A Mobile Health Intervention for Inner City Patients with Poorly Controlled Diabetes: Proof-of-Concept of the TExT-MED Program

Abstract

Objective:

Numerous mobile health (mHealth) interventions are being developed to aid in management of complex chronic medical conditions. However, the acceptance of mHealth programs by low-income, bilingual populations has not yet been evaluated. The Trial to Examine Text-based mHealth for Emergency department patients with Diabetes (TExT-MED) program is a text message–based mHealth program designed specifically for resource-poor patients with diabetes. We conducted a prospective proof-of-concept trial to assess satisfaction and preliminary effectiveness of the TExT-MED program.

Research Design and Methods:

A consecutive sample of adult patients in the emergency department with diabetes and a text message–capable mobile phone was enrolled in the TExT-MED program. Participants received three text messages daily for 3 weeks in English or Spanish in the following domains: educational/motivational, medication reminders, healthy living challenges, diabetes trivia, and links to free diabetes management tools.

Results:

Twenty-three patients with diabetes (median hemoglobin A1c, 8.9%) were enrolled in TExT-MED. In the week before TExT-MED, 56.5% of subjects reported eating fruits/vegetables daily versus 83% after, 43.5% reported exercising before versus 74% after, and 74% reported performing foot checks before versus 85% after. Self-efficacy, measured by the Diabetes Empowerment Scale—Short Form, improved from 3.9 to 4.2. Scores on the Morisky Medication Adherence Scale improved more dramatically from 3.5 to 4.75. Ninety percent of participants indicated they would like to continue the program, and 100% would recommend the program to family or friends.

Conclusions:

This pilot trial of the TExT-MED program demonstrated increased healthy behaviors, improved diabetes self-efficacy and medication adherence, and received excellent satisfaction scores in resource-poor, inner city patients with diabetes.

Introduction

D iabetes is an epidemic that disproportionately affects minorities and the underserved^1

–8 (see Table 17 in Kung et al.⁷). These individuals often lack access to primary care, and many overuse the emergency department (ED) for both acute and chronic care.^9
–11 In order to help bridge the gap between urgent visits to the ED and connection to follow-up care, we developed a mobile health (mHealth) text message–based intervention to improve healthy behaviors, medication adherence, and self-efficacy in low-income patients with diabetes.

mHealth refers to the utilization of mobile phones and other personal mobile devices as a platform on which to provide public health or medical interventions. Currently, the term encompasses a broad array of applications and technologies ranging from highly interactive Web-based social media platforms to avatars representing a patient's current and predicted future health status, to simple broadcast text message systems. Programs such as “Text4Baby,” a text-based intervention designed to promote maternal and child health, and other text message–based mHealth programs have been developed as adjunct management tools for patients with both acute and chronic diseases, including human immunodeficiency virus and diabetes.^12

–15 However, most existing diabetes mHealth programs are highly complex, involve bidirectional remote monitoring of glucose levels, and require expensive smartphones to access applications (apps) and the mobile Web. National data confirmed by our local pilot work show that 80% of Spanish-speaking patients have and use short message service message–capable mobile phones, but very few have costly smart phones.¹⁶

To date no mHealth interventions have been reported describing use in a bilingual, underserved population in the United States with diabetes. mHealth has the potential to be a significant breakthrough for resource-poor, Spanish-speaking patients if it is low cost, accessible, and easy to use. For mHealth to be effective for resource-poor patients it must be technologically familiar (short message service messages) and easily scalable. To this end, we have developed the Trial to Examine Text-based mHealth for Emergency department patients with Diabetes (TExT-MED) program. TExT-MED is designed to motivate, educate, and empower inner city patients with diabetes. This is a culturally sensitive, low-cost, bilingual, evidence-based, and largely unidirectional text message–based mHealth program requiring only a basic mobile phone to use. We conducted a prospective proof-of-concept study to assess satisfaction and preliminary effectiveness of the TExT-MED program.

Research Design and Methods

Study setting and patient selection

This prospective study was conducted in the ED at the Los Angeles County Hospital at the University of Southern California, which is a public safety-net hospital, and the ED has over 170,000 patient visits annually. A trained research assistant approached a consecutive sample of patients with a documented medical history of diabetes in 8-h blocks for a 7-day period. The research assistant explained the purpose of the study, described the TExT-MED intervention, and obtained written informed consent. Subjects were eligible for inclusion if they were (1) ≥18 years old, (2) had diabetes, (3) had a text message–capable mobile phone, (4) knew how to receive text messages, and (5) spoke and read English or Spanish. Exclusion criteria for the trial included (1) overt psychosis, (2) inability to provide informed consent, and (3) pregnancy. Enrolled subjects were given a $50 subject participation stipend to cover the potential costs associated with the text messages and the follow-up visit. Institutional Review Board approval for this study was obtained prior to study initiation.

Study procedure and data collection

TExT-MED is a technologically simple solution using unidirectional daily text messages used to deliver health messages and triggers to promote knowledge, healthy eating, exercise, self-efficacy, and medication adherence. In this study, subjects received three text messages daily (9 a.m., 12 p.m., and 6 p.m.) on their mobile phone for 3 weeks. Messages were delivered by a fully automated, Web-based program developed by EPG Technologies (Van Nuys, CA). EPG staff have undergone National Institutes of Health Human Subjects training, and their program met Institutional Review Board and HIPAA privacy and security rules. Messages were sent in English or Spanish according to patient preference. The full schedule of messages organized by message type is detailed in Figure 1. The first text message was delivered while the subject was still in the ED to ensure technical success and read: “Thank you for participating in the TExT-MED Program. Together, we will conquer your diabetes.” Subsequent text messages were divided into the following categories:

1. Educational and motivational (two per day). These messages were largely adapted from the National Diabetes Education Program (NDEP). We chose to adapt materials from the NDEP to text message format as they are in the public domain, bilingual, non-branded, written at a 5^th grade reading level, and available in both English and Spanish.¹² Information from the NDEP falls into the following content areas: Blood Glucose Control, Blood Pressure, Cholesterol, Controlling Diabetes, Foot Care, Healthy Eating, Heart Disease, Physical Activity, Recipes, Social Support, and Teens and Diabetes. Most NDEP key points are already written in short statements that were easily translatable to 160 7-bit character text messages. Through an iterative process involving physician specialists in endocrinology and emergency medicine and a certified diabetes educator, we selected messages representing all NDEP content areas focusing on specific areas of knowledge deficit we have identified in previous studies with Los Angeles County Hospital at the University of Southern California ED patients (e.g., signs of high and low blood sugar, wound care management, and elements of a healthy diet).¹⁸ Sample educational and motivational text messages included “Having diabetes can lead to a heart attack or stroke—but it doesn't have to,” “Controlling your blood glucose, blood pressure, and cholesterol—can mean a longer and healthier life,” and “Eat more fruits, vegetables, beans, and whole grains and less salt and fat.”

2. Medication reminders (three per week). These messages were developed to provide a trigger that would result in increasing patient adherence with prescribed medications. The message read: “Remember to take your medications every day as prescribed by your doctor.”

3. Healthy living challenges (two per week). These messages were designed to provide patients a concrete, attainable goal for the day. The challenges were based on healthy living principles set forth by the Amereican Diabetes Association and the NDEP. Sample challenge messages included “Challenge! Don't drink any soda or juice today. Only drink water or milk,” “Challenge! Eat one meal today that has only vegetables,” and “Challenge! Look at food labels and find a snack that has less than 100 calories.”

4. Trivia (one per week). These messages were derived from NDEP materials, designed to educate and motivate, and presented in question form. This was the only portion of the TExT-MED program that gave the participants an option to reply to received messages. Sample trivia texts included “What is a normal blood sugar level?”

5. Phone links (one per week). These messages provided patients with a phone number to call if they wanted to receive a free gift to help them manage their diabetes. This provided them with another way to interact with us that was not text message based. Sample offered gifts included a diabetes book and a pedometer.

FIG. 1. Weekly schedule of text message types.

	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
9 a.m.	Medication Reminder	Healthy Living Challenge	Medication Reminder	Trivia	Medication Reminder	Healthy Living Challenge	Phone Link
12 p.m.	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate
6 p.m.	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate	Educate/Motivate

Measures

Subjects provided demographic, clinical, and biometric data at the index ED visit. Hemoglobin A1c (HbA1C) was measured at the index visit using the Afinion AS-100 (Axis-Shield PoC AS, Oslo, Norway) capillary point-of-care HbA1C meter, which reports National Glycohemoglobin Standardization Program–aligned values. At both the index visit and upon completion of the 3-week trial, patients were asked questions about their engagement in healthy behaviors over the preceding 7 days (specifically, diet, exercise, and foot care), completed the Diabetes Knowledge Test (DKT) developed by the Michigan Diabetes Research Training Center, the Diabetes Empowerment Scale—Short Form (DES-SF), and the eight-item Morisky Medication Adherence Scale (MMAS).^19

–22 In addition, at the exit interview subjects were asked if they completed each of the six individual challenges and were then given a series of Likert-type questions assessing various aspects of the TExT-MED program including (1) perceived efficacy of the program, (2) opinions on the different types and frequency of text messages, (3) the appropriateness of the level of the text message content, and (4) their willingness to continue with or recommend the TExT-MED program to family or friends.

Analysis

The analysis is largely descriptive with point estimates and appropriate 95% confidence limits displayed. No formal statistical test of hypothesis was conducted. Analyses were performed with Stata version 10.1 (StataCorp, College Station, TX). The DKT, DES-SF, and MMAS were scored according to previous reports.^19

–22 Likert-type questions were dichotomized into a positive response if “strongly agree” or “somewhat agree” was selected and a negative response if “neutral,” “somewhat disagree,” or “strongly disagree” was selected. We did not conduct a formal sample size calculation for this proof-of-concept study. Rather, we chose enroll patients in the study over a 1-week period based on locally available resources.

Results

Participants

In the 1-week enrollment period, we identified 83 ED patients with diabetes. Of these, six were excluded because of inability to speak English or Spanish, and three were excluded because of inability to provide informed consent secondary to altered mental status. Of the remaining 74 patients, 51 did not have a text message–capable phone or indicated they did not know how to receive text messages. This left 23 patients eligible, and all 23 of these were enrolled. Mean age of subjects was 45 years, 60% were male, 70% were Latino, and 30% opted to received messages in Spanish. Further characteristics of the study population are detailed in Table 1.

Table 1.

Characteristics of Study Participants

Demographic	Value
Number	23
Age (years) [mean (SD, range)]	45.4 (7.5, 36–63)
Gender (male)	61%
Ethnicity
Latino/Hispanic	70%
Non-Latino/non-Hispanic	30%
Language preference
English	70%
Spanish	30%
BMI (kg/m²) [mean (SD, range)]	32.1 (5.4, 23.4–45.4)
HbA1C [median (SD, range)]	8.9% (2.2, 5.7–14.1%)
Diabetes type
Type 1	9%
Type 2	70%
Not sure	21%

BMI, body mass index; HbA1C, hemoglobin A1c.

Main results

Median HbA1C on enrollment was 8.9%, indicating poor glycemic control. At the end of the 3-week trial, 20 of the 23 (87.0%) subjects returned to complete the exit interview and follow-up questionnaire. Globally, the TExT-MED program was associated with improved health behaviors: in the week before TExT-MED, 56.5% of subjects reported eating fruits/vegetables daily versus 83% after, 43.5% reported engaging in any form of exercise before versus 74% after, and 74% reported performing any foot checks in the week before the intervention versus 85% after. Seventy percent of patients performed at least four of the six health challenges. Self-efficacy, as measured by the DES-SF, improved modestly during the trial (from 3.9 to 4.2). Scores on the MMAS improved more dramatically, from 3.5 to 4.75. When the aspects of the program requiring bidirectional interactions were examined, only two patients called in response to the phone link messages, and only eight patients responded to any of the trivia messages. Scores on the DKT did not change meaningfully (from 8.7 at baseline to 8.4 post-intervention).

Satisfaction with TExT-MED program

Overall, subjects reported high levels of satisfaction with the TExT-MED program. Specifically, 90% of participants indicated they would personally like to continue the program, and 100% would recommend the program to family or friends. Similarly, 90% of participants agreed that text messages were an effective way of communicating, that the messages were understandable, and that they liked receiving them every day of the week. Table 2 gives further details of the satisfaction responses.

Table 2.

Reported Satisfaction with the TExT-MED Program

Statement	Participants who agreed (n=20)
Using text messages is a good way to teach	90%
I have enjoyed the TExT-MED program	90%
I was able to understand all the messages	100%
The number of text messages per day was just right	70%
I liked getting text messages on the weekdays	80%
I liked getting text messages on the weekends	80%
The text messages came at times that were good for me	95%
I would like to continue receiving text messages	90%
I would recommend this program to a friend or family member	100%
I was motivated by the challenges	80%
I liked the weekly trivia questions	80%
The medication reminders helped me remember to take my meds	90%
The number of medication reminders was just right	90%

TExT-MED, Trial to Examine Text-based mHealth for Emergency department patients with Diabetes.

Conclusions

mHealth technologies have garnered significant interest and enthusiasm from public health advocates, policy makers, and the technology industry. The range of mHealth interventions being developed and tested is wide and includes simple unidirectional text-based systems to virtual worlds in which current behaviors are projected forward in time onto avatars to demonstrate the future effects of present lifestyle choices. The clinical and public health efficacy of mHealth is only beginning to be rigorously assessed, but initial results are encouraging.^12

–15 However, the effect of mHealth programs on resource-poor, inner city populations is wholly unknown. The TExT-MED program was designed specifically for resource-poor patients with the belief that a successful mHealth intervention in this population should use a basic technology that they already have and are familiar with (text messages) and that the information provided should be simple and easy to understand. To be a sustainable on a system-wide scale and widely disseminated, it should be low cost, easy to implement, and readily adaptable.

Patients with diabetes who use the ED for care may represent an excellent target for brief, scalable, diabetes-specific interventions such as TExT-MED. Over the last decade, there has been a 5.6% relative annual increase in the proportion of ED visits that were diabetes-related.¹¹ Recent studies show that 11–27% of ED patients suffer from diabetes and that poor glycemic control is common (in our sample, median HbA1C was 8.9%).^9,10 Furthermore, it has been reported that ED patients with diabetes, particularly those seen in a public safety-net system, are often poorly educated, have limited English language skills, are in poor physical health, and have insufficient access to traditional medical homes.¹⁰ Unfortunately, the ED is overcrowded and ill prepared to provide chronic disease management for these patients. Our proof-of-concept study in an urban ED shows that TExT-MED was highly accepted as patients were engaged and reported overwhelmingly positive satisfaction scores. It is unusual to have 100% of eligible patients agree to participate in a study, but we feel this level of enthusiasm speaks to the enormous potential that mHealth has to impact this population. The intervention also appeared to be effective in our sample as healthy behaviors, diabetes self-efficacy, and medication adherence increased. It is interesting that these positive behavior changes occurred without an increase in diabetes-specific knowledge, suggesting that improved knowledge may not be required to impact outcomes. In addition, all patients completed at least four of the six provided healthy living challenges. The trivia texts and phone links were less engaged (only two patients engaged in the phone link), which supports the concept that unidirectional text message programs may be more readily accepted in this patient population.

Ours is the first study we are aware of describing a bilingual mHealth program for low-income patients with diabetes. Other groups have conducted studies aimed at improving diabetes control via text-based mHealth in different patient populations. Franklin et al.¹² conducted a study of 92 pediatric patients with type 1 diabetes in Scotland, sending one or two text messages daily consisting of personalized insulin injection regimens, healthy eating tips, and exercise goals for the week; the “Sweet Talk” intervention showed improved HbA1C at 1 year only when combined with intensive insulin therapy, but “Sweet Talk” alone improved diabetes self-efficacy and medication adherence. Kollman et al.¹⁴ published results of a pilot study requiring patients to send daily blood glucose values via text messages to a central monitoring center, and patients would receive personalized text messages in the evening based on reported values. Although there was a trend towards decreased HbA1C and average blood glucose, their system is highly involved and bidirectional and driven primarily by active subject engagement. Kim¹³ designed a randomized control trial of 60 patients with type 2 diabetes in Korea in which patients in the intervention group sent daily blood glucose values to a Web server and based on these data would receive individualized management strategies via text messages; although effective in decreasing HbA1C by 1.15% at 12 weeks, this system was also bidirectional, involved using the Internet, and required ongoing physician involvement to create management strategies.

A relative strength of this study and the TExT-MED program is that it demonstrates the potential of simple, scalable, unidirectional text messaging to encourage positive behavior change. Moreover, the study was successful despite being conducted in the ED, a clinical area not typically associated with chronic disease management. This study is limited by its small sample size, uncontrolled design, and short follow-up period, which prevented us from evaluating for a measurable change in HbA1C. Another important consideration that may limit the generalizability of these findings is that some patients may incur a cost associated with the receipt of text messages. Researchers and program designers should be cognizant of this when these programs are widely distributed. Of note is that in our study all subjects already had unlimited text message plans prior to enrollment. Additionally, in our study 31% of subjects with diabetes had a text message–capable phone and knew how to receive text messages. This limits the potential reach of similar mobile phone–based interventions. However, 31% of patients with diabetes still represents over 8 million individuals in the United States who could benefit from mHealth.

In conclusion, our proof-of-concept text-based program for low-income, resource-poor patients with diabetes resulted in excellent satisfaction and modest improvement in self-care behaviors after 3 weeks. These results suggest TExT-MED has the potential to impact a hard-to-reach patient population. We are conducting a larger, randomized controlled trial to confirm these findings.

Footnotes

Acknowledgments

This study was supported by a grant from the McKesson Foundation as a part of their mobilizing for health initiative. S.A. conceived of the project, helped design the study, participated in the development of the text message curriculum, and contributed to the writing and revision of the manuscript. A.L.P. helped design the study, participated in the development of the text message curriculum, and contributed to the writing and revision of the manuscript. C.A. helped design the study, participated in the development of the text message curriculum, and assisted with collection of data. M.M. helped design the study, performed all statistical analyses, and helped write and revise the manuscript.

Author Disclosure Statement

The authors have no conflicts of interest and no financial or other ties to report.

References

Centers for Disease Control and Prevention: Survey Data. 2003 Hyattsville, MD: U.S. Department of Health and Human Services, 2003.

Centers for Disease Control and Prevention: Survey Data. 2004 Hyattsville, MD: U.S. Department of Health and Human Services, 2004.

Centers for Disease Control and Prevention: Survey Data, 2005. Hyattsville, MD: U.S. Department of Health and Human Services, 2005.

Centers for Disease Control and Prevention: Survey Data, 2006. Hyattsville, MD: U.S. Department of Health and Human Services, 2006.

Centers for Disease Control and Prevention: Survey Data, 2007. Hyattsville, MD: U.S. Department of Health and Human Services, 2007.

Centers for Disease Control and Prevention: Survey Data, 2008. Hyattsville, MD: U.S. Department of Health and Human Services, 2008.

Kung

, Hoyert

, Xu

, Murphy

. Deaths: final data for 2005. Natl Vital Stat Rep, 2008; 56,10:1–120.

Lawrence

. Diabetes in Hispanic American youth. Diabetes Care, 2009; 32,Suppl 2:S123–S132.

Menchine

, Arora

, Camargo

, Ginde

. Prevalence of undiagnosed and suboptimally controlled diabetes by point-of-care HbA1c in unselected emergency department patients. Acad Emerg Med, 2011; 18:326–329.

10.

Menchine

, Vishwanath

, Arora

. Prevalence, health and demographics of emergency department patients with diabetes. West J Emerg Med, 2010; 11:419–422.

11.

Menchine

, Wiechmann

, Peters

, Arora

. Trends in diabetes-related visits to US EDs from 1997 to 2007. Am J Emerg Med, 2011 May 11[Epub ahead of print]10.1016/j.ajem.2011.02.028.

12.

Franklin

, Waller

, Pagliari

, Greene

. A randomized controlled trial of Sweet Talk, a text-messaging system to support young people with diabetes. Diabet Med, 2006; 23:1332–1338.

13.

Kim

. A randomized controlled trial of a nurse short-message service by cellular phone for people with diabetes. Int J Nurs Stud, 2007; 44:687–692.

14.

Kollmann

, Riedl

, Kastner

, Schreier

, Ludvik

. Feasibility of a mobile phone-based data service for functional insulin treatment of type 1 diabetes mellitus patients. J Med Internet Res, 2007; 9:e36.

15.

Lester

, Ritvo

, Mills

, Kariri

, Karanja

, Chung

, Jack

, Habyarimana

, Sadatsafavi

, Najafzadeh

, Marra

, Estambale

, Ngugi

, Ball

, Thabane

, Gelmon

, Kimani

, Ackers

, Plummer

. Effects of a mobile phone short message service on antiretroviral treatment adherence in Kenya (WelTel Kenya1): a randomised trial. Lancet, 2010; 376:1838–1845.

16.

Livingston

. Latinos, Digital Technology. PEW Internet & American Life Project. February 9, 2011. www.pewhispanic.org/files/reports/134.pdf. 2012 February 6.

17.

Centers for Disease Control and Prevention: What is the NDEP. ndep.nih.gov/. 2011 August 8.

18.

Arora

, Marzec

, Gates

, Menchine

. Diabetes knowledge in predominantly latino patients and family caregivers in an urban emergency department. Ethn Dis, 2011; 21:1–6.

19.

Fitzgerald

, Funnell

, Hess

, Barr

, Anderson

, Hiss

, Davis

. The reliability and validity of a brief diabetes knowledge test. Diabetes Care, 1998; 21:706–710.

20.

Anderson

, Fitzgerald

, Gruppen

, Funnell

, Oh

. The Diabetes Empowerment Scale—Short Form (DES-SF) Diabetes Care, 2003; 26:1641–1642.

21.

Morisky

, Ang

, Krousel-Wood

, Ward

. Predictive validity of a medication adherence measure in an outpatient setting. J Clin Hypertens (Greenwich), 2008; 10:348–354.

22.

Morisky

, Green

, Levine

. Concurrent and predictive validity of a self-reported measure of medication adherence. Med Care, 1986; 24:67–74.