Study Design
This is a cross-sectional study of participants hospitalized with IDU-associated infections at four hospitals in Maine in counties deemed high risk for HIV/hepatitis outbreaks (Van Handel et al., 2016). A convenience sample of n=101 participants was prospectively recruited from January 1, 2019 through March 18, 2020. Study enrollment was halted due the COVID-19 pandemic.
Inclusion and exclusion criteria
Criteria for study enrollment included: 1) inpatient infectious disease consultation for a primary diagnosis of an IDU-associated infection such as infective endocarditis, skin/soft tissue infection, osteomyelitis, HIV or viral hepatitis, or whose chart has been reviewed by the ID antibiotic stewardship team and found to have an IDU-associated infection; 2) age 18-65; 3) EHR-reported or self-reported injection drug use and/or recent stigmata (e.g., injection sites on physical exam); 4) English speaking; and 5) ability to provide informed consent. Exclusion criteria included intubation, suicidal/homicidal ideation, or if the individual showed signs of psychotic symptoms. Data were collected through Audio Computer-Assisted Self-Interview (ACASI) ("ACASI: Audio Computer-Assisted Self-Interview Software,") survey and medical record review.
Measurements
Outcome
The primary outcome was SSP utilization which was defined as 1) having reported using an SSP in the past 3 months or 2) responding to the question about most common ways the participant got to an SSP in the past 3 months.
Variables
The main independent variable was driving distance to closest SSP. Driving distance from the closest SSP to the participant's address was calculated in miles with an online map tool. If the participant was experiencing homelessness or if the participant's address was not available in electronic health record (EHR), driving distance was calculated between the closest SSP and the self-reported place/zip code centroid where the participant most frequently lived or slept in the past 90 days.
Some variables were collected through self-report. Self-report demographic and health variables included gender, history of incarceration, willingness to take pre-exposure prophylaxis for HIV/discussed pre-exposure prophylaxis with provider, condomless sex, and homelessness. Self-report variables regarding substance use included overdose history and injectable and non-injectable drug(s) of choice. Severity of opioid use disorder was measured using the Short-inventory of Problems-Modified for Drug Use (SIP-DU) (Alterman, Cacciola, Ivey, Habing, & Lynch, 2009). The Bacterial Skin Index Risk Score (BIRSI) score, which includes questions about alcohol pad and sterile water use, handwashing, rotating injection sites, injecting subcutaneously or in the muscle (“skin/muscle popping”), and clean needle use, was used as a continuous score to measure risk of skin and soft tissue infections (Phillips & Stein, 2010). Syringe disposal variables (i.e. disposal of needle/syringes in SSPs, public places, etc.) were also collected through self-report. Unhealthy alcohol use was categorized using the AUDIT-C score (Saunders, Aasland, Babor, de la Fuente, & Grant, 1993). Naloxone uptake was defined by self-report use of naloxone on another person, or having naloxone used on themselves. Self-report variables around SSP utilization included barriers to access, reasons for re-use of needles/syringes and other drug injection equipment.
Additional variables such as having a primary care provider (PCP), prior infectious complications, history of sexually transmitted disease and viral hepatitis were collected by a combination of self-report and EHR data. Hepatitis C (HCV) exposure was defined as positive if the participant self-reported history of HCV, HCV found in EHR, HCV antibody positive and no HCV RNA available. HCV infection was defined as HCV RNA detectable or self-reported history of HCV treatment. Hepatitis B (HBV) infection was defined as self-reported chronic HBV, HBV listed in EHR, or positive HBV DNA. HIV infection was defined as self-reported HIV infection, positive HIV antigen/antibody or HIV noted in EHR. Vaccinations for Hepatitis A, HBV, HCV, and Tdap were collected via self-report and EHR. Rurality was categorized as either rural (small or isolated rural) or urban (large rural or metropolitan) using the Rural-urban commuting area (RUCA) codes (Agriculture, 2019). Other variables were collected through EHR review exclusively. Such variables included insurance status, infectious disease diagnosis, Charlson co-morbidity index (Charlson, Pompei, Ales, & MacKenzie, 1987), prescribed MOUD prior to admission. MOUD prior to admission was defined as buprenorphine, buprenorphine/naloxone, naltrexone, or methadone on the pre-admission medication list.
Statistical analyses
Descriptive analyses were performed to characterize injection knowledge, attitudes and practices. The primary outcome was past 3-month SSP utilization, and the main independent variable was driving distance to closest SSP. Data were compared between subgroups using t tests for continuous data; chi square tests or Fisher’s exact tests were used as appropriate for categorical data. Logistic regression analyses were performed to identify factors associated with SSP utilization. Potential covariates were chosen a priori based on clinical knowledge and literature review. Bivariate unadjusted odds analyses were performed testing the following variables: gender, insurance, employment status, homelessness, overdose history, PCP, insurance, condomless sex, driving distance, MOUD, trouble accessing SSP, HIV, HBV, HCV, SIP-DU and BIRSI scores. Gender, homelessness, history of overdose, having primary care physician and distance to SSP were included in the final multivariable regression model based on statistical significance of p<0.05 in the bivariate, unadjusted odds analyses. SAS Enterprise Guide version 7.1 was used for the analysis.