The questionnaire response rate was 72.4% (113/156). As shown in Table 2, the participating pharmacists had a wide range of experience. Most respondents were community pharmacists (81.4%, 92/113) and had completed a fourth-year curriculum (77%, 87/113).

Respondents’ research activities are shown in Fig. 1. For the experience of presentation at conferences, there were 64 respondents (57%, 64/113) in “once or less,” 34 respondents (30%, 34/113) in “once every few years,” and 15 respondents (13%, 15/113) in “at least once every year.” Of these respondents, 15% (5/34) in “once every few years” and 60% (9/15) in “at least once a year” had experience of paper publication.

Relationship between frequent presentations in conferences and publications as respondents

The influencing background for low research activities, defined as respondents who answered “once or less” for the opportunity to present the results of research activities at the conference, is shown in Table 3. Univariate analysis revealed that male sex (p = 0.007), workplace at community pharmacies (p = 0.001), absence of mentors for research activities (p = 0.008), absence of mentors for writing a research paper (p = 0.009), not obtaining certifications (p = 0.002), lack of membership in academic societies (p < 0.001), and opportunities to read papers less than once per month (p = 0.001) were associated with low research activities. In the multivariate logistic regression analysis, the adjusted odds ratios for workplaces at community pharmacies, did not obtain certifications, and did not have membership in academic societies were 4.72 [95% confidence interval [CI]: 1.105–20.122, p = 0.036], 8.96 (95%CI: 1.662–48.290, p = 0.011), and 4.46 (95%CI: 1.791–11.115, p = 0.001), respectively.

The influencing background for pharmacists who can serve as mentors in research activities, defined as those who have published papers as the first author, is shown in Table 4. Univariate analysis revealed that workplaces at general hospitals/clinics (p < 0.001), the presence of mentors for research activities (p < 0.001), the presence of mentors for writing a research paper (p < 0.001), being a member of an academic society (p = 0.002), and opportunities to read papers more than once per month (p < 0.001) were involved in pharmacists who can serve as mentors for research activities.

In multivariate logistic regression analysis, the adjusted odds ratios for the presence of mentors for writing a research paper and opportunities to read papers more than once /month were 8.07 (95%CI: 1.356–48.038, p = 0.022) and 34.3 (95%CI: 6.151–191.28, p < 0.001), respectively.

In the DT analysis, pharmacists who did not belong to academic societies (78%, 46/59) or members of one or two societies who did not have certification (100%, 5/5) were seldom involved in research activities. In contrast, pharmacists belonging to more than three academic societies were highly involved in clinical research (12%, 2/17) (Fig. 2). The accuracy of the DT model is 77.0%. The sensitivity analysis also confirmed that the same DT models were constructed.

Decision tree model for predicting the pharmacists with low research activity. The low research activities were defined as responders who answered “once or less” for the opportunity of presentation on the results of research activities in conference

As shown in Fig. 3, pharmacists who read papers more than once per month and had a mentor nearby were likely to publish papers as the first author and serve as mentors for research activities (73%, 11/15). The accuracy of the DT model is 93.8%.

Decision tree model for predicting the pharmacists who become future mentors for research activities. The pharmacists who could become future mentors were identified as those who had published papers as first authors

In the subgroup analysis (Fig. 4), community pharmacists who were not affiliated with any academic societies and had not obtained any certifications (100%, 4/4) or those affiliated with one or more academic societies (79%, 44/56), exhibiting a high proportion of low effort. In contrast, pharmacists who were not affiliated with academic societies but held qualifications showed a lower proportion of low effort, at 34% (11/32). The accuracy of the DT model was 75.0%. A DT model for pharmacists who could serve as mentors in research activities could not be constructed, as there were only six pharmacists who had published papers as first authors.

Decision tree model for predicting community pharmacists with low research activity. This decision tree model focused on community pharmacists (n = 92) as part of a subgroup analysis. Low research activity was defined as responders who answered “once or less” regarding their opportunity to present the results of research activities at conferences