Abstract
Free full text
Current Science on Consumer Use of Mobile Health for CVD Prevention
Part 1: Background
Although mortality for cardiovascular disease (CVD) has declined for several decades, heart disease and stroke continue to be the leading causes of death, disability and high healthcare costs. Unhealthy behaviors related to CVD risk (e.g., smoking, sedentary lifestyle, and unhealthful eating habits) remain highly prevalent. The high rates of overweight and obesity, type 2 diabetes mellitus (T2DM), the persistent presence of uncontrolled hypertension, lipid levels not at target, and approximately 18% of adults who continue to smoke cigarettes pose a formidable challenge for achieving improved cardiovascular health.1,2 It is apparent that the performance of healthful behaviors related to the management of CVD risk factors has become an increasingly important facet of the prevention and management of CVD.3
In 2010, the American Heart Association (AHA) made a transformative shift in their strategic plan and added the concept of cardiovascular (CV) health.2 To operationalize this concept, the Association targeted four health behaviors in the 2020 Strategic Impact Goals, reduction in smoking and weight, healthful eating, and promoting regular physical activity. Three health indicators also were included: glucose, blood pressure (BP), and cholesterol. Based on the AHA Life’s Simple 7 metrics for improved CV health, less than one percent of adults in the US follow a healthful eating plan, only 32% have a normal body mass index (BMI) and over 30% have not reached the target levels for lipids or BP. NHANES data revealed that persons who met six or more of the CV health metrics had a significantly better risk profile (hazard ratio of 0.49 for all-cause mortality), compared to individuals who had achieved only one metric or less.2 The studies reviewed in this statement targeted these behaviors (e.g., smoking, physical activity, healthful eating, and maintaining a healthful weight) and CV health indicators (e.g., blood glucose, lipids, BP, BMI) as the primary outcomes in the clinical trials testing mobile health (mHealth) interventions.
eHealth, or digital health, is the use of emerging communication and information technologies, especially the use of the internet to improve health and health care.4 mHealth, a subsegment of eHealth, is the use of mobile computing and communication technologies (e.g., mobile phones, wearable sensors) for health services and information.4,5 mHealth technology uses techniques and advanced concepts from an array of disciplines, e.g., computer science, electrical and biomedical engineering, and medicine and health-related sciences.6 Mobile devices that permit collection of data in real time are increasingly ubiquitous, enabling researchers to assess multiple behaviors in various contexts and thus inform the development of interventions to prompt behavior change. Technology-supported behavioral health interventions are designed to engage individuals in health behaviors that prevent or manage illness, and they have led to fundamental changes in health practices.7 In addition to permitting more frequent and convenient community-based assessment of health parameters, these technology-mediated tools support exchange of health information among consumers and between consumers and health providers, enable health decision-making, and encourage positive health behaviors including self-management and health promotion.8,9 Consequently, mobile health technologies are becoming more prevalent, and their use will continue to grow,10 consistent with the Institute of Medicine’s (IOM) call to increase the design and testing of health technologies.11
The ubiquity of mobile devices presents the opportunity to improve health outcomes through the delivery of state-of-the-art medical and health services with information and communication technologies.12 Due to their diverse capabilities and advanced computing features, smartphones are often considered pocket computers.6 In addition to these devices that can inform and communicate, there are wearable sensors, which can be worn for short or extended periods and monitor activity or physiological changes (e.g., exercise, heart rate, sleep). These sensors can provide data in real time or save the data to a device for later uploading and review.
The Food and Drug Administration (FDA) has a public health responsibility to oversee the safety and effectiveness of medical devices. However, these apply only to applications (apps) that are accessory to regulated medical devices (e.g., ones that diagnose a condition). Many mobile apps are not medical devices, meaning they do not meet the definition of a device under section 201(h) of the Federal Food, Drug, and Cosmetic Act (FD&C Act)), and FDA does not regulate them. Some mobile apps may meet the definition of a medical device but because they pose a lower risk to the public, FDA intends to exercise enforcement discretion over these devices. Most of the mHealth apps on the market at this time fit into these two categories.13,14
Numerous innovations in health information technology are empowering individuals to assume a more active role in monitoring and managing their chronic conditions and therapeutic regimens as well as their health and wellness.15 These advances are increasingly accepted by the public.16 Unlike the initial digital divide that placed computer use and internet access beyond the reach of many older, disabled, and low-income individuals, mobile devices have been widely adopted across demographic and ethnic groups especially those most in need of health behavior interventions.17,18 This trend is confirmed in the 2014 statistics from the Pew Research Center’s Internet and American Life Project that showed that 81% of households with an income above $75,000/year owned a smartphone and nearly half (47%) of those with an annual household income less than $30,000 owned a smartphone.19 The highest smartphone ownership was among Hispanic and African Americans, at 61% and 59%, respectively. Of those with phones who use the internet, 34% mostly use their phones, rather than a desktop or laptop, to access online programs.20
Mobile devices offer great promise for improving the health of the populous. Most smartphones include basic functionalities, e.g., video streaming, email, internet access and high quality imaging. These developments in wireless technology and the shift to mobile devices are demanding a re-examination of technology as it currently exists within the healthcare infrastructure.6 However, the pace of science in evaluating these apps is incongruent with the business and industry sectors and the consumer demands. There are concerns that health-promoting smartphone apps being developed fail to incorporate evidence-based content and that rigorous testing to provide efficacy data is trailing behind their adoption.21–24 However, a systematic review of the literature suggests a positive impact of consumer health informatics tools on select health conditions, e.g., there were intermediate outcomes such as knowledge, adherence, self-management and change in behaviors related to healthful eating, exercise and physical activity, but not obesity.25 Another review suggests that smartphone apps are useful tools at the point of care and in mobile clinical communication as well as in remote patient monitoring and self-management of disease.26
Recent papers have reviewed the latest technological advances in digital social networks related to health27 and wireless devices for cardiac monitoring.28 What is missing in the scientific literature is a report on the health-related mobile technologies specifically focused on CVD prevention. In particular, it is important to investigate the degree to which these CVD focused technologies include best content and have been evaluated for their effectiveness. In the absence of such data, clinicians may be hesitant to recommend or endorse any program to their patients and thereby potentially miss an opportunity to improve their engagement in healthful behaviors.
The aims of this scientific statement are to: 1) review the literature on mHealth tools available to the consumer in the prevention of CVD (e.g., dietary self-monitoring apps, physical activity and BP monitors), 2) provide the current evidence on the use of the vast array of mobile devices such as use of mobile phones for communication and feedback, smartphone apps, wearable sensors or physiologic monitors that are readily available and promoted to the public for monitoring their health, and 3) provide recommendations for future research directions. The goal is to provide the clinician and researcher a review of the current evidence on using mHealth tools and devices when targeting behavior change and CV risk reduction as well as improved CV health. The paper is divided into sections by the behaviors or health indicators included in the AHA’s Simple 7 program, e.g. achieving a healthful weight, improving physical activity, quitting smoking, achieving blood glucose control, managing BP and also lipids to achieve target levels. Within each section, the recent evidence for studies using mHealth approaches are reviewed, the gaps identified and directions for future research are provided.
While the majority of studies reported on the use of mobile devices, e.g., basic mobile phones that support the use of text messaging (SMS) or smartphones that provide internet access, several reported on interventions delivered via the internet, e.g., studies reporting on increased physical activity or blood pressure management. The writing group made the decision to include these studies as there is an increasingly greater proportion of people accessing the internet via mobile devices, as noted in a Pew report in February 2014, 68% of adults access the internet with mobile devices.29 This figure has likely increased in the past year. Moreover, in some of the designated areas of cardiovascular risk, there were few studies reporting on the use of mHealth supported interventions.
Part 2: Review of the scientific literature on mHealth tools related to CVD prevention
Search strategy.
We conducted a literature search that included the following terms: mHealth; mobile health; mobile phone; mobile device; mobile technology; mobile communication; mobile computer; mobile PC; cell phone; cellular phone; cellular telephone; handheld computer; handheld device; handheld technology; handheld PC; hand held computer; hand held device; hand held technology; hand held PC; tablet device; tablet computer; tablet technology; tablet PC; smartphone; smart phone; iPad; Kindle; Galaxy; iPhone; Blackberry; iPod; Bluetooth; short message service; SMS; pocket PC; pocketPC; PDA; personal digital assistant; Palm Pilot; Palmpilot; smartbook; mobile telephone; messaging service; MP3 player; portable media player; podcast; email; e-mail; electronic mail; electronic message. Search terms used within the technology or clinical topic (e.g., diabetes) groups were divided with OR, while the search terms between the technology and clinical topic were connected with AND. Within each subsection the key terms used in the search for a given clinical topic are identified. The search was limited to the last 10 years (2004 – 2014) and studies reported in the English language. We limited our review to studies enrolling adults, except for smoking cessation where we included adolescents. We included studies conducted in the U.S. and in developed countries. We also briefly discuss key systematic reviews or meta-analyses in each topic area, except in management of dyslipidemia.
Use of mHealth to Improve Weight Management
Obesity causes or contributes to a myriad of physical and mental health conditions, such as CVD, T2DM and depression, which either individually or collectively, represent the leading causes of morbidity and mortality in the US.30–32 Over 35% of US adults ages ≥20 years are obese33 and more than 1 in 4 Americans have multimorbidity,34,35 which is associated with high healthcare use and costs, functional impairment, poor quality of life, psychological distress, and premature death.36–40 Sustained weight loss of 3–5% can delay or possibly prevent T2DM41,42 and significantly improve CVD risk factors (e.g., abnormal glucose, elevated blood pressure).43–46 However, effective treatments for obesity that are accessible to consumers, affordable for diverse socioeconomic groups, and scalable at a population level are lacking.
The 2013 obesity treatment guideline by the AHA and the American College of Cardiology (ACC) and Obesity Society recommended that clinicians advise overweight and obese individuals who would benefit from weight loss to participate for ≥6 months in a comprehensive lifestyle program characterized by a combination of a reduced-calorie intake, increased physical activity, and behavioral strategies.47 The guideline panelists found evidence of moderate strength supporting the efficacy of electronically-delivered, comprehensive lifestyle programs that include personalized feedback from a trained interventionist, defined as programs delivered to participants by internet, email, mobile texting, or similar electronic means. Therefore, it was recommended that electronically-delivered interventions are an acceptable alternative to in-person interventions, although it was recognized that the former may result in smaller weight loss than the latter.
Use of mHealth in Weight Management Interventions.
This review is limited to technology-supported lifestyle behavioral interventions for weight loss. Readers are referred to numerous systematic reviews of more traditional internet-, email-, and telephone-based lifestyle interventions for weight loss.48–53 Overall, weight management interventions have employed a range of mobile technologies,50,54–58 including short message service (SMS), smartphone applications, handheld personal digital assistants (PDAs), and interactive voice response (IVR) systems.56,59,60 Numerous network-connected devices have also been used,50,54 including e-scales and wireless physical activity monitoring devices61. Use of mobile devices and their functionality (e.g., SMS and multimedia messaging service [MMS], mobile internet, and software apps in weight loss interventions have increased exponentially in recent years. In this section, we focus on the latest evidence on mobile technology interventions for weight loss.
With few exceptions,62 most interventions have used a single, pre-determined technology channel and did not give participants the option of either choosing between channels or using multiple channels simultaneously (which has become commonplace for commercial applications). Most technologies have been created in research settings, although at least one published study used a commercially available app.61 The majority of these trials were primarily focused on efficacy testing and it was unclear whether these interventions used strategies designed to promote user engagement (e.g., employing established design principles, conducting usability testing, and/or undergoing iterative development and testing). Additionally, a key translational challenge is that many commercial apps have not been tested empirically, and many apps with empirical data are not commercially available.
Review of evidence for efficacy of mHealth-based weight loss interventions.
We conducted an electronic literature search using Medline (PubMed), CINAHL, and PsychInfo in June 2014 and extended back to 2004. Search terms for this topic included: Overweight; obese; obesity; body mass; adiposity; adipose; weight loss; weight gain. Only original studies with human subjects with a primary outcome of weight loss and published in English were included. Of 184 references identified, 169 were excluded based on the review of title (n=19), abstract (n=121), and full text (n=29). Fourteen references were eligible for this review, including 10 studies conducted among US adults, and 2 among adults outside of the US.
Table 1 includes the studies reviewed and provides details regarding study design, intervention, sample characteristics and primary outcomes. Five of the 8 US RCTs63–67 reported significantly more weight loss in the intervention group than in the control or comparison group. The testing and use of mobile technologies varied a great deal and combinations of mHealth components and tools were often very specific to a particular study. Five investigators used text messaging63,66,68–70 in studies that ranged from 8 weeks to one year in duration. Patrick permitted the participant to set the frequency of the SMS (25 times/day) and found a significant difference in weight loss between the two groups at four months while Napolitano observed better weight loss in the Facebook + SMS than the Facebook alone group at 8weeks.66 Only one study68, which used SMS and MMS 4 times/day in a 12-month study, did not observe a significant difference in weight loss. Two of the SMS studies were conducted outside of the US. Carter observed better weight loss at 6 months in the group receiving SMS and Haapala demonstrated similar results at 12 months. While none of the US studies using SMS reported positive findings beyond nine months, the Finnish study69 showed that a SMS intervention could result in significantly greater weight loss than no intervention for up to 12 months.
Table 1
Study Cited, Design, Outcome, Setting, Quality Rating | Sample Characteristics, Group Size, Baseline BMI, Study Retention | Study Groups & Components | Technology used | Intervention Duration, # of Intervention Contacts, Intervention Adherence, Interventionist | Primary Outcome: Mean Weight Loss (kg, kg/m2, or % change) |
---|---|---|---|---|---|
Haapala et al, 200969 Design: 2-group RCT Outcome: wt.Δ and waist circumference Δ Setting: Community Country: Finland | N = 125 Int1: n = 62 Int2: n = 63 Women: 77.4% Mean age (SD): Int1: 38.1 (4.7) yrs. Int2: 38.0 (4.7) yrs. BMI: Int1: 30.6 (2.7) kg/m2 Int2: 30.4 (2.8) kg/m2 Retention: Int1: 73% Int2: 65% | Int1: SMS (for personalized feedback) and study website (for tracking and information) Diet: cut down on unnecessary food intake and alcohol PA: increase daily physical activity Behavior: self-mon and reporting of wt. via SMS or study website Int2: Wait list control No Intervention | Mobile phone, SMS, study website | Duration: 1year Contacts: Int1: Real-time when participants reported wt. via text messaging Int2: No intervention contact Intervention Adherence: Mean number (SD) of wt. reporting via SMS or study website per week 3 mos. Int1: 8.2 (4.0) 6 mos. Int1: 5.7 (4.6) 9 mos. Int1: 3.7 (3.5) 12 mos. Int1: 3.1 (3.5) Interventionist: Int1: Automated Int2: NA | ITT (LOCF) 12 mos.
wt.Δ, kg, M (SD) Int1: −3.1 (4.9) Int2: −0.7 (4.7) p = .008 Waist Circumference Δ, cm, M (SD) Int1: −4.5 (5.3) Int2: −1.6 (4.5) p = .002 |
Patrick et al., 200963 Design: 2-group RCT Outcome: wt. Δ Setting: Community Country: US | N = 78 Int1: n = 39 Int2: n = 39 Mean age (SD): 44.9 (7.7) yrs. Women: 80% White: 75% African American: 17% BMI: Int1: 32.8 (4.3) kg/m2 Int2: 33.5 (4.5) kg/m2 Retention: Int1: 67% Int2: 67% | Int1: Mobile Phone Weight Loss Program Diet goal: 500 kcal/day reduction PA: Increase from baseline Behavior: Self-mon weekly wt. using mobile phone; time/frequency of tailored SMS set by Ps (2–5 times/daily), monthly phone calls by coach Int2: Mail Diet: No intervention PA: No intervention Behavior: Monthly mailings (healthful eating, PA and wt. loss) | Mobile phone SMS and MMS | Duration: 4 mos. Contacts: Int1: daily SMS and MMS, frequency set by Ps Int2: 4 monthly mailings Intervention adherence: Int1: 100% adherence to responding to all messages requesting a reply; by week 16, approximately 66%. Int2: NR Interventionist: Int1: Health coach + automated Int2: NA | LOCF imputation 4 mos. wt.Δ, kg, M (SE) Int1: − 2.10 (0.51) Int2: − 0.40 (0.51) p = .03 Completers only: Int1: − 2.46 (0.64) Int2: − 0.47 (0.64) p = .04 |
Turner-McGrievy et al., 200964 Design: 2-group RCT Outcome: wt. Δ Setting: Community Country: US | N = 78 Int1: n = 41 Int2: n = 37 Mean age (SD): Int1: 37.7 (11.8) yrs. Int2: 39.6 (12.2) yrs. | Int1: Social Cognitive Theory-based wt. loss podcast Diet: increase fruit and vegetable intake, decrease fat intake PA: increase from baseline Behavior: encourage tracking wt., calories, and exercise | Podcast via MP3 player or computer for Int1 & Int2 | Duration: 12 wks. Contacts: Int1: 2 podcasts/wk. (mean length 15 min) Int2: Same as Int1 (mean length 18min) | ITT (BOCF) 12 wks. wt.Δ, kg, M (SD) Int1: − 2.9 (3.5) Int2: − 0.3 (2.1) p <.001 |
Women: Int1: 68% Int2: 81% White: Int1: 85% Int2: 78% BMI: Int1: 31.8 (3.2) kg/m2 Int2: 31.4 (4.1) kg/m2 Retention: Int1: 90% Int2: 92% | Int2: Non-theory-based wt. loss podcast Diet: avoid overeating PA: NR Behavior: NR | Intervention adherence: Mean (SD) number of podcasts listened to, (n =24), Int1: 17.5 (8.1) Int2: 16.6 (7.5) p <0.67 Interventionist: Int1: Automated Int2: Automated | BMI Δ, kg/m2, M (SD) Int1: − 1.0 (1.2) Int2: − 0.1 (0.7) p <.001 | ||
Shuger et al., 201167 Design: 3-group RCT Outcome: wt. Setting: Community Country: US | N = 197 Int1: n = 49 Int2: n = 49 Int3: n = 49 Int4: n = 50 Mean age (SD): 46.9 (10.8) yrs. Women: 81.7% White: 66.8% African American: 32.1% BMI: Int1: 33.0 (5.0) kg/m2 Int2: 33.2 (5.4) kg/m2 Int3: 33.1 (4.8) kg/m2 Int4: 33.7 (5.5) kg/m2 Retention: At 4 mos.: 70% At 9 mos.: 62% | Int1: Group-based behavioral wt. loss program + armband Diet: adopt healthful eating pattern PA: Increase PA + armband Behavior: self-mon of daily meal, lifestyle activity, and emotion/mood + weekly weigh-in and coach-directed sessions for weigh loss support and maintenance Int2: Armband alone Diet: adopt healthful eating pattern PA: Increase PA + armb000and Behavior: self-mon of daily meal, lifestyle activity, and emotion/mood + real-time feedback on energy expenditure, min. spent in mod and vig PA, and steps/day Int3: Group-based behavioral wt. loss program alone Diet: Same as Int1 + emphasis on wt. loss PA: Same as Int 1 Behavior: Same as Int 1 + weekly weigh-in and coach-directed sessions for weigh loss support and maintenance Int4: Self-directed wt. loss program following an evidence-based manual Diet: adopt healthful eating pattern PA: increase PA Behavior: self-mon of daily meal, lifestyle activity, and emotion/mood. | Bodymedia™ armband with a real-time wrist watch display and a personalized wt. management solutions web account | Duration: 9 mos. Contacts: Int1: Same as Int2 and Int3 Int2: Real-time when participants uploaded armband and recorded daily energy intake and body wt. to the website Int3: 14 weekly group sessions during the first 4 mos.; 6 1-on-1 phone counseling sessions during the final 5 mos. Int4: 1 self-directed wt. loss manual Intervention adherence: NR Interventionist: Int1: Health coach + automated Int2: Automated Int3: Health coach Int4: NA | ITT (how handled missing data NR) Baseline Wt., kg, M (SE) Int1: 100.32 (2.97) Int2: 101.15 (2.95) Int3: 101.84 (2.95) Int4: 102.22 (2.97) n.s.d. among 4 groups 4 mos. Wt., kg, M (SE) Int1: 96.83 (2.99) Int2: 98.48 (2.97) Int3: 100.74 (2.99) Int4: 101.23 (3.03) p NR 9 mos. Wt., kg, M (SE) Int1: 93.73 (2.99) Int2: 97.60 (2.99) Int3: 99.98 (3.00) Int4: 101.32 (3.05) Int1 vs. Int4: p=0.04 Int2 or Int3 vs. Int4: p NR |
Burke et al., 201274; Burke et al., 201175 Design: 3-group RCT Outcome: % wt.Δ at 6 and 12 mos. Setting: community/academic center Country: US | N = 210 Int1: n = 68 Int2: n = 70 Int3: n = 72 Mean age (SD): 46.8 (9.0) yrs. Women: 84.8% White: 78.1% Median BMI, (IQR): 33.09 (6.89) kg/m2 Retention: Int1: 86.8% Int2: 84.3% Int3: 86.1% | Int1: PDA only Diet: 1200–1800/day calorie goal based on wt. and gender; ≤25% of total calories from fat PA: Increase by 30 minutes semi-annually to 180 minutes by 6 mos. Behavior: Self-mon using PDA Int2: PDA with daily tailored feedback message Diet: Same as Int1 PA: Same as Int1 Behavior: Self-mon using PDA and receiving automated daily feedback regarding calories or fat intake. Int3: paper diary Diet: Same as Int1 PA: Same as Int1 Behavior: Self-mon using paper diary and a nutritional reference book | PDA with dietary and PA self-mon program, daily remotely delivered feedback message in real time to Int2 group | Duration: 24 mos. Contacts: Int1: weekly group sessions for mo. 1–4, biweekly for mo. 5–12, and monthly for mo. 13–18, 1 session during the last 6 mos. Int2: Same as Int1 Int3: Same as Int1 Intervention adherence: ≥30% adherent to dietary self-mon at 18 mo Int1: 19–20% Int2: 19–20% Int3: 8% Interventionist: Int1: Dietitians and exercise physiologists Int2: Dietitians and exercise physiologists + automated Int3: Dietitians and exercise physiologists | ITT (0.3 kg/mo. was added to previous observation) 6 mos. % wt.Δ, %, M (SD) Int1: − 4.88% (6.20) Int2: − 6.58% (6.77) Int3: − 4.59% (5.66) n.s.d. 24 mos. % wt.Δ, %, M (SD) Int1: − 1.18% (8.78) Int2: − 2.17% (7.04) Int3: − 1.77% (7.23) n.s.d. |
Shapiro et al., 201268 Design: 2-group RCT Outcome: % wt.Δ Setting: Community Country: US | N = 170 Int1: n = 81 Int2: n = 89 Mean age (SD): 41.9 (11.8) yrs. Women: 65% White: 64% BMI: Int1: 32.4 (4.2) kg/m2 Int2: 32.0 (4.0) kg/m2 Retention: Int1: 79% Int2: 89% | Int1: e-newsletter + SMS and MMS + website Diet: 500/day kcal reduction goal PA Goal: 12,000 steps/d with a gradual increase of 750 steps per wk, then encourage increase PA time or walk at a faster pace Behavior: Self-mon daily step count and weekly wt., automated personalized feedback on progress via mobile phone, accessing health tips, recipes, food and PA logs, wt. chart on a website Int2: e-newsletter control Diet: Same as Int1 from e-newsletters only PA: Same as Int1 from e-newsletters only Behavior: No intervention | Mobile phone SMS and MMS | Duration: 12 mos. Contacts: Int1: SMS and MMS 4 times/d, monthly e-newsletters Int2: monthly e-newsletters Intervention adherence: Responses to SMS. Int1: knowledge testing questions: 60%, the first pedometer steps questions: 51%, and the first wt. questions: 55% Int2: NA Interventions: Int1: Automated Int2: NA | Imputation via MICE At 6 mos. (primary) Wt..Δ,lb., M (SE) Int1: − 3.72 lb. (9.37) Int2: − 1.53 lb. (7.66) p = .110 12 mos. (secondary) Wt..Δ,lb., M (SE) Int1: −3.64 lb. (12.01) Int2: −2.27 lb. (9.39) p = .246 |
Turner-McGrievy and Tate, 2011,72Turner-McGrievy et al., 201373 Design: 2-group RCT Outcome: wt.Δ | N = 96 Int1: n = 47 Int2: n = 49 Mean age (SD): Int1: 42.6 (10.7) yrs. | Int1: Podcast + mobile group Diet: reduction of ≥500 kcal/day, decrease dietary fat to <30% of total energy, limit added sugar, increase fruit and vegetable consumption PA: goal minimum of 30 min/d of mod- | App on mobile phone | Duration: 6 mos. Contacts: Int1: Same as Int2 + daily contacts with coaches and group members via mobile app Int2: 2 15-min podcasts/wk. x’s 3 | ITT (BOCF): 3 mos. % wt.Δ, %, M (SD) Int1: − 2.6% (3.5) Int2: − 2.6 % (3.8) n.s.d. |
Setting: Community Country: US | Int2: 43.2 (11.7) yrs. Women: Int1: 77% Int2: 73% White: Int1: 75% Int2: 78% BMI: Int1: 32.9 (4.8) kg/m2 Int2: 32.2 (4.5) kg/m2 Retention: Int1: 89% Int2: 90% | vig PA by week 4 Behavior: Same as Int2 + self-mon diet, PA using mobile app, social support group members via Tweets app. Int2: Podcast group Diet: Same as Int1 PA: Same as Int1 Behavior: overcoming barriers and problem-solving, self-mon diet using book with calorie and fat gram content | mos., 2 mini podcasts /wk. x’s 3 mos. Intervention adherence: Podcasts (n = 24) downloaded, % 0–3 mos. Int1:68% Int2: 60.4% 4–6 mos. Int1: 37.5% Int2: 34.1% % adherence to self-mon diet: 0–3 mos. and 4–6 mos.: Int1: 41.4% 24.3% Int2: 34.3% 18.6% Percent adherence to recording PA0– 3 mos. and 4–6 mos. Int1: 34.3% 21.4% Int2: 37.1% 22.8% Interventionist Type: Int1: Study coordinator Int2: NA | 6mos. (primary) % wt.Δ, %, M (SD) Int1: − 2.7% (5.6) Int2: − 2.7% (5.1) n.s.d. | |
Carter et al., 201370 Design:3-group RCT Second outcome: wt.Δ, BMIΔ, %body fatΔ Setting: Community Country: UK | N = 128 Int1: n = 43 Int2: n = 42 Int3: n = 43 Women: 77.3% Mean age (SD): Int1: 41.2 (8.5) yrs. Int2: 41.9 (10.6) yrs. Int3: 42.5 (8.3) yrs. White: Int1: 100% Int2: 92.9% Int3: 83.3% BMI Int1: 33.7 (4.2) kg/m2 Int2: 34.5 (5.6) kg/m2 Int3: 34.5 (5.7) kg/m2 Retention: | Int1: Apps on mobile phone + SMS + internet forum (for social support) Diet: NR PA: NR Behavior: wt. loss goal setting, self-mon. daily calorie intake, PA, and wt., instant or weekly feedback via SMSs to enhance self-efficacy and reinforce positive behaviors Int2: Website + internet forum (for social support) Diet: NR PA: NR Behavior: goal setting and self-mon. Int3: Paper diary + Internet forum (for social support) Diet: NR PA: NR Behavior: goal setting and self-mon. | app on mobile phone, SMS | Duration: 6 mos. Contacts: Int1: Instant and weekly Int2: No intervention contact Int3: No intervention contact Intervention Adherence: Mean days of dietary self-mon. Int1: 92 (67) Int2: 35 (44) Int3: 29 (39) p <0.001 Interventionist Type: Int1: Automated Int2: NA Int3: NA | ITT (BOCF) 6 mos. (not powered to detect significance) wt.Δ, kg, M (95% CI) Int1: −4.6 (−6.2, −3.0) Int2: −1.3 (−2.7, 0.1) Int3: −2.9 (–4.7, −1.1) p NR (Int1 vs. Int2: p < .05; Int1 vs. Int3 p =.12) BMI Δ, kg/m2, M (95% CI) Int1: −1.6 (−2.2, −1.1) Int2: −0.5 (−0.9, 0.0) Int : −1.0 −1.6, −0.4) p NR % Body fat Δ, %, M (95% CI) Int1: −1.3 (−1.7, −0.8) Int2: −0.5 (−0.9, 0.0) Int3: −0.9 (−1.5, −0.4) p NR |
Int1: 93% Int2: 55% Int3: 53% | |||||
Napolitano et al., 201366 Design: 3-group RCT Outcome: wt.Δ Setting: Academic setting Country: US | N = 52 Int1: n = 17 Int2: n = 18 Int3: n = 17 Mean age (SD): 20.5 (2.2) yrs. Women: 86.5% White: 57.7% African American: 30.8% Hispanic: 5.8% Asian: 1.9% BMI: 31.36 (5.3) kg/m2 Retention: Int1: 100% Int2: 89% Int3: 100% | Int1: Facebook Diet: calorie target based on wt. PA: target ≥ 250 min of mod intensity exercise per week Behavior: self-mon, planning, stress management, social support, special occasion tips, relapse prevention Int2: Facebook + SMS and personalized feedback Diet: Same as Int1 PA: Same as Int1 Behavior: Same as Int1, sent self-mon data via SMS, received daily SMS on self-mon of calorie, PA, and wt. goals, received weekly summary reports via Facebook link, and selected a “buddy” for support. Int3: Wait list control No intervention | Mobile phone, SMS, social media | Duration: 8 wks. Contacts: Int1: 8 weekly Facebook sessions Int2: Same as Int1, daily SMSs Intervention adherence: Responses to SMS Int1: NA Int2: self-mon SMS 68.5%, general monitoring SMS 79.8% Int3: NA Interventionist Type: Int1: NA Int2: Automated Int3: NA | ITT (ways to deal with missing data NR)
4 wks. wt.Δ, kg, M (SD) Int1: − 0.46 kg (1.4) Int2: − 1.7 kg (1.6) Int3: 0.28 kg (1.7) p = <.01 post-hoc contrasts showed Int2 was significantly different from G1 (p < 0.05) and G3 (p ≤ .001) 8 wks. (primary) wt.Δ, kg, M (SD) Int1: − 0.63 kg (2.4) Int2: − 2.4 kg (2.5) Int3: − 0.24 kg (2.6) p <.05 post-hoc contrasts showed Int2 was significantly different from G1(p < 0.05) and Int3 (p < .05) |
Spring et al., 201365 Design: 2-group RCT Outcome: wt.Δ Setting: Veterans Affairs medical center Country: US | N = 70 Int1: n = 35 Int2: n = 35 Mean age (SD): 57.7 (11.9) yrs. Women: 14.5% White: 69.6% Minorities: 30.4% BMI: Int1: 36.9 (5.4) kg/m2 Int2: 35.8 (3.8) kg/m2 Retention: Int1: 83% Int2: 80% | Int1: standard + connective mobile technology system Diet: Same as Int2, calorie reduction was wt. loss based. PA: Same as Int2, goal - 60 min/d of mod-intensity PA with 25% increase if previous goal met Behavior: Wt. loss phase (1–6 mos.): Same as Int2, self-mon and regulating food intake and PA using PDA daily first 2 wks., then weekly until 6 mos., personalized feedback from coach every 2 wks via 10–15 min phone call; Maintenance phase (7–12 mos.): Same as Int2, recorded and transmitted data biweekly during 7–9 mos. and 1 week per month during 10–12 mos. Int2: standard-of-care Diet: 18 MOVE! Group sessions | PDA | Duration: 12 mos. Contacts: Int1: bi-weekly group sessions mos. 1–6, monthly mos. 7–12 Int2: Same as Int1 Intervention adherence: Mean number of MOVE! sessions attended Int1: 6.2 (34%) out of 18 sessions Int2: 5.9 p=0.54 Mean (SD) number of treatment calls received by Int1: 8.9 (2.8) Interventionist Type: Int1: Dietitians, psychologists, or physicians Int2: Dietitians, psychologists, or | ITT, ways to deal with missing data NR 3 mos. wt.Δ, kg, M (95% CI) Int1: - 4.4 kg (−2.7,−6.1) Int2: - 0.86 kg (−0.04, −1.8) p<.05 6 mos. wt.Δ, kg, M (95% CI) Int1: −4.5 kg (−2.1, −6.8) Int2: −1.0 kg (0.7, −2.5) p<.05 9 mos. wt.Δ, kg, M (95% CI) Int1: − 3.9 kg (−0.8, −6.9) Int2: − 0.9 kg (1.1, −2.9) p<.05 |
PA: 18 MOVE! Sessions Behavior: Wt. loss phase (1–6 mos.): 12 bi-weekly MOVE! Sessions, self-mon encouraged; Maintenance phase (7–12 mos.): 6 monthly MOVE! Support group sessions | physicians + paraprofessional coach | 12 mos. wt.Δ, kg, M (95% CI) Int1: − 2.9 kg(−0.5, −6.2) Int2: − 0.02 kg (2.1, −2.1) n.s. | |||
Systematic Reviews and Meta-Analysis | |||||
Siopis, et al. (2014)56 Design: Meta-analysis of 6 RCTs Outcome: mean wt.Δ Setting: NR | N ranged from 51–927 Retention:47%−96% | Int1: SMS Int2: group session diet/exercise intervention or no intervention | mobile phone, SMS | Duration: 8wks.- 12mos. Intervention Adherence: NR | Pooled wt.Δ, kg, M (95%CI) Int1: −2.56 (−3.46, −1.65) Int2: −0.37 (−1.22, −0.48) Meta-regression results: Int mean wt. Δ 2.17 kg higher than Int2 group (95% CI = 3.41 to −0.93, p =.001) |
Note: : P or Ps= participant(s), N = total sample, n = subgroups, Int = Intervention group, wt. = Weight, wk. = week, mo.= month, wks. = weeks, mos. = months, yrs. = years, BOCF = baseline observation carried forward, IQR = interquartile range; ITT = intention to treat, LMM = linear mixed model, LOCF = last observation carried forward, MICE = multivariate imputation by chained equations, MMS = multimedia messaging service, automated = without a clinician who generates, tailors, or modifies the output; NA = not applicable, NR = not reported, n.s. = not significant, n.s.d. = not significantly different, PDA= personal digital assistant, SMS = short message service, self-mon = self-monitoring, Δ = change or difference, BMI= body mass index,
Shuger reported on a study that tested the Bodymedia™ armband for monitoring daily physical activity with a wrist watch display with or without a behavioral intervention67 and compared it to two groups not using the armband. Only the armband plus group intervention achieved significantly greater weight loss than the self-directed control group at 9 months. Two investigators71 used PDAs for self-monitoring. Burke delivered daily, tailored feedback messages via the PDA to one of the groups and found no difference in weight loss at 2 years while Spring delivered personalized feedback by phone and observed significantly different weight loss at 6 months but the effect was not sustained at 12 months.
Turner-McGrievy et al.64 reported that a theory-based podcast delivered via MP3 players or computers led to significantly greater weight loss than a non-theory-based weight loss podcast at 12 weeks. Building on this study, the investigators72,73 conducted a follow-up study to compare the incremental effect of adding to the theory-based podcast mobile apps for self-monitoring and communication with a health coach and group members. However, the addition did not result in significantly greater weight loss than the podcast alone at 6 months.72,73
Two adult RCTs were conducted outside of the US (see Table 1). The 6-month UK study70 compared a self-directed smartphone app for goal setting and self-monitoring plus automated tailored feedback via text messaging with a website control and a paper diary control. Compared to the two other groups, the smartphone group achieved significantly greater mean weight loss at 6 months. The 1-year Finish study69 was the only mobile technology intervention reviewed in the 2013 AHA/ACC obesity treatment guideline. It tested a weight loss intervention via text messaging for instructions, self-monitoring, and automated personalized feedback vs. a no-intervention control group among overweight or obese adults. While none of the US adult studies reported positive findings beyond 9 months, the Finish study showed that a text messaging intervention could result in significantly greater weight loss than no intervention up to 12 months (i.e., intermediate term).
Also include in Table 1 is the only meta-analysis56 to date that focused on text messaging interventions for weight loss and showed that the pooled mean weight change was significantly better in intervention participants than in the control conditions. However, both intervention and control subgroups were heterogeneous and the funnel plot suggested a possible publication bias.
Khaylis and colleagues identified 5 key components to efficacious technology-based weight loss interventions: use of a structured program, self-monitoring, feedback and communication, social support, and individual tailoring.59 These components were found in the mobile technology interventions shown to produce greater weight loss than a randomized control group, although the extent and nature of the implementation of each component varied across studies. Additionally, all of the effective mobile interventions focused on calorie-reduced healthy eating, increased physical activity, and behavior change, which is consistent with the 2013 AHA/ACC guideline recommendation for comprehensive behavioral weight loss interventions.47 Evidence from the reviewed RCTs suggests that these technologies may be effective when used alone or in conjunction with traditional weight loss intervention delivery modalities (e.g., telephonic coach feedback or in-person group sessions or websites) to achieve modest weight loss of clinical significance in the short term.
Recommendations for consumers and healthcare practitioners.
During the past decade, the mHealth field has made great strides developing efficacious mobile weight loss approaches. Indeed, mobile interventions can produce weight loss in motivated populations, albeit at a lower magnitude relative to traditional treatment approaches. The characteristics of successful mobile interventions are quite comparable to their offline counterparts: the largest weight losses are produced by comprehensive, multicomponent interventions that are personally tailored, promote regular self-monitoring, and involve a qualified interventionist.59 The accumulated evidence, while limited, supports intervention delivery through a range of technology channels (including web, SMS, e-mail, telephone, and IVR), with limited variability in the magnitude of weight loss outcomes.
Standard behavioral weight loss treatment is delivered by a trained healthcare professional to promote calorie-controlled healthy eating, increased physical activity, and behavior change in in-person group or individual sessions of a prescribed frequency and duration. It is encouraging that sufficient evidence derived mainly from studies of internet-, email-, and telephone-based interventions has accrued to buttress the 2013 AHA/ACC obesity treatment guideline recommending electronically-delivered comprehensive weight loss programs encompassing personalized coach feedback as an acceptable, albeit possibly less effective, alternative to standard in-person treatment.
Our review finds that self-monitoring and automated personalized feedback are common features in the contemporary mobile weight loss interventions. Based on consistent findings from multiple RCTs of fair and good quality, the evidence is strong for short-term weight loss benefits in adults from text messaging interventions for self-monitoring and feedback when supported by other methods (e.g., coach phone calls, websites, or private peer groups via social media)63,66 or incorporated into an existing comprehensive lifestyle program,65 with some evidence suggesting sustained intervention effectiveness through 12 months.47 Importantly, there is no evidence to suggest that SMSs as a stand-alone intervention are effective. One RCT in the UK showed the effectiveness of a self-directed smartphone app as a standalone intervention in overweight and obese adults,70 although the translatability of the results to US adults is unclear due to a lack of research.
Until more evidence emerges, health practitioners looking to implement or recommend mHealth interventions to their overweight and obese patients should ensure that programs and tools they recommend include established evidence-based content and components of a comprehensive lifestyle intervention (i.e., calorie-controlled healthy eating and increased physical activity with specified goals, and behavioral strategies) and facilitate adoption of evidence-based weight loss behaviors (e.g., self-monitoring, personalized feedback, and social support from coaches or peers). In the context of these programs, mobile technologies, in particular SMS/MMS messaging and smartphone apps, may be the primary intervention modality while supported by other methods (e.g., websites or phone calls). At present, no recommendations can be made for US consumers regarding the effectiveness of text messaging as a stand-alone intervention for weight loss or the effectiveness of any particular smartphone app.
Gaps and Recommendations for Future Research.
There is great need for studies that explore mobile interventions in diverse contexts, particularly general consumer samples and in clinical practice settings. While great strides have been made, we do not have answers to the questions that consumers are most likely to ask: whether commercial mobile weight loss apps are efficacious. We know little about the efficacy of the more than 1000 apps that purport to help consumers lose weight. Moreover, few, if any, research-tested apps have been widely disseminated or commercialized. Academic-industry partnerships are needed from the intervention development stage through formative evaluation to confirmatory research and then dissemination and implementation.
The research literature investigating mobile weight loss interventions remains in its infancy, with many important questions yet to be answered. Indeed, we know little about how to best integrate mobile interventions into the primary care setting, where they might serve as adjuncts to weight loss counseling delivered by primary care or other providers such as dietitians or nurses. There are potentially significant opportunities to explore the integration of mobile technologies, given health system changes associated with the Affordable Care Act and the 2011 Center for Medicare and Medicaid Services decision to reimburse qualified providers for delivering intensive behavioral treatment for obesity. We need to also expand the range of populations that have been studied. Thus far, we know the least about those populations with the highest obesity rates, and who bear the greatest burden of obesity associated disease -- racial/ethnic minorities and the socioeconomically disadvantaged.76 This fails to deliver on the promise of digital health approaches, which have potential for extending the reach of intervention approaches. Despite their higher levels of mobile phone ownership and utilization,77,78 early evidence suggests that high risk populations experience suboptimal weight losses,76 as is often observed with traditional treatment approaches.
More work is necessary to assess and improve the magnitude of weight loss outcomes produced by mobile interventions as well as long term maintenance of weight loss. A particular priority is identifying strategies to promote sustained user engagement. Indeed, across a large number of studies weight loss outcomes have been shown to be largely dependent on the level of participant engagement.53,62 Unfortunately declining engagement and attrition (often as high as 40–50%) are characteristic of digital health interventions.79 Mobile interventions developed in research settings might benefit from leveraging the iterative design and testing conventions that are commonly used in the commercial market to promote user engagement. Further, the most successful trials have combined interventionist support with a mobile intervention. We know much less about the efficacy of standalone mobile interventions, those that have the greatest potential for broad dissemination.
At present, there is considerable variability in the technologies, intervention components, design, and delivery schedules of mobile interventions. We know little about which technologies or intervention components, or combinations thereof, are best equipped to produce clinically meaningful weight loss. There does not appear to be substantial variability in the magnitude of weight loss outcomes in the mHealth approaches we reviewed. We have identified the following gaps and directions for future research:
The applications of mobile technology for weight loss have been limited in conceptualization and narrow in implementation. Future mobile technology weight loss interventions should build on the best evidence of the efficacious core components of comprehensive lifestyle programs.
Text messaging has been the primary delivery format researched to date; however, it is only one of a growing number of mobile delivery formats (e.g., smartphone apps, wearable sensors that synchronize data with smartphones). We need to address the many pitfalls in the current mHealth approaches, e.g., absence of theoretical basis, limited application of the best practices in technology design, low usage of empirically-supported behavioral strategies, and limited scientific rigor, by engaging in transdisciplinary collaboration and inclusion of the end-users, the clinicians and patients in all phases, from the intervention development to implementation.
Use mixed methods research to elucidate the frequency, timing and duration of various mobile delivery formats which can enhance the usability and acceptability of technology.
Future work needs to focus on comparative effectiveness research using alternative designs, for example, equivalence and noninferiority trials. Also, we need to use more flexible study designs that are able to provide answers within a shorter time frame than the conventional 5-year clinical trial when testing a delivery mode that will become obsolete before the end of the trial.80
Finally, we need to capitalize on the currently available technologies that permit collection and transmission of data in real time to better learn about the behaviors and moods of individuals in their natural setting, referred to as ecological momentary assessment, which can inform the development of interventions that can be delivered in real time and thus provide support when individuals are in need of it.81,82
Use of mHealth Interventions to Increase Physical Activity
Regular physical activity is important in improving cardiovascular health. The Centers for Disease Control (CDC), the American College of Sports Medicine (ACSM), and the American Heart Association recommend that adults participate in 30 minutes or more of moderate-intensity physical activity on most days of the week. 83According to the 2008 Physical Activity Guidelines for Americans84, adults should avoid inactivity or extended periods of sedentary activity, do at least 150 minutes of moderate-intensity activity weekly, and do muscle-strengthening activities on at least 2 days per week.85 Sustained physical activity has many health benefits, such as decreasing the risk for premature death, T2DM, stroke, some forms of cancer, osteoporosis, and depression.86 There is sufficient evidence that physical activity can help reduce CVD risk factors, such as high blood pressure.86
Physical activity in the US has significantly declined over the past two decades. Since the late 1980s, the proportion of adult women who report no leisure-time activity has increased from 19.1% to 51.7% and the proportion of adult men reporting no leisure-time activity rose from 11.4% to 43.5%.87 The participation in leisure-time activity is lowest in African Americans, with over 55% not meeting the guidelines, followed by those identifying as Hispanic or Latino, with over 54% not meeting guidelines.88 Over 66% of those who have not completed high school do not meet the 2008 Physical Activity Guidelines for Americans.88
Review of evidence for efficacy of mobile technology-based interventions to promote physical activity.
We searched PubMed using the terms physical activity; physically active; walk; aerobic; sport; lifestyle; sedentary. The literature search yielded 1490 studies. 1415 were excluded based on the review of title (n=797), abstract (n=528), or full text (n=122). Of the 122 that did not qualify based on full text review, articles were excluded for the following reasons: 44 were focused on diabetes, 39 on weight loss and 41 for not meeting RCT criteria. Therefore, 41 articles were eligible for the current review, 12 were literature reviews of physical activity for CVD prevention, 15 were studies validating technology, and 14 were RCTs that are detailed in Table 2. The literature search yielded studies reporting on numerous types of technology that can be used for increasing physical activity; texting or SMS messaging on mobile phone (n=3), pedometer (n=1), email (n=1), and internet (n=9). Several studies included a combination of technologies.
Table 2.
Study Cited, Design, Primary Outcome, Setting, Quality Rating | Sample Characteristics, Group Size, Baseline BMI, Study retention | Study Groups & Components | Technology used | Intervention Duration, # of Intervention Contacts, Intervention Adherence, Interventionist | Primary Outcome |
---|---|---|---|---|---|
Plotnikoff et al., 200592 Design: 2-group Outcome: Mets/min Setting: Workplace Country: Canada | N=2121 Int1: n=1566 Int2: n=555Mean age (SD): Int1: 44.9 (6.2) yrs. Int2: 45.0 (6.4) yrs. Women: 73.5% White: NR Mean BMI (SD): Int1: 27.2 (5.7) kg/m2 Int2: 27.0 (5.7) kg/m2 Retention: not reported | Int1: received one physical activity and one parallel nutrition message per week for 12 weeks. Int2: received no weekly messages | Duration: 12 wks. Contacts: Int group received a total of 24 messages over the 12 wks. Intervention adherence: NR Interventionist: Int1:NR Int2:NR | Completer’s analysis (n=2074) 12 wks. PA, mean MET/min. Int1: 683.68 Int2: 592.66 p <.01 | |
Hurling et al., 200789 Design: Randomized, stratified Controlled trial Outcome: Δ in MPA (METs/wk) Setting: Community Country: UK | N= 77 Int1: n=47 Int2: n=30 Mean age (SD): Int1: 40.5 (7.1) yrs. Int2: 40.1 (7.7) yrs. Women: 66% Mean BMI (SD): Int1: 26.2 (2.8) kg/m2 Int2: 26.5 (4.1)kg/m2 Retention: 100% | Int: 9 weeks of tailored solutions for barriers, mobile phone and email reminders to exercise, message board, real-time feedback via internet. Int2: verbal advice on recommended PA levels | Internet, mobile device, email | Int lasted 9 wks. Study duration: 12 wks. Contacts: Int1: Not specified Intervention adherence: 85% of Int1 Ps logged onto website in first 4 weeks, 75% logged in during the last 5 weeks. Only 33% of participants accessed all components of the system Interventionist: Int1: Automated Int2: NR | ITT, ways to deal with missing data NR 12 wks. Accelerometer data, MPA, METs/wk., M(SE) Int1: 5.39(0.01) Int2: 5.34(0.01) p = .02 |
Spittaels et al., 2007102 Design: 3-group RCT Outcome: Total PA Setting: Workplace Country: Belgium | N=526 Int1: n=174 Int2: n=175 Int3: n=177 Mean age (SD): 39.5 (8.5) yrs. Women: 31% Mean BMI (SD): 24.4 (3.3) kg/m2 White: NR Retention: 72% | Int1: Online-tailored physical activity advice + stage-based reinforcement e-mails Int2: Online-tailored physical activity advice only Int3: Online non-tailored standard physical activity advice | Internet, email | Duration: 6 mos. Contacts: Online-tailored PA advice + email group received 5 emails over 8 weeks. Intervention adherence: Int1 group, 77% of Ps read the emails they received Interventionist: Int1:NR Int2:NR Int3: NR | Completer’s analysis (n=379) 6 mos. Total PA, min/wk., M(SD) Int1: 776 (540) Int2: 682(452) Int3: 708(514) n.s.d. |
Dunton et al., 200893 Design: 2-group RCT Outcome: Δ in walking time and MVPA Setting: Community Country: US | N=156 Int1: n=85 Int2: n=71 Mean age (SD): Int1: 42.8 (12.8) yrs. Int2: 42.8 (10.5) yrs. Women: 100% White: 65% Retention: 85% | Int1: Individually tailored PA plans via internet, strategies to overcome barriers via internet, 10 weekly follow- up emails Int2: Waitlist | Email, Internet (Website-Women’s Fitness Panner) | Duration: 3 mos. Contacts: 3 mos. access to website, 10 weekly follow-up email newsletters Intervention adherence: Int1: 6% reported not receiving weekly emails, 23% opened all emails, 8% opened none. 8% visited the website >10 times Int2: 11% reported not receive the weekly newsletters Interventionist: Int1:NR Int2: NR | ITT (MRCM; HGLM) 3 mos.. Δ in walking time, mean min./wk. Int1: 69 min. Int2: 32 min. p=.035 (one-tailed) MVPA Δ, mean min/ week Int1: + 23 min./ wk. Int2:−25 min/ wk. p=.045 (one-tailed) |
King et al, 200895 Design: 2-group RCT Outcome: minutes/week of PA Setting: Community Country: US | N=37 Int1: n=19 Int2: n=18 Mean age (SD): Int1: 60.7 (6.8) yrs. Int2: 59.6 (7.6) yrs. Women: 43% Caucasian: 78.5 % Retention: 100% | Int1: PDA programmed to monitor PA levels twice per day for 8 weeks. Daily and weekly individualized feedback, goal setting, and support. Int2: written PA educational materials | PDA | Duration: 8 wks. Contacts: Int1: Daily contacts for 8 weeks Intervention adherence: Int1 Ps completed an average of 68% of the PDA entries over the 8 wks. Interventionist: Int1: NR Int2: NR | 8 wks. PA, min/week., M(SD) Int1: 310.6 (267.4) Int2: 125.5 (267.8) p=.048 |
Ferney et al., 200994 Design: 2-group RCT Outcome: min/wk. of PA Setting: community Country: Australia | N=106 Int1: n=52 Int2: n=54 Mean age (SD): Int1: 51.7 (4.1) yrs. Int2: 52.2 (5.0) yrs. Women: 72% White: NR Retention: 88% | Int1: Ps received access to a neighborhood environment-focused website, received tailored information for increasing PA through emails Int2: access to a motivational-information website, received non-tailored emails | Email and internet | Duration: 26 wks. Contacts: Both groups: received 11 emails over the 26 weeks. Weeks 1–4: weekly emails Weeks 5–12: bi-weekly emails Weeks 13–26: monthly emails Intervention adherence: 13 % of Ps used self-monitoring tool and 25% sent email to the activity counselor in Int group Interventionist: Int1: NR Int2: NR | ITT (BOCF) 26 wks. Total PA Δ, mean min/wk. Int1: + 57.8 min/wk. Int2: + 13 min/wk. Interaction effect: p < .05 |
Fjeldsoe et al., 201094 | N= 88 | Int1: a face-to-face PA goal- | SMS | Duration: 13 weeks | ITT |
Design: 2-group RCT Outcome: Δ in MVPA and walking time Setting: Community Country: Australia | Int1: n=45 Int2: n=43 Mean age (SD): Int1: 28 (6) yrs. Int2: 31 (6) yrs. Women: 100% Education level < 10 yrs.: 17% Retention: 69% | setting consultation, phone consultation, a goal-setting magnet, 3–5 personally tailored SMS/wk and a nominated support person who received SMS per week. Int2: face-to-face information session | Contacts: Int1: and 6 wks.: face-to-face PA goal-setting consultation 42 tailored SMS on behavioral and cognitive strategies: 0–2 wks.: 5/wk. 3–4 wks.: 4/wk. 5–12 wks.: 3/wk. 11 weekly ‘goal check’ SMS Int2: no contacts apart from reminder telephone calls to confirm their 6-and 13-wk assessments Intervention adherence: 13 wks. : 84% of Int1 group meeting MVPA goal 10 wks.: 24% response to SMS 6 wks., 64% of Int2 Ps remaining in the trial (n = 36) reported reading the SMS and then storing them, 33% reported reading the SMS and then deleting them, and one P (3%) reported deleting the SMS without reading them Interventionist: Int1: trained behavioral counselor + automated Int2: trained behavioral counselor | 13 wks. Δ in MVPA duration, M(SE) min/wk. Int1: 18.26 (24.94) Int2: 16.36 (25.53) p =.26 Δ in walking duration, M (SE) min/wk. Int1: 16.67 (13.33) Int2: 0.34 (13.64) p =.005 | |
Richardson et al., 2010103 Design: 2-group RCT Outcome: step count Setting: community Country: US | N= 324 Int1: n=254 Int2: n=70 Mean age (SD): 52.0 (11.4) yrs. Women: 66% White: 86% mean BMI (SD): 33.2(6.2) kg/m2 Retention: 76% | All Ps wore pedometers and had access to individually tailored messages, weekly goals. Int1: access to post and read messages from other Ps Int2: no access to message board | Internet; Pedometers | Duration: 16 wks Contacts: Int1: received access to a community message board (reading and posting comments to group) for the 16-wk Int Intervention adherence: 65% of Int Ps were active in the community. Int1 group uploaded pedometer data on 87% of days, Int2 group uploaded pedometer data on 75% of days | ITT(BOCF) 16 wks. Step counts, steps/day, M(SD) Int1: 6575 (3127) Int2: 5438 (2667) p = .20 |
Interventionist: Int1: NR Int2: NR | |||||
Aittasalo et al., 2012104 Design: 2-group RCT Outcome: Δ in walking time Setting: Community Country: Finland | N=241 Int: n=123 Con: n=118 Int1: Mean age (SD): Int1: 44.1 (9.4) yrs. Int2: 45.3 (9.1) yrs. Women:71% BMI >25: 63% Int2: Mean age: 45.3 yrs. Women:66% BMI >25: 76% Retention: 77% | Int1: One group meeting, log-monitored pedometer use, six email messages Int2: no intervention | Pedometers and Email | Duration: 12 mos. Contacts: 6-mo treatment duration, 1 email per mo., pedometer use daily Intervention adherence: 60% of Int Ps used pedometer regularly; 37% reported using pedometer irregularly for 6 mos. Emails reached 99% of Ps, 80% reported reading the messages Interventionist: Int1: NR Int2: NR | Completer’s analysis (n=164) 12 mos. Total walking, min/wk, M(SD) Int: 521 (468) Con: 395 (319) n.s. % of Ps walking stairs Int1: 88% Int2: 86% OR (95%CI): 2.24 (0.94 to 5.31) % of Ps walking for leisure Int1: 87% Int2: 76% OR (95%CI): 2.07 (0.99 to 4.34) |
Reid et al., 201290 Design: 2-group RCT Outcome: steps/day; Δ in MVPA Setting: Community Country: Canada | N=223 Int1: n=115 Int2: n=108 Mean age (SD): 56.4 (9.0) yrs. Women: 16.7% Mean BMI (SD): 29.3(4.8) kg/m2 Retention: 69% | Int1: personally tailored physical-activity plan upon discharge from the hospital, provided access to a secure website for activity planning and tracking, 5 online tutorials, and email access with an exercise specialist. Int2: consisted of PA guidance from an attending cardiologist | Internet | Duration: 12 mos. Contacts: 5 online tutorials over a 6-mo. period and email contact with an exercise specialist for Int1 group Intervention adherence: Mean # of online tutorials completed by Int1 Ps: 2.7 of max 5,61.7% of Ps completed ≥ 3 of 5 tutorials. 37 Int1 Ps emailed exercise specialist ≥ once. Interventionist: Int1: exercise specialist Int2: cardiologist | ITT(multiple imputation of missing values) 12 mos. Step counts, steps/day, M (SD) Int1: 7392 (3365) Int2: 6750 (3366) p = .023 Δ in MVPA, min/wks., M (SD) Int1: 201.4(179.8) Int2: 163.4(151.3) p = .047 |
Bickmore et al., 201391 Design: 2-group RCT Outcome: steps/day Setting: Community Country: US | N=263 Int1: n=132 Int2: n=131 Mena age (SD): 71.3 (5.4) yrs. Women: 61% White: 37% High school diploma or less: 51% | Int1: Mobile tablet computers with touch screens for 2 mos., directed to connect pedometers to tablet, interact with a computer-animated virtual exercise coach daily. Next 10 mos. given opportunity to interact with coach in a kiosk in | Internet via tablet with virtual exercise coach, pedometer | Duration: 12 mos. Contacts: Ps interact with coach in a clinic kiosk between mo.2 and mo. 10 Intervention adherence: Int1 group interacted with coach 35.8 (19.7) times during the 60day intervention phase, and | Completer’s analysis (n=200) 2 mos.: Steps/day, M Int1: 4,041 Int2: 3,499 p = .01 Completer’s analysis (n=128): 12 mos.: |
Retention: 86% | clinic waiting room. Int2: pedometer that only tracked step counts for an equivalent period of time. | accessed the kiosk an average of 1.0 ( 2.9) times during the 10mos follow-up. Interventionist: Int1: NR Int2: NR | Steps/day, M Int1: 3,861 Int2: 3,383 p = .09 | ||
Gotsis et al., 2013105 Design: randomized crossover design Outcome: Days/week of PA Setting: Community Country: US | N=142 Int1: n=64 Int2: n=43 Mean age (SD): 35.6 (9.5) yrs. Women: 67.6% Asian: 18% Hispanic: 28% Retention: 61% | Int1: Diary+Game group: additional features:(1) rewards, (2) virtual character, (3) choosing virtual locations for wellness activities, (4) collecting virtual items, (5) wellness animations by spending points, (6) virtual character wellness activities as updates Int2: Diary group: (1) posting updates of PA, (2) private messages, (3) history posted, and (4) viewing display of physical activities | Internet, social networking | Duration: 13 wks. Contacts: follow-up visits at 5–8 weeks and at 10–13 weeks Intervention adherence: Ps accessed the website every other day, with the number of total logins ranging from 1–102 (mean 38.00, SD 22.31) Interventionist: Int1: NR Int2: NR | Completer’s analysis (n=87) 13 wks. Δ in PA, days/wk., M Int1: 3.43 Int2: 0.88 p =.08 |
Kim et al, 201397 Design: 2-group RCT Outcome: steps/day Setting: community Country: US | N= 45 Int1: n=30 Int2: n=15 Mean age (SD): Int1: 69.3 (7.3) yrs. Int2: 70.6 (7.5) yrs. Women: 80% African American: 100% Mean BMI (SD): Int1: 30.2 (7.0) kg/m2 Int2: 31.4 (7.4) kg/m2 Retention: 80% | Int1: pedometer and manual to record steps plus motivational SMS 3 times/day, 3 days/wks. for 6 weeks. Int2: pedometer and manual to record steps but not SMS | SMS | Duration: 6 weeks Contacts: 3 x’s/day for 3 days/wks. for 6 wks. Intervention adherence: NR Interventionist: Int1: Automated Int2: NR | Completer’s analysis (n=36 6 wks. Δ in step count, M Int1: +679 Int2: +398 p<.05 |
King et al., 2014106 Design: 2-group RCT Outcome: Δ in MVPA Setting: Community Country: US | N=148 Int1: n=73 Int2: n=75 Mean age (SD): 60 (5.5) yrs. Women: NR White: NR Mean BMI (SD): 29.5 (5.4) kg/m2 Retention: 78% | Int1: 12-mo. home-based moderately vigorous physical activity (MVPA), primarily walking, program delivered via a trained telephone counselor (human advice arm\) Int2: a similar program delivered via an automated, computer interactive telephone system (automated | Automated, computer interactive telephone system | Duration: 12 mos., study 18 mos. Contacts: Ps in both groups received one 30-min in-person, one-on-one instructional session followed by a similar number of advisorinitiated telephone contacts across the 12- month period Intervention adherence: NR | ITT(BOCF) 18 mos. Int1: 167.0 +/− 135.6 Int2: 145.2 +/− 134.5 p =.41 |
advice arm). | Interventionist: Int1: : trained phone counselor Int2: automated | ||||
Systematic Reviews and Meta-Analysis | |||||
Bort-Roig et al, 201496 Design: Systematic review of 26 RCTs | Not provide Among the 26 reviewed articles, 17 implemented and evaluated a smartphone-based intervention, 5 used single group pre–post designs and 2 studies used pre–post designs relative to a control or comparison group. | Interventions that used smartphones to influence PA | Smartphone | Not given | Four studies (three pre-post and one comparative) reported physical activity increases (12–42 participants, 800–1,104 steps/day, 2 weeks-6 months), and one casecontrol study reported physical activity maintenance (n = 200 participants; >10,000 steps/day) over 3 months |
Note: P or Ps= participant(s), N = total sample, n = subgroups, Int = Intervention group, 1, 2, etc. Con = control group, RCT = randomized control trial, Δ = change or difference, mo. = month, mos. = months, wk. = week, wks. = weeks, year = yr., years = yrs., Baseline = 0, SMS = short message service, MMS = multimedia messaging service, PA = physical activity, MVPA, moderate- to vigorous-intensity PA, NA = not applicable, NR = not reported, n.s., = not significant, n.s.d = not significantly different, PDA = personal digital assistant; automated = without a clinician who generates, tailors, or modifies the output; ITT = intention to treat, HGLM=Hierarchical Generalized Linear Model, MRCM = Multilevel random coefficient modeling, BOCF = baseline observation carried forward, MICE = multivariate imputation by chained equations, LMM, linear mixed model, LOCF, last observation carried forward
Nine of the 14 studies reported significant increases in physical activity in the intervention group when compared to the control condition.89–97 Overall, the technology that was used most often to increase physical activity was the internet through websites, online tutorials, or networking opportunities. Many of the programs that used the internet also used other forms of technology, including pedometers and feedback messages via email. Of the 9 studies that used the internet as the main intervention component, 5 reported significant differences between groups in increasing physical activity.89–91,93,94 The outcomes differed in each study and included step counts, increases in moderately vigorous physical activity (MVPA), increases in moderate physical activity, and minutes/week of physical activity. Two of the 14 studies examined the use of SMS,96,97 and both reported significant differences between the intervention and control or comparative conditions.96,97 Two additional studies reported testing the use of messages, either through a PDA or email and found significantly greater increases in physical activity in the intervention group compared to the control or non-intervention group.92,95
A systematic review of 26 studies published in 2014 by Bort-Roig et al. examined the use of smartphones to influence physical activity.98 Only five studies in the review assessed interventions for physical activity and four reported an increase in steps/day. However, the studies were limited by small samples with only one study having a sample size greater than 50. A systematic review of 11 studies by Buchholz et al. in 2013 reported that fewer than 10 RCTs using SMS to target physical activity had been conducted across 7 countries and found that a small number of studies had examined the use of SMS for promotion of physical activity.99 The median effect size for differences in change scores between intervention and control groups for the studies was 0.50, but ranged from 0.20 to greater than 1.00. There was no evidence to suggest why there were such vast differences in the effect size.
One area that is growing in acceptance among consumers is active video gaming or exergaming. The studies using this technology had some methodological limitations and thus were not included in this review, however a systematic review by Peng et al., reported that laboratory studies have demonstrated that this technology is capable of providing light-to-moderate physical activity.100 However, only three studies in that review supported gaming as an effective tool to significantly increase physical activity or exercise attendance.
Most mobile technology interventions that have been reported in the published literature allow users to self-monitor physical activity by manually entering exercise bouts or total accumulated activity. However, more technologically sophisticated approaches for physical activity monitoring are rapidly becoming widely available. Physical activity tracking devices, also referred to as “wearables”, have become highly prevalent among consumers for self-monitoring daily activity. Most of these devices include accelerometers that capture users’ duration and intensity of physical activity.101 Some devices also include GPS functionality that can capture the location of exercise sessions. Originally designed to be worn on the hip, wearables can now be placed comfortably in a range of locations (e.g., wrist, ankle, arm, shoe). The majority of modern smartphones similarly include accelerometers and gyroscopes, allowing them to provide functionality similar to wearable devices. A host of third-party software applications have emerged to leverage this technology and some mobile operating systems include physical activity tracking as a default functionality. There is emerging evidence that combining physical activity tracking devices with group behavioral treatments will produce larger weight loss outcomes than either the device, or group treatment alone.67
Gaps and Recommendations for Future Research.
A large number of smartphone applications exist that are designed to monitor, track, and promote physical activity, as well as wearable devices (FitBit, JawBone) but none of these apps have been compared to the established methods of objectively measuring physical activity, such as accelerometers and thus have no empirical basis. Over 20% of US adults are tracking their health with some form of technology and 1 in 5 adults with a smartphone have at least one health application. The most popular health applications (38% of downloads) are those related to exercise, pedometer use, and heart rate monitoring.107 However, none of the studies identified in this review tested these wearable monitors. Therefore, it is recommended that future studies include the use of these commercially available devices in RCTs to determine their efficacy in improving physical activity. Since there is an absence of established accuracy and efficacy data for these consumer wearable devices, no guidelines exist on the use of these physical activity trackers. One study examining the accuracy and precision of these devices reported that most wearable devices yielded reasonably accurate reporting of energy expenditure, within about 10–15% error, when compared to a portable metabolic analyzer.108 Thus, rigorous RCTs with diverse populations are needed to establish an empirical basis for the use of the apps and mobile devices for improving physical activity or reducing sedentary activity.
The realm where there seems to have been a prolific explosion of wearable devices and trackers is physical activity; however, compared to some of the other health related areas, the research conducted to date is limited. The following list outlines the gaps and recommendations for future research in the area of mHealth interventions for promoting physical activity.
Little is known about the use of wearable consumer devices, although many adults are using this technology. Therefore, large scale randomized trials of diverse populations need to be conducted to test the effectiveness of this technology in increasing physical activity or reducing sedentary behavior.
Health related apps are amongst the most popular downloads, yet are not being rigorously tested. Therefore, commercially available apps that are downloaded by the public need to be validated and examined for efficacy and acceptability, as well as sustainability of engagement. Only then can we provide the consumer with evidence for their use.
Similarly, additional testing is recommended on the use of exergaming to increase PA levels in both children and adults
Use of the internet was the platform tested most often for the delivery of technology targeting increased physical activity. Thus, use of other platforms need to be tested for promoting physical activity, e.g., SMS or more recently developed approaches that can be delivered on a smartphone or tablet.
Use of mHealth for Smoking Cessation
Tobacco use remains the most significant preventable risk factor for CVD. The AHA Task Force on Risk Reduction noted that approximately a third of CVD deaths are attributable to smoking and that a substantial and rapid decrease in risk results from smoking cessation.109 Although there are a number of effective pharmacologic and behavioral interventions for smoking cessation, the delivery of these interventions has been inconsistent. Practice guidelines for smoking cessation incorporate the five A’s: Ask, Advise, Assess, Assist, and Arrange.110 Although most healthcare providers report asking about smoking and advising their patients to quit, they are much less likely to assess willingness to quit, assist with cessation, or arrange follow-up.111 Given the limitations of smoking cessation delivered by health professionals, technologies have been leveraged to facilitate the delivery of smoking cessation interventions. Early approaches utilized the internet to deliver these interventions112 and current and reputable internet interventions such as smokefree.gov are available.113 The advent of mobile technologies provides potential delivery advantages over internet interventions via desktop or laptop computers.
Smoking urges occur frequently throughout the day in response to various triggers, and indoor smoking bans have moved smoking behavior outside, away from computers used at work and home. Mobile devices, therefore, are more likely to be available when smokers experience the urge to smoke and can deliver interventions at these times. These mobile devices also offer the promise of “Just-In-Time Adaptive Interventions” (JITAI) that adapt interventions based on context and potentially preempt smoking behavior by anticipating when urges are likely to occur.114
Unfortunately, current commercially available mobile apps for smoking cessation have generally failed to deliver empirically-supported interventions or to make optimal use of the capabilities of mobile phones. A series of studies by Abroms and colleagues115,116 have shown that most commercially available smoking cessation apps do not adhere to practice guidelines for smoking cessation. Some of these practice guidelines, developed for delivery by healthcare professionals, may not be appropriate criteria for mobile interventions. For instance, should a smoking cessation mobile app ask about smoking or is it reasonable to assume that if a user has downloaded a quit smoking app that he/she is a smoker? Additionally, some empirically-supported approaches that are amenable to a computerized intervention such as scheduled gradual reduction of smoking117 may not have been included in the practice guidelines due to the difficulty of delivery by healthcare professionals. Even with these caveats, commercially available mobile apps for smoking cessation are generally incomplete and lack empirical basis. Abroms and colleagues have documented that although some smoking cessation apps are downloaded more than a million times per month, smartphone apps adhere, on average, to only about a third of the practice guidelines for smoking cessation interventions.
Although most commercially available smoking cessation apps are incomplete or lack empirical support, there has been considerable research on the efficacy of mobile interventions for smoking cessation that we review below. Unfortunately, until quite recently these empirically-tested interventions developed by smoking cessation researchers were not commercially available. Much of the initial research on SMS for smoking cessation occurred outside of the U.S. and the programs, if available, are available only in those countries. Additionally, researchers developing these mobile smoking cessation programs often did not partner with commercial entities capable of marketing the program once evaluated; however, there are recent examples of commercially available programs developed by researchers.118,119
Review of evidence for efficacy of mobile technology-based interventions to promote smoking cessation.
We searched PubMed for the years 2004 to 2014, using the terms quit smoke; stop smoke; stopped smoke; ceased smoke; smoking cessation; cigarette smoke; smokeless tobacco; smoker; tobacco cessation; tobacco use; nicotine replacement; nicotine gum; nicotine lozenge; nicotine nasal; nicotine patch; nicotine inhalant. These terms were cross-referenced with the mobile technology terms described previously. This search resulted in 286 identified articles. Of these, most (211) were not relevant to mobile technologies for smoking cessation. These were predominately internet-based interventions or studies that used mobile technologies for recruitment or measurement purposes, but not for intervention. Of the remaining 85 publications, 14 were RCTs of mobile technologies for smoking cessation, and these trials are described in Table 3. The remaining reports of mobile technologies for smoking cessation included a range of studies including descriptions of design and development; reports of feasibility, acceptability, and usability data, uncontrolled trials; and various systematic reviews. For completeness, two Cochrane meta-analyses of this area120,121 are included in the table
Table 3
Study Cited, Design, Primary Outcome, Setting, Quality Rating | Sample Characteristics, Group Size, Study retention | Study Groups & Components | Technology Used | Intervention Duration, # of Intervention Contacts, Intervention Adherence, Interventionist | Primary Outcome |
---|---|---|---|---|---|
Rodgers, A., et al., 2005122
2-group RCT Primary Outcome: 6 wk. abstinence Secondary outcomes: 12 wk. and 26 wk. abstinence Setting: Community Country: New Zealand | N=1705 Int1: n = 853 Int2: n = 852 Women 58.5% Median age (IQR): 22 yrs. (1930) European ethnicity: 63.0% Maori: 20.8% Pacific Islander: 3.5% Other: 12.7% Baseline Fagerstrom Score (median (IQR)) 5(3–6) Mean (SD) of number of CPD was 15 (3). Average previous quit attempts: 2/person Lost to follow up: 6 wks.: Int1:46 Int2: 35 Retention: 95.2% 26 wks.: Int1: 261 Int2: 179 Retention: 74.2% | Int1: quit day established within 30 days, received personalized texts. Ps received free SMSs for one mo. after quit date. Int2: texts related to appreciation for participating, Ps received one month of free SMSs upon completion (not dependent on quit status). Neither group was advised to cease using other resources for quitting smoking. SMS was an add-on to standard treatment. | SMS | Duration: 26 wks. (6 mo.) Contacts: Int1: 5 SMS/day for the first 5 wks. then 3 SMS wkly until end of 6 mo. Int2: one SMS every 2 wks. Follow-up via phone at 6, 12, and 24 wks. for both groups Interventionist: Int1: Automated, tailored SMS Int2: Automated, non-tailored SMS | ITT (assuming missing = smoking) Abstinence (%) 6 wks. Int1: 239 (28%) Int2: 109 (13%), RR 2.2 (95% CI: 1.79–2.70, p <.001) 12 wks. Int1: 247 (20% Int2: 160 (29%) RR 1.55 (95% CI: 1.30 to 1.84), p <.001 26 wks. Int1: 216 (25%) Int2: 202 (24%) RR 1.07(95% CI 0.91 to 1.26), p = n.s. Of 83 Int1 and 42 Int2 selfreported abstainers at 6 wks. asked to provide saliva for cotinine assay, bioverification confirmed abstinence in: Int1: 17 (20.5% Int2: 6 (14.3%) RR 2.84 (95% CI: 1.12–7.16), p = .02 |
Brendryen et al., 2008123
Design: 2-group RCT Outcome: 1, 3, 6, 12 month 7 day no puff self-report abstinence Setting: Community Country: Norway | N=290 Int1: n=144 Int2: n=146 Mean age (SD): Int1: 39.5 (11.0) yrs. Int2: 39.7 (10.8) yrs. Women: 50% Mean (SD) cigarettes smoked per day: Int1: 16.6 (7.2) Int2: 17.6 (7.0) College degree: Int1: 49% Int2: 52% Mean (SD) Nicotine Dependence: Int1:4.5 (2.3) Int2:4.6 (2.2) Retention:77.9% Int1: 81.9% Int2: 74.0% | Int1: Happy Endings (HE) group: received HE (internet and mobile phone smoking cessation program) Int2: received 44-pg self-help book | Email, web pages, IVR, SMS, Craving hotline | Duration: 12 mos. Contacts: 1, 3, 6, and 12-mo abstinence reports Intervention adherence: Number of Web & phone responses 1mo: Int1: n= 139 Int2: n = 127 3 mos: Int1: n=135 Int2: n= 131 6 mos: Int1: n=124 Int2: n=120 12 mos: Int1: n=131 Int2: n=123 Discontinued treatment Int1: n=57 (47%) Interventionist: Int1: Automated Int2: booklet | ITT, Missing assumed = smoking 7-day no puff point abstinence 1mo: Int1: 42% Int2: 17% p = .001 3mo: Int1: 35% Int2: 16% p= .001 6mo: Int1: 29% Int2: 14% p = .002 12mo: Int1: 33% Int2: 23% p = .07 Complete case analysis Repeated point abstinence: 1+3mos: Int1: 30% Int2: 12% p =.001 1+3+6mos: Int1: 24% Int2:7% p =.001 1+3+6+12mos: Int1: 20% Int2: 7% p =.002 |
Brendryen et al., 2008124
Design:2-group RCT Outcome: 1, 3, 6, 12 month 7 day no puff self-report abstinence Setting: Community Country: Norway | N=396 Int1: n= 197 Int2: n= 199 Mean age (SD): Int1 35.9(10.0) Int2: 36.4 (10.5) Women: Int1: 50.8% Int2: 19.8% College degree- Int1: 42.1% Int2: 39.7% FTND: Int1 4.8 ± 2.2 Int2 4.9 ± 2.2 Cpd- Int1: 18.3 ± 5.9 Int2: 18.1 ± 5.8 Pre-cessation self-efficacy- Int1: 4.9 ± 1.3 Int2: 5.1 ± 1.3 Retention: Int1: 88% Int2: 84% | Int1: Happy Endings Internet and cell-phone-based smoking cessation program, 400+ contacts by email, webpages, IVR and SMS Int2: 44 pg. self-help booklet Both groups offered NRT | Email, web pages, IVR, SMS | Duration: 54 wks Contacts: 1, 3, 6, and 12-mo abstinence reports NRT adherence: Int1: 93% Int2: 87% P = n.s. Discontinued treatment: Int1: n=45 (23%) Interventionist: Int1: Automated Int2: NA | ITT, Missing assumed = smoking 7-day no puff point prevalence abstinences: 1mo: Int1: 50.3% Int2: 29.6% p = .001 3 mos.: Int1: 44.7% Int2: 28.6% p = .001 6 mos.: Int1: 37.1% Int2: 21.6 P = .001 12mo: Int1: 37.6% Int2: 24.1% p = .005 |
Free et al., 2009125
Design: 2-group RCT Outcome: 4 week and 6 month self-reported abstinence Setting: Community Country: UK | N =200 Mean age (SD):36 (9) yrs. Women: 38% Median # cigarettes smoked: 20/day Manual occupations: 33% Retention: 92% | Int1: received SMS smoking cessation program (txt2stop) comprised of motivation messages and behavioral-change support. Int2: received SMS messages unrelated to quitting smoking | Mobile Phone SMS | Duration: 6 mos. Contacts: Int1 group received daily SMS starting at randomization with a countdown to quit day and then five messages per day for 4 wks. after the quit day. Intervention continued with a maintenance package of 3 SMS per wk. for 26 wks. Int2 group received simple, short, generic SMS every 2 wks. Intervention adherence: Response rate at 4wks was 96%, 6wks: 92% Interventionist: Int1: Automated Int2: Automated | Completer sample, self-report point prevalence abstinence 4wks Int1: 26% Int2: 13% P = 0.02 RR 2.08 (95% CI 1.11 to 3.89), 6 mos. Int1: 8.5% Int2: 6.7% P = 0.6 |
Free et al., 2011126
Design: Single-blind 2-group RCT Outcome: 6 month biochemically verified smoking abstinence Setting: Community Country: UK | N=5800 Int1: n = 2911 Int2: n = 2881 Women: 45% Mean age (SD): Int1: 36.8 (11.0) Int2: 36.9 (11.1) White: Int1: 89% Int2: 88% Previous quit attempts (1–5 times): Int1:,74% Int2: 76% Fagerstrom score ≤5: Int1: 60%) Int2: 60% Retention: 95% | Int1: SMS txt2stop mobile phone smoking cessation program. Set quit date w/in 2 wks., received 5 SMS/day first 5 wks., then 3/wk. for next 26 wks. Participants can text back “crave”, “lapse”, and receive supportive instant message Int2: received SMS unrelated to quitting, every 2 wks., short, SMS related to the importance of participation. | Mobile phone SMS | Duration: 6 mos. Contact: 4 wks. and 6 mos. Intervention adherence: Received entire intervention Int1: n=2509 Int2: n=2734 Interventionist: Int1: Automated Int2:Automated | ITT, missing data multiply imputed 6 mos. Self-reported continuous abstinence biologically verified by postal salivary cotinine or in person exhaled carbon monoxide: Int1: 10.7% Int2: 4.9% p<.0001 |
Whittaker et al., 2011127
Design: 2-group RCT Outcome: 6 month self-reported continuous abstinence Setting: Community Country: New Zealand | N = 226 Predominantly Maori Int1: n = 110 Int2: n = 116 Mean age (SD): Int1: 27.5 (9.5) Int2:16.6 (7.8) Women: Int1 = 53% Int2 = 42%; Retention: Int1 = 63% Int2 = 78% | Int1: quit date prompt and 2 SMS per day, video messages regarding cessation Int2: quit date prompt and 2 SMS per day, video | SMS and Video messaging to mobile phones; internet | Duration: 12 weeks of Contacts: 1–3 messages per day, reducing to alternating days during maintenance Intervention adherence: 29% used the text “crave” function; 16% used the text “relapse” function requesting assistance Interventionist: Int1: Automated Int2: Automated | ITT. Missing assumed = smoking 6 mos. Continuous abstinence Int1: 26.4% Int2: 27.6% p=n.s. |
Naughton et al., 2012128
2-group RCT Outcomes: 12-week self-reported and cotinine-validated 7-day point prevalence abstinence and cognitive determinants of quitting Feasibility and acceptability of a tailored self-help SC intervention for pregnant smokers (MiQuit) Setting: Community Country: UK | N= 207 Pregnant Int1: n = 102 Int2: n = 105 White: 100% <21 weeks gestation Mean age (SD): Int1: 27.2 (6.4) yrs. Int2: 26.5 (6.2) yrs. 12 week Retention: Int1: n=86 (84%) Int2: n=89 (85%) | Int1: MiQuit sent a four-day, colored, tailored, self-help leaflet via mail and also received tailored SMS Int2: received a non-tailored leaflet via mail Received no tailored SMS, but did receive assessment SMS at 3 and 7 wks. | SMS | Duration: 11 wks. Total contacts: one four-page leaflet for both intervention groups; 2 assessment SMS, one at 3 wks. and one at 7 wks. 3-month follow-up for acceptability, cognitive determinants of quitting, and smoking outcomes. Int1 also received approximately 80 tailored SMS over 11 wks. 0, 1, or 2 SMS were sent daily at various times over 11 wks. Feasibility: 94% (81/86; 95% CI 89%−99%) of MiQuit participants and 80% ( 71/89; 95% CI 71%88%) of controls received both SMS and the leaflet Acceptability: 9% (95% CI 4%15%) of MiQuit participants opted to discontinue SMS Interventionist: Int1: Automated tailored SMS Int2: Automated assessment SMS | ITT. Missing assumed = smoking 12 wks. Self-reported abstinence: Int1 22.9% Int2 19.6%; OR = 1.22, 95% CI 0.62–2.41; p=n.s. Cotinine-validated abstinence Int1 12.5% Int2 7.8%; OR 1.68, 95% CI 0.66–4.31, p =n.s Process outcomes: Int1 more likely to: set a quit date (p= .049), higher levels of self-efficacy (p= .024), harm beliefs (p= .052), and determination to quit (p= .019) |
Ybarra, M., et al., 2012129
2-group RCT Primary Outcome: Bioverified sustained abstinence at 3 mos. Setting: Community Country: Ankara, Turkey | N=151 Int1: n = 76 Int2: n = 75 Mean age (SD) Int1: 36.1 (9.5) yrs. Int2:35.6 (10.3) yrs. Women: Int1: 46.1% Int2: 32.0% Mean CPD (SD): Int1: 18.7 (7.2) Int2: 20.4 (9.2) Fagerstrom score mean (SD): Int1: 4.8 (2.3) Int2: 4.9 (2.5) Retention: Int1: n=46 (61%) Int2: n = 51 (68%) | Int1: 6-wks daily messages aimed at quitting skills. Messages automated except for 2 days and 7 days post quit day in which RAs manually assigned Ps to content “paths” based on whether they had relapsed or had maintained quitting. Int2: 7-page brochure | SMS | Duration: 3 months Int duration: 6 wks Contacts: Int1: Varied by P (dependent upon stage of change and whether relapse occurred. Range is from 912013146.) Int2: no SMS Each group had in-person visits at baseline, 4 wks after quit day, and at 3-mo. F/U Intervention Adherence: NR Interventionist: Int1: Automated + RA manually assigned to content path Int2: NA | ITT Missing assumed = smoking 3-mo. cessation bioverified by carbon monoxide Int1:11% Int2: 5% p= n.s. Secondary outcome: Smoking < 20 Int1: 17% Int2:0% p=0.02 |
Borland et al., 2013130
Design: 5-group RCT Outcome: self-reported continuous abstinence at 6 months Setting: Community | N=3530 Int1: n = 809 Int2: n = 756 Int3: n = 785 Int4: n= 758 Int5: n = 422 | Five conditions: Int1: QuitCoach personalized tailored internet-delivered advice program Int2: onQ, an interactive automated SMS program Int3: an integration of both | Internet and SMS | Duration: 7 mos. Contacts: Int. lasted 7 mos., follow-up surveys at 1 mo. and 7 mos. | ITT assuming missing = smoking, LOCF, and Completer analysis 6-mo sustained abstinence: Int1: 9.0% Int2: 8.7% |
Country: Australia |
Mean age (range): 42.1 (18–80) yrs. Women: 60% Currently smoking: 87.4% Average # cigarettes smoked: 16.9/day Retention: 86.5% | QuitCoach and onQ Int4: a choice of either internet or SMS alone or the combined program Int5: received minimal Int and was offered a simple information website | Intervention adherence: Used intervention: 42.5% Tried it: 14.6% Did not use: 43% Interventionist: Int1: Automated Int2: Automated Int3: Automated Int4: Automated Int5: NA | Int3: 8.7% Int4: 9.1% Int5: 6.2% p = n.s. | |
Buller et al., 2013131
Design: Randomized pretest-posttest two-group design Outcome: 7 day point prevalence self-reported abstinence at 6 weeks, 30 day point prevalence abstinence at 12 weeks Setting: Community Country: US | N=102 Mean age (SD): Int1: 25.5 (NR) yrs. Int2: 24.3 (NR) yrs. Women: Int1: 45% Int2: 57% White: Int1: 70% Int2: 76% Cigarettes smoked per day: Int1: 16.8 Int2: 17.1 Attempted to quit in the past yr: Int1: 66% Int2: 71% Retention: 67% | Int1: Smokers received smartphone application (REQ-Mobile) with interactive tools Int2: assigned to the onQ group which received a SMS system | Smart phone application (REQ- Mobile), SMS system (onQ) | Duration: 12 wks. Contacts: Pretest, 6-wk posttest, and 12-wk posttest smoker reported smoking status Intervention adherence: 60% used allocated service Interventionist: Int1: Interactive online Int2: automated SMS | ITT assuming missing = smoking and completer analyses 6 wks (completer analysis, n=66) 7-day point prevalence abstinence Int1: 30% Int2: 58% p = 0.03 12 wks. ITT 30 day point prevalence abstinence Int1: 18% Int2: 31% p = n.s. completers 30 day point prevalence abstinence Int1: 27% Int2: 46% p = n.s. |
Haug et al., 2013132
Design: 2-group cluster randomized design Outcome: 7-day self-reported abstinence at 6 months Setting: Vocational schools Country: Switzerland | N = 755 in 178 classes Int1: n = 383 in 88 classes Int2: n = 372 in 90 classes Mean age (SD): Int1: 18.2 (2.4) yrs. Int2: 18.3 (2.2) yrs. Women: 49% Smoking status: Occasional = 29%; Daily = 71% Retention at six months: Int1: 79.3% Int2: 71.0% | Int1: Online assessment, weekly SMS assessment, 2 weekly tailored messages, integrated quit day and relapse prevention Int2: No intervention | SMS to mobile phones | Duration: 3 months Contacts: 3 SMS per week Intervention Adherence: 2.4% unsubscribed Mean number of replies to weekly assessment: 6.5 out of a possible 11 possible replies Interventionist: Int1: Automated Int2:Assessment | ITT with 30 imputed data sets mos. day self-reported abstinence Int1: 12.5% Int2: 9.6% OR: 1.03 (0.59 to 1.79), p=n.s. |
Shi et al., 2013133
Design: 2-group cluster randomized design Outcome: 7-day self-reported abstinence at 12 weeks Setting: Vocational schools Country: China | N = 179 in 6 schools Int1: n = 92 in 3 schools Int2: n = 87 in 3 schools Mean age (SD): Int1 = 17.6 (NA) yrs. Int2 = 16.9 (NA) yrs. Women: Int1 = 7% Int2 = 2% Smoking status: Occasional = 29% Daily = 71% Retention at 12 weeks: Int1: 83% Int2: 53% | Int1: Tailored daily SMS based on transtheoretical model Int2: Smoking cessation pamphlet | SMS to mobile phones | Duration: 12 weeks Contacts: daily SMS Intervention Adherence: 87 participants completed the intervention, receiving a median 129 messages and sending a median 32 messages Interventionist: Int1: Automated daily SMS Int2: NA | ITT assuming missing=smoking 12 wks. 7 day self-reported abstinence: Int1: 14% Int2: 8% OR: 1.8 (0.7 to 4.2) |
Ybarra, M.L., et al., 2013118
Design: 2-group RCT Primary outcome: 3-mo.continuous abstinence, verified by significant other Setting: Community Country: New Zealand | N=164 Int1: n = 101 Int2: n = 63 Mean age (SD): Int1: 21.6 (2.1) yrs. Int2: 21.6 (2.1) yrs. Women Int1: 44% Int2: 28% White: Int1: 65% Int2: 41% Retention at 3 months Int1: 81 of 101, (80%) Int2: 51 of 63 (81%) | Int1: 6-wk SMS (Stop My Smoking) intervention provided tailored SMS based on relapse status and quit day date. Included buddy support and craving support Int2: attention-matched control group with similar number of SMS as intervention, but aimed at improving sleep and physical activity. Not tailored to quit day status. Buddy support and craving support not available | SMS | Duration: 3 months Intervention: 6 wks. Contacts: two F/U appts. 2013 one at 6 wks. and one at 3 mo. Varying # SMS sent per day to each group, which was dependent on time point in the study. Interventionist: Int1: Automated SMS + buddy Int2: Automated SMS | ITT with missing assumed = smoking ITT 4 wks. Quit rate: Int1: 39% Int2: 21%, OR = 3.33, 95% CI: 1.48, 7.45 3 mos. Quit rate: Int1: 40% Int2: 30%, OR = 1.59, 95% CI: 0.78, 3.21 |
Abroms et al., 2014119
Design: 2-group RCT Outcome: 6 month biochemically validated point prevalence abstinence Setting: Community Country: US | N = 503 Int1: n = 262 Int2: n = 241 Mean age (SD): 35.7 (10.7) yrs. Women: 66% Average # cigs/day: 17.3 Retention: 76% at six months | Int1: Interactive SMS timed and tailored around the user’s quit date. Int2: receive smokefree.gov site until site included SMS, then changed to Clearing the Air website | SMS; internet | Duration: 3 months push SMS followed by 3 months of SMS on request Contacts: 2 SMS per day on average but up to 5/day around quit date Intervention adherence: 85% received at least 1 SMS Mean of 28 SMS received of those who received at least one Interventionist: Int1: Iinteractive SMS Int2: Automated | ITT, missing assumed=smoking point-prevalence abstinence at 6 months bio-verified by saliva cotinine) Int1: 11.1% Int2: 5.0% Relative risk: 2.22, 95% CI (1.16, 4.26) p<0.05 |
Systematic Reviews and Meta-Analysis | |||||
Whittaker, R., et all, 2009120
Meta-analysis of MEDLINE, EMBASE, Cinahl, PsycINFO, The Cochrane Library, the National Research Register, and ClinicalTrials register Outcome: self-reported point prevalence abstinence | Four trials split into two analyses N1 = 1905 Int1: n1 = 954 Con1: n1 = 951 N2 = 696 Int2: n2 = 348 Con2: n2 = 348 Included smokers of any age who wanted to quit and used any type of mobile phone-based intervention. Retention range for all four studies Int: 69–92% Con: 79–92% | Four studies included (in 5 papers) Used the Mantel-Haenszel Risk Ratio fixed-effect method in which there was no evidence of substantial statistical heterogeneity as assessed by the I(2) statistic | Analysis 1 = SMS Analysis 2 = SMS plus internet | Analysis 1 = studies were 6 mos. duration Analysis 2 = studies were 12 mos. duration Intervention contacts varied by study Intervention Adherence: NR | Analysis 1 = When the studies were pooled, Significant increase in short-term self-reported abstinence(RR 2.18, 95% CI: 1.82.65) Analysis 2 = When the data from the internet and mobile phone programs were pooled, there were significant increases in short- and long-term self-reported quitting (RR 2.03, 95% CI 1.40–2.94) |
Whittaker, R., et al., 2012121
Meta-analysis of the Cochrane Tobacco Addiction Group Specialized Register Outcome: 6 mos. Smoking abstinence, allowing 3 lapses or 5 cigarettes | 5 randomized or quasi-randomized trials N= 9100 Int1: n = 4730 Int2: n = 4370 Retention at 6 mos.: varied across studies. Int1: 68–94% Int2: 78–97% | Used the Mantel-Haenszel Risk Ratio fixed-effect method. There was substantial statistical heterogeneity as indicated by I(2) statistic I(2) = 79% | 3 studies used SMS, which was adapted over the course of the studies for different populations and contexts. One multi-arm study used SMS intervention and an internet QuitCoach separately and in combination. One Study used video messaging delivered via mobile phone | Study duration:≥6 mos. Adherence rates: NR | Mobile phone interventions increase long-term quit rates compared to control programs at 6 mos. (RR 1.71, 95% CI: 1.471.99, > 9000 participants) |
Note: CDS = Cigarette Dependence Score, CI = Confidence Interval, CO = Carbon Monoxide, CPD = Cpd, FNDS = Fagerstrom nicotine dependence scale, Int: Intervention group, Con: Control group, IVR= interactive voice response, aOR = adjusted Odds Ratio, OR = Odds Ratio, NRT = Nicotine Replacement Therapy, mo = month, mos. = months, RA = research assistant, NA=not applicable, P = participant, Ps = participants, RCT = Randomized Controlled Trial, SC = Smoking Cessation, SD = Standard Deviation, SMS = Short Message Service, wk = week, wks = week, NR = not report, Automated = without a clinician who generates, tailors, or modifies the output; ITT = intention to treat, LOCF = last observation carried forward, Δ = change or difference
SMS for Smoking Cessation.
Most of the research on mobile smoking cessation interventions has focused on text messaging as the delivery medium. Why SMS when there are so many other delivery mediums on today’s smartphone? First, many of the early studies using mobile phones for smoking cessation122,134 predate the advent of the smartphone; hence SMS was one of the few functions available on feature phones for the delivery of interventions. Second, SMS is a relatively inexpensive development environment that will run on any cell phone whereas a smartphone app needs to be developed for each operating system (e.g. Android, iOS) and updated with each operating system update. Third, although smartphone use is increasing dramatically and is now above 50 percent in the U.S.19, smartphone use was reported as lowest by adults in lower socioeconomic groups.135 Smoking rates are disproportionately higher in lower socioeconomic groups136 that remain predominately feature phone users. Recent PEW statistics, however, show that Hispanics and African Americans have higher rates of smartphone use than Whites, indicating the demographic shift in mobile phone use that could make smartphone apps a viable medium for cessation interventions targeting minorities.20
Cochrane Meta-analysis.
Controlled studies of mobile phone programs for smoking cessation have been summarized in a Cochrane meta-analysis 120 and updated. 121 The details of studies reviewed in these two meta-analyses are listed in Table 3. For both reviews, the primary outcome was smoking abstinence of six months or longer and included both sustained and point prevalence abstinence and both self-reported and biochemically validated smoking status. However, the number of studies reviewed was four and five, respectively, and there was considerable effect heterogeneity across studies.
The initial Cochrane review 120 identified nine articles relevant to smoking cessation via mobile phones in which the mobile intervention was a core component, not just an adjunct to an internet or in-person program. Of these, four were small, non-randomized feasibility trials, and two had insufficient follow-up for inclusion. Of the four studies included in the meta-analysis, two assessed the same text messaging program delivered in two different countries122,125 and the remaining two trials 123,124 evaluated a combined internet and mobile phone intervention. The four studies lacked long-term follow-up or biochemical validation in more than a small subsample of participants, but all four studies showed significantly greater abstinence at six months compared to controls (see Table 3 for details),.
In the 2102 update of the Cochrane review, the two Norwegian studies were subsequently excluded due to the considerable non-mobile aspects of the intervention, but three trials published since the initial review were added: a video messaging mobile phone intervention 121; a web-based quit coach and text messaging intervention130 and a large scale evaluation of an SMS or text messaging intervention 126. Pooled across these five total studies, the RR was 1.71. Among the studies reviewed in the Cochrane update, the large and well-controlled United Kingdom study by Free and colleagues126 accounted for over half (50.45%) of subjects in the meta-analysis. In this single blind trial, 5800 smokers willing to quit were randomly assigned to either a mobile phone text program (txt2stop) that included behavior change support and motivational messages or to a control group that received SMSs unrelated to quitting smoking. Based on biochemically verified continuous abstinence at six months, quit rates were significantly greater in the txt2stop (10.7%) than the control group (4.9%), and the abstinence rates were similar when those lost to follow-up were treated as smokers. Since the Cochrane update in 2012, there have been a number of RCTs of smoking cessation programs delivered via mobile phone technologies, and these are listed in Table 3.
Special Populations.
There are limited intervention options for pregnant smokers. In a preliminary trial comparing smoking cessation programs in pregnant smokers128, there were no significant differences in self-reported smoking abstinence between groups who received educational materials and tailored SMS. Further research is needed to identify minimal risk interventions that are effective for pregnant smokers.
Likewise, low-education young adults are a particularly vulnerable population for smoking that warrant additional research on both prevention and cessation interventions. Two recent studies132, 133 compared the effectiveness of technology-based smoking cessation interventions to educational pamphlets in adolescent vocational students. Neither study reported significant differences in self-reported abstinence after intervention between groups receiving text-messaging interventions or paper-based educational materials, however the sample sizes in these two studies may have been inadequate to detect differences.
Recent Studies in U.S.
Ybarra and colleagues first studied an SMS program delivered in Turkey 129 and more recently studied the effects of their SMS intervention in a study of young adult smokers in the U.S. 118 Although the intervention produced significantly higher abstinence rates at four weeks, these differences were not sustained at 3 months.
Abroms and colleagues recently published a controlled trial of Text2Quit, an automated, tailored, and interactive text messaging program for smoking cessation.119 In contrast to many previous programs which primarily push out texts, the Text2Quit program is interactive and prompts users to track smoking and report cravings. Via keyword texts, users have the ability to reset quit dates, request help with a craving, get program and data summaries, and indicate if they have slipped and smoked. Mailed saliva cotinine verified point prevalence abstinence at 6 months, showed an 11% abstinence rate for intervention vs. 5% for controls. In contrast to the studies and programs in the earlier Cochrane reviews, this study was conducted in the U.S. and evaluated a program that is commercially available to smokers in the U.S.
Gaps and Recommendations for Future Research.
There is substantial evidence that mobile phone apps for smoking cessation, particularly SMS programs, are effective for smoking cessation. The effects found for mobile phone smoking cessation interventions are comparable to the effects found from other smoking cessation interventions, including nicotine replacement therapies.137 The considerable heterogeneity of this evidence, however, suggests that not all text messaging programs are created equal, and that there is considerable individual variability in response to these programs. Therefore, although these text messaging programs have sufficient empirical support to be recommended to patients interested in quitting smoking, the selection of text messaging intervention may matter. Unfortunately, many of these empirically-supported text messaging programs were developed and evaluated outside of the U.S. and are not available, commercially or otherwise, to U.S. smokers. This lack of U.S. access to proven text messaging programs is beginning to change. Abroms and colleagues recently published evidence for their Txt to Quit program which is commercially available.119
Although there are hundreds of smartphone applications for smoking cessation commercially available, there is considerable evidence that these applications have a limited empirical basis115,116 and we could find no published study testing the efficacy of any of these commercially available smartphone apps. In the one study that compared smartphone apps to text messaging131, text messaging produced better quit rates. While it is clearly premature to recommend any smartphone application for smoking cessation at this time, smartphone applications hold potential future promise as smoking cessation interventions. Smartphones provide a range of potential features and functions not available via text messaging modalities that have not been adequately leveraged to date for smoking cessation. For example, movement and location sensors in smartphones could be used to learn the contexts in which users smoke and deliver interventions preemptively before the urge to smoke occurs114. Sensors connected to smartphones, such as carbon monoxide monitors 138, provide objective measures of smoking status. Another promising approach that builds on the use of smartphones is ecological momentary interventions (EMIs), an approach to the delivery of interventions to people during their everyday lives (i.e. in real time) and in natural settings (i.e. real world).139 This approach is gaining increasing attention as a potential approach across multiple behavioral domains and was tested in an earlier study for smoking cessation with significantly higher quit rates in the intervention group than the control group at 6 and 12 weeks but was not sustained at 26 weeks.122,139 Ongoing studies are testing this approach with the currently available smartphone technology.
One critical but inadequately researched area is how to engage smokers to initiate the use of these mobile phone SMS programs. The well-controlled, population-based, multi-arm trial of Borland and colleagues 130 had less than half of the intervention participants engage with the intervention on even a minimal basis. The follow-up study by Riley and colleagues in which participants were assisted with program initialization 140 was due to the findings of an earlier trial in which 37% of the participants who completed baseline measures but failed to initialize the SMS program on their own.134 Bock and colleagues conducted focus groups on preferences for a SMS-based smoking cessation program from potential users. Participants recommended including social networking components, greater control of program output via online profile, and more interactive text messaging features. In parallel with research on the efficacy of these mobile phone programs for those who engage with them, research on how to engage smokers and keep them engaged in these programs also needs to occur.
Summary and recommendations.
Smoking cessation via mobile phone intervention is a relatively young area of research with only 10 years of published literature. Within this short period, however, a number of large and well-controlled studies have shown that SMS programs produce approximately double the abstinence rates of minimal intervention control conditions. Despite this success, the failure rate from these programs is still unacceptably high (approximately 90% fail to quit at six months) and the heterogeneity of effect across studies suggests that certain varieties of SMS interventions may work better than others, and in certain populations differentially from others. Until more is known on optimal intervention components of SMS for smoking cessation, and on which smokers are more likely to benefit from these approaches, the current literature is only able to support that SMS interventions should be considered along with other efficacious smoking cessation interventions for smokers trying to quit.
Use of mHealth for Self-Management of Diabetes
Diabetes occurs in 9.3% of the US population (29.1 million persons). Of increasing concern is the number of US adults with undiagnosed diabetes (8.1 million) or pre-diabetes (86 million).141 CVD and stroke are serious complications of diabetes. The majority of US adults 18 years and older with diabetes have CVD risk factors, including high blood pressure (71%), high cholesterol (65%)141, and obesity (70%).142 Although death rates for heart attack and stroke have decreased, adults with diabetes are twice as likely to be hospitalized and die from these diseases as people who do not have diabetes. Because people with diabetes are living longer, the prevalence of obesity is not abating, and the rate of diagnosed new cases is increasing (7.8–12.0 per 1000 in 2012 depending on age), scientists expect that the number of people with diabetes and CVD to continue to rise. However, since the rates of survival after heart attack and stroke continue to improve, more persons with diabetes will continue to live into older age with comorbidities of CVD and diabetes. According to the joint statement of the American Heart Association (AHA) and the American Diabetes Association (ADA), glycemic control in diabetes management for both type 1 and type 2 diabetes is important in risk reduction for CVD events. A1c is the clinical measure of glycemic control and the self-monitoring of blood glucose (SMBG) is done by the consumer (patient). A general population target of A1C <7% is recommended for clinician consideration and health plan targets, but an individualized approach to glycemic control at the patient level is suggested. It is important to note that the consumer (patient) role in glycemic control requires problem-solving and daily decisionmaking about multiple behaviors (eating, activity, monitoring, and medication taking) and the healthcare provider role is collaborating with the patient to prescribe the appropriate diabetes medication(s) and monitoring the impact.143
Consumer/patient perspective.
There are thousands of mobile applications for supporting diabetes self-management, primarily serving as tracking and reference apps. Few have been evaluated and even fewer have demonstrated outcomes.131 However, less than one percent of mobile applications have been evaluated through research. It can be hoped that increased federal and private foundation investments in mHealth, and behavioral, clinical, and health system interventions in combination with new regulatory requirements, will provide consumers and providers with evidence of effectiveness or what works.
A number of pharmacologic and lifestyle interventions for diabetes management have been confirmed by multiple RCT’s, however only 48.7% of patients meet the A1c, blood pressure, and lipid goals for diabetes care and only 14.3% meet these 3 measures and also the goals for tobacco use.131 The National Standards for Diabetes Self-Management Education/Support, jointly published by the American Diabetes Association (ADA) and the American Association of Diabetes Educators (AADE) incorporate the AADE7 self-care behaviors (physical activity, healthy eating, taking medication, monitoring, self-management problem-solving, reducing risks, and healthy coping) as essential behaviors for improving diabetes self-management. 144–146
Mobile technologies for diabetes self-management can be categorized in the following way: SMS apps via mobile phone, diabetes medical devices (e.g., blood glucose meters, insulin pumps) with connectivity to smartphone apps, and bi-directional data sharing between patients and providers using smartphones. This classification did not exist when most of the reviewed articles were published. Interventions delivered via mobile technologies and directed at consumers may be supported by behavior change theories or principles, e.g., self-efficacy theory. However, most studies have limited theoretical foundations or lack an empirical basis. Moreover, health care providers lack knowledge about what apps are available or how to evaluate them and thus are hesitant to recommend them.13
Although large, primary care, RCTs of mobile diabetes management are limited, smaller studies addressing feasibility, usability and acceptability have generally identified the following components as essential to successful diabetes management: personalized engagement, provision of actionable feedback for consumers, and connection with providers and/or health care systems. Additional contributors to usability include mobile technologies to support community health workers and peer supported self-care behaviors.147
Review of evidence for efficacy of mobile technology-based interventions to promote self-management of diabetes.
We searched PubMed for the years 2004 to 2014, using the terms type 2 diabetes; NIDDM; maturity onset diabetes; adult onset diabetes; non-insulin dependent; noninsulin dependent; slow onset diabetes; stable diabetes; hyperinsulinemia; hyperinsulinism; insulin resistance; hyperglycemia; glucose intolerance; metabolic syndrome; metabolic X syndrome; dysmetabolic syndrome; metabolic cardiovascular syndrome. These terms were cross-referenced with the mobile technology terms described previously. This search resulted in 242 identified articles. Of these, 83 were not relevant to the use of mobile technology with diabetes, 159 were reviewed further. Of these 159 references were identified, 142 were excluded based on review of title, abstract, and full text. Similar to other sections of this review, mobile technologies may target multiple behaviors singly or in combination to improve numerous clinical and behavioral outcomes. Therefore, for this review we focused on studies with change in the clinical metric of HbA1c as the primary outcome, considered the gold standard in diabetes improvement. Seventeen articles were eligible for this review and ten of these were international studies.
The types of mobile technologies used for diabetes self-management research interventions include mobile platforms, with diabetes specific software apps or SMS. Table 4 provides the detail of the RCTs using these mobile tools that we reviewed.
Table 4
Study Cited, Design, Outcome, Setting, | Sample Characteristics, Group Size, Baseline HbA1c, Study Retention | Study Groups & Components | Technology used | Intervention Duration, # of Intervention Contacts, Intervention Adherence,Interventionist | Primary Outcome: HbA1c (%, or %change) |
---|---|---|---|---|---|
Kim, H. S., 2007154,163 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N=60 Int1: n = 30 Int2: n = 30 Mean age (SD) : Int1: 46.8 (8.8) yrs. Int2: 47.5 (9.1) yrs. Women: 56.9% HbA1c (%), M (SD): Int1: 8.1 (1.7) Int2: 7.6 (1.1) Retention: 85% | Int1: Ps tracked their blood glucose levels and medications on a web portal, and received weekly feedback from a diabetes nurse Int2: usual care | SMS with web based tracking of glucose levels | Duration: 6 mos. Contacts: Int1: weekly feedback via SMS Int2: 1–2 times during the 6 mos. Intervention adherence: NR Interventionist: Int1: Diabetes nurse Int2: Clinician | Completer’s analysis (n=51) 3 mos. HbA1c (%), M (SD): In1t: 6.9 (1.0) Int2: 7.7 (0.9) p<.05 6 mos. HbA1c (%), M (SD): Int1: 7.0 (1.4) Int2: 7.7 (0.9) Group x time: p=.008 |
Faridi, Z., et al. 2008152 Design: 2-group RCT Outcome: Δ in HbA1c Setting: Community Country: US | N=30 Int1: n = 15 Int2: n = 15 Mean age (SD): 56 (9.7) Women: 63% White: NR HbA1c (%), M (SD): Int1: 6.4 (0.6) Int2: 6.5 (0.7) Retention: 13% | Int1: 1-day training and 3-mos intervention using the Novel Interactive mobile-phone technology for Health Enhancement (NICHE) system (transmits glucometer and pedometer data to online server which then transmits tailored feedback to Ps via text messaging). Int2: continued standard diabetes self-management and tracked step count with pedometer. | Internet and SMS | Duration: 3 mos. Contacts: 1-day training workshop on NICHE device; Ps required to upload once daily glucose and pedometer data daily and receive tailored SMS messages Intervention adherence: 13.3% completely adherent; 26.7% adherent for 1–2 months; 26.7% adherent for 1 week; 33.3% did not transmit any information. Interventionist: Int1: Nurse practitioners Int2: NR | ITT 3 mos. HbA1c Δ, %, M (SD) Int1: −0.1 (0.3) Int2: 0.3 (1.0) p=NS |
Kim, H.S., et al., 2008 and Kim, S.I. et al., 2008155,164 Design: 2-group RCT Outcome: HbA1c (%) Setting: Outpatient clinic Country: South Korea | N=40 Int1: n =20 Int2: n = 20 Mean age (SD) : Int1: 45.5(9.1) yrs. Int2: 48.5(8.0) yrs. Women: 52.9% White: NR HbA1c (%), M (SD): Int1: 8.1(1.9) Int2: 7.6(0.7) Retention: 85% | Int1: Ps recorded daily glucose values in web portal. Received weekly SMS feedback from diabetes educator Int2: usual care | SMS feedback based on web-based tracking portal | Duration: 12 mos. Contacts: Int1: weekly feedback via SMS Int2: contact at 3 and 6 mos. Intervention adherence: NR Interventionist: Int1: Diabetes physician + diabetes educator Int2: Diabetes physician + diabetes educator | Completer’s analysis (n=34) 6 mos. HbA1c (%), M (SD): Int1: 7.1 (1.5) Int2: 7.7 (0.5) Group x time: p=.04 12 mos. HbA1c (%), M (SD): Int1: 6.7 (0.8) Int2: 8.2 (0.5) Group x time: p=.02 |
Yoon, K., et al., 2008165 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N=60 Int1: n = 30 Int2: n = 30 Mean age (SD) : Int1: 46.8(8.8) yrs. Int2: 47.5(9.1) yrs. Women:56.9 % HbA1c (%), M (SD): Int1: 8.1(1.7) Int2: 7.6(1.1) Retention:85.0% | Int1: completed self-mon blood glucose levels, entered values and medication info on a webpage; this information used to tailor recommendations to Ps. Tailored messages sent via SMS and internet weekly. Medication adjustments communicated to the Ps’ physician. Int2: met with endocrinologist in person at an outpatient clinic and was given basic information | SMS and Web | Duration: 12 mos. Contacts: baseline and post-test assessments; blood draws at baseline, 3, 6, 9, and 12 mos. Int1 group had 52 messages over one year Int2: same assessment time points as Int1, but in-person contact at outpatient clinic was variable per Ps Intervention adherence: Assessment Attendance (completed post-test), %: Int1: 83.3% Int2: 86.7% Interventionist: Int1: Physicians and nurses Int2: Endocrinologist | Completer’s analysis (n=51)
12 mos. HbA1c (%), M (SD): Int1: 6.8 (0.8) Int2: 8.4 (1.0) p=.001 |
Istepanian, R.S.H., et al., 2009166 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: UK | N= 137 Int1: n = 72 Int2: n = 65 Mean age (SD) : Int1: 60 (12) yrs. Int2: 57 (13) yrs. Women: NR White: 34% HbA1c (%), M (SD): Int1: 7.9 (1.5) Int2: 8.1 (1.6) Retention: 64% | Int1: SMBG via Bluetooth upload, data were reviewed by research team, analysis sent via mail to Ps and PCP. Ps had hotline access to research team for questions Int2: standard care | Glucometer adapted to send data via Bluetooth to mobile phone | Duration: 9 mos. Contacts: Int1 P’s blood glucose measurements transmitted wirelessly; research clinicians sent letters to Ps and their providers with treatment recommendations Intervention adherence: NR Interventionist: Int1: Clinicians Int2: Clinicians | ITT
Mean 9 mos. HbA1c (%), M (SD): Int1: 7.9 (NR) Int2: 8.2 (NR) p=.17 Completer’s analysis (n=87) HbA1c (%), M (SD): Int2: 7.8 (NR) Int2: 8.4 (NR) p=.06 |
Rodríguez-Idígoras, M.I., et al., 2009167 Design: 2-group RCT Outcome: HbA1c Setting: Community Country: Spain | N = 328 Int1: n = 161 Int2: n = 167 Mean age (95%CI): Int1: 63.3 (61.6, 65.0) Int2: 64.5 (63.0, 66.1) Women: 48% White: NR HbA1c (%), M (95%CI): Int1: 7.6 (7.4, 7.9) Int2: 7.4 (7.2, 7.6) Retention: 91% | Int1: Ps provided mobile phone and tele-assistance system (DIABECOM,) using real-time transmission of blood glucose results, with immediate reply when necessary, and telephone consultations, Int2: standard clinical care | Mobile phone, teleassistance system | Duration: 12 mos. Contacts: Ps made average of 3calls/month; average of 2.6 reminder/follow-up calls from call center. Intervention adherence: Use of teleassisstance system (%): Int1: 62% Int 2: NA Interventionist: Int1: Physician and a nurse specializing in diabetes and diabetes education Int2: NR | ITT (n=321) 12 mos. HbA1c (%), M (95%CI) Int1: 7.4 (7.2, 7.6) Int2: 7.4 (7.1, 7.6) p=.34 |
Yoo, H.J., et al., 2009160
Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N= 123 Int1: n = 62 Int2: n = 61 Mean age (SD) : Int1: 57.0 (9.1) yrs. Int2: 59.4 (8.4) yrs. Women: 47.2% HbA1c (%), M (SD): Int1 7.6 (0.9) Int2: 7.4 (0.9) Retention: 90.2% | Int1: Ubiquitous Chronic Disease Care (UCDC) system using mobile phones and web-based interaction. UCDC included device attached to mobile phone that transmitted blood glucose data. Ps received SMS reminder to check blood glucose, also tips via SMS 3 x’s/day. Physicians could follow the Ps’ data and send individualized messages as needed. Int2: Usual Care. Ps visited according to usual schedule and received usual care in the outpatient setting | SMS and internet | Duration: 3 mos. Contacts: Int1: two alarms daily to remind pts to measure blood glucose values and blood pressure as well as one alarm daily for weight. Additionally, each Ps received at least three SMS daily Int2: Dependent upon usual care routine Each Ps was seen at baseline and at 3mo to collect anthropometric as well as laboratory data Intervention adherence: Int1: sent in glucose readings 1.84 ± 0.31 times per day with a compliance rate of 92.2 ± 15.4% Blood pressure readings sent in 1.72 ± 0.32 times per day with a compliance rate of 86.0 ± 16.2% Weight measurements were sent in 0.87 ± 0.20 times per day with a compliance rate of 87.4 ± 20.1% Interventionist: Int1: Automated Int2: Physician | Completer’s analysis (n=111) 3 mos. HbA1c (%), M (SD): Int1: 7.1 (0.8) Int2: 7.6 (1.0) Group x time p=.001 |
Kim, C., et al., 2010153 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N=100 Int1: n = 50 Int2: n = 50 Mean age (SD) : Int1: 47.8 (9.6) yrs. Int2: 49.0 (10.7) yrs. Women: 50% White: NR HbA1c (%), M (SD): Int1: 9.8 (1.3) Int2: 9.8 (1.2) Retention: 92% | Int1: received daily insulin dose adjustments via SMS based on logged data sent via mobile phone to website Int2: self-adjusted basal insulin according to daily self-monitored capillary FBG measurements using glucometers | SMS with web tracking | Duration: 12 wks. Contacts: Ps dose adjustments were reviewed by the investigator at 4- and 8-wk clinical visits. Intervention adherence: # checks of blood glucose monitoring: Int1: 51.8 (16.1) checks Int2: 42.2(13.2) checks p = .002 Interventionist: Int1: Automated Int:2: NR | Completer’s analysis (n=92) 12 wks. HbA1c (%), M (SD): Int1: 7.4 (0.7) Int2: 7.8 (0.8) p=.02 Δ in weight (kg), M (SD): Int1: 2.4 (3.0) Int2: 2.2 (2.8) p=.65 |
Noh J.H., et al., 2010158 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N=44 Int1: n = 24 Int2: n = 20 Mean age (SD) : Int1: 42.5 (10.6) yrs. Int2: 42.3 (7.6) yrs. Women: 22.5% White: NR HbA1c (%), M (SD): Int1: 9.0 (2.3) Int2: 8.6 (1.2) Retention: 90.9% | Int1: electronic Management of Diabetes (eMOD ), a web-based ubiquitous information system, for mobile phone users along with a website for Internet users to provide diabetes education. Int2: educational books with similar contents as eMOD website | eMOD mobile and web application for diabetes education | Duration: 6 mos. Contacts: all Ps visited their physicians every 2 mos. Intervention adherence: Int1: eMOD system was accessed via computer 160 times during the study period Interventionist: Int1: Physicians Int2: Physicians | Completer’s analysis (n=40) 6 mos. HbA1c (%), M (SD): Int1: 7.5 (1.4) Int2: 8.1(0.3) p=.23 |
Carter, E.L., et al., 2011168 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: US | N = 74 Mean age (SD): Int1: 52 (NR) Int2: 49 (NR) Women: 64% African American: 100% HbA1c (%), M (SD): Int1: 9.0 (NR) Int2: 8.8 (NR) Retention: 64% Int1: n = 26 Int2: n = 21 | Int1: Ps were provided laptop with peripherals (scale, BP cuff, glucometer) with automatic transmission to internet; biweekly video conferencing with nurse; access to internet-based self-management module with tailored action plan, health education module and social networking module Int2: standard clinical care | Internet, wireless scales, BP cuffs, and glucometers | Duration: 9 mos. Contacts: Ps weigh daily, check BP weekly, SMBG three times daily; biweekly 30-minute video conferences with telehealth nurse to develop tailored action plan Intervention adherence: NR Interventionist: Int1: Nurse Int2: NR | Completer’s analysis (n=47) 9 mos. HbA1c (%), M (SD) Int1: 6.8 (NR) Int2: 7.9 (NR) p<.05 Δ in weight (lb), M: Int1: −73.0 Int2: −58.1 p<.05 Δ in systolic BP, M: Int1: −7 Int2: −8 p>.05 Δ in diastolic BP, M: Int1: −15 Int2: −14 p>.05 |
Lim, S., et al., 2011156 Design: 3-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N=154 Int1: n = 51 Int2: n = 51 Int3: n = 52 Mean age (SD) : Int1: 67.2 (4.1) yrs. Int2: 67.2 (4.4) yrs. Int3: 68.1 (5.5) yrs. Women: 55.8% HbA1c (%), M (SD): Int1: 7.8 (1.0) Int2: 7.9 (0.9) Int3: 7.9 (0.8) Retention: 93.5% | All Ps were standardized with diabetes education. Int1: usual-healthcare: SMBG + SMS feedback Int2: SMBG Int3: usual care | SMS, Gluco Dr Supersensor, AGM- 2200, Allmedicus | Duration: 6 mos. Contacts: All Ps visited the outpatient clinic every 3 mos. for an interview conducted by their physician and provided a blood sample Intervention adherence: Frequency of SMBG, number/week: Int1: 10.5(5.1) Int2: 8.2(4.2) Int3 2.4(3.3) Interventionist: Int1: Automated + specialized diabetes management team consisting of welltrained professionals, including diabetologists, nurses, dietitians, and exercise trainers, organized and directed patient education Int2: Specialized diabetes management team consisting of well-trained professionals, including diabetologists, nurses, dietitians, and exercise trainers, organized and directed patient education Int3: NR | Completer’s analysis (n=144) 6 mos. HbA1c (%), M (SD): Int1: 7.4 (1.0) Int2: 7.7 (1.0) Int3: 7.8 (1.0) p<.05 (Int1 vs. Int2 and Int1 vs. Int3) |
Quinn, C. et.al, 2008169
Design: RCT Outcome: Δ in HbA1c (%) Setting: Primary care practices Country: US | N=30 Int1: n=NR Int2: n=NR Mean age (SD): 51.04 (11.03) yrs. Women: 65% African American: 62% | Int1: mobile phone-based diabetes management software system used with web-based data analytics and therapy optimization tools Int2: Usual care by PCP | Mobile phone | Duration: 3 mos. Total contacts: Baseline, 3-mo followup Intervention adherence: NR Interventionist: Int1: Health care providers Int2: PCP | Complete analysis (n=26) Δ in HbA1c (%), M (95%CI): Int1: 2.03% Int2: 0.68% (P < 0.02, one-tailed) |
Quinn, C., et al., 2011151 Design: Cluster- RCT Outcome: Δ in HbA1c (%) Setting: Primary care practices Country: US | N = 213 Int1: n = 62 Int2: n = 38 Int3: n = 33 Int4: n = 80 Mean age (SD) : Int1: 53.2(8.4) yrs. Int2: 52.8(8.0) yrs. Int3: 53.7(8.2) yrs. Int4: 52(8.0) yrs. Women: 44.2% White: 52.8% HbA1c (%), M (SD): Int1: 9.2(1.7) Int2: 9.3(1.8) Int3: 9.0(1.8) Int4: 9.9(2.1) Retention:76.5 % | Int1: coach-only (CO). Ps received educational and motivational messages after putting data into the phone. P also received supplemental electronic messages within the application, generated by “virtual educators” based on longitudinal data trends Int2: coach PCP portal (CPP) Same as CO, except PCP was able to view raw data and discuss with the Ps Int3: coach PCP portal with decision-support (CPDS): Same as CO, except PCPs received Ps analyzed data that summarized patient’s glycemic and metabolic control, adherence to medication, self-management skills, related to evidence-based guidelines and standards of care. Int4: usual care | App designed for DM management, web portal | Duration: 12 mos. Total Contacts: Baseline, 12-mo follow-up. Charts reviewed for HbA1c at 3, 6, and 9 mos. Intervention adherence: NR Interventionist: Int1: Coach Int2: Coach diabetes educators+ PCP Int3: Coach diabetes educators+ PCP Int4: Usual care clinicians | Completer’s analysis (n=163) 12 mos. Δ in HbA1c (%), M (95%CI): Int1: −0.7 (−1.1, −0.3) Int2: −1.6 (−2.3, −1.0) Int3: −1.2 (−1.8, −0.5) Int4: −1.9 (−2.3, −1.5) p=.001 (Int4 vs. Int1) p=.02 (Int2 vs. Int1) p=.40 (Int3 vs. Int1) Δ in systolic BP (mmHg), M (95% CI): Int1: 2 (−3, 7) Int2: 4 (−4, 11) Int3: 2 (−6, 10) Int 4: −2 (−6, 3) p>0.05 Δ in diastolic BP (mmHg), M (95% CI): Int1: 1 (−2, 4) Int2: 2 (−2, 7) Int3: −2 (−6, 3) Int4: −1 (−4, 2) p>0.05 |
Orsama, A.L., et al., 2013159 Design: 2-group RCT Outcome: Δ in HbA1c (%) Setting: Community Country: Finland | N= 55 Int1: n = 29 Int2: n = 26 Mean age (SD) : Int1: 61.5(9.1) yrs. Int2: 62.3(6.5) yrs. Women:45.8 % HbA1c (%), M (SD): Int1: 7.1(1.5) Int2: 6.9(1.6) Retention:87.3 % | Int1: Ps participated in remote patient reporting of health status parameters and linked health behavior change feedback (called Monica) Int2: received standard of care including diabetes education and healthcare provider counseling. | Internet, mobile phone | Duration: 10 mos. Contacts: Int1: Ps received real time feedback Intervention adherence: NR Interventionist: Int1: Automated + healthcare provider Int2: healthcare provider | Completer’s analysis (n=48) 10 mos. Δ in HbA1c (%), M (95%CI): Int1: −0.40 (−0.67, −0.14) Int2: 0.04 (−0.23, 0.30) p=.02 Δ in weight (kg), M (95%CI): Int1: −2.1 (−3.6, −0.6) Int2: 0.4 (−1.1, 1.9) p=.02 Δ in systolic BP (mmHg), M (95% CI): Int1: −13.5 (−21.3, −5.8) Int2: −17.1 (−24.3, −9.9) p=.51 Δ in diastolic BP (mmHg), M (95% CI): Int1: −7.3 (−10.9, −3.8) Int2: −9.5 (−12.9, −6.2) p=.38 |
Forjuoh, S. N., et al., 2014170 Design: 4-group RCT Outcome: Δ in HbA1c Setting: Community Country: US | N=376 Int1: n = 101 Int2: n = 81 Int3: n = 99 Int4: n = 95 Mean age (SD): 57.6(10.9) Women: 55% White: 64% HbA1c (%), M (SD): Int1: 9.4 (1.7) Int2: 9.3 (1.6) Int3: 9.2 (1.4) Int4: 9.2 (1.6) Retention: 70% | Int1: CDSMP--Chronic Disease Self-Management Program Int2: PDA Int3: CDSMP+PDA Int4: Usual care | PDA with Diabetes Pilot Software | Duration: 12 mos. Contacts: Int1: 6 week, 2.5 hr/wk. classroom-based program for diabetes self-management Int2: Diabetes Pilot software on a PDA (with training; software tracks glucose, BP, medications, physical activity and dietary intake) Intervention adherence: Attendance (4 of 6 sessions): Int1: 75.6% Int3: 72.7% # entries/yr: Int2: 342 Int3: 359 Interventionist: Int1: NR Int2: NR Int3: NR Int4: NR | Completer’s analysis (n=263) 12 mos.
HbA1c Δ, %, M (SD) Int1: −0.7 (NR) Int2: −1.1 (NR) Int3: −0.7 (NR) Int4: −1.1 (NR) p=.77 |
Systematic Reviews and Meta-Analysis | |||||
Liang, X., et al., 2011161 Design: Metaanalysis of 22 clinical trials Outcome: Δ in HbA1c (%) | N=1657 Mean age (SD): 44 (18) yrs. Women: 45% White: NR | Studies on impact of mobile phone intervention on diabetes self-management | SMS to deliver blood glucose test results and self-management information | Duration: median 6 mos. (range 3–12 mos.) | Median 6 mos. Pooled Δ in HbA1c (%), M (95%CI): −0.5 (−0.3, −0.7), indicating that the reduction of HbA1c value was 0.5% lower in mobile Int groups compared with other Int groups Subgroup analysis showed greater Δ in HbA1c in type 2 than type 1 DM (−0.8% vs. −0.3%, p=0.02) |
Pal et al., 2013162 Design: metaanalysis of data from 11 trials Outcome: HbA1c (%) | N = 3578 Mean age: 46–67 yrs. Time since dx: 6 −13 yrs. | Assess the effects on health status and health-related quality of life of computer-based diabetes self-management interventions for adults with T2DM | Computer-based interventions | Duration: ranged from1–12 mos. | Based on 2637 Ps; 11 trials: Pooled effect on HbA1c: 0.2% (95% CI: −0.4 to −0.1) p = .009 Based on 280 Ps; 3 trials. The effect size on HbA1c was larger in the mobile subgroup: mean difference in HbA1c −0.5% (95% CI −0.7 to −0.3); p < .00001 |
Note: P or Ps= participant(s), N = total sample, n = subgroups, Int = Intervention group, 1, 2, etc. Con = control group, RCT = randomized control trial, Δ = change or difference, mo. = month, mos. = months, wk = week, wks = weeks, years = yrs., Baseline = 0, SMS = short message service, NA = not applicable, NR = not reported, DM = Diabetes Mellitus, PCP = primary care physician, SMBG = Self-monitoring blood glucose, T2DM= type 2 diabetes mellitus, Self-mon = self-monitoring, BP=blood pressure; automated = without a clinician who generates, tailors, or modifies the output.
When evaluating interventions, we considered a HbA1c reduction of at least 0.3% as a clinically meaningful treatment effect148 and 1% decrease in HbA1c as a clinically meaningful indicator of reduced risk of diabetes complications based on the DCCT and UKPDS clinical trials.149,150 One US study151 reported a significantly greater HbA1c decrease in the intervention group than in the control group. Quinn evaluated a mobile phone software application with a patient and provider web portal.151 The average HbA1c decline over the one-year intervention was 1.9% for the intervention group versus 0.7% for the control group, a difference of 1.2% (p<0.001). Among four studies152–155 using SMS alone and SMS with web-tracking, three studies reported significant change in HbA1c.153–155 Six studies used a mixture of technologies for the intervention, including mobile phones, Internet, web portals, SMS, and/or glucose meters that provided messaging156–160
We also include in Table 4 a systematic review by Liang161 and a Cochrane Review.162 The systematic review included 22 trials. The meta-analysis of 1657 participants showed that mobile phone interventions for diabetes self-management reduced HbA1c values by a mean of 0.5% over a median of 6 months follow-up. A subgroup analysis of 11 studies of T2DM patients reported significantly greater reduction in HbA1c compared to studies of those with type 1 diabetes [0.8 (9 mmol/mol) vs. 0.3% (3 mmol/mol); P = 0.02]. The authors reported that the effect of the mobile phone intervention did not significantly differ by other participant characteristics or intervention strategies. The Cochrane Review reported on computer-based diabetes self-management interventions for adults with T2DM in 4 studies. The interventions addressed in this review included those using computer-based software applications that were based on user input (touch screen or other clinic support), desktop computer-based and mobile phone-based interventions. The Cochrane review also included other outcomes besides HbA1c, e.g., health related quality of life, death from any cause, depression, adverse effects, and economic data.162 A review of eleven studies by Pal provided data for a meta-analysis from which the authors reported pooled results indicating a small, statistically significant difference in outcomes between the intervention and comparison groups, mean difference −0.21 (95% CI −0.4 to −0.1.162 However, for 8 of the reviewed studies, they reported a significant mean difference in the HbA1c change for mHealth interventions compared to control condition ranged from 0.01 to −0.8 (−1.45, 0.15).
An early review of evidence on barriers and drivers to the use of interactive consumer health information technology (IT) by the elderly, persons with chronic conditions or disabilities, and the underserved concluded that questions remain as to the 1) optimal frequency of use of systems by patients, providers and 2) whether the success of interventions depends on repeated modification of the patient’s treatment regimen or ongoing assistance with applying a static treatment plan.171 A recent review focused on the effect of mobile phone interventions for glucose control in diabetes.161 This meta-analysis of 22 studies with 1657 participants showed that mobile phone interventions significantly reduced HbA1c by a mean of 6mmol/mol or 0.5% over a median follow-up of six months. Among the studies that we reviewed (Table 4), duration of interventions in the studies varied from 3 to 18 months. However, it should be noted that most clinical trials we reviewed examined change in HbA1c during a three-month intervention and very little was reported about the engagement and persistence of use with the technology. Participants randomized to the intervention arms of the trials received enhanced clinical attention and may also have received diabetes management supplies. Therefore, it may be inaccurate to assume that an intervention group’s significant change in HbA1c at three months is attributable to technology versus other nonspecific benefits of participation, especially considering a report from a 2011 survey showed that 26% of downloaded health apps are used only once and 74% are abandoned by the 10th use.172
The use of heterogeneous interventions (mobile phones, SMS and/or internet based) makes it more difficult to determine the effect of any single technology component on HbA1c. As suggested in other reviews162,173 of studies with different technology-based approaches (e.g., automatic SMS messages versus personalized feedback) a single component of technology may impact different behaviors in ways not clearly distinguishable when intervention components are combined. Authors of two systematic reviews concluded that interventions were more likely to be successful if they selected and combined theory-based behavior change strategies162,174, including interactive components that involve tracking, personalized feedback and peer support.
Gaps and Recommendations for future directions.
Few studies focus on high risk, underserved or minority populations. Most studies do not report on changes in anti-hyperglycemic medications during the intervention which may impact change in HbA1c. Without that information, it is difficult to determine if changes in lifestyle behavior or changes in medications contributed to the effectiveness of the mobile intervention It is possible that reports of the follow-up secondary analyses of such studies has not been published, or that our search missed them. The reviewed studies did not report intervention dose or receipt, i.e. number of SMS messages or push notifications sent and opened by participants. Only one study151 reported differences in HbA1c change as a function of different baseline A1c levels which may be important for understanding who will most benefit when targeting specific populations, including older adults. Similar to other sections in this paper, we recommend that future studies address the need to identify specific behaviors that may impact glucose management singly or in combination.
We recommend:
that technology development and/or intervention development be considered to meet the needs of specific population groups: a) older adults with age-related changes such as vision or touch, b) minorities needing culturally sensitive intervention content or materials and approaches-, and c) low-income adults who may have inconsistent access to mobile technologies and supplies to support diabetes management.
that studies evaluate technology-supported glucose management for periods longer than 3 months to determine sustainability of engagement and the long-term effects of mHealth interventions in maintaining behavior changes.
That studies include clinical, technical, and behavioral factors that may influence initial engagement and ongoing use of mHealth and its associated impact on outcomes.
That studies examine other outcomes related to improved diabetes management such as quality of life and acceptability of mHealth devices.
Finally, we recommend that future studies examine the relationships among use of mHealth interventions, HbA1c change, and health care utilization and costs, including consumer and provider costs. As more public and private insurers reimburse for the cost of mHealth interventions, evaluation of claims data from these populations may add to our understanding of cost effectiveness.
Using mHealth to Improve Hypertension Care
Hypertension is a highly prevalent chronic medical condition that is a major risk factor in CVD. The risk for CVD events such as stroke or myocardial infarction doubles for every 20 mm Hg increase in systolic and 10 mm Hg diastolic blood pressure.175 Best practices for treatment of hypertension include a combination of pharmacotherapy with preventive lifestyle counseling for exercise, healthful eating and smoke-free living. 175 Despite widespread initiatives to treat hypertension and availability of antihypertensive medications, less than 50% of people in the US have controlled blood pressure.142 This is thought to be due largely to sub-optimal adherence to self-care.176
Strategies to improve self-care and adherence have been explored. Face-to-face counseling has been shown to be associated with reductions of 3–8 mm Hg in systolic blood pressure (SBP) among patients with hypertension.176 Team-based hypertension care, with partnership between a primary care physician and other professionals, such as nurses, pharmacists, or community health workers has been shown to increase the percentage of patients with controlled blood pressure by 12%.177 Still, costs of such care models prevent dissemination and sustainability.
The rapid growth of the internet and mobile telecommunication offers unprecedented opportunity to improve patient access to and engagement with hypertension care.178,179 In general, they follow the premise that patients might spend only a few hours a year with a physician or nurse, but they spend 5000 waking hours each year engaged in choices that affect their health.180 These eHealth programs can be delivered by the Internet, email, SMS, or similar electronic means to engage patients in remote blood pressure, medication and behavior monitoring as well as provide patients relevant education, counseling and motivational support.
One example of an mHealth intervention that has become accepted as beneficial to the management of hypertension is self-measured blood pressure (SMBP) monitoring. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure recommend SMBP monitoring as an adjunct method in the management of hypertension.175 The AHA recommends SMBP for evaluation of most patients with known or suspected hypertension to assess response to treatment and possibly improve adherence.181 Still, much remains unknown about what other mHealth interventions are effective in improving hypertension care.
Review of evidence for efficacy of mobile technology-based interventions to promote blood pressure control.
Our review focuses on mHealth intervention effects on SBP specifically given its association with cardiovascular outcomes. We searched PubMed for the years 2004 to 2014, using the following terms, hypertension; hypertensive; antihypertensive; anti-hypertensive; pre-hypertensive; high blood pressure; elevated blood pressure; increased blood pressure; systolic blood pressure; diastolic blood pressure. These terms were cross-referenced with the mobile technology terms described previously. This search resulted in 316 identified articles and 125 were reviewed further but were not relevant to mobile technologies. The studies we selected for a detailed review (see Table 5) were published between 2008 and 2014 and permitted patients some form of electronic platform to assist with self-monitoring and/or support for hypertension. We focused on interventions that offered some additional feature beyond simply SMBP monitoring, and also included internet-based studies since an increasing number of people access the internet on mobile devices.16 We divided the review to describe individual studies organized by the primary form of mHealth used to deliver the intervention followed by existing systematic reviews. Here we provide details on the salient studies and what was learned from the review.
Table 5.
Study Cited, Design, Outcome, Setting, Quality Rating | Sample Characteristics, Group Size, Baseline Blood Pressure, Study Retention | Study Groups & Components | Technology used | Intervention Duration, # of Intervention Contacts, Intervention Adherence, Interventionist: | Primary Outcome: Mean SBP (mmHg or %change) |
---|---|---|---|---|---|
Green et al., 2008182 Design: 3-group RCT Outcome: % of Ps with BP <140/90 mm HG Setting: 10 medical centers Country: US | N = 778 Int1: n = 258 Int2: n = 259 Int3: n = 261 Mean age (SD): Int1: 59.5 (8.3) yrs. Int2: 59.3 (8.6) yrs. Int3: 58.6 yrs. (8.5) yrs. Mean (SD) mm/Hg: Int1: 152.2 (10.0) Int2: 152.2 (10.4) Int3: 153.3(10.6) Women %: Int1: 45.9% Int2: 55.9% Int3: 54.7% White %: Int1: 86.1% Int2: 79.3% Int3: 82.9% Retention: Int1: 95.0% Int2: 90.8 Int3: 95.8% | Int1: home BP monitoring, secure email, refilling medications, viewing medical record, health library, links to resources. Int2: I1 + q2week pharmacist interaction via web (action plan) and secure messaging Int3: HTN pamphlet | Home BPM equipment, web site, secure messaging | Duration: 12 mo. Contacts: Int1 and Int2: Office BP measurement at baseline and 12 mo; Int3: Office BP measurement at baseline and 12 mo. Intervention adherence: PCP visits: Int1: 2: 3.2 Int2: 1: 3.0 Int3: 3.2 Int1: 2..4 (4.6) Int2: HC messages: 22.3 (10.2) Int3: 3.3 (7.4) Int1: 3.8 (5.0) Int2: HC phone calls:7.5 (9.3) Int3: 4.0 (4.8) Interventionist: Int1: Pharmacist Int2: Pharmacist Int3: NA | Completer’s analysis (n=730) 12mos. Achieved <140/90 target: Int1: 36% Int2: 56% Int3: 31% p<.001 Mean SBP Δ, mm Hg Int1: −8.2 Int2: −14.2 C: −5.3 p< .001 |
Madsen et al., 2008183 Design: 2-group RCT Primary outcome: office-based SBP Setting: Primary care setting Country: Denmark | N = 236, Int1: n = 113 Int2: n = 123 Mean age (SD): Int1: 55.0 (11.7) yrs. C: 56.7 (11.6) yrs. Mean (SD) SBP mm/Hg: Int1: 153.1(13.2); Int2: 152.2 (13.7) | Int1: Ps self-monitored BP 3x/week x 3 mos., then 1x/week x 3 mos. transmission via. secure website with internet recommendations and PDA messages to P Int2: informed about study but no active intervention | Home BP monitoring equipment, PDA, emails | Duration: 6 mos. Contacts: Int1: Office BP measurement at baseline and 6 mo; continuous education and medication adjustment Int2: Office BP measurement at baseline and 6 mo. Intervention adherence: | Completer’s analysis (n=223) 6 mos. Mean SBP Δ, mm Hg Int1: −11.9 Int2: −9.6 p = n.s. Achieved BP target: Int1: 60% |
Women %: Int1: 51.3% Int2: 48.0% White %: not reported Retention: Int1: 93% Int2: 96% | No data on adherence to home BP monitoring Interventionist: Int1: General practitioner Int2: NA | Int2: 38% p<.001 | |||
Cottrell et al., 2012184 Design: Quasi-experimental (nonrandomized) Outcome: SBPΔ Setting: 10 General Practitioner (GP) Groups Country: UK | N = 488 Int1: n= 124 Int2: n = 364 Mean age (range): Int1: 59 (25–86) yrs. Int2: 60 (36–87) yrs. Mean SBP (range) mm/Hg: Int1: 146 (82–194) Int2: 136 (87–197) Women %: Int1: 40% Int2: 40% White %: not specified Retention: 41% | Int1 Ps self-mon BP. SMS results to a secure server. Reminders to check BP and recommendations to contact GP were sent to Ps as SMS as needed. Results reviewed by GP or nurse at least weekly. Int2: informed about study but no active intervention | Secure server, SMS | Duration: minimum 3 mos. or until BP controlled; maximum 6 mos. Contacts: Int1: daily measurement of home BP, with daily reminders if no BP value received. Int2: BP abstracted from clinic chart Intervention adherence: Continued 3–6 mos.: 37 (30%) Completed 3 months and stopped: 51 (41%) Interventionist: Int1: General practitioner or nurse Int2: NA | Completer’s analysis Int1: n=89; Int2: n=NR 0–3 mos. Mean SBP Δ, mm Hg Int1: −8 Int2: +1 0–3 mos. (Ps meeting criteria #2) Mean SBP Δ, mm Hg Int1: −15.88 Int2: − 11.42 p NR |
Kiselev et al, 2012185 Design: 2-group RCT (unblinded) Outcome : % Ps BP Setting: Single center Cardiology practice Country: Russia | N = 199 Int1: n = 97 Int2: n = 102 Mean age (SD): Int1: 49 (11) yrs. Int2: 51 (11) yrs. Mean SBP (SD): | Int1: Ps self-mon BP and other values requested from server by SMS and Ps responses submitted by SMS. If weekly average BP not at target, Ps invited via. SMS or phone for office visit or | Secure internet-based web site, SMS | Duration: 12 mos. Contacts: Int1: SMS requests and reminders sent to Ps on a variable frequency (daily to semiannually) on factors related to BP control and med adjustments. No minimum | Completer’s analysis (n=164) 12 mos. % Ps achieving BP goal Int1: 77% Int2: 12% p < .001) |
Int1: 153.4 (9.6) mm Hg Int2:158.2 (9.9) mm Hg (p < 0.05) Women %: Int1: 45% Int2: 50% White %: not specified Retention: Int1: 64% Int2: NR | telephone consultation. Int2: standard of care drug therapy and lifestyle recommendations, Ps encouraged to check BP at home | frequency of office visits. Int2: no reminders sent. frequency of office visits determined by physician, but must be at least every 6 mos. Intervention adherence: Ps withdrawn if didn’t respond to SMS for 1 mo. Int1: 18 (51%) withdrew due to loss of interest; 12 (34%) due to technical difficulties; 5 (15%) due to unknown reasons. Int2: not reported Interventionist: Int1: Physician Int2: Physician | Mean SBP Δ, mm Hg Int1: −23.7 Int2: −6.9 p NR | ||
Logan et al., 2012186 Design: 2 group Pilot RCT Outcome: 7-day ambulatory SBP Setting: 5 General Practitioner (GP) Groups Country: Canada | N = 110, Int1: n = 55 Int2: n = 55 Mean age(SD): Int1: 63.1 (9.0) yrs. Int2: 62.7 (7.8) yrs. Mean (SD) SBP mm/Hg: Int1: 142.6 (10.2) Int2: 142.7 (10.9) Women %: Int1: 38% Int2: 51% White %: Int1: 71% Int2: 60% Retention: | Int1 + self-care: messages tailored to BP reading. Alerts to provider re: abnormal SBP; auto voice messages when non-adherent to BP readings; printouts of summary BP to doctors. Interventionist: Physician Int2: Home BP monitor, measure 2x/week in AM and 2x/week in PM | Bluetooth, Blackberry smartphone software, home BP monitor | Duration: 1 yr. Contacts: Int1: Avg. alert to Ps 1.82 (3.69); alerts to MDs (0.09 (0.35) Int2: only at assessments, Office BP measurement at baseline and 1 yr. Intervention adherence: readings per week=10.8 (6.7); decline in % adherent per week =−1.8 Interventionist: Int1: primary care physician Int2: primary care | Completer’s analysis (n=105) 12 mos. SBP Δ, mm Hg, Mean (SD) Int1: −9.1 (15.6) Int2: −1.5 (12.2) p<.005 Achieved <130/80 target: Int1: 51% Int2: 31% p<.05 |
Int1: 96.4% Int2: 92.8% | physician | ||||
Nolan et al., 2012187 Design: RCT Outcome: SBP Δ Setting: 3 sites Country: Canada | N = 387 Randomized: Int: n = 194 Con: n =193 Actual exposure (analyzed sample): IntInt1: n = 97 Int2: n =63 Int3: n =227 Mean age (95% CI): Int1: 55.7 (54.3–57.0) yrs. Int2: 57.0 (55.2–58.8) yrs. Int3: 56.7 (55.7–57.7) yrs. Mean SBP mm/Hg: Int1: 143.3 Int2: 134.6 Int3: 139.6 Women %: Int1: 72.2 Int2: 61.9 Int3: 52.9 White %: NR Retention %: Int1: 76.8 Int2: 81.9 | Int1: e-counseling on recommendations for diet, exercise, smoke-free living based on stage of change (≥8 emails over 4 months) Int2: received Heartline e-newsletters from the Heart and Stroke Foundation that contained general information and advice for heart-healthy living | Duration: 4 mos. Contacts: Int1: Mo1: weekly e-mails Mos 2: bi- weekly e-mails Mos. 3 & 4: monthly e-mails Intervention adherence: BP readings in mo.1=17%; mo.#6=7% Interventionist: Int1: NR Int2: NR | ITT analysis: no significant difference between groups on Δ in primary outcomes. Per protocol analysis was conducted with 3 groups according to whether Ps received ≥8 emails, 1–7 emails or 0 e-mails (control). 4 mos. Mean SBP Δ, mm Hg Int1: −8.9 (−11.5 to − 6.4) Int2: −5.8 (−9.1 to −2.6) Int3: −5.0 (−6.7 to −−3.3) p =.03(Int1 vs. Int3) | |
Piette et al, 2012188 Design: 2-group RCT Outcome: SBP Setting: 8 clinics Country: Honduras and Mexico | N = 200: Int1: n = 99 Int2: n = 101 Mean age (SD): Int1: 58.0 (1.3) yrs. Int2: 57.1 (1.1) yrs. | Int1: BP readings; automated feedback through IVR (med adherence, salt intake, BP checks), e-mail alerts for health workers, elect to enroll | Electronic home BP monitor, IVR, emails to providers | Duration: 6 weeks Intervention contacts with clinicians: unmeasured. Office BP measurement at baseline and 6 wks. | Completer’s analysis (n=181) 6 wks. SBP Δ, mm Hg, Mean (SD) Int1: −10.7 (2.3) |
Mean initial SBP (SD): Int1: 153.2 (2.1) mm Hg Int2: 150.0 (2.0) mm Hg Women %: Int1: 66.3% Int2: 68.4% White %: NR Retention: Int1: 90% Int2: 91.1% | family/friend to get summaries of P status and support messages Int2: Usual primary care. | Int1 adherence: 67% completed phone calls, 20% received call from clinician due to auto emails. Interventionist: Int1: Automated phone calls Int2: NR | Int2: −6.4 (2.4) p<.09 Achieved BP target: Int1: 57% Int2: 38% p<.001 | ||
Watson et al., 2012189 Design: Cluster RCT Outcome: SBP Δ Setting: 6 worksites Country: US | N = 404 patients Int1: n = 197 Int2: n =207 Sites: Int1: 3 Int2: 3 Mean age (SD): Int1: 49.5 (8.0) yrs. Int2: 48.4 (8.0) yrs. Mean (SD) SBP mm/Hg: Int1: 134 (14) Int2: 132 (14) Women %: Int1: 21.3% Int2: 25.1% White %: Int1: 86% Int2: 87% Retention: Int1: 95.4% Int2: 98.5% | Int1: Ps self-mon BP, automatically transmitted data to a central server. Data were displayed on a self-management web site. Ps logged onto the web site ≥1 times/wk.. The web site allowed Ps to track BP, access educational material, & receive automated, tailored messages. Int2: Ps received training of BP self-mon, but did not receive any feedback. | Home BPM, modem, website | Duration: 6 mos. Contacts: not recorded Intervention adherence: BP readings in mo. #1=17%; mo.#6=7% Interventionist: Int1: Automated messages Int2: NA | ITT (how to handle missing data NR)
6 mos. Achieved SBP target: Int1: 21.3% Int2: 16.4% (p=0.04) Mean SBP Δ, mm Hg 0.49 p =.8 |
Magid DJ, et al. 2013190 | N = 348 Int1: n = 175 | Int1: provided home BP cuff, enrolled in | Web-enabled software for home | Duration: 6 mos. | Completer’s analysis (n=326) |
Design: 2-group RCT Outcome: proportion Ps achieved goal BP Setting: 10 Kaiser Permanente Clinics Country: US | Int2: n= 173 Mean age (SD): Int1: = 60 (11.3) yrs. Int2: 59.1 (10.9) yrs. Mean SBP (SD): Int1: 148.8 (16.2) mm Hg Int2: 145.5 (14.5) mm Hg Women %: 40% White %: 83% Retention: Int1: 93% Int2: 95% | Heart360 web program, met with clinical pharmacy specialist for medication adjustment, provided lifestyle counseling. Both groups received written educational materials on managing BP, diet, PA, instructed to follow-up with PCP. | BP monitoring (Heart360) | Contacts: Int1: Ps self-measure BP 3 times/wk., uploaded values into Heart360 web site. Pharmacist made medication adjustments via telephone or secure email to S and to PCP via. EMR Reminders for BP upload automated phone call. Intervention adherence: Median time to follow-up was 182 days in both groups. Int1 group: 70% Ps adherent (uploading values at least once a week >80% of study duration) Clinic visits No. (%): Int1: 3.3 (2.5) Int2: 3.1 (2.3) Telephone contacts: Int1: 5.3 (4.5) Int2: 3.5 (3.8) Email contacts: Int1: 6.0 (5.5) Int2: 2.4 (3.2) Interventionist: Int1: Clinical pharmacy specialist Int2: Primary care physician | 6 mos. % achieved SBP goal Int1: 54.1% Int2: 35.4% Adjusted risk ratio 1.5; 95%CI: 1.2–1.9 Mean SBP Δ, mm Hg Int1: −20.7 Int2: −8.2 p NR |
McKinstry B, et al. 2013191 Design: 2-group RCT Outcome: SBPΔ Setting: 20 Primary Care Practices Country: Scotland | N = 401 Int1: n = 200 Int2: n= 201 Mean age (SD): Int1: 60 (11.3) yrs. Int2: 59.1 (10.9) yrs. Mean SBP (SD): Int1: 148.8 (16.2) mm | Int1: self-monitor BP initially twice in AM, once in evening for 1st week, then weekly; used Bluetooth-enabled BP cuff with automated responses based on BP control and healthcare team review and recommendations | Electronic home BP monitor sent BP reading via Bluetooth to cellular, then transmitted via SMS to secure website. | Duration: 6 mos. Contacts: Int1: automated response to patient based on BP control every 10 readings or weekly; healthcare team review at least weekly Mean PCP visits (SD): | Completer’s analysis (n=359) 6 mos. Mean SBP Δ, mm Hg Int1: −6.0 Int2: −2.2 p = .0002 |
Hg Int2: 145.5 (14.5) mm Hg Women %: Int1: 38.3% Int2: 41% White %: not specified Retention: Int1: 97.5% Int2: 98.5% | Int2: standard of care BP management | Int1: 3.66 (2.67) Int2: 2.6 (2.52) (p value for Δ between groups = 0.0002) Intervention adherence: Compliance with BP checks in Intervention: median of 76 BP readings; 89% of Ps completed > 90% of expected minimum # readings. Interventionist: automated messages, medication changes by physician Int2: doctor or practice nurse | |||
Rifkin, et al. 2013192 Design: 2-group RCT (2:1 ratio) Outcome: SBP Δ Setting: VA, CKD & HTN clinic Country: US | N = 43 Int1: n = 28 Int2: n = 15 Mean age (SD): Int1: 68.5 (7.5) yrs. Int2: 67.9 (8.4) yrs. Mean daytime ambulatory SBP (SD):I: Int1: 149 (16.2) mm Hg Int2: 147 (8.6) mm Hg Women %: Int1: 7% Int2: 0% White %: Int1: 75% Int2: 73% Retention: Int1: 93.3% Int2: 88.2% | Int1: self-monitor BP using Bluetooth-enabled BP monitor, weekly phone calls for out of range BP readings (pharmacist counseling) Int2: home BP monitoring, standard of care BP management | Electronic home BP monitor; home health hub (Bluetooth, internet), secure web site to view BPs | Duration: 6 mos. Contacts: Int1: 2.7 over 6-mo. 1.9 med changes per patient. Intervention adherence: 29 readings per month; 78% of Ps used cuff 4x/month for 6 mos. Int2: 20% brought BP records to med visit. Interventionist: Int1: Physicians and pharmacist Int2: Physicians | Completer’s analysis (n=43) 6 mos. Mean SBP Δ, mm Hg Int1: : −13 Int2: −−8.5 p = .32 |
Thiboutot et al., 2013193 | N = 500 patients Int1: n = 282 Int2: n =218 | Int1: automated web site with tailored messages based on self-report BP; | Internet website | Duration: 12 mos. Contacts: | ITT (LMM) 12 mos. |
Design: Cluster RCT Outcome: SBP Δ Setting: 54 physician practices Country: US | Sites: Int1: 27 Int2: 27 Mean age (SD): Int1: 59.6 (12.1) yrs. Int2: 61.6 (11.4) yrs. Mean (SD) SBP mm/Hg: Int1: 132.7 (14.9) Int2: 132.4 (15.2) Women %: Int1: 58.5% Int2: 56.4% White %: Int1: 75.5% Int2: 74.3% Retention: Int1: 84% Int2: 83% | suggestions for questions to ask PCP Int2: website with general prevention service info unrelated to HTN care. | Int1 &Int2: Office visits at baseline, 12 mo. Intervention adherence: 34.8% used website ≥ once each of 12 mos.; 82.2% used website at least once. Interventionist: Int1: Automated messages Int2: NA | Achieved target: Int1: 71.3% Int2: 65.6% p=.31 Mean SBP Δ, mm Hg Int1: −4,4 Int2: −3.5 p<0.88 | |
Cicolini G et al, 2014194 Design: 2-group RCT (unblinded) Outcome: SBPΔ Setting: Single-center Hypertension Primary Care Center Country: Italy | N = 203, Int1: n = 102 Int2: n = 101 Mean age (SD): Int1: 59.8 (15) yrs. Int2: 58.3 (13.9) yrs. Mean SBP (SD): Int1: 150 (11) mm Hg Int2: 153 (12) mm Hg (p < 0.12) Women %: Int1: 50% Int2: 48% White %: not specified Retention: Int1: 97% Int2: 98% | Int1: 1-hr. education program on risk factors and healthy lifestyle plus weekly email alerts and phone calls from a nurse care manager. Int2: 1-hr. education program on risk factors and healthy lifestyle. | Email reminders | Duration: 6 mos. Contacts: Int1: weekly email reminders Both groups: follow-up visits at 1, 3, and 6 mos. Daily selfassessment form of adherence to treatment. Intervention adherence: Mean PCP visits (SD): Int1: 3.66 (2.67) Int2: 2.6 (2.52) (p = 0.0002 for Δ between groups) Compliance with therapy dose (%): Int1: 100 % Int2: 96.9 % Compliance with therapy hours (%) I: 91% C: 96.9% Interventionist: Int1: Nurse care manager Int2: Nurse care manager | Completer’s analysis (n=198) 6 mos. Mean SBP Δ, mm Hg Int1: −14.9 (8.1) Int2: −10 (11.6) p < .001 |
Systematic Reviews and Meta-Analysis | |||||
Uhlig et al., 2013195 Design: Systematic review and meta-analysis Outcome: SBPΔ Setting: no setting restrictions Country: no language restrictions | Prospective comparison studies with at least 8 wks. follow-up. Analysis 1: SMBP+ support vs. usual care; 25 studies; Analysis 2: SMBP+ support vs SMBP; 13 studies; | Support included educational materials, letters to Ps and providers on treatment recommendations, Web resources, phone monitoring with electronic transmission of BP data, telecounseling, behavioral management, medication management with decision support, | Only one study used: web-based pharmacist counseling | Analysis 1: 5 quality A studies Analysis 2: too heterogeneous Intervention Adherence: NR Study duration: 8 wks. | Analysis 1 12 mos. Mean SBPΔ: −2.1 to −8.3 mm Hg Analysis 2: NR |
nurse or pharmacist visits, calendar pill packs, and adherence contracts | |||||
Liu et al., 2013196 Design: Systematic review and meta-analysis Outcome: SBPΔ Setting: no setting restrictions Country: no language restrictions (56% in USA) | Prospective comparison studies testing preventive e-counselling or advice using Web sites or e-mails to modify exercise or diet as a means of improving blood pressure control of at least 8 weeks duration. 13 studies N= 2221 Mean age: 55 yrs. (range 18–89) Mean SBP (SD):1 36 (6.4) Women %: 44 White %: not specified Retention: 53–94% | e-counselling or advice using websites or emails to modify exercise or diet as a means of improving BP control. | Internet, email The internet-based interventions were primarily self-guided, and access was through desktop and mobile devices | Mean intervention duration (SD): 5.6 (3.6) mos. 8 of the 13 studies being short-term (< 6 mos.) and 5 being long-term (6–12 mos.) Intervention Adherence: NR | Pooled: Mean SBP Δ, mm Hg: Int: −3.8 (95% CI −5.63 to −2.06) Pooled effect size: − 0.27 (95% CI: −0.4 to −0.1) Longer interventions vs. shorter interventions, effect size on SBP: 0.44 (95% CI, −0.58 to −0.31) vs.−0.23 ( 95% CI, −0.36 to −0.10) ≥5 vs. <5 behavioral change techniques effect size on SBP: −0.46 (95% CI-0.60 to −0.33) vs. −0.19 (95% CI, −0.33 to −0.06). |
P or Ps= participant(s), N = total sample, n = subgroups, Int = Intervention group, 1, 2, etc. Con = control group, RCT = randomized control trial, Δ = change or difference, mo. = month, mos. = months, wk = week, wks = weeks, year = yr, Baseline = 0, CI = confidence interval, SMS = short message service, MMS = multimedia messaging service, essentially small pictures, EMR = electronic medical record;NA = not applicable, NR = not reported, n.s., = not significant, n.s.d = not significantly different, PDA = personal digital assistant, BP = blood pressure, BPM = blood pressure monitoring, HTN=hypertension, SBP=systolic blood pressure, IVR= interactive voice response; ;automated = without a clinician who generates, tailors, or modifies the output.
SMS:
There were 3 RCTs that utilized text messaging as the primary intervention modality.184–186 The details of these studies are provided in Table 5. The three studies had methodological limitations including poor retention. Two of the studies185,186 reported significant differences in blood pressure reduction between the treatment conditions; however, all studies reported results using the completers’ analysis approach rather than ITT.
There were 3 RCTs that used email as the primary intervention modality.111,183,187 These studies ranged from 4 to 6 months and all had high retention. The frequency of the email contact was not specified in Nolan’s study and was frequent in the other studies. Madsen183 augmented the email information exchange with messages sent via a PDA. Ciccolini111 and Nolan187, using a completers’ analysis, reported a significant difference between groups in blood pressure changes while Madsen, using ITT analysis, did not find a difference in blood pressure between groups but observed that a significantly higher proportion of the intervention group achieved the target blood pressure.
A single study was found that used IVR as the primary intervention modality and was conducted in Honduras and Mexico.188 Participants received weekly information on medication adherence and salt intake tailored to their BP through the IVR. There was only a trend for a significant difference in BP reduction from the control group, which may be due to only 67% completion of the IVR calls.
There were 2 RCTs that used a website as the primary intervention modality.189,193 Watson et al. enrolled 500 adults from primary care practice offices in central Pennsylvania. The web-based intervention provided feedback on reported blood pressure and advice; however, only 35% of intervention participants used the website at least once monthly.193 Watson et al. enrolled 404 adults with HBP from 6 worksites for a 6-month study that included a website that displayed SMBP readings, provided education and custom messages based on BP reports.189 Adherence to SMBP was low overall, with only 17% of intervention participants reporting SMBP in month 1 and 7% at month 6. Neither of these studies demonstrated significant reductions in BP and no differences between intervention and control conditions. It was not stated but it is possible that the adherence was so low because participants might not have had the capability to access the website via a mobile device.
Mixed use of mHealth strategies:
There were 3 RCTs that used a mixture of mHealth modalities to deliver the intervention. Green et al.182 used web-access, including secure email, medical record, health library and links to resources vs. web+ pharmacist or a control (usual care) condition. Only the web+ pharmacists group reduced SBP significantly better than the other conditions at 12 months. It also resulted in increases in secure messaging between patient and provider/pharmacist and more antihypertensive medication classes being added. McKinistry et al. compared 6 months of SMBP with access to a website with graphical displays of SMBP data, and optional automated SMS or emails with feedback on their blood pressure control to a control condition.191 Adherence to uploading BP data was high but the number of participants opting for SMS and/or emails was not reported. The mean reduction in SBP in the intervention group was significantly higher than in the control group. Also, there were more outpatient care visits and antihypertensive medications prescribed in the intervention group. Magid et al. recruited patients from primary care clinics and randomized them to a control condition or an intervention using the Heart360 web site to upload their SMBP, IVR reminders if patient did not enter blood pressure data, and phone calls and emails from a pharmacist recommending antihypertensive pharmacotherapy changes.190 While Magid used some components of mHealth in this intervention, it does not appear that the phone component was based on mobile devices. Adherence to uploading BP data was high and there were higher rates of emails and phone calls with pharmacists in the intervention group. Results revealed a significant mean reduction in SBP in the intervention group at the end of the 6-month study. All three of these studies provided some combination of patient educational resources, timely delivery of BP data to providers, and personalized messages to patients. The positive results of the three trials may suggest that a combination of such strategies or modes of intervention delivery may be needed to engage patients. Whether these multimodal technology-based approaches can provide the same or better results than team-based in-person care for a lower cost remains unclear; the data at this time suggest that further investigations are warranted. What is clear from this review of studies targeting improved BP control as well as in other sections that addressed interventions targeting behavior, is that mHealth or digital health has no defined taxonomy or classification of interventions delivered by the existing technology. Thus it is difficult to summarize the study outcomes by such a classification of interventions. The current state of science suggests they all have an important place in targeting improved CV health.
In summary, 8 of the 12 studies detailed in Table 5 were conducted outside of the US and two of those were conducted in Canada so results may not be generalizable to all healthcare systems. Most, but not all of the studies, used self-monitoring of BP and used those data for reporting and receiving feedback. Eight of the studies reported a significant difference between the treatment conditions, but only three of the 12 studies used an ITT approach in analyzing the results. Instead, most studies reported results only in experimental subjects who were compliant with the mHealth technology used. This approach not only inflates the results and compromises randomization but also raises questions about the generalizability to a broad hypertensive population, particularly elderly or disabled patients, such as those with stroke, who may have difficulty using the technology.
Systematic Reviews:
We identified two systematic reviews and meta-analyses examining studies testing mHealth interventions for BP control.195,196 Using the same quality assessment tool for systematic reviews and meta-analyses as in the AHA/ACC guideline, both systematic review studies were rated good quality (indicating a study with the least level of bias and results deemed valid). Both reviews focused on a slightly different topic, but shared common features. Uhlig et al.195 focused on SMBP monitoring with or without additional support and Liu focused on internet-based interventions for blood pressure control.196 Both reviews suffered from heterogeneity across studies in SMBP equipment used, intervention modality and components, participants, and blood pressure endpoints, precluding direct comparisons across studies. Neither of the systematic reviews exclusively included RCTs. Both reviews focused on intervention comparison with usual care or no intervention, whereas only Uhlig examined comparison with an active control (SMBP self-monitoring), and only Liu attempted to determine which of the intervention characteristics were associated with better outcomes.
Uhlig et al.195 reviewed 25 studies that compared SMBP plus support to usual care. Among the 5 quality A studies that compared SMBP + support to usual care, there was a net reduction in SBP of−2.1 to −8.3 mm Hg. The type of support offered varied greatly across studies, and included only one that used an mHealth support intervention.182 Uhlig et al. also examined 13 trials comparing SMBP versus SMBP plus support and found no evidence to support the benefit of SMBP + support on top of SMBP alone. Liu et al. examined 13 studies, which compared internet-based counseling interventions on blood pressure control in pre-hypertensive and hypertensive patients, 11 of which were RCTs. They found that e-counseling interventions significantly reduced daytime SBP by 3.8 mm Hg (95% CI −5.63 to −2.06). The authors found that longer interventions (6–12 months duration) were associated with greater effects on SBP. They also found trends of greater effects when interventions used multiple behavioral techniques and were proactive with patients (as opposed to reactive or passive). The one study to specifically explicate a theoretical framework was conducted by Nolan RP et al., 2012.187
A significant limitation of the current evidence for the use of mHealth for blood pressure control is that the majority of studies (8 of12) were conducted for 6 months or less, with no studies beyond a12month duration. Given that hypertension is a chronic condition that requires long-term medical care, a sustained benefit of mHealth interventions beyond a few months should be demonstrated before this technology becomes widely accepted. While adherence with some mHealth technologies over the short-term may have been demonstrated, the ability to maintain the consumers’ interest and active use of these tools over the long term requires further study.
Gaps and Limitations.
The results of the studies described above indicate that mHealth interventions in general show promise in reducing SBP in patients with hypertension, but with large variability in behavioral targets, intervention components, delivery modalities, and patient engagement. Although behavioral targets for blood pressure control include routine monitoring of blood pressure, healthful dietary intake, physical activity, medication adherence, smoking cessation and stress management, among others, Existing mHealth interventions have not identified how best to address these behaviors. This includes questions about which behaviors need to be addressed to change blood pressure and in whom, as well as whether to address them at the same time or sequentially, ad lib or scheduled.
Essential components of mHealth interventions to promote blood pressure control also largely remain unknown, but likely include similar behavioral techniques shown in in-person counseling interventions to be effective, including self-monitoring, goal-setting, and problem solving. Evidence197 suggests that education alone is not effective to change behavior, and results from Liu suggest that multiple techniques are likely to be more effective than fewer. Other unknowns include whether mHealth intervention components should be proactive or reactive, expert driven (protocol driven, prescriptive messaging) or user driven (collaborative protocol with supportive messaging), and whether there are specific behavioral theories that are more useful than others to guide intervention components.
The modalities used to deliver mHealth interventions for blood pressure control included web-based, email, SMSs and IVR. The best modality may never be known given the rapid pace of change in information and communication technologies. As well, the best delivery modality may vary between individuals as well as within individuals given their location and setting. Individual consumer factors, such as age, access to the internet, and learning preferences, may determine the successful use of specific tools in a specific individual. Therefore, consumer preference may ultimately determine the most effective method of delivery in an individual patient.
Patient engagement with mHealth for blood pressure control remains limited by the fact that the majority of studies (8 of12) were conducted for 6 months or less, with no studies beyond a12-month duration. Given that hypertension is a chronic condition that requires long-term medical care, a sustained benefit of mHealth interventions beyond a few months will likely be needed to show meaningful health benefits. As well, estimates of patient engagement with mHealth interventions over these short periods were not possible given the heterogeneity of studies reviewed and components useful to maintain engagement specifically were not studied.
Aside from these current scientific limitations and unknowns of how best to use mHealth interventions to improve blood pressure control, the commercial availability of evidence-based mHealth interventions for patients and providers is scarce. For example, the only one we were aware of as being commercially available is the Heart360 website. However the current usage rates of such programs remain unknown and consumer adherence outside of monitored RCTs may be difficult to predict. Adoption of mHealth interventions for blood pressure control in health systems may additionally be hindered by security and patient privacy policies regarding transmission of identifiable patient data, non-existent current reimbursement for eHealth interventions, and IT interface difficulties.
Suggestions for future research:
Identify behavioral targets that are tailored to an individual that will have the greatest effect on BP control and, if multiple behaviors, how to best attempt to change those behaviors.
Leverage existing knowledge of effective intervention components for BP control from in-person counseling-based studies and adapt them to mHealth platforms, while using the unique aspects of mHealth platforms to innovate components.
Utilize delivery modalities that are currently used by individuals, meet the needs of their various lifestyles and preferences, and work across mHealth platforms. This includes trials testing mHealth interventions from a broader consumer base (e.g. elderly, disabled, etc.).
Study techniques to optimize lasting patient engagement beyond 6 months duration, including strategies such as gamification and contingency management (incentivization).
Conduct trials comparing mHealth strategies to effective, yet possibly more costly in-person counseling interventions.
Use of mHealth in Management of Dyslipidemia
Dyslipidemia affects nearly one in five to ten Americans.198 Despite ready access of health care providers to evidence-based cholesterol-management goals and potent, well-tolerated medical therapy, management of hyperlipidemia remains suboptimal.199 A large body of evidence accumulated over the last two decades supports the link between dyslipidemia and atherosclerosis200–202, and the clinical benefits of statin therapy in the treatment of lipoprotein abnormalities. This evidence provides the basis for a number of consensus based guidelines9,203–209 for optimizing lipid levels in adult and pediatric populations. However, despite the wide dissemination of these guidelines, hyperlipidemia remains prevalent and suboptimally treated in the US.210
There are several reported potential barriers to the implementation of these evidence-based treatment guidelines into clinical practice, including provider and patient knowledge, attitudes, and behaviors; provider-patient communication issues; and system-based issues such as costs and the lack of organized systems of care around the recognition and treatment of hyperlipidemia.211,212 Thus a multimodal approach affecting providers, patients, provider-patient communication, and care-delivery systems is likely needed to translate these guidelines into clinical practice and maximize the use of mHealth technology. Other barriers may include the unknown cost of delivering mHealth interventions to achieve optimal lipid control.
In 2012, the U.S. Department of Health and Human Services (DHHS)213 proposed a challenge seeking new mobile technology applications to help consumers assess their heart health risk, identify places to measure their BP and cholesterol, and use the results to partner with their health care professional to develop a treatment plan to improve their heart health. The new app would be part of a broader education effort in support of the Million Hearts™ initiative214, a public-private effort of the DHHS that targets the prevention of a million heart attacks and strokes through clinical and community prevention programs.215 In response to this challenge, the Marshfield Clinic developed the HeartHealth Mobile app which allows users to obtain a health risk assessment based on several individual health factors such as blood cholesterol and blood pressure values.
Currently, the majority of tools available for information delivery, education, motivation and self-monitoring in dyslipidemia are contained within more comprehensive materials targeting overall CVD risk reduction.194,215–224 Some of these materials provide the basis for the development of CVD risk scores and web-based score calculators that are available for patients and providers. For example, the Framingham Risk Score was developed using predictive equations based on over 5000 men and women who were 30–74 years old at baseline and were followed for cardiovascular events for 12 years.225 This score is sex-specific and incorporates information on age, BP, total cholesterol, LDL cholesterol, diabetes and smoking as predictors of CHD.226
Recently a Task Force of the American College of Cardiology (ACC) and the AHA published a set of guidelines aimed to reduce CVD risk.227 The purpose of the guidelines was to define provider practices that meet the needs of patients, however these were not meant as a replacement for clinical judgment. While the guidelines had a relatively limited scope and focused on selected critical questions, they were based on the highest quality evidence available. The guidelines were derived from evidence accumulated from RCTs, meta-analyses, and observational studies that were evaluated for quality. A CVD 10-year and lifetime risk calculator was devised, which is sex specific and incorporates information on age, race, total cholesterol, HDL cholesterol, systolic BP, treatment for high BP, diabetes and smoking. A downloadable spreadsheet and web-based risk calculator are available on the American Heart Association website.228
Review of evidence for efficacy of mobile technology-based interventions to promote management of dyslipidemia.
We searched PubMed for the years 2004 to 2014 using the terms, anticholesteremic; cholesterol inhibitor; cholesterol level; cholesterol lowering; dyslipidemia; elevated cholesterol; HDL cholesterol; high density lipoprotein; hypercholesterolaemia; hypercholesterolemia; hyper-triglyceride; LDL cholesterol; lipoprotein cholesterol; low density lipoprotein; total cholesterol; triglyceride; high cholesterol. We reviewed 24 articles in detail reporting on the use of mobile technology to manage dyslipidemia as one of the goals. The majority of studies evaluated usability, feasibility, efficacy and adherence to cholesterol improvement programs using technology-based tools or strategies, such as email, text messaging, and websites. Of note, several studies aimed at reducing diabetes or hypertension complications also included lipids as a secondary outcome160,194,221,229–232, of these, only three were of sufficient quality to include in the paper. Because of the limited number of studies using mHealthas part of the intervention to target improved lipids, we included studies reporting lipid as secondary outcome in Table 6, the study by Yoo et al.160 is also reported in the diabetes sections.
Table 6.
Study Cited, Design, Primary Outcome, Setting, Quality Rating | Sample Characteristics, Group Size, Baseline lipids, Study Retention | Study Groups & Components | Technology used | Intervention Duration, # of Intervention Contacts, Intervention Adherence, Interventionist | Secondary Outcome |
---|---|---|---|---|---|
Kang JY, et al. 2010231 Design: 3-group RCT Outcome: reduction of diabetes risk factors. Setting: Community Country: South Korea | N = 125 Int1: n = 25 Int2: n = 25 Int3: n = 75 Age: Int1: 47.47 (5.79) yrs. Int2: 45.61 (6.06) yrs. Int3: 45.84 (5.17) yrs. Mean total cholesterol: Int1: 195.48 ( 31.12) mg/dL Int2: 222.32(31.59) mg/dL Int3: 204.04 ( 32.10) mg/dL Mean LDL: Int1: 121.70(34.62) mg/dL Int2: 135.20(31.91) mg/dL Int3: 135.72 (31.39) mg/dL Mean HDL Int1: (13.37) mg/dL Int2: 44.64(13.66) mg/dL Int3: 49.87 (13.80) mg/dL Retention: 98.4% | Int1: 1-yr face-to-face counseling (5 times over 12 weeks), 10 emails over 30 wks, repeat assessment at 2 yrs. Int2: 2-yr face-to-face counseling (5 times over 12 weeks, 10 emails over 30 weeks in year 1, repeated in year 2;repeat assessment at 2 yrs. Int3: provided general health info at baseline, repeat assessment at 2 yrs. | email messaging | Duration: 2 yrs. Contacts: Int1: 15 intervention contacts Int2: 30 intervention contacts Intervention adherence: NR Interventionist: Int1: trained staff Int2: trained staff Int3: NA | Completer’s analysis (n=123) 24 mos. Total chol Δ, mg/dL, M (SD) Int1: −0.09 (27.42) Int2: −11.12 (19.56) Int3: 5.75(25.61) Int1 vs. Int2 p>.05 Int1 vs. Int3 p>.05 Int2 vs. Int3 p<.05 LDL Δ, mg/dL, M (SD) : Int1: −6.65(21.99) Int2: −5.32(26.64) Int3: −11.41(26.90) Int1 vs. Int2 p>.05 Int1 vs. Int3 p>.05 Int2 vs. Int3 p>.05 HDL Δ, mg/dL, M (SD) : Int1: −2.78 (5.79) Int2: −3.28 (10.08) Int3: 0.67(8.25) Int1 vs. Int2 p>.05 Int1 vs. Int3 p>.05 Int2 vs. Int3 p<.05 |
Dekkers et al. 2011218 Design: 3-group RCT Outcome: reduction in CV risk factors (waist, skinfold, blood pressure, total cholesterol, aerobic fitness level, body weight, BMI) Setting: Workplace intervention Country: Netherlands | N=276 Int1: n = 91 Int2: n = 93 Int3: n = 92 Mean age: 44.0 (9.2) yrs. Women: 30.8% BMI: 29.7 (3.1) kg/m2 Total chol: 4.9 (0.8) mmol/l Retention: Int1: 54%, Int2: 54%, Int3: 65% | Int1: Internet ALIFE@Work, a distance-counseling lifestyle intervention program by phone Int2: Internet ALIFE@Work, a distance-counseling lifestyle intervention program by internet Int3: Usual care (self help materials on overweight, physical activity and healthful diet brochures) | Internet or mobile phone | Duration: 6 mos. intervention, 2 yrs. follow up Contacts: Int1Phone calls every 2 wks Int 2 self-paced Maximum 10 counseling contacts during 6-mo Intervention adherence: Used modules (%) Int 1: 93.2% Int 2: 87.5% Counseled on all modules (%) Int 1: 64% Int 2: 17% Interventionist: Int1: Counselors (dieticians, physical activity specialists) Int2: Counselors (dieticians, physical activity specialists) Int3: NA | Completer’s analysis (n=141) 24 mos. Total chol Δ, mg/dL, M difference (95% CI) : Int1 vs. Int3: 0.0 (−0.3, 0.3) Int2 vs. Int3: −0.1 (−0.4, 0.2) |
Yoo et al. 2009160 Design: 2-group RCT Outcome: HbA1c (%) Setting: Community Country: South Korea | N = 123 Int1: n = 62 Int2: n = 61 Mean age (SD) : Int1: 57.0 (9.1) yrs. Int2: 59.4 (8.4) yrs. Women: 47.2% BMI: 25.6 kg/m2 Total chol: 4.6 mmol/l Retention: Int1: 91% Int2: 89% | Int1: Ubiquitous Chronic Disease Care (UCDC) system using mobile phones and webbased interaction. UCDC included device attached to mobile phone that transmitted blood glucose data. Ps received SMS reminder to check blood glucose, also tips via SMS 3 x’s/day. Physicians could follow the Ps’ data and send individualized messages as needed. Int2: Usual Care. Ps visited according to usual schedule and received usual care in the outpatient setting | SMS and internet | Duration: 3 mos. Contacts: Int1: two alarms daily to remind pts to measure blood glucose values and blood pressure as well as one alarm daily for weight. Additionally, each P received at least three SMS messages daily Int2: Dependent upon usual care routine Each P was seen at baseline and at 3-mo to collect anthropometric as well as laboratory data Intervention adherence: Int1: sent in glucose readings 1.84 ± 0.31 times per day with a compliance rate of 92.2 ± 15.4% Blood pressure readings sent in 1.72 ± 0.32 times per day with a compliance rate of 86.0 ± 16.2% Weight measurements were sent in 0.87 ± 0.20 times per day with a compliance rate of 87.4 ± 20.1% Interventionist: Int1: Automated Int2: NR | Completer’s analysis (n=111) Total chol Δ, mmol/l, M: Int1: −0.5 p<0.001 Int2: 0.0 p=0.882 p=0.011 LDL chol Δ, mmol/l, M Int1: −0.4 p<0.001 Int2: −0.1 p=0.628 p=0.025 |
Note: P or Ps= participant(s), N = total sample, n = subgroups, Int = Intervention group, 1, 2, etc. Con = control group, RCT = randomized control trial, Δ = change or difference, mo. = month, mos. = months, wk = week, wks = weeks, year = yr, Baseline = 0, cellular = cell or mobile phone, HER = electronic health record, NR = not reported, chol = cholesterol, LDL = low density cholesterol, HDL = high density cholesterol, TC = total cholesterol; automated = without a clinician who generates, tailors, or modifies the output.
Some of the existing publications were focused on design, rationale and testing accuracy of tools with no lipid outcomes available at this time.216,233–235. Others were focused on small studies with inconclusive results219,229,236, or pilot feasibility studies that did not provide adequate results.221,223 Only one peer-reviewed publication addressed the topic of consumer use of technology as standalone tools specifically for lipid disorders (Table 6).237 The vast majority of publications did not meet criteria for inclusion in the tables due to the absence of a control group, or the fact that they did not include lipids as a primary outcome. However, there were a number of promising studies in this group of papers. In a quasi-experimental study, Park et al. showed that the use of a website and SMS improved total cholesterol, HDL cholesterol, LDL cholesterol, and triglycerides.224 Onescu et al. described a device that works with a smartphone camera to measure cholesterol.235 This device, if accurate and easily used, might show promise as a technology to allow self-monitoring of one’s serum cholesterol; this should be evaluated in future research. Studies reported in design papers by Redfern et al. and Chow et al. show promise in the future management of dyslipidemia.233,238 RCTs testing interventions that specifically target lipid reduction are needed, as there are no existing studies in this area. One study used electronically monitored medication blisters and a reminder system and reported that total cholesterol improved; however the study ended early.230 Supplying patients with smartphones with Bluetooth enabled blood pressure, glucometers, and a website for tracking showed decreases in total cholesterol; however, these studies did not include a control group.221,223
In a search of the Android and Apple App Stores for the term cholesterol, over 400 apps were found. This wide array of apps included information delivery, education, motivation, self-monitoring, lifestyle, drug therapies, and alternative therapies. However, none of the apps have been critically evaluated, and their development was not evidence-based. The absence of an empirical base for these apps is a major deficit in this potential area of treatment for such a highly prevalent condition.
Research has shown that education of consumers and self-management interventions can be beneficial for patients. Advances in information technology and consumer health-related mHealth are emerging as promising tools for facilitating management of dyslipidemia, e.g. home lipid testing using a smartphone, educational smartphone apps, and web portals for patients and providers. Although there is suggestive evidence of some benefit to their use, the amount of evidence-based literature in this area remains surprisingly low. High quality adequately powered trials are required to evaluate the role of mHealth-based interventions in dyslipidemia. Due to a lack of adequately tested tools, guidelines for use cannot be provided.
Gaps and Recommendations for future directions.
The paucity of well-controlled trials for the use of mHealth interventions specifically for lipid disorders is remarkable, considering the prevalence of dyslipidemia in the general population. One critical but inadequately researched area is how to engage patients and providers to initiate the use of mHealth devices in education, evaluation, self-monitoring and self-management of dyslipidemia. This first step may lay the groundwork for creation of treatment tools using mobile technology. Additional research is needed in how providers wish to approach the consumer in managing their dyslipidemia. It is possible that the use of other health-related apps such as mHealth apps focused on lifestyle behaviors could be used in this population and as indicated in a previous section, tools for the self-management of diabetes.
Summary of Representation of the Studies Reviewed.
Our review included a total of 69studies that investigated the use of mobile technologies to reduce CVD risk behaviors, which included 10 RCTs targeting weight loss, 14 on increasing physical activity, 14 aiming to improve smoking cessation, 15 on blood glucose management, 13 on hypertension management, and just three targeting lipid management. The majority was RCTs, for completeness we also included systematic reviews and meta-analyses in each topic area except dyslipidemia where none existed. Overall, the studies had samples that were mixed in ethnicity with a large portion being comprised of Whites, although one study on increasing physical activity and one on diabetes had 100% African American samples. Females made up the majority of many samples; however, one study conducted in a VA setting had 93% male representation. The smoking cessation studies enrolled younger individuals, usually 18 to 45 years of age, compared to most other studies that included participants up to 70–75 years of age. The geographic distribution was quite variable; weight loss studies were conducted in the US while studies focused on dyslipidemia were conducted outside of the US. Several studies focused on smoking cessation were conducted in Europe and seven of the15 RCTs in diabetes were done in South Korean and two were conducted in Europe. Thus, our evidence base also has limitations related to general representation resulting in limited knowledge on the effectiveness of augmenting traditional patient care with the use of mHealth-supported strategies in males, minority or underserved populations, and for specific risk behaviors (smoking) or conditions (diabetes) among US populations. Other limitations identified across the studies are the continued reliance on internet-based or SMS for interventions and the somewhat limited use of advances in mHealth strategies; however, this may be due to delays in publication. From a methodological perspective, several studies did not use ITT in their primary analysis and thus biased the results of their studies. Use of completer’s analyses was most evident in the studies focused on physical activity, blood pressure and dyslipidemia. And finally, several studies in the areas outside of weight loss used relatively brief interventions.
These limitations beg for some innovative changes in intervention studies using mHealth. First, a more rigorous approach to the analytic methods used and, second, inclusion of more diverse samples from an ethnic, socioeconomic and gender perspective. Finally, we need to use more adaptive and diverse methods in the testing of the rapidly changing mHealth devices and strategies and use approaches that can optimize the intervention designs and provision of efficacy data in a period shorter than the conventional 5-year RCT.239–241 Identifying the most effective features in a shorter time frame also will reduce costs and ensure the incorporation of the most effective components early in the development phase.
Part 3: How mHealth tools can improve health care delivery when partnered with health care providers
It is well established that a significant proportion of the CVD burden is preventable. Compared to pharmacological treatments for acute events and to secondary prevention, reductions in the prevalence of CVD risk factors have resulted in greater reductions in CVD-related mortality.242 However, the amount of information that must be conveyed and the support that is necessary to counsel and motivate individuals to engage in behaviors to prevent CVD is far beyond what can be accomplished in the context of face-to-face clinical consultations or through traditional channels such as patient education leaflets.243 The use of mHealth or mobile technologies has the potential to overcome these limitations and transform the delivery of health-related messages and ongoing interventions targeting behavior change. Moreover, the employment of monitoring devices (e.g., blue-tooth enabled BP monitors and blood glucose monitors) permits the sharing of important patient self-management parameters with health care providers in real time, as well as the delivery of feedback and guidance to the patient when they need it. Furthermore, the use of mHealth tools for monitoring provide the clinician data that far exceeds what can be measured in the brief clinical encounter and also reflects the status of physiological or behavioral measures in the person’s natural setting.
Part 4: Recommendations for Future Research
The development of drug and device therapies typically follows a standard path: Molecules that show promise in pre-clinical lab and animal testing are then evaluated in Phase1 human studies that provide an initial assessment of the agent’s safety. Those that survive Phase 1evaluation go onto larger Phase 2 and Phase 3 clinical trials where the impact of the therapy is evaluated in progressively larger populations. Only those that are found to be clearly efficacious and safe in these rigorous evaluations are then eligible for regulatory approval and release to the general population. Even once on market, drugs and devices often undergo further monitoring to assure that the findings seen in controlled trials are consistent with those seen in broader, more diverse patient populations and community settings.
In contrast, mHealth applications are often developed quite in the course of weeks to months as opposed to years for drugs. And, once developed, have traditionally not been regulated by governmental agencies. As such, these health apps may be offered to the public with limited to no information on the accuracy of their content; whether they are based on proven learning theory or behavioral interventional strategies; or whether they have undergone formal effectiveness and safety evaluation. While one may suggest that mHealth technologies do not require such careful scrutiny, there are arguments for such investigation. First, just like drugs or medical devices, these mobile technologies and applications have the potential to either improve health, or to be ineffective, or even to cause unanticipated harm. Second, without rigorous evidence behind them, it becomes difficult or impossible for care guidelines to recommend them or for clinicians to promote them. Third, the market is rapidly being flooded with these applications. Without evidence supporting the comparative usefulness of these, it is nearly impossible for the consumer (or clinician) to decide which to use. Finally, if a consumer who is motivated to modify his or her lifestyle selects an unhelpful product due to lack of information, there is a true lost opportunity and a chance to improve health has been missed.
The specific sections above reviewed current mobile applications and technologies for the treatment of obesity, and encouraging regular physical activity, smoking cessation, control of hypertension and dyslipidemia and for the treatment of diabetes. These literature searches uncovered a wide variety of products that have been developed. However, the reviews also identified the paucity of published empirical evaluation of their effectiveness. To date, many devices have no published evaluation, and those that have undergone evaluation are often limited to measuring customer satisfaction and user sustained engagement. While such intermediate measures are important, they fall far short of actually determining whether the users of these products had clinically meaningful changes in biologic parameters.
Several common themes were noted among each reviewed area. There were consistent concerns voiced regarding the designs of evaluative studies of mobile technologies. Often studies employed a pre-post design without concurrent controls, or better yet, randomized comparison group. Without such controls, a true measure of product effectiveness is likely to be over-estimated. Similarly, many studies relied on self-report that must again lead to overestimation of effectiveness in un-blinded evaluations. Additionally, many trials elected not to use an intention to treat perspective and thus again may overinflate the benefits of the intervention among those that used and stuck with the product.
To date, mobile technologies have been generally evaluated in motivated individuals and selected settings. These idealized conditions also will lead to exaggerations of the typical effectiveness that might be seen had the product been evaluated in general community practice or among diverse or underserved populations. Most studies were also of short duration leading to lingering questions regarding the products sustainability or ‘stickiness’. In particular, the fields of obesity and physical activity interventions are littered with interventions that work acutely but fail to support durable lifestyle change. And perhaps most challenging, studies to date have almost uniformly evaluated a single technology vs standard care and there have been almost no head to head studies comparing how various technologies compare relative to one another.
Beyond consistent questions regarding product safety and effectiveness, our review of the field found almost no studies that analyzed how products worked or user input in its development. Specifically, formative work had not defined which component(s) in an intervention are pivotal to success, or whether the products impact varies depending on the mode of use or delivery. Without these data, it is difficult to anticipate whether a similar but slightly different mobile technology would too be likely to be effective. Finally, these reviews pointed out the need for more implementation studies evaluating how to best incorporate these technologies (once proven) into a broader collaborative model of care.
Until such information is available, mobile app developers will continually face questions and doubts from the public, providers and payers. Just like any other product that claims to improve health, groups will want to know: Does the product work best when used in certain settings or among specific patient groups? Does the app duplicate or potentiate impact when it is combined with other traditional interventions (such as in person counselling)? If a specific mobile technology is found effective, in what cases can these findings be generalized among similar technologies in the class? Are the effects seen durable or does the intervention’s impact wane over time? And are there any unintended consequences associated with the device and program it’s used in?
Producing this evidence must be a shared responsibility. In the future, manufacturers will likely come under increasing pressure from regulatory agencies to produce evidence of effectiveness before marketing. Insurers are also likely to demand proof of durable effectiveness before they are willing to cover these services. However, the sole responsibility for generating evidence should fall not only on the product developers but the researcher and clinical communities too must help to generate these needed data. Our review of the evidence to date, even with its flaws and limitations, clearly demonstrates the great potential mobile technologies can have to aid in lifestyle modification. Thus, clinicians should not conclude that mobile technologies are generally ‘unproven’ and thus can be ignored. It must be recalled that the current absence of evidence should not be used as evidence of an absence of effectiveness. Instead, we need to embrace the challenge of producing this needed evidence regarding how effective these new technologies are and how we can best adopt them into our practice to promote better patient health.
Acknowledgements:
We would like to acknowledge the contribution of individuals who assisted with the literature review and development of this manuscript: Mary Lou Klem, PhD, Annabel Kornblum, MPH, and Heather Alger, PhD.
Contributor Information
Lora E. Burke, University of Pittsburgh.
Jun Ma, Palo Alto Medical Foundation Research Institute.
Kristen M.J. Azar, Palo Alto Medical Foundation Research Institute.
Gary G. Bennett, Duke University.
Eric D. Peterson, Duke Clinical Research Institute.
Yaguang Zheng, University of Pittsburgh School of Nursing.
William Riley, NIH.
Janna Stephens, Johns Hopkins.
Svati H. Shah, Duke University Medical Center.
Brian Suffoletto, Independent Contributor.
Tanya N. Turan, Medical University of South Carolina.
Bonnie Spring, Northwestern University.
Julia Steinberger, University of Minnesota.
Charlene C. Quinn, University of Maryland School of Medicine.
References
Full text links
Read article at publisher's site: https://doi.org/10.1161/cir.0000000000000232
Read article for free, from open access legal sources, via Unpaywall: https://www.ahajournals.org/doi/pdf/10.1161/CIR.0000000000000232
Citations & impact
Impact metrics
Citations of article over time
Alternative metrics
Smart citations by scite.ai
Explore citation contexts and check if this article has been
supported or disputed.
https://scite.ai/reports/10.1161/cir.0000000000000232
Article citations
Unravelling the shift: exploring consumers' adoption or resistance of E-Pharmacy through behavioural reasoning theory.
BMC Public Health, 24(1):2789, 11 Oct 2024
Cited by: 0 articles | PMID: 39394074
A roadmap for empowering cardiovascular disease patients: a 5P-Medicine approach and technological integration.
PeerJ, 12:e17895, 30 Aug 2024
Cited by: 0 articles | PMID: 39224824 | PMCID: PMC11368085
Review Free full text in Europe PMC
Applications and outcomes of implementing telemedicine for hypertension management in COVID-19 pandemic: A systematic review.
PLoS One, 19(8):e0306347, 01 Aug 2024
Cited by: 0 articles | PMID: 39088489 | PMCID: PMC11293715
Review Free full text in Europe PMC
Safety and adherence to medications and self-care advice in oncology (SAMSON): pilot randomised controlled trial protocol.
BMJ Open, 14(7):e079122, 23 Jul 2024
Cited by: 0 articles | PMID: 39043598 | PMCID: PMC11268069
Digitalization of hypertension management: a paradigm shift.
Naunyn Schmiedebergs Arch Pharmacol, 15 Jun 2024
Cited by: 0 articles | PMID: 38878087
Review
Go to all (283) article citations
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Implications of American College of Cardiology/American Heart Association (ACC/AHA) Cholesterol Guidelines on Statin Underutilization for Prevention of Cardiovascular Disease in Diabetes Mellitus Among Several US Networks of Community Health Centers.
J Am Heart Assoc, 6(7):e005627, 03 Jul 2017
Cited by: 6 articles | PMID: 28673901 | PMCID: PMC5586289
Primary prevention of cardiovascular diseases in people with diabetes mellitus: a scientific statement from the American Heart Association and the American Diabetes Association.
Circulation, 115(1):114-126, 27 Dec 2006
Cited by: 423 articles | PMID: 17192512
Self-Care for the Prevention and Management of Cardiovascular Disease and Stroke: A Scientific Statement for Healthcare Professionals From the American Heart Association.
J Am Heart Assoc, 6(9):e006997, 31 Aug 2017
Cited by: 168 articles | PMID: 28860232 | PMCID: PMC5634314
Review Free full text in Europe PMC
Progress in prevention: the other epidemic.
J Cardiovasc Nurs, 24(6):426-428, 01 Nov 2009
Cited by: 1 article | PMID: 19858951