Participants will be individuals aged 18 years or older who are medical providers or administrators in the DCJ or in the surrounding community, people formerly incarcerated in the DCJ, representatives of community-based organizations serving those affected by HIV or with criminal justice involvement, and members of the public health department. Of note, due to restrictions on recording individuals at the DCJ and due to the benefit of addressing the postincarceration transition, the individuals we recruit will be formerly incarcerated rather than currently incarcerated. Formerly incarcerated individuals will be referred by their health care providers in community clinics or can self-refer in response to Institutional Review Board (IRB)-approved flyers posted in community-based service organizations. Health care professionals will be recruited at routine staff meetings and via email invitations. Administrators will be contacted about participation via email or personal contact from members of the study team.

We will conduct 20 to 30 semistructured interviews, which, based on our previous work, should be sufficient to reach saturation (ie, the point at which no new information is being uncovered through additional interviews). Individual interviews are preferred for formerly incarcerated individuals, jail staff, and members of community organizations due to the sensitive nature of questions and the convenience of scheduling.

Separately, we will also conduct a focus group with 10 to 12 members of the HIV/AIDS Re-Entry Coalition (HARC) in Dallas County during one of their quarterly teleconferenced meetings. The HARC is a group comprised of individuals from diverse community organizations who regularly collaborate to enhance the delivery of health and supportive services to individuals who have been recently released from incarceration. Feedback is especially valuable from these group discussions given the breadth and length of experience of the members. Lastly, we will conduct 2 focus groups during a summit event in which we will invite PrEP-providing organizations to compare their experiences and challenges in providing PrEP and related services amid the dynamic funding landscape and to assess opportunities for collaboration.

Semistructured interview and focus group guides will be tailored to each target group, and topics will reflect the range of factors relevant to PrEP implementation across different socioecological levels, as shown in Figure 1. All participants will also be asked to provide feedback on how to adapt and optimize the components of our implementation strategy for local populations, settings, resources, and norms. All interviews and focus groups will be audio recorded and conducted by research staff experienced in qualitative methods.

We will conduct both a rapid qualitative analysis during data collection and a systematic thematic analysis when all data have been collected. The purpose of the rapid analysis is to promote consistent data capture, facilitate communication among the members of the research team about findings, and monitor progress toward saturation [69]. The thematic analysis will systematically categorize patterns in findings across interviews within a group (eg, patients) and across groups (eg, patients, jail providers, and health care leaders), using iterative thematic analysis in accordance with best practice techniques [70,71]. All data will be managed in NVivo 12.0 (QSR International). The study’s principal investigators in conjunction with a study co-investigator who is an established qualitative methodologist will jointly develop a thematic codebook based on a review of the rapid analysis data and 20% of the interview fieldnotes, with codes mapped to the socioecological model presented as part of Figure 1. Trained research staff will double-code all interview transcripts, resolving discrepancies through discussion and conferring with the qualitative co-investigator as needed. A synthesis of the major findings will be presented to stakeholder groups to confirm appropriateness [72].

The Parkland Center for Clinical Innovation (PCCI) has developed and validated an automated matching algorithm to extract data from the DCJ EHR (PEARL) and match it to data from Parkland’s EHR (Epic) from 2010 to the present. Records will be matched based on multiple variables, including patient name, preferred name, date of birth, gender, sex assigned at birth, race or ethnicity, and social security number as available. We will exclude any individuals younger than 18 years. Individuals may have multiple encounters in both the jail and Parkland EHRs due to repeat jail incarcerations or medical visits, and these data will be incorporated as summary variables (eg, number of prior episodes of gonorrhea) with statistical adjustments for repeated measures in an individual. This combined dataset will provide rich data for identifying PrEP candidates and allow us to develop an HIV prediction model. This component of the study will involve the collection of existing data and the development of tools and protocols, and therefore, a waiver of consent will be obtained. Furthermore, matching of records using the identifiers above will be performed behind the firewall of the PCCI, and all datasets will be deidentified prior to sharing with the rest of the research team.

We will use our prior models for predicting incident HIV [73,74] to identify candidate variables for our HIV prediction model. Subsequently, the use of temporal validation will assess how well our model will perform prospectively, dividing our dataset into a derivation dataset (2015-2021) and a validation dataset (2022-present). We will train our model to predict incident HIV diagnoses, defined as positive HIV screening and confirmatory tests or new HIV diagnosis codes without prior evidence of HIV in the EHR [73-75]. We will have ample data on new HIV diagnoses to train our models, as Parkland has a robust HIV testing program and diagnoses more new cases of HIV than any other entity in Dallas [76]. Jail EHR data may be more limited in scope because the jail generally provides short-term episodic care. However, given high recidivism rates [77], many incarcerated individuals will have multiple data points in the jail EHR to inform HIV risk prediction, including data on prior HIV/STI testing and substance use.

We will compare the performance of parsimonious versus full models in the validation dataset based on their ability to discriminate between patients with and those without incident HIV using the C-statistic or area under the receiver operator characteristic curve and the Hosmer-Lemeshow goodness of fit test. Additionally, we will use multiple complementary approaches for estimating HIV risk to capture the greatest number of individuals with indications for PrEP. This will include anyone who identifies as transgender, requests MSM housing, has had a recent STI, or opts-in for HIV/STI testing, in addition to those flagged by our EHR model. We will compare EHR models against these criteria to measure the added value of each approach (eg, number of individuals identified by our model as likely to benefit from PrEP who are missed by patient-reported data and vice versa).

In conjunction with our risk prediction model, the findings from our aim 1 stakeholder interviews will directly inform our protocol for identifying persons who are at increased risk for HIV and likely to benefit from PrEP. This will include elements, such as which practitioners receive EHR alerts about potential PrEP candidates and the content of staff trainings for collecting and acting on patient-reported data and EHR alerts. We will select a final process for identifying candidates for PrEP discussions based on sensitivity (ie, missing as few persons at increased risk as possible), feasibility, and efficiency, while taking into account priorities identified by stakeholders in aim 1.

A key consideration for sexual health discussions in the jail setting is privacy. Feedback in aim 1 will inform how and where these discussions will take place. Providers may also desire additional training in assessing sexual risk, as knowledge of PrEP among correctional nursing staff and clinical providers is vital for successful implementation [78]. Furthermore, the approach to incarcerated individuals about sensitive topics, such as sexual health and substance use, will be informed by best practices, including having these screenings conducted one-on-one by medical staff and partnering with community-based organizations [79]. Any paper materials given to patients will be adapted to the American Medical Association’s recommended 6th grade reading level and worded to avoid disclosing the health status.

Individuals who are identified as being at increased risk for HIV and who test negative for HIV will be considered PrEP candidates. We will develop a PrEP decision-support tool to be used with the diverse populations present in the jail, informed by stakeholders’ preferences from aim 1. This will include presenting information on sexual health and HIV prevention for multiple populations (eg, transgender people and cisgender women), local details on HIV epidemiology, and ways to afford and access PrEP. Individuals identified as PrEP candidates will meet with medical providers or navigators (depending on stakeholders’ preferences) to review HIV prevention options and our patient decision-support tool. Patients will be informed that they will not be charged for these visits, as for all care related to communicable diseases in the DCJ. At this visit, the provider or navigator will assess the patient’s knowledge of and interest in PrEP. Together with the patient, they will work through the decision-support tool. If the patient plans to pursue PrEP care after incarceration, the navigator will initiate the navigation protocol.

Those interested in PrEP will receive intensive navigation that includes both in-person and electronic referrals. Similar to existing navigation protocols (eg, for hepatitis C [80]), the navigator will initiate this process while the individual is still in jail. The navigator will obtain detailed patient contact information, identify and address the patient’s health priorities and potential barriers to care, and create a personalized step-by-step process for accessing PrEP services after release. Patients will also be able to contact the navigator after jail release directly by phone or SMS text messaging.

Multiple PrEP providers in the Dallas community employ peer outreach workers who accompany patients to their PrEP visits. We will work with these providers to develop a protocol for seamless hand-off from the jail-based navigator to the community-based peer. For electronic referrals, we will integrate stakeholders’ preferences for details regarding the transfer of information to community PrEP sites, including what information to communicate, how to communicate, and the timing of communication. Findings from our aim 1 qualitative data will help us pinpoint solutions to barriers or competing priorities to PrEP that patients identify. In some cases, we anticipate that multiple referrals will be necessary for patients to successfully access PrEP.

The project principal investigators will conduct a 2-session interactive group training for frontline DCJ practitioners to introduce our PrEP implementation strategy. Session 1 will be adapted from trainings on culturally competent sexual health care, HIV, and PrEP [81], including clinical indications for PrEP and barriers and facilitators to its adoption in the jail and local community. Session 2 will introduce the rationale and practical aspects of using patient-reported and EHR data to identify PrEP candidates, as our prior qualitative work suggests that providers are more likely to trust, adopt, and use HIV risk assessment tools if they understand how they work [82]. The session will also review how providers and navigators can engage patients in the shared decision-making process and will include demos of PrEP decision-support tools. Lastly, session 2 will cover PrEP navigation processes, including PrEP resources in Dallas County, and navigation protocols. Postsession evaluations will be conducted for rating quality and usefulness.

Our strategy will include the following detailed steps that can be visualized as part of the continuum (Figure 2): (1) identify, (2) engage, and (3) navigate.

In the “identify” step, when individuals enter the jail, a jail EHR record will be created (or an existing record will be opened for those with prior DCJ incarceration). Our automated algorithm will match the jail EHR to the Parkland EHR and generate an individual HIV risk score. For people with predicted HIV risk above a prespecified threshold that maximizes sensitivity without causing alert fatigue among providers, an automated prompt will appear in the jail EHR for providers to discuss PrEP and order HIV/STI testing. As illustrated in Figure 2, providers will also be prompted to order HIV/STI tests for any patients who report risk factors at intake or during a clinical assessment (eg, transgender identity and requests to be housed with MSM) or who request testing directly. Prompts will persist until orders are completed as a “nudge,” a behavioral economic intervention to maximize ordering [83], and results are entered into the EHR.

In the “engage” step, positive HIV tests will prompt referrals to the jail HIV clinic for rapid initiation of antiretroviral therapy according to the current practice, and STIs will be treated based on CDC guidelines [84]. Persons with negative HIV tests will meet one-on-one with a provider or PrEP navigator and receive counseling informed by our PrEP decision-support tool. The navigator we hire will be experienced in working with CL-involved people. Counseling will be person-centered and culturally tailored, with positive sexual health framing.

In the “navigate” step, for patients interested in PrEP, the navigator will review a list of PrEP clinic locations and assist with selecting a provider. Once an individual is released from the DCJ, the navigator will facilitate PrEP linkage by contacting that individual (by phone or SMS text messaging) and providing personalized care navigation. A hotline will be available for released individuals to call for assistance with PrEP referrals. The navigator will continue to follow referred patients for 3 months after any PrEP initiations, recording these contacts and clinical outcomes in REDCap, a secure online data capture system [85].

For this aim of the study, the patient navigator will obtain verbal consent from patients to proceed with a discussion on sexual health and to contact them after release from jail. The navigator will use a script to inform patients about the goals of the project, including the metrics to be collected, and reinforce that the meeting with the navigator is voluntary and will not affect the legal status of the patients.

We will track quantitative and qualitative metrics related to our implementation strategy during a 24-month period of iterative assessment and refinement. Our study metrics map directly to the EPIS framework at key steps in the PrEP care continuum (Figure 2). As shown in Table 1, we will assess quantitative effects, including the number of patients who attend a community PrEP visit, initiate PrEP, and persist with PrEP use for ≥3 months, which is a robust HIV prevention outcome [7]. We will track the rates of new and previously known HIV diagnoses and STIs in the DCJ through EHR chart review.

Feasibility refers to the extent to which a practice change can be successfully used or carried out within a specific setting [86]. We define feasibility as >65% of eligible patients being offered HIV/STI testing and one or more PrEP discussions, our patient PrEP decision-support tool, a PrEP navigation visit, or a PrEP referral in jail.

Penetration refers to the extent to which a practice is integrated within a specific setting [86]. For this metric, we will track the proportion of providers who discuss PrEP or refer to a navigator for at least one eligible patient.

Acceptability refers to the perception among setting stakeholders that a practice is agreeable, palatable, or satisfactory [86]. We will calculate the proportions of patients who accept offers for one or more HIV testing, STI testing, PrEP discussion, navigator contact, or PrEP referral, with acceptance of >80% of test offers and >65% of discussion offers considered acceptable based on prior studies [35,87-89]. We will evaluate differences in feasibility and acceptability among patients by age, sex, gender identity, and race or ethnicity to identify imbalances in the offering and uptake of testing and referrals.

Additionally, on a monthly basis, we will measure the proportion of individuals who test negative for HIV, the proportion of this subgroup who receive notifications of their HIV test results, and the proportion of all those tested who are candidates for PrEP based on US Public Health Service practice guidelines for PrEP [90].

Among these PrEP candidates, we will assess how many are released to the community and how many are successfully contacted by the jail navigator after release. We will track the proportion of candidates attending ≥1 PrEP visit after jail release, the proportion prescribed PrEP, and the proportion using PrEP at 3 months after initiation (using direct patient assessments by the navigator). We will complete a run chart, which we will adapt from quality improvement methodologies [91], to evaluate trends in the PrEP care cascade [92] on a monthly basis during the 24-month intervention period and the additional 6-month sustainment period.

We will also track STIs and new and previously known HIV diagnoses in the jail among priority groups and the general jail population. STI positivity will be calculated as positive tests per the total number of tests completed. New HIV diagnosis rates will be calculated as confirmed positive test results in persons without a prior diagnosis of HIV per the total number of tests completed within a given time period. We will use an interrupted time series analysis using monthly data points to compare rates during the 24 months before and 24 months after our intervention is implemented. Our primary analysis will involve a comparison of the longitudinal rates before and after the intervention, and a mixed effects model for interrupted time series will be used as a secondary analysis. We will conduct linear regression analyses to determine if there is a significant difference in new HIV diagnosis rates before and after the intervention.

Sustainability refers to the extent to which a new practice is maintained or institutionalized within a setting’s ongoing operations [86]. We will assess how well the strategy components are maintained as part of existing jail processes for 6 months after the formal 24-month study period, including tracking retention in PrEP care for up to 6 months (through navigator follow-up). Our research team includes dissemination consultants, who are medical providers working in jails in Austin, TX; Chicago, IL; and Boston, MA, and who will offer input on refining our strategy components for ongoing use at the DCJ and the potential transfer and integration into other jail settings.

For PrEP linkage, current rates of linkage are essentially zero. Thus, in lieu of a pre/post comparison, we will report the proportion of patients linked to PrEP care during the implementation period with 95% CIs. Power analysis is performed for the interrupted time series analysis of monthly HIV positivity rates, assessing the difference in the mean between the 24 months before and after implementation. Over the preimplementation period, an anticipated 250 tests will be performed monthly with 3 new HIV diagnoses per month. In the postimplementation period, of 500 HIV tests per month, an increase to 9 new diagnoses per month is anticipated due to the focus on individuals identified by the prediction model as being at increased HIV risk. Furthermore, we assume a correlation of 0.2 among monthly observations. Based on a mixed effects Poisson regression model, we can detect the difference in the mean HIV-positive rate between the 2 periods with >80% power at a 2-sided type I error of 0.05. We will also perform subgroup analyses, comparing testing rates by sex, age group, and race or ethnicity.

We will also evaluate the qualitative effects of our implementation strategy through structured observations and discussions at the DCJ to assess changes (positive or negative) in knowledge and opinions of PrEP, workflow, and cross-agency collaboration in line with Dallas County’s goal of a coordinated HIV response [93]. Observations will be conducted by individuals with postgraduate training in qualitative methods who have undergone additional training with the DCJ and Parkland Health and who are approved to conduct observations under the protocol approved by the IRB at UT Southwestern Medical Center. Observations will be focused on specific interactions pertinent to the research questions (eg, between the navigator and person soon to be released). Qualitative data on effectiveness and implementation outcomes will be rapidly analyzed in real-time using structured analysis of field notes from observations and discussions. The resulting iterative findings will be presented to a study advisory panel every 4 months to assess evolving attitudes and practices surrounding HIV testing and PrEP (ie, acceptability), gather perceptions of feasibility and penetration of the intervention, and solicit feedback about the ways to improve outcomes.