Introduction

Hypertension (HT) is the leading preventable cause of cardiovascular disease and death globally [1]. In Hong Kong, about 30% of the adult population has hypertension, and majority (> 80%) are managed in government-funded general outpatient clinics (GOPCs) in the primary healthcare system [2]. Patients with optimal HT control also attend the GOPCs every 3–4 months for blood pressure (BP) measurement and refill of medications, which places an enormous burden on the healthcare system in Hong Kong. To better allocate healthcare resources, telemedicine systems have the potential to reduce the number of doctors’ consultations by automatically confirming optimal disease control (i.e., BP control in patients with HT) and refilling medications. Well-implemented telemedicine systems can help healthcare systems cope with larger patient volumes and facilitate better allocation of resources [3].

The use of telemedicine has dramatically increased in the last few years due to the 2019 coronavirus disease (COVID-19) pandemic and is likely to remain the main channel for patients’ encounter worldwide [4]. Furthermore, transferring home BP monitoring (HBPM) data and intensifying treatments by telemedicine to the physician-in-charge can improve systolic BP (SBP) control by approximately 4 mmHg because it also encourages self-monitoring and self-management. Additionally, it promotes behavioral changes, treatment compliance, and self-efficacy [3, 5,6,7,8].

Although telemedicine typically represents more intensive treatments and has been examined in patients with suboptimal control of HT in previous randomized controlled trials (RCTs), there is currently a relative lack of studies examining the safety, acceptability, and feasibility of using automatic telemedicine systems to reduce physician consultations [5, 9, 10]. Furthermore, telemedicine was not found to be cost saving for patients with HT in a recent review [11]. A telemedicine app and system (‘HealthCap’) has been developed to record HBPM, provide automatic feedback in response to different BP levels, transfer BP data to the case physicians, and confirm optimal BP control on HBPM (details under methods). We hypothesized that when optimal control of BP is confirmed on HealthCap, the index physician’s consultation can be safely deferred, and medications can be automatically prescribed.

Although we believe HealthCap can reduce consultations with the physician, be timesaving for patients, and allow better resource allocation, a pilot RCT is needed because the feasibility for a full-scale RCT and patients’ acceptability using HealthCap are not known, and telemedicine is currently understudied, especially with telemedicine rarely being practiced in Southeast Asia. A systematic review in 2018 that explored barriers to implementation of telemedicine did not find any Southeast Asia or Chinese studies [12]. The feasibility of implementing telemedicine systems differs geographically due to different cultures, patients’ demographics (age, education, and level of e-health literacy), and expectations [12]. In Hong Kong, implementation could be difficult because patients with HT are old and less educated, which are known barriers [10].

For the primary outcome, we hypothesized that an RCT examining HealthCap was feasible (by examining recruitment and retention rates) and acceptable. For secondary outcomes, we hypothesized that reducing physicians’ consultations with HealthCap would be safe. Furthermore, we hypothesized that patients randomized to HealthCap would have similar BP control (primary outcome in future definitive RCT), have a reduced number of doctors’ consultations, and have non-inferior self-efficacy and treatment compliance when compared with usual care at 6 months. Participants randomized to HealthCap were interviewed to understand their experiences, including any difficulties.

Methods

This pilot RCT was approved by the joint CUHK-NTEC Clinical Research Ethics Committee (Ref no.: 2020.294) and registered in the clinical trial registry (ClinicalTrials.gov no.: NCT04542564). Participants were randomized in a 1:1 allocation ratio to receive telemedicine (HealthCap) or usual care.

Intervention: HealthCap

HealthCap was developed by an expert team in primary care and with patients’ feedback. It is commercially available in English and Chinese on the Apple and Google platforms. It prompts regular HBPM (every 1–2 weeks), records patients’ HBPM readings, formulates the mean BP values in the past 7 and 30 days, and provides automatic feedback to patients. For example, patients are diagnosed with elevated BP when their mean SBP or diastolic BP are consistently 136–179 mmHg and 86–119 mmHg, respectively. These patients are encouraged to monitor BP frequently and to book an early medical appointment if elevated BP persists. For very high BP (BP ≥ 180/120 mmHg), HealthCap advises double-checking and emergency treatments. These threshold levels are in accordance with local and international recommendations [13]. Furthermore, BP readings and their mean values are automatically transferred to the physician’s office.

In the current project, HealthCap reminded patients to take dual BP readings both in the morning and evening for 1 week prior to the index consultation (Appendix 1). These data were then sent to a password protected computer in the respective clinic and the physician-in-charge retrieved BP data 1–2 days before the index consultation. This measurement algorithm was used in previous studies and was the only one included in the Hong Kong Primary Care guidelines [14]. If the 7-day HBPM mean was optimal (i.e., < 135/85 mmHg), other parameters were checked by an online questionnaire and were confirmed by the physicians, these included: (i) self-reported good drug compliance and no drug side effects, (ii) absence of symptoms of complications (i.e., chest pain and hemiplegia), and (iii) no other health complaints that needed a consultation. If the answers were all negative, medications were prepared in the clinic, and the index physician appointment was deferred for three months. However, if any of the answers were positive or 7-day HBPM mean was suboptimal (i.e., ≥ 135/85mmHg), the patient was asked to consult the doctor in-person as planned. The doctor would assess the patient and was advised to step up the medications according to the Hong Kong Primary Care guidelines [14], as appropriate.

Usual Care

In Hong Kong, patients with well-controlled HT are routinely seen every 3–4 months. Participants in usual care who used to conduct HBPM were not excluded from the RCT because HBPM is beneficial, and it was unethical to withhold HBPM from patients. However, no extra training or advice was provided. They were also asked not to download or use any HT mobile app during the study period.

For any other health complaint, patients had unlimited access to private clinics, GOPCs, and emergency departments. These were not limited to the participants in either group of the trial.

Participants

The participants were recruited from two large GOPCs (Lek Yuen and Fanling Clinics) in Hong Kong. They were patients receiving antihypertensive medications with well-controlled HT defined as BP < 135/85 mmHg on out-of-office BP measurements (ambulatory blood pressure monitoring (ABPM) or HBPM) [13]. Out-of-office measurements are more reproducible and predictive to cardiovascular outcomes than office BP [15, 16]. Patients were excluded if they had severe mental illnesses that impaired their ability to use HealthCap and if they required regular face-to-face consultations (e.g., other diseases such as asthma or diabetes). As a pilot trial, we aimed to recruit 50 patients, similar to other pilot trials [17]. The detailed inclusion and exclusion criteria are presented in Appendix 2.

Randomization

Stratified randomization according to age (≥ 65 years or < 65 years) and education level (primary school or below versus secondary education or above), with blocks of four or six was used to achieve a 1:1 allocation ratio. The aim was to achieve similarity with respect to sociodemographic characteristics between the two groups because age and educational levels are important barriers to telemedicine implementation [11]. The randomization sequence was generated by an independent statistician and sealed in light-opaque envelopes. The envelopes were opened only after confirmation of eligibility and a consent form was signed. Although the nature of the intervention prohibited blinding of the doctors and patients after allocation (an inherent difficulty shared by app-based intervention RCTs), the outcome assessor(s) and statistician were blinded to the allocation [18].

Outcome Measures

As a pilot RCT, the primary outcomes included the rates of recruitment and retention. Good feasibility was defined as the rate of recruitment of 50 patients within 6 months by two recruiting doctors and a retention rate of ≥ 80%. Acceptability was further assessed by interviewing 21 patients in the telemedicine group at 6 months (end of the study).

Appendix 3 provides a detailed description of the secondary outcomes. In summary, the secondary outcomes included daytime/night-time/24-hour BP parameters from ABPM, office BP, health service utilization (e.g., number of visits to the emergency department and GOPC), number and types of anti-HT medications, body mass index, serum creatinine, fasting glucose and lipid levels, self-efficacy, medication and diet adherence, exercise level, and health and e-health literacy (see Appendix 3 for the validated tools used to assess these outcomes). The validated device PhysioQuant (Envitec, Belgium) was used to measure ABPM, following international guidelines and using a fixed period to define daytime (6am to 10pm) and night-time (10pm to 6am) [26]. BP was measured every 30 min during daytime and 60 min during night-time and the readings were considered valid if there was > 70% of valid readings overall, > 20 valid awake and > 7 valid asleep BP readings in 24-hour intervals; ABPM was performed on a routine working day and on the non-dominant arm [26]. All outcomes were collected at recruitment and at 6 months. Sociodemographic information including age, sex, marital status, educational level, smoking status, and alcohol use (assessed by the validated AUDIT questionnaire) were collected on recruitment [19].

Statistical Analysis

Baseline characteristics were compared using two-sample t-tests for continuous variables and the chi-square test for categorical variables. The recruitment and retention rates were calculated. Data from validated scales were regarded as continuous outcomes if the data were normally distributed. Data that were not normally distributed were categorized. Continuous variables were examined with analysis of variance, and the effects of telemedicine on ABPM were examined using analysis of covariance, with BP as the baseline and treatment group as the covariate, following the intention-to-treat principle. Categorical secondary outcomes were analyzed using the Fisher exact test or the Chi-square test as appropriate. Subgroup analysis was conducted for baseline characteristics that were unbalanced between the two groups.

Data from the interviews were analyzed using thematic analysis, and the interviews were audio-recorded, transcribed, and thematically coded. The coding was independently cross-checked for reliability by two researchers (SW/EKPL).

Results

Participants’ Characteristics and Rate of Recruitment/Retention (Primary Outcome)

The mean age (± SD) of our participants was 59.9 (± 8.4) years. Most of the participants were women (53.1%), had an education level lower than high school (85.7%), had a household income of < HKD$ 20,000 (~ USD$ 2,565) (64.5%), and were non-smokers (79.6%). The baseline characteristics were similar between both groups, except that more female patients were recruited in the telemedicine group (70% versus 36%) (P = 0.015) (Table 1).

Table 1 Baseline characteristics of the telemedicine and usual care groups
Full size table

Forty-nine participants were recruited within 6 months; one participant did not show up on the recruitment days. The rate of recruitment was thus about eight participants per month. Only one participant dropped out immediately after recruitment because of family objection (Fig. 1). The retention rate was 98%, with 55% (27/49 participants) of participants receiving ABPM at the study endpoint. No adverse event was observed in either group.

Fig. 1
figure 1

Consort diagram

Full size image

Other Outcomes

Participants in the telemedicine group had fewer visits to GOPCs (0.8 versus 2 consultations; P < 0.001) and had higher low-density lipoprotein (LDL) (3.5 versus 3.0 mmol/L; P = 0.037) (Table 2) compared to those in the usual group. Participants in both groups had similar BP on ABPM (daytime SBP: 128.2 mmHg [telemedicine group] versus 126.9 mmHg [usual care group], p = 0.41) and office SBP (125 mmHg [telemedicine group] versus 129.8 mmHg [usual care group], p = 0.089), e-health literacy, self-efficacy, drug use, other healthcare utilization and medication, exercise, and diet adherence (Tables 2 and 3). Subgroup analyses yielded similar results (see Appendix 4).

Table 2 Outcomes about GOPC utilization, blood pressure, medication change and physical test in intervention and usual care groups at follow-up
Full size table
Table 3 Outcomes from validated scales, health utilization and drug utilization in intervention and usual care groups at Follow-Up
Full size table

Patients’ Interview

Participants perceived the telemedicine system as safe, convenient, and time efficient. The telemedicine system encouraged HBPM, educated patients about the threshold levels of elevated BPs, and reassured participants when BP was normal. Some participants observed that home BP was more reliable than office BP and preferred telemedicine. Some participants also believed that using telemedicine allowed for better allocation of resources to patients in need (Appendix 5).

However, the study participants perceived that older age and forgetfulness in measuring home BP were barriers to the telemedicine system. They found that the randomization procedures were acceptable, but some participants found ABPM inconvenient and did not understand why it was needed (Appendix 5). The qualitative results were in accordance with the quantitative results, including the high retention rate (Table 4).

Table 4 Side-to-side table for qualitative and quantitative results
Full size table

Discussion

Main Findings

The telemedicine system was well accepted by patients and safe; participants from both groups had no adverse events and had similar BP and physical outcomes, except LDL levels that were higher in the telemedicine group. The difference in LDL could be explained by an outliner in the telemedicine group, whose LDL increased from 1.9 mmol/L to 4.3 mmol/L, and no significant difference between two groups was detected after excluding this outlier. The recruitment and retention rates of this pilot RCT were high. However, many patients defaulted the ABPM. Because this trial was conducted during the COVID-19 outbreak and was self-funded, ABPM could not be arranged timely at the end-of-study visits and patients paid two extra visits for the ABPM on a working day without incentives. This was unavoidable because personnel allocation was tight in the Hong Kong healthcare system during the pandemic and extra workloads (e.g., vaccination administration) were imposed on the clinics. However, we showed that patients from both groups had similar office BP at 6 months (study endpoint). Therefore, to conduct a more definitive RCT to ascertain the effectiveness, acceptability, and safety of telemedicine services, we suggest: (i) offer incentives for ABPM that can improve patient compliance, (ii) hire a research assistant to conduct ABPM at the end-of-study visit (in contrast to overburdening the existing clinic system), (iii) explain the necessity of ABPM as the most reliable reference standard of BP measurement [16], and (iv) include other BP measurement methods as outcome measures, including office BP and home BP, which are more acceptable to patients [20].

Our results are confirmed by previous studies which showed that telemedicine was well accepted by patients with HT [21]. However, while previous studies used telemedicine as an intensification of treatment and was expensive to build and maintain, the current study showed that telemedicine can represent less frequent patient visits and may potentially be cost-saving [22, 23]. Therefore, in contrast to previous studies that showed improved BP in patients with suboptimal BP, our study provided preliminary strong evidence that BP control was non-inferior in the telemedicine group in patients with optimal HT control [22].

Clinical and Research Implication

As telemedicine systems are very heterogeneous in design, every telemedicine system should be scientifically validated prior to widespread implementation in healthcare systems [23]. However, most commercially available mobile apps and telemedicine systems have not been validated [24]. Our results showed that the HealthCap telemedicine system was safe and non-inferior to usual care, although this needs to be confirmed in a full-scale, parallel-group, and non-inferiority RCT. Such a trial can have a longer study period (i.e., 1 year) and include a formal cost-effectiveness analysis. Furthermore, patients commented that they felt empowered by HealthCap (e.g., more HBPM monitoring), but our quantitative results did not find an increase in self-efficacy scores. This could be due to our limited sample size and should be investigated further.

Our results show that the telemedicine system is safe and acceptable to patients and can be implemented in healthcare systems. However, patients should be allowed to consult their doctors if suboptimal BP is detected, or new problems arise. Although telemedicine systems are generally accepted by physicians, some physicians are concerned about the additional workload [21]. The current version of HealthCap telemedicine system still requires the physician or nurse to manually confirm responses to the three safety questions (see Method). If these steps can be fully automated in the telemedicine system, this would further reduce the workload and costs.

Strength and Limitations

Our pilot trial showed that the HealthCap telemedicine system is safe and accepted by patients while maintaining adequate BP control. This is one of the first studies to show use of telemonitoring in Southeast Asian patients and may replace consultations for BP monitoring. Our RCT was pre-registered, and the randomization process was adequate, with qualitative results in agreement with the quantitative results.

This study has a few limitations. As a pilot trial aimed at investigating the feasibility of a definitive RCT, our sample size did not allow us to provide a definite conclusion [25]. However, a pilot study is recommended prior to a larger scale trial by international guidance, and the current study provides important data (e.g., retention and recruitment rates) for RCTs planning [25]. Patients’ non-adherence to ABPM may also be a barrier to interpretation of the ABPM results, although office BP in both groups was similar. Although we interviewed patients from the telemedicine group (21 of 24 patients), we did not interview the physicians-in-charge because only two physicians were involved. A definitive RCT will likely involve more physicians and physician interviews. Our RCT only included patients with well-controlled HT and excluded patients with other diseases that required regular doctor assessment. Therefore, our results had limited external validity to patients with multimorbidity, but we argue that the reduction of physicians’ contact with these patients was not appropriate. Furthermore, optimal BP control was defined as < 135/85 mmHg, but a lower target (< 130/80 mmHg) may be considered in patients with HT complications, such as chronic kidney diseases. HealthCap may further evolve to individualize BP targets. However, patients with comorbidities who require regular consultations were excluded from the current RCT. Moreover, future studies may also extend the evaluation of telemedicine services to other stakeholders, such as patient caregivers and policymakers.

Conclusion

In conclusion, the HealthCap telemedicine system is feasible. It was generally accepted by the patients and appeared safe with no reported side effects or increased healthcare utilization in the telemedicine group. Replacing physicians’ consultations with an automated telemedicine system does not appear to lead to inferior BP control. These results need to be further verified in an adequately powered RCT.