Author + information
- Received March 30, 2017
- Revision received June 19, 2017
- Accepted June 19, 2017
- Published online September 13, 2017.
- Sanjeev P. Bhavnani, MDa,
- Srikanth Sola, MDb,
- David Adams, RCS, RDCSc,
- Ashwin Venkateshvaran, PhDb,
- P.K. Dash, MDb,
- Partho P. Sengupta, MD, DMd,∗ (, )
- ASEF-VALUES Investigators
- aScripps Clinic and Research Foundation, San Diego, California
- bSri Sathya Sai Institute of Higher Medical Sciences, Whitefield, Bangalore, India
- cDuke University School of Medicine, Durham, North Carolina
- dWest Virginia University Heart & Vascular Institute at West Virginia University School of Medicine, Morgantown, West Virginia
- ↵∗Address for correspondence:
Dr. Partho P. Sengupta, West Virginia University Heart & Vascular Institute, West Virginia University School of Medicine, 1 Medical Center Drive, Morgantown, West Virginia 26506.
Objectives This study sought to determine whether mobile health (mHealth) device assessments used as clinical decision support tools at the point-of-care can reduce the time to treatment and improve long-term outcomes among patients with rheumatic and structural heart diseases (SHD).
Background Newly developed smartphone-connected mHealth devices represent promising methods to diagnose common diseases in resource-limited areas; however, the impact of technology-based care on long-term outcomes has not been rigorously evaluated.
Methods A total of 253 patients with SHD were randomized to an initial diagnostic assessment with wireless devices in mHealth clinics (n = 139) or to standard-care (n = 114) in India. mHealth clinics were equipped with point-of-care devices including pocket-echocardiography, smartphone-connected-electrocardiogram blood pressure and oxygen measurements, activity monitoring, and portable brain natriuretic peptide laboratory testing. All individuals underwent comprehensive transthoracic echocardiography to assess the severity of SHD. The primary endpoint was the time to referral for therapy with percutaneous valvuloplasty or surgical valve replacement. Secondary endpoints included the probability of a cardiovascular hospitalization and/or death over 1-year.
Results An initial mHealth assessment was associated with a shorter time to referral for valvuloplasty and/or valve replacement (83 ± 79 days vs. 180 ± 101 days, p <0.001) and was associated with an increased probability for valvuloplasty/valve replacement compared to standard-care (34% vs. 32%; adjusted hazard ratio: 1.54; 95% CI: 0.96 to 2.47, p = 0.07). Patients randomized to mHealth were associated with a lower risk of a hospitalization and/or death on follow-up (15% vs. 28%, adjusted hazard ratio: 0.41; 95% CI: 0.21 to 0.83; p = 0.013).
Conclusions An initial mHealth diagnostic strategy was associated with a shorter time to definitive therapy among patients with SHD in a resource-limited area and was associated with improved outcomes. (A Randomized Trial of Pocket-Echocardiography Integrated Mobile Health Device Assessments in Modern Structural Heart Disease Clinics; NCT02881398)
The transformative potential for cellular technologies, expanding internet connectivity, and the development of innovative mobile health (mHealth) devices to improve health care delivery in resource-limited areas is promising (1,2). As these areas begin to leverage new digital infrastructures for health care, several key factors have emerged. These include: 1) to identify the heuristic factors and evaluative methods that lead to appropriate use of new technologies; 2) to determine the integration of device-based findings into existing informational systems and health records; 3) to demonstrate the patterns of effective use at the point-of-care; and 4) to identify those patterns that lead to earlier diagnostic and treatment decisions (3,4). In the aggregate, an emphasis on the deterministic approaches of mHealth must include pragmatic device use and outcomes-based assessments as new technology-based health care initiatives are organized (5,6).
Recent shifts in the global burden of cardiovascular diseases have led to an increasing prevalence in resource-limited areas with more than 25 million deaths in these regions predicted by 2030 (7). This problem is further compounded with resource-limited areas receiving a disproportionately low allocation of global resources ranging from the availability of appropriate diagnostic tests to sufficiently trained health care professionals (8). Portability, lower cost, and simple-to-use form factors are among the design features of mHealth that may be well suited to bridge these inequalities, and to mobilize care from hospital- and clinic-based encounters to the practitioner at the point-of-care and in remote locations (9). Although attractive from a technological perspective, the impact of mHealth used as a practitioner-based clinical-decision support tool on long-term outcomes has not been rigorously evaluated (10).
Therefore, the objective of the present study was to compare the outcomes of mHealth with smartphone-connected devices and pocket-echocardiography on medical decision making among patients with rheumatic and structural heart disease (SHD) in a health care system of a resource-limited area.
The study was performed under the ASEF-VALUES (American Society of Echocardiography Foundation–Valvular Assessment Leading to Unexplored Echocardiographic Stratagems) program—a philanthropic and educational initiative to explore health care solutions for patients with SHD using new technologies. Within the program was a nested, single-site, randomized trial conducted at the Sri Satya Sai Institute of Higher Medical Sciences (SSSIHMS), a charitable, free-of-charge, tertiary-care, and teaching institution in Bangalore, India that exclusively provides care to the underserved, sees more than 20,000 SHD patients per year, and performs 1,100 percutaneous valvuloplasties and 1,000 valve replacements on an annual basis.
The primary study sponsors were the ASEF and SSSIHMS. General Electric Healthcare (Bangalore, India) provided local instruments and logistical support. Additional device support was provided by CoreSound Imaging (Raleigh-Durham, North Carolina) and iHealth (San Francisco, California). Five cardiologists and 12 sonographers from 12 academic medical centers across the United States, 15 cardiologists and cardiothoracic surgeons from SSSIHMS, and 30 cardiologists from across India participated in the study.
The study participants were outpatients with a new or an established diagnosis of SHD. The definition of SHD included valvular disease, left/right ventricular failure and congenital heart defects, and included adult, pediatric, and pregnant patients. We decided a priori to include SHD patients with a prior valvuloplasty or valve replacement. Exclusions included neonatal patients and those with an unstable hemodynamic status. All subjects provided written informed consent in their native language.
Trial organization, randomization, and masking
Consecutive subjects were randomly assigned to an initial evaluation with mHealth or to standard care. Study subjects were evaluated in either 1 of 10 (5 mHealth, or 5 standard care) clinical sites all located at SSSIHMS. Each individual clinics that were used for a patient encounter after randomization. We decided to create mHealth and standard-care sites within 1 hospital to minimize variability with the initial clinical encounter (mHealth or standard care) after randomization, and separated these clinics to reduce any potential bias introduced by using mHealth devices. All mHealth clinics were equipped with the same devices and used the institutions electronic medical record to standardize workflow the data generated during the trial encounter. The standard-care clinics were designed with pragmatic intention and to mimic usual care practice patterns from a group of physicians across India. To minimize confounding resulting from the same physician participating in the clinical assessments as well as conducting procedures, physicians performing mHealth or standard care assessments did not participate in procedural interventions assessments and vice versa.
Following the clinical encounter, with mHealth or standard-care, all study participants underwent a comprehensive transthoracic echocardiogram (TTE) for the severity of SHD. The TTE was performed on the same day and was interpreted either on the same day or within 24 hours in both randomized arms (Online Figure 1). The decision to complete TTE on the same day as randomization was to: 1) eliminate the time to diagnostic testing with TTE as a variable that is commonly observed in resource-limited areas; 2) minimize the time to TTE as a factor resulting in treatment delays; 3) determine the yield of diagnostic information provided by mHealth or standard care on the referral rate for treatment (valvuloplasty or valve replacement) and; 4) assess the impact of mHealth and standard care on medical decision making at the time of enrollment and on follow-up. A randomization schedule was created (based on hospital outpatient estimates) that on a given day outpatient numbers would not exceed a maximum 400 patients, and was formulated using a simple randomization scheme (random number generator SPSS version 23.0 [IBM Corporation, Armonk, New York]).
To ensure concealment of allocation that would otherwise introduce selection bias despite randomization, randomization was performed by study staff not involved in the study and was concealed until the primary and secondary endpoints were analyzed. Given the diagnostic and treatment procedures in the present study, it was important to differentiate those cardiologists who performed the initial diagnostic assessment (mHealth or standard care) from those cardiologists and surgeons who performed valvuloplasty or valve replacement. The purpose of the initial assessment was to allow participating physicians to make clinical decisions based on mHealth or standard-care findings. The treatment plan and referral for intervention was generated by these cardiologists conducting the initial assessment and was formulated through the aggregate of diagnostic information available at the time of enrollment. For mHealth it was history and the findings on activity monitoring, pocket ultrasound, smartphone electrocardiograph (ECG) and point-of-care brain natriuretic peptide (BNP) (the latter if applicable), and for standard care the usual physical examination findings and diagnostic tests on follow up. Subsequently, operating interventional cardiologists and surgeons (different than those performing the initial assessment) were blinded to a study subject’s group allocation; however, they reviewed the findings on TTE for diagnostic accuracy at the time of planned percutaneous intervention or surgical procedures.
Initial mHealth assessments
Each mHealth clinic was equipped with wireless mHealth devices that were selected to assess functional and structural abnormalities at the point-of-care (Online Figure 2) including: 1) pocket echocardiography (VScan, General Electric, Whitefield, India); 2) vital signs with smartphone-connected oximetry and blood pressure monitors (iHealth, San Francisco, California); 3) 6-min walk test with a trial-axial activity monitor (Ozeri, San Diego, California); 4) cardiac rhythm abnormalities were classified by a smartphone-connected-iECG (AliveCor, San Francisco, California); and 5) point-of-care testing with fingerstick B-type natriuretic peptide (Alere Triage, Gurgoan, India).
All participating physicians and study staff received training on the use of mHealth devices before initiation of the program. Point-of-care echocardiographic examinations were performed using the VScan, a pocket-sized device. Scans were performed by local physicians to execute a protocol consisting of 11 standard views including color-flow Doppler images of all valves (11), and were trained by ASE sonographers similar to the training methodology used in the ASEF-VISION (Value of Interactive Scanning for Improving Outcomes of New Learners) study (6). By design, ASE sonographers did not participate in pocket-echocardiographic image acquisition and local physicians were independent when using pocket-echocardiography in the mHealth assessment and for clinical decisions. The VScan is a handheld-sized device (135 × 73 × 28 mm) that weighs 400 g and has an 8.9-cm (diagonal) display with a resolution of 240 × 320 pixels. The device uses a phased-array transducer (1.7 MHz to 3.8 MHz) and displays gray scale images with a sector width of 75° and color Doppler images with a fixed sector width of 30°. Current-generation devices do not have the capabilities of spectral Doppler or M-mode imaging. Qualitative assessments (mild, moderate, or severe) of chamber size, volumetric estimations and severity of valvular stenosis, regurgitation, and left and right ventricular dysfunction were interpreted at the time of the examination. Left ventricular ejection fraction was stratified into normal ≥55% or low if it was <55% by visual estimation and the presence of valvular abnormalities (regurgitant or stenotic) and severity (mild, moderate, or severe), or mitral stenosis (progressive, severe, very severe) were recorded according to ASE recommended definitions. The severity of regurgitant lesions was based on 2-dimensional findings (atrial or ventricular enlargement, hyperdynamic left ventricle) and qualitative color Doppler findings (width of vena contracta and jet area), whereas the severity of stenotic lesions was based on 2-dimensional findings of valve opening and leaflet mobility, thickness, and calcification.
An assessment of functional capacity was quantified by a tri-axial activity monitor that calculates the number of steps, duration of activity (minutes), distance (meters) and gait speed (miles per hour) to quantify the level of activity on a 6-min walk test (6MWT). The 6MWT was conducted by nurses trained in device use, data acquisition, and was performed along a fixed distance of 50 m marked at 10-m intervals (12). Internal quality control of activity measurements to fixed distances were performed daily with study staff as volunteers. Error >10% required removal and replacement of the device (none required in the study). To quantify the New York Heart Association (NYHA) functional class, the following calculations were applied: gait speed (mph) was automatically calculated from the distance walked in meters over the duration of activity achieved. NYHA functional class was stratified by gait speed into the following categories: NYHA functional class I ≥2.2 mph, NYHA functional class II = 1.5 to 2.2 mph, NYHA functional class III ≤ 1.5 mph, and NYHA functional class IV symptoms at rest (13).
Smartphone-connected oxygen and blood pressure monitors provided assessments of oxygen saturation, and blood pressure at rest and with exertion. Oxygen saturation was measured by photoplethysmography and blood pressure with oscillometric measurements and automated inflation. The resulting measurements are displayed on a tablet application.
A smartphone-connected iPhone-ECG (iECG) was used to determine the heart rate and cardiac rhythm (2). In general, the iECG produces a single-lead ECG when held with the left and right fingers (lead I) or placed directly on the chest wall for a precordial lead. The ECG recording was transmitted to the tablet using frequency modulation of the electrical signal to ultrasound. Capture of this sound signal on the tablet’s microphone produces a real-time cardiac rhythm on the tablet’s display. Diagnostic findings were classified as an atrial arrhythmia (atrial fibrillation or atrial flutter, supraventricular tachycardia, or ventricular arrhythmias) or bradyarrhythmia (second- or third-degree atrioventricular block, or sinus node dysfunction).
Natriuretic peptide levels
Fingerstick B-type natriuretic peptide (BNP) was performed among select patients with findings of severe mitral stenosis, nonequivocal symptoms on functional assessments. The test is a rapid, point-of-care fluorescence immunoassay used to measure BNP in K2 Ethylenediaminetetraacetic acid anticoagulated whole blood droplets. Results are displayed within 15 minutes on a miniaturized, portable, and battery powered device (14).
Initial standard-care assessments
The standard-care clinics used available resources including a 12-lead ECG, radiographs, and laboratory testing as required. Participating cardiologists who have experience treating patients with SHD conducted the initial clinical examination, interpreted all point-of-care assessments, made preliminary medical and surgical treatment decisions, and determined the frequency of follow-up visitations.
Transthoracic echocardiography and cloud-based and paperless reporting
After enrollment, all subjects underwent a comprehensive transthoracic echocardiographic examination (General Electric Vivid-E9, Philips Healthcare-ie33) performed by onsite ASE sonographers according to ASE guidelines (15,16). Local and ASE cardiologists interpreted all echocardiographic studies by using a paperless and cloud-based system as previously reported (5) (Online Appendix). mHealth devices, with the exception of pocket-echocardiography and point-of-care BNP, were connected via Bluetooth to a tablet computer (Samsung Galaxy Ta, Samsung Electronics, Seoul, South Korea). The mHealth clinics were designed to operate without WiFi and used only local power supply. The institutional electronic medical record (Enterprise Manage, Computer Science Corporation, Bangalore, India), a standards-based, modular application running on Oracle11G database, was modified with templates to input all mHealth and standard-care findings to facilitate paperless data collection.
The primary outcome was the time to treatment with valvuloplasty or valve replacement over 12-months after the initial mHealth or standard-care assessment. Secondary outcomes included the occurrence of a cardiovascular hospitalization and/or death on follow-up. The primary investigators at SSSIHMS adjudicated all clinical endpoints and determined the necessity for percutaneous or surgical treatment. Outcomes were obtained at the time of a procedure or were determined by telephone, text message, or by community health worker visitation to the home.
Our hypothesis was that an initial assessment with mHealth would result in a shorter time to treatment. We performed a post hoc power calculation derived from the number of participants enrolled and time to primary endpoint. Based on the observed means and SDs across trial arms, our sample size had a power of 80% even if the type I error rate was as small as 5.22 × 10-14. The full relationship between type I and type II error rates is shown in Online Figure 3. All analyses were intention-to-treat based on randomized treatment allocation. Descriptive analyses of continuous variables are described as means and SDs and categorical variables as frequencies and percentages, and were compared using the Student t test or the Mann-Whitney test, or the chi square or Fisher exact tests where appropriate, respectively. Statistical comparisons of the randomized groups were based on a time-to-first-event (primary outcome of the rate of valvuloplasty/valve replacement and secondary outcome of a cardiovascular hospitalization/death) that was reported with mean differences and the Cox proportional hazard model. All outcomes were adjusted for the presence of moderate or severe mitral and/or aortic valve disease, a history of SHD, and a prior valvuloplasty/valve replacement. Relative risks were expressed as means and adjusted hazard ratios (AHRs). Additionally, the outcome of a hospitalization and/or death was also computed using valvuloplasty/valve replacement as a time-dependent covariate. Cumulative event rates were calculated for each randomized group as a function of time from randomization with the use of the Kaplan-Meier method. Pre-specified sub-group analyses were performed according to relevant demographic variables: age (younger than or older than 40 years), sex, presence of symptoms, history of SHD using the Cox model, and the interaction of individual mHealth devices on outcomes using chi-square test associations. All analyses reported 95% confidence intervals (CIs) where appropriate. The protocol received ethics and Institutional Review Board committee approval and was registered with ISRCTN #61479659 and NCT02881398.
All 253 study subjects were recruited and randomized over 3 days between August 10, 2014, and August 12, 2014. Follow-up was completed in October 2015 and data analysis performed in January 2016 (Figure 1). A comprehensive mHealth assessment was performed in 139 subjects and a standard-care assessment in 114. Data was available from all devices with the exception of 5 activity assessments that were deleted on the activity monitoring devices.
Characteristics at baseline
The mean age of the study population was 39 ± 14 years and 42% (107 of 253) of participants were women. The mean follow-up time was 337 ± 116 days. The study population had a substantial burden of disease with an established diagnosis of SHD with or without prior valvuloplasty/valve replacement observed in 46% (118 of 253). Nearly one-half of the study population were symptomatic with 46% (116 of 253) exhibiting NYHA functional class II to III symptoms. Pertinent clinical characteristics were well balanced with no statistically significant differences between randomized arms (Table 1). Transthoracic echocardiographic findings are listed in Online Table 1. The presence of SHD with mitral stenosis, mitral regurgitation, aortic stenosis, or aortic regurgitation were frequently observed in 57%, 42%, 23%, and 32% of transthoracic echocardiographic studies, respectively. The prevalence of severe mitral or aortic valve disease was similar between randomized groups.
Pertinent findings on mHealth devices are listed in Table 2. Activity parameters were significantly lower among study subjects with an activity duration of less than 6 min (n = 40) compared to subjects able to complete 6 min of activity (n = 94) (Online Figure 4). The iECG was interpretable with high diagnostic certainty in 98% of rhythm strips. Ninety-six percent of pocket-echocardiographic studies were graded to have good/excellent image quality and showed adequate diagnostic correlation to transthoracic echocardiography for moderate/severe valvular stenosis/regurgitation (areas under the curve of 0.74 and 0.79, respectively) (Online Figure 5). An example of an initial mHealth assessment is shown in Figure 2.
Primary end points
Overall, 34% (85 of 253) of the study population underwent treatment with valvuloplasty or valve replacement on follow-up. The total number of procedures was 119 comprised of 26 valvuloplasties, 27 aortic and 53 mitral valve replacements and 13 dual aortic and mitral valve replacements with 8 patients receiving both valvuloplasty and valve replacement. The occurrence of the individual types of procedures within the composite primary outcome was similar between randomized groups (Figure 3).
At 12 months, a similar treatment rate was observed between the mHealth and standard-care groups (34% [95% CI: 26% to 42%] vs. 32% [95% CI: 25% to 42%], mean difference 0.5% [95% CI for difference between study arms of −11% to +12%, p = 0.52]). Compared to standard care, a shorter duration from enrollment to primary outcome was observed with mHealth (83 ± 79 days vs. 180 ± 101 days, mean difference −96% [95% CI: −136 days to −56 days], p <0.001) with twice as many participants randomized to mHealth undergoing treatment at 90 days (20% vs. 10%). On follow-up, 51 subjects (20%) experienced a cardiovascular hospitalization and 3 died (1%). The occurrence of a hospitalization and/or death was lower in the mHealth than in the standard-care arm (15% [95% CI: 9% to 21%] vs. 28% [95% CI: 2% to 36%], mean difference −13% [95% CI of −23% to −3%, p = 0.012]).
Study subjects randomized to mHealth were more likely to undergo treatment with valvuloplasty and/or valve replacement (AHR 1.54 [95% CI: 0.96 to 2·47, p = 0.07]) (Figure 4A) compared to standard care and was associated with a lower hazard of hospitalization and/or death on follow-up (AHR 0.41 [95% CI: 0.21 to 0.83, p = 0.013]) (Figure 4B). Modeling the occurrence of the primary outcome as a time-dependent covariate, the probability of a cardiovascular hospitalization and/or death was lower in the mHealth arm as compared to standard-care (AHR 0.31 [95% CI: 0.15 to 0.63], p = 0.001).
Pertinent sub-group analyses of relevant demographic cohorts can be found in Figure 5. Within the mHealth study arm, an incremental correlation to the primary outcome was observed with a X2 association of 2.7 (p = 0.23), 11 (p <0.001), and 32 (p = 0.001), resulting from the presence of an abnormal finding on the iECG, <6 min of walking observed on activity monitoring, and severe valvular disease shown on pocket echocardiography, respectively.
The principle findings from ASE-VALUES are the following: 1) compared to standard care, an initial diagnostic strategy with mHealth was associated with a shorter time to referral for valvular interventions and a lower probability of a hospitalization or death among a community cohort of SHD patients; 2) for a similar severity of SHD, an incremental effect on outcomes was observed with mHealth; and 3) point-of-care mHealth devices used to assess the severity of symptoms, structural, and functional abnormalities can be used at the point of care as clinical decision support tools.
The ASE digital global health programs have aimed to maximize the yield of pocket echocardiography. The ASEF-REWARD (Remote Echocardiography with Web Based Assessments for Referrals at a Distance) study (5) was the first investigation to determine the feasibility of cloud-computing– and internet-based echocardiographic image transfer. In a remote region of India, 1,000 individuals with symptoms of SHD were imaged with pocket echocardiography within 48 h. The digitized studies were uploaded within 4 min and interpreted by a global consortium of 75 cardiologists within 12 h. Results of complex SHD were delivered back to local physicians for clinical decisions providing a novel mechanism for accessing expert consultation at the point of care. To standardize image reporting, the follow-up ASEF-VISION (Value of Interactive Scanning for Improving Outcomes of New Learners) study (6) was performed in which sonographers in the United States trained clinicians in India using a web-based imaging and educational platform resulting in improvements in image acquisition and interpretation. ASEF-VALUES aimed to advance these findings and to determine the impact pocket echocardiography and mHealth on outcomes in SHD.
The design of the present investigation required referral of patients with SHD; therefore, we observed a high incidence of complex valve disease. Greater than 50% of our participants exhibited severe valvular disease with 1 of 4 patients with combined mitral and aortic valve pathologies. For example, 38% of the population exhibited severe rheumatic-mitral stenosis (mean mitral valve area and transmitral pressure gradient of 1.1 ± 0.4 cm2 and 9.2 ± 4.7 mm Hg, respectively) and were identified as a high-risk cohort at risk of hospitalization and death. These findings are representative of a cross-sectional sample of patients commonly seen in endemic regions where patients often present late in the disease process and with advanced symptoms (5,17,18). Within India, a 6- to 24-month waiting period for cardiac procedures and cardiothoracic surgery is commonly observed in nonprofit and government-run hospitals (19). In these health systems, determination of severity of disease and prioritization of sicker patients is of paramount importance. Similar to that observed in the standard-care arm of the present study, longer waiting periods for treatment or interventions commonly result in increased morbidity and mortality. Various socioeconomic factors also contribute to long treatment delays including patient refusal for surgery, poor health literacy, and a lack of social support as additive factors and has been shown in 50% of patients with SHD in rural India (20).
Compared to standard care, an initial assessment with mHealth was associated with a shorter time to treatment of 3 months (89 days vs. 180 days) with twice as many participants in the mHealth arm undergoing treatment at 90 days (20% vs. 10%). Overall, our treatment rate of 34% at 12 months is dramatically shorter than the national average (19). Upon multivariate and time-dependent analysis, an initial mHealth assessment was associated with improved outcomes within both the overall mHealth cohort and among those mHealth subjects who underwent valvuloplasty or valve replacement on follow-up. The latter suggests that earlier treatment is associated with improved outcomes. We postulate the following reasons for these observations. The first is due to an improved characterization of SHD with mHealth that provided a comprehensive assessment of the severity of valvular abnormalities at the time of enrollment. Such point-of-care diagnostics likely facilitated timely clinical decisions and referral for treatment among those patients with severe disease (18,21). The yield of pocket echocardiography showed moderate to high diagnostic accuracy compared to comprehensive echocardiography for qualitative determination of SHD lesions, and is consistent with the diagnostic accuracy observed in the ASEF-REWARD study (5). The second is due to a quantified approached of symptoms and functional limitations that facilitated medical decision making. Within the mHealth arm, activity monitoring reclassified 37% of participants to a different NYHA functional class with 30% reclassified to a higher and 6% to a lower NYHA functional class, respectively. Third, a shorter notification rate—the rapid availability of diagnostic information from the point of care and input into electronic medical records—may have improved care coordination, facilitated medical decisions such as early initiation of diuretics for heart failure or rate control and anticoagulation for atrial fibrillation, and may have resulted in shorter follow-up for higher-risk patients (22,23). Taken together, point-of-care diagnostics, functional assessments, and the availability of digitally acquired patient data likely influenced clinical decisions and referral of those patients that were more likely to benefit with earlier interventions (18).
Because of sheer numbers of individuals at risk, mHealth utilization in resource-limited areas must be designed in pragmatic and cost-effective approaches. Recent progress with digital health in these regions has included using text messaging to improving medication compliance among patients with HIV and to encourage lifestyle changes among individuals at risk for diabetes (24,25), genome sequencing to diagnose tuberculosis at the point of care (26), and electronic intensive care units to streamline care for an acute coronary syndrome (27). In contrast to most chronic diseases, the diagnosis of SHD requires specialized diagnostics including echocardiography and trained users to accurately perform and interpret the severity of structural abnormalities. Training health care workers to perform portable echocardiographic studies and to shift imaging to community-based screening programs are potentially scalable methods to address the shortage of technically proficient health care personnel (28,29). In this context, Engelman et al. (30), Mirabel et al. (31), and Ploutz et al. (32) have provided seminal results and address “task-shifting” with trained nurses proficient in the acquisition of echocardiographic images to diagnose SHD. When compared to transthoracic echocardiography, nurse-based screening with portable echocardiographic devices showed a diagnostic accuracy of 85% for the detection of SHD in more than 4,000 at-risk children. Such training may emerge as a practical and potentially transformative method to increase the yield of portable imaging in endemic areas.
Several limitations must be noted when interpreting our findings. We observed unequal sample sizes for 2 main reasons: 1) using an a priori randomization schedule used for daily enrollment versus over the enrollment period; and 2) a simple (or unrestricted) rather than a restricted (i.e., permuted block) randomization method. The simple randomization method allows for random variation in sample sizes important for pragmatic trials and to minimize bias particularly in non–double-blinded studies. In such designs, equal randomization is not necessarily required (33,34). Despite this finding, baseline demographics were well balanced and the overall treatment rates at 12 months were equal between randomized groups, suggesting that bias was minimized when analyzing the effectiveness and safety of mHealth. A multiple-arm trial and blinded assessment of pocket echocardiography compared to TTE was not performed as this would be ethically unacceptable because Doppler measurements cannot be performed on pocket devices for accurate hemodynamic assessment required for interventional/surgical referral of cases. Although our study was small in size, its randomized comparison of initial testing strategies, use of available technologies, broad representation of a real-world community cohort, and the use of hard clinical events as outcome measures enhances internal and external validity, and may represent potentially reproducible methods for mHealth use in other underserved areas.
Compared to standard care, an initial testing strategy with mHealth was associated with a shorter referral time for treatment among symptomatic patients with advanced SHD, and was associated with improved health outcomes in an endemic area with a high burden of disease. These data have important implications for the use of pocket echocardiography and smartphone-connected mHealth devices at the point of care as clinical decision support tools in the health care system of resource-limited areas.
COMPETENCY IN MEDICAL KNOWLEDGE: Newly developed mHealth and smartphone-connected technologies have emerged as potentially transformative innovations to improve health care delivery and for public and global health benefit.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: Despite significant progress in prevention and screening, rheumatic and SHD remain leading causes of morbidity and mortality in resource-limited areas. In such areas, the intersection of expanding digital infrastructures with near ubiquitous cellular phone use and internet connectivity, combined with new devices such as portable and wireless mHealth devices, handheld-ultrasound, and lab-on-a-chip technologies may bridge common health care disparities by providing the necessary diagnostic information for health care practitioners to formulate clinical decisions at the point of care.
TRANSLATIONAL OUTLOOK 1: This is the first study to compare the impact mHealth device assessments such as pocket echocardiography and smartphone electrocardiography to the standard of care on treatment rates and outcomes among patients with advanced SHD in the health system of a developing nation. In doing so, this study identifies the capacity of mHealth to identify high-risk patients and those patients who may derive benefit with earlier medical therapies and surgical interventions.
TRANSLATIONAL OUTLOOK 2: The present investigation provides an understanding for how technology-enabled care in such regions can be delivered. Integrating mHealth findings into existing health information technology systems and using diagnostic information provided by mHealth to improve risk stratification are among the important heuristic factors that can improve patient outcomes. New integration methods should remain a focus of future studies evaluating the outcomes of technology-enabled care in resource-limited areas.
The authors thank the American Society of Echocardiography Foundation for program organization, strategic planning, and funding; General Electric Healthcare, iHealth, and CoreSound Imaging for their generous contributions of devices and information technology resources that were integral to the execution of this investigation; and Hemant Kulkarni, MD, for statistical support. They also thank the participants of this study and for their willingness to contribute towards these efforts.
For a complete list of investigators, a description of training procedures for mHealth devices, supplemental figures and tables, please see the online version of this paper.
Supported by The American Society of Echocardiography Foundation. Dr. Bhavnani has received an educational and research grant from the Qualcomm Foundation to Scripps Health; is a consultant to Proteus Digital; and is an advisory board member to iVEDIX, WellSeek, and Misceo. Dr. Sola has received a research grant from General Electric Healthcare (outside of this investigation). Dr. Sengupta has received research grants from Heart Test Labs and Echo Sense Ltd; and is a consultant to TeleHealth Robotics, Intel, Hitachi Aloka, and Heart Test Labs. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
A complete list of investigators in the ASEF-VALUES study is provided in the Online Appendix.
Maurice Enriquez-Sarano, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- American Society of Echocardiography Foundation
- brain natriuretic peptide
- mobile health
- structural heart disease
- Sri Sathya Sai Institute of Higher Medical Sciences
- transthoracic echocardiogram
- Received March 30, 2017.
- Revision received June 19, 2017.
- Accepted June 19, 2017.
- 2017 American College of Cardiology Foundation
- Agarwal S.,
- LeFevre A.E.,
- Lee J.,
- et al.,
- WHO mHealth Technical Evidence Review Group
- Piette J.D.,
- List J.,
- Rana G.K.,
- et al.
- Pulignano G.,
- Del Sindaco D.,
- Di Lenarda A.,
- et al.
- Maisel A.,
- Barnard D.,
- Jaski B.,
- et al.
- Nishimura R.A.,
- Otto C.M.,
- Bonow R.O.,
- et al.,
- American College of Cardiology/American Heart Association Task Force on Practice Guidelines
- Zühlke L.,
- Karthikeyan G.,
- Engel M.E.,
- et al.
- Evangelista A.,
- Galuppo V.,
- Méndez J.,
- et al.
- Agarwal A.
- Bansal M.,
- Sengupta P.P.
- Saxena A.
- Engelman D.,
- Kado J.H.,
- Reményi B.,
- et al.
- Mirabel M.,
- Bacquelin R.,
- Tafflet M.,
- et al.
- Ploutz M.,
- Lu J.C.,
- Scheel J.,
- et al.
- Bacchetti P.,
- Deeks S.G.,
- McCune J.M.