Examining the Factors Affecting PrEP Implementation Within Community-Based HIV Testing Sites in Florida: A Mixed Methods Study Applying the Consolidated Framework for Implementation Research

Abstract

HIV testing/counseling is a critical point during which non-clinical staff could intervene, discuss and/or refer clients for pre-exposure prophylaxis (PrEP). This analysis investigated the contextual factors affecting PrEP implementation within HIV testing sites. Two generalized linear mixed models were conducted to estimate PrEP implementation as a function of constructs from the Consolidated Framework for Implementation Research (CFIR). Qualitative interviews were analyzed thematically. Data integration occurred via joint analysis and triangulation. Constructs from the CFIR domain Characteristics of Individuals did not predict PrEP implementation when controlling for demographic characteristics; qualitative data signaled divergent findings in PrEP knowledge. Within the CFIR domains Inner and Outer Settings, relevant priority and available resources predicted PrEP implementation; qualitative data confirmed the importance of available resources and provided insight into the impact of cosmopolitanism and leadership. Addressing the contextual factors that affect PrEP implementation may help HIV testing staff to better implement PrEP programs.

Introduction

Pre-exposure prophylaxis (PrEP) significantly reduces HIV acquisition when taken as prescribed [1-3]. Despite approval by the Food and Drug Administration and being recommended for populations at higher than average risk for HIV exposure [4-9], the uptake of PrEP has been slower than needed to End the HIV Epidemic [10]. Medical providers have an important role in PrEP programs [11, 12], but there are limitations to relying on medical providers as the only staff responsible for PrEP implementation; among these limitations are the discomfort or insufficient knowledge [13, 14] some providers have regarding PrEP, and the struggle some providers have discussing sexual health with their patients [15]. Additional sites—such as STI clinics, pharmacies, and community-based organizations—could be utilized to disseminate PrEP-related education and support [16-20].

HIV testing sites offer alternative locations in which PrEP implementation (i.e. PrEP education and/or referral to a place where a PrEP prescription is available) could occur [16]. Many HIV testing sites can be accessed regardless of insurance coverage and operate on a walk-in basis. Additionally, staff providing HIV testing already speak with clients about their sexual risk behaviors and HIV prevention [21]. Thus, HIV testing sites may be an important place to reach clients who may meet the indications for PrEP.

Research regarding HIV testing and PrEP has focused on the client—the person receiving the HIV test and related counseling [22-25]. Studies have examined the frequency in which clients already on PrEP have continued to get tested for HIV, and if CDC testing guidelines for PrEP are met [22, 23]. Researchers have recruited clients from within HIV testing sites and investigated the interest these clients have in PrEP [25]. One study assessed the role of HIV testing sites in linkage to PrEP, but focused only on the perspective of the clients [26]. While the possible utility of PrEP referrals during HIV testing has been noted [27], PrEP implementation studies focusing on the role of staff who provide HIV testing are limited [28].

One approach to understanding program implementation is use of implementation science theory. Damschroder and colleagues [29] synthesized implementation theories and frameworks and identified five overarching domains affecting implementation: Intervention Characteristics, Outer Setting, Inner Setting, Characteristics of Individuals, and Process. Collectively, these domains and their related constructs make up the Consolidated Framework for Implementation Research (CFIR) [29]. The CFIR helps researchers and practitioners understand implementation processes and where key changes could be made to improve PrEP implementation. In this context, the CFIR domain Characteristics of Individuals can explain how characteristics of HIV testing staff may affect their PrEP implementation behaviors, influencing training development and hiring practices. Other domains, such as the Inner Setting and Outer Setting, provide information on the contextual factors affecting PrEP implementation within HIV testing sites.

Using the CFIR, this study examines how the intrapersonal characteristics of HIV testing staff (Characteristics of Individuals), organizational (Inner Setting) and community (Outer Setting) characteristics impact PrEP implementation by staff performing HIV testing. Two research questions are addressed:

1. What characteristics of the HIV testing staff are associated with PrEP implementation during HIV testing (i.e. CFIR Domain Characteristics of the Individual)?

2. What Inner and Outer setting factors are associated with PrEP implementation during HIV testing?

Methods

Study Design

This study utilized a mixed methods concurrent triangulation design [30] to examine the factors affecting PrEP implementation during HIV testing in Florida. Participants included staff who conduct HIV testing. Eligibility criteria included: (1) working or volunteering at an organization that provides community-based publicly funded HIV testing in Florida; (2) having provided HIV testing/counseling within the past 3 months and (3) being at least 18 years old. Participants were ineligible if they were unable to complete a computer-based survey without assistance. The guiding framework for the study was the CFIR.

Human Protections

This study was approved by the Institutional Review Board of the University of South Florida. All participants were provided with an electronic informed consent prior to initiating the survey.

Data Collection

Quantitative data were collected from February to May, 2018 using an online survey (approximately 15–20 min in length) via Qualtrics [31]. The primary investigator of the study contacted each publicly funded testing site via email with a request to share the survey with staff who perform HIV testing and counseling. Organizations were contacted up to four times [32].

After completing the online survey, participants had the option to provide their name and contact information; via a raffle, three of the 150 participants received a $50 gift card. Participants were also asked if they were willing to be contacted to complete a semi-structured interview. These participants were purposefully sampled to ensure data from participants with a wide range of experiences in PrEP implementation. Participants who took part in the qualitative interview received a $20 gift card.

Measures

Quantitative Measures

Quantitative study measures included constructs within the CFIR domains Characteristics of Individuals, Inner Setting, and Outer Setting, as well as PrEP implementation behaviors occurring within HIV testing sites. The study outcome was PrEP implementation. In this setting, PrEP implementation is conceptualized as the inclusion of PrEP in HIV counseling and/or PrEP referrals. Additionally, PrEP was referred to as a once a day pill—Truvada, as other forms of PrEP were not yet approved.

The study outcome “PrEP implementation” was determined based on prior research [28]. The three groups of PrEP implementation were defined as: (1) UNIVERSAL (42% of the sample; highly likely to discuss PrEP with their clients, regardless of the client’s risk of HIV exposure); (2) ELIGIBILITY DEPENDENT (33% of the sample; likely to discuss PrEP if they perceived their client to be at high risk for HIV); and 3) LIMITED (25% of the sample; sometimes, but not systematically, discuss PrEP with clients).

Model 1

Two constructs within the CFIR domain Characteristics of Individuals were included in analysis: (1) knowledge and beliefs about the intervention (PrEP); and (2) self-efficacy. Knowledge and beliefs about the intervention was measured using two composite variables. Knowledge was measured using six items adapted from previous PrEP knowledge scales (α = 0.66) [33, 34]. Beliefs about the intervention were measured using a 10-item scale on personal beliefs and attitudes related to PrEP (α = 0.78) [34]. Self-efficacy was measured as a composite variable derived from a six item scale [34] that measured how comfortable participants were with a variety of behaviors related to PrEP implementation within HIV testing sites (e.g., discussing PrEP with a client; α = 0.88).

Model 2

Within the CFIR domain Outer Setting, three constructs were included in the analysis: patient needs and resources, cosmopolitanism, and external policies and incentives. Patient needs and resources [the degree to which clients would benefit from increased opportunities for HIV prevention, including PrEP; how the needs of clients are prioritized by the organization in which the participant (i.e., staff performing HIV testing) works] was measured as a composite score of four items adapted from the Organizational Readiness for Change Assessment (ORCA; α = 0.78) [35, 36]. Cosmopolitanism was measured as a composite score of three items, created based on the definition of cosmopolitanism [36] (e.g., knowledge of local clinics or organizations to which they could refer clients if the client was interested in PrEP; α = 0.83). External policies and incentives [external pressures to implement an intervention, e.g., influence of external recommendations and guidelines, such as those from the CDC and WHO] was measured as a composite score of three items from the Organizational Change Manager (OCM; α = 0.77) [37].

Within the CFIR domain Inner Setting, the constructs structural characteristics, culture, tension for change, compatibility, relative priority, learning climate, and available resources were included. Structural characteristics describe the structure of an organization. Three individual items were used to measure structural characteristics in analysis: size of the organization, if the organization has offices in more than one county or state, and if the organization specializes in serving Lesbian, Gay, Bisexual, and/or Transgender (LGBT) populations. Culture [organizational norms and values] was measured by a composite score determined by five items adapted from the ORCA (α = 0.77) [35, 36]. Tension for change [the degree to which participants felt a change was needed to effectively reduce the spread of HIV] was measured using two items adapted from the OCM (α = 0.66) [36, 37]. Compatibility [how well the intervention (here, PrEP implementation) fits within an organization [36]] was measured by two items adapted from the ORCA [35] and OCM [37] (α = 0.66). Relative priority [importance of PrEP implementation in relation to the other tasks] was measured by one item “Talking about PrEP during HIV testing is less important than talking about other HIV prevention methods.” Available resources was measured by one item describing the availability of physical materials related to PrEP that could assist staff and/or be disseminated to clients. Learning climate [if an organizational culture allows for mistakes during the learning process and empowers staff to believe they can make a difference by implementing the intervention [36]] was measured by two items based on the CFIR definition for learning climate (α = 0.77). A full list of items used to measure CFIR constructs can be found in Online Appendix A.

Qualitative Measures

Qualitative interviews were derived from the CFIR Interview Guide Tool [36] and used to provide context regarding the constructs most important to PrEP implementation within HIV testing sites. Based on a literature review and expert feedback, the research team entered CFIR domains and constructs into the Interview Guide Tool, which provided a draft semi-structured interview guide. Questions were adapted so that the “intervention” being referenced was listed as PrEP/PrEP Implementation (e.g. to what extent do you network with people who provide HIV testing and counseling outside of your organization [cosmopolitanism]; what kind of information exchange do you have with others outside your organization, either related to PrEP, or more generally about HIV testing? [cosmopolitanism]). Interviews were audio recorded and transcribed.

Sample Size

A total of 150 participants from 48 unique organizations were used in analysis. For the qualitative sample, saturation was met after 22 interviews and determined when no new themes emerged from the data.

Data Analysis

Quantitative data were exported from Qualtrics [31] into SPSS v.24 [38]. Descriptive statistics and bivariate analyses were conducted. To account for clustering (i.e. the groups of participants working within the same organization), generalized linear mixed models [39] with multinomial distribution, logit link, and robust variance estimator were used to estimate PrEP implementation as a function of key CFIR variables from the domains Characteristics of Individuals, Inner Setting, and Outer Setting.

To address the first research question (What characteristics of the HIV testing staff are associated with PrEP implementation during HIV testing [i.e. CFIR Domain Characteristics of the Individual]?), PrEP implementation was estimated as a function of PrEP knowledge, PrEP beliefs, and PrEP-related self-efficacy—variables from the CFIR domain Characteristics of Individuals. The following participant characteristics were controlled: gender, age, race, HIV status, and prior personal experience with PrEP.

To address the second research question (What Inner and Outer setting factors are associated with PrEP implementation during HIV testing?), PrEP implementation was estimated as a function of CFIR constructs from the domains Inner Setting and Outer Settings. Outer Setting predictors included: patient needs and resources, cosmopolitanism, and external policies and incentives. Inner Setting predictors included: structural characteristics (offices in multiple counties or states; organization size; specializes in serving LGBT populations), culture, tension for change, compatibility, relative priority, learning climate, and available resources.

Qualitative interview transcripts were verified by the primary author to ensure correct transcription, imported into MaxQDA [40], and analyzed thematically [41]. Using the codebook provided within the CFIR technical assistance website as a starting point, an initial codebook was created based on CFIR constructs [36]. Emerging codes and adaptations were integrated into the codebook by two researchers based on an initial review of transcripts. Two researchers trained in qualitative data analysis coded the same transcript independently before discussing revisions and edits for the codebook. An additional transcript was coded by both researchers to refine the codebook. Following agreement on the codebook, four transcripts were coded to calculate inter-rater reliability (IRR). After discussion of interpretation and clarifications of codes, IRR was attempted with four new transcripts and reached with K = 0.86. The primary author coded the remaining transcripts (n = 12).

Data Integration and Interpretation

Data integration was included in the study design, methods, and interpretation. During data collection and analysis, the quantitative data were used to recruit qualitative participants via connecting [30, 42]. Moreover, the qualitative and quantitative data were used iteratively to build upon each other for data interpretation [43], and the quantitative and qualitative data sets were merged [42, 43]. A joint display of data was used with all analyses performed in MaxQDA [43].

Results

Participant and Organization Characteristics

A total of 150 participants from 48 organizations (M = 3.1 participants per organization [1–17]; SD = 3.5); took part in the survey. Participants ranged in age from 20 to 73 years, with an average age of 41.4 (± 14.1) years. About half of the participants identified as male (53.7%), heterosexual (48.3%), and/or white (51.3%). Participants also described the agencies in which they worked. 28.7% of participants described their organization as small (under 20 staff members), 30% as moderately sized (21–50 staff members), and 41% as large (more than 50 staff members). Half of participants indicated that their organization had offices in multiple counties or states (50.3%). Many (73.3%) stated that their organization specialized in serving LGBT clients. Participants served clients from a range of geographic areas: 4.7% exclusively rural, 37.3% exclusively urban, and 58% a mix of both rural and urban. Participant and organization characteristics can be found in Table 1.

Table 1.

Participant characteristics (N = 150)

	n (%)
Age (range)	41.4 ± 14.1 (20–73)
Racec
White	77 (51.3)
Black/African American	35 (23.3)
Asian	6 (4.0)
American Indian/Alaskan Native	4 (2.7)
Native Hawaiian/Pacific Islander	1 (0.7)
Hispanic ethnicity	50 (33.3)
Gendera
Male	80 (53.7)
Female	65 (43.6)
Transgender female to male	2 (1.3)
Another gender	1 (0.7)
Sexual orientationb
Heterosexual (Straight)	72 (48.3)
Homosexual (Gay/Lesbian)	61 (40.9)
Bisexual	7 (4.7)
Another sexual orientation	5 (3.4)
Ever taken PrEPe	20 (13.4)
Currently on PrEPf	14 (9.6)
HIV statuse
Living with HIV (HIV positive)	29 (19.5)
Not living with HIV (HIV negative)	120 (80.3)
Organization size
Small (under 20 staff)	43 (28.7)
Medium (21–50 staff)	45 (30.0)
Large (more than 50 staff)	62 (41.3)
Organization reach (multiple counties/states; yes)	75 (50)
Residence of clients served
Rural clients	7 (4.7)
Urban clients	56 (37.3)
Mix of rural and urban	87 (58.0)

^aNo participant identified as transgender male to female. One participant selected “prefer not to answer”. One participant skipped the question. N = 149

^bTwo participants indicated “not sure” and two indicated “prefer not to answer”. One participant skipped the question. N = 149

^cParticipants could select all races/ethnicities that applied. Some participants did not select a race, and only selected Hispanic ethnicity

^dOne participant skipped the question N = 149

^eOne participant selected “prefer not to answer” N = 149

^fOne participant selected “prefer not to answer”. Three participants skipped the question N = 146

Characteristics of HIV Testing Staff Affecting PrEP Implementation

In bivariate analyses (Table 2), significant differences were found between UNIVERSAL and LIMITED participants in knowledge, beliefs, and self-efficacy. No statistically significant differences were found between the ELIGIBILITY DEPENDENT and UNIVERSAL participants, nor ELIGIBILITY DEPENDENT and LIMITED participants in bivariate analyses. In multivariate analyses (Table 2), neither knowledge, beliefs, nor self-efficacy were significant when controlling for participant characteristics (e.g. race, sexual orientation). Integration of major qualitative and quantitative findings can be found in Table 4.

Table 2.

Model 1: PrEP Implementation as a function of HIV testing staff implementing the intervention [CFIR Domain: Characteristics of Individuals]

	Universal (Ref: Limited)		Eligibility dependent (Ref: Limited)		Universal (Ref: Eligibility Dependent)
	Unadjusted OR [95% CI]	Adjusted OR [95% CI]	Unadjusted OR [95% CI]	Adjusted OR [95% CI]	Unadjusted OR [95% CI]	Adjusted OR [95% CI]
Self-efficacy	2.67* (1.42–5.03)	2.06 (0.95–4.73)	1.48 (0.82–2.69)	1.45 (0.65–3.20)	1.81 (0.96–3.39)	1.42 (0.68–2.99)
Beliefs	2.47* (1.17–5.19)	2.08 (0.67–6.50)	1.27 (0.57–2.83)	1.20 (0.36–4.05)	1.93 (0.99–3.77)	1.73 (0.76–3.94)
Knowledge	1.43* (1.10–1.86)	1.36 (0.96–1.94)	1.13 (0.85–1.50)	1.12 (0.74–1.70)	1.26 (0.98–1.62)	1.21 (0.87–1.68)

Probability distribution: Multinomial; Link Function: Generalized Logit

Model controlled for: Gender; age; race; sexual orientation, HIV status; and prior personal experience with PrEP

^*Indicates significance at p < 0.05

Table 4.

Summary of major findings

Quantitative	Integrated findings	Qualitative
	Available resources
Bivariate: More resources in UNIVERSAL (vs. LIMITED & ELIGIBILITY DEPENDENT) Multivariate: More resources in UNIVERSAL (vs. ELIGIBILITY DEPENDENT)	Having available resources (e.g. materials, time, or other needed items to implement PrEP) increases the likelihood that HIV testing staff will speak to clients about PrEP. Use of the DH1628 created an opportunity to transition to a conversation about PrEP	PrEP-related resources emerging in the qualitative data included: referral cards, physical information about PrEP, and use of the DH1628 (required form completed with each HIV test)—all acted as facilitators for PrEP implementation. The client lacking time was a deterrent to staff initiating conversations about PrEP
	Knowledge and beliefs
Bivariate: Greater knowledge in UNIVERSAL (vs. LIMITED); Greater pro-PrEP beliefs in UNIVERSAL (vs. LIMITED) Multivariate: NS	A more nuanced understanding of the knowledge required to effectively talk to clients about PrEP may be needed. Participants indicated their personal beliefs about PrEP when discussing if, and how, they spoke to clients about PrEP	Most participants stated factual information about PrEP; some participants had incorrect knowledge. PrEP-related beliefs were salient across implementation groups; key themes included beliefs that: side effects of PrEP were too numerous/severe, additional effectiveness research is needed, and clients may misuse PrEP
	Patient needs & resources
Bivariate: NS Multivariate: NS	HIV testing staff are aware of the needs of their clients, but the ways in which this impacts their behaviors differs	The needs of clients were salient across implementation groups; key themes of client needs/resources included: perceived costs for visits or medication, client knowledge, side effect concerns, provider stigma/lack of PrEP knowledge
	Cosmopolitanism
Bivariate: NS Multivariate: NS	Cosmopolitanism was not significant in quantitative analysis. The conceptualization of cosmopolitanism was rooted in the types and strength of ties, as well as trust between organizations; these may be a critical next step to understanding referral patterns	Participants discussed the need for/ways in which they networked with other organizations to facilitate PrEP referrals including vetting organizations before referring clients, as well as having existing relationships and memorandums of understanding with referral partners
	Relative priority
Bivariate: Greater relative priority in UNIVERSAL (vs. LIMITED) Multivariate: Greater relative priority in UNIVERSAL (vs. LIMITED)	Reiterating organizational priorities for PrEP implementation/referral to all staff providing HIV testing may be important in practice	Relative priority often occurred alongside Leadership Engagement with PrEP implementation. Participants discussed PrEP implementation in relation to other duties and promotion of other methods of prevention (e.g. exclusively promoting condoms)
	Structural characteristics
Bivariate: UNIVERSAL more likely to specialize in serving LGBT populations (vs. LIMITED) Multivariate: NS	Apart from specializing in serving LGBT populations, other structural characteristics (e.g. organization size) were neither significant nor salient; additional structural characteristics should be investigated	Participants discussed their organization having specific services for persons who are LGBT, trainings for LGBT-appropriate services, organization of community events, and advocacy efforts

NS not significant

Self-efficacy, Knowledge, and Beliefs [CFIR Domain: Characteristics of Individuals]

Within the CFIR domain Characteristics of Individuals, the construct most salient in the qualitative data was the knowledge and beliefs HIV testing staff have regarding PrEP and PrEP implementation. Regardless of PrEP implementation group, participants mentioned their personal knowledge and beliefs regarding the side effects of PrEP, such as one participant who stated:

But it [PrEP] also has side effects. It’s a strong medication, antiretroviral for HIV. These people, they don’t have HIV. They’re negative. They’re taking it as a prevention. They have side effects in the liver, in the kidneys. Also, they said on the pamphlet that in younger people, [PrEP can] diminish the density of the bones (Participant 103).

While quantitatively PrEP knowledge was high across all PrEP implementation groups, a few statements in the qualitative findings that reflected some staff performing HIV testing had inadequate, or incorrect, knowledge of PrEP, such as:

Yeah, and oftentimes, it [PrEP] being a shot, and it staying in your system, we’ve received concern from clients that … because they’ll be stuck with those side effects for whatever, a month. (Participant 102).

Cost (e.g. “Obviously it’s [PrEP] 700, 800, even $1000 a month, just for the medication… Then of course any doctor’s bills, any lab work et cetera (Participant 119)”) and concerns about clients not using condoms (e.g. “I see too many young men running around, saying, ‘Oh, I’m taking PrEP now, I don’t have to wear a condom’…what about HPV, syphilis, gonorrhea? (Participant 111)”) were also commonly mentioned among participants. Less prevalent concerns included the perception that limited research is available regarding PrEP and fear of client non-adherence.

Organizational and Extra-Organizational Characteristics Affecting PrEP Implementation [Inner Setting and Outer Setting]

In bivariate analyses (Table 3) several constructs from the CFIR domains Inner Setting and Outer Setting were significant in explaining membership in a particular PrEP implementation group, including: available resources, tension for change, compatibility, learning environment, culture, and external policies. However, in multivariate analyses (Table 3), fewer Inner Setting and Outer Setting constructs predicted group membership. Compared to the LIMITED group, participants in the UNIVERSAL group were more likely to have high scores in relative priority (aOR 1.65 [1.09–2.50]), indicating that participants had the perception that PrEP implementation was a high priority within their organization, even when compared to the many tasks that organizations providing HIV testing perform. Compared to participants in the ELIGIBILITY DEPENDENT group, participants in the UNIVERSAL group were more likely to indicate the availability of resources to assist in PrEP implementation (aOR 1.973 [1.197–3.253]). No constructs were statistically significant when comparing participants in the ELIGIBILITY DEPENDENT group to participants in the LIMITED group.

Table 3.

Model 1: PrEP Implementation as a function of organizational and extra-organizational characteristics [CFIR Domains: Inner Setting and Outer Setting]

	Universal (Ref: Limited)		Eligibility dependent (Ref: Limited)		Universal (Ref: Eligibility Dependent)
	Unadjusted OR [95% CI]	Adjusted OR [95% CI]	Unadjusted OR [95% CI]	Adjusted OR [95% CI]	Unadjusted OR [95% CI]	Adjusted OR [95% CI]
Patient needs	1.64 (0.81–3.32)	0.78 (0.29–2.10)	1.50 (0.92–2.45)	0.92 (0.40–2.15)	1.14 (0.59–2.23)	0.86 (0.49–1.53)
Cosmopolitanism	1.41 (0.88–2.27)	0.63 (0.35–1.16)	1.14 (0.81–1.60)	0.73 (0.40–1.35)	1.26 (0.84–1.87)	0.90 (0.65–1.25)
Resources	2.13* (1.18–3.83)	1.31 (0.75–2.29)	0.98 (0.73–1.32)	0.69 (0.45–1.06)	2.14* (1.27–3.61)	1.97* (1.20–3.25)
Tension for change	1.99* (1.14–3.47)	1.35 (0.77–2.36)	1.72* (1.03–2.86)	1.25 (0.62–2.53)	1.17 (0.78–1.76)	1.11 (0.692–1.781)
Compatibility	4.84* (1.93–12.17)	1.46 (0.46–4.64)	2.91* (1.42–5.97)	2.55 (0.99–6.55)	1.72 (0.71–4.13)	0.57 (0.21–1.58)
External policies	1.76* (1.01–3.08)	1.45 (0.73–2.89)	1.27 (0.84–1.92)	0.98 (0.56–1.72)	1.41 (0.90–2.21)	1.52 (0.97–2.38)
Learning climate	4.89* (1.79–13.40)	2.56 (0.760–8.64)	2.86* (1.40–5.85)	1.45 (0.54–3.92)	1.72 (0.55–5.37)	1.69 (0.37–7.65)
Relative priority	1.73* (1.19–2.50)	1.65* (1.09–2.50)	1.23 (0.87–1.73)	1.177 (0.741–1.871)	1.393 (0.993–1.955)	1.368 (0.942–1.986)
Culture	1.96 (0.99–3.87)	1.88 (0.80–4.41)	1.95* (1.11–3.46)	2.26 (0.98–5.24)	1.00 (0.52–1.95)	0.80 (0.41–1.56)
Offices in multiple counties or states (Yes; Ref = No)	1.15 (0.37–3.60)	1.14 (0.25–5.20)	1.33 (0.54–3.30)	1.20 (0.33–4.29)	0.89 (0.35–2.29)	1.04 (0.35–3.11)
Organization size (Ref: Large)	REF	REF	REF	REF	REF	REF
Small	1.26 (0.37–4.34)	1.56 (0.41–5.93)	0.79 (0.29–2.15)	0.89 (0.24–3.27)	1.65 (0.60–4.53)	1.80 (0.60–5.42)
Medium	3.25 (0.80–13.17)	3.79 (0.81–17.77)	2.69 (0.69–10.52)	3.01 (0.61–14.86)	1.19 (0.45–3.17)	1.10 (0.37–3.30)
Specializes in serving LGBT populations	2.89* (1.02–8.18)	2.33 (0.62–8.78)	1.95 (0.84–4.52)	2.12 (0.60–7.49)	1.54 (.58–4.09)	1.02 (0.29–3.58)

Probability distribution: Multinomial; Link Function: Generalized Logit

^*Indicates significance at p < 0.05

Leadership [CFIR Domain: Inner Setting]

Leadership emerged as an important factor in the initiation of some PrEP programs (i.e. leadership engagement under the CFIR domain Inner Setting). This is reflected below.

When I was at [Former Organization Name], I know we were talking about doing a PrEP program, but there was only 1 or 2 people who really knew a lot about PrEP. … Here at [Current Organization Name] they made a requirement, like, ‘Look, if you’re going to be a part of this organization, if you’re going to be a doctor who supports our programs, this is part of your job.’ … The leadership really stepped up, so I think that really made a difference (Participant 106).

Others discussed leadership being involved in administrative decisions, such as creating memorandums of understanding:

… we renew them every year, but some of them they’ve been there for five years, six years. So it just depends. But the Clinical Director is the one that creates the MOUs [memorandum(s) of understanding; for PrEP referral] with these partnering agencies (Participant 116).

Availability of Resources [CFIR Domain: Inner Setting]

Some participants described PrEP as a natural extension of services from HIV care to HIV prevention. The availability of resources was a facilitator to PrEP implementation in both multivariate analysis (above) and in qualitative analyses. In qualitative findings, available resources consisted of time, written referral policies, physical handouts and brochures related to PrEP, and other physical materials used in HIV prevention (e.g. condoms). The availability of physical materials, especially those containing information about PrEP that could be provided to the client, but more importantly assisted participants in discussing PrEP with their clients. One participant described how available resources assisted them in linking clients to PrEP as follows:

Our testing kits, we have PrEP brochures and referral cards and all that, so they would have all that information. A lot of times we do testing at off-site locations. We actually have a box that has all sort of brochures and referral information…More and more PrEP is part of that linkage to care for HIV negative individuals. (Participant 104).

The importance of these materials was salient, as another participant stated: “Yeah, we have all the material there. Like a little business card specifically for the PrEP clinic with the number that is directly to someone who works there. (Participant 105).” Lastly, participants focused on utilization of the state-required form that must be completed with each HIV test performed, the DH1628, referred to by participants as “the 1628”. This form was a resource that created a natural transition to talking about PrEP. For example:

In the 1628, which is the state form that you fill out when you’re doing an HIV test, there is a section on PrEP and nPEP, so we talk, I make sure that I talk to clients both about PrEP, and PEP—post exposure prophylaxes, and how to get it, if they ever need it. (Participant 104).

Another participant reiterated this view:

The form that we use [DH1628] asks them if they’ve ever taken PrEP or PEP, so that really is … people most of the time are like, what’s that? So it’s really a way for us to talk to them about it… (Participant 107).

In qualitative analyses, participants discussed how the Outer Setting influenced their PrEP implementation behaviors. The most salient themes within the Outer Setting were related to the needs and resources of the clients served by the organization. Although the perspectives of the clients were not directly captured in this study, the perceived needs and resources of these clients did impact the behaviors of the staff who perform HIV testing. Participants across all implementation groups referenced the needs and resources of the clients they served; however, participants in the UNIVERSAL group discussed these barriers to the greatest extent.

Cosmopolitanism [CFIR Domain: Outer Setting]

Cosmopolitanism [36] was expressed in several ways. Some participants noted the formal role of partnerships, extending beyond leadership and into community engagement, such as one participant who described how memorandums of understanding (MOUs) were present within her organization and aided in PrEP implementation.

We go to these huge meetings where we all talk about barriers to health and what lacks in the community and stuff like that, and if a doctor’s like ‘Hey, these are the services I have, you can always send your clients over. We’re doing this for free. Or we do this and we’re low income cost.’ That’s how the MOU [memorandum of understanding] is kind of drawn to place. So, it’s honestly knowing people (Participant 116).

Quantitatively, cosmopolitanism was not statistically associated with PrEP implementation. Similar findings were found in qualitative interviews – where participants discussed some degree of networking regardless of implementation group. Some participants described that relationships were established at planning meetings and often occurred because both organizations had a relationship with the state-run health department. One participant, when asked to describe if he felt other organizations in his community were familiar with PrEP, highlighted the role the health department has in networking between community-based organizations (CBOs):

… from going to consortium meetings when other CBOs are in the room, if they’re absorbing what the health department is sending out, the information that the health department is sending out, at the same rate that my CBO is, then all the CBOs should have at least minimum knowledge to link clients to PrEP services (Participant 119).

While cosmopolitanism is important for organizations implementing any program that may require outside referrals, in this setting it may be most important for organizations without onsite medical providers. One participant who was located at a community-based organization that did not have medical providers on staff stated:

… we don’t have a health provider on site, but we do have a list of providers that we know for sure that provide PrEP in the area. Now the health departments are also rolling it out, so they’re also going to be doing it, and there’s a couple organizations that are the HIV resource centers that are now implementing their own RN’s that can prescribe PrEP. So we go through the list, and tell them a couple more…(Participant 121).

Participants also used external networking (i.e. cosmopolitanism) to understand how other organizations worked—specifically, which organizations may be appropriate, or inappropriate, for partnerships. Such partnerships were particularly important because participants were hesitant to refer their clients to an unknown entity. One participant, stated the following:

… ’Cause I don’t want to send someone to an unknown area and then them just get shut down because I don’t know how everyone else is, like if I send them to someplace that I’ve never heard of or never met, ’cause usually we have a contact at every agency that we call for these reasons. They usually pick up and know too, kind of like a mini counseling session with the client over the phone and kind of be like, ‘Hey, okay you can come in at this time.’ And they set up the initial appointment (Participant 116).

Here, the participant discussed the importance of not only cosmopolitanism, but developing trust within networking relationships. By trusting the services that another organization provided, the participant could refer the client without fear of the client having a negative experience. Integration of major qualitative and quantitative findings can be found in Table 4.

Discussion

Guided by the CFIR [29], this study investigated the role of Characteristics of Individuals (i.e. characteristics of the staff performing HIV testing), Inner Setting (i.e. organizational characteristics) and Outer Setting (i.e. characteristics of the community the organization serves) in PrEP implementation via a mixed methods concurrent triangulation design [30]. In model 1, we sought to understand how characteristics of the HIV testing staff affected their PrEP implementation behaviors; this knowledge can help in preparing training materials or programs for HIV testing staff. After performing a multinomial generalized linear mixed model, and controlling for demographic characteristics, knowledge, beliefs, and self-efficacy were not statistically significant. In model 2, we sought to understand how Inner and Outer Setting characteristics impacted the PrEP implementation behaviors of HIV testing staff; this knowledge is important because to effectively implement PrEP, knowledge is needed on how best to equip community-based HIV testing sites, and the role that the surrounding community has in such implementation [12]. In a multilevel multinomial model, only relative priority and available resources remained significantly associated with PrEP implementation. Triangulation with the qualitative findings highlighted the importance of available resources and provided insight into the role of cosmopolitanism and patient needs and resources in PrEP implementation.

In this study, knowledge, self-efficacy, and beliefs were captured under the domain Characteristics of Individuals. While none of these constructs were statistically significant, statements reflecting PrEP knowledge occurred regardless of implementation group in the qualitative findings. Based on quantitative findings, participants had relatively high rates of PrEP knowledge. PrEP knowledge among HIV testing staff has been understudied; however, research on PrEP knowledge among providers has indicated high rates of PrEP knowledge among HIV care providers compared to their counterparts specializing in general medicine [13, 14]. Perhaps knowledge may not play a significant role in PrEP implementation during the HIV testing process because, similar to HIV specialists, those working in HIV testing are knowledgeable in HIV prevention, regardless of their views on PrEP. Previous research has been inconsistent regarding the association between PrEP knowledge and PrEP prescribing practices; some researchers have found that PrEP knowledge is associated with a greater likelihood of prescribing PrEP [33], and others have noted that PrEP knowledge alone was unlikely to result in an increase in prescribing practices [14]. This same line of reasoning may be true among staff providing HIV testing, as knowledge may not be enough to increase rates of actual referrals. Moreover, given that some participants scored high on the PrEP knowledge scale, but also reflected incorrect knowledge about PrEP in the qualitative findings, future research into a more nuanced scale to measure PrEP-related knowledge among non-clinical staff may be needed.

PrEP beliefs have been found to affect prescribing PrEP among providers and were hypothesized to also play a role in PrEP implementation occurring during the HIV testing process. Among HIV care providers, the most important factor triggering screening for PrEP was the patient being in a serodiscordant relationship [44], especially those who were MSM in serodiscordant partnerships. Eligibility driven specifications were also mentioned among participants of this study, particularly in the ELIGIBILITY DEPENDENT group, who targeted their PrEP discussions based on presumed client eligibility. The concerns related to PrEP mentioned by our participants were similar to the concerns mentioned among medical providers, including increased sexual risk behavior or a lack of adherence among clients [15, 33, 44-46].

Prior findings indicated that race and sexual orientation may be important to understand PrEP implementation; finding staff who identified with a sexual or racial minority group were more likely to be in the UNIVERSAL group and take part in PrEP implementation regardless of a client’s perceived risk of HIV exposure [28]. Although demographic characteristics were not highlighted in these analyses, race and sexual orientation remained predictors of PrEP implementation even when other characteristics of the HIV testing staff (i.e., Characteristics of Individuals such as knowledge, beliefs, and self-efficacy) were added to the model. The persistent significance of race and sexual orientation in predicting PrEP implementation may reinforce the idea that groups disproportionately impacted by HIV may feel a greater need to share information that could reduce the spread of the virus.

Cosmopolitanism was prominent in the qualitative findings, although not significant in the quantitative findings. Cosmopolitanism may be important across implementation groups and thus may not be a predictor of group membership. When studying PrEP implementation, cosmopolitanism may be particularly important because without a sense of which external organizations are providing PrEP, or financial assistance programs, participants may not know to whom to refer clients. Additionally, several participants also noted the importance of trusting external organizations prior to any referrals being made. It may be important to draw a distinction regarding the strength of these external ties—such as if formal partnerships (e.g. memorandums of understanding, etc.) are in place, or if the participant is only merely aware of the resources provided by other organizations outside of the one in which they work. This distinction in future quantitative and qualitative research may be helpful to understand the types of networks necessary to realize the full potential of PrEP implementation.

Available resources was significant in quantitative findings and salient in qualitative interviews. Quantitatively, a greater availability of PrEP resources was found among participants in the UNIVERSAL group compared to participants in the ELIGIBILITY DEPENDENT group. Physical materials describing PrEP may be an important factor for PrEP implementation. In qualitative findings, the importance of the DH1628 form required by the health department was particularly salient. In 2016, prior to the time of this study, questions regarding the use of antiretroviral medications for PrEP and/or PEP were added to the DH1628 [47]. Participants stated that they were required to fill out the DH1628 for each client they test, and this form provided an easy transition to discuss PrEP.

These findings also provide insight into how a PrEP referral typically looks during HIV testing and counseling. Participants differentiated between simply handing a client materials about organizations providing PrEP vs. helping a client to make the initial appointment. HIV testing guidelines in Florida require that community-based staff providing HIV testing should be able to assist clients with linkage to other related services such as PrEP—going beyond a referral or distribution of materials [47]. Based on these findings, this requirement may not be met at all community-based HIV testing sites. Follow through of referrals was not studied here, but these varying methods of PrEP referral may play an important role in PrEP initiation and continued use.

As with any study, limitations exist. Although not all constructs and domains of the CFIR need to be included in every study [29], it is possible that constructs or domains that were not included could be important to PrEP implementation within HIV testing sites. Measurement of CFIR constructs was determined based on available literature [36]; however, at the time of this study many scales had not been previously validated for this use. Despite the scales not being previously validated, all scales had adequate internal consistency within the study. Additionally, some scales were adapted for the population being studied (i.e. staff performing HIV testing). For example, an adapted version of a scale originally intended to measure behavioral skills [34] was used to measure self-efficacy. It is possible that different scales could be more appropriate for the given population. Lastly, this study only collected data from a convenience sample of staff who perform HIV testing in Florida; other types of staff, or those working in other geographic areas, may have different experiences. Furthermore, it is possible that staff taking part in the study were different than those who opted not to take part; for example, 73% of staff identified their organization as being focused on LGBTQ health. Future research should investigate other decision makers, such as administrative staff, as well as the clients receiving an HIV test.

Despite these limitations, this study also has many strengths. The study investigated the Characteristics of HIV testing staff, Inner Setting, and Outer Setting components related to PrEP implementation within HIV testing sites—a relatively understudied context. The study design was a mixed methods triangulation design and included concurrent analysis of the quantitative and qualitative data, allowing for integration and contextualization of the findings. This study also took place in Florida, a state with high HIV prevalence and incidence, that is also geographically and ethnically diverse [48].

Conclusions

Overall, these findings improve our understanding of how PrEP is implemented within HIV testing sites and provides evidence for the potential of PrEP implementation within this setting. Future research can build upon these findings to best optimize PrEP implementation within HIV testing sites. Additionally, given the possible limited role that knowledge, self-efficacy, and PrEP beliefs may have in differentiating PrEP implementation groups, organizational characteristics may be a particularly important consideration when designing a PrEP program.