This was a prospective, nonrandomized, longitudinal study performed in five provinces of northwest China. The inclusion criteria were aged ≥18 years and no serious cognitive disease or mental disorder. The exclusion criteria were aged <18 years with severe cognitive impairment or mental illness. By 2013, 56,542 people from Shaanxi, Gansu, Ningxia, Inner Mongolia, and Shanxi provinces, including 15 cities and 61 counties (districts, banners), had joined the health management platform, which is for personal use only. This platform includes modules such as exercise management and diet management. Complete relevant data were available for 51,486 people. The included study population participated in an annual medical checkup and completed a questionnaire every 2 years.

After collecting the physical examination data of the study participants in 2013, they received internet-based health management and follow-up. The platform is mainly divided into three modules: “understanding my health,” “improving my health,” and “health improvement effectiveness.” Physical examinations and the questionnaire were administered after 2 years (in 2015) and 4 years (in 2017). Hypertension was based on systolic blood pressure (SBP)≥140 mmHg and/or diastolic blood pressure (DBP)≥90 mmHg on repeated clinical measurements or a previous history of hypertension.

Based on the physical examination in 2013, we divided the newly identified patients with hypertension into two groups according to good and poor BP control. Good BP control was determined according to SBP<140 mmHg and DBP<90 mmHg on repeated clinical measurements; otherwise, the individual was considered to have poor BP control [18].

The health management platform is for private use only. Anyone can log in to their own account to access health records on the platform according to each physical examination to help users understand their own health status. Relevant experts then formulate an individualized diet, exercise, and other relevant opinions for individuals according to their health status, and carry out health management. After 3 months of health management, the personal health status is further assessed. Health records include individual basic information (eg, age, gender, height, weight) and clinical indicators (eg, BP, blood glucose, blood lipids) from the physical examination in that year, along with family history, diet, and exercise.

The system automatically formulates 12 stages of progressive exercise prescription according to the health status of an individual. Users can choose relevant sports as recommended and manually record the time and date of sports. One week later, the rationality of the recommendations is evaluated by comparing the exercise performed in the previous week with the recommended exercise prescription.

The system platform automatically formulates meal prescriptions for individuals based on the submitted health questionnaire information. Users who want to create personalized recipes according to their personal dietary preferences only need to modify and replace the foods in the expert-recommended recipes. By recording at least 2 days of meal diaries every week, the system can analyze the dietary nutrition status, identify potential problems in meals, and make appropriate recommendations.

The research data were derived from the physical examination data of participants from the five provinces in northwest China and from the questionnaire survey. The health management for participants included dietary guidance; physical exercise guidance; living habits guidance; basic knowledge of hypertension, diabetes, and hyperlipidemia; and other aspects.

Basic characteristics, including age, sex, and family history of hypertension, were collected at the first physical examination. Individual parameters were collected by questionnaire after each physical examination, including health-related behaviors (smoking, drinking, physical activity, sedentary time [hours], sleeping time [hours], quality of sleep [very good, fair, not good, very bad]), dietary intake (cereals and potatoes; fish, livestock, poultry, meat, and eggs; milk and dairy; soybean and nuts; vegetables; fruit). Physical activity was rated according to the International Physical Activity Questionnaire scale [19]. Dietary intake was rated according to the 2016 Chinese dietary guidelines [20], divided into lower than the recommended intake, recommended intake, and higher than the recommended intake. The psychological status of participants was evaluated by the General Health Questionnaire (GHQ-12) [21]. Compared with the results in 2013, the improvement of the results in 2017 was considered to indicate an improvement in relevant risk behaviors following the health management plan.

The clinical measurement results of participants were imported into an epidata database after each physical examination. An automatic sphygmomanometer was used to measure the BP of the left arm (the measurement position was flush with the heart) in a sitting position after the subject rested quietly for at least 5 minutes. The measurement was repeated at intervals of at least 5 minutes and the average of the two readings was recorded. Venous blood samples were collected during fasting, and fasting blood glucose was measured by the glucose oxidase method according to routine operating procedures. Total cholesterol, triglyceride, and other blood lipid indicators were measured on an empty stomach. All the onsite measurements including laboratory tests were validated for internal quality control according to clinical standards.

This study was ethically reviewed and approved by the Biomedical Ethics Committee of Peking University (IRB00001052-0816). All participants in the study provided informed consent. All data were anonymized. No subsidy was provided to the participants.

Descriptive analysis was used to describe the basic characteristics of participants. The raw data were processed using the 99-quantile capping method to remove outliers. Analysis of variance was used to analyze the annual changes in the clinical indices of 8576 subjects and newly identified patients with hypertension. The χ² test or analysis of variance was used to analyze the annual changes of health-related behavior and diet as appropriate. Patients with new-onset hypertension were divided into groups according to their BP control after 4 years of health management, and the factors that may affect control were evaluated by the χ² test. Backward elimination was used to find the influencing factors of hypertension control and a binary logistic model was established. This model was selected since BP control is more meaningful than a cure for chronic diseases such as hypertension. All statistical analyses were performed on R software (version 4.1.2) and a P value <.05 was considered statistically significant. If the area under the receiver operating characteristic (ROC) curve was greater than 0.5, the result was considered to be meaningful.

The baseline survey included 51,486 individuals, 42,347 (82.25%) of whom were healthy and 9139 (17.75%) of whom had hypertension (new onset+current onset). Among the healthy population, there was a slight majority of men. Among the patients with hypertension, there was a large majority of men (>80%). The average age of the patients with hypertension was higher than that of healthy participants. In addition, the intake of milk, fruits, and vegetables in the population at baseline was deemed to be seriously insufficient (Table 1).

Of all subjects screened, a total of 8567 individuals screened in 2013, 2015, and 2017 were deemed to be eligible for inclusion in the study. After 4 years of health management, the smoking cessation rate increased significantly (P<.001), the amount of exercise increased significantly (P<.001), and other health-related behaviors also improved to varying degrees. The improvement of various behaviors showed a trend of strengthening with the increase of health management time. In terms of clinical indicators, low-density lipoprotein cholesterol levels of the participants increased (P=.005) and high-density lipoprotein cholesterol levels decreased (P=.007), although the change of indicators was small, which may be related to their age. There was no significant difference in other clinical indices measured over time. The awareness rate of hypertension among eligible participants in 2013 was 26.48% (Table 2). After 4 years of internet-based health management, the awareness rate of hypertension among this population increased significantly to 30.82%. The control rate of hypertension showed the same increasing trend. In 2013, the control rate of patients with hypertension was only 6.13%, which significantly increased to 22.61% after 4 years of health management.

In the physical examination of 2013, 1008 new patients with hypertension were found. After 4 years of health management, their smoking cessation rate increased significantly (P=.03), the amount of exercise increased slightly (P=.08), and other health-related behaviors also improved to varying degrees. The improvement of various behaviors also showed the trend of strengthening with the increase of health management time. In terms of clinical indicators, the DBP decreased considerably (P<.001) and the SBP decreased slightly (P=.13). However, there was no significant change detected in other clinical indices. The improvement of various indicators showed a trend of continuous improvement with an increase in health management time (Table 3).

After 4 years of health management in the patients with new-onset hypertension (N=1008), 195 (19.35%) patients had good BP control and 813 (80.65%) patients had poor BP control. Statistical analysis showed that younger patients had better control than older patients (P=.05) and women achieved better control than men (P=.03). In addition, those who changed their intake of cereals and potatoes, or fish, poultry, meat, and eggs according to the guidance of health management achieved better BP control (P<.001). As shown in Table 4, in comparison with patients who did not follow health management guidance, those who changed their milk and dairy intake or fruit intake according to health management guidance achieved better BP control (P<.001). Improving physical activity was also found to be beneficial for BP control in this population (P<.001). Moreover, reducing sedentary time emerged as a favorable factor in reducing BP (P=.01). Those who improved their mental state according to the guidance of health management also had better BP control (P=.004). There were no significant differences in the remaining indicators according to good or poor BP control at the end of the study period (Table 4).

To clarify the influencing factors of hypertension control, we performed regression analysis. Binary logistic regression analysis showed a significant correlation between health-related behavior and the ability to achieve BP control in patients with hypertension. Achieving BP control was more difficult for patients who did not follow the guidance of health management to change fish, livestock, poultry, meat, and egg intake; fruit intake; increase physical activity; and improve their psychological state. The area under the ROC curve was 0.6787 (Table 5).

Long-term internet-based health management has a good effect on the BP control of patients with hypertension, especially on regulating SBP. Moreover, this control effect is long-lasting and not easy to rebound. In addition, this study found that being young; female; more likely to follow the guidance of health management to control BMI within a reasonable range; adjust intake levels of cereals and potatoes, fish, eggs, poultry, livestock, milk, and fruit; and appropriately increase physical activity have a significant effect on the control of BP in patients with hypertension.

After 4 years of health management, 195 (19.35%) of the 1008 patients with hypertension had achieved good BP control. In 2018, the BP control rate of Chinese patients with hypertension (age≥18 years) was reported to be much lower at only 11.0% [22], demonstrating a potential benefit of health management. McManus et al [23] also showed that internet-based health management has a good effect on BP control in patients with hypertension. In addition, Hu et al [13] showed that face-to-face health management can effectively help patients to control their BP during the implementation of management, but rebound readily occurs after the end of management. However, due to the long-term nature of internet-based health management, this rebound can be effectively avoided.

This study found that changing dietary intake, appropriately increasing physical activity, adjusting the mental state, and controlling BMI according to health management guidance can effectively increase the control rate of BP in patients with hypertension. In addition, the rate of BP control was higher in female patients and in younger patients. Atik et al [24] and Li et al [25] showed that a lower BMI helps to control BP in patients with hypertension. Shim et al [26] and Duchame-Smith et al [27] pointed out that diet may be an important factor in BP control in patients with hypertension. Some bioactive ingredients may explain the cardioprotective effects of diet, such as vitamins, essential elements, dietary fiber, and phytochemicals [28-31]. The potential mechanisms of action may involve antioxidation; anti-inflammation; and regulation of blood sugar, blood lipids, and BP [32,33]. These previous results are consistent with the findings of this study.

Strengths of our study include the consistency of the findings in this study and other cohort studies, together with the fact that internet-based health management has strong universality. Moreover, we combined logistic regression to analyze the related factors of BP control in patients with hypertension and also analyzed dietary components in more detail than in previous studies.

However, some limitations should be noted. To date, this research has been carried out over a relatively short time and therefore the longer-term effects are unknown; however, our research will continue to clarify the effect of internet-based health management on BP control in patients with hypertension. In addition, our study population was only recruited from five provinces in northwest China, which is still small compared with the whole country. However, the incidence rate of hypertension in these provinces represents the medium level for nationwide statistics [34], which can better represent the whole country. In the future, we will try our best to implement internet-based health management across the country. In addition, we did not assess changes of treatment rate. Although treatment rate is very important in chronic disease research, this was beyond the focus of this study. Moreover, many environmental factors can influence BP measurement [35,36]. However, in this study, all participants were sitting in a quiet environment to measure the BP of the left arm (the measurement position was flush with the heart), and the difference in BP caused by measurement error between participants was small. Our clinical parameters were self-reported by the subjects according to the physical examination in that year. Although this may cause some bias, the parameters we collected are relatively conventional and therefore the bias is not expected to be substantial. Another limitation is that our study only focused on the influence of dietary behavior factors on BP, without considering drug factors. Finally, we have not evaluated some new promising biomarkers of cardiovascular risk prediction and atherosclerosis, such as high-sensitivity C-reactive protein, fibrinogen, matrix metalloproteinases, and myeloperoxidase [37].

Our findings confirm that internet-based health management has a significant effect on BP control in patients with hypertension, which can be maintained over the long-term. However, at present, the attention to internet-based health management remains insufficient, and there is still a long way to go to comprehensively popularize this management strategy.