|
Preexposure Prophylaxis Indication Criteria Underidentify
Black and Latinx Persons and Require Revision
|
|
|
Download the PDF here
Download the PDF here
PrEP screening tools predominantly composed of risk behavior data seem destined to perform poorly for Black MSM.....patients with private insurance, White patients, and MSM were more likely to be on PrEP, relative to HIV diagnoses
Among patients at an urban federally qualified health center with a sexual health focus, we found that PrEP:HIV ratios differed by insurance status, race/ethnicity, and gender and orientation. Specifically, patients with private insurance, White patients, and MSM were more likely to be on PrEP, relative to HIV diagnoses.
This work supports and extends previous research suggesting that the current CDC guidelines regarding PrEP are insufficient.3,9
In particular, our data showed that current guidelines under-identify transwomen and poorly identify ciswomen who could benefit from PrEP; there were also large racial/ethnic gaps in indication:HIV ratios. Identification relies to a great degree on provider-patient communication around intimate behaviors, some of which can be stigmatizing, as well as overcoming issues of medical mistrust.18-21
we should not ignore disparities in uptake. Insurance status was clearly associated with PrEP:indication ratios and many patients face financial barriers not only for PrEP prescriptions but also for the associated testing and follow up.28-31
In conclusion, we found important differences in PrEP:HIV by race/ethnicity, gender and orientation, and insurance, driven by both identification of PrEP candidates according to CDC guidelines and uptake. Indication:HIV ratios differed by race/ethnicity, gender and orientation, and insurance, suggesting that current PrEP guidelines do not ameliorate these disparities. Of particular note, specific indications are lacking for transgender individuals. Furthermore, we suggest that better patient-provider communication about sexual health in a nonstigmatizing context can improve PrEP access for those most in need, while structural changes to remove financial and other barriers are also needed.
CONCLUSIONS
Pyra et al. note that their work extends arguments regarding the insufficiency of current PrEP indication guidance and emphasize that improving access should not contribute to any further stigmatization. It is noteworthy that the CDC conducted some of the initial PrEP trials and provided some of the earliest guidance regarding PrEP. The US Food and Drug Administration was one of the first bodies to approve a medication for PrEP. This type of leadership is encouraging and merits confidence that current challenges with PrEP indication guidance are likely to be addressed in short order. Recently, the CDC predicted that in the absence of intervention 1 in 2 Black MSM, 1 in 5 Latinx MSM, and 1 in 11 White MSM will acquire HIV in their lifetimes. Given the high safety and efficacy of PrEP, it is hard to justify communication with any member of these groups that would not support PrEP use.
Black race was associated with lower ratios in all models, with the most extreme difference among transwomen for PrEP:HIV; Latino/a MSM, but not transwomen, had significantly lower PrEP:HIV and indication:HIV ratios.
The overall PrEP:HIV ratio was 15.8-there were 15.8 patients on PrEP for each incident diagnosis over the 4-year study period (Table 2). The observed PrEP:HIV ratio was highest for Whites (25.9) and lowest for Blacks (7.9); by insurance status, it was highest for those with private insurance (27.1) and lowest for those with Medicaid (9.3). Among key populations, White MSM had the highest PrEP:HIV ratio (26.3), and Black transwomen had the lowest (4.6). After adjustment, the PrEP:HIV ratio among Blacks was one third that of White patients (adjusted odds ratio [AOR] = 0.31; 95% confidence interval [CI] = 0.24, 0.41; Table 3); among Latino/a patients, it was roughly two thirds (AOR = 0.63; 95% CI = 0.48, 0.83). Compared with those with private insurance, the PrEP:HIV ratio for patients with any other insurance was between 22% and 39%. The PrEP:HIV ratio among transwomen was 54% that of MSM (AOR = 0.54; 95% CI = 0.38, 0.79), after we adjusted for other patient characteristics. There was no significant association between age and PrEP:HIV ratios.
----------------------------------------
Preexposure Prophylaxis Indication Criteria Underidentify Black and Latinx Persons and Require Revision
AJPH March 2020- Aaron J. Siegler is with the Department of Behavioral Sciences and Health Education, RollinsSchool of Public Health, Emory University, Atlanta, GA
HIV preexposure prophylaxis (PrEP) provides more than 99% protection against HIV transmission when taken as indicated. Side effects of PrEP are low, but costs of the medication and related clinical care are high. Efforts to end the HIV epidemic hinge on PrEP, with both mathematical models and real-world data suggesting that bringing PrEP to scale among groups with high HIV incidence will substantially affect the HIV epidemic.1 The US Centers for Disease Control and Prevention (CDC) provides guidance for PrEP indication to assist clinicians in determining who should be provided PrEP.
In this issue of AJPH, Pyra et al. (p. 370) analyze data on PrEP provision and PrEP indication from a federally qualified health care center in Chicago, Illinois. The authors present data in the form of several ratios: PrEP users divided by incident HIV diagnoses (PrEP:HIV, also termed PrEP-to-need),2 PrEP users divided by persons for whom PrEP is indicated (PrEP:Indication), and persons for whom PrEP is indicated divided by HIV diagnoses (Indication:HIV). The goal in using these ratios is to provide information regarding disparities in who is receiving PrEP. Pyra et al. found that Blacks, Latinx, and transwomen had substantially lower PrEP:HIV and Indication:HIV ratios than did White and cisgender men populations. Relative to their need as determined by incident HIV infection, these groups were less likely to be indicated for PrEP and also less likely to receive it.
Previously, cohort data from two studies found that behavioral- and sexually transmitted infection-based PrEP indication criteria, such as the CDC guidelines, perform poorly in predicting incident HIV infection among Black men who have sex with men (MSM).3,4 In fact, one of the cohort studies found that race alone better predicted HIV incidence than did any of the risk screening tools,3 supporting the concept that sexual networks confer risk more than does individual behavior. This aligns with a meta-analysis that found risk behavior of Black MSM to be lower than risk behavior of White MSM,5 in stark contrast to HIV incidence, which is substantially higher among Black MSM than White MSM.
When this previous evidence is considered, PrEP screening tools predominantly composed of risk behavior data seem destined to perform poorly for Black MSM. The work of Pyra et al. provides clinic-level data supporting this conclusion. The implications are potentially enormous: clinicians strictly following CDC guidelines will exclude or deemphasize PrEP for a number of Black and Latinx persons who are at high risk of acquiring HIV, unintentionally increasing racial/ethnic disparities in HIV incidence. This also leads to a potential inefficient allocation of resources: organizations funded by CDC to conduct PrEP outreach to highly affected populations such as Black MSM may then proceed to not recommend PrEP to members of populations that are excellent candidates but lack a guidelines-based indication.
Some have suggested using abbreviated PrEP behavioral guidelines to facilitate clinical use, demonstrating that such criteria performed well for members of the iPrEx study (the Preexposure Prophylaxis Initiative trial).6 Yet, as the authors of that study acknowledge, such indications would likely perform poorly for Black MSM. Alternatives include individually tailored indications based on machine-learning models.7 This approach has the advantage of a high ability to predict incident infection that could alleviate disparities in indication but also the disadvantage of feasibility of scale-up and complexity of patient communication. Another possible approach would be to consider prescribing PrEP for all members of high-prevalence or high-incidence groups. Additional thought would be needed on potential cut points and how to approach other groups with greater heterogeneity of outcomes.
Pyra et al. sought to incorporate some of the additional detail in the CDC PrEP indication guidance that notes clinicians should consider the local epidemiological context. They used a criterion that prescribed PrEP for all persons in zip code areas with 2% or more HIV prevalence. This did not resolve observed PrEP ratio disparities. It is important to note that the authors did not use such prevalence thresholds in at-risk populations. For instance, a 2% or even 10% HIV prevalence threshold assessed for MSM or transwomen populations would result in universal indication for all members of each group, resolving PrEP ratio disparities.
The article by Pyra et al. has several limitations. It is from a single health center that serves sexual and gender minority populations, making generalizability challenging. Yet the poor performance of PrEP indication criteria is consistent with both cohort data and our expectations for indication criteria that consist mainly of risk behavior data. As the authors note, PrEP indications in the data set may be underreported because of data limitations of electronic medical records. Last, as the authors acknowledge, the study used an ecological design with cross-sectional data, a design that does not allow direct exploration of whether lower PrEP indication caused lower PrEP prescribing. Self-referral into PrEP has been found to be common in other studies, and some clinicians may not use PrEP indication criteria. Nonetheless, given that many clinicians report being aware of and using CDC PrEP guidance, it seems likely that it has some impact on shaping PrEP discussions between providers and their patients.
CONCLUSIONS
Pyra et al. note that their work extends arguments regarding the insufficiency of current PrEP indication guidance and emphasize that improving access should not contribute to any further stigmatization. It is noteworthy that the CDC conducted some of the initial PrEP trials and provided some of the earliest guidance regarding PrEP. The US Food and Drug Administration was one of the first bodies to approve a medication for PrEP. This type of leadership is encouraging and merits confidence that current challenges with PrEP indication guidance are likely to be addressed in short order. Recently, the CDC predicted that in the absence of intervention 1 in 2 Black MSM, 1 in 5 Latinx MSM, and 1 in 11 White MSM will acquire HIV in their lifetimes. Given the high safety and efficacy of PrEP, it is hard to justify communication with any member of these groups that would not support PrEP use.
----------------
![table4](../images/020720/table4.gif)
Correlations of HIV Preexposure Prophylaxis Indications and Uptake, Chicago, Illinois, 2015-2018
Objectives. To examine gaps in identification of preexposure prophylaxis (PrEP) candidates, uptake, and use of PrEP by populations most likely to seroconvert.
Methods. At a federally qualified health center in Chicago, Illinois, we used electronic medical records, prescription data, and our best approximation of Centers for Disease Control and Prevention PrEP guidelines to determine how many patients were indicated for PrEP relative to HIV diagnoses (indication:HIV), how many were on PrEP relative to indications (PrEP:indication), and how many were on PrEP relative to HIV diagnoses (PrEP:HIV). We compared these ratios across age, gender and orientation, race/ethnicity, and insurance.
Results. Overall, there were 32 indications per incident diagnosis and 16 patients on PrEP per incident diagnosis. In adjusted models, Whites had higher indication:HIV and PrEP:HIV ratios compared with Blacks, men who have sex with men had higher indication:HIV and PrEP:HIV ratios compared with transwomen but lower PrEP:indication ratios, and uninsured patients had higher indication:HIV but lower PrEP:indication and PrEP:HIV ratios compared with those with insurance.
Conclusions. PrEP use, relative to HIV diagnoses, differs by important patient characteristics. While improved guidelines will address some of the disparity, better approaches for determining PrEP candidates and more normalized patient-provider communication are needed to ensure better PrEP access to all individuals at high risk for HIV.
DISCUSSION
Among patients at an urban federally qualified health center with a sexual health focus, we found that PrEP:HIV ratios differed by insurance status, race/ethnicity, and gender and orientation. Specifically, patients with private insurance, White patients, and MSM were more likely to be on PrEP, relative to HIV diagnoses. By looking at indication:HIV and PrEP:indication ratios separately, we were able to better understand the source of these disparities. For instance, differences in PrEP:HIV ratios by insurance status appeared to be driven by differences in uptake; differences by race/ethnicity and gender and orientation seem to be driven more by identification (Table 3).
Previous work on PrEP:HIV ratios reported a national value of 1.812; our overall PrEP:HIV ratio was much higher, given that the data came from sexual health clinics focused on HIV prevention. There is no ideal target PrEP:HIV ratio. High need ratios may reflect, in some populations, HIV prevention success, such that the denominator of incident infections is smaller12; it may also signify uptake among those with little HIV risk. Regardless, we do not propose a cap or limit to need ratios.14 Patients may seek out PrEP without divulging all of their potential risks, and there are secondary benefits to PrEP use, including normalizing to PrEP (at the community level) and continued contact with the health care system (at the individual level).15-17
This work supports and extends previous research suggesting that the current CDC guidelines regarding PrEP are insufficient.3,9 In particular, our data showed that current guidelines under-identify transwomen and poorly identify ciswomen who could benefit from PrEP; there were also large racial/ethnic gaps in indication:HIV ratios. Identification relies to a great degree on provider-patient communication around intimate behaviors, some of which can be stigmatizing, as well as overcoming issues of medical mistrust.18-21 It has been suggested that normalizing sexual health and PrEP education, as a part of all clinical encounters, may help improve identification.5,15 Of note, while we found that need ratios were much higher among primary care patients, the gaps between White versus Black and MSM versus transgender patients remained, suggesting care with an established provider may not necessarily improve communication; additional research concerning communication, particularly with vulnerable and stigmatized populations, is needed.
Others have suggested that PrEP indicators with only individual behaviors, rather than markers of social and network risk, will increase racial/ethnic differences in identification and, subsequently, use.3,5,22,23 We attempted to account for this, by including zip code-level HIV prevalence as an indicator. However, race/ethnicity and gender and orientation disparities remained. This suggests that home neighborhood HIV prevalence may not capture the relevant data about HIV risk.24,25
Elegant arguments have been made against using risk assessments to determine PrEP candidacy.26 Even after we controlled for the presence of a documented indication, disparities in need by race/ethnicity, gender and orientation, and insurance remained; it appears that PrEP access is limited for patients with fewer resources and more stigma. We therefore need an approach that not only identifies those most likely to benefit from PrEP but also works against inherent biases in the system.27 And, importantly, we need to improve access in a way that does not further stigmatize those who need it most.11,15
Finally, we should not ignore disparities in uptake. Insurance status was clearly associated with PrEP:indication ratios and many patients face financial barriers not only for PrEP prescriptions but also for the associated testing and follow up.28-31
There are limitations to this study. PrEP indications were underreported, as clinicians may have knowledge about their patients that was not documented or was documented in ways that were not extractable from the EMR, as we assessed in the chart review. In addition, patients may have received STI diagnoses at other clinics. Therefore, we were not able to perfectly replicate the CDC guidelines. However, the PrEP:HIV ratios are independent of documented indications. We did not exclude patients who may be medically ineligible for PrEP, though we expect this number to be small. Different populations access care at our clinics for different reasons; for instance, many transgender patients come for hormone therapy, while many cisgender patients come for STI or HIV testing. We included chlamydia as an indicator for ciswomen, though it is not included in the CDC guidelines on PrEP, which may explain why the identification ratios for ciswomen were so high, although results were only slightly lower in the sensitivity analysis that removed chlamydia. Finally, indication:HIV and PrEP:indication ratios were based on a simplified framework that patients must first be identified (or self-identify) for PrEP and then decide whether they can or want to start PrEP; while the process in reality is more nuanced, we believe it is a useful approach to understanding disparities in PrEP use.
There were also many strengths to this work. We were able to examine PrEP:HIV ratios by insurance status, race/ethnicity, and age; we also had a large enough population to provide important data on MSM and transgender patients. Furthermore, by having access to EMR data, we were able to disaggregate differences in need ratios attributable to identification versus uptake and thus deepen our understanding of PrEP uptake.
In conclusion, we found important differences in PrEP:HIV by race/ethnicity, gender and orientation, and insurance, driven by both identification of PrEP candidates according to CDC guidelines and uptake. Indication:HIV ratios differed by race/ethnicity, gender and orientation, and insurance, suggesting that current PrEP guidelines do not ameliorate these disparities. Of particular note, specific indications are lacking for transgender individuals. Furthermore, we suggest that better patient-provider communication about sexual health in a nonstigmatizing context can improve PrEP access for those most in need, while structural changes to remove financial and other barriers are also needed.
HIV preexposure prophylaxis (PrEP) as a daily pill was approved for use by the US Food and Drug Administration in 2012.1 The US Centers for Disease Control and Prevention (CDC) created guidelines for providers in 2013, with the latest update in 2017.2 These guidelines include suggested questions regarding potential sources of HIV risk and recommend taking routine sexual histories for all patients, and also include a table of summary guidances. However, there are concerns that these guidelines are not well aligned with the actual probability of acquiring HIV3,4 and may be restricting PrEP use for some key populations.5,6 Of note, although the report reviews the evidence regarding transwomen and PrEP and concludes that all people with a sexual risk for HIV should have access to PrEP, none of the indication categories specifically include transgender individuals.
PrEP access has increased rapidly but unequally in the United States7; according to CDC data through 2016, there were more than 20 000 White male PrEP users, but fewer than 5000 Black male PrEP users and slightly more than 1000 female PrEP users nationally, with no estimates for transgender individuals.7
There is evidence of racial/ethnic disparities,8,9 and uptake seems to be highest among men who have sex with men (MSM), with some groups feeling that PrEP is not for them.10,11 PrEP:HIV ratios (also known as PrEP-to-need ratios), while having no absolute meaning, have been used to understand trends in PrEP use; by presenting the number of PrEP users compared with the number of HIV incident infections, these ratios help identify areas that have relatively lower PrEP use.12
We aimed to apply PrEP:HIV ratios to various demographic patient characteristics to better understand gaps in PrEP use for all patients, with a particular focus on cis-MSM and transgender women, 2 key populations at increased risk for HIV. We also extended the PrEP:HIV ratio by examining 2 component ratios linked to the CDC guidelines-identification of PrEP candidates (indication:HIV) and uptake of PrEP relative to those indicated (PrEP:indication).
RESULTS
Data were available for 45 906 unique patients, three quarters of whom had local zip codes with HIV prevalence data (Table 1). Patients were young, with more than half aged 18 to 34 years; 49% were non-Hispanic White, 21% were non-Hispanic Black, and 18% were Latino/a. A third used private insurance, 16% used public aid, and 51% were self-pay or uninsured. Most patients identified as cismen (61%), with 7% transwomen. Among the subset with a local zip code, the median HIV prevalence was 1.5%. Over the study period, 456 patients acquired HIV, with 389 among MSM and 43 among transwomen.
PrEP:HIV Ratios
The overall PrEP:HIV ratio was 15.8-there were 15.8 patients on PrEP for each incident diagnosis over the 4-year study period (Table 2). The observed PrEP:HIV ratio was highest for Whites (25.9) and lowest for Blacks (7.9); by insurance status, it was highest for those with private insurance (27.1) and lowest for those with Medicaid (9.3). Among key populations, White MSM had the highest PrEP:HIV ratio (26.3), and Black transwomen had the lowest (4.6). After adjustment, the PrEP:HIV ratio among Blacks was one third that of White patients (adjusted odds ratio [AOR] = 0.31; 95% confidence interval [CI] = 0.24, 0.41; Table 3); among Latino/a patients, it was roughly two thirds (AOR = 0.63; 95% CI = 0.48, 0.83). Compared with those with private insurance, the PrEP:HIV ratio for patients with any other insurance was between 22% and 39%. The PrEP:HIV ratio among transwomen was 54% that of MSM (AOR = 0.54; 95% CI = 0.38, 0.79), after we adjusted for other patient characteristics. There was no significant association between age and PrEP:HIV ratios.
Indication:HIV and PrEP:Indication Ratios
To better understand the disparities in the PrEP:HIV ratios, we examined both the indication:HIV and PrEP:indication ratios (Table 2). Overall, the indication:HIV ratio was 32.4, meaning there were 32 patients with a documented PrEP indication for each incident diagnosis, and the PrEP:indication ratio was 0.49, meaning that for every 100 patients with an indication, 49 patients were on PrEP. White patients had both the highest indication:HIV and PrEP:indication ratios, while Blacks had the lowest for both. In the adjusted models, the indication:HIV ratio of Black patients was only one third that of White patients (AOR = 0.31; 95% CI = 0.24, 0.40; Table 3); Latino/a patients also had a significantly lower indication:HIV ratio (AOR = 0.45; 95% CI = 0.35, 0.58). By insurance status, uninsured patients had the highest indication:HIV ratio (57.6) but the lowest PrEP:indication ratio (0.18). In the adjusted models, uninsured patients had an indication:HIV ratio that was more than twice that of privately insured patients (AOR = 2.53; 95% CI = 2.00, 3.21), but PrEP:indication was only 13% of private insurance uptake; patients with Medicaid and Medicare, however, had significantly lower indication:HIV ratios compared with those with private insurance (Table 3). Finally, compared with MSM, transwomen had a very low indication:HIV ratio (7.2 vs 23.3) but a higher PrEP:indication ratio (1.38 vs 0.69). Indeed, the PrEP:indication ratio for transwomen was twice that of MSM, but in adjusted models, the indication:HIV ratio was only 50% that of MSM (Table 4).
Men Who Have Sex With Men and Transwomen
We examined PrEP:HIV and indication:HIV ratios among MSM and transwomen separately (Table 4). Age was not a significant predictor in any model, though direction of the odds ratios differed between MSM and transwomen. Black race was associated with lower ratios in all models, with the most extreme difference among transwomen for PrEP:HIV; Latino/a MSM, but not transwomen, had significantly lower PrEP:HIV and indication:HIV ratios. Finally, the decreased PrEP:HIV ratio for all other patients compared with those privately insured remained, though the size of the association was larger for transwomen. All patterns were the same when we compared indication:HIV ratios between MSM and transwomen; however, as seen in the larger analysis, uninsured MSM actually had a higher indication:HIV ratio compared with MSM with private insurance.
|
|
|
|
|
|
|