Key Points

Question听 What is the association between prescribing clinician specialty and patients not picking up (prescription reversal/abandonment) their initial preexposure prophylaxis (PrEP) prescription?

Findings听 In this cross-sectional study of 37鈥�003 patients who were prescribed PrEP that used claims data, primary care practitioners (PCPs) prescribed to the largest proportion of patients, followed by other specialty clinicians and infectious disease (ID) specialists. Compared with PCPs, patients of ID specialists and other clinicians had lower and higher odds of reversals and abandonments, respectively.

Meaning听 The study results suggest that PCPs are a critical entry point for PrEP care, and non-ID clinicians may especially benefit from greater training, clinical support, and systems for adherence monitoring to enhance PrEP care.

Importance听 Clinicians are a key component of preexposure prophylaxis (PrEP) care. Yet, no prior studies have quantitatively investigated how PrEP adherence differs by clinician specialty.

Objective听 To understand the association between prescribing clinician specialty and patients not picking up (reversal/abandonment) their initial PrEP prescription.

Design, Setting, and Participants听 This cross-sectional study of patients who were 18 years or older used pharmacy claims data from 2015 to 2019 on new insurer-approved PrEP prescriptions that were matched with clinician data from the US National Plan and Provider Enumeration System. Data were analyzed from January to May 2022.

Main Outcomes and Measures听 Clinician specialties included primary care practitioners (PCPs), infectious disease (ID), or other specialties. Reversal was defined as a patient not picking up their insurer-approved initial PrEP prescription. Abandonment was defined as a patient who reversed and still did not pick their prescription within 365 days.

Results听 Of the 37鈥�003 patients, 4439 (12%) were female and 32鈥�564 (88%) were male, and 77% were aged 25 to 54 years. A total of 24鈥�604 (67%) received prescriptions from PCPs, 3571 (10%) from ID specialists, and 8828 (24%) from other specialty clinicians. The prevalence of reversals for patients of PCPs, ID specialists, and other specialty clinicians was 18%, 18%, and 25%, respectively, and for abandonments was 12%, 12%, and 20%, respectively. After adjusting for confounding, logistic regression models showed that, compared with patients who were prescribed PrEP by a PCP, patients prescribed PrEP by ID specialists had 10% lower odds of reversals (odds ratio [OR], 0.90; 95% CI, 0.81-0.99) and 12% lower odds of abandonment (OR, 0.88; 95% CI, 0.78-0.98), while patients prescribed by other clinicians had 33% higher odds of reversals (OR, 1.33; 95% CI, 1.25-1.41) and 54% higher odds of abandonment (OR, 1.54; 95% CI, 1.44-1.65).

Conclusion听 The results of this cross-sectional study suggest that PCPs do most of the new PrEP prescribing and are a critical entry point for patients. PrEP adherence differs by clinician specialties, likely due to the populations served by them. Future studies to test interventions that provide adherence support and education are needed.

There are approximately 1.2 million individuals in the US who are living with HIV, with around 34鈥�800 incident HIV cases as of 2019.^1,2 Preexposure prophylaxis (PrEP) has been proven to be associated with a reduced risk of HIV infection by 99% when taken as prescribed,³ but uptake has been suboptimal. The US Centers for Disease Control and Prevention have suggested that an additional 1.2 million adults in the US have an indication for PrEP⁴; however, only approximately 224鈥�000 received a PrEP prescription in 2019.^5,6 Even after being prescribed PrEP, many patients do not or are unable to pick up their prescriptions from their pharmacies. When a patient does not pick up an insurer-approved PrEP prescription at the pharmacy point of sale, the pharmacy returns it to stock and registers it as a prescription reversal. Our previous research found that among those who were newly prescribed PrEP in the US, 19% did not pick up their first prescription from the pharmacy, an indicator of primary nonadherence; the risk of HIV acquisition among these patients was more than 3 times higher than those who did pick up their initial prescription.⁷ We also found that of patients whose prescriptions were reversed, most (71%) abandoned their prescription, not picking it up at all within a year after the date of prescription.⁷ While barriers at the pharmacy point of sale, including cost and stigma,^5,6 may be associated with an individual PrEP reversal,⁷ there may be further upstream factors that are also associated with PrEP primary adherence.

As demonstrated in other chronic illnesses, clinicians play a key role in whether patients initiate or continue with a prescription.^8-11 Regarding PrEP uptake, there remain opportunities and challenges specific to the type of clinician (eg infectious disease [ID] specialist, primary care specialist) who is prescribing PrEP to patients.¹² Primary care practitioners (PCPs) may be well positioned as PrEP prescribers, as they see patients early in the PrEP decision-making process and most often see HIV-negative patients who have factors associated with HIV seroconversion. PCPs have long-standing relationships with patients that might be associated with better adherence to treatment suggestions.¹³ PCPs may alternatively refer their patients to ID specialists for further PrEP care. ID specialists may be well positioned to provide HIV risk and adherence counseling for PrEP given the nature of their work in providing adherence support for antiretroviral therapy and specialized training in preventing sexually transmitted infections.¹⁴ However, ID specialists may not be as straightforward to access as PCPs due to referral processes that may be required by insurance. The types of health insurance plans and geography may play a role in which type of clinician a patient sees and to whom they are referred.

Yet, to our knowledge, no prior studies have quantitatively described PrEP uptake by clinician specialty in the form of prescription reversal and abandonment, which are key markers of successful PrEP initiation. This may be important to understand when identifying ways in which clinicians can help improve PrEP initiation and reduce HIV incidence. Therefore, our aim was to understand how clinician specialty may be associated with PrEP prescription reversal and abandonment.

We evaluated 2015 to 2019 pharmacy claims data on HIV PrEP prescriptions in the US from the Symphony Health Solutions Integrated Data Verse (Source Healthcare Analytics, LLC [SHA]) and matched that to clinician characteristics data from the National Plan and Provider Enumeration System (NPPES) File. Data were analyzed from January to May 2022. The SHA proprietary database has claims for more than 274 million patients across all US states and insurance types. An estimated 80% to 85% of all PrEP prescription claims are captured in this dataset,¹⁵ which includes claims paid by public and private payers, assistance programs, and cash and covers the full prescription lifecycle from adjudication through final pick-up status from the patient.⁷ The claims dataset only excludes integrated delivery networks (eg, Kaiser Permanente, Veterans Affairs hospitals, US Department of Defense) that do not make data available outside of their networks. For a subset of patients, SHA includes patient-level demographic characteristics through a partnership with KBM Group. All claims were deidentified; thus, the study received institutional review board exemption as well as a waiver of consent.

We analyzed medical and pharmacy claims for patients with index PrEP prescription claims for tenofovir disoproxil fumarate/emtricitabine (TDF/FTC), but not tenofovir alafenamide/emtricitabine claims because it was not approved by the US Food and Drug Administration for PrEP during the study period. Tenofovir alafenamide/emtricitabine is a newer version of tenofovir that is also delivered in tablet form, but it is taken in a much lower dosage than TDF/FTC. We excluded rejected claims because they were not associated with the study outcomes; rather, they represented issues with insurance prior authorization or administrative processing at the pharmacy.

We included patients with a newly prescribed insurer-approved PrEP prescription from 2016 to 2018 with no PrEP use during the previous year starting in 2015 and followed up through 2019. We considered the first PrEP prescription during this period as the patient鈥檚 index prescription. We determined this eligible PrEP claim through previously established algorithms^16-18 on which we have previously published.⁷ To ensure that we captured TDF/FTC for PrEP rather than HIV treatment, postexposure prophylaxis, or hepatitis B virus, we excluded people who had a TDF/FTC prescription 28 days or less or more than 91 days or who had an International Classification of Diseases, Ninth Revision (ICD-9)/ICD-10 diagnosis code for HIV or hepatitis B virus and had any claims for HIV medications during the year before or 30 days after the index prescription (based on the US Centers for Disease Control and Prevention definition).¹⁶ We did not have access to patient zip codes. Instead, we used clinician zip codes as a proxy for creating zip code鈥搇evel covariates and limited our sample to patients who lived in the same state as their clinician鈥檚 practice.

We also excluded anyone with a PrEP claim within the 365 days before the index prescription, as our focus was on patients with newly approved claims, and those with no pharmacy activity within 365 days after the index date.¹⁹ Finally, for patients with multiple claims with differing statuses being approved or reversed on the same date, we prioritized the status of approved vs reversed.

We defined clinician specialty by the health care network taxonomy in the NPPES file, which covers most US clinicians.^20,21 For clinicians without a clear specialty from the NPPES files, we used the SHA data to supplement specialty information. After including information from claims, 0.1% of clinicians were missing specialty information and thus were excluded. We reclassified clinician specialty into primary care, ID, or other specialties. PCPs were considered any physicians or advanced practice clinicians (APCs) with family practice or internal medicine specialty codes associated with their NPPES health care network taxonomy. This was the reference group. ID physicians included adult and pediatric physicians. The other specialty category included any clinician who prescribed PrEP but was not classified as a PCP or an ID specialist (eg, emergency medicine physicians, obstetrics/gynecology physicians (OB/GYNs), and APCs from various specialties). We excluded any clinician specialties that did not have prescribing rights (eg, registered nurses, naturopaths, and students).

We evaluated 2 binary outcomes associated with PrEP prescription fills. Prescription reversal was defined as when a patient did not pick up their insurance-approved PrEP initial prescription, leading the pharmacy to reverse the prescription claim back to the insurer. A reversal is not defined using any specific period; it is predetermined in the claims as reversed or not based on each pharmacy鈥檚 policy. Abandonment was defined as when a patient who initially had a prescription reversal continued to not pick up the insurance-approved initial prescription for 365 days.

Patient-level covariates included sex, age, race and ethnicity (reported from the KBM Group), household income, education, Charlson Comorbidity Index score, and insurance type. Zip code鈥搇evel geographic covariates included rural/urban status and Area Deprivation Index tertile.²² Patient race and ethnicity, income, and education data were linked to claims data from external sources available to the claims data vendor.

First, we assessed differences in the characteristics of patients by clinician specialty using Pearson 蠂² tests. Next, we compared the unadjusted prevalence of each outcome by clinician specialty. Then, we imputed for missing data on household income and education using multiple imputation using chained equations with 30 imputed datasets. Using the imputed datasets, we calculated the adjusted odds of reversals and abandonment using logistic regression with robust standard errors, controlling for covariates. Finally, we conducted a sensitivity analysis on the entire sample of patients using the imputed datasets, including those who did not live in the same zip code as their clinician. Analyses were conducted using Stata (StataCorp), and statisitical significance was set at 伪鈥�=鈥�.05.

The sample included 37鈥�003 patients treated by 16鈥�337 clinicians. Among all patients, 24鈥�604 (66.5%) received prescriptions from PCPs, 3571 (9.7%) from ID specialists, and 8828 (23.9%) from a clinician from another specialty (eTable 1 in Supplement 1). eTable 1 in Supplement 1 also details the most prevalent prescribing specialty types classified as other specialty clinicians listed after PCPs and ID specialists. PCPs, ID specialists, and other specialty clinicians prescribed to a mean (SD) of 2.3 (5.4), 2.6 (3.7), and 2.2 (6.5) patients, respectively, during the 3 years of the study, with the rate of prescription remaining relatively steady during this time.

As shown in Table 1, the highest percentage of new PrEP claims was among patients aged 25 to 34 years. Many patients who received a prescription from a PCP were male, non-Hispanic, had an income of $100鈥�000 or greater, had an associate degree or more, had commercial insurance, and/or had clinicians in areas of low deprivation compared with those prescribed by ID specialists and other specialists. Many patients prescribed by ID specialists were more likely to be non-Hispanic, Black, have an income of $30鈥�000 or less, have some college education, have Medicare, and receive care in urban zip codes and/or in areas with high deprivation compared with those who received prescriptions from PCPs and other specialists. Many patients prescribed by other clinicians were more likely to be female, be of other or unknown races, and/or have Medicaid compared with PCPs and ID specialists.

The Figure displays the prevalence of reversals and abandonments by prescribing clinician specialty. The results display that other specialty clinicians have the highest prevalence of reversals and abandonments among the clinician categories. The prevalence of reversals for patients treated by PCPs, ID specialists, and other specialty clinicians was 18%, 18%, and 25%, respectively; for abandonments, it was 12%, 12%, and 20%, respectively.

After adjusting for covariates of age, sex, race and ethnicity, household income, education, comorbidity, insurance type, rural/urban status, and area-level deprivation (Figure; Table 2), patients who were prescribed PrEP by an ID specialist had lower odds of reversing (odds ratio [OR], 0.90; 95% CI, 0.81-0.99) or abandoning (OR, 0.88; 95% CI, 0.78-0.98) their prescription compared with patients who received their prescriptions from PCPs. However, patients who were prescribed PrEP by other specialty clinicians had higher odds of reversing (OR, 1.33; 95% CI, 1.25-1.41) or abandoning (OR, 1.54; 95% CI, 1.44-1.65) their prescription compared with patients who received prescriptions from their PCPs. In our sensitivity analysis with the sample of patients who met the inclusion criteria except for the same state restriction, results were similar or slightly attenuated compared with the main analysis (eTable 2 in Supplement 1).

To our knowledge, this cross-sectional study is among the first studies to characterize variations in PrEP uptake in the context of initial prescription reversals and abandonments. We found that almost two-thirds of the US population鈥揵ased sample of 37鈥�003 patients received a PrEP prescription from a PCP, less than one-tenth from an ID specialist, and (surprisingly) about one-quarter from other specialty clinicians, largely consisting of advanced practice clinicians of unspecified specialties and clinicians in emergency medicine, women鈥檚 health, and more. Other studies using claims data from different vendors provided similar estimates, with 68% of patients seeing PCPs and 14% of patients seeing ID specialists.¹³

Nevertheless, our study found that 19.7% and 14.0% of the PrEP prescriptions were reversed and abandoned, respectively. Absolute differences in prevalences of reversals and abandonments between differing clinician types were small. However, after controlling for differences in confounders, patients who were prescribed PrEP by an ID specialist had 10% lower odds of reversals and 12% lower odds of abandonment for their initial PrEP prescription compared with patients who receive prescriptions from PCPs. While this difference in odds between PCPs and ID specialists is small to moderate, the greater difference was when comparing patients seeing PCPs with patients seeing other specialty clinicians. Patients who received prescriptions from other specialty clinicians had 33% higher odds of reversals and 54% higher odds of abandonment of their initial PrEP prescription.

The study results suggest that PCPs are an important initial access point for PrEP, given that they do the bulk of PrEP prescribing; however, the odds of reversals and abandonments were slightly lower for patients seeing ID specialists and substantially higher for patients seeing other specialties. Given that the prescriptions written by each clinician were insurer-approved in our study, differences in clinician resources and the motivation to go through prior authorization approvals are unlikely to explain our results. Instead, our results may be because patients who have high readiness/motivation or who are high risk for HIV may be more likely to be referred to ID specialists or possibly due to the specialty training that ID specialists receive compared with PCPs or other specialists.

Regarding the first possibility, while we did not have information about patients鈥� HIV risk, ID clinicians in our sample were more likely to prescribe PrEP to individuals from high HIV risk groups, including patients who are aged 18 to 24 years, Black, Hispanic, and/or have economic disadvantages.^23-25 These findings suggest that ID specialists may prescribe to an at-risk population for HIV that may have greater motivation, readiness, or need to pick up PrEP. Additionally, because PCPs are often an entry point into the PrEP decision-making process, those who see PCPs may having differing readiness and certainty around taking PrEP compared with those who see ID specialists.²⁶ Individuals purposefully seeking sexual health services may be referred or specifically seek an ID specialist from whom they then receive PrEP care. Therefore, there may be overall differences in the populations seeing PCPs vs ID specialists.

Previous literature also supports the hypothesis that the training that ID specialists receive may be associated with lower PrEP reversals and abandonments. One study detailed that only 26% of PCPs were comfortable prescribing PrEP compared with 76% of ID specialists, which suggests a potential training gap in PrEP care.²⁷ This study has also shown that PCPs would be more willing to prescribe PrEP if they had the skills and information to be confident in doing so.²⁷ Increasing PrEP-specific training opportunities for PCPs could be a way to enhance their ability to address patients鈥� PrEP-specific needs. Additionally, when patients receive prescriptions from ID specialists at HIV clinics, these clinics may have access to support services, like care coordination, peer navigators, and specialty pharmacies with automatic delivery,^26,28,29 that might help patients reduce the odds of reversing or abandoning their prescription. However, further studies are needed to understand the mechanisms through which ID specialists have reduced reversals and abandonments. Overall, our findings may indicate a need for additional training for PCPs in providing PrEP care, including education around PrEP content, sexual history taking, and navigating insurance barriers, as well as additional structural supports, such as care navigators.²⁶

Our finding of other specialty clinicians having higher odds of reversals and abandonments along with the prevalence of other specialty prescribing PrEP is novel because much of the literature does not consider them in the PrEP care conversation. However, prescribing by other clinicians is prevalent and by pharmacists is increasing; thus, they must be considered in this conversation.³⁰ For example, 1 study found 33% greater odds of abandonment for patients prescribed PrEP by a nurse practitioner or physician鈥檚 assistant compared with those who received a prescription from a physician.³¹ Previous studies have shown that emergency medicine clinicians and OB/GYN clinicians are similarly uncomfortable with their lack of PrEP content knowledge, uncertainty around referrals, and/or time-related barriers to discussing PrEP.^32,33 OB/GYN clinicians are often engaged in sexually transmitted infection care, and increasing their PrEP training may provide an opportunity for advancing PrEP care. This implies that PrEP education and resources, such as care navigators, are necessary for other specialists as well, and we cannot exclude them from the PrEP care conversation. Therefore, general PrEP training for other clinicians along with training specific to these unique contexts in which PrEP is prescribed is important to improve PrEP care. Patients seeing other specialty clinicians may especially benefit from adherence monitoring and support given the substantially higher reversal and abandonment compared with PCPs and ID specialists.

Our study had several limitations. First, we were limited by the SHA dataset in its nontrivial proportion of missingness for race and ethnicity and socioeconomic status variables; however, most national pharmacy claims datasets do not include demographic characteristics, making this dataset our best opportunity for including these characteristics as confounders, despite its missingness. Additionally, we imputed for missing values of education and income and analyzed the multiply imputed data in our main analysis. We found that results from the multiple imputed data were similar to those of results from analyses that used a missing data indicator to account for missingness, and our conclusions remained the same. Next, we were limited in that APCs with no listed specialty but who were in primary care or ID may have been misclassified as other specialty; nevertheless, this misclassification means that estimates for other specialty may be more similar to results for primary care and ID and that the results underestimate the actual differences between the groups. Another limitation was that the primary care and specialty categories may not be 100% accurate. For instance, a physician in internal medicine could not be doing primary care, and a physician in a listed specialty could be doing primary care. However, this classification scheme has been previously used by others and demonstrated to be an approximation of actual clinical practice.³⁴ Additionally, in this dataset, we did not know if those who saw an ID specialist saw a PCP or specialty clinician first because we did not have encounter information. We only had the prescribing clinician, so we did not have information on referral patterns. Another limitation we acknowledge is our assumption that the clinician and patient zip codes were to be similar enough that the clinician zip code could be a proxy for the patient zip code. However, this only affected our classification of rural/urban and Area Deprivation Index status, so we conducted our analysis without these geographic variables as confounders; our results were almost identical to those in our main analysis. We were also limited by our lack of knowledge around the pharmacy type that filled the PrEP prescription and clinics associated with ID physicians. There could be mail-order or hospital pharmacies. ID physicians could also be associated with large sexually transmitted infection or Ryan White Program鈥搒upported clinics. All of these have support services that could be associated with reduced odds of reversal. Therefore, including these pharmacies/clinics may underestimate reversals for ID clinicians without those support services, and the rate for ID clinicians might otherwise be higher and more similar to rates for PCPs. Finally, we were limited in that we did not have data on the motivations patients had for taking PrEP, resulting in potential residual confounding. Participants who are seeing ID specialists may be doing so because they have a higher perceived HIV risk and are seeking sexual health services; thus, they may have higher motivation to pick up PrEP. Also, the type of patient who sees an ID specialist may have already decided to take PrEP; thus, they may have greater motivation to adhere to the prescription schedule. This may contribute to the lower odds of reversals for ID specialists.

PCPs are a critical PrEP entry point for patients, as they do the majority of prescribing; however, many different types of clinicians are also involved in PrEP care and may have different patient populations, training, and clinical supports that pattern patient pick up of PrEP. While differences in reversals and abandonments by clinician specialty reflect that patients who are motivated toward PrEP may be more likely to be referred to ID specialists for care, our results may also support the need for additional structural supports and education for PCPs and other non-ID specialty clinicians. These supports may further equip them to provide enhanced care to PrEP-indicated patients and broaden the range of clinicians who feel equipped to prescribe PrEP, potentially improving overall PrEP use in the US.

Accepted for Publication: June 26, 2024.

Published Online: August 19, 2024. doi:10.1001/jamainternmed.2024.3998

Corresponding Author: Lorraine T. Dean, ScD, 615 N Wolfe St, Baltimore, MD 21205 (ldean9@jhu.edu).

Correction: This article was corrected on October 14, 2024, to fix errors in Table 1.

Author Contributions: Ms Bakre had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Bakre, Chang, Doshi, Goedel, Chan, Nunn, Dean.

Acquisition, analysis, or interpretation of data: Bakre, Chang, Doshi, Saberi, Nunn, Dean.

Drafting of the manuscript: Bakre, Chang, Doshi, Saberi, Nunn, Dean.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Bakre, Chang.

Obtained funding: Nunn, Dean.

Administrative, technical, or material support: Doshi, Nunn.

Supervision: Doshi, Goedel, Nunn, Dean.

Conflict of Interest Disclosures: Dr Chang reported being an employee of Janssen Scientific Affairs, LLC and holding stock in Johnson & Johnson. Dr Doshi reported grants from National Institutes of Health during the conduct of the study as well as grants from Janssen, Merck, and Spark Therapeutics and personal fees from AbbVie, Acadia, Janssen, Merck, Otsuka, and Takeda outside the submitted work. No other disclosures were reported.

Funding/Support: This work was supported by National Institutes of Health grants R21NR018387, R01NR017573, and R25MH083620-11.

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.