The initial pilot study, TeleClinical Care (TCC)-Pilot (ACTRN12618001547235), recruited patients between February 2019 and March 2020 from 2 hospitals in Sydney, New South Wales [25].

Patients were eligible if they were being discharged after an admission for either HF or acute coronary syndrome, were aged ≥18 years, and owned a smartphone. Patients were excluded if they were unable or unwilling to provide informed consent, unable to operate the app or attend in-person follow-up, or traveled overseas in the first 30 days after discharge. Patients were either randomized to usual care or TCC plus usual care. The intervention arm involved measuring blood pressure, heart rate, and weight daily to input into the TCC app on their smartphone, as well as wearing a fitness band which tracked minutes of activity per day. If an input was outside of the defined limits, an alert was emailed to the monitoring team (a cardiologist and a cardiac nurse practitioner) who reviewed the alert and contacted the patient, and if deemed necessary, their general practitioner or cardiologist would alter management. The primary objective was to examine the efficacy of the TCC model compared with usual care on the incidence of 30-day hospital readmission rates.

The TCC-Cardiac RCT (ACTRN12621000754842) began recruitment in July 2021 from 8 public hospitals across regional and metropolitan New South Wales [26]. Patients were eligible if they were being discharged after an admission with acute myocardial infarction (AMI) or decompensated HF, provided written consent, and were aged ≥18 years. They were excluded if they had a cognitive impairment, terminal illness, limited English to operate the app or communicate with staff, enrolled in another active study, or were determined to not be able to adhere to study requirements. Patients were stratified into 1 of 3 cohorts depending on their access to technology. Cohort 1 were randomized to receive either the TCC-Cardiac model of care or usual care, cohort 2 participants were randomized to TCC-text or usual care, and cohort 3 was a registry of patients with no mobile phones receiving usual care. The TCC-Cardiac model of care collected the same data as the TCC intervention arm, as well as oxygen saturations. Additionally, the app provided medication reminders and measurement of compliance, a virtual exercise program, and education to reinforce health behaviors such as physical activity, smoking cessation, and healthy eating. The follow-up of this data was the same as the TCC intervention arm. The TCC-text intervention cohort received automated messages tailored to their diagnosis and smoking status with the aim to prompt positive behavioral change. The primary objective was the 6-month unplanned hospital readmission rate in cohort 1 patients randomized to TCC-Cardiac as an adjunct to usual care versus usual care alone. These RCTs form the basis of this study as outlined in Figure 1.

In both trials, screening involved visually inspecting phones to confirm it was an eligible smartphone, with a smartphone being defined as a mobile phone that has internet access and is capable of downloading, installing, and operating apps. If the patient did not have a compatible smartphone, then they were either categorized as having an incompatible phone (either unable to download or operate the TCC app, or a phone that does not have internet capability) or no phone.

The TCC-Pilot data were grouped in the pre–COVID-19 cohort, while the TCC-Cardiac data were grouped in the post–COVID-19 cohort. Patients were excluded if smartphone ownership was not recorded, they did not live in Sydney, or they were not admitted secondary to HF or CAD. Data regarding smartphone ownership from TCC-Cardiac were used for this study until February 2023. Subsequently, supplemental patient demographic and medical history data were collected from the electronic medical record. The Socio-Economic Indexes for Areas Index of Relative Socio-Economic Advantage and Disadvantage was determined based on the patient’s postcode. Poor mobility was a clinical judgment based on their medical history, including high falls risk or assistance required to mobilize. Patient risks and comorbidities were compared according to phone ownership using the chi-square and Kruskal-Wallis tests for categorical and continuous variables, respectively. Logistic regression was used to estimate the odds ratios for smartphone ownership adjusted for age and sex. All analyses were conducted using Stata (version 15.1; StataCorp).

The study was reviewed and approved by the South-Eastern Sydney Local Health District Human Research Ethics Committee (2022/ETH01643). As this was a retrospective audit, consent was not required. All data were deidentified.

This study is currently the largest study of smartphone ownership in an inpatient population. In this population of patients with AMI, unstable angina, and decompensated HF, smartphone ownership was just over 50%. It was less common among older adults; patients with comorbidities; and those with markers of lower socioeconomic status, such as lack of private health insurance. This study highlights that mHealth interventions could fail to address the health care disparities they set out to overcome and are likely excluding the most vulnerable population. Understanding smartphone ownership patterns is vital to tailoring mHealth interventions and improving accessibility.

This study is not alone in highlighting the inequity of mHealth interventions due to low smartphone ownership rates. Similarly, in a survey of 100 outpatients with HF in the United States, 93% owned a mobile phone, with only 68% owning a smartphone [27]. In contrast, in a 2018 survey of 50 US patients (mean age 64.5, SD 8.3 years; 32% women), the study by Sohn et al [28] found that 90% of patients with HF owned a smartphone and 49% owned a tablet. However, the study population consisted of mainly White, male patients aged 50 to 80 years, which is a younger and less diverse population than our own. More recent studies have also found varying results depending on geographical location and setting examined. A study of a low socioeconomic district in South Africa investigating mHealth interventions for HIV found that of the 788 individuals surveyed, 86% owned a personal mobile phone but only 43% of these were smartphones [29]. A study of 949 dialysis patients in the United States found that 81% owned smartphones or other internet-capable devices, with 72% using the internet [30]. Rates of smartphone ownership vary significantly between patient populations; however, the trend has not been increasing in recent years when comparing studies. Our hypothesis that the COVID-19 pandemic would increase smartphone ownership appears to be false in the target population; however, the collected data present a valuable insight into smartphone ownership patterns in the surveyed population.

This is contrasted with studies looking at smartphone ownership in the general population. The Health Information National Trends Survey, a nationally representative cross-sectional survey of civilian, noninstitutionalized adults aged >18 years in the United States, found that of the 5438 responders, 87.8% were smartphone owners, 8.3% had incompatible mobile phones only, and 3.9% had no phone [31]. This cohort had equal male to female respondents with an average age of 49 years and 63.5% identifying as White. In a subset of this population who identified as having or being at risk for cardiovascular disease, there was a smaller but still high prevalence of smartphone ownership (73%) [32]. This is similar to the Pew Research Center [33] cross-sectional survey of 5626 people living in the United States which found that 98% of people owned a mobile phone, with 91% having smartphones. Similarly, the respondents to this survey were predominantly aged <65 years and White; information on gender was not collected. Clearly, our inpatient population is not reflective of the general population in terms of age, sex, or smartphone ownership. Drawing broadly from these studies in conjunction with our own confirms that inpatients have lower rates of smartphone ownership than the general population.

A deeper analysis of smartphone ownership patterns within the patient population is required to understand the potential reach of the current and future mHealth interventions, as well as to ensure equity in these interventions. Many studies agree that there is a digital divide between younger and older adult patients, with younger people being more likely to own a smartphone [27,29-32,34]. This is especially highlighted by the Pew Research Center [33] survey which found 98% of people aged 18 to 29 years owned smartphones, but this decreased to 79% in those aged >65 years. The study by Yao et al [35] highlights in their scoping review that socially disadvantaged groups have more challenges in accessing digital health technologies, which may lead to more severe health inequities. This is supported by multiple studies showing that people who are employed, female [29], and have higher levels of education are more likely to own a smartphone [30-32,36]. Our study supports this with higher rates of private health insurance, a marker of higher socioeconomic status, being associated with smartphone ownership. Greater health literacy has also been associated with smartphone ownership [34,37]. Our study is the first to identify a detailed list of conditions associated with lower smartphone ownership; specifically, a history of GORD, poor mobility, valvular heart disease, atrial fibrillation, and HF. It is proposed that having a history of GORD is associated with not owning a smartphone, given it is known to be correlated with low socioeconomic status. However, the study by Na and Sheu [31] identified that not owning a smartphone was associated with chronic conditions such as being deaf, heart disease, lung disease, and cancer.

mHealth is a useful tool for addressing health disparities through equitable access to health information; however, there remain health inequities caused by the adoption of digital health technologies in health care. The determinants of access to mHealth are complex and often interrelated. Given that smartphones are a relatively new technology, a significant barrier is the lack of perceived usefulness and ease of use of the technology system preventing patients from prioritizing smartphone ownership [38]. There are both technology factors, such as lack of network infrastructure particularly in regional areas (eg, power, network coverage, and internet access) and the availability of smartphones, as well as patient factors (eg, age, sex, race, rurality, educational attainment, income, poor health conditions, health literacy, and education level) that contribute to not owning a smartphone.

Smartphone ownership in older adults is the most well-researched factor. The study by Wildenbos et al [39] identified 4 domains that were barriers to mHealth uptake in this population: cognition, motivation, physical ability, and perception. Learning to use mobile technology can be a challenging task due to little prior knowledge [40], lack of confidence [41], and physical impairments such as poor vision and fine motor skills [42,43]. Using mobile technology for mHealth interventions is an added complexity as the individual requires digital health literacy, a complex competency that includes health literacy, digital literacy, and the ability to make decisions on the basis of digital health data [44]. The reasons for lack of smartphone ownership are many and varied; however, it is the first barrier to accessing mHealth interventions.

Given almost half of the target population cannot access mHealth interventions in this study, strategies need to be devised to increase access to these apps to reduce hospitalization rates [11]. The study by Yao et al [35] suggests government agencies and medical institutions should provide help with resources to improve accessibility of mHealth. The aim would be to recoup the cost of these devices by avoiding hospital admissions, with mobile phones and tablets being available for AUD $200 (US $142.51) and data plans for AUD $180 (US $128.26) per year, compared with the average cost of a hospital bed in New South Wales Health of AUD $1660 (US $1182.83) per day [45], with cardiology and coronary care beds exceeding this. This suggests that the cost of a mobile phone is diminutive in comparison to readmission, a hypothesis that should be evaluated in future research. Companies should be encouraged to donate tablets and smartphones to hospitals, and software developers should be encouraged to maintain backward compatibility of their apps with older versions of devices and mobile operating systems. This would improve the accessibility to such services for socially disadvantaged groups. If a low rate of smartphone ownership is present in a population, app-based interventions may not be feasible, and alternative methodologies may be more equitable such as telephone or SMS text message–based services. Once accessibility has been overcome, the design of the intervention needs to be curated with consideration of the target group, preferably with their input during the design and implementation stages [35,46]. This will maximize accessibility and acceptability for all groups who use the app, reducing disparities. To ensure perceived usefulness, this needs to be explicitly communicated to the patient during the recruitment phase. Finally, new users need to be adequately supported when adopting novel mHealth interventions, especially when there is evidence supporting their efficacy.

Although this is the largest study of smartphone ownership in inpatients, there are limitations. First, the study excluded patients with cognitive impairment or limited English proficiency. These patient populations often face the highest barriers to digital literacy and health care access, thus potentially underestimating the digital divide. Second, a significant number of patients were excluded from the post–COVID-19 cohort due to missing phone ownership details. The phone must have been visualized to be included in the study; this is beneficial as it ensures accuracy but also confounds the data as patients who did not bring their phone to hospital were excluded from the study. Third, it was difficult to assess the patients’ sociodemographics completely given employment status was not available. The participants were primarily located at 1 local health district in a relatively affluent area in Sydney as shown by a median decile Socio-Economic Indexes for Areas Index of Relative Socio-Economic Advantage and Disadvantage of 9 out of 10 in both cohorts. However, this tool does not account for the nuances of this area where social housing is spread among the wealthy. As such, private health insurance status was chosen to represent socioeconomic status as a reflection of discretionary spending. Fourth, the poor mobility metric was based on clinical judgment, which is subjective and may introduce bias. Fifth, as acknowledged, there are many factors affecting smartphone ownership; however, only age and sex were adjusted for. Finally, only patients with an admission diagnosis of HF, AMI, and unstable angina were included, and therefore we are unable to extrapolate to other patient populations.

mHealth interventions were introduced to eliminate health disparities and have the potential to improve chronic disease management. However, inequality in smartphone ownership in the highest risk comorbid patients as seen in this cohort has the potential to exacerbate rather than eliminate these disparities. Interestingly, this study showed that there was no improvement in smartphone ownership during the COVID-19 pandemic. Until smartphone ownership is commonplace in patient populations, these groups will continue to be systemically excluded. Health care providers who plan to use mHealth interventions in their practice need to ensure accessibility to all and have contingency plans if participants do not own a smartphone, such as providing a phone or tablet, or using other means to provide the same service.