Scolaris Content Display Scolaris Content Display

Intervenciones educativas por aplicaciones móviles de salud para la insuficiencia cardíaca

Contraer todo Desplegar todo

Antecedentes

La insuficiencia cardíaca (IC) es una enfermedad crónica con un impacto significativo en la calidad de vida y presenta muchos desafíos para las personas a las que se les diagnostica, debido a un régimen diario aparentemente complejo de autocuidado que incluye fármacos, control de peso y síntomas, identificación de signos de deterioro y seguimiento e interacción con múltiples servicios sanitarios. La educación es vital para comprender la importancia de este régimen y para adherirse a él. Tradicionalmente, la educación se ha impartido a los pacientes con insuficiencia cardíaca de forma presencial, ya sea en un entorno comunitario u hospitalario, utilizando materiales impresos o presentaciones en vídeo/DVD. En una época en la que la tecnología evoluciona rápidamente y se utilizan teléfonos inteligentes y tabletas, la tecnología basada en la asistencia sanitaria por móvil (definida por la Organización Mundial de la Salud como las tecnologías móviles e inalámbricas para lograr objetivos de salud) es una forma innovadora de impartir educación en salud que tiene la ventaja de poder llegar a los pacientes que no pueden o no quieren acceder a los programas y servicios tradicionales de educación sobre la insuficiencia cardíaca.

Objetivos

Revisar y cuantificar sistemáticamente los posibles efectos beneficiosos y perjudiciales de la educación impartida mediante aplicaciones móviles de salud para los pacientes con insuficiencia cardíaca.

Métodos de búsqueda

Se realizó una extensa búsqueda en bases de datos y registros bibliográficos (CENTRAL, MEDLINE, Embase, CINAHL, PsycINFO, IEEE Xplore, ClinicalTrials.gov y en la Plataforma de registros internacionales de ensayos clínicos de la OMS (ICTRP), utilizando términos para identificar la IC, educación y aplicaciones móviles de salud. Se realizaron búsquedas en todas las bases de datos desde su creación hasta octubre de 2019 y no se impuso ninguna restricción en cuanto al idioma de publicación.

Criterios de selección

Se incluyeron estudios si se realizaron como un ensayo controlado aleatorizado (ECA), con la participación de adultos (≥ 18 años) con un diagnóstico de IC. Se incluyeron ensayos que comparaban la educación en salud por móvil, como los programas educativos por Internet y basados en la web para su uso en teléfonos inteligentes y tabletas (incluidas las aplicaciones) y otros dispositivos móviles, mensajes de SMS y programas educativos impartidos por medios sociales, versus la atención habitual de la IC.

Obtención y análisis de los datos

Dos autores de la revisión, de forma independiente, seleccionaron los estudios, evaluaron el riesgo de sesgo y extrajeron los datos de todos los estudios incluidos. Se calculó la diferencia de medias (DM) o la diferencia de medias estandarizada (DME) para los datos continuos y el odds ratio (OR) para los datos dicotómicos con un intervalo de confianza (IC) del 95%. Se evaluó la heterogeneidad mediante la estadística I2 y se evaluó la calidad de la evidencia utilizando los criterios GRADE.

Resultados principales

En esta revisión, se incluyen cinco ECA (971 participantes) de intervenciones de educación en salud impartidas mediante aplicaciones móviles para pacientes con IC. El número de participantes en el ensayo osciló entre 28 y 512. La edad media de los participantes estuvo entre 60 y 75 años, y el 63% de los participantes en todos los estudios eran hombres. Los estudios procedían de Australia, China, Irán, Suecia y los Países Bajos. La mayoría de los estudios incluyeron a participantes con IC sintomática, NYHA clase II ‐ III.

Tres estudios abordaron el conocimiento de la IC, revelando que el uso de programas educativos impartidos mediante móviles no mostró evidencia de una diferencia en el conocimiento de la IC en comparación con la atención habitual (DM 0,10; IC del 95%: ‐0,2 a 0,40; P = 0,51; I2 = 0%; tres estudios, 411 participantes; evidencia de baja calidad). Un estudio que evaluó la autoeficacia informó que ambos grupos de estudio tenían altos niveles de autoeficacia al inicio del estudio y una incertidumbre en la evidencia de la intervención (DM 0,60; IC del 95%: ‐0,57 a 1,77; P = 0,31; un estudio, 29 participantes; evidencia de muy baja calidad). Tres estudios evaluaron el autocuidado de la IC utilizando diferentes escalas. No se agruparon los estudios debido a la naturaleza heterogénea de las medidas de resultado, y la evidencia es incierta. Ninguno de los estudios informó eventos adversos. Cuatro estudios examinaron la calidad de vida relacionada con la salud (CdVRS). Hubo incertidumbre en la evidencia para el uso de la educación impartida mediante aplicaciones móviles en cuanto a la CdVRS (DM ‐0,10; IC del 95%: ‐2,35 a 2,15; P = 0,93; I2 = 61%; cuatro estudios, 942 participantes; evidencia de muy baja calidad). Tres estudios informaron sobre la hospitalización relacionada con la IC. El uso de educación en salud impartida mediante aplicaciones móviles puede lograr poca o ninguna diferencia en las hospitalizaciones relacionadas con la IC (OR 0,74; IC del 95%: 0,52 a 1,06; P = 0,10; I2 = 0%; tres estudios, 894 participantes; evidencia de baja calidad). Se disminuyó la calidad de los estudios debido a las limitaciones en el diseño y la ejecución de los mismos, la heterogeneidad, los amplios intervalos de confianza y a que había menos de 500 participantes en el análisis.

Conclusiones de los autores

Se encontró que el uso de intervenciones de educación en salud ofrecidas mediante aplicaciones móviles para pacientes con IC no muestra evidencia de una diferencia en el conocimiento de la IC; incertidumbre en la evidencia de autoeficacia, autocuidado y calidad de vida relacionada con la salud; y puede resultar en poca o ninguna diferencia en las hospitalizaciones relacionadas con la IC. La identificación de los estudios que se están llevando a cabo actualmente y los que están en espera de clasificación indican que se trata de un área de investigación de la que surgirá nueva evidencia a corto y largo plazo.

PICO

Population
Intervention
Comparison
Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Intervenciones de educación para la insuficiencia cardíaca administradas mediante teléfonos inteligentes, tabletas e Internet

Pregunta de la revisión

¿Cuál es la evidencia de ensayos controlados aleatorizados para las intervenciones educativas ofrecidas mediante aplicaciones móviles de salud sobre el conocimiento de la insuficiencia cardíaca (IC), el autocuidado y la autoeficacia a las personas con IC en comparación con los métodos tradicionales para educar al paciente?

Antecedentes

La educación es vital para el autocuidado (actividades que las personas realizan con la intención de mejorar su salud, prevenir enfermedades, limitar enfermedades y restaurar su salud) en la IC. Tradicionalmente, se ha educado en salud a las personas con insuficiencia cardíaca de forma presencial, utilizando materiales impresos o presentaciones en vídeo/DVD. En una época en la que la tecnología evoluciona rápidamente y se utilizan teléfonos inteligentes y tabletas, la tecnología basada en la asistencia sanitaria por móvil es una nueva forma de proporcionar educación en salud, con la ventaja de poder llegar a las personas que no pueden o no quieren acceder a los programas y servicios tradicionales de educación en materia de salud.

Fecha de la búsqueda

Se encontraron estudios mediante búsquedas realizadas en octubre de 2019.

Características de los estudios

En esta revisión, se incluyen cinco ensayos controlados aleatorizados (971 participantes) de intervenciones de educación en salud mediante aplicaciones móviles para personas con IC. La edad media de los participantes osciló entre los 60 y los 75 años y el 63% de los participantes eran hombres. Los estudios procedían de Australia, China, Irán, Suecia y los Países Bajos.

Resultados clave

Cinco estudios abordaron el conocimiento de la IC; se encontró que el uso de programas de educación en salud por móvil no mostró evidencia de una diferencia en el conocimiento de la IC en comparación con la atención habitual. Un estudio que evaluaba la autoeficacia informó de incertidumbre de la evidencia para la educación en salud por aplicaciones móviles en comparación con la atención habitual. Tres estudios evaluaron el autocuidado de la IC usando diferentes escalas. No se combinaron los estudios, debido a las diferencias entre las medidas de desenlace, y la evidencia es incierta. Los estudios no informaron acerca de ningún efecto secundario de las intervenciones. Cuatro estudios examinaron la calidad de vida relacionada con la salud y mostraron incertidumbre en la evidencia entre la educación impartida por aplicaciones móviles de salud y la atención habitual. Tres estudios informaron sobre las tasas de hospitalización relacionadas con la IC. El uso de educación en salud impartida por aplicaciones móviles puede lograr poca o ninguna diferencia en las hospitalizaciones relacionadas con la IC.

Calidad de la evidencia

Se calificó la calidad de la evidencia de muy baja a baja, debido a las limitaciones en el diseño y la ejecución del estudio y a que hubo menos de 500 participantes en el análisis.

Conclusión

No hay evidencia de que exista una diferencia en el conocimiento de la IC con el uso de intervenciones de educación en salud por aplicación móvil para las personas con IC. Hay incertidumbre en la evidencia para la autoeficacia, el autocuidado y la calidad de vida relacionada con la salud. Puede haber poca o ninguna diferencia en las hospitalizaciones relacionadas con la IC, en comparación con la atención habitual. Por "atención habitual" se entiende en este caso la inscripción en un programa de control de la insuficiencia cardíaca (clínico o domiciliario). Esta es un área de investigación de la IC de la que surgirá más evidencia a corto y largo plazo.

Authors' conclusions

Implications for practice

In a real‐world setting there are notable factors to consider when implementing patient education interventions. User preferences and the acceptability of the intervention are paramount, along with consideration of patient factors such as cognition, which will have a significant impact on the use and effect of the intervention. Provision of information alone, in any format, without consideration of the ability of the patient to access, read and interpret that information will be insufficient.

Cognitive impairment is common in older HF patients and presents complex challenges for education (Cameron 2017). Prevalence of cognitive impairment is estimated to be between 50% to 80% in the heart‐failure population depending on age (Vogels 2007). Evidence supports the need for tailoring of treatment, strengthened with clear communication to reduce readmission rates, mortality and functional decline of older people (Hickman 2015). mHealth education interventions could improve communication and tailoring of care which can be customised to cognition, language and healthcare settings for each patient or group of patients.

At this point in time, there is no evidence of a difference in HF knowledge and little to no difference in HF‐related hospitalisation. There is uncertainty in the currently available evidence on self‐efficacy, self‐care and health‐related quality of life which will need to be confirmed with further research.

Implications for research

There are substantial implications for future research into the use of technology to facilitate patient education and support for people with heart failure. We experienced some challenges synthesising and meta‐analysing data from the included studies, due to inconsistencies in outcomes, measurement tools and scales used across the study populations. Future studies should look to measure heart‐failure education, self‐efficacy, self‐care and health‐related quality of life using validated tools and scales. Heart‐failure hospitalisation, medication adherence and patient acceptance of the intervention and adherence to it would also be valuable outcomes to assess. Taxonomies, such as those by Krumholz 2006 provide a clear guide. It would also be beneficial if future studies could report details of the theoretical underpinnings of their education intervention and relationship to behaviour change, as this will allow for better comparisons between studies and consideration of factors leading to heterogeneity (Michie 2011). As education should be provided in the context of integrated disease management, it will be valuable to consider the setting and care delivered alongside the provision of education.

Summary of findings

Open in table viewer
Summary of findings 1. mHealth education intervention compared to usual care in heart failure

mHealth education intervention compared to usual care in heart failure

Patient or population: Adults with heart failure
Setting: Hospital and community
Intervention: mHealth education intervention
Comparison: Usual care

Outcomes

Tools

Mean length of follow‐up (months)

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with mHealth education intervention

Heart failure knowledge Analysis 1.1

Dutch HF Knowledge Questionnaire

56 (39.6)

The mean heart failure knowledge was 12.7

MD 0.10 higher
(−0.20 lower to 0.40 higher)

411
(3 RCTs)

⊕⊕⊝⊝
LOWa,d

Higher score equals a better outcome

Heart failure self‐efficacy Analysis 1.2

Self‐efficacy for Managing Chronic Disease Scale

28

The mean heart failure self‐efficacy was 7.4

MD 0.60 (−0.57 lower to 1.77 higher)

29 (1 RCT)

⊕⊝⊝⊝
VERY LOWc,d,e

Higher score equals a better outcome

Heart failure self‐care Analysis 1.3

European Heart Failure Behaviour Scale (EHFBS)

Self‐Care Heart Failure Index (SCHFI)

56 (39.6)

One study showed an improvement in the SCHFI maintenance subscale in the intervention group (MD 9.9, 95% CI −3.6 to −23.6), whereas the usual‐care group decreased over time (MD −3.5, 95% CI −10.3 to 1.3).

Two studies reported improvements on the EHFBS for the intervention groups.

411
(3 RCTs)

⊕⊝⊝⊝
VERY LOWa,d,f

Due to the high heterogeneity observed in the analysis, we decided not to pool the studies for this outcome

Adverse events

No studies reported this outcome

Health‐related quality of life Analysis 1.4

Kansas City Cardiomyopathy Questionnaire

Minnesota Living with HF Questionnaire

84

The mean health‐related quality of life was 51.7

MD −0.10 lower
(−2.35 lower to 2.15 higher)

942
(4 RCTs)

⊕⊝⊝⊝
VERY LOWa,b

Higher score equals a better outcome

Heart failure‐related hospitalisations Analysis 1.5

84

Study population

OR 0.74
(0.52 to 1.06)

894
(3 RCTs)

⊕⊕⊝⊝
LOW a,e

275 per 1000

219 per 1000
(87 to 454)

*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: Confidence interval; MD: mean difference; OR: Odds ratio; RCT: randomised controlled trial.

GRADE Working Group grades of evidence
High certainty: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

aDowngraded by one level for 'limitations in study design and execution', as the studies were rated at high risk of bias in multiple domains (performance and detection, attrition bias or other bias).
bDowngraded by one level for 'inconsistency' due to substantial heterogeneity (I2 = 61%).
cDowngraded by one level for 'limitations in study design and execution', as the study was rated at high risk of bias in multiple domains (performance, detection and other bias).
dDowngraded by one level for 'imprecision' as there were fewer than 500 participants in the analysis.

eDowngraded by one level for 'imprecision' as the confidence intervals were wide.

fDowngraded by one level for 'inconsistency' due to high heterogeneity that precluded meta‐analysis (I2 = 86%).

Background

Heart failure (HF) is a chronic disease impacting on quality of life and leading to adverse outcomes (Atherton 2018). Living with a life‐limiting, chronic illness such as HF requires significant changes in lifestyle, such as changes to diet, managing and adhering to several medications, caring for oneself through appropriate exercise and energy conservation (Hammash 2017; Jaarsma 2017). Decades of research have underscored that education is necessary to promote a desired health or behavioural outcome. There are many different methods of delivering health education to people with HF, which may include (but are not limited to) one‐on‐one or group consultations with a nurse or allied health professional (Baptiste 2014), internet‐based interventions, smartphone and tablet apps, or printed reading materials (Baker 2011; Seto 2012). The effectiveness of each method of delivering health education, in particular, using mHealth education tools and platforms for people with heart failure, has not yet been examined in an extensive systematic review or meta‐analysis.

The use of eHealth (defined by the World Health Organisation (WHO) as "the use of information and communication technologies for health") (WHO 2011) has the potential to provide innovative solutions to health issues and is a key 'enabling' technology to improve care and the experience of care for those living with chronic conditions. However, there are significant societal and professional constraints associated with its use, including legal, ethical and data protection issues (Cowie 2016). In addition, health professionals may be resistant to such innovations, particularly if the evidence for the impact on quality of care is less than robust (Cowie 2016). Based on their recent position statement on eHealth, the vision of the European Society of Cardiology is to play a proactive role in all aspects of eHealth by helping to develop, assess and implement effective innovations to support cardiovascular health and health‐related activity across Europe (Cowie 2016). This vision is delivered through investment in research, education, training and advocacy (Cowie 2016). The ESC e‐Cardiology Working Group position paper highlighted the importance of patient‐education programmes, digital health workflow design, uniform European‐wide digital health legislation, data standardisation and interoperability assurance and the role of the digital health industry, health‐insurance industry, patient organisations and professional organisations in addressing the challenges in digital health implementation in Europe (Frederix 2019).

Similarly, the American Heart Association has highlighted the importance of integrating the practice of telehealth into traditional healthcare delivery systems and barriers to its effective implementation across broad populations of patients and providers (Schwamm 2017). Their ultimate goal is to increase the access of telehealth technologies to patients with cardiovascular diseases and stroke through partnerships with other organisations to achieve the following objectives:

  1. ensure that coverage mandates exist in all states so that third‐party payers must offer specific, evidence‐based telehealth interventions as covered services;

  2. ensure all properly‐trained providers are deemed eligible providers without restricted networks that would limit reimbursement by the provider;

  3. encourage development of simpler, less expensive technology platforms to keep the patient burden and costs to healthcare systems low;

  4. ensure that large e‐health record systems incorporate telehealth and make it compatible with traditional health records;

  5. encourage development of improved education for providers to increase adoption; and

  6. ensure adoption of telehealth does not sacrifice quality in the name of cost savings (Schwamm 2017).

Description of the condition

Heart failure is defined by the National Heart Foundation of Australia and Cardiac Society of Australia and New Zealand: guidelines for the prevention, detection, and management of heart failure in Australia 2018 as "a complex clinical syndrome with typical symptoms and signs which generally occur on exertion, but can also occur at rest. It is secondary to an abnormality of cardiac structure or function that impairs the ability of the heart to fill with blood at normal pressure or eject blood sufficient to fulfil the needs of the metabolising organs" p. 1136 (Atherton 2018). It is characterised by typical symptoms such as breathlessness, ankle swelling and fatigue which may be accompanied by signs, e.g. elevated jugular venous pressure, pulmonary crackles and peripheral oedema (Ponikowski 2016). Heart‐failure patients often experience many of these symptoms at the same time, which eventually lead to decreased functional status, frequent hospitalisation and a diminished quality of life (Hoekstra 2013). A 2016 update from the American Heart Association estimated that 5.7 million people in America have heart failure and more than 915,000 new diagnoses occur every year (Mozaffarian 2016). Globally, the prevalence of HF is estimated to be 2% to 3% (Ambrosy 2014; Bui 2011; Dunlay 2017), and increases with age. At least 20% of hospital admissions among people older than 65 years are due to HF (Dunlay 2017).

The goal of HF management is to relieve symptom burden, prevent hospitalisation, and improve survival (Ponikowski 2016). Over the last few decades, the use of effective treatment, including pharmacological agents, non‐surgical devices, as well as disease‐management programmes, have reduced the risk of hospitalisation and sudden cardiac death, and improved overall mortality among people with HF (Al‐Khatib 2018; Kristensen 2018; Nochioka 2018; Velazquez 2018). However, the prognosis of HF still remains poor (Al‐Khatib 2018; Schmidt 2016). Approximately 30% of people hospitalised with HF are readmitted within 30 to 90 days (Ambrosy 2014; Pandolfi 2017), and non‐adherence is a common cause of admission. Among 54,322 admissions from 236 hospitals in the USA, non‐adherence to medication or dietary requirements accounted for 10.3% of HF admissions (Ambardekar 2009). There is high‐quality evidence to support HF disease‐management programmes (case management‐type interventions led by a specialist HF nurse), with benefits seen in HF‐related and all‐cause readmissions and all‐cause mortality (Takeda 2019). Patient education is central to all disease‐management programmes (Ambardekar 2009; Angermann 2012; Bekelman 2015; Clark 2015; Jonkman 2016), but effective methods for delivering patient education are yet to be identified.

Description of the intervention

Traditionally, patient education has been conducted in a face‐to‐face manner using printed or written materials, with perhaps the use of a video to support key messages. Information communication technology is now commonplace in both the developed and developing worlds (WHO 2011) and as healthcare rationalisation reforms reduce or restrict face‐to‐face services, using information communication technology to assess, educate, support and interact with patients is becoming more common. Using technologies such as smartphone and tablet apps, SMS messages and social media‐delivered education programmes will become more commonplace in the future. To date, many patients already expect these innovations to support modern healthcare delivery by facilitating a more personalised person‐centred care (Cowie 2016).

The interventions included in this review fall under the description of mHealth. mHealth is defined by the WHO as " the use of mobile and wireless technologies to support the achievement of health objectives" (WHO 2011).

mHealth‐delivered education interventions provide patients with disease‐specific information through a mobile platform. This mobile platform could be a tablet, a mobile phone or a laptop. Technology‐based mobile patient‐education interventions have been used for people living with many different conditions, such as depression (Pinto 2013), diabetes (Heinrich 2012; Pal 2014), and breast cancer (Jibaja‐Weise 2011). An example of such an intervention in heart failure is the 'Heart Failure Matters' (heartfailurematters.org) by the European Society of Cardiology. The interventions we examine in this review provide education to people with heart failure, with the purpose of increasing their knowledge and understanding of what heart failure is, symptoms of heart failure, and how these can be managed and prevented from exacerbation, and the importance of adhering to any prescribed therapies such as medications, diet or exercise. The intervention are aimed at the patient, but caregivers may also access the information provided to the patient.

Technological innovations are very heterogeneous and lack shared definitions, which limits the possibility of comparing and evaluating different programmes. This review addresses a critical gap by examining the effectiveness of technology‐based HF patient‐education tools and platforms. We examine mHealth education for people with HF, focusing on interventions which are delivered using technologies such as smartphone and tablet apps, SMS messages and social media‐delivered education programmes. We exclude TV programmes, videos (other than those incorporated into the included technologies), telephone calls, interactive voice response systems and paper‐based education materials, where these are used in isolation from one of the listed technologies. We provide a description of the mode, method, and frequency of the intervention, as well as conceptual underpinnings of the intervention, methods of instructional design and processes of promoting intervention fidelity. We identify both the structural and functional mechanisms of effective interventions to facilitate broad translation.

How the intervention might work

Self‐management is “the engagement of a person in activities which protect or promote their own health, manage their symptoms and the impact which their condition has on their life and also adhere to any prescribed therapies” (Gruman 1996). Self‐management can be learned through health‐education interventions, and targets improving knowledge about the disease, treatment, and outcomes (health literacy), so that the person is better informed and capable of engaging in activities which protect or promote their own health (e.g. fluid restriction), manage their symptoms (e.g. breathlessness and fatigue), the impact which heart failure has on their life, and also adherence to any prescribed medications. A range of patient, provider and healthcare system factors influence HF management. As a consequence, self‐care and patient outcomes and technological innovation are promising in promoting access to health education and self‐management strategies (Omura 2017).

Engaging clinicians and patients as partners in care is increasingly recognised as critical, and has led to an increased emphasis on shared decision‐making (Ting 2014). Implicit in this process is engaging the patient as an active partner, and this can challenge traditional paradigms of health education, where the patient is a passive recipient of care. In promoting effective self‐management, the capacity to tailor and target the approaches to an individual's needs is important. Innovative health education using eHealth and mHealth technologies may provide greater opportunities for patient and clinician engagement, as well as being better tailored to patient preferences and needs, be these cognitive, language, or sensory disabilities. These individual factors include understanding how factors such as health literacy, numeracy, cognitive capacity, cultural needs and health‐seeking behaviours contribute to both the learning experience and the capacity to adopt health information and education, and progress to self‐management (Davidson 2013; Dennison 2011; McNaughton 2013). Due to the comorbidity burden in heart failure, and the fact that it is a condition of the elderly (Dunlay 2017), factors such as cognition and psychomotor skills influence the capacity to both engage in and benefit from technological innovation in information delivery and promotion of behaviour change (Holden 2013). Recent research in neuroscience, cognitive and developmental psychology has also increased our understanding of how to actively engage learners in achieving a desired outcome. Cognitive impairment is an important consideration in understanding both the design and the feasibility and acceptability of educational interventions (Cameron 2017; Uchmanowicz 2017). Appreciating the diverse range of physical, social, psychological, economic and cultural factors that contribute to heart failure outcomes is important to promote both the efficacy and effectiveness of education interventions. As drivers of health outcomes are strongly mediated by social determinants of health, it is important to appraise health‐education interventions within this context (Marmot 2017).

In addition, mHealth‐delivered interventions can provide timely access to health information and communication with healthcare professionals to support self‐management and decision‐making (WHO 2011). In this review, we examine whether mHealth‐delivered education interventions work, which is important not only in implementing evidence‐based recommendations, but also in advancing the science of self‐care and self‐management. This is a novel and important focus for this review and is guided by a standardised taxonomy using criteria such as theoretical underpinning, mode of instructional design, and mode of delivery and capacity for interaction. Appreciating how eHealth‐ and mHealth‐based assisted learning incorporate concepts such as grounded cognition, spaced learning, and simulation in educational media is important, and both underscore the complexity and capability of intervention development (Koong 2014). Focusing solely on the medium of technology (such as SMS or social media support) in evaluating the efficacy of interventions is simplistic and less likely to be effective in translating research into practice.

Why it is important to do this review

Disease management has been shown to significantly improve patient outcomes and reduce costs of care for HF (Angermann 2012; Bekelman 2015). Although improvements in care and outcomes have been achieved through disease management, the range of patient, provider and healthcare‐system factors that influence HF management often remain unaddressed (Omura 2017), and disparities in HF outcomes are growing. As a consequence, the overall prognosis of HF remains poor (Schmidt 2016).

Health care is rapidly evolving, and information technology, which is evolving ever more rapidly, offers great promise as a tool to promote effective communication and learning that is highly individualised to patients' circumstances and needs in the developing and developed worlds (WHO 2011). Patient education is central to all disease‐management programmes (Ambardekar 2009; Angermann 2012; Bekelman 2015; Clark 2015; Jonkman 2016; Takeda 2019). It is not clear which of the various methods for delivering patient education are effective. A text messaging intervention to improve HF self‐management in a largely African‐American population has shown a positive impact on outcomes (Nundy 2013). Similarly, an education and coaching programme integrated with telehealth has also shown improved outcomes (Stut 2015). The ever‐increasing role of technology and access to health apps, social media and web pages, not all of which may have been underpinned by evidence or evaluated in any way, warrants the need to conduct this systematic review. The survey of global mHealth use undertaken by the WHO in 2011 identified effectiveness and cost effectiveness as key barriers to mHealth implementation (Cowie 2016).

Implementation of effective, tailored, technology‐driven patient‐education interventions that promote self‐management in heart failure may promote access to heart‐failure care across the trajectory and reduce growing health disparities.

In the current setting of increasing use of information communication technology worldwide, mHealth‐delivered interventions for patient education are becoming more available (Cowie 2016; Schwamm 2017). We do not believe that age is a barrier to the application of these interventions, as smartphones and tablet computers are becoming more common (Cowie 2016; Schwamm 2017). Although limited research exists to fully understand the use of information‐communication technology amongst typically older people with cardiovascular illness, a survey of 123 patients (mean age 51 years) at an outpatient cardiopulmonary clinic in a large tertiary hospital in Australia indicates that for this sampled population, information‐communication devices are becoming a part of everyday life, with most survey respondents engaging regularly with a computer (83%), mobile telephone (97%) and the internet (86%) (Disler 2015). Accessing health information online was also a common activity, with 74% regularly consulting online health information sites (Disler 2015). Although this study was undertaken at a single site, it does provide some insight into the patterns of use amongst people for whom the technologies examined in this review are being designed.

The National Heart Foundation of Australia and Cardiac Society of Australia and New Zealand guidelines strongly recommend educating patients and carers about HF and self‐management to decrease hospitalisation and mortality (Atherton 2018). This should begin soon after diagnosis, must be patient‐centred, appropriate to their level of health literacy, culturally appropriate and revised throughout the person's life (Atherton 2018). Similarly, the European Society of Cardiology guidelines have also emphasised the importance of education for self‐care (Ponikowski 2016). It has recommended that patients must be provided with sufficient up‐to‐date information to make decisions on lifestyle adjustment and self‐care (Ponikowski 2016). The American Heart Association guidelines also highlight that education must be tailored individually and must consider relevant comorbidities that may influence retention of information (Mozaffarian 2016).

Objectives

To systematically review and quantify the potential benefits and harms of mHealth‐delivered education for people with heart failure.

Methods

Criteria for considering studies for this review

Types of studies

We include only randomised controlled trials (RCTs). We include studies reported as full text and those published as abstract, as long as sufficient information was available to determine study protocols and outcomes. We excluded case reports, case‐control and cross‐sectional studies.

Types of participants

We included adults (≥ 18 years) with a diagnosis of chronic heart failure as in the National Heart Foundation of Australia and Cardiac Society of Australia and New Zealand and European Society of Cardiology 2016 guidelines (see also Description of the condition). We excluded studies which target general cardiac disorders rather than HF specifically. We excluded participants with comorbidities such as dementia which impair their ability to engage in education interventions. We excluded studies of people with end‐stage HF or other terminal conditions, or those receiving advanced HF therapies (e.g. left ventricular‐assist devices or heart transplant).

Types of interventions

We include trials comparing mHealth‐delivered education such as internet and web‐based education programmes for use on smartphones and tablets (including apps) and other mobile devices, SMS messages and social media‐delivered education programmes, versus usual HF care, which may include verbal discussion and/or provision of paper‐based education materials about HF, medications and self‐management of HF. Interventions could be one‐off (a single session) or could occur at regular intervals or on demand by the participant. We considered 'usual care' for each study to ensure that this did not include mHealth.

We exclude interventions that do not incorporate the technologies listed above, but rather only provide education by the following methods: TV programmes, videos (other than those incorporated into the included technologies), telephone calls, interactive voice response systems and paper‐based education materials, where these are used in isolation from one of the included technologies.

We exclude interventions where both control and intervention participants received different care other than the provision of education, which may include enrolment in a heart‐failure management programme (clinic‐ or home‐based), telemonitoring or structured telephone support programme (where signs and symptoms of HF and clinical assessments were communicated and responded to) (Inglis 2015a; Inglis 2011), cardiovascular rehabilitation, exercise programme, or enrolment in a drug trial.

Types of outcome measures

Primary outcomes

  1. Heart‐failure knowledge using validated tools

  2. Self‐efficacy using validated tools

  3. Self‐care using validated tools

  4. Adverse events

Secondary outcomes

  1. Health‐related quality of life using validated tools

  2. Heart failure‐related hospitalisations (number of participants hospitalised at least once)

  3. Medication adherence

  4. Cost effectiveness

Since publication of the protocol we have amended the primary study outcomes. We have removed all‐cause mortality as an outcome, and have added self‐care using validated tools.

We include studies that reported at least one of the following primary or secondary outcomes: heart failure knowledge, self‐efficacy or self‐care using validated tools; health‐related quality of life using validated tools; or heart failure‐related hospitalisations; or medication adherence. We have reported participant feedback on the intervention (acceptance and satisfaction) and cost of the intervention where available for studies that met the conditions for inclusion based on the primary and secondary outcomes given above.

We have nominated the primary outcomes (#1, #2, #3) as these are the most direct measures of HF knowledge, self‐care and self‐efficacy (a person’s belief that they can achieve a given task). We selected the secondary outcomes because these are commonly reported in such studies and may be indirect measures of participant knowledge.

Search methods for identification of studies

Electronic searches

We identified trials through systematic searches of the following bibliographic databases on 14 October 2019:

  • Cochrane Central Register of Controlled Trials (CENTRAL), through the Cochrane Register of Studies (CRS Web);

  • Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily and MEDLINE (Ovid, 1946 to October 11 2019);

  • Embase (Ovid, 1980 to 2019 week 41);

  • CINAHL (EBSCOHost, 1937 to 14 October 2019);

  • PsycINFO (Ovid, 1806 to week 41 2019).

We adapted the preliminary search strategy for MEDLINE (Ovid) (Appendix 1) for use in the other databases. We applied the Cochrane sensitivity‐maximising RCT filter (Lefebvre 2011) to MEDLINE (Ovid) and adaptations of it to the other databases, except CENTRAL.

We searched all databases from their inception to October 2019, and imposed no restriction by language of publication.

Searching other resources

We checked reference lists of all relevant primary studies and review articles for additional references. We also conducted a search of IEEE Xplore (ieeexplore.ieee.org/Xplore/home.jsp), ClinicalTrials.gov (www.ClinicalTrials.gov) and the WHO ICTRP Search Portal (apps.who.int/trialsearch/) in October 2019. We present the search terms for these in Appendix 2.

Data collection and analysis

Selection of studies

Two review authors (SCI, SA) independently screened titles and abstracts of all the potentially includable studies we identified from the search using Covidence (Mavergames 2013), and coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. If there were any disagreements, we asked a third review author (HD) to arbitrate. RP assisted with screening a smaller number of titles. We retrieved the full‐text study reports/publication and two review authors (SCI, SA) independently screened the full text, identified studies for inclusion, and identified and recorded reasons for exclusion of the ineligible studies. We resolved any disagreement through discussion or, if required, we consulted a third review author (HD). We identified and excluded duplicates and collated multiple reports of the same study, so that each study rather than each report is the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Moher 2009) and 'Characteristics of excluded studies' table.

Data extraction and management

We used a data‐collection form for study characteristics and outcome data which we had piloted on at least one study in the review. We extracted the following study characteristics.

  1. Methods: study design, duration of study, number of study centres and location, study setting, withdrawals, and date of study.

  2. Participants: N, mean age, age range, sex, severity of condition, diagnostic criteria, baseline left ventricular ejection fraction, inclusion criteria, and exclusion criteria.

  3. Interventions: intervention, comparison.

  4. Outcomes: primary and secondary outcomes specified and collected, and time points reported.

  5. Notes: funding for trial, and notable conflicts of interest of trial authors.

Two review authors (SCI, SA) independently extracted outcome data from included studies. We resolved disagreements by consensus or by involving a third review author (HD). One review author (SA) transferred data into Review Manager 5 (RevMan 2014) file. We double‐checked that data were correctly entered by comparing the data presented in the review with the study reports. A second review author (HD) spot‐checked study characteristics for accuracy against the trial reports.

Assessment of risk of bias in included studies

Two review authors (SCI, SA) independently assessed risks of bias for each study, using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We resolved any disagreements by discussion or by involving another review author (HD). We assessed the risks of bias according to the following domains.

  1. Random sequence generation.

  2. Allocation concealment.

  3. Blinding of participants and personnel.

  4. Blinding of outcome assessment.

  5. Incomplete outcome data.

  6. Selective outcome reporting.

  7. Other potential bias.

In accordance with the Cochrane 'Risk of bias' assessment tool, we graded each potential source of bias as high, low or unclear, and provided a quote from the study report together with a justification for our judgement in the 'Risk of bias in included studies' table (Sterne 2017). We summarised the 'Risk of bias' judgements across different studies for each of the domains listed. Where information on risk of bias relates to unpublished data or correspondence with a trialist, we note this in the 'Risk of bias in included studies' table.

When considering treatment effects, we have taken into account the risks of bias for the studies that contributed to that outcome.

Assessment of bias in conducting the systematic review

We conducted the review according to this published protocol and reported any deviations from it in this review.

Measures of treatment effect

We analysed dichotomous data as odd ratios (ORs) with 95% confidence intervals (CIs), and continuous data as mean differences (MDs) or standardised mean differences (SMDs) with 95% confidence intervals.

We had planned to narratively describe skewed data reported as medians and interquartile ranges, but none of the included studies reported outcomes in this format.

Unit of analysis issues

We did not anticipate any unit‐of‐analysis issues with the included studies. Two of the included studies reported one study time point while one study reported three time points. The first time point was used for analysis. One study had three arms. We used only data from two arms for analysis, as the third arm did not meet the review inclusion criteria. One study used a measurement tool which runs in the opposite direction. We modified the result of this study by multiplying the mean scores by −1 to ensure that both studies run in the same direction.

Dealing with missing data

We contacted investigators in order to verify key study characteristics and to obtain missing numerical outcome data where possible (e.g. when a study is presented as abstract‐only). We used the Revman calculator to compute any missing standard deviations from other information in the publication.

Assessment of heterogeneity

We used the I2 statistic to measure heterogeneity among the trials in each analysis (Deeks 2017). We investigated potential sources of heterogeneity where I2 was greater than 40%. We reported similarities between interventions, participants, design, and outcomes in the Included studies subsection. We checked forest plots visually for signs of heterogeneity.

Assessment of reporting biases

We had planned to creat funnel plots, but due to the small number of included studies these are unlikely to be useful for exploring possible small‐study biases for the primary outcomes.

Data synthesis

We have undertaken meta‐analyses only where this is meaningful, i.e. if the interventions, technologies, participants, and the underlying clinical questions are similar enough for pooling to make sense. We planned to use a random‐effects model, but used a fixed‐effect model due to the population sizes of the included studies. We used inverse variance as the method of analysis.

'Summary of findings' and quality of the evidence

Two of the review authors (SCI, SA) also assessed the quality of evidence according to GRADE (Atkins 2004) by constructing a 'Summary of findings' table for the main outcomes using the GRADEPro tool (GRADEproGDT 2015; Schünemann 2017). We reported the following outcomes in summary of findings Table 1: heart failure knowledge; self‐efficacy; self‐care; adverse events; health‐related quality of life; heart failure‐related hospitalisations.

Subgroup analysis and investigation of heterogeneity

Owing to the small number of included studies, we have not been able to stratify findings by mode of delivery or intensity of the intervention, nor to perform any subgroup analyses as we had planned.

We had planned to carry out the following subgroup analysis: studies where mean/median participant age is more than 70 years, but we could not undertake this due to the small number of included studies.

Sensitivity analysis

Owing to the small number of included studies we have not performed sensitivity analyses as planned, to include only studies with a low risk of bias.

Reaching conclusions

We based our conclusions only on findings from the quantitative or narrative synthesis of included studies for this review. Our implications for research suggest priorities for future research and outline the remaining uncertainties in the area.

Results

Description of studies

Results of the search

We identified 7896 records in total, including the following from each database:

  • CENTRAL n = 4762

  • MEDLINE (OVID) n = 853

  • EMBASE (OVID) n = 1446

  • CINAHL Plus with Full text (EBSCO) n = 659

  • PsycINFO (OVID) n = 176

We found additional references by searching the following:

  • IEEE Xplore n = 56

  • ClinicalTrials.gov n = 59

  • WHO ICTRP Search Portal n = 28

  • Personal files and contacts = 1

We excluded 515 duplicate references. We screened 7525 titles and abstracts and excluded 7299 irrelevant records. We retrieved full‐text reports for the remaining 226 records. After reading the full texts, we excluded 161 studies (188 references) as they did not meet the review eligibility criteria. We have provided primary reasons for exclusion in the Characteristics of excluded studies table and in Figure 1. We also searched several systematic reviews and meta‐analysis (Flodgren 2015; Konstam 2010; Kotb 2015; Kraai 2011; Krauskopf 2019; Kuijpers 2013; Lazkani 2016; Malik 2014; Mallidi 2011; McDermott 2013; Nakamura 2014; Pandor 2013; Peate 2013; Radhakrishnan 2012; Van Spall 2017; Verheijden Klompstra 2011) found among the full‐text reports for additional references (see Additional references). One study is awaiting classification owing to insufficient information on design, intervention and analysis (see Characteristics of studies awaiting classification). Seventeen studies are classified as ongoing, as they have yet to start recruitment or finalise the analysis (see Characteristics of ongoing studies). Finally, we included five trials (15 references) in the quantitative synthesis.


Study flow diagram.

Study flow diagram.

Included studies

We include five full‐text peer‐reviewed studies of mHealth‐delivered education in the form of web‐based self‐care education programmes for use on smartphones, computer/laptop and tablets, compared to usual care in the meta‐analysis (see Characteristics of included studies).

  • Trials ranged in size from 28 participants in Bashi 2016 to 512 participants in Li 2016;

  • Mean age of participants ranged from 60 years to 75 years;

  • Mean percentage of men was 63%, ranging from 57% to 78%;

  • The studies originated from Australia (1), China (1), Iran (1), Sweden (1) and The Netherlands (1);

  • The studies included participants with symptomatic heart failure, NYHA Class II ‐ III.

Bashi 2016 randomised 29 participants (15 to the intervention group and 14 to the control group). One participant in the intervention group was lost to follow‐up, leaving 28 participants included in the analysis. Hagglund 2015 randomised 82 participants. Forty‐two received the mHealth intervention and 40 were allocated to usual care. Ten participants withdrew consent before the system was installed and were therefore not included in the analysis. Seventy‐two participants were included in the final analysis.

Two studies (Hagglund 2015; Wagenaar 2019) reported their funding source. Hagglund 2015 received funding from the Swedish National Quality Registry and Care Ligo which provided the OPTILOGG systems used in the study and paid a small stipend for each included participant. Wagenaar 2019 received funding from a 'CareWithinReach' foundation.

The length of intervention and follow‐up varied between the studies. The duration of intervention in three studies (Bashi 2016; Hagglund 2015; Jalali 2018) was three months, while another (Wagenaar 2019) lasted 12 months. Three studies had one follow‐up time point (Bashi 2016‐ one month; Jalali 2018‐ three months and Li 2016‐ six months) compared to two studies with multiple follow‐up time points (Hagglund 2015‐ three and six months and Wagenaar 2019‐ three, six and 12 months, respectively).

The type of intervention also varied between the studies. Two studies (Bashi 2016; Wagenaar 2019) were web‐based educational interventions to support self‐care which required access to a website, while participants in Hagglund 2015 received a tablet device containing information about HF and lifestyle advice according to the guidelines. Two studies (Jalali 2018; Li 2016) used SMS messages to deliver information about HF, signs and symptoms and lifestyle advice. In terms of usual care, participants in two studies received either an information sheet with advice on HF treatment (Hagglund 2015) or comprehensive educational information on topics such as medication, nutrition, exercise and psychosocial issues from the HF outpatient clinic (Bashi 2016). 'Usual care' in Wagenaar 2019 also received HF education but in conjunction with up‐titration of HF medication, optimising adherence and personalised lifestyle advice by a HF nurse in face‐to‐face consultations. If necessary, additional telephone consultations were also provided.

Excluded studies

We excluded most studies for the following reasons:

  • Ineligible intervention (not mHealth education): n = 110;

  • Ineligible study design (not an RCT): n = 40;

  • Ineligible outcome measures (did not report the primary or secondary outcomes of interest): n = 3

  • Ineligible comparator (not usual care): n = 2

  • Opinion piece/review paper: n = 2

  • Different participant population: n = 5

  • Ineligible indication n = 1

See also Characteristics of excluded studies tables.

Risk of bias in included studies

We present in Figure 2 a graphical summary of 'Risk of bias' assessments performed by review authors for the five included studies, based on the seven 'Risk of bias' domains. Figure 3 provides a summary of 'Risk of bias' results for each included study. We provided reasons for judgements in the Risk of bias in included studies tables. For clarification, we provided quotes in these tables.


Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.


Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Using the Cochrane criteria, we rated all the studies as having unclear risk of selection bias and rated no studies as having high risk of selection bias. Three of the studies did not report the method of sequence generation (Bashi 2016; Hagglund 2015; Jalali 2018) and two studies used computerised block randomisation (Li 2016; Wagenaar 2019). Allocation concealment was not clearly reported for any of the included studies (Bashi 2016; Hagglund 2015; Jalali 2018; Li 2016; Wagenaar 2019).

Blinding

We assessed blinding as being at unclear risk of performance and detection bias in two studies (Hagglund 2015; Jalali 2018), as no detail was provided on blinding of participants or the person responsible for outcome assessment. The healthcare professionals delivering the intervention were aware of the allocation schedule in two studies (Li 2016; Wagenaar 2019) and the primary researcher who collected the outcome data was not blinded in one study (Bashi 2016); this contributed to our assessment of a high risk of bias. Although an independent committee assessed one outcome in one of the studies (Wagenaar 2019), it is unclear whether the remaining outcomes were assessed by a blinded outcome assessor.

Incomplete outcome data

We assessed one study as having unclear risk of bias for outcome reporting (Bashi 2016). Two studies were at high risk owing to dropouts or incomplete data (Hagglund 2015; Li 2016). We rated studies as having high risk of bias if dropout rates were uneven between groups and if we suspected that the reason for dropout was related to group allocation. We also rated studies as having high risk of bias if investigators did not report how they dealt with the dropouts (e.g. ITT analysis, last observation carried forward). We rated two studies (Jalali 2018; Wagenaar 2019) as low risk, as the final analysis included all participants i.e. an ITT analysis was presented.

Selective reporting

We rated three studies at unclear risk of bias for selective reporting (Bashi 2016; Hagglund 2015; Li 2016), as no published protocol or pre‐registered trial registry entry was available for these studies to compare with the published results. A published protocol was available for two studies (Jalali 2018; Wagenaar 2019) and all prespecified outcomes in the published protocol were reported in the published results. We therefore rated them at low risk of bias.

Other potential sources of bias

We rated other potential sources of bias as low risk for three studies (Hagglund 2015; Li 2016; Wagenaar 2019), as we detected no other biases. We rated one study (Bashi 2016) at high risk, as it was a pilot trial and a full trial is not being done. We judged the remaining study to be at unclear risk (Jalali 2018).

Effects of interventions

See: Summary of findings 1 mHealth education intervention compared to usual care in heart failure

A 'Summary of findings' table for mHealth‐delivered education intervention compared to usual care is presented in the summary of findings Table 1.

Primary outcomes

Heart failure knowledge

Three studies addressed HF knowledge in the short term (up to three months) (Bashi 2016; Hagglund 2015; Wagenaar 2019), revealing that the use of mHealth‐delivered education programmes demonstrate no evidence of a difference in HF knowledge compared to usual care (mean difference (MD) 0.10, 95% confidence interval (CI) −0.20 to 0.40; 411 participants; low‐quality evidence; Analysis 1.1). Heterogeneity was low (I2 = 0%). See summary of findings Table 1.

Heart failure self‐efficacy

One study (Bashi 2016) assessed self‐efficacy, and reported that both study groups reported high levels of self‐efficacy at baseline, but found uncertainty in the evidence between the intervention (MD 0.60, 95% CI −0.57 to 1.77; 1 trial, 29 participants; very low‐quality evidence; Analysis 1.2) and control groups. See summary of findings Table 1.

Heart failure self‐care

Due to the heterogeneity (I2= 86%) observed in the analysis, we decided not to pool the studies for this outcome (Analysis 1.3). It is difficult to explain the reasons for this heterogeneity. It may be due to the use of two different measures to assess this outcome. We therefore decided to narratively describe the outcomes of each study. The certainty of evidence is very low for this outcome.Three studies evaluated HF self‐care. Two studies used the European Heart Failure Behaviour Scale (Hagglund 2015; Wagenaar 2019).

Bashi 2016 used the Self‐Care Heart Failure Index (SCHFI). Bashi 2016 showed an improvement in the SCHFI maintenance subscale in the intervention group (MD 9.9, 95% CI −3.6 to −23.6), whereas the usual‐care group decreased over time (MD −3.5, 95% CI −10.3 to 1.3). No significant between‐group differences were found (P = 0.94).

In Hagglund 2015, the intervention group demonstrated an increase in self‐care compared to the usual‐care group (median 17, IQR 13 to 22 versus median 21, IQR 17 to 25; P < 0.05). The mean score of the usual care‐plus‐HFM group (intervention) also showed higher self‐care scores compared to usual care (73.5 versus 70.8 (difference 2.7, 95% CI 0.6 to 6.2) (Wagenaar 2019).

Adverse events

No studies measured and reported adverse events.

Secondary outcomes

Health‐related quality of life

Four RCTs provided data for this outcome (Hagglund 2015; Jalali 2018; Li 2016; Wagenaar 2019). The measurement tool used in one study (Hagglund 2015) was the Kansas City Cardiomyopathy Questionnaire (KCCQ), with higher scores indicating better health status. A mean difference over time of 5 points on the overall summary scale reflects a clinically‐significant change in heart failure status. The other studies (Jalali 2018; Li 2016; Wagenaar 2019) used the Minnesota Living with HF Questionnaire (MLWHFQ) to assess quality of life. Lower scores in this tool indicate better quality of life. To ensure that both scales run in the same direction, we modified the mean MLWHFQ score. The result shows uncertainty in the evidence for the use of mHealth in HRQoL compared to usual care (MD −0.10, 95% CI −2.35 to 2.15; 4 studies, 942 participants; very low‐quality evidence; Analysis 1.4). Heterogeneity was 61%. See summary of findings Table 1.

Heart failure‐related hospitalisation

Three studies reported on HF‐related hospitalisation (Hagglund 2015; Li 2016; Wagenaar 2019). The use of mHealth delivered‐education may result in little to no difference in HF‐related hospitalisation with 69/444 events in the mHealth group and 89/450 events in the usual care group (odds ratio (OR) 0.74, 95% CI 0.52 to 1.06; 3 studies, 894 participants; low‐quality evidence; Analysis 1.5). Heterogeneity was 0%. See summary of findings Table 1.

Medication adherence

No studies measured and reported medication adherence.

Cost

Only one study (Wagenaar 2019) reported cost: "The mean costs per patient were €4,865 and €5,741 per quality‐adjusted life years for HFM website + usual care and usual care, respectively. The net‐monetary benefit was positive (larger than 0) for HFM versus usual care".

Discussion

Summary of main results

This review included five studies with a total of 971 participants. We were unable to reach any conclusions on the applications of mHealth education interventions in clinical practice for people with heart failure due to the limited number of included studies and the methodological limitations within the studies. Nevertheless, the review has some important implications for future research and for the application of mHealth‐delivered intervention in clinical practice. First, as it would be unethical to deny control‐group participants education of any sort, mHealth‐delivered interventions were compared to traditional or usual care education. While neither study showed evidence of a difference in HF knowledge, this finding might be explained as a ceiling effect, as participants in two of the studies had high baseline scores on the Dutch HF knowledge scale (Bashi 2016; Wagenaar 2019). There was uncertainty in the evidence for mHealth‐delivered interventions for self‐efficacy, self‐care and health‐related quality of life, due to the limited number of studies included in the meta‐analysis and inconsistencies in outcomes, measurement tools and scales used across the study populations. We found little to no difference in HF‐related hospitalisation. In situations where regular face‐to‐face interventions are not possible, as in the case of people living in rural and remote areas, mHealth‐delivered education interventions offer a good alternative for HF patient self‐care.

Secondly, the versatility of the technology means that the mHealth approach to patient education is complex. All of the interventions reported in the reviewed studies included multiple components, related to education, self‐care and self‐monitoring tools, and most required Internet connectivity. While technology offers many advantages, it is of no benefit to patients unless they actually use it. One means of assessing whether an intervention is acceptable to participants is by reviewing treatment adherence, i.e. the extent to which participants actually use the learning system. Treatment adherence was addressed by Hagglund 2015, who reported an adherence rate of 88% (IQR 78% to 96%) of the days the participants were equipped with the system, while Bashi 2016 reported that only 50% of the participants accessed the system and 28% had no record of access at all.

Hagglund 2015 included the contact details of local HF nurses and doctors at the end of the web‐based resources. In Wagenaar 2019, emails were sent to encourage participants to use the web‐based resources. Bashi 2016 also used email to motivate participants, and suggested that future studies should use face‐to‐face interaction for this purpose. Because learning is a human interaction, patient‐educator interaction is important in facilitating behaviour changes and self‐care, and maintaining motivation for self‐monitoring (Abraham 2008; Suhling 2014; Treskes 2018). The intrinsic, limited interaction between human and computer must be considered in clinical application of mHealth interventions.

Patient education is about transferring HF knowledge to patients and helping them to develop self‐care skills. Knowledge is only one of the many factors influencing self‐efficacy (Bandura 2011). Beyond information‐giving, this also involves confidence‐building. The intervention itself must be pedagogically sound and not just a different way of delivering traditional printed materials. To build participants’ self‐efficacy, Bashi 2016 used avatars as role models. The use of an avatar can mimic natural and innate human interaction through facial expressions, body language and speech (An 2013; Anam 2016). Hagglund 2015 explained how telemonitoring and technology could support patient self‐care, while Wagenaar 2019 only had a protocol and abstract available, so there were no discussions of any theoretical underpinnings of their intervention to support learning and self‐care. In clinical practice, nurses and clinicians must realise that patients still need support and encouragement to build confidence if they are to translate the knowledge acquired from various resources into day‐to‐day decision‐making and self‐care activities.

There is a high prevalence of cognitive impairment in older people living with heart failure (Ambrosy 2014). It is crucial that future educational interventions include screening for cognitive impairment and focus on optimising care and outcomes in individuals living with both HF and cognitive impairment (Cameron 2017). Research and validation assessment scales demonstrate the feasibility of undertaking research in HF populations with cognitive impairment. Despite this, the system lacks the information required both to optimise care for this population group and to support caregivers to continue this care using mhealth interventions.

Overall completeness and applicability of evidence

The accuracy of the findings of this review and meta‐analysis is based on the studies which met our inclusion criteria. Future updates of this review will incorporate new data along with the findings of studies which are currently underway but not yet completed, or are only available as a conference abstract or awaiting classification.

Quality of the evidence

Review authors rated the certainty of evidence for all comparisons using the five GRADE considerations (study limitations, consistency of effect, indirectness, imprecision and publication bias (Schünemann 2017). We created a 'Summary of findings' table. Certainty assessment ranged from very low to low.

Limitations in study design or execution (risk of bias)

For the comparison of mHealth‐delivered education intervention versus usual care for heart failure knowledge, self‐efficacy,self‐care, health‐related quality of life and heart failure‐related hospitalisations, we downgraded the certainty of evidence by one level for strong suspicions of performance and detection, attrition and other bias. The strong suspicion of detection bias is associated with the primary researcher collecting the data unblinded. Suspicion of attrition bias was due to uneven dropout rates between groups, with the reason for dropout suspected to be related to group allocation. We suspected other bias, as one study was only a pilot trial and the full trial is not being performed.

Inconsistency of results

We downgraded the certainty of evidence for health‐related quality of life by one level for inconsistency of results (I2 = 61%). We also downgraded the certainty of evidence for self‐care by one level for inconsistency, because the very high unexplained heterogeneity precluded meta‐analysis (I2 = 86%).

Indirectness of evidence

All included trials addressed the main review question (PICO): use of mHealth in the form of web‐based education programmes on smartphones and tablets compared to usual care in men and women with HF, compared to usual care. We therefore did not downgrade any outcome in any comparisons for indirectness of evidence.

Imprecision

We downgraded the certainty of evidence by one level for heart failure knowledge and self‐efficacy, as there were fewer than 500 participants included in the analysis. We also downgraded the certainty of evidence for self‐efficacy and heart failure‐related hospitalisation as the confidence intervals were wide.

Publication bias

For all outcomes, we did not downgrade the certainty of evidence for publication bias, as we did not detect it, although the small number of included studies may have prevented this.

Potential biases in the review process

Our review has adhered to Cochrane methodology, and all review authors and personnel have at all times tried to avoid or minimise any biases in the review process. We undertook an extensive search of databases and additional sources and applied no restrictions on language within the search process. We therefore believe that we have identified and included in this review all potentially relevant studies. We translated possibly relevant and non‐relevant non‐English full‐text study reports into English, to finalise the eligibility process. Furthermore, at least two review authors systematically extracted and managed trial data.

The searches captured one study currently awaiting classification and 17 currently ongoing studies. The currently ongoing studies have not yet reported any findings and only a study protocol or conference abstract was available. We contacted authors of studies available as a conference abstract or study protocol, in order to identify a full peer‐reviewed publication for the study. We received no response from some studies which we classify as ongoing or awaiting classification, despite multiple attempts to contact the authors.

Agreements and disagreements with other studies or reviews

One systematic review has been published on this topic (Cajita 2016). Although this review appears similar, there are important differences in the inclusion criteria from our review, particularly the inclusion of a monitoring system, which typically included a blood‐pressure measuring device, weighing scale, and an electrocardiogram recorder, and also accepting quasi‐experimental trials. Our review focuses solely on mHealth education such as internet‐ and web‐based education programmes for use on smartphones and tablets (including apps) and other mobile devices, SMS messages and social media‐delivered education programmes. Differences in the inclusion criteria of Cajita 2016 on this topic limit the possibility of directly comparing our findings with theirs. However, the outcomes of our review along with Cajita 2016 highlight the uniqueness and importance of our findings. Cajita 2016 includes 10 studies (nine randomised controlled trials and one quasi‐experimental trial) comparing mobile health technology as part of a monitoring system versus usual care, while our systematic review only includes five randomised controlled trials. Nonetheless, the overall findings have been consistent in confirming that the impact of mHealth‐based education interventions on HF knowledge, self‐care and health‐related quality of life was mixed, and highlighting the need for further research. The identification of studies currently underway and those awaiting classification indicate that this is an area of HF research for which further evidence will emerge in the short and longer term.

Strengths and weaknesses of this review

Weaknesses of this review are due to inadequate reporting by some studies, which has precluded classification of risks of bias as either low or high risk, leading to some of the studies being rated at unclear risk of bias.

Study flow diagram.

Figuras y tablas -
Figure 1

Study flow diagram.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figuras y tablas -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figuras y tablas -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Comparison 1: mHealth education intervention vs. usual care, Outcome 1: Heart failure knowledge

Figuras y tablas -
Analysis 1.1

Comparison 1: mHealth education intervention vs. usual care, Outcome 1: Heart failure knowledge

Comparison 1: mHealth education intervention vs. usual care, Outcome 2: Self‐efficacy

Figuras y tablas -
Analysis 1.2

Comparison 1: mHealth education intervention vs. usual care, Outcome 2: Self‐efficacy

Comparison 1: mHealth education intervention vs. usual care, Outcome 3: Heart failure self‐care

Figuras y tablas -
Analysis 1.3

Comparison 1: mHealth education intervention vs. usual care, Outcome 3: Heart failure self‐care

Comparison 1: mHealth education intervention vs. usual care, Outcome 4: Health‐related quality of life

Figuras y tablas -
Analysis 1.4

Comparison 1: mHealth education intervention vs. usual care, Outcome 4: Health‐related quality of life

Comparison 1: mHealth education intervention vs. usual care, Outcome 5: Heart failure‐related hospitalisations

Figuras y tablas -
Analysis 1.5

Comparison 1: mHealth education intervention vs. usual care, Outcome 5: Heart failure‐related hospitalisations

Summary of findings 1. mHealth education intervention compared to usual care in heart failure

mHealth education intervention compared to usual care in heart failure

Patient or population: Adults with heart failure
Setting: Hospital and community
Intervention: mHealth education intervention
Comparison: Usual care

Outcomes

Tools

Mean length of follow‐up (months)

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with mHealth education intervention

Heart failure knowledge Analysis 1.1

Dutch HF Knowledge Questionnaire

56 (39.6)

The mean heart failure knowledge was 12.7

MD 0.10 higher
(−0.20 lower to 0.40 higher)

411
(3 RCTs)

⊕⊕⊝⊝
LOWa,d

Higher score equals a better outcome

Heart failure self‐efficacy Analysis 1.2

Self‐efficacy for Managing Chronic Disease Scale

28

The mean heart failure self‐efficacy was 7.4

MD 0.60 (−0.57 lower to 1.77 higher)

29 (1 RCT)

⊕⊝⊝⊝
VERY LOWc,d,e

Higher score equals a better outcome

Heart failure self‐care Analysis 1.3

European Heart Failure Behaviour Scale (EHFBS)

Self‐Care Heart Failure Index (SCHFI)

56 (39.6)

One study showed an improvement in the SCHFI maintenance subscale in the intervention group (MD 9.9, 95% CI −3.6 to −23.6), whereas the usual‐care group decreased over time (MD −3.5, 95% CI −10.3 to 1.3).

Two studies reported improvements on the EHFBS for the intervention groups.

411
(3 RCTs)

⊕⊝⊝⊝
VERY LOWa,d,f

Due to the high heterogeneity observed in the analysis, we decided not to pool the studies for this outcome

Adverse events

No studies reported this outcome

Health‐related quality of life Analysis 1.4

Kansas City Cardiomyopathy Questionnaire

Minnesota Living with HF Questionnaire

84

The mean health‐related quality of life was 51.7

MD −0.10 lower
(−2.35 lower to 2.15 higher)

942
(4 RCTs)

⊕⊝⊝⊝
VERY LOWa,b

Higher score equals a better outcome

Heart failure‐related hospitalisations Analysis 1.5

84

Study population

OR 0.74
(0.52 to 1.06)

894
(3 RCTs)

⊕⊕⊝⊝
LOW a,e

275 per 1000

219 per 1000
(87 to 454)

*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: Confidence interval; MD: mean difference; OR: Odds ratio; RCT: randomised controlled trial.

GRADE Working Group grades of evidence
High certainty: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

aDowngraded by one level for 'limitations in study design and execution', as the studies were rated at high risk of bias in multiple domains (performance and detection, attrition bias or other bias).
bDowngraded by one level for 'inconsistency' due to substantial heterogeneity (I2 = 61%).
cDowngraded by one level for 'limitations in study design and execution', as the study was rated at high risk of bias in multiple domains (performance, detection and other bias).
dDowngraded by one level for 'imprecision' as there were fewer than 500 participants in the analysis.

eDowngraded by one level for 'imprecision' as the confidence intervals were wide.

fDowngraded by one level for 'inconsistency' due to high heterogeneity that precluded meta‐analysis (I2 = 86%).

Figuras y tablas -
Summary of findings 1. mHealth education intervention compared to usual care in heart failure
Comparison 1. mHealth education intervention vs. usual care

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Heart failure knowledge Show forest plot

3

411

Mean Difference (IV, Fixed, 95% CI)

0.10 [‐0.20, 0.40]

1.1.1 Dutch Heart Failure Knowledge Scale (higher score = complete HF knowledge)

3

411

Mean Difference (IV, Fixed, 95% CI)

0.10 [‐0.20, 0.40]

1.2 Self‐efficacy Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

1.2.1 Self‐Efficacy for Managing Chronic Disease Scale (high score= more confident in self‐care)

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

1.3 Heart failure self‐care Show forest plot

3

Std. Mean Difference (IV, Fixed, 95% CI)

Totals not selected

1.3.1 European Heart Failure Self‐care Behaviour Scale (high score= better self‐care)

2

Std. Mean Difference (IV, Fixed, 95% CI)

Totals not selected

1.3.2 Self‐care Heart Failure Index (high score= adequate self‐care)

1

Std. Mean Difference (IV, Fixed, 95% CI)

Totals not selected

1.4 Health‐related quality of life Show forest plot

4

942

Mean Difference (IV, Fixed, 95% CI)

‐0.10 [‐2.35, 2.15]

1.4.1 Kansas City Cardiomyopathy Questionnaire (high score= better health status)

1

82

Mean Difference (IV, Fixed, 95% CI)

12.10 [3.08, 21.12]

1.4.2 Minnesota Living with Heart Failure Questionnaire (high score= better quality of life)

3

860

Mean Difference (IV, Fixed, 95% CI)

‐0.91 [‐3.23, 1.42]

1.5 Heart failure‐related hospitalisations Show forest plot

3

894

Odds Ratio (M‐H, Fixed, 95% CI)

0.74 [0.52, 1.06]

Figuras y tablas -
Comparison 1. mHealth education intervention vs. usual care