Effectiveness of Interventions To Enable Hypertension Medication Adherence In Low-And Middle-Income Countries: A Systematic Review And Meta-analysis

Background: In recent decades low- and middle-income countries (LMICs) are witnessing an increase in hypertension and thus becoming a signicant public health issue due to associated Cardiovascular disease (CVD) outcomes. Antihypertensive medication adherence is crucial to controlling blood pressure; therefore, this systematic review aimed to evaluate the effectiveness of non-pharmacological interventions on improving blood pressure control and medication adherence in patients with hypertension in LMICs. Methods: We searched the following databases for relevant literature published between Jan 2005 – Dec 2020: PubMed, EBSCOhost included Academic Search; CINAHL and MEDLINE complete; PubMed; WEB of Science; Cochrane Central Register of Controlled Trials; Cochrane Database of Systematic Reviews and Google Scholar. Cochrane risk of bias tool (RoB 2) was used to appraise included studies critically, and the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) was used to measure the quality of evidence. We conducted a meta ‐ analysis using DrSimonian-Laid's random-effect model at 95% condence intervals (CIs). The secondary outcomes of interests were synthesised descriptively as changes in BP adherence outcomes. Results: We identied 14 eligible randomised controlled trials that presented blood pressure (BP) effectiveness and medication adherence among BP patients aged between 18-75 years. The overall quality of evidence with the majority of trials was moderate. Meta weighed effect (SBP) for 12/14 studies was -4.74 (95% CI:-6.07 to -3.47) and I 2 = 57%. Out of 14 eligible studies, (86%) suggested a signicant improvement in the proportion of patients with controlled blood pressure (BP < 140/90mmHG) with a positive effect on secondary outcomes such as quality of life. Conclusion: Non-pharmacological interventions could be effective in managing hypertension. In recommending the need to investigate the feasibility of non-pharmacological evidence in specic LMIC settings, focus should be on an intervention strategy consisting of an educational intervention directed toward the patients, health professionals and organisation. Considering heterogeneity, randomised trials that are well-designed with larger sample sizes are encouraged in LMICs` to help policymakers make well-informed decisions on hypertension management. Systematic review Registration: PROSPERO registration number: controlled trials. Exclusion included studies reporting on other chronic illnesses, not reporting non-pharmacological effectiveness, and BP medication adherence outcome. keywords: hypertension or blood LMICs, adherence compliance, patient, Self-monitoring, health reminder systems, organizational, delivery Data pooled from all 14 studies showed heterogeneity across studies for systolic and diastolic blood pressure. Random effects of meta-analysis results show that systolic blood pressure was reduced by -7.61(-10.47, -4.74), with four exceptions where systolic blood pressure was slightly lower in the intervention group. Heterogeneity across studies was statistically signicant ( I 2 = 92%, P = 0.00) ( Fig. 3). The pooled mean difference for diastolic blood pressure was − 3.48(-5.52, -43). Heterogeneity for DBP was statistically signicant I 2= 89%, P = 0.00) (Fig. 4). When we excluded two studies (Alhalaiqa, 2012 and Mirniam, 2019) SBP random effect was − 4.74 (95% CI:-6.07 to -3.47) and I 2 = 57%. (Fig. 5) and DBP (-4.56–1.27) and I 2 = 78% (Fig. 6). Additional results showing funnel plots are included in the supplementary le material (Additional le 5–8).


Introduction
Hypertension is one of the leading risk factors for CVD worldwide [1][2][3][4]. Low-and middle-income countries have seen an increase in hypertension in recent decades, thus, making it a signi cant public health issue due to the associated adverse health outcomes such as the increased incidence of CVD accounting for 75% of the global deaths in LMICs (82% attributed to coronary heart disease (CHD), stroke and heart failure) [5][6][7][8]. Despite some LMICs offering universal access to hypertension medication, including Namibia, adherence to chronic medication remains a problem [6, [9][10][11]. Recent data shows that CVD in Namibia accounted for 21% of annual deaths and a reported prevalence of blood pressure (BP) among women and men aged 35-64 at between 44-57% [12,13]. Thus, it would appear that the prevalence of hypertension in Namibia is high compared to the global and African regional prevalence of 22% and 27%, respectively [14]. Non-adherence to antihypertensive medication is common and contributes to poor BP control and adverse health outcomes [15]. While it is substantial to understand the barriers and enablers of non-adherence, it is equally important to explore different interventions' effectiveness on improving BP control and BP medication adherence [15][16][17].
Non-pharmacological interventions to encourage BP medication adherence include educational interventions directed towards the patients, educational interventions directed towards the health professional, appointment reminder systems/models, and Organisational interventions [10,[18][19][20][21][22][23]. These Interventions intentionally target non-adherence contributing factors at different levels: patients, health care workers, and the health care system [24,25]. Many Higher-income countries have conducted extensive research on hypertension effective interventions [26,27]. However, we cannot say the same about LMICs, especially in sub-Saharan Africa [28].
Bene ts associated with blood pressure lowering include a reduction in many complications such as stroke (35-40%), heart attack (20-25%), and heart failure (over 50%) [22,23]. Achieving the WHO global target of a 25% relative reduction in the risk of premature mortality from NCDs by 2025 and the SDG target of a one-third reduction in premature deaths from NCDs by 2030 requires hypertension management interventions. Therefore, this systematic review aims to evaluate the effectiveness of non-pharmacological interventions on improving blood pressure control (Primary Outcome) and medication adherence (Secondary Outcome) in patients with hypertension in LMICs. We anticipate that the results from this study will provide current evidence on effective hypertension strategies targeted at controlling hypertension and CVD in LMIC settings.

Methods
A study protocol was developed prior to the conduct of this review. The protocol is registered in the PROSPERO international prospective register of systematic reviews and is accessible via the link below: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020172954. The Centre for Reviews and Dissemination (CRD) guideline for systematic reviews in health care guided this study, following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [31,32]. Completed PRISMA checklist (see Additional le 1).

Identifying the research question
The identifying research question was the effectiveness of interventions to enable Hypertension Medication Adherence in LMICs.

Eligibility criteria
A Complete inclusion and exclusion criteria form part of this review protocol, registered with (PROSPERO; CRD42020172954).
We included studies published from 2005 which met the following criteria: Hypertensive patients over 18 years of age and above, regardless of sex and ethnicity with a diagnosis of primary hypertension in a speci ed geographically de ned LMIC setting; randomised controlled trials (RCTs) of hypertensive patients which evaluated non-pharmacological effectiveness of the following interventions to improve BP medication adherence: educational interventions directed to the patient, educational interventions directed to the health professional, self-monitoring, models on appointment reminder systems, organisational interventions aimed at delivery care.
We excluded RCTs` not reporting evidence from LMIC settings published before 2005 and non-randomised controlled trials. Exclusion criteria also included studies reporting on other chronic illnesses, not reporting non-pharmacological effectiveness, and BP medication adherence outcome.
Search strategy for the identi cation of relevant studies and information sources After consulting with an information scientist at the Department of Library Services, Faculty of Health Sciences, and the University of KwaZulu-Natal on literature search approaches, search strategies were developed based on the Cochrane Handbook of Intervention Guidelines and the Cochrane Central Register of Control Trials. To nd relevant literature, we used the following keywords: hypertension or blood pressure, LMICs, adherence or compliance, control or monitoring, Educational, patient, Self-monitoring, health professional, reminder systems, organizational, delivery care. The Boolean search terms (AND OR) and MeSH terms were employed. A search of the following databases was performed in March 2019 and repeated in December 2020: PubMed, EBSCOhost included Academic Search; CINAHL and MEDLINE complete; PubMed; WEB of Science; Cochrane Central Register of Controlled Trials; Cochrane Database of Systematic Reviews and Google Scholar. Additionally, we searched Grey literature from university dissertations and theses from institutional repositories to retrieve relevant articles. We searched for randomised controlled trials conducted from January 2005 -December 2020 to capture the recent evidence on the effectiveness of hypertension interventions available in LMIC's. There was no language restrictions applied to the search.

The search management
After a rigorous, intensive search, we exported relevant articles to the Endnote X9 library. Following the import of relevant articles by title relevance, we removed duplicates before the abstract screening stage. We then conducted a selection of abstracts and full articles using eligibility criteria utilizing abstract and full title screening forms developed in Google (see Additional le 2) for identi ed articles and the date of each literature searching of each database. Independent reviewers (ON and PN) did the article screening. The third screener (SN) was involved in resolving discrepancies. Subsequently, we assessed disagreement using Cohen's Kappa coe cient (κ) statistic on Stata 13.0SE (StataCorp College Station, TX, USA (Additional le 3).

Data extraction and quality assessment
We used a data extraction tool developed in Google form based on a checklist provided in the Cochrane handbook for systematic review for intervention. The data extracted related to main study characteristics, including author, setting, type of randomised trial, intervention and control group details, and percentage of patients enrolled in intervention and control groups(see Table 1 at the end of manuscript for full characteristics of the studies). Meta-analysis data extracted included SBP/DBP changes from different non-pharmacological interventions (Additional le 4).
The methodological quality of included studies was assessed using the Cochrane risk of bias tool Current version RoB 2, which include aspects of randomisation process; deviations from intended interventions; missing outcome data; measurement of the outcome; selection of the reported result, and overall bias (reported as Low, Some concerns or High) ( Figure 2 ) [33]. Data extraction and quality assessment were conducted by two reviewers' (ON and SN), and the third author resolved discrepancies through discussions and consensus. The quality of evidence was assessed according to the domains speci ed in the grading of recommendation, assessment, development, and evaluation (GRADE) guidelines [34].

Data synthesis and analysis
The primary outcome of this study was to determine the effectiveness of interventions to enable hypertension medication adherence in Low-and Middle-Income Countries. Based on the nature of our research question, we synthesised the data by presenting changes in mean systolic blood pressure (SBP) and mean diastolic blood pressure (DBP). Systolic and diastolic blood pressure was analysed and calculated as mean difference (MD) and pooled to produce an overall random effect size between intervention and control groups. Heterogeneity among trials was conducted using the χ2 test at 95% con dence interval and I 2 statistic value. Forest and funnel plots were presented to show the results for the primary outcome The secondary outcome aim was to determine hypertension medication adherence between intervention and control groups. We synthesized the data descriptively by showing the proportion of patients with medication adherence between intervention and control groups and effect sizes for medication adherence between the two groups. Medication adherence indicators synthesized include quality of life, and knowledge, beliefs, and retention in clinical care, number of hospitalisations, and cost-effectiveness of interventions. We adopted the de nitions used by individual studies for anti-hypertension treatment and adequate control of blood pressure. [35].
Data was captured in Microsoft Excel and analysed in MetaXL version 5.3 (EpiGear International, Sunrise Beach, and Queenland, Australia).

Literature search and study characteristics
The initial database search found 162 906 potentially eligible studies, of which 161 991 were excluded during title screening. Following title screening, 915 met eligibility criteria and were included in full title screening yielding 14 de nitive studies and considered for meta-analysis (see Fig. 1 for the whole screening process). The inter-rater reliability score for full-article screening showed 61.34% agreement versus 92.00% expected by chance which constitutes a moderate agreement between the screeners (Kappa statistic = 0. 09 and p-value < 0.05). In addition, McNemar's chisquare statistic suggests that there is no statistically signi cant difference in the proportions of yes/no answers by reviewers with a p-value > 0.05 (Additional le 3).
All studies included and considered for meta-analysis are from LMICs conducted between 2005-2020 among patients diagnosed with hypertension ranging from 18 to 75 years. Additionally, most studies included parallel randomised controlled trials (n = 13) except one study, a 3arm randomised trial conducted in a low resource setting in Cape -Town, South Africa. Studies recruited patients mostly from rural settings and low resource settings (n = 7), and ve were conducted from government hospitals, including a community pharmacy (n = 1). The main reason for exclusion was non-randomised trials (n = 8), not LMIC's = 14; evaluated other chronic diseases apart from hypertension (n = 23) nonpharmacological (n = 5) and the rest were other reasons with studies which included older patients > 75 years and unclear methodology and outcome (n = 36). All studies included were published in English. Included articles all evaluated Educational intervention towards patients, however (7/14) articles evaluated more than one intervention of which it was either Educational intervention directed towards the health professional(2/14), Appointment reminder systems/models (7/14), or Organisational interventions (2/14). Appointment reminder systems encompassed principles of reminding patients when medicines were ready for collection or about scheduled clinic appointments information (2), organizational intervention aimed at delivery care-home BP monitoring and audit (2). Models on appointment reminder systems-reminder (3) box on taking medication.
The primary care providers were medical and nursing staff, community pharmacists and health workers, family support. Community interventions were delivered in community hospitals, community centres, or patients' homes (see Table 1).

Quality of included studies
Eleven (11) studies were graded to be Low risk, and three as having some concerns. All studies described randomisation methods; however, among the studies with some concerns, two had issues with the randomisation process where allocation sequence and concealment of participants enrolled and assigned to intervention was not indicated. The article with concerns of outcome measurement did not indicate whether the outcome assessment would have been in uenced by the intervention received. The rest of the studies did not indicate any concerns with deviations from intended interventions, missing outcome data, issues with measurement of the outcome, and selection of the reported results (Fig.  2). Overall, the GRADE quality of evidence was moderate (Additional le 10).

Other secondary outcome results
In addition to the proportion of controlled hypertension reported from all 14 studies, a total number of BP medication adherence indicators from the included studies were synthesized as follows: articles that evaluated the quality of life (7); level of hypertension knowledge (2); beliefs on hypertension (2), retention to care (2); and the number of hospitalisations (2). Some studies reported more than one indicator. Results reporting on the quality of life display better overall health status in the intervention group(s) than those in the control group, similarly, with knowledge and beliefs. An increase in the knowledge level in the intervention groups was reported to modify patients' beliefs about medicines positively. Beliefs about medication in the intervention group have moved away from the view that medications are intrinsically harmful, towards one that recognises the bene ts of medication. Retention in care, as well as hospital admissions, also improved. There was no signi cant difference in mean health care costs per patient between groups.

Discussion
Our systematic review ndings reveal that Non-pharmacological interventions are effective in controlling blood pressure. We included studies from the following LMICs: Jordan, South Africa low resource settings, Turkey, India, Argentina, Bangladesh, Pakistan, Sri Lanka, Iran-Isfahan, Pakistan, China, and South Korea. We observed a statistically signi cant reduction in SBP and DBP .among the following interventions: Educational intervention towards patients, Educational intervention directed towards the health professional, Appointment reminder systems/models, and Organisational interventions. Educational interventions directed to the patient and health professional and Organisational interventions showed the high-moderate quality of evidence. These interventions can be pilot tested in LMIC's to establish suitable country speci c BP strategies in anticipation to help meet Sustainable Development Goal (SDG) target 3.4 by reducing premature mortality from non-communicable diseases (NCDs) by a third by 2030. When we restricted analysis (12/14 studies), effects remained signi cant with low heterogeneity for SBP and DBP.
Secondary outcomes, including quality of life, level of hypertension knowledge, beliefs on hypertension, retention to care, and the number of hospitalisations, showed a signi cant controlled blood pressure (BP < 140/90mmHG). The aforementioned secondary outcomes are well outlined in the WHO model of hypertension medication adherence as contributing factors of BP medication adherence.
We believe that this is the rst study to estimate aggregate BP effects of randomised controlled trials conducted in LMICs only. Studies included in this systematic review were conducted between 2005-2020 from low resource settings providing recent evidence on the effectiveness of BP interventions and medication adherence. The variability in heterogeneity can be observed when two studies (Alhalaiqa, 2012 and Mirniam) are excluded showing a signi cant reduction in SBP and DBP (Fig. 5 &6)). These studies had the weakest effect in the intervention groups, which could have been caused by high heterogeneity in random-effect analysis. Overall, we attribute heterogeneity across studies to different study periods, sample sizes, and pooling of multi-component interventions between studies. Though our ndings had moderate evidence on average primarily because of inconsistency and publication bias, heterogeneity was expected because it is common in behavioral research [36]. Similar results on heterogeneity have been reported in several studies in High-income countries that evaluated the effectiveness of BP interventions [26,27]. Considering the current burden on hypertension in LMICs, our results have heightened the need for scaling up interventions to improve hypertension outcomes and meet global SDG in achieving universal health coverage in LMICs [37].
Meta-analysis studies on the effectiveness of BP interventions have been conducted in developed countries [26,27]. Interventions for hypertension show a more signi cant effect when administered in combination [36]. In agreement with signi cant effects of BP interventions when administered in combination is a Cochrane systematic review of 72 randomised trial studies which stated that educational interventions alone were unlikely to be effective [20]. Authors such as Anderson LJ et al. (2020) and Ampofo AG (2020) have stated that education interventions and reminders, including regular follow-ups, were found to be effective at improving patient adherence [38,39]. Health care workers play an essential role in making sure that these interventions are effective. Results from our study have shown a signi cant controlled blood pressure among Educational interventions directed to the patient-led by health care workers, especially pharmacists or community health care workers. According to European and Canadian guidelines, Pharmacists should be the standard care of hypertension as they are valuable in team-based care (TBC) [40,41]. The quality of our study ndings was moderate on average, showing the reliability of our results. High-quality evidence was reported in a meta-analysis that evaluated non-pharmacological interventions, which concluded that Dietary Approach to Stop Hypertension (DASH) could be the most effective in hypertension management [27]. The high-quality evidence reported from DASH intervention could have been the result of evaluating more than 20 interventions from both High income and low-income countries included. The geographical settings of HIC and LMICs might have had an in uence as well, in comparison to our study, which only focused on LMICs.
Our central ndings that non-pharmacological interventions effectively control hypertension could positively impact non-adherence, subsequently reducing CVD's. Patients can get a better understand on hypertension management, which will eventually in uence their awareness and behaviour towards BP medication adherence. More so, help clinicians make better decisions in clinical practice. The focus of this review was on RCT, which is the focused gold standard for evidence of the effectiveness of health [42]. In addition to the pooled change in systolic and diastolic blood pressure, the association between intervention and control group warrants health care workers to consider non-pharmacological interventions.
Hypertension management interventions are essential for achieving the global target of a 25% relative reduction in the risk of premature mortality from NCDs by 2025 and the SDG target of one-third reductions in premature deaths from NCDs by 2030. Thus, looking at our ndings, nonpharmacological interventions for hypertension adherence could avoid over half of CVD events and complications in LMICs. Future research should focus more on evaluating the feasibility of BP non-pharmacological interventions in a speci c LMIC setting. Speci c LMIC interventions may enable the development of public policy measures required to establish, improve and upgrade community health services to cope with the increasing burden of chronic diseases [43]. More so, helping to identify the optimal and effective way to organise and deliver care to hypertensive patients in LMIC's [19,20]

Conclusion
Our study ndings conclude that non-pharmacological interventions are effective. Educational interventions directed to the patient and health professional and Organisational interventions could be the most effective in managing hypertension. There is compelling evidence that multicomponent interventions delivered by pharmacists or health workers may be the most effective in hypertension management. The present and already stated interventions also positively affect outcomes of BP adherence which include quality of life, level of hypertension knowledge, beliefs, retention to care and number of hospitalisations.. Investigation of the feasibility of this evidence at a country level is necessary for LMIC to provide more reliable and effective interventions covering the country-speci c needs having considered her health care system. A multi-component intervention involving patients, health professionals, and the health care system can be considered for future LMICs. Further well-designed randomised trials with larger sample sizes are encouraged to help policymakers make well-informed decisions on hypertension management. The authors would like to thank the University of KwaZulu-Natal (UKZN) for providing the platform to set up and conduct this research study.

Funding
There was no funding source for this study. All authors had full access to all the data included in the study, and the corresponding author had nal responsibility for the decision to submit for publication.

Availability of data and materials
All data reported and analysed in this paper was only from the published literature; hence most of the data and study materials are available in the public domain. Other datasets generated and analysed are available in the article and appendix.

Authors` contributions
The study has been conceptualized and designed by ON, BS and TPM-T. The initial draft of the study has been prepared by ON and PN, SN contributed to the abstract screening and full title screening. ON, BS, SJ-N and TPM-T contributed to data analysis. All authors critically reviewed the draft. All authors approved the nal version of the manuscript.  The methodological quality of results   Forest plots for Diastolic blood pressure after exclusion of two articles

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.