Adherence with blood pressure monitoring wearable device among the elderly with hypertension: The case of rural China

Abstract Introduction Wearable blood pressure (BP) monitor devices are increasingly adopted owing to the promotion of hypertension management program. However, little is known about the adherence and its associated factors in older adults (OAs) with hypertension. Methods The aim of this study was to determine factors associated with adherence to wearable BP monitor. In total, 212 OAs with hypertension in a remote rural area in China were asked to wear a BP monitor over 1‐month period. The following information on associated factors for adherence was collected, including demographic characteristics, cardiovascular health measurements, technology fluency, the Compliance of Hypertensive Patients' Scale, and the Health‐related Quality of Life Survey. As a result, the mean age of the 212 hypertension participants was 71.32 years (SD = 6.81). Results During the 30‐day survey, 50.94% of the participants with daily recoded BP wristband data were assigned as “BP device users.” The binary logistic regression model revealed that lower lifestyle compliance, lower medication compliance and higher total hypertension compliance were significant predictors for adherence. However, there were no differences concerning cardiovascular health factors, technology fluency, and health‐related quality of life between device users and nonusers. Conclusion Individuals reporting a higher level of total compliance in hypertension management were more likely to wear a BP monitor device among OAs with hypertension. In addition, further research is needed to determine how wearable mHealth technology can be used to develop better hypertension self‐management education programs for this population.

To the best of our knowledge, studies have rarely focused on the influencing factors with the adherence with wearable BP monitor device in a remote rural area in China.
To date, several epidemiologic studies have assessed the compliant wearable device users (Glynn et al., 2015;Mair et al., 2012;Murray et al., 2010). Yet, little information is available about the specific features contributing to adherent device users in the community-based elderly population. To this end, the present study was designed to evaluate the differences in sociodemographic characteristics, cardiovascular health factors, technological fluency, hypertension compliance, and health-related quality of life between device users and nonusers in the elderly. The ultimate goal of our study was to provide potential future directions for the mHealth technology in this field.

| Study population
Individuals with confirmed hypertension were recruited for a selfmonitoring intervention program for hypertension control in a remote mountainous district of Hubei province from April 2018 to July 2018 (the disposition of participants was shown in Figure 1). The inclusion criteria were shown as follows: (a) living in the Xuanen area; (b) aged over 60 years; (c) systolic BP ≥140 mm Hg and/or diastolic BP ≥90 mm Hg or a history of antihypertensive medication; and (4) communication proficiency to carry out study tasks. Exclusion criteria were as follows: (a) cognitive dysfunction; (b) without smartphones to use the device; and (c) hypertension emergency or other urgency.
Participants completed the study in the following orders: (a) registration and signing written informed consent; (b) measurement of BP, height, weight, waistline, and hipline; (c) survey instrument completion; (d) mHealth device training; and (e) debriefing. The collection of BP data continued throughout the 30-day period.
The study was approved by the institutional review boards of Wuhan University, and written informed consent was obtained from all participants.

| BP data collection
A 12-hr ambulatory BP monitor was used to assess the BP of participants. The device was pilot-tested from December 2017 to February 2018 among a similar elderly population in Wuhan, China (N = 98). The accuracy and reliability of the device were validated.
Based on previous studies, the monitor was designed to automatically record BP at 1-hr interval during awake hours (Celler, Argha, Varnfield, & Jayasena, 2018). The BP monitoring system also included a secure online account to track CVD-related health factors ( Figure 2). A hub was installed in an accessible location at each community. The wristband device could automatically upload the recorded BP to central hub location through a smartphone application daily during the 1-month study. Pressing the device button would prompt the smartphone to display BP results. The system would terminate the transmission of data if there was no Internet connection. Upon receipt of the ambulatory BP monitor, participants were trained on the usage of the wristband and provided with a written instruction manual. More details of the research process were reported elsewhere Zhang, Xu, et al., 2019;. Each community had a "super user," a participant volunteer who helped community participants with hub issues and communicated with investigators.
The uploaded BP data were collected and the device usage of participants was monitored by investigators during the 30-day study period.
All BP wearable devices were provided for free.

| Definition of adherence
Adherence to BP monitor was defined as daily recording of BP data during the 30-day assessment period. "BP device user" (coded "1") referred to participants with daily uploaded BP results over the study period. While participants without daily recorded BP data were noted as "BP device nonuser" (coded "0").
F I G U R E 2 Secure data collection system for blood pressure monitor data

| Computer-email-web fluency
Utilization of mobile phone and Internet technology was evaluated using the adapted Computer-Email-Web (CEW) Fluency Scale developed by Bunz (2004). The CEW is one of the most widely used scales to measure levels of technology fluency in OAs (Bunz, Curry, & Voon, 2007). The generally adapted CEW is a 17-item scale, all of which are rated on a 5-point Likert scale ranging from 1 (no fluency) to 5 (high fluency) based on three subscales (computer skills, email skills, and web skills). Higher scores indicated higher levels of fluency. In this study, the Cronbach's alpha of the CEW was 0.985.

| Hypertension compliance
Hypertension compliance was measured using the Compliance of Hypertensive Patients' scale (CHPS) developed by Kyng and Lahdenper (2010). CHPS is one of the most commonly used instruments to evaluate hypertension compliance in OAs (Lahdenperä, Wright, & Kyngäs, 2003). To be specific, CHPS, a 15-item scale, could be used to assess the level of compliance in the following six aspects: Intention (4 items), Lifestyle (3 items

| Health-related quality of life
Health-related quality of life (HRQOL) was determined using the 12-Item Short-Form Health Survey (SF-12), which was originally developed as a short alternative to the SF-36. The SF-12, a widely used generic questionnaire, has been widely validated to measure HRQOL in OAs (Bentur & King, 2010;Jakobsson, 2007). It is measured by two summarized components: a physical component score (PCS) and a mental component score (MCS). Both PCS and MCS scores range from 0 to 100 according to the scoring program described by Ware, Kosinski, and Keller (1996). Higher scores of each component indicated higher levels of life quality. In this study, the Cronbach's alpha of the SF-12 was 0.828.

| Data analysis
SPSS software version 21.0 was used for statistical analysis.
Descriptive baseline sociodemographic characteristics of the participants were presented as percentages. Independent t test (for continuous variables) and chi-squared test (for categorical variables) were utilized for comparison between user and nonuser groups.
Afterward, a binary logistic regression analysis was constructed to determine the significant variables associated with the adherence to device use. The variables selected for regression model were based on all variables from the bivariate analysis with p ≤ .05. A two-sided p ≤ .05 was considered as statistically significant.

| RE SULTS
The baseline characteristics of the users and nonusers were shown in Table 1. In total, 70.67% participants completed all survey and measurements at baseline. Of the 212 subjects, 110 (51.89%) were female, with the mean age of 71.32 years (standard deviation: 6.81, range: 60-98). During the 30-day study period, 50.94% of the participants uploaded BP wristband data to central hub location at least once daily. The family monthly income (p = .040) and number of concomitant diseases (p = .043) were significantly different between users and nonusers.
The cardiovascular-related health factors for both users and nonusers were shown in Table 2. To be specific, body mass index (BMI; p = .008) and waist circumference (p = .005) were significantly different between users and nonusers, while no significant differences were identified on systolic blood pressure (SBP), diastolic blood pressure (DBP), or hip circumferences between the two groups.
As depicted in Table 3, all participants reported some level of technology fluency other than "Not at all" on technology fluency for all three subscales. No differences were identified in computer skills (p = .439), email skills (p = .453), web navigation skills (p = .540), or the cumulative score (p = .453) between the two groups.
As shown in Table 4, intention (p < .001), lifestyle (p < .001), attitude (p = .001), medication use (p = .003), and the total hypertension compliance score (p < .001) were statistically significant in hypertension compliance. However, there was no significant difference on responsibility (p = .962) or smoking and drinking (p = .753) between the two groups.
As shown in Table 5, both PCS (p = .597) and MCS (p = .532) scores were not significantly different.
The results from binary logistic regression models were summarized in Table 6. The higher levels of adherence to BP monitor were significantly found in those with lower lifestyle compliance, lower medication compliance and higher total hypertension compliance, while family monthly income, number of concomitant disease(s), BMI, waist circumference, intention, or attitude failed to account for significant variance in this model.

| D ISCUSS I ON
The study was aimed to explore the differences in sociodemographic characteristics, cardiovascular-related health factors, technology fluency, hypertension compliance, and health-related quality of life between users and nonusers. Our results revealed that individuals reporting lower compliance with healthy lifestyle and medication use, but higher total hypertension compliance, were more likely to use mHealth BP monitor over the 30-day period. In addition, half of  (2010) has indicated that "health lifestyle" is not only related to medication adherence behavior, but also associated with self-management behaviors including diet, exercise, and weight control. However, our results also suggested that OAs with lower "health life" compliance and medication were more likely to adopt the BP monitor. It might be due to the fact that the study population generally lived with their children and took care of their grandchildren or performed other important tasks for their families. OAs shared their concerns by communicating with family members to cope with the challenges of BP management under the guidance of family members. In a previous study by Chang, Lu, Yang, and Pin (2016), social support exerts a positive effect on users' engagement and adherence with activity trackers. Furthermore, Piette et al. (2015) have asserted that incorporation of caregivers into a mHealth patient monitoring intervention leads to improved medication adherence and improved quality of life among patients with depression. Our findings also underscored the importance of improving hypertension compliance when designing an effective intervention for hypertension control.
Therefore, further research is warranted to identify the role of hypertension compliance in device adoption in population with more diverse backgrounds.
In addition, we also found technology fluency as an insignificant factor in the regression model on device adherence prediction, which was consistent with the study by Yingling et al. (2017) on African American population, indicating that lower technology did not appear to impede adoption of mHealth. This relationship attributing similar access to wearable device for both users and nonusers is consistent with the growing widespread mHealth device adoption ownership (Coughlin & Smith, 2016). However, in other trials (Angladamartínez, Roviraillamola, Martinconde, Sotocamomblona, & Codinajané, 2016;Karhula et al., 2015), difficulty in operating technology could result in a lack of device adherence. Other research groups also tried to identify specific age-related barriers hindering OAs' interaction with multiple mHealth device, including mobile application, smartphones, and wearable activity monitors (Baig, Gholamhosseini, Moqeem, Mirza, & Lindén, 2017;Foster & Sethares, 2014;Wildenbos, Peute, & Jaspers, 2018). All these inconsistent results, however, underlie the necessity for a more rigorously designed study to tease out the true relationship between technology fluency and device adherence.
Surprisingly, our findings also suggested no differences in Likewise, Yingling et al. (2017) has found that lower SES participants may be more likely to interact with a mHealth system than higher SES individuals. Nevertheless, more research is required to explore the path model connecting sociodemographic, quality of life and user adoption adherence among population with more diverse backgrounds. Although the adoption of mHealth in OAs lags behind of younger individuals consistently (Searcy et al., 2019), a comprehensive understanding of the differences between device users and nonusers can identify barriers in device engagement and user adherence, thereby providing information to better allocate intervention resources in this population. As Burke et al. (2012) ranked "identifying strategies that improve self-monitoring adherence" as an important factor in moving forward for mHealth-based weight loss efforts. Our study was characterized by the assessment of user adoption of wearable BP monitoring device among OAs in a resource-limited rural area.
We observed that participants with higher levels of hypertension compliance also had high levels of device adherence. Thus, mHealth nonusers were more likely to be suboptimal hypertension compliance. The desire of OAs to engage the intervention is the main motivational inertia barrier, which is more difficult to overcome. Reaching these suboptimal participants may optimize mHealth tools in early intervention by healthcare setting to promote more engagement.
Limited technology usage and fluency should be noted to reduce potential disparities in the application of BP monitoring device to population-based interventions in resourced-limited settings. Our findings revealed that lower technology fluency was unlikely to impede integration of this mHealth BP monitoring system among OAs, which is consistent with the premise of the "new digital divide," where technology may not be the most prominent along SES lines (Atienza & Patrick, 2011