Text messaging to engage friends/family in diabetes self-management support: acceptability and potential to address disparities

Abstract

Objective

Explore acceptability of engaging family/friends in patients’ type 2 diabetes (T2D) self-management using text messaging.

Materials and Methods

Participants (N = 123) recruited from primary care clinics for a larger trial evaluating mobile phone support for T2D completed self-report measures and a hemoglobin A1c test and then had the option to invite an adult support person to receive text messages. We examined characteristics and reasons of participants who did/did not invite a support person, responses to the invitation, and feedback from patients and support persons.

Results

Participants were 55.9 ± 10.1 years old, 55% female, 53% minority, and 54% disadvantaged (low income, less than high school degree/GED, uninsured, and/or homeless). Participants who invited a support person (48%) were slightly younger, more likely to be partnered, and reported more depressive symptoms and more emergency department visits in the year prior to study enrollment as compared to participants who did not (all p <.05). Participants’ reasons for inviting a support person included needing help and seeing benefits of engaging others, while reasons for not inviting a support person included concerns about being a “burden” and support person’s ability or desire to text. Support persons reported the texts increased awareness, created dialogue, and improved their own health behaviors.

Discussion

Patients inviting a support person had higher need and thus may stand to benefit most. Most support persons were open to engagement via text messages.

Conclusion

Across race and socioeconomic status, text messaging may engage support persons to increase health-related support—particularly for patients with higher levels of need.

Trial Registration

Clinicaltrials.gov NCT02409329.

INTRODUCTION

Successful provision of patient-centered care in the current health care context includes consideration of family and social support networks.^1,2 Family/friends provide regular support for chronic disease self-management to over 65 million (29%–39%) Americans.^3,4 Support persons can help patients access and use health technology, provide self-care reminders, manage prescription refills, aid in health care appointment attendance, and provide support with regimen-related decision-making and problem-solving. Technology creates opportunities to meaningfully engage and inform support persons in accordance with patients’ needs and desires; it may also be particularly useful for leveraging out-of-home and long-distance supporters,³ who report high willingness to be helpful but need more information about patients’ health/treatment regimens to do so.⁵ Most efforts to engage supporters with technology center on patient portal functionality including allowing support persons proxy or delegate access with patient invitation and permissions.^6–9 Meaningful support person engagement via patient portals may improve care and outcomes,^10,11 but it may also exclude disadvantaged and vulnerable patients most in need of support.¹² Disparities in access to¹³ and use of patient portals have been consistently documented.^12,14–18 Portals require user-initiated behavior (set up an account, remember username/password, and log-on) suggesting that support persons who access and use portals are already quite engaged in patients’ health and are technologically savvy. Because of existing portal disparities, efforts to engage support persons via portals may disproportionately benefit patients with more resources who already have support for their care, leading to intervention-generated (or exacerbated) inequalities.¹⁹

In contrast, mobile health (mHealth) technologies can be designed to automatically push out notifications/alerts to support persons in response to patient inputs, providing timely information and nudges for practical self-care help.²⁰ The commercial market has provided mHealth solutions allowing patients to engage support persons as accountability partners. Examples include Medisafe,²¹ an app allowing users to identify support persons to receive notifications when they miss a medication dose, and the Dexcom mobile app²² which allows users to grant others’ access to view their continuous glucose monitoring data. However, few studies have examined characteristics of patients who choose to engage a support person in their health using any type of technology.^23,24

Specifically, text messaging is emerging as a promising platform for supporting chronic disease management and reducing disparities,^25–27 in part due to high rates of utilization²⁸ across socioeconomic and minority groups of all ages.^29,30 Texting interventions have improved adherence to medications and self-care, clinical outcomes, and led to reduced emergency department utilization and cost savings in outpatient populations, particularly among high-risk patients.^31–36 The most effective texting interventions—particularly among disadvantaged or vulnerable patients—link automated texts with human interaction.³⁷ Despite these successes, little is known about patients’ and support persons’ thoughts on this use of automated text messaging,^38,39 and no study has compared characteristics of patients who choose to engage a support person using text messaging with those who do not.

Objective

We sought to examine disparities in adults’ decision to invite a friend/family member to receive text messages as a support person for type 2 diabetes (T2D) self-care in a sample overrepresenting racial/ethnic minorities and disadvantaged patients. We also sought to examine: the acceptability of inviting a support person via automated, tailored text messages; patients’ reasons for inviting/not inviting a support person; and support persons’ receptivity to and feedback about these communications.

MATERIALS AND METHODS

We recruited adults with T2D from primary care clinics in Nashville, Tennessee – including 13 Federally Qualified Health Center locations and 3 Vanderbilt University Medical Center locations – from May 2016 to December 2017 to participate in a randomized trial evaluating mobile phone-delivered diabetes self-care support.⁴⁰ Vanderbilt recruitment oversampled patients with no or public health insurance only and racial/ethnic minorities.⁴⁰ Participants (N = 512) were randomized to 1 of 3 arms: enhanced treatment as usual, daily text support,⁴¹ or daily text support plus FAMS (Family-focused Add-on for Motivating Self-care).³⁹ FAMS provides patient participants monthly phone coaching and the option to enroll a support person to receive text messages for 6 months. Here we focus exclusively on participants randomized to receive FAMS and their decision to/not to invite a support person to participate, support person enrollment outcomes, and feedback about support person involvement from both patients and enrolled support persons at the end of the intervention experience.

Participant eligibility and enrollment

Eligible patients had T2D (self-reported and confirmed by electronic medical record [EMR] review/provider); were prescribed at least 1 daily diabetes medication; were responsible for taking their medicine without a caregiver; owned a cell phone; could read, speak, and text in English; and received outpatient care at a partnering clinic. Patients whose most recent hemoglobin A1c (HbA1c) value in their EMR was <6.8% were excluded, as were patients with auditory limitations, the inability to orally communicate, or who failed a brief cognitive screener. Upon enrollment, participants completed informed consent and a survey in person with a trained research assistant at their clinic or (less frequently) by phone, pen-and-paper, or online. Participants without a recent (<3 weeks) HbA1c in their EMR completed 1 upon enrollment via clinic-administered venipuncture or using a mail in kit.^42,43

Intervention

FAMS^39,40 is based on the family systems theory concept of feedback loops⁴⁴ which describes the sustainability of patient-initiated behavior change as a function of responses from friends/family, which either serve to increase (helpful) or decrease (harmful) likelihood of the change being repeated and sustained over time. FAMS includes goal setting and text message support for patient-initiated health behavior change(s) with skills training to prepare the patient to manage friends/family members’ responses to their efforts to meet the goal (ie, to increase the likelihood of helpful responses and decrease the likelihood or impact of harmful responses). Support person text messages aim to increase dialogue and create safe opportunities for the patient to practice the skills learned in coaching, thereby potentiating the likelihood of a helpful response from the enrolled support person.

FAMS participants receive monthly coaching sessions which include setting a diet or exercise goal, discussing family/friends’ involvement in T2D self-care efforts, and a skill-building exercise designed to help them garner support from family/friends or overcome family/friend related barriers to goal attainment and communicate effectively about needed social support for their goal. Upon completing each of the first 4 (of 6) coaching sessions, participants were asked if they’d like to invite a support person to receive texts about their goal and general information about how to support them with T2D. Support persons had to be at least 18 years old, speak and read in English, have a mobile phone, and verbally consent to participate when contacted by study staff; there were no restrictions on relationship type or cohabitation. Patients could invite a second support person if the first was not reachable, ineligible, or declined.

Patients received 1-way and interactive daily texts for 6 months. Enrolled support persons received 1-way texts every other day. Messages were tailored to preferred time of day and to patients’ name/nickname, gender pronouns, and self-care goals set during monthly coaching sessions and sent by MEMOTEXT, an algorithmic communications and data management platform supporting personalized user outputs and inputs via text messaging. Support person texts were designed to increase T2D knowledge, support for T2D, and dialogue about patients’ goals and support needs (eg, “It's hard to make healthy lifestyle changes by yourself. Help Bob by working on your health goals together.”). Patients were compensated for study enrollment and follow-up assessments plus $10 for each completed phone coaching session (with a possible total of $165). Support persons were compensated for the study enrollment survey ($25) and for completing a follow-up interview ($25).

Data

We used REDCap⁴⁵ to store survey data, track monthly coaching, participants’ decisions regarding support person invitation, and support persons’ enrollment decisions. We operationalized “inviting a support person” as providing study staff with contact information (first name and phone number) and permission to contact the support person. Select data were transferred automatically from REDCap to MEMOTEXT via an application-programming interface for the tailoring and delivery of messages.

Upon enrollment, patients completed self-report measures including demographic information and measures of health literacy (Brief Health Literacy Screen^46,47), family and friend involvement in diabetes self-care (Family and Friend Involvement in Adults’ Diabetes⁴⁸), diabetes self-efficacy (Personal Diabetes Self-Management Scale, 4 item version⁴⁹), depressive symptoms (Personal Health Questionnaire-8⁵⁰), dietary behavior (Personal Diabetes Questionnaire dietary subscales⁵¹), physical activity (MET-minutes per week, International Physical Activity Questionnaire short form^52,53) and medication adherence (Adherence to Refills and Medications Scale for Diabetes⁵⁴). Participants also completed single items asking about self-rated health⁵⁵ and number of emergency department visits and hospitalizations in the 12 months prior to study enrollment. Support persons who enrolled completed a brief self-report survey by phone with study staff which included demographic information.

We invited a subset of participants to participate in an exit interview; all participants who completed their final coaching session between July and November 2017 (n = 44) were invited. Because less is known about the perspectives of support persons on involvement in diabetes support interventions via text messages, all support persons were invited to participate in structured exit interviews to provide feedback about FAMS.

Patient interviews (lasting around 15–20 minutes) were completed by phone and audio recorded with consent, de-identified, and transcribed verbatim for analyses. Herein we focus on responses to questions related to support person invitation/involvement (Appendix A, questions 7–9). Support persons completed post-intervention interviews via phone or online using a unique REDCap link. Interviews completed by phone were audio recorded with consent and then transcribed into REDCap free-text boxes by study staff. Interviews were structured and included close-ended questions (“On a scale of 1 to 10, 1 being not at all and 10 being very much, how much did the text messages … motivate you to talk with the patient about his/her diabetes? … improve your ability to support the patient in reaching his/her health goals?”), each followed by open-ended questions (“Can you tell me why you chose that number?”). In addition, we asked support persons if their knowledge about T2D and its demands had increased since beginning FAMS and the sources of any new knowledge (ie, the texts, conversations with the patient, both, or another source).

Analyses

We used SPSS version 25 and NVivo 11 for analyses. We used nonparametric tests of difference (Chi-square tests for independence and Mann-Whitney U tests) to compare patients who invited a support person and those who did not on characteristics in Table 1. We also assessed potential selection bias in qualitative interviews by comparing interviewed (n = 34) and noninterviewed participants (n = 89) on Table 1 characteristics using nonparametric tests of difference.

Table 1.

Participant characteristics by decision to invite a support person

Mean (SD) or %	Did not invite a support person (n = 64)	Invited a support person (n = 59)	p value
Age, years	57.6 (10.2)	54.2 (9.9)	.031
Racial/Ethnic Minority	56.3%	49.2%	.431
Gender, male	45.3%	44.1%	1.00
Income < $25 000 USDa	45.3%	40.7%	.604
Education, years	14.8 (3.0)	14.1 (2.5)	.284
≤ High school degree/GED	34.9%	40.7%	.512
Homelessb	9.4%	13.6%	.465
Uninsured	20.3%	18.6%	.816
Married/Partnered	31.3%	55.9%	.010
Diabetes Duration, years	10.9 (7.2)	9.8 (6.9)	.449
Insulin Status, taking insulin	42.2%	49.2%	.553
Health Literacy (BHLS)	13.0 (2.7)	12.5 (2.7)	.225
Low (< 4 on any item)	35.9%	42.4%	.465
Family/Friend Involvement (FIAD)
Difference score (helpful – harmful)	0.03 (0.1)	0.4 (0.1)	.027
Diabetes Self-efficacy (PDSMS4)	14.1 (3.7)	13.4 (3.2)	.245
Depressive Symptoms (PHQ8)	4.8 (3.8)	6.7 (5.2)	.042
Dietary behavior (PDQ)
Problem eating behaviors	3.4 (1.0)	3.5 (1.0)	.285
Use of dietary information	2.7 (1.5)	3.0 (1.5)	.391
Physical Activity (MET-min/week)	1539.5 (1434.9)	1699.5 (1795.2)	.842
Medication Adherence (ARMS-Drev)	39.7 (4.0)	39.6 (3.5)	.729
Self-rated health, Poor/Fair	42.2%	55.9%	.128
ED visits in past year, #	0.6 (1.2)	1.0 (1.2)	.017
Hospitalizations in past year, #	0.6 (1.2)	0.4 (0.9)	.439
HbA1c, %	8.3 (1.8)	8.7 (1.9)	.140

Note: Non-parametric tests of difference, Mann-Whitney U Tests, and Chi-square Tests were used.

Abbreviations: ARMS-D^rev, Adherence to Refills and Medications Scale-Diabetes, reverse coded; BHLS, Brief Health Literacy Screen; ED, emergency department; FIAD, Family and Friend Involvement in Adults’ Diabetes, difference score = helpful involvement score – harmful involvement score; PDQ, Personal Diabetes Questionnaire; PDSMS4, 4-item Perceived Diabetes Self-Management Scale; PHQ8, Personal Health Questionnaire-8; SD, standard deviation; USD, United States dollar.

^aMissingness on income (n = 10) equally distributed across decision to invite a support person. $25 000 is roughly the 25th percentile for annual household incomes in the Nashville area, 2017.

^bUS Department of Health and Human Services, Section 330(h)(5)(A): “a homeless individual includes an individual whose primary residence during the night is a supervised public or private facility that provides temporary living accommodations and an individual who is a resident in transitional housing.” In addition, “an individual may be considered to be homeless if that person is ‘doubled up,’ a term that refers to a situation where individuals are unable to maintain their housing situation and are forced to stay with a series of friends and/or extended family members” (HRSA/Bureau of Primary Health Care, Program Assistance Letter 99-12, Health Care for the Homeless Principles of Practice).

We coded transcripts using thematic analyses—identifying, organizing, and interpreting themes identified in the data.⁵⁶ For patient interviews, initial open coding of all transcripts was performed by author EB and a research assistant to identify text about patients’ decisions and experiences around inviting a support person. A codebook was constructed based on coders’ preliminary read of transcripts, in conjunction with existing literature on family support in diabetes and applied to 5 transcripts to clarify definitions and resolve discrepancies. Then all transcripts were coded independently, with one-third coded by both coders to evaluate inter-rater reliability (kappa = 0.85 pooled across codes on these 11 transcripts). EB conducted comparative analyses by patient’s decision to/not to invite a support person.

For support person interviews, descriptive statistics were used in conjunction with thematic analysis. Author SL conducted open coding on responses to open-ended questions. Author LSM reviewed themes and associated quotes and disagreements were reconciled. We compared themes from support persons who rated the text messages high versus low on either of the 2 scaled response questions (from 1 = not at all to 10 = very much) using the median split (7 was the median and modal response for both scaled questions). Responses to open-ended questions about the text messages motivating dialogue and improving ability to support the patient elicited the same themes and often respondents repeated/reiterated the same content for each. Therefore, we combined themes across those questions for presentation.

RESULTS

Of 126 participants randomly assigned to receive FAMS, 3 were withdrawn before having an opportunity to invite a support person; the remaining 123 were included in analyses. Patients’ average age was 55.9 ± 10.1 years, 44.7% were male and 43.1% were married/partnered. Over half (52.9%) were racial/ethnic minorities including 39.8% African American and 5.7% Hispanic. Forty-three percent reported annual household incomes < $25 000; 19.7% were uninsured, 23.8% had public insurance only; and 11.4% were homeless. Fifty-four percent met 1 or more of the following criteria for disadvantaged: household annual income < $25 000 (< 25^th percentile for median household income in Nashville area, 2017⁵⁷), no high school degree/GED, uninsured, or homeless (defined in Table 1). Of the participants who were invited to complete an exit interview, 77.3% (34 of 44) did so (n = 10 who declined were younger, p < .05, and more likely to be male, p < .01). Interviewed (n = 34) and noninterviewed (n = 89) patient participants had similar characteristics except interviewees reported more problem eating behaviors (3.8 ± 1.1 vs 3.3 ± 1.0, p = .02). We invited all enrolled support persons to participate in an interview and interviewed 83.0%.

Patients’ decision to invite/not invite a support person

Nearly half (48.0%) of the 123 FAMS participants invited a support person. Most did so at the first (59%) or second (20%) coaching session. As compared to participants who did not invite a support person (n = 64), participants who did (n = 59) were younger (U = 1461, p = 0.03), more likely to be married/partnered (χ² = 6.653, p = 0.01); they also reported more depressive symptoms (U = 1488, p = 0.04), more emergency department visits in the year prior to study enrollment (U = 1468, p = 0.02), and more helpful than harmful family/friend involvement in their diabetes self-care (U = 1453, p = 0.03; Table 1). There were no differences in race/ethnicity, gender, income, education, health literacy, diabetes duration, insulin status, health-related quality of life, diabetes self-efficacy, dietary behavior, physical activity, diabetes medication adherence, number of hospitalizations in the year prior to study enrollment, or baseline HbA1c. An additional 17 patients expressed interest in inviting a support person but never provided study staff with contact information (Figure 1), indicating 61.8% were interested in inviting a support person.

Figure 1.

Participant preferences for inviting a support person to participate and invitation outcomes.

*Study-related reasons include participant wanted to invite a support person too late in the intervention period, participant wanted to invite a support person who was enrolled in the study as a patient participant.

Of the subset of participants interviewed, 61.8% (n = 21) invited a support person. Themes in their reasons included needing help with diabetes management (n = 5), seeing potential benefits of engaging others (n = 6), and having an existing positive relationship with the support person (n = 6) (Table 2). For instance, patients acknowledged needing help with their diabetes management as a reason for inviting a support person: “I knew I needed to get better … [get] my health under control.” Patients also described their decision to invite a support person based on potential benefits (“The more that I keep [diabetes] forefront in my thoughts and the more positive influences that I have, I thought would be better for me”) or because the support person could help them stay on track with their diabetes management (“I thought it would be helpful for [the support person] also to learn some stuff to help me”). Some patients noted that they chose to invite a someone with whom they already shared a close relationship and/or was already involved in T2D self-care.

Table 2.

Reasons for inviting/not inviting a support person to receive text messages: Themes and associated quotes from patient participants

Theme	Demographics
Reasons for inviting a support person
Needing help with diabetes management
I didn’t know what was going on with diabetes … they told me that I had diabetes, but they didn’t explain to me how it goes, how to do this, how to eat this, you know, exercise and all that.	NHB Female
[T]he things I learned about my medicine, where I could get my medicine, to talk to my doctor about—this is stuff I’ve never done, talk to my doctors about my medicines or the pharmacy. I was just taking the medicine. I knew it was for diabetes, but I guess I didn’t know. I did not know what I know now.	NHW Female
I knew I needed to get my better … [get] my health under control.	NHW Male
Potential benefits of engaging others
Just to, like, have some type of encouragement, somebody to kind of bounce it off of with me while I was going through it.	NHB Female
It helps to remind me to stay on task and keep my goals.	NHB Male
[T]he more that I keep [diabetes] forefront in my thoughts and the more positive influence that I have, I thought would be better for me.	NHW Female
I thought it would be helpful [for the support person] also to learn some stuff to help me.	NHW Male
Existing positive relationship with SP
We are a team … and I always tell my wife—I look at her and say, “Here we go down the middle of the road together again.” And just letting her know that we are together jointly in this journey we are on … She was happy to be involved in my situation.	NHB Male
[Support person] and I get along good. And we do things together, whether she’s here or not. Like, if I’m not with her, we’ll walk together (laughs)—while I’m on the phone, we’ll walk … we even have each other’s keys to our apartments (laughs) … since she kind of does supportive behaviors already, it just made sense for you to do it.	NHW Female
My wife was already involved. And I think bringing her into the fold even more just got her more involved. I think it’s just the right thing to do. I mean, she was already helping.	NHW Male
Reasons for not inviting a support person
Additional “burden” or stress for support person
This takes up time. And some people don’t like to be disturbed, ya know? … I didn’t want to recommend anybody because I know some of my friends … like their privacy. Some of them don’t like being on the telephone.	NHB Female
I think it would be too much … both of my roommates have some health issues. I just couldn’t think it was a thing that I should add to their stressors, you know what I mean.	NHW Female
Everybody’s busy. They don’t have time to deal with what I’m going through.	NHB Female
Concerns about support persons’ ability/desire to receive texts
[My mother] doesn’t know a smart phone from a hole in the ground … she is extremely computer and phone illiterate.	NHW Female
I thought about my son, but he’s on a limited phone plan, and so I didn’t want to use up his text messages … and my daughter gets so many text messages that she would never see them anyway. And my sister hates texting, so I guess she wouldn’t be willing to do that.	NHW Female
Patients’ preference for autonomy or privacy
[When asked about why she chose not to invite a support person:] Then I’d have to tell them that I’m a diabetic, right?	NHW Female
I’m not really a people person. So, I think to try to take things up on my own.	NHB Male
I’m kind of private, and so I just don’t want everybody involved.	NHB Female
Not having anyone to invite
I don’t really have anybody here that would be interested in doing that. I don’t know … I mean, we’re retired. So, it’s me and my husband here … it’s not his thing.	NHW Female

Abbreviations: NHB, non-Hispanic Black; NHW, non-Hispanic White.

Of the subset of participants interviewed, 38.2% (n = 13) did not invite a support person. Themes in patients’ reasons included not wanting to “burden” or add stress to the support person (n = 4), concerns about support persons’ ability/desire to receive texts (n = 4), patients’ preference for autonomy or privacy (n = 3), and not having anyone to invite (n = 2) (Table 2). Some patients declined to invite a support person because they perceived it would add additional burden to their family member or friend—these concerns sometimes came alongside stating preference to handle things on their own or to maintain privacy. When asked why she chose not to invite a support person 1 participant responded, “Then I’d have to tell them that I’m a diabetic, right?” Some were concerned family members or friends were unfamiliar with cell phone technology or expressed they might not like texting.

We asked all patients how we might increase others’ interest in inviting a support person. Suggestions included emphasizing the importance of family/friends in diabetes self-care and highlighting potential benefits—specifically that support person involvement could improve the relationship. Stated examples included improved communication, the opportunity to work towards something together, and mutual support. Patients also recommended sharing success stories to encourage others to invite a support person. Lastly, patients suggested emphasizing that the text messaging can be stopped at any time and that patients can change their support person if desired.

Support person enrollment

Sixty-five support persons were invited by 59 patients, and 53 enrolled (Figure 1). Only 3 of 65 support persons declined to participate when contacted by study staff, stating lack of interest or feeling too busy. Table 3 shows characteristics of enrolled support persons. Although there were not gender differences in patients’ choice to invite a support person, most support persons (77.4%) were female. None of the enrolled support persons withdrew from the study, however, 2 texted “stop” to cease receiving text messages (each after 4 months of texts).

Table 3.

Characteristics of enrolled support persons

n = 53	Mean ± SD or %
Gender, female	77.4%
Age, years	46.8 ± 12.3
Diagnosed with diabetes (type 1 or 2)	13.2%
How informed are you about diabetes and what it takes to manage it?
Not at all/A little	20.8%
Somewhat	39.6%
A lot	39.6%
Race/Ethnicity
Non-Hispanic White	47.2%
Non-Hispanic Black	43.4%
Hispanic	5.7%
Other race/ethnicity	3.7%
Education
< High School Diploma/GED	7.5%
High School Diploma/GED	28.3%
≥ Some College	64.2%
Proximity to Patient
Cohabitating	54.7%
Same Town	22.7%
Same State (different town)	9.4%
Out of State	13.2%
Relationship with Patient
Spouse/Partner	50.9%
Adult child	18.9%
Other relative (Sibling/Cousin/Parent)	18.9%
Friend	11.3%

Abbreviation: GED, general education diploma.

Support persons’ feedback

Nearly half (45.5%) of the interviewed support persons reported knowing more about T2D management than before participating. Of these, 15.7% reported learning via text messages and 75.0% reported learning via both texts and conversations with the patient. Most support persons (77.2%) reported that they talked to the patient about diabetes more than before.

On a scale from 1 = not at all to 10 = very much, 63.6% rated the texts ≥ 7 for motivating them to talk with the patient about T2D and 61.4% rated the texts ≥ 7 for improving their ability to support the patient. Support persons rating the texts ≥ 7 reported 3 common themes pertaining to the helpfulness of the text messages: creation of dialogue, increased awareness about T2D management, and improvement of support persons’ own health behavior. Support persons rating the texts < 7 cited their existing communication about/support for diabetes management prior to the intervention and/or already knowing the information provided in the texts. Themes and associated quotes corresponding to in-text descriptions are shown in Table 4.

Table 4.

Themes and associated quotes from support persons

Theme	Demographics	Cohabitating
Support persons rating the texts high for motivating dialogue and improving support
Created Dialogue
It helped me get an approachable conversation started about diabetes.	Hispanic Female	No
Otherwise I might not have said something as much as I do.	NHW Female	Yes
It gave me a more … how should I put it … more specific things to talk about. Instead of talking about it generally, I was able to pinpoint.	NHB Female	No
It was a good reminder to get that dialogue going. It was very helpful. Mainly the text messages that reminded me to cook healthy or make healthy plans with him, it did open dialogue for us to go shopping for healthier foods.	NHW Female	Yes
Increased Awareness
[T]hat's just not something we used to talk about. But whenever we talk now we try to talk about that. It made me more aware I guess.	NHB Female	No
Because a lot of stuff I didn't know about I was able to share with her and I think that at some point there were messages that were given that she didn't know herself so it benefited the both of us.	NHB Male	No
Gave me more awareness and to think about her more often and think about what she is dealing with.	NHW Female	No
Because I mean every time I get a text I try to provide that support towards her.	NHB Male	No
Improved Health Behaviors
[Now,] we help each other with health.	NHW Female	Yes
Because it encouraged me to talk to her and, for me, to watch what I ate.	NHB Female	No
Support persons rating the texts low for motivating dialogue and improving support
Communication already present
When she first found out that she was a diabetic, we sat down and talked about it … the dos and don’ts and … what you can do to help me. So I kind [of] already knew some stuff.	NHB Female	No
Well, because, the questions they were asking were things that I already knew, that I already been doing. So I'm like, oh, I've already done this 'ya know.	NHB Female	Yes
That’s something that we do anyway. I didn't need the motivation; it’s a way of life. We're always reading up on something whether I get a text message or not. 9 times out of 10, we're already doing what those text messages are saying.	NHB Female	Yes

Abbreviations: NHB, non-Hispanic Black; NHW, non-Hispanic White.

DISCUSSION

In a sample overrepresenting disadvantaged and minority adults with T2D, we found text messaging to be an acceptable approach for engaging friends/family in adults’ chronic disease self-management. Nearly two-thirds expressed interest in inviting a support person to receive text messages, and nearly half did so. We found some evidence that patients’ who invited a support person had higher need for support—younger patients with more emergency department visits in the prior year and more depressive symptoms were more likely to invite a support person. This finding was supported by qualitative evidence that participants who invited a support person reported needing help with their health as a motivating factor for inviting a support person. We did not find disparities commonly cited in the use of web-based tools (eg, race, income, education, or health literacy^58–60) in patients’ choice to invite a support person to receive text messages. Among patients who did not want to invite a support person, concerns about the support persons’ perception of and response to being invited were common. By contrast, support persons rarely declined to participate (3 of 65 invited) or stopped texts early (2 of 53 enrolled), indicating patients’ social networks may be more willing to be involved in their health than patients assume. Notably, only 2 patients cited concerns for privacy as a reason not to invite a support person. Other studies (not specific to support person involvement) have also reported that privacy concerns about text messaging are rare.^61,62 In contrast to portals which provide access to information in patients’ EMR, text content can be limited to specific conditions or types of information, resulting in fewer privacy concerns from patients.

To our knowledge, only 2 other studies^23,24 have examined willingness of adults to engage friends/family in their health/health care using technology, each in predominantly white, but socioeconomically diverse, samples. Each of these studies sent a noncohabitating support person email feedback about patients’ adherence per their responses to weekly interactive voice response calls. Both found patients with inadequate health literacy and lower incomes were more likely to enroll with a support person than patients who did not. In comparison, our study sent texts to support persons without restrictions on cohabitation and found similar, though slightly higher, acceptability of this approach in our sample (48% versus 39%²³ and 42%²⁴). Collectively, these findings indicate patients inclined to invite a support person may stand to benefit most from the option to do so. Thus, interventions engaging support persons with text messaging have potential to reduce disparities, whereas evidence suggests portal and web-based tools to date lack such potential.^12–18

Our findings regarding support persons’ experiences are consistent with a small pilot study of TExT-MED + FANS, a texting intervention among Latino/a adults with diabetes who used emergency department services.³⁸ Support persons indicated high acceptability of receiving texts and, as we found with FAMS, reported both increased communication about health and changing their own health behaviors alongside the patient. However, patients were required to enroll with a support person; thus, by permitting choice, our study provides new insight into characteristics of patients who choose not to invite a support person to receive texts and their reasons.

Other strengths of our approach include recruiting from multiple primary care clinics (including community clinics and academic medical center clinics) and oversampling minorities and patients with lower socioeconomic status who are often underrepresented in studies examining engagement with health information technologies. Because this study was part of a larger randomized controlled trial,⁴⁰ participants did not know they would be offered the option to invite a support person to receive texts, thereby reducing selection bias for the findings reported here as compared to studies advertising/requiring support person involvement. Use of mixed methods captured multiple aspects of the process of support person invitation and enrollment alongside feedback from patients and support persons. Following best-practices in concurrent mixed-methods research,⁶³ we interviewed a subsample of participants to reach thematic saturation, selected for interview invitations only as a function of when they completed the intervention to reduce selection bias, and found no differences between interviewed and noninterviewed participants on any measured characteristics or the decision to invite a support person. Limitations include lacking interview data from invited support persons who chose not to enroll or could not be reached for an exit interview (including the 2 who elected to stop texts early) and/or the friends and family of patients who chose not to invite anyone—each of whom may have different viewpoints—and use of self-report measures of self-care behaviors and emergency department visits/hospitalizations. FAMS text content was specific to T2D self-care support, so generalizability to other conditions should be explored. FAMS content was designed to avoid telling support persons how the patient is doing (eg, adherence, HbA1c results) to reduce potential for nagging/conflict;³⁹ however, this did not come up in patients’ considerations for inviting a support person. Finally, findings may not generalize to efforts initiated by clinics or health systems.

CONCLUSION

Text messaging can be used to engage existing support networks for patients who need/want help managing their health and appears to be well-received by support persons. In contrast to patient portals, text messaging for these purposes may have few disparities and can engage support persons who are not already involved in the patients’ health. Providing opportunities for patients to initiate text messages to support persons can enhance patient- and family-centered care, particularly for adults with chronic diseases for whom most disease management occurs outside clinic visits. Texts may also provide increased support without the privacy and confidentially concerns associated with giving full EMR access/view. Finally, future research should explore potential for text messaging to be used in conjunction with delegate features of a patient portal (eg, sending support persons a text when an out-of-range lab value appears, there is a new visit summary or follow-up appointment scheduled) to leverage benefits of both modalities.

FUNDING

This research was supported by the National Institutes of Diabetes and Digestive and Kidney Diseases, grant numbers K01-DK106306 and R01-DK100694. Simone Laing was supported by the Vanderbilt Diabetes Research and Training Center, Summer Research Training Program (T35-DK007383). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

AUTHOR CONTRIBUTIONS

LSM served as principal investigator, oversaw all aspects of the study design and execution, and led writing of the manuscript. EMB led all qualitative analyses and wrote related portions of the manuscript. SL executed quantitative analyses and some qualitative analysis with oversight from LSM and EB. KJH was lead interventionist and research coordinator for the intervention and developed intervention content and study processes. CAB served as a senior mentor throughout the development of the intervention and conceptualization of the manuscript. All authors read and provided feedback on manuscript drafts.

Appendix A. Patient Participant Interview Questions

1. What was the most useful thing you did in coaching?

2. What was the least useful thing you did in coaching?

3. What did you think about the number of coaching sessions?

   1. We tried to do coaching about once per month. Were the coaching sessions too often, not often enough, or just right?

4. Is there anything you wish had been discussed in coaching?

5. In your opinion, what role do family and friends play in your diet?

   1. What role do family and friends play in your physical activity?
   2. What role do family and friends play in your diabetes?

6. While you were in coaching, how did your thoughts change about the role of family and friends in your health behaviors?

7. We offered for everyone to invite a support person to receive text messages, but some people do not. Why did you/did you not invite a support person to participate?

8. [If they did invite a support person] Did anything surprising or unexpected happen as a result of inviting a support person? [probe for positive and negative]

9. In your opinion, what could we do to get more people interested in inviting a support person?

CONFLICT OF INTEREST STATEMENT

The authors have no competing interests to declare.