Design and setting.
Data for this analysis was collected via a cross-sectional online survey exploring pharmacist’s roles in opioid use safety [27], among a representative sample from the four more populous states of Australia (Victoria, NSW, Queensland and WA), representing 88% of the Australian population [28]. Results were reported according to the STROBE cross sectional reporting guidelines [29].
Sampling of pharmacies
A comprehensive list of community pharmacies in each state was obtained using two publicly available pharmacy marketing lists, Maven Marketing and Australian Marketing List. These lists were combined, and duplicates and non-community pharmacies were removed. This resulted in a total of 1981 pharmacies identified in NSW, 1237 in Queensland, 1469 in Victoria and 633 in WA.
Participants and procedures
Recruitment occurred throughout August-October 2023. Minimum sample size was calculated using ‘Raosoft” [30]. The total population of n = 5320 identified community pharmacies was used in the calculation with a confidence level of 99% and a margin of error of 5%, with an estimated response rate of 50%, indicated from a previous study [31]. Minimum sample size was calculated at n = 359. Approximately 500 pharmacies per state were randomly selected. The total population of pharmacies was divided into subsets by state, and each subset was randomised using the Microsoft Excel formula “= rand() ”. Pharmacies were contacted via telephone, with 1955 pharmacies contacted in total, based on estimated response rate to provide the required sample size (see Fig. 1) [32]. Once contacted, the pharmacist in charge was invited to take part in the survey, with only one pharmacist per pharmacy eligible to participate to remove issues of clustering. At least three attempts were made, on different days and times, to contact the pharmacist in charge at each pharmacy. Upon speaking to the pharmacist in charge, if they agreed to participate, a link to access the survey was sent to the provided email address.
The online survey was administered via Monash University Qualtrics platform. From the date of the initial email, reminder emails were sent at the one- and two-week marks to prompt participation. The participating pharmacist was required to read the associated study information sheet and provide online consent before they commenced the survey. Pharmacists who completed the survey could opt in to enter a draw to win one of two iPads, with contact details collected via an unlinked Qualtrics form.
Measures
The current study examined a range of potential correlates of stocking naloxone related to the five broad factors outlined below, which were selected based on existing literature [24–26, 33, 34]. As most covariates were recorded as binary responses, other covariates were recoded and collapsed into binary categories for consistent analysis:
1. Pharmacist characteristics
Gender
Male or female. Originally recorded as male, female or non-binary. Non-binary was excluded from analysis due to small response rate (n = 3).
Years of experience as a pharmacist
<15 years or 15 years or more. Originally recorded as a continuous variable where participants entered years of experience manually. These were divided into binary variables using mean number of years’ experience as an indicator of centrality (Mean = 14.88).
2. Pharmacy characteristics
Pharmacy state
NSW, VIC, QLD or WA.
Geographic Location
Capital city/urban and rural/remote. Originally recorded as Capital city, urban, rural or remote. Collapsed to represent urban vs rural areas.
Pharmacy type
Chain/banner or independent/other. Originally recorded as single independent pharmacy, small chain or banner group (2–9 branches), large chain or banner group (≥ 10 branches), or other. To make findings comparable to existing literature [24, 35], independent pharmacy and other were combined into one variable, while small and large chain or banner pharmacy were combined into a second variable.
3. Frequency of opioid dispensing
How often opioid prescriptions were dispensed
Less than once a day, or once or more per day. Originally recorded as once per week, multiple times per week, around once per day, several times and day, and more than 10 times a day. These were collapsed into binary variables based on mean and mode response (Mean = 2.79, Mode = 3.00)
4. Provision of pharmacy services:
Whether the pharmacy offered or supplied the following services
naloxone, needle and syringe program and opioid agonist treatment (OAT) (methadone and buprenorphine for opioid use disorder). All recorded as binary variables (‘yes’ if provided, ‘no’ if not provided)
5. Pharmacists’ comfort
Pharmacists comfort in discussing overdose prevention with patients who use prescription or illicit opioids: Comfortable or Not Comfortable. Originally collected on a 4-point Likert scale [36] with options ‘very comfortable’, ‘comfortable’, ‘uncomfortable’ and very uncomfortable’ collapsed to binary variables.
Statistical analysis:
Descriptive statistics were used to explore and describe sample characteristics as well as participants’ comfort discussing naloxone. A multivariate probit regression analysis with an adjusted coefficient was used to explore the correlates of stocking naloxone and pharmacist and pharmacy characteristics. The final model included 10 variables that included both pharmacist and pharmacy characteristics. A McNemar test was used to test the difference in pharmacists’ comfort levels discussing overdose prevention with patients who take prescription versus illicit opioids. All statistical tests used a p value of 0.05 to determine significance. All analysis was completed using SPSS V28.