Table 4.

Estimates and percent change in the estimate for the mediator effects of the emotional distress variables on longitudinal medication adherence over 36 months

Emotional distress factors	Without mediators: estimate	With mediators
		Treatment satisfaction		Specific necessity		Specific concerns		Self-efficacy
		Estimate	%Δ	Estimate	%Δ	Estimate	%Δ	Estimate	%Δ
PHQ-8 total score	−0.35	−0.33	5.16	†	†	−0.31	11.94*	†	†
PHQ-8 somatic symptom score	−2.28	−2.17	4.51	†	†	−2.00	12.07*	†	†
PHQ-8 cognitive-affective score	−2.19	−2.05	6.50	†	†	−1.92	12.34*	†	†
DDS average score	−2.19	−2.17	1.18	−2.25	−2.69	−1.85	15.50*	−2.27	−3.59
DDS emotional burden	−1.27	−1.22	3.22	−1.34	−5.74	−0.87	31.46*	−1.36	−7.70
DDS regimen-related distress	−2.48	−2.50	−0.71	−2.51	−1.12	−2.30	7.42	†	†
DDS interpersonal distress	−1.23	−1.15	6.34	−1.27	−3.62	−1.01	17.25*	−1.28	−4.64

Data are shown first unadjusted (no mediators) and then adjusted for potential mediators one at a time, with the percent change (%Δ) (% mediation) of the emotional distress effect. All results are based on linear mixed-effects models with random effect for participant with adjustment for age, sex, race/ethnicity, randomized treatment group, and longitudinal duration of T2DM, BMI, and HbA_1c. All coefficients have P values <0.0001.

*Factors with a percent change (mediation) of at least 10%.

†No mediation, as the predictor was not associated with the proposed mediator; because “control over diabetes” and DDS physician-related distress did not show any mediation, they were not included in the table.