Optimizing STI/HIV care for safety net patients in South King County, Washington: Implementation science methods for planning efficient and equitable service provision

Abstract

Sexually transmitted infections (STIs) are associated with significant health and financial burden in the United States, affecting an estimated 20% of the population per year. Bacterial STIs and HIV disproportionately impact low-income, immigrant, racial, and ethnic minority populations nationally and in King County (Washington). While health surveillance data reveal obvious STI disparities, it is unclear how to address care gaps in ways that are locally feasible and acceptable to both providers and care recipients. The objective of this project was to leverage implementation science methods to identify optimal strategies for improving STI/HIV testing and treatment services in South King County, the cities and neighborhoods located south of Seattle which experience the highest regional bacterial STI rates and high and growing HIV rates. Given South King County hosts a high proportion of clients living below the poverty line, this study specifically focused on opportunities to improve safety net services, which provide medical care to clients irrespective of their ability to pay, often accepting Medicaid as an insurance and providing sliding scale services to medically uninsured clients. This project includes three studies which, together, identify priorities and potential actions for improving STI/HIV care in South King County from the perspective of providers and clients. In the first study we conducted a landscape analysis of STI/HIV service delivery, using an exploratory sequential mixed-method design (qual -> QUAN). We first conducted 12 key-informant interviews with local service providers and applied rapid thematic analysis guided by the Consolidated Framework for Implementation Research (CFIR 2.0) to identify barriers and facilitators to care. Insights from these interviews informed the development of a quantitative prioritization survey, which was administered to 31 STI subject matter experts to rank the importance of noted barriers and the feasibility and effectiveness of proposed implementation strategies to address them. We identified key challenges related to limited healthcare accessibility, misconceptions about STIs among safety net clients, healthcare workforce shortages, and provider knowledge gaps relating to a shifting STI landscape. Prioritized implementation strategies included mass media campaigns to improve STI knowledge among the public, increased funding for safety net providers to implement operational changes that improve accessibility of care, and use of technology to expand and streamline STI testing. Notably, no feasible strategies were identified for addressing the barrier of high client out-of-pocket costs. In the second study, we conducted a discrete choice experiment (DCE) with potential safety net clients in South King County, to identify preferences for service delivery attributes that could promote client accessibility and utilization of STI/HIV services. Clients completed 12 choice questions, exploring their preferences related to five STI/HIV service and facility-level attributes including: cost, travel time, type of clinical location, appointment hours, and appointment scheduling method. We fit an effects-coded Hierarchical Bayes model to estimate average attribute-level preference weights and overall attribute relative importance scores. We also conducted subgroup analyses to examine differences in preferences by survey language of choice and type of health insurance. A total of 250 clients completed interviews, 190 of whom completed surveys in English and 60 in Spanish. Most clients were either enrolled in Medicaid (42%) or held no insurance at the time of the survey (33%). Cost was the most important attribute overall to clients, accounting for 42% of the impact on choice in where to access services and remaining the top priority across all subgroups. The next most important factors were travel time, with a preference for shorter commutes, followed by type of facility location. Participants demonstrated strong preferences for traditional clinical settings such as general health clinics and sexual and reproductive health clinics, over services offered at community locations such as libraries or through health fairs. Our findings also suggest that, to a lesser degree, shifting hours of operation to allow for evening or weekend appointments and improving online scheduling can further align services with client preferences. In our third study, we use an optimization modeling technique to explore ideal safety net locations to bolster STI/HIV services within South King County. We used a Maximum Covering Location Problem (MCLP), which aims to maximize demand coverage within a distance or time threshold that is hypothesized to be geographically accessible. We expanded traditional MCLP models to also account for facility-level attributes, such as clinic hours, when defining accessibility and estimating utilization. Our objective function maximized the number of STI positive cases within South King County who have access to STI services. We used simulations based on data from our discrete choice experiment to estimate utilization rates for different facility-models. We also ran a series of one-way sensitivity analyses to test variability in model inputs. Our model identified three key locations for expanding service delivery, resulting in an overall coverage rate of 83%, meaning that the three selected locations were estimated to be accessible to 83% of clients diagnosed with STIs. Sensitivity analyses explored the impact of using different data inputs and model constraints, resulting in similar solutions. Prioritized locations were primary care clinics with expanded hours of operation. Results from our model could be incorporated into healthcare priority setting processes, to aid stakeholders in their decision-making processes about strategic locations for driving health outcomes under resource constraints. This dissertation contributes to the literature by providing a use case for the practical application of implementation science frameworks and methods for local health departments. Study findings specify useful information about determinants of STI/HIV safety net service delivery, alongside feasible and effective implementation strategies for addressing key challenges. Additionally, we generate information about relative preferences of clients when accessing STI services, allowing safety net providers and public health practitioners to tailor care to clients. Lastly, we demonstrate a practical use case for applying DCE results to identifying locations for bolstering care that maximize client preferences and estimated utilization of care. Learnings from this work were grounded in participant ideas and feedback and provided an opportunity to explore alignment between client and providers in terms of priorities.