A systematic review is conducted to identify effective interventions that improved adherence to antihypertensive drugs among patients with coronary heart diseases (CHDs). Primary studies designed to measure interventions to improve adherence on antihypertensive drugs in patients with CHD were included. Three online databases, COCHRANE, EMBASE and MEDLINE, were searched for primary studies published in English from 2005 to 2019. Studies were screened independently for eligibility. Cochrane risk-of-bias tool and the Newcastle-Ottawa Scale were used for quality assessment of randomised controlled trials (RCTs) and non-randomised studies, respectively. Of the 2000 entries identified, seven articles, including one cross-sectional study and six RCTs, met the inclusion criteria and were reviewed. One of the articles evaluated two interventions, so eight interventions were evaluated in total. Quality of the included studies was generally high, with the cross-sectional study rated as having good quality under Newcastle-Ottawa Scale, while four and two RCTs were rated as having low and some risk of bias under Cochrane risk-of-bias tool, respectively. Six of the intervention programmes were considered effective. An intervention was considered effective if it is associated with a significant (p≤0.05) and non-trivial (Cohen’s d≥0.2) improvement in compliance-related outcomes such as in terms of the Morisky 8-item Medication Adherence Scale. Medication education, disease education, health education, constant reminders and medications dispensed using blister packs were identified to be effective in improving patients’ compliance to medications. The importance of the continuity of interventions was also established. It is recommended that health service institutions should provide continuous education programmes, daily reminders and regular follow-ups for their patients who have CHD. It is recommended that further research ought to be carried out by using only one intervention in each trial with a standardised outcome measure, or using factorial designs, so that further cost-effectiveness evaluation of each intervention can be carried out independently, leading to the formulation of a comprehensive, optimised intervention programme for patients with CHD taking antihypertensives.
- coronary heart disease
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Cardiovascular diseases (CVDs) are a group of disorders related to heart and blood vessels. Coronary artery disease, peripheral vascular disease, deep vein thrombosis, pulmonary embolism, hypertension and stroke are the major types of CVD, with coronary heart disease (CHD) making up two-thirds of CVD mortalities.1 CVD has been the leading cause of death (followed by malignant neoplasm and respiratory disease) around the world for the past 30 years. In 2017, CVD accounted for 17.79 million deaths, equivalent to 31.8% of all deaths around the world.2 Despite a significant drop in the death rate of CVD in developed countries in recent years, CVD remains a major challenge to public healthcare systems around the world. By 2030, the number of deaths of the patients with CVD will continue to rise to 23.6 million, accounting for 23.5% of deaths globally.3
According to the WHO, medication adherence is defined as ‘the degree to which the person’s behaviour corresponds with agreed recommendations from a healthcare provider’.4 Effective management of CVD, CHD in particular, requires patient adherence to long-term medications and lifestyle modifications. Some major drugs classes for the medical treatment for CHD include antiplatelets (eg, aspirin and clopidogrel), nitrates (eg, sublingual glyceryl trinitrate (GTN)) and antihypertensives (eg, beta blockers and calcium channel antagonists). A typical first-line treatment for stable angina includes low dose aspirin, statin, sublingual GTN and a beta blocker. Hypertension is the most important modifiable risk factor for CVD, thus proper control of patients’ hypertension via good drug compliance is important in the proper management of patients’ condition.5
Despite slight improvements in medication adherence being reported in the USA,6 medication non-adherence remains a major challenge in disease management. The WHO estimates that 50% of patients do not adhere to their medication prescription,7 a systematic review with patient data from 15 countries also reported a 45.2% non-adherence to antihypertensive medication.8 Previous studies have identified a significant relationship between medication non-adherence and hospitalisation rate and mortality in patients with CVD.9–11 Medication non-adherence not only worsens patients’ health outcomes, it also increases overall medical costs due to increased inpatient hospital stay and emergency department visits and also false clinical diagnosis of refractory hypertension.12 A number of modifiable factors to medication non-adherence have been identified in previous research including forgetfulness, concern about the side effects of medications, health literacy and other practical barriers. Other non-modifiable factors include advanced age, sex and diminished functional capacities.13 Interestingly, the role of polypharmacy in non-adherence of medication is yet to be elucidated as previous studies provided inconsistent results concerning the association between number of medication and non-adherence.14 Lower adherence can also be attributed to choices of medication, adherence rate ranges from 53% to 72% across nine different classes of antihypertensive drugs.8 Adherence to prescribed medication regimen may increase pharmaceutical spending, but the medical cost of an adherent patient is offset by the decrease in utilisation of acute inpatient medical service or even surgical interventions.15 Overall, enhancing patients’ adherence to medications improves patient health outcomes and relieves the burden of local healthcare services.
Despite research regarding drug adherence of patients with CHD has been well documented globally,16 there is a lack of comprehensive comparison on the effectiveness of various types of intervention that improve adherence to antihypertensive medications in patients suffering from CHD. This systematic review aims at identifying interventions that influence decisions of the patients with CHD on their adherence to antihypertensive medications; so as to give future suggestions on methods that reduce the risks caused by patients’ non-adherence to antihypertensive drugs.
Retrieval of primary data and searching strategy
An electronic search was carried out in three online databases: MEDLINE, COCHRANE and EMBASE. Primary studies reported in English in the recent 15 years (from 2005 to 2019) were included. A searching strategy with a combination of medical subject headings and text word for keywords was used to locate potential studies. The keywords were related to (1) interventions to improve adherence (eg, ‘compliance’, ‘adherence’, ‘patient education’), (2) antihypertensive drugs (eg, calcium channel blockers, diuretics, beta-blockers, ACE inhibitors) and (3) ischaemic heart diseases (eg, ‘myocardial infarction’, ‘CABG’, ‘angioplasty’). The details regarding the items of our searching strategy are included in online supplemental appendix 1.
Inclusion and exclusion criteria for the selection of studies
Studies were retrieved from online databases. Using predefined selection criteria, the electronically searched entries were reviewed independently by four investigators (JH, WYN, HYM, WYL). Investigators were divided into two groups and a double search and selection independent of each other was carried out to further narrow down potential studies. The studies were then retrieved online and evaluated in detail. Discrepancies between investigators were resolved after discussion.
For inclusion criteria, primary trials or research published in the recent 15 years (2005–2019) in English journals were included. They must be related to ‘adherence intervention’, ‘antihypertensive drug’ and ‘ischaemic heart disease’ concurrently. The outcome measure of the trial must include a direct measurement of drug adherence (such as pill counts or using established adherence assessment scales or questionnaires).
For exclusion criteria, non-original articles, protocols, editorials, commentaries, correspondences and reviews were excluded. Journals not published in English were also excluded. Only studies published in English are included as it has been shown that the risk of introducing selection bias into the systematic review is low as the majority of studies are written in English.17 Studies conducted prior to 2005 were excluded as the WHO has updated its definition of CVD and its public health burden in 2004.18
Our systematic review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.19 Figure 1 summarises the screening and selection processes in the PRISMA format.
Information from each article was extracted by five investigators (YCT, PYBL, MFW, JH, KYL). Investigators were divided into two groups (a group of two and a group of three), and data extraction was carried out with two groups independent of each other, using a standardised protocol and then recorded in a pre-formulated table, and the information it contains has been presented in tables 1 and 2, figure 2 and online supplemental appendix 2 of this review. Details such as patient information, sample size, study design, selection criteria, interventions, outcome measure and quality assessment were included in the review process.
Data evaluation and quality assessment
Numerous assessment instruments are available for evaluating medication adherence, including medication possession ratio, pill count, Medication Adherence Questionnaire and 8-item Morisky Medication Adherence Scale (MMAS-8). However, there is no single measurement that can be considered as the gold standard, and different instruments have been adopted in different studies.
In order to evaluate whether an intervention is effective, regardless of the instrument used for adherence assessment, the following three criteria were used:
The compliance, regardless of the method of measurement, in the experimental group (ie, the group receiving the intervention) must be higher than that of the control.
The observation in the first criteria must be statistically significant, and p≤0.05 was used as the standard for statistical significance.
The effect size of the intervention should not be too small. Cohen’s d value has been used to evaluate this criterion, and by convention, any intervention with d value <0.2 is considered as ‘trivial’ and should not be considered effective. For interventions with d values ≥0.2, all of them are considered to meet this criterion successfully and would be further classified into having small, medium, large, very large and huge effects with cut-offs of 0.2, 0.5, 0.8, 1.2 and 2.0, respectively,20 to facilitate further analysis. If the d value of an intervention cannot be calculated directly, it will be inferred from its OR, and the formula of ‘d value=ln(OR)×(√3/π)’ would be used.21
An intervention will be regarded as effective only if all of the three criteria are met.
Quality assessment was carried out in the form of risk of bias analysis. Version 2 of the Cochrane risk-of-bias tool (RoB2)22 was used for randomised controlled trials (RCTs) and the Newcastle-Ottawa Scale23 was used for non-randomised studies. Covering different domains of a clinical trial including study design, conduct and reporting, RoB2 was used to assess the risk of bias in RCTs. The risk of bias was appraised by the RoB2 algorithm after answering the required signalling questions as listed.
A total of 2000 entries were identified from the database with the aforementioned searching strategy. Fifty papers were chosen after screening and seven articles were chosen for systematic review. Out of the seven articles, six studies were RCTs while one study was a cross-sectional study.
Study design and patient characteristics
Table 1 summarises the characteristics of each of the selected studies. The median sample size in the seven studies was 806 patients and the range was 159–1757. Two trials were small-scale studies (<200 patients). A total of 5550 patients with multiple ethnicities including African American, White, Spanish, Chinese, Indian and Norwegian were recruited. Only adults were recruited in all studies, and the study conducted by Lee et al focused on elderly patients with a mean age of 78 (SD 8.3).24 Of all studies, the mean age of patients is 63.8 with the majority of male participants. All studies included patients experiencing cardiovascular events including CHD and acute myocardial infarction or had at least one risk factor of cardiovascular events such as hypertension. Six studies recruited patients during hospitalisation and one study recruited patients in the community.
Overview of intervention types and outcomes
Variation of interventions existed among trials. Five studies examined the impact of multifaceted patient education on medication adherence.24–28 Four of these studies involved clinical healthcare professionals while one study recruiting trained health workers to provide education sessions. For their design of the education programme, individualised medication and health education are the cornerstone of most of the included studies focusing on education.24 26 27 Two studies emphasise the collaboration of pharmacists and patients in improving patient’s medication adherence with personalised medication education24 27 while one study explored the role of physician and patients’ family members in enhancing medication adherence with postdischarge educational intervention. Two studies involved nurses or community healthcare workers in providing intensive health education to patients, for example, weekly lectures on preventing possible complications after discharge. One trial studied the utilisation of advanced medication reminder application with interactive function in increasing a patient’s medication adherence,29 the use of reminder is also the part of pharmacists’ intervention programme using blister pack. One trial suggested novel ideas to enhance patients’ drug compliance, by comparing comprehensive cardiac rehabilitation programmes and exercise-based cardiac rehabilitation programmes.30
Four of the selected articles adopted MMAS-8 and all demonstrated positive outcomes with 7.5%–24.5% increase in the number of adherent patients.
Three studies assessed the patients’ compliance by self-reported medication counts, yet the exact calculation varied among the three studies. Data from two of the trials demonstrated that their interventions could enhance patients’ adherence by 5%–26.4%. However, the study conducted by Muñiz et al showed no statistically significant differences that could be identified between the intervention and control groups for antihypertensives including beta blockers and angiotensin receptor blockers.26
Interventions performed and their outcomes in details
A variety of interventions were implemented in different studies, and most studies used an intervention ‘package’ that consists of several items. Most intervention packages were found to be effective in improving drug compliance, while some did not show any statistically significant improvements. In table 2, all the items within the intervention packages of each study have been listed. The outcomes of these packages have also been included, together with their statistical significance (p values) and effect sizes (Cohen’s d values). A conclusion regarding whether each of the packages is effective has been drawn based on the three criteria stated in the Methods section and stated in table 2.
Among the total of eight interventions evaluated from seven studies, six interventions (in studies by Lee, Xu, Du, Santo, Xavier and Peersen) have been considered as effective based on the three criteria listed in the Methods section. Three, one, one and one studies were found to have small, medium, large and huge effect sizes, respectively, by Cohen’s d value. Two interventions (in studies by Muniz and Peersen) have been considered ineffective as they were statistically insignificant and had trivial effect sizes.
The quality of all the selected articles was evaluated by either RoB2 or Newcastle-Ottawa Scale as mentioned above. Quality scores of the selected studies were generally high. One observational study was rated 8/9 in Newcastle Ottawa Scale. Four of the randomised trials were considered having low risk of bias using RoB2. However, two randomised trials conducted by Muñiz et al and Xu et al, respectively, were shown to contain some risk of bias in terms of randomisation process, deviation from intended intervention, and measurement of outcome. Figure 2 shows the risk of bias of each RCT included in this review for each of the five domains of RoB2, as well as their overall judgement. The details of our evaluation of each RCT for each RoB2 domain can be found in online supplemental appendix 3, in which the answers to each signalling question of RoB2 are included.
This systematic review mainly found out four major types of interventions that could improve the drug compliance of patients with CHD taking antihypertensives. They include (1) continuous education programmes that involve medication education, disease education and health education; (2) daily reminders, which can either be digital or traditional; (3) regular follow-ups and (4) dispensing medications using blister packs. The following analysis would explain this principal finding.
It was likely that education could improve compliance. However, its implementations had substantial impacts on its effectiveness. Education was included in six intervention programmes among five of the studies,24–27 30 but some of these interventions improved compliance while some did not. Four of the educational intervention programmes showed promising results in improving compliance,24 25 27 30 while the other two were ineffective.26 30 After observing the ways in which education was carried out, this finding could be well explained. For successful plans, education was offered regularly for multiple times throughout the whole disease management process. In a study with a successful educational programme, education took place during every follow-up visit, that is, every 2 months.24 Another successful programme had education offered to patients with low adherence during the once-a-month pharmacist follow-up.27 Another one had education offered even more frequently as patients took part in weekly lectures.25 The details of the education programme in the last study were, however, not specified.30 On the other hand, in ineffective plans, education was given on a one-off basis that lacks continuity. One unsuccessful education intervention had information material given once only during the interview 2 weeks after discharge.26 Another one had education offered only during the one day ‘heart school’.30 Summarising from these, we could observe that a regular and continuous approach to patient education was more effective than education that was offered only once. We hypothesised that the frequency of education sessions would increase medication adherence, and further exploration on this issue warrant more testing by RCTs. It is worth noting that most research studies are backed by research funds for recruiting extra personnel and resources to provide patient education. For example, nurses, physicians, pharmacists and community health workers are invited to work with patients in the intervention group. Even if we saw a positive impact of patient education on medication adherence, whether those healthcare professionals will devote their time and efforts to continue those practices outside the research context is doubtful. Also, most of the studies did not provide the cost of the intervention programme, making it difficult to compare the cost-effectiveness of different intervention designs.
Organising from the effective interventions, education was a big item which had three dimensions: (1) medication education, which was about the drug itself, including treatment objective, mechanism of action, route and frequency of administration, as well as adverse effects24 25 27; (2) disease education, which was about the aetiology, pathogenesis, clinical manifestations, diagnosis, complications, treatment and prevention of the disease suffered by the patient25 27 and (3) health education, which involved appropriate lifestyle management (eg, exercising and diet) for improving general health and slowing down disease progression.25 27 Medication education increased the clarity of instructions such that non-compliance caused by misunderstandings and uncertainty could be minimised. Disease education and health education might increase patients’ health awareness, and thus increase their motivation to carry out health promoting behaviours, which includes compliance to drug regimens. In line with a review by Nieuwlaat et al, RCTs investigating medication adherence often offer a large variety of complex interventions, assessed by multiple outcomes with inaccurate adherence measure.31 Owing to the heterogeneity of the studies, and the fact that some studies did not clearly specify the specific elements included in their education programmes, it was impossible to delineate and compare the individual effects and contributions by each of the education dimensions.
It was shown that a constant reminder for taking drugs was important for improving compliance. It could either be in the form of a mobile application,29 or as a drug calendar displayed at home.28 Both of the two studies which used this method resulted in successful improvement of compliance.28 29 One study also showed that patient interaction with the reminding tool did not show significant changes to compliance.29 Thus, it was concluded that the way by which the reminder was given was not important, be it digital or traditional, interactive or not. Simply having a constant reminder by itself was already very sufficient for improving compliance.
Regular follow-up visits or contacts may help improve compliance. In general, the total number and frequency of follow-up visits were higher for studies with effective intervention programmes. Among the six successful intervention programmes, the total number of follow-up visits done for four of them were 3,24 24,27 625 and 6,30 respectively, giving an average of 9.75 for studies with effective interventions (two programmes have been excluded as the follow-up visit numbers are not available).29 30 The frequency of follow-up visits in the four studies mentioned above was also decent, being every 2 months,24 every 1 month,27 on average every 2.4 months during first year,25 and every 2 months,28 respectively, giving an average of 7.25 times per year, that is, once every 1.66 months. Conversely, for the two interventions that were ineffective, one of them only had one follow-up visit (2 months after discharge),26 showing significant inferiority compared with the successful intervention programmes; while information of follow-up visits was not available for the other ineffective intervention.30 Hence, adequate number of frequent follow-up visits could be helpful in improving compliance.
In addition, the importance of continuity of the interventions was established from two studies.24 27 In one study,24 if the intervention was discontinued after 6 months, the patients’ compliance rate would drop significantly, and six more months after the discontinuation, it would reach a level not much higher than the initial baseline level before any intervention begins. Conversely, a high compliance was maintained if the interventions were continued. Thus, it was essential for the interventions to be carried out continuously during the whole course of management. In another study,27 it was observed that the benefits of the interventions become more significant as the interventions were continued for a longer period of time. This was observed from the fact that extending the intervention period from 6 months to 2 years resulted in both an increased magnitude of compliance improvement, as well as a larger statistical significance regarding the effectiveness of the intervention.
Analysing the effect sizes of the interventions, the interventions by Lee et al and Xu et al stood out as they got ‘huge’ and ‘large’ effect sizes, respectively, according to their Cohen’s d values, as opposed to the other four effective interventions with lower effect sizes (three had ‘small’ effect size and one had marginally ‘medium’ effect size).24 27 The ‘large’ effect size of Xu’s intervention can be explained by the points listed above, as it is a comprehensive package containing continuous coverage of all educational dimensions: medication education, disease education and health education. It also has the highest frequency (once a month) and number (24) of follow-ups.27 On the contrary, the ‘huge’ effect size of Lee’s intervention has drawn our attention since the package only contained three items.24 The education item only included the dimension of medication education, while the follow-up item was also not too strong as there were only three follow-up visits. The only item that could be used to explain its ‘huge’ effect size would be dispensing medications using adherence aids such as blister packs, and this item was unique to this package and did not appear in any of the other seven intervention packages evaluated in this systematic review. In Lee’s study, the ‘adherence aid’ was described to be some blister packs with medications custom-packaged into them based on the patients’ pharmaceutical regimen, which means that each blister pack contained all the pills that the patient needed to take in a daily dosing time.24 To sum up the findings from effect size evaluation, it was likely that packaging medications in blister packs can substantially improve medication adherence.
There are several limitations in this study, mainly caused by the lack of studies related to our topic, as well as the heterogeneity between different studies, regarding the intervention combinations, study methods and the ways of assessing outcomes and so on.
First of all, only a few studies investigating cardiovascular drugs included a direct measurement of compliance (no matter in terms of percentage of pills taken or using established scales). In contrast, most studies evaluated drug effectiveness and compliance-improvement interventions using other outcome measures, which were mainly biochemical, such as changes in blood pressure or blood lipid levels. Since these levels could be contributed by many factors other than compliance (including the effectiveness of the drug by itself, patients’ lifestyle modifications, as well as the unique disease progression of each individual patient), studies using these outcome measures that could not indicate compliance directly and reliably were excluded. As a result, only seven studies met our inclusion criteria in terms of outcome measure, which might hamper the comprehensiveness and informativeness of our review.
Second, the heterogeneity between various studies should be noted. For instance, when measuring compliance, some of the studies used the Morisky Medication Adherence Scale, while others used pill counts and so on. These methods were not standardised, and the definitions of ‘having high compliance’ were different, making parallel comparison between different studies difficult. Besides, the intrinsic variations between the studies in terms of age group, the underlying CVD, as well as the drugs taken, also made comparison difficult and reduced the clinical utility of this review.
Third, many of the studies used a predefined package of interventions containing several intervention items, instead of using single interventions. For instance, in one study, patients in the intervention group were given a ‘direct pharmacist intervention’, which in fact was a package that contains seven distinct intervention items.27 When an intervention package was proven effective, there was no clue whether all of the items, or only some of them, contributed to its success. Any possible synergistic effects between the intervention items were also unknown. Besides, since the packages provided in various studies overlapped in some elements but were not fully identical, together with the different durations of intervention and other areas of heterogeneity, it would be rather impossible to delineate the effectiveness of individual intervention items for the formation of any ‘optimal’ management plan.
Fourth, for intervention items other than the four mentioned in the principal finding (ie, continuous education, daily reminders, regular follow-up and dispensing medications using blister packs), most of them were only used in one single study, making comparison and confirmation more difficult. In most instances, they were also used together with other intervention items, such that the individual effects of these intervention items would be hard to delineate. Thus, it might be rather risky and unreliable to draw conclusions on them based on the limited information.
Conclusion and further recommendations
Our systematic review recommended that health service institutions should provide continuous education programmes, daily reminders, medications dispensed using blister packs and regular follow-ups for their patients who have CHD, so that their drug compliance, and by extension their health outcomes, could be improved.
There is a lack of systematic studies on how different interventions might impact patients’ compliance. Moreover, there is a lack of uniformed approach to address the change in compliance as a result of such interventions. Thus, it is recommended that further primary research regarding this topic in the future can be carried out by using a single intervention each time and measuring patient compliance using a standardised outcome measure, and this will allow delineation and comparison on the individual effects of every single intervention. Alternatively, methods such as factorial designs, despite not commonly used in clinical research, have emerged and been proven to be feasible and efficient for studying clinical interventions containing multiple components.32 It is possible to analyse the individual effects of each item within each intervention package by this method, but a well-established protocol regarding the classification and characterisation of each intervention component has to be made beforehand. After knowing the individual effects of each intervention item by either of the two approaches, cost-effectiveness evaluation can be carried out for the formation of a comprehensive, optimised intervention programme for patients with CHD taking antihypertensives.
Continuous education programs, which include elements of medication education, disease education and health education, are helpful for increasing compliance.
Constant reminders, regular follow-ups and dispensing medications using blister packs are helpful for increasing compliance.
Further research should be carried out by using only one intervention in each trial with a standardised outcome measure, so that further cost-effectiveness evaluation of each intervention can be carried out independently.
Current research questions
What should be included in medication-related patient education?
What is the most effective way of reminding patients to take medicine?
How to create a comprehensive postdischarge management programme for patients with coronary heart disease?
Lee JK, Grace KA, Taylor AJ. Effect of a pharmacy care program on medication adherence and persistence, blood pressure, and low-density lipoprotein cholesterol: a randomized controlled trial. JAMA. 2006;296(21):2563-71.
Xu N, Wang C, Wan J, Liu X, Li Z, Chen M. Effectiveness of pharmacist intervention in the management of coronary artery disease after index percutaneous coronary intervention: a single centre randomized controlled trial. Int J Clin Exp Med. 2019;12(6):7760-5.
Du L, Dong P, Jia J, et al. Impacts of intensive follow-up on the long-term prognosis of percutaneous coronary intervention in acute coronary syndrome patients – a single centre prospective randomized controlled study in a Chinese population. Eur J Cardiovasc Prev Rehabil. 2016;23(10):1077-85.
Xavier D, Gupta R, Kamath D, et al. Community health worker-based intervention for adherence to drugs and lifestyle change after acute coronary syndrome: a multicentre, open, randomised controlled trial. Lancet Diabetes Endocrinol. 2016;4(3):244-53.
Santo K, Singleton A, Rogers K, et al. Medication reminder apps to improve medication adherence in coronary heart disease patients (MedApp-CHD): a randomised clinical trial. Eur Heart J. 2018;39(suppl_1).
Multiple choice questions
Medication non-adherence can be caused by forgetfulness, side effects of medications and poor health literacy.
Intervention programs used in most selected studies involve a package consisting of more than one intervention items.
All selected studies used the 8-item Morisky Mediacom Adherence Scale (MMAS-8) for outcome measure.
Within the three dimensions of education aiming to improve compliance, medication education is confirmed to be more important than disease education and health education.
To improve compliance, an interactive reminding tool is superior to one that is non-interactive.
Patient consent for publication
The authors would like to thank the School of Public Health of the University of Hong Kong for its Health Research Project.
JH, YCT and PYBL contributed equally.
Contributors JH, YCT and PYBL contributed equally as first author. The authors of this research contributed equally to the study design. JH, PYBL and DC contributed equally to formulation of searching strategy. JH, WYN, HYM and WYL contributed equally to eligibility screening of data entries. JH, YCT, PYBL, MFW, WYN and KYL contributed to data extraction from eligible studies and the compilation of them. YCT contributed to data interpretation. JH, YCT, PYBL and MFW wrote the manuscript, and the authors of this research are involved in the revision of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.