PrEP Implementation Behaviors of Community-Based HIV Testing Staff

A Mixed-Methods Approach Using Latent Class Analysis

DeAnne Turner, PhD, MPH; Elizabeth Lockhart, PhD, MPH; Wei Wang, PhD; Robert Shore, PharmD; Ellen M. Daley, PhD, MPH; Stephanie L. Marhefka, PhD


J Acquir Immune Defic Syndr. 2020;83(5):467-474. 

In This Article


Participant Characteristics

A total of 150 staff providing HIV testing (participants) from 48 organizations were included in analysis. Participants were an average of 41.4 years old, ranging in age from 20 to 73 (Table 1). The sample was diverse in terms of both gender (53.7% male; 43.6% female) and sexual orientation (48.3% heterosexual; 40.9% homosexual). Just over half of the participants indicated their race as white, while just over 20% indicated being black. Most participants worked fulltime at their organization (78.1%); however, some worked part time (10.1%) or volunteered (11.8%). Notably, nearly 20% (19.3%) of participants were living with HIV.

Qualitative interview participants (N = 22) were similar to the larger sample, with slight variations. These participants were slightly younger, and more likely to be white and/or male than the full sample. Qualitative participants were an average of 37.4 years (range: 21–64), 63.6% white, 63.6% male, 54.5% heterosexual, and 18.2% were living with HIV.

Classes of PrEP Implementation

A total of 4 LCAs were conducted—each containing 1, 2, 3, or 4 classes, respectively (Table 2). The 1-class and 4-class models each had poor model fit; the 2-class and 3-class models were further investigated. Although the 2-class model had the best model fit as determined by the BIC criteria, the 3-class model had the better fit when considering LMR and theoretical interpretation. Based on consideration of fit statistics and theoretical relevance, a 3-class LCA was selected and used for all further analyses.

As a model approaches one, it approaches full delineation of latent classes, referred to as entropy;[45] entropy for the LCA was 0.914. After reviewing the classes, or grouping of participants, each group was assigned a label to be used throughout analysis and data reporting. The name of each label was determined based on the characteristics of their latent class group. The PrEP implementation behaviors of the latent groups were often reiterated in the qualitative data.

Class 1: Universal PrEP Implementation

Class 1 (42%; n = 62) included staff performing HIV testing who were PrEP advocates; these participants were highly likely to talk about PrEP with their clients, regardless of client eligibility. This group was labeled Universal because participants indicated talking about PrEP with all or most of the clients they see, as well as handing out PrEP-related materials (such as pamphlets, flyers, and written contact information for PrEP-friendly providers) to clients regularly. In qualitative interviews, some members of the Universal group stated that they believed everyone should be aware of PrEP, regardless of their current level of risk. In addition, participants in the Universal group often confirmed discussing PrEP with clients on a frequent basis during the qualitative interviews, such as "I'm obligated as somebody who's doing HIV prevention and testing to talk to somebody about this (PrEP). (Participant 113)" or "I think it's great because it's a resource that everyone should know about. Regardless if you don't use it (PrEP) or use it, you know, if you know someone that can use it, you can always recommend them to it (Participant 116)." This theme was less present in the Eligibility Dependent group and not present in the Limited group.

Class 2: Eligibility Dependent PrEP Implementation

Class 2 (33%; n = 48) included staff who often talked to clients about PrEP, but were most likely to discuss the prevention method if they felt their client was eligible, thus referred to as Eligibility Dependent. This group also provided physical materials to clients, but not as consistently as participants in the Universal group. Eligibility Dependent participants often highlighted the need for a more selective discussion of PrEP based on perceived side effects or risk criteria, such as "Like I said, if you're married and you have one partner, then you're taking PrEP to damage your liver or your kidneys, and so on and so on. It doesn't make sense (to use PrEP). (Participant 103)" or "We have many people that we know are in serodiscordant relationships, so we want to help them (Participant 107)." These participants weighed the pros and cons of PrEP for their client and determined whether PrEP should be discussed based on their client's level of risk.

Class 3: Limited PrEP Implementation

Class 3 (25%; n = 37) was the most discriminate group with regards to PrEP implementation—these participants sometimes spoke to clients about PrEP, but not as systematically as those in the Universal or Eligibility Dependent groups, thus referred to as Limited. This group was also the least likely group to provide physical information about PrEP, such as brochures or referral cards. In LCA, the Limited group was found to have inconsistent discussions about PrEP with clients. This lack of systematic rationale for PrEP implementation was captured by the varied qualitative responses within this group. This group discussed concerns regarding PrEP but did not have salient qualitative rationale for why they did, or did not, discuss PrEP with clients.

A full description of categorical response distribution can be found in Table 3.

Variation in Group Membership Based on Participant Characteristics

In bivariate analyses (Table 4), significant differences existed between the Eligibility Dependent participants and the Limited participants in age, race, sexual orientation, and ever taken PrEP status. Significant differences were found between Universal and Limited participants in sexual orientation. No statistically significant differences were found between the Eligibility Dependent participants and the Universal participants in bivariate analyses.

In multivariate analyses, only race and sexual orientation remained significant predictors of the PrEP implementation group (Table 4). Participants who identified as a racial minority were less likely to be in the Limited group {compared with both the Universal [3.78 (1.48–9.68)] or Eligibility Dependent [6.79 (2.31–19.98)] groups} compared with their white counterparts; however, race did not differentiate the odds of group affiliation when comparing the Eligibility Dependent vs. Universal groups. Sexual orientation played a role in differentiating between the Limited vs. Eligibility Dependent and Universal groups—participants who identified as a sexual minority were less likely to be in the Limited group compared to the Eligibility Dependent [8.41 (2.26–31.30)] or Universal [4.85 (1.47–15.97)] groups; however, sexual orientation did not differentiate between Eligibility Dependent vs. Universal groups. Age, gender, ever having taken PrEP, and HIV status did not impact the odds of being in a specific PrEP implementation group.