**3. Methods**

Data from 2009, 2014, and 2019 National Pharmacist Workforce Surveys were analyzed [38–40]. Data in 2009 and 2014 were collected using a mailed questionnaire to a random sample of licensed pharmacists (3000 in 2009 and 5200 in 2014) obtained from a national data warehouse. In 2019, an electronic survey of 96,100 licensed pharmacists obtained from the National Association of Boards of Pharmacy Foundation was used. Responses from 1200 pharmacists in 2009, 1382 in 2014, and 4766 in 2019 were used for analysis. Two continuous variables were the primary focus of this study: (1) percent time spent in medication providing and (2) percent time spent in the patient care services at each respondent's primary place of employment. Respondents reported the proportion of time they spent in each of the activities. These were two of the six work activities included in each survey, which were defined as:

**Medication providing**: professional services associated with preparing, distributing, and administering medication products, including associated consultation, interacting with patients about the selection and use of over-the-counter products, and interactions with other professionals during the medication dispensing process.

**Patient care services**: professional services not associated with medication dispensing for assessing and evaluating patient medication-related needs, monitoring and adjusting patients' treatments to attain desired outcomes, and other services designed for patient care.

**Business**/**organization management**: managing personnel, finances, and operations.

**Research**/**scholarship**: discovery, development, and evaluation of products, services, and/or ideas.

**Education**: teaching, precepting, and mentoring of students/trainees/technicians.

**Other**: any activities not described in the above categories.

Data were extracted from each database and analyzed for this report. Two variables (percent time in medication providing and percent time in the patient care services) were utilized for conducting a two-step cluster analysis, with IBM SPSS version 24.0 statistical software (IBM Corp., Armonk, NY, USA). The two-step cluster analysis uses a scalable cluster algorithm. The first step of the analysis is to 'pre-cluster' each case (a record) into many small sub-clusters through a sequential clustering approach. The second step of the analysis is to 'cluster the sub-clusters' resulting from step one into the final cluster solution using an agglomerative hierarchical clustering method. The log-likelihood distance measure (a probability-based distance) is applied for each step of the analysis so that both continuous and categorical variables can be used if so desired [41].

For inclusion in cluster analysis, respondents needed to report both their percent time devoted to medication providing and to patient care services. Respondents who reported that they were: (1) retired, do not practice pharmacy at all, (2) employed in a career not related to pharmacy, or (3) unemployed were not asked the work activity questions and, thus, not included for analysis. Respondents who were included for analysis were those who reported that they were: (1) practicing as a pharmacist, (2) employed in a pharmacy-related field or position, or (3) retired, but still working in a pharmacy or employed part-time as a pharmacist.

Our primary goal was to identify pharmacist segments and describe them using descriptive statistics within the context of the new roles for pharmacists and new work systems that were mentioned in the introduction of this paper. Thus, after pharmacist segments were identified, they were compared across several demographic variables using Chi-Square and Analysis of Variance statistics.
