Participants complete study assessments, receive text messages advising how to access study A1c results, receive quarterly newsletters on healthy living with diabetes, and have access to a helpline for study- or diabetes medication-related questions.

The intervention consists of REACH individually-focused text messaging, plus family-focused phone coaching session, goal-focused text messaging, and the option to invite a family member/support person to receive text messaging.

as measured by International Physical Activity Questionnaire-Short form [metabolic equivalent minutes (MET-minutes) per week] where more MET-minutes per week indicates more physical activity

as measured by hemoglobin A1c (HbA1c, %) We may be under-powered to for these comparative analyses.

In addition to the REACH text messages tailored to user's individual barriers to adherence, participants will receive FAMS components (monthly phone coaching and text messages supporting a goal set in coaching, plus the option to invite a family member/support person to receive text messages) for 6 months. After six months, participants in this arm will receive REACH text messages only.

All participants will also receive text messages advising how to access their study A1c test results, receive quarterly newsletters on healthy living with diabetes, and have access to a Helpline for study- and diabetes medication-related questions.

Helpline and A1c results: Participants complete study assessments, receive text messages advising how to access study A1c results, receive quarterly newsletters on healthy living with diabetes, and have access to a helpline for study- or diabetes medication-related questions.

REACH + FAMS: The intervention consists of REACH individually-focused text mes

Participants assigned to the control group will complete measures at each time point and maintain care as usual (i.e., medical treatment and physician monitoring).

All participants will receive text messages advising how to access their study A1c test results, receive quarterly newsletters on healthy living with diabetes, and have access to a Helpline for study- and diabetes medication-related questions.

Helpline and A1c results: Participants complete study assessments, receive text messages advising how to access study A1c results, receive quarterly newsletters on healthy living with diabetes, and have access to a helpline for study- or diabetes medication-related questions.

Total of all reporting groups

as measured by the Summary of Diabetes Self-Care Activities (SDSCA) Medications subscale; response options are days in the last week ranging from 0 to 7, with 7 representing perfect adherence

as measured by the Adherence to Refills and Medications Scale for Diabetes (ARMS-D); reverse coded such that higher scores indicate better adherence on a scale from 11 to 44

as measured by the Personal Diabetes Questionnaire diet subscale "Use of information for decision making"; possible range 1-6, higher indicates more use of dietary information (better)

Personal Diabetes Questionnaire diet subscale "Problem eating behavior"; possible range 1-6 with higher indicating more problem eating behavior (worse)

as measured by International Physical Activity Questionnaire-Short form [metabolic equivalent minutes (MET-minutes) per week] where more MET-minutes per week indicates more physical activity

Participants assigned to the control group will complete measures at each time point and maintain care as usual (i.e., medical treatment and physician monitoring).

All participants will receive text messages advising how to access their study A1c test results, receive quarterly newsletters on healthy living with diabetes, and have access to a Helpline for study- and diabetes medication-related questions.

Helpline and A1c results: Participants complete study assessments, receive text messages advising how to access study A1c results, receive quarterly newsletters on healthy living with diabetes, and have access to a helpline for study- or diabetes medication-related questions.

as measured by the Summary of Diabetes Self-Care Activities (SDSCA) Medications subscale; response options are days in the last week ranging from 0 to 7, with 7 representing perfect adherence

We compared participants receiving the REACH intervention (combining REACH and REACH + FAMS) with participants not receiving REACH (in condition Helpline and A1c Results).

as measured by the Adherence to Refills and Medications Scale for Diabetes (ARMS-D); reverse coded such that higher scores indicate better adherence on a scale from 11 to 44

We compared participants receiving the REACH intervention (combining REACH and REACH + FAMS) with participants not receiving REACH (in condition Helpline and A1c Results).

as measured by the Personal Diabetes Questionnaire diet subscale "Use of information for decision making"; possible range 1-6, higher indicates more use of dietary information (better)

We compared participants receiving the REACH intervention (combining REACH and REACH + FAMS) with participants not receiving REACH (in condition Helpline and A1c Results).

Personal Diabetes Questionnaire diet subscale "Problem eating behavior"; possible range 1-6 with higher indicating more problem eating behavior (worse)

We compared participants receiving the REACH intervention (combining REACH and REACH + FAMS) with participants not receiving REACH (in condition Helpline and A1c Results).

as measured by International Physical Activity Questionnaire-Short form [metabolic equivalent minutes (MET-minutes) per week] where more MET-minutes per week indicates more physical activity

We compared participants receiving the REACH intervention (combining REACH and REACH + FAMS) with participants not receiving REACH (in condition Helpline and A1c Results).

as measured by hemoglobin A1c (HbA1c, %) We may be under-powered to for these comparative analyses.

We compared participants receiving the REACH intervention only with those receiving REACH + FAMS and with participants not receiving REACH (in condition Helpline and A1c Results). This analysis was subsetted to participants with baseline HbA1c greater than or equal to 8.5% with the goal of improved power.

In addition to the REACH text messages tailored to user's individual barriers to adherence, participants will receive FAMS components (monthly phone coaching and text messages supporting a goal set in coaching, plus the option to invite a family member/support person to receive text messages) for 6 months. After six months, participants in this arm will receive REACH text messages only.

All participants will also receive text messages advising how to access their study A1c test results, receive quarterly newsletters on healthy living with diabetes, and have access to a Helpline for study- and diabetes medication-related questions.

Helpline and A1c results: Participants complete study assessments, receive text messages advising how to access study A1c results, receive quarterly newsletters on healthy living with diabetes, and have access to a helpline for study- or diabetes medication-related questions.

REACH + FAMS: The intervention consists of REACH individually-focused text mes

Participants assigned to the control group will complete measures at each time point and maintain care as usual (i.e., medical treatment and physician monitoring).

All participants will receive text messages advising how to access their study A1c test results, receive quarterly newsletters on healthy living with diabetes, and have access to a Helpline for study- and diabetes medication-related questions.

Helpline and A1c results: Participants complete study assessments, receive text messages advising how to access study A1c results, receive quarterly newsletters on healthy living with diabetes, and have access to a helpline for study- or diabetes medication-related questions.

Any REACH vs. Helpline and A1c results. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. For the HbA1c model, we allowed for a three-way interaction between time, treatment group, and baseline HbA1c because there was descriptive evidence that the effect was modified by baseline HbA1c values. Multiply imputed data (m=1,000) using chained equations.

Adjusted for baseline values with cubic splines (3 knots).

A priori threshold p<.05.

15-month point estimate

We performed an omnibus test of the treatment effect at 3, 6, 12, and 15 months using a robust variance-covariance based Wald statistic. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

Any REACH vs. Helpline and A1c results. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

Adjusted for baseline values with cubic splines (3 knots).

A priori threshold p<.05.

15-month point estimate

We performed an omnibus test of the treatment effect at 3, 6, 12, and 15 months using a robust variance-covariance based Wald statistic. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

Any REACH vs. Helpline and A1c results. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

Adjusted for baseline values with cubic splines (3 knots).

A priori threshold p<.05.

15-month point estimate

We performed an omnibus test of the treatment effect at 3, 6, 12, and 15 months using a robust variance-covariance based Wald statistic. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

Any REACH vs. Helpline and A1c results. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

Adjusted for baseline values with cubic splines (3 knots).

A priori threshold p<.05.

15-month point estimate

We performed an omnibus test of the treatment effect at 3, 6, 12, and 15 months using a robust variance-covariance based Wald statistic. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

Any REACH vs. Helpline and A1c results. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

Adjusted for baseline values with cubic splines (3 knots).

A priori threshold p<.05.

15-month point estimate

We performed an omnibus test of the treatment effect at 3, 6, 12, and 15 months using a robust variance-covariance based Wald statistic. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

To evaluate if REACH only and REACH+FAMS had similar effects on HbA1c relative to control, we subsetted to participants with baseline HbA1c greater than or equal to 8.5% to maximize our power for these analyses. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

12-month point estimate

Testing superiority of REACH only relative to control. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

To evaluate if REACH only and REACH+FAMS had similar effects on HbA1c relative to control, we subsetted to participants with baseline HbA1c greater than or equal to 8.5% to maximize our power for these analyses. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

6-month point estimate

Testing superiority of REACH_FAMS relative to control. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.

To evaluate if REACH only and REACH+FAMS had similar effects on HbA1c relative to control, we subsetted to participants with baseline HbA1c greater than or equal to 8.5% to maximize our power for these analyses. We used generalized estimating equations (GEE) with a working-exchangeable correlation structure and identity link, adjusting for baseline and allowing a time-treatment interaction. Multiply imputed data (m=1,000) using chained equations.

12-month point estimate

Testing superiority of REACH+FAMS relative to control. We used ordinary least squares linear regression with Huber-White heteroscedasticity-consistent standard errors to obtain point estimates and 95% CIs for the intervention effect at each time.