 patients and by the end of October 2020, 268 patients were being followed by the PrEP Coordinators. The number of patients during this timeframe fluctuated from 261 to 287. Patients had a mean age of 35.8 years (SD=16.2). They were 48.9% cis-gender men, 14.9% transgender women, and 9.0% cis-gender women; 31.7% Latino, 20.5% White, and 12.3% African American.

Throughout January–October 2020, the number of patients in the active (mean=196; range=177–211; SD=12.3), referred (mean=31; range=26–41; SD=4.4), and LTFU/needs support (mean=41; range=28–58; SD=9.0) statuses were essentially unchanged (Figure).

Figure.

Changes in PrEP use among SFDPH primary care clinics from January to October 2020 (missing data for June and July 2020).

Models with a linear trend (i.e., straight line only) for months were preferred to the corresponding models containing cubic splines (i.e., curvature) for active (linear=481.16; CS=489.76), referred (linear=338.12; CS=349.19), and LTFU/needs support (linear=381.03; CS=382.94) patients. Consistent with the Figure, no statistically significant time trends were observed for active (RR=1.002, p=0.82), referred (RR=0.98, p=0.23), and LTFU/needs support (RR=0.93, p=0.22) patients. Similar findings were obtained when comparing April–October to January–March for active (RR=1.04; p=0.43) and LTFU/needs support (RR=0.82; p=0.11), though a non-statistically significant trend was observed for referred patients (RR=0.78; p=0.053) such that there were fewer referrals in April–October than in January–March.

Discussion

Our data indicate no reduction in PrEP use by patients in SFDPH primary care clinics following the SARS-CoV-2 pandemic shelter-in-place restrictions. This is in contrast to reports showing a substantial decline in PrEP use during this time^1,5. We believe that this sustained use of PrEP was due to ongoing and multi-pronged intervention efforts to communicate with patients via telehealth and text (particularly during a time when patients’ communication with their healthcare team was difficult), modify CDC guidelines to minimize real or perceived exposure risk to patients and clinic staff, and offer an alternative dosing strategy to tailor PrEP use to the changing needs of the patients.

In prior studies, the main reason for discontinuation of PrEP has been the reduction in frequency of sex and reduced sex partners^6,7. However, it is possible that the patients in our study continued to use PrEP because they did not experience changes in their sexual behaviors despite shelter-in-place restrictions. Other potential reasons for PrEP continuation include the individual’s desire to avoid undermining pill-taking habits, maintain reassurance of being protected from HIV, and the potential belief that PrEP may offer some protection against COVID-19². Importantly, reductions in frequency of sex or number of sex partners are likely temporary and will change as individuals develop “quarantine fatigue¹⁵” or as physical distancing guidelines ease. Future cohort studies should examine their data for “rebound” of sexual behaviors to baseline levels.

Unlike other studies that have used pharmacy refill data to determine PrEP use⁵, we preformed direct patient outreach. Pharmacy refill data is limited in that it assumes that patients only pick up medications from one pharmacy, it is not a measure of medication ingestion, and does not capture patients using event-based dosing. These data are difficult to obtain outside of closed healthcare systems. While it is possible that patients over reported their PrEP use, given the stability of the PrEP cohort over time, this is a less likely explanation for the lack of PrEP discontinuations during the pandemic.

Limitations of our study included the inability of PrEP-Rx to retrieve historical data resulting in the loss of two months of data (June and July), inability to separate patient categories (e.g., LTFU and needs support), or retrieve patient demographics from each month. However, given the relative linearity of the data, it is unlikely that these months were substantively different from the other months. All of the clinical sites were primary care clinics in San Francisco; therefore, results may not be generalizable to other non-primary care clinics or clinics outside of San Francisco. Lastly, we were limited in our ability to assess individuals’ sexual behaviors or examine reasons for ongoing PrEP use.

Due to the success of this program in maintaining HIV prevention measures during the pandemic and healthcare providers’ support for the program¹⁶, we are examining paths to integrate and sustain this intervention in SFDPH’s HIV prevention services. Future studies should examine a similar PrEP panel management strategy in other locations where HIV prevention services are needed.