This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.
The treatment of many chronic illnesses involves long-term pharmaceutical therapy, but it is an ongoing challenge to find effective ways to improve medication adherence to promote good health outcomes. Cardioprotective medications can prevent the enlargement of harmful clots, cardiovascular symptoms, and poor therapeutic outcomes, such as uncontrolled high blood pressure and hyperlipidemia, for patients with coronary heart disease. Poor adherence to cardioprotective medications, however, has been reported as a global health concern among patients with coronary heart disease, and it is particularly a concern in China.
This study aimed to evaluate the efficacy of a mobile health (mHealth) intervention using 2 mobile apps to improve medication adherence and health outcomes.
A randomized, placebo-controlled, 2-arm parallel study was conducted in a major university-affiliated medical center located in Chengdu, China. Participants were recruited by flyers and health care provider referrals. Each participant was observed for 90 days, including a 60-day period of mHealth intervention and a 30-day period of nonintervention follow-up. The study coordinator used WeChat and Message Express to send educational materials and reminders to take medication, respectively. Participants used WeChat to receive both the educational materials and reminders. Participants in the control group only received educational materials. This study received ethics approval from the Duke Health Institutional Review Board (Pro00073395) on May 5, 2018, and was approved by West China Hospital (20170331180037). Recruitment began on May 20, 2018. The pilot phase of this study was registered on June 8, 2016, and the current, larger-scale study was retrospectively registered on January 11, 2021 (ClinicalTrials.gov).
We recruited 230 patients with coronary heart disease. Of these patients, 196 completed the baseline survey and received the intervention. The majority of participants were married (181/196, 92.4%), male (157/196, 80.1%), and lived in urban China (161/196, 82.1%). Participants’ average age was 61 years, and half were retired (103/191, 53.9%). More than half the participants (121/196, 61.7%) were prescribed at least 5 medications. The mean decrease in medication nonadherence score was statistically significant at both 60 days (t179=2.04,
The proposed mHealth intervention can improve medication adherence and health outcomes, including systolic blood pressure and diastolic blood pressure.
ClinicalTrials.gov NCT02793830; https://clinicaltrials.gov/ct2/show/NCT02793830 and ClinicalTrials.gov NCT04703439; https://clinicaltrials.gov/ct2/show/NCT04703439
Coronary heart disease (CHD) is a heart and blood vessel disease related to atherosclerosis in which plaque builds up in the walls of the coronary arteries, narrowing the arteries and restricting the flow of blood [
In China, however, poor adherence to cardioprotective medications has been reported as a public health concern [
After these pilot studies, mobile phone–based mHealth interventions have been conducted in the field of cardiovascular medicine in China [
This study used 2 mobile apps: WeChat (Tencent Inc) and Message Express (Bluemobile.zt). WeChat is the most widely used messaging app in China [
This unblinded, 2-arm, parallel randomized controlled trial was conducted between May and December 2018 at the Cardiology Department of West China Hospital, located in Chengdu. West China Hospital is a major university-affiliated hospital that serves more than 10,000 outpatients a day [
This study received ethical approval from the Duke Health Institutional Review Board (Pro00073395) and West China Hospital (20170331180037).
The interventions were refined based on the pilot study [
Brief description of the 2 apps.
Participants’ demographic characteristics were recorded, including age, gender, ethnicity, weight, height, marital status, job status, education, medical insurance status, living area, number of prescribed medications, and family income. A REDCap survey hyperlink was sent to each participant to collect their demographic characteristics and track their health outcome variables at 7 time points: enrollment (ie, baseline); 15, 30, 45, and 60 days after enrollment (ie, at the end of the intervention); and 75 and 90 days after enrollment (ie, at the end of follow-up). Health outcome variables included medication nonadherence score, heart rate (HR), systolic blood pressure (SBP), and diastolic blood pressure (DBP). The primary outcome (ie, medication nonadherence score) was measured using a validated 3-item, 5-point Likert scale, the Voils Extent Scale [
All analyses were performed using SAS 9.4. Categorical variables such as gender, ethnicity, marital status, job status, medical insurance, and living area were reported as frequency (n) and percentage (%). Numerical variables including age, weight, height, and number of prescribed medications were reported as mean (SD). The chi-square test and independent
The mean (SD) for all outcome variables (ie, medication nonadherence, HR, SBP, and DBP) in each group was computed and graphed to show trends at the 7 time points described above. Differences in outcomes between the experimental and control groups at the critical time points (ie, baseline, 60 days, and 90 days) were determined using the
In this study, we recruited 230 participants and randomly assigned 116 to the experimental group and 114 to the control group (see
CONSORT (Consolidated Standards of Reporting Trials) flow diagram.
Baseline characteristics of participants at enrollment.
Variable | All participants (N=196) | Experimental group (n=103) | Control group (n=93) | ||
|
|
|
|
.86 | |
|
Male | 157 (80.1) | 83 (80.6) | 74 (79.6) |
|
|
Female | 39 (19.9) | 20 (19.4) | 19 (20.4) |
|
Age (years), mean (SD) | 61 (11) | 61 (11) | 62 (11) | .48 | |
Height (cm), mean (SD) | 164.8 (8.0) | 165.5 (8.0) | 164.1 (8.0) | .23 | |
Weight (kg), mean (SD) | 67.6 (11.3) | 68.2 (11.9) | 66.9 (10.5) | .45 | |
|
|
|
|
.98a | |
|
Han | 184 (93.9) | 96 (93.2) | 88 (94.6) |
|
|
Zang | 6 (3.1) | 3 (2.9) | 3 (3.2) |
|
|
Yi | 1 (0.5) | 1 (1.0) | 0 (0) |
|
|
Hui | 1 (0.5) | 0 (0) | 1 (1.1) |
|
|
Mongolian | 1 (0.5) | 1 (1.0) | 0 (0) |
|
|
Other ethnic minorities | 3 (1.5) | 2 (1.9) | 1 (1.1) |
|
|
|
|
|
.25 | |
|
Married | 181 (92.4) | 93 (90.3) | 88 (94.6) |
|
|
Widowed, separated, divorced, or single | 15 (7.6) | 10 (9.7) | 5 (5.4) |
|
|
|
|
|
.31a | |
|
Employed | 65 (34.0) | 39 (39.0) | 26 (28.6) |
|
|
Unemployed | 4 (2.1) | 3 (3.0) | 1 (1.1) |
|
|
Farmer | 19 (10.0) | 8 (8.0) | 11 (12.1) |
|
|
Retired | 103 (53.9) | 50 (50.0) | 53 (58.2) |
|
|
|
|
|
.36 | |
|
Primary school or lower | 31 (16.1) | 16 (15.8) | 15 (16.3) |
|
|
Middle school | 43 (22.3) | 27 (26.7) | 16 (17.4) |
|
|
High school | 47 (24.3) | 26 (25.7) | 21 (22.8) |
|
|
Noncollege postsecondary | 28 (14.5) | 11 (10.9) | 17 (18.5) |
|
|
College or above | 44 (22.8) | 21 (20.8) | 23 (25.0) |
|
|
.82a | ||||
|
<5 | 75 (38.3) | 41 (39.8) | 34 (36.6) |
|
|
5-9 | 116 (59.2) | 59 (57.3) | 57 (61.3) |
|
|
≥10 | 5 (2.5) | 3 (2.9) | 2 (2.1) |
|
Covered by medical insurance | 186 (96.9) | 97 (96.0) | 89 (97.8) | .69a | |
|
.90b | ||||
|
Complete coverage | 41 (27.9) | 22 (30.1) | 19 (25.7) |
|
|
Some coverage | 89 (60.5) | 40 (54.8) | 49 (66.2) |
|
|
No coverage | 17 (11.6) | 11 (15.1) | 6 (8.1) |
|
|
|
|
|
.82 | |
|
Urban | 161 (82.1) | 84 (81.6) | 77 (82.8) |
|
|
Rural | 35 (17.9) | 19 (18.4) | 16 (17.2) |
|
|
|
|
|
.36 | |
|
Living alone | 14 (7.1) | 9 (8.7) | 5 (5.4) |
|
|
Living with family or relatives | 182 (92.9) | 94 (91.3) | 88 (94.6) |
|
|
.78 | ||||
|
<1 year | 24 (12.0) | 11 (10.7) | 13 (14.0) |
|
|
1-4 years | 109 (56.0) | 58 (56.3) | 51 (54.8) |
|
|
≥5 years | 63 (32.0) | 34 (33.0) | 29 (31.2) |
|
|
.44 | ||||
|
Daily | 152 (80.9) | 83 (83.8) | 69 (77.5) |
|
|
Occasionally | 25 (13.3) | 12 (12.1) | 13 (14.6) |
|
|
Never | 11 (5.8) | 4 (4.0) | 7 (7.9) |
|
|
.12b | ||||
|
<¥54,000 | 81 (52.3) | 48 (59.3) | 33 (44.6) |
|
|
¥54,001-¥90,000 | 34 (21.9) | 13 (16.1) | 21 (28.4) |
|
|
¥90,001-¥120,000 | 18 (11.6) | 11 (13.6) | 7 (9.5) |
|
|
>¥120,000 | 22 (14.2) | 9 (11.1) | 13 (17.6) |
|
|
|
|
|
.89 | |
|
Good | 70 (35.7) | 37 (35.9) | 33 (35.5) |
|
|
Fair | 101 (51.5) | 54 (52.4) | 47 (50.5) |
|
|
Bad | 25 (12.8) | 12 (11.7) | 13 (14.0) |
|
aThe Fisher exact test was used due to small values in cells.
bThe Mann-Whitney
The medication nonadherence score (
HR decreased (
The proportional rate of participants with normal SBP (
Although no baseline characteristics were significantly different between the experimental group and the control group, HR and blood pressure can be influenced by body weight [
The difference in the rate of change of SBP (
Comparison of changes in medication nonadherence score between groups (SE 2).
Comparison of baseline and unadjusted changes in medication nonadherence and health outcomes between the 2 arms.
|
Experimental group (n=103), mean (SD) | Control group (n=93), mean (SD) | |||
|
|
||||
|
Baseline | 6.85 (1.85) | 7.03 (2.05) | 0.64 (194) | .52 |
|
Day 60 | –1.21 (2.59) | –0.42 (2.63) | 2.04 (179) | .04 |
|
Day 90 | –1.58 (2.49) | –0.08 (3.15) | 3.48 (155) | <.001 |
|
|
||||
|
Baseline | 73.93 (12.03) | 73.16 (9.76) | –0.49 (183) | .63 |
|
Day 60 | –1.46 (12.68) | –1.95 (9.03) | –0.28 (148) | .78 |
|
Day 90 | –1.88 (12.88) | –1.32 (8.98) | 0.32 (145) | .75 |
|
|
||||
|
Baseline | 129.5 (14.37) | 125.2 (15.10) | –2.04 (188) | .04 |
|
Day 60 | –2.14 (16.20) | 2.72 (16.07) | 1.92 (161) | .06 |
|
Day 90 | –2.87 (15.10) | 4.38 (14.89) | 3.12 (165) | .002 |
|
|
||||
|
Baseline | 78.71 (11.73) | 76.03 (15.98) | –1.30 (162) | .20 |
|
Day 60 | –2.46 (12.49) | 1.87 (14.06) | 2.07 (160) | .04 |
|
Day 90 | –1.57 (12.21) | 2.92 (13.99) | 2.21 (164) | .03 |
Comparison of changes in heart rate between groups (SE 2).
Comparison of changes in systolic blood pressure between groups (SE 2).
Comparison of changes in diastolic blood pressure between groups (SE 2).
Results of a mixed-effects model with a random intercept and slope to measure between-group differences in the trajectory of changes in outcomes.
Item | Medication nonadherence | Diastolic blood pressure (mmHg) | Systolic blood pressure (mmHg) | Heart rate (bpm) | ||||
|
Parameter |
Parameter |
Parameter |
Parameter |
||||
Intercept | 5.11 | <.001 | 84.86 | <.001 | 93.48 | <.001 | 56.66 | <.001 |
Group (reference: control) | –0.23 | .37 | 2.05 | .19 | 3.39 | .06 | 0.68 | .63 |
Gender (reference: male) | 0.29 | .33 | –2.50 | .15 | 0.23 | .92 | 2.00 | .26 |
Age | 0.002 | .85 | –0.22 | <.001 | 0.26 | .001 | 0.001 | .99 |
Education | 0.11 | .17 | 0.17 | .72 | 0.41 | .50 | –0.21 | .66 |
Weight | 0.013 | .27 | 0.11 | .12 | 0.21 | .02 | 0.21 | .003 |
Time (days) | 0.001 | .74 | 0.04 | .14 | 0.06 | .09 | –0.01 | .33 |
Group × time | –0.02 | <.001 | –0.05 | .004 | –0.08 | <.001 | –0.01 | .74 |
The aim of this study was to assess if a mobile phone–based mHealth intervention could improve medication adherence and relevant health outcomes (eg, blood pressure and HR) among patients with CHD in comparison to a control group that received general educational materials over a period of 2 months. We found that our mHealth intervention increased medication adherence and had a lasting effect in improving medication adherence among patients with CHD, even though no intervention was given after 60 days. We also found that our mHealth intervention improved health outcomes by lowering SBP and DBP, and that this effect continued for 30 days after the intervention. Unlike SBP and DBP, our mHealth intervention did not significantly lower HR, although mean HR consistently remained within the normal range. After dichotomizing SBP, DBP, and HR into binary variables, we found that there was no significant difference between the proportional rate of participants with normal SBP, DBP, and HR at baseline, 60 days, or 90 days between the intervention and control groups.
A mixed-effects model for continuous outcomes showed that differences in the rate of change of medication nonadherence, SBP, and DBP were statistically significant between the 2 groups after controlling for the effects of group, time, the group-by-time interaction, and some baseline variables that can influence HR and blood pressure, such as body weight, gender, age, and educational level. Similarly, the generalized mixed-effects model with a logit link used for dichotomized binary outcomes showed that after controlling for the effects of baseline body weight, gender, age, educational level, group, time, and the group-by-time interaction, our mHealth intervention could significantly increase the proportional rate of participants with normal SBP. In summary, these key findings demonstrate that our mHealth intervention could significantly increase medication adherence and the proportional rate of participants with normal SBP, as well as lower SBP and DBP, in patients with CHD.
In this study, participants were defined as dropping out if (1) they filled out the baseline survey but deleted the study coordinator’s WeChat account at any time point before their completion of the study; or (2) they told the study coordinator that they wanted to withdraw from the study. Participants were defined as lost to follow-up if they filled out the baseline survey and kept the study coordinator’s WeChat account, but did not fill out later surveys and did not respond to reminders and WeChat calls made by the study coordinator. This could happen at any time point during the study. In this study, 9 participants were lost to follow-up and 6 participants dropped out, which together accounts for 7% of the data. Due to the small amount of missing data, we did not perform missing data imputation. We compared the baseline variables between participants who dropped out or were lost to follow-up with those who completed the study, and found that none of the baseline variables were significantly different between these groups (
In this study, the ratio of male to female patients with CHD was 4:1, which is the same ratio as in our pilot study conducted in 2017. The higher proportion of male participants reflects the higher prevalence of CHD in men in China; Chinese men are more likely to engage in risky behaviors, such as smoking and drinking [
This study contributes important knowledge about mHealth as a tool to improve medication adherence, but it has several limitations. First, all participants were recruited at a major university-affiliated hospital; many of them were middle-aged male urban residents and were thus not representative of the general Chinese population. Second, health outcomes such as SBP, DBP, and HR were self-reported by participants. There may therefore have been discrepancies regarding these measurements. Third, during this study, some participants did not provide their health outcome data in a timely manner. To collect these data, some participants had to be reminded by WeChat messages and phone calls. These messages and phone calls might have been covariates that could influence the participants’ medication-taking behaviors, but we did not consider their influence. Finally, the mHealth intervention was not automated and lasted for only 60 days, a comparatively short period of time. To adopt this intervention in real clinical settings, which have thousands of patients, and to create long-term effects, it will be necessary to use automated methods. Finally, this study was unblinded. All participants, data collectors, and data analysts were aware of which treatment arms participants had been assigned to. To increase rigor and reduce bias, future large multisite studies should consider a double-blind design.
The treatment of many chronic illnesses involves long-term pharmaceutical therapy. Nevertheless, it is an ongoing challenge to find effective ways to improve medication adherence and other health behaviors to improve health outcomes. For patients with CHD, cardioprotective medications can prevent the enlargement of harmful clots [
Although some studies have been conducted in China on using mobile apps (eg, WeChat) to improve blood pressure and self-management behavior [
The treatment of many chronic illnesses involves long-term pharmaceutical therapy, but it is an ongoing challenge to find effective ways to improve medication adherence and promote good health outcomes. In this study, we examined an mHealth intervention to remind patients with CHD to take their cardioprotective medications. Our results demonstrate that it is feasible to conduct an mHealth intervention to improve medication adherence and health outcomes, represented by measures including SBP and DBP. In summary, mHealth interventions that are constructed using evidence-based content show promise to help increase medication adherence and improve health outcomes.
Comparison of changes of the proportional rates of normal SBP.
Comparison of changes of the proportional rates of normal DBP.
Frequency and proportion of normal blood pressures at multiple time points [n (%)].
Comparison of changes of the proportional rates of normal heart rate.
Results of the mixed-effects model for proportional rate change of normal health outcomes.
Baseline characteristics of participants who completed the study and those who did not [n (%)].
CONSORT-EHEALTH (V 1.6.1) checklist.
coronary heart disease
diastolic blood pressure
heart rate
mobile health
systolic blood pressure
This study was supported by the Duke University Global Health Institute (2018 Duke Global Health Doctoral Certificate Fieldwork Grant); Duke University Graduate School (2017 International Dissertation Research Travel Award); and the Duke University School of Nursing (2018 PhD Student Pilot Study Fund).
ZN contributed to conceptualization, methodology, investigation, and formal analysis and wrote the original draft. BW contributed to conceptualization, methodology, validation, supervision, writing, reviewing, and editing. QY contributed to conceptualization, methodology, validation, formal analysis, writing, reviewing, and editing. LLY contributed to conceptualization, methodology, validation, writing, reviewing, and editing. CL contributed to methodology, project coordination, writing, reviewing, and editing. RJS contributed to conceptualization, methodology, validation, supervision, writing, reviewing, and editing.
None declared.