Characteristics of Health-related Text Messages Preferred by Medically Underserved African-American Patients with Diabetes

Introduction Text messaging (TM) is increasingly used by the U.S. medical practices and healthcare delivery systems, but little is known about preferences of medically underserved minority patients for TM supporting improved self-care decisions. We sought to determine the characteristics of text messages and TM programs preferred by African-American patients with diabetes in medically underserved areas. Methods This convergent mixed methods study employed a self-administered survey and focus group interviews. Quantitative and qualitative data were collected simultaneously, analyzed separately, and merged to provide a holistic view of the TM characteristics patients preferred. Participants (N = 36) were recruited from a medically underserved area in Memphis, Tennessee. Focus group data were uploaded into the NVivo qualitative data analysis software program, and main themes were identified. Standard frequencies were calculated for survey responses. Results Participants ranged in the age of 22-74 years (M = 54.1; SD = 14.6) were predominantly female (77.8%), African-Americans (88.9%), and had at least a high school education (91.7%). A majority used mobile phones for sending (69.4%) and receiving (72.2%) text messages. Participants wanted to receive daily (44.4%) or weekly (47.2%) text messages from their healthcare provider (61.1%), or a motivational message program (33.3%). They preferred actionable messages with a positive tone and wanted options to customize message type, content, and frequency according to their preferences, goals, and needs. Discussion Medically underserved African-American diabetes patients want customized text messages that are practical, actionable, encouraging, and from their doctor. Healthcare providers seeking to develop patient-centered TM programs for medically underserved minority patients should personalize and tailor messages according to patient preferences, health goals, and self-care needs.


Introduction
African-Americans are nearly twice as likely to be diagnosed with diabetes as Whites and more likely to suffer from complications associated with diabetes [1]. African-Americans are also more likely to live in medically underserved areas with inadequate primary care and support for effective diabetes self-care [2][3][4]. Nationwide, only 43% of all patients with diabetes, 50% with hypertension, and 80% with hyperlipidemia have reached their recommended goals for average blood sugar, blood pressure, and cholesterol, respectively [5][6]. Control rates are even lower in medically underserved African-American communities, resulting in higher rates of hospitalization, complications, and death [4,7].
Health-related text messaging (TM) is increasingly being used in practices and health systems across the country to engage patients with chronic disease in self-care [8]. Research suggests that strategies aimed at incorporating self-care interventions into the primary care infrastructure can significantly improve the management of the chronic disease [9][10][11]. Preliminary research suggests that TM may be a particularly low-cost and effective approach for supporting self-care in primary care settings [12][13][14][15][16]. Furthermore, TM is practical, widely available and is less costly than other electronic forms of physician-patient communication [12].
However, despite the growing interest in determining TM efficacy and effective best practices in primary care, little is known about patient preferences for TM program characteristics. Previous work in this area is preliminary at best, and although studies have assessed interest, acceptance, and effectiveness of a TM program, the patient perspectives on the types of TM they desire is lacking [12,14,16]. Even less is known about the perspectives of medically underserved health-seeking populations. Humble and colleagues demonstrated that African-American diabetic patients in medically underserved areas are very interested in receiving text messages from their doctor's office and were particularly interested in lab results, explanations of results, and diabetes information [12]. However, evidence regarding the most effective TM frequency, content, length, and level of interactivity are lacking.
This study combined quantitative and qualitative assessment of patient perspectives on the characteristics of text messages and primary care-based TM programs. We sought to determine those characteristics most likely to be effective in supporting improved self-care decisions by medically underserved African-Americans with diabetes. The goal of the qualitative component was to obtain more in-depth information from patients on their preferences regarding TM. We hypothesized that participants will express interest in receiving individualized motivational text messages from health providers about managing diabetes and they will have preferences for the TM program characteristics.

Methodological framework
Quantitative data were collected using a self-administered survey, while qualitative data were collected using focus group interviews. Survey and interview questions were designed to assess patient interest in a motivational TM program, as well as the types of messages they preferred.
The pragmatist paradigm provided the framework for this mixed-methods study, which employed a convergent design [17]. This design involves the concurrent collection and independent analysis of quantitative and qualitative data using the analytic approaches best suited to the research questions and supporting the assumptions underlying the quantitative and qualitative approaches. The results of these separate analyses are merged and compared. The pluralistic nature of pragmatism supports the concurrent collection, independent analysis, and merging of quantitative and qualitative data in this study. In the practice-oriented pragmatist paradigm, the focus is on research questions, problems to be addressed, and what will work best in clinical settings [18]. Therefore, using parallel collection and analysis of quantitative and qualitative data and subsequent merging of data sets, a more holistic and real-world view can be obtained of the potential TM characteristics most likely to be effective [19][20].

Recruitment and participants
Participants were recruited from the medically underserved and majority (96.0%) African-American Memphis suburb of Whitehaven, TN through the Methodist Le Bonheur Healthcare South Hospital and its affiliated clinics. Median household income of the study population is $37,016 with approximately 28% living beyond the poverty level [8]. Potential participants included adult patients with diabetes and their family members. Level of diabetes control through average blood sugar (A1c) or self-report was not assessed or considered as inclusion criteria. Recruitment strategies included distribution of flyers at clinics, letters to patients, and telephone calls. All participants were at least 18 years of age, spoke English, and were capable of answering survey questions and participating in a focus group interview. The same participants (N = 36) participated in both survey and focus group data collection. Study participants indicated their consent by answering text message survey questions and taking part in focus group discussions. The study, TM survey, and focus group interview guide were all approved by The University of Tennessee Health Science Center Institutional Review Board.

Survey and focus group interview guide development
The TM survey and focus group questions were developed by the research team using standardized questions and data from peer-reviewed articles [12,14,[16][17]. Qualitative, quantitative, and medication adherence research experts assisted in refining the survey and interview guide to ensure accuracy and stringency.
The TM survey was designed to gather demographic characteristics, data about participant technology use, and interest in a TM program supporting diabetes self-care. Health literacy was assessed using Chew's validated one-question screening instrument and employed a 5-point Likert Scale [21][22]. The survey had 27 questions in four categories: literacy (1), demographics (7), mobile phone use and texting (6), and interest in specific health topics or types of text messages [13]. The survey (Appendices, Figures 1-2) included sample text messages to assess the preferred message topic, type, length, and degree of personalization. The focus group interview guide (Appendices, Figures 3-7) was designed to collect information about participant experiences in diabetes self-care management and text message preferences.

Survey administration and focus group interviews
Four focus groups were conducted between June 2015 and April 2016 in hospital and clinic conference rooms at Methodist Le Bonheur Healthcare South Hospital and Christ Community Health Services. An experienced qualitative researcher trained the focus group facilitators and assistants using the guide by Krueger and Casey [23]. A medical student conducted the first three focus groups, and a faculty member facilitated the fourth one. The facilitator explained the purpose and scope of the study. Participants then completed the 27-question selfadministered paper-and-pencil survey. Survey completion took about 10-15 minutes, and the focus group immediately followed. Focus group sessions were audio-recorded and lasted 90, 40, 40, and 75 minutes, respectively.
The initial focus group was conducted with members of the Diabetes Wellness and Prevention Coalition Patient Advisory Council (PAC), a patient advocacy group. Members of the PAC were newly diagnosed and chronic diabetic patients, caregivers of diabetic patients, and community members. The PAC focus group served as the pilot. The second focus group consisted of diabetic hospital employees and caregivers of diabetic patients, while the third and fourth focus group participants were diabetic patients.

Data analysis
Survey results were analyzed using descriptive statistics. Focus group recordings were transcribed verbatim, de-identified to protect participant confidentiality, and uploaded into NVivo 11 qualitative data analysis software (QSR International, Melbourne, Australia). Two study researchers read and re-read the transcripts and used thematic analysis to identify common categories and themes [24]. Researchers integrated the quantitative and qualitative data using a joint display to visually represent and integrate of survey responses with corresponding focus group transcriptions by content area [17][18].

Results
All participants (N = 36) completed both the survey and participated in one of the four focus groups. Participant demographics are reported in Table 1.

Quantitative findings
Results from the descriptive statistical analysis of the TM survey are presented in Tables 1 and  2. Participants wanted to improve their self-care in the areas of healthy eating (77.8%), physical activity (75.0%), and taking medications correctly (55.6%). The majority (94.4%) reported using a mobile phone, with 47.2% using a smartphone. Participants' use of mobile phones to send and receive text messages was similar (69.4% and 72.2%, respectively). The frequency of TM ranged from daily (50.0%) to weekly (19.4%) and monthly (2.8%). Ten individuals did not use their mobile phones for TM.  Participant preferences for text message delivery, content, length, and message interactivity are shown in Table 2. The majority preferred for motivational text messages to be delivered daily or weekly. Although one-third preferred messages from a TM program, almost two-thirds preferred text messages from their health provider. Educational, trivia, and motivational text messages were the most preferred types of messages.

Qualitative findings
Focus group discussions yielded six major themes: 1) diabetes self-care struggles; 2) use of mobile phones; 3) perceived advantages of TM and TM programs; 4) perceived barriers of TM and TM programs; 5) desired characteristics of TM programs; and 6) undesired characteristics of TM programs ( Table 3).

Themes and Categories
Quotes The texting is less intrusive than a call, so people will respond quicker.
If I call they are gonna talk too long, so I just text…because everybody has friends that just want to go on and on. And on and on.
Straight to the point.

Diabetes Self-care Struggles
Participants disclosed and discussed their struggles with diabetes self-care surrounding discipline, lack of education, fear related to glucose monitoring and injecting insulin, financial difficulties, and time limitations. Study participants frequently cited lack of discipline as their primary obstacle to achieving their self-care goals. In particular, they cited difficulties in finding the discipline to exercise, eat healthy foods, monitor glucose, take medication properly, and establish good sleep habits. Participants reported that they wanted to know what to eat, how much to eat, and how to prepare food. They also identified financial difficulties as an impediment to maintaining a healthy diet or adhering to medications. Participants also expressed concerns about paying for health insurance, medication, and equipment. Others said they were unaware of the exact nature of their illness and associated complications. Fear of needles was a barrier for insulin adherence. Participants suggested that a TM program could help them overcome these struggles by providing helpful reminders, as well as information about grocery shopping, food preparation, exercises that fit individual lifestyles and schedules, and local resources to help those with financial difficulties obtain healthy foods and medications.

Use of Mobile Phone
Participants said mobile phones were a convenient way to keep in contact with family members, friends, and church members. Some participants also use their mobile phones to complete errands, while others use it as a way to check in with loved ones during the day when away from home. Participants reported that younger adults are more adept at texting, while older adults are often newer to texting. One participant suggested offering an optional TM lesson for those who chose to enroll in a TM program.

Perceived Advantages of TM
Participants identified many advantages of TM and a potential TM program, including convenience, access to information, the private nature of messaging, and helpfullness in developing a supportive community. Many participants identified TM as convenient because it prevents lengthy conversations, is less intrusive than a phone call, and allows the user to respond at his/her own discretion. Furthermore, they identified a TM program as a potentially effective method to enhance diabetes patient education both for enrolled patients and for their family or friends. Participants mentioned that TM would also allow program participants to share knowledge with loved ones. Privacy was an important concern and participants agreed that a TM program would provide a level of privacy. Lastly, while most participants were able to identify advantages of TM and a potential TM program, one participant did not identify any advantages.

Perceived Barriers of TM
Participants identified several possible TM barriers: lack of texting experience, inconvenience, compromised safety, lack of community, visual impairment, and cost. Discussions revealed participants' concern that TM could lead to bothersome messages, and could, therefore, be less convenient than other forms of communication. Several participants voiced safety concerns about texting and driving. When comparing a TM program to a support group, one participant noted that TM can be isolating. A visually impaired participant reminded the group that many diabetes patients are blind. Lastly, a potential barrier identified during group discussions was affordability, as the cost of adding TM to a mobile phone plan might deter would-be participants.

Preferred Characteristics of TM Program
Participants preferred a customizable and educational program that features uplifting, instructional, simple TM that considers the age of the recipient and incorporates images and reminders. According to participants, a customizable program should include a voicemail option for the visually impaired. Other options identified were the frequency of TM, level of interactivity, message content, message length, level of personalization, and timing of TM to meet their needs. They differed on the degree of personalization preferred, with some preferring medication reminder text messages that were highly personalized (e.g. using patient name and medication details) and others preferring generic medication reminders. Participants also wanted to have an "opt out" option and the ability to change program settings. Many desired a program that would provide information and educational opportunities about exercise, diet, and diabetes complications. They also noted that newly diagnosed diabetic patients are suddenly faced with the task of navigating a new lifestyle and often learn about complications when it is too late. In terms of the types of TM, participants desired what one participant called a "mom message," a message with instructions and reminders. Others wanted uplifting messages or messages with pictures that would be motivating. Additionally, participants listed product solicitation, poor timing, and too many messages as potentially undesirable TM features. Table 4 provides a joint display that integrates the quantitative and qualitative data and demonstrates the convergence of themes. The table highlights how the focus group interviews enabled participants to provide in-depth information on survey topics. For example, when discussing diabetes self-care struggles, focus group participants identified financial difficulties, lack of discipline and lack of education as key barriers. In contrast, the survey focused more narrowly on diabetes self-care activities related to healthy eating, physical activity, and medication use. Similarly, while survey questions about TM focused primarily on text message frequency, focus group discussions noted importance of timing (e.g. avoiding messages in the middle of the workday).

Limitations
It is possible that the sample of patients selected for this mixed-methods study were not representative of the target population of medically underserved African-American patients with uncontrolled diabetes. The level of diabetes control among participants was not assessed. However, all participants had experience living with diabetes, were African American, and lived in medically underserved areas in the Mid-South. Previous research indicates that diabetes is poorly controlled in approximately 50% of this population [5]. In addition, the study is potentially subject to Berkson's bias, since the sample was not taken from the general population, but from a subpopulation served by a healthcare delivery system. However, this healthcare delivery system serves over 70% of African-American patients in the targeted medically underserved ZIP codes.

Survey findings Qualitative findings Mixed methods meta-inferences
Barriers to Self-care improvement areas: Healthy

Discussion
Many studies have explored the use of health-related text messages to influence patient selfcare decisions and health behaviors, but few studies have specifically assessed the types of messages preferred by African-American patients with chronic illness in medically underserved areas [8,13]. This study is among the first to show these patients were most interested in receiving text messages that are personalized and tailored to their interests and self-care needs. Participants' interest in handling their disease management and improving their health outcomes is consistent with the concept of ownership of diabetes as described by Shearer and colleagues [25]. We found that a majority of the medically underserved patients in this study sample already use mobile phones for personal text messaging and are highly interested in receiving health-related text messages from their doctor's office. Furthermore, both quantitative and qualitative findings demonstrated that medically underserved African-American patients with diabetes want support in managing their own health and believe that a TM program could help them overcome their diabetes self-care struggles.
Our study's finding that patients prefer tailored TM programs that are personalized to individual needs is consistent with research indicating that tailored TM programs are more likely to motivate patients [26][27]. For example, Horner and colleagues found that a one-sizefits-all approach led to frustration and disengagement as study participants found the messages impersonal and unrelated to their specific self-care struggles. The current study is also consistent with studies showing that patients are more likely to read messages specific to their health goals and, therefore, more likely to succeed in improving their health [26,28]. Previous studies demonstrated that personalized information can be gathered and utilized to create a TM database for each participant that is updated in real-time to reflect patients' changing preferences over time. For example, Job and colleagues checked in with participants in a weight loss program at 12 weeks, updated their preferences and progress, and changed content, leading to increased patient accountability and engagement with the TM program [27]. Similarly, our study showed that many patients want to be able to change TM preferences over time. Future studies are needed to establish ideal levels of personalization and best approaches for incorporating patient preferences over time.
The current study found high levels of individual variability in patient preferences regarding frequency and timing of text messages, consistent with previous research suggesting that messages sent too frequently or at inconvenient times can elicit negative reactions. For example, Smith and colleagues found that adolescents who received obesity-related text messages three times a week developed negative feelings about the TM program and started avoiding or ignoring messages [26]. However, studies in other populations have shown high receptivity to daily text messages [15,29]. Recent studies that have tailored text message frequency and timing to participant preference have demonstrated high patient engagement over time [27]. Previous studies indicate that patient preferences for message timing may depend on the message type. A large and varied TM database may also aid participant engagement as objections to frequent messages may be due to boredom from similar messages or being accustomed to messages making them less effective [8,29]. Frequent check-ins and updating of participant preferences may also help negate the apathy that results from repeated messages [11,27].
Our study provides evidence that engaging diabetes patients in medically underserved areas could benefit from TM content being tailored to individual self-care struggles and goals. Our findings are consistent with research suggesting that text messages that are tailored to individual challenges and goals may do more to remind patients of their desire for change and improvement [26]. We found that patients may want to identify secondary or even tertiary health goals in order to be able to focus on more than one goal at a time. This is especially applicable to diabetes management as patients may struggle with more than one area of care. Personalization and the ability to change TM content is important because individual needs vary and are likely to change over time based on disease stability, progression, or symptoms.
Lastly, the current study demonstrates that text message tone is critically important for effectively engaging African-American diabetes patients in medically underserved areas. Similar to Smith and colleagues, we found that patients disliked messages that had a negative tone or were perceived as shaming [30]. Our findings are consistent with previous studies suggesting that patients like practical and actionable messages that keep them on track and accountable [27]. In addition, the current study suggests that motivational and inspirational messages may be particularly effective in encouraging continued patient engagement, regardless of whether they relate to personal health goals. Similarly, text messages that incorporate encouraging images or emoticons [26] may provide a more natural and casual exchange, thus enhancing engagement.

Conclusions
Primary care practices and health systems seeking to develop patient-centered TM programs for medically underserved minority patients with chronic diseases should strive to tailor healthrelated text messages to meet participants' personal health needs and interests. Assessing patients' self-care struggles, health goals, mobile phone use, perceptions of TM, and desired outcomes of the TM program during enrollment in a TM program can lead to a customized TM program that supports active participation, patient satisfaction, and responsible decisionmaking for better health. This study suggests that tailored TM programs may provide a patientcentered, effective, and well-received approach for supporting self-care for underserved African-American diabetic patients. Our results suggest that the most effective TM programs should have the capacity to personalize messages to meet the needs of individual participants.

Appendices
interview. The requirement to secure a signed consent form is waived under 45 CFR 46.117(c) (2). Willingness of the subject to participate will constitute adequate documentation of consent. Your application has been determined to comply with proper consideration for the rights and welfare of human subjects and the regulatory requirements for the protection of human subjects. This letter constitutes full approval of your application (version 1.0) and surveys, stamped approved by the IRB on 06/22/2015 for the above referenced study. . Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work.
Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.