Conceptual framework
This study was based on the EMI framework [17,53]. This framework shifts the locus of evidence production away from universally generalizable knowledge, which is common in traditional biomedical research. Instead, EMI prioritizes a more contextualized scientific process in which data and conclusions are generated through localized public health interventions serving immediate, applied needs. Therefore, the purpose of this analysis is not to present the hypothetically universal experience of naloxone distribution, but rather to examine one location in-depth to understand the forces that directly impacted service delivery and naloxone utilization. The application of the framework to the current investigation can be summarized using the six central tenets of EMI. In applying these principles in the Results section, “Programmatic Context” follows “Quantitative Results” for each set of variables analyzed.
- Material-discursive Process: Naloxone distribution in Pittsburgh is not expected to be the same as anywhere else, yet there is value in understanding the local context. State policies and local drug supply considerations are made when interpreting quantitative data.
- Emergent, Contingent, Multiple effects: Applied to this study, participant behaviors were expected to change over time. Overdose response practices naturally evolved over a 17-year period, instead of assumed to be static, as in shorter studies.
- Practice-based Matter-of-concern: Of central relevance is how the concept of naloxone distribution was interpreted by program staff and locally adapted. For example, the program adapted to the COVID pandemic, and as new naloxone products and street drugs shifted. Therefore, contemporaneous contextual details are provided allow quantitative data to be interpreted with fidelity.
- Practice of Implementation: How the intervention was delivered is of equal importance to other outcomes (e.g., biomedical or pharmacological). Therefore, logistical considerations and site expansion rationales are provided in detail, especially in ways that impacted participant recruitment and training of participants, and ultimately, the quantitative data.
- Performative Work of Science: Administrative data were collected first and foremost for service delivery, and the scientific knowledge generated from their review is an added benefit. While data were collected with the intention of analysis, the questions asked of participants were also designed to gather information on reversals that would reveal opportunities for counselling and behavior change at the point of care.
- Equality of Knowledge: Program staff’s experience of service delivery is of equal explanatory value as quantification of administrative records. Program staff were included in each step of the analysis process, and their experiences are recorded in the Results section, and they are co-authors of this manuscript.
The Equality of Knowledge principle, a recursive process for knowledge generation was applied, starting with whole-team generation of the research questions. The data analyst (ND) generated tabular and graphical representations of time trends for batches of variables. The team then assembled to discuss patterns, aberrations, policy impacts, public health implications, and topics for further investigation, including new research questions based on discussions of programmatic context. After the initial discussion, the analyst would prepare follow-up tables, developing statistical methods as the inquiry warranted, and refine time trend graphs, which were then presented at the following meeting. This recursive process was applied to each set of variables in the dataset until all variables had been analyzed and discussed. In addition to the five research questions elaborated in the pre-registration, the recursive process resulted in three additional research questions described above.