Background

Hypertension is the major contributor to global burden of cardiovascular morbidity and mortality [1]. Currently, more than 1.4 billion of the world’s population have hypertension [2] and this figure is expected to rise to 1.6 billion by the year 2025 [3]. The cardiovascular and cerebrovascular complications of hypertension are the most important causes of non-communicable diseases (NCD) related morbidities and mortalities [4]. As hypertension is a preventable risk factor, collaborated actions can prevent the development of complications [5].

Meta-analysis of observational studies in Ethiopia estimated the prevalence of hypertension to be between 20 and 30% [6, 7]. According to WHO, 39% of all deaths in Ethiopia are due to NCDs of which 16% is attributed to cardiovascular diseases (CVD) [8]. Uncontrolled hypertension is one of the major causes of heart failure, chronic renal failure, and ischemic and hemorrhagic strokes which impose severe financial and service burdens on health systems [9, 10]. The control of hypertension within a target goal of blood pressure (BP) plays a critical role in reducing associated CVD. However, hypertension remains inadequately controlled in clinical practice [11, 12]. This would increase the burden of CVD on the health system. The proportions of patients treated for hypertension with uncontrolled BP reported across the country vary substantially. However, these data have not been meta-analyzed to provide pooled estimate of the prevalence of uncontrolled BP among treated hypertensive patients. Therefore, the aim of this study is to examine the prevalence of uncontrolled BP among treated hypertensive patients in Ethiopia. Determining the prevalence will help to comprehend the magnitude of the problem and develop strategies to reduce the imposed burden of CVD.

Methods

Study protocol

The Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) was used in the identification of records, screening of titles and abstracts accompanied by evaluation of eligibility of full texts for final inclusion [13]. The study protocol is registered at PROSPERO with reference number ID: CRD42018116336 and the published methodology is available from: http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42018116336.

Data sources and search strategy

Literature search was done from PubMed/Medline, EMBASE (Ovid® interface) and Google Scholar. Advanced search strategies were used to retrieve relevant findings, by restricting the search for studies on human and published in English. HINARI interface was used to access articles published in subscription based journals and indexed in Science-Direct and Wiley online library. Gray literatures from organizations and online university repositories were accessed through Google Scholar. Key words and indexing terms were used to retrieve articles that were published from 2000 onwards. The key words used for searching were “hypertension”, “high blood pressure” [MeSH] and “Ethiopia”. Boolean operators (AND, OR) were also used in the identification of records. The search was conducted from February 1 to 14, 2019 and all published and unpublished articles available online from January 1, 2000 till the day of data collection were considered.

Screening and eligibility of studies

ENDNOTE reference manager software version 9.2 (Thomson Reuters, Stamford, CT, USA) was used. With the help of the reference manager, duplicate records were identified, recorded and removed. Due to variation in reference styles from different sources, some references were managed manually. Thereafter, two authors (FA and BH) independently screened the titles and abstracts with predefined inclusion criteria. Two authors (MS and BM) independently collected full texts and evaluated the eligibility of them for final inclusion. In each case, the third author played a critical role in solving discrepancies that arose between two authors and in coming to a final consensus.

Inclusion and exclusion criteria

Predefined inclusion-exclusion criteria were used to screen titles and abstracts; and evaluate full texts for eligibility. Observational studies addressing hypertension control among treated adult hypertensive patients in Ethiopia were included. Literatures published from 2000 onwards in the English language were considered. Articles with irretrievable full texts (after requesting full texts from the corresponding authors via email and/or ResearchGate), records with unrelated outcome measures, articles with missing or insufficient outcomes were excluded.

Data extraction

Data abstraction format was prepared in Microsoft Excel. Two authors (FA and BH) independently extracted data related to study characteristics (study area, first author, and year of publication, study design, population characteristics, and sample size) and outcome of interest (hypertension control).

Quality assessment of studies

The internal and external validity of included studies was assessed by using the Johanna Briggs institute (JBI) critical appraisal checklist for studies reporting prevalence data. Based on the checklist, the studies were graded out of 9 points (Table 1). Scores of the two authors (MS and BM) in consultation with third author (FA) (in case of disagreement between the two authors’ appraisal result) were taken for final decision. Studies with the number of positive responses (yes) greater than half of the number of checklists (i.e., score of five and above) were included in the systematic review and meta-analysis.

Table 1 Quality assessment of studies using JBI’s critical appraisal tools designed for prevalence studies
Full size table

Outcome measurements

The primary outcome measure in this meta-analysis is the prevalence of uncontrolled hypertension in Ethiopia. It is aimed to assess the pooled estimates of uncontrolled hypertension among treated hypertensive patients in the country. The sample size was intentionally adjusted to response rates in individual study to reduce bias in calculating the overall prevalence.

Data processing and statistical analysis

A format prepared in Microsoft Excel was used to extract data from the included studies. The data was then exported to STATA software, version 15.0 for analyses. The percentage of variance attributable to study heterogeneity was assessed using I2 statistics. To ascertain variation in true effect sizes across population, Der Simonian and Laird’s random effects model was applied at 95% confidence level. The event rate (proportion) was calculated out of 1 and standard error of Logit event rate was also added with the help of Comprehensive Meta-analysis (CMA) (Biostat, Englewood, New Jersey, USA) version-3 software. CMA was also used for publication bias assessment by using the Begg and Rank correlation as well as Egger’s regression tests. Funnel plots of standard error and precision with Logit event rate was used to present the publication bias assessment. A p-value less than 0.05 (one tailed) was used to declare significance.

Results

A total of 426 studies were identified through the search of electronic databases including PubMed/Medline, EMBASE, and Google Scholar. Eight other articles were identified through reference tracing and other sources. After removing 82 duplicates through ENDNOTE reference manager and manual tracing, a total of 352 records were screened using their titles and abstracts. Then, full text assessment of 26 potentially relevant articles resulted in 13 studies that passed the eligibility criteria and quality assessment and hence included in the systematic review and meta-analysis (Fig. 1).

Fig. 1
figure 1

A PRISMA flowchart showing the selection process of the articles

Full size image

Study characteristics

A total of 13 studies with 5226 hypertensive patients were included in this systematic review and meta-analysis. Among the included hypertensive patients, 2534 were found to have uncontrolled BP. Eleven of the included studies used cross-sectional study design [14,15,16,17,18,19,20,21,22,23,24] while the remaining two were cohort in design [25, 26]. Almost all the included studies were hospital based except for one which was conducted at health centers [18]. The year of publication of included studies ranged from 2014 to 2018. The prevalence of uncontrolled hypertension ranged from 11.42% [22] in Gondar university hospital to 69.94% in Zewditu memorial hospital, Addis Ababa [24] (Table 2).

Table 2 Descriptive summary of studies on hypertension control in Ethiopia included for systematic review and meta-analysis
Full size table

Study outcome measures

Primary outcomes

The pooled prevalence of uncontrolled hypertension in Ethiopia from the 13 studies describing control of BP among treated hypertensive patients was 48% (95% confidence interval [CI]: 36, 61%). When random effects model was assumed for this meta-analysis, a high degree of heterogeneity was observed across studies as evidenced by the I2 statistics (I2 = 99.01%, P < 0.001) (Fig. 2). Univariate meta-regression model showed that sampling distribution is not a source of heterogeneity (regression coefficient = 0.000, p-value =0.92) (Fig. 3).

Fig. 2
figure 2

Forest plot depicting the pooled estimate of the prevalence of uncontrolled hypertension among treated hypertensive patients in Ethiopia

Full size image
Fig. 3
figure 3

Univariate meta-regression model using sample size

Full size image

Sensitivity and subgroup analyses

Sensitivity analysis was conducted by excluding outliers from the analysis. However, there was no significant change on the degree of heterogeneity even when outliers were excluded from the analysis. Therefore, all the studies that passed the quality assessment were included for the meta-analysis. A subgroup analysis was conducted based on the year of publication of the studies. The result of the subgroup analysis revealed that the prevalence of uncontrolled hypertension was highest in 2016 (63%; CI: 60, 67%) followed by 2015 (59%; CI: 53, 65%) (Fig.4). Univariate meta-regression revealed that year of publication is also not a source of heterogeneity (regression coefficient = 0.005, p-value =0.88) (Fig. 5).

Fig. 4
figure 4

Sub-group analysis based on publication year of studies

Full size image
Fig. 5
figure 5

Univariate meta-regression model using publication year of studies

Full size image

Publication bias

Publication bias was assessed by using funnel plots of standard error with logit effect size (event rate). The analysis showed that there is no evidence of publication bias on the included studies. This is confirmed by Egger’s regression test (one-tailed), p = 0.09 and Begg’s correlation test (one tailed), p = 0.15 (Fig. 6).

Fig. 6
figure 6

Funnel plot illustrating publication bias of included studies with Logit event rate and standard error

Full size image

Discussion

A total of 13 institution based studies with 5226 hypertensive patients were included in this systematic review and meta-analysis. In the current study, the pooled prevalence of uncontrolled hypertension among hypertensive patients on treatment in Ethiopia was 48% (CI: 36, 61%). This finding indicated that almost half of hypertensive patients who were following their treatment in health institutions (hospitals and health centers) in Ethiopia did not achieve a target BP, proven to reduce CVD risk associated with hypertension. The prevalence of uncontrolled hypertension in this study is close to the proposed WHO target control rate [1]. The result of the subgroup analysis showed the increment in the prevalence of uncontrolled hypertension from 2014 to 2017. This clearly shows the quality of health service provided for patients with hypertension. Additionally, the national burden of cardiovascular and cerebrovascular diseases, chronic renal failure and the associated morbidity and mortality are expected to rise with the uncontrolled BP [27]. Moreover, a study has shown that treated hypertensive patients but not having control were at increased risk of all cause, CVD specific, heart disease-specific or cerebrovascular disease specific mortality [37] and a holistic approach of patient care including pharmacists to manage patients drug therapy should be used [38].

Conclusion

The prevalence of uncontrolled hypertension was high in Ethiopia. This is alarming as uncontrolled hypertension is associated with an increased risk of cardiovascular complications. This would impose additional burden on the health care system of the country, which is struggling to contain communicable diseases. The prevalence of uncontrolled hypertension is increasing over the years. This evidence suggests that double burden diseases are increasingly affecting Ethiopia. In light of this evidence, policy makers and health care professionals working in the area should implement interventional strategies focusing on achieving an optimal BP among treated hypertensive patients.

Limitation of the study

The study has extensively addressed all relevant data regarding hypertension control among treated hypertensive patients in Ethiopia. However, there are certain limitations to mention. The studies included for the meta-analysis used different cut-off point to define control of BP as there was change in guideline recommendation regarding optimal BP. Additionally, the number of BP measurement used to define uncontrolled hypertension across the included studies was inconsistent.