BACKGROUND: By 2045, the global population of people with diabetes (PWD) is projected to reach 783 million. Health education tools are needed to enhance glycemic management among PWD. PURPOSE: This study evaluated the effects of a diabetes conversation map (CM) intervention on glycated hemoglobin (HbA1c) and examined the mediating roles of health behavior improvements among PWD in Taiwan. METHODS: A large randomized controlled trial (N = 602) investigated: (1) whether an additional 1-hour, theory-driven CM intervention (N = 300), compared with usual shared-care service only (N = 302), could significantly better improve PWD's HbA1c at 3-month posttest, and (2) whether the CM intervention's effects on HbA1c reduction were mediated through desired changes in diet and exercise health behaviors between pretest and 3-month posttest. RESULTS: Multivariate linear autoregression analysis demonstrated that controlling for baseline levels, the CM group exhibited significantly lower HbA1c (β = -0.101) at 3-month posttest than the control group. In addition, the CM group showed significantly greater improvements in both diet (β = 0.261) and exercise (β = 0.239) health behaviors between the pretest and 3-month posttest, compared with the control group. Further mediation analysis revealed that the CM intervention's effects on HbA1c reduction were primarily mediated through improvements in diet (β = -0.126, P <
.002), rather than improvements in exercise (β = -0.023, P = .465). CONCLUSIONS: Integrating brief, single-session CM interventions, as described in this study, into existing diabetes shared-care frameworks may effectively enhance diet and exercise health behaviors and thereby improve HbA1c management among PWD.