Pharmacotherapy for hypertension‐induced left ventricular hypertrophy

Summary of findings 1. Antihypertensive therapy versus placebo or no treatment for hypertension‐induced left ventricular hypertrophy

Outcome	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Certainty of the evidence (GRADE)
Antihypertensive therapy versus placebo or no treatment for hypertension‐induced left ventricular hypertrophy
Patients or population: adults (18 years of age or older) with hypertension‐induced left ventricular hypertrophy Setting: outpatients Intervention: antihypertensive pharmacological therapy (either monotherapy or in combination) Comparison: placebo or no treatment
Outcome	With placebo/no treatment^a	With antihypertensive therapy	Relative effect (95% CI)	No. of participants (studies)	Certainty of the evidence (GRADE)
All‐cause mortality Mean follow‐up: 3.5 to 4.3 years	136 per 1000	139 per 1000 (101 to 190)	RR 1.02 (0.74 to 1.40)	930 (3 studies)	Very low^b
Cardiovascular events Mean follow‐up: 3.5 to 4.3 years	115 per 1000	125 per 1000 (89 to 178)	RR 1.09 (0.77 to 1.55)	930 (3 studies)	Very low^b
Total serious adverse events Mean follow‐up: 3.5 to 4.3 years	481 per 1000	491 per 1000 (428 to 558)	RR 1.02 (0.89 to 1.16)	930 (3 studies)	Very low^c
Hospitalisation for heart failure Mean follow‐up: 3.5 to 4.3 years	125 per 1000	103 per 1000 (71 to 146)	RR 0.82 (0.57 to 1.17)	915 (2 studies)	Very low^b
Withdrawal due to adverse events Mean follow‐up: 3.5 years	49 per 1000	151 per 1000 (83 to 277)	RR 3.09 (1.69 to 5.66)	522 (1 study)	Very low^d
The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI:* confidence interval; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.
^aControl group estimates come from pooled estimates of control groups. ^bDowngraded two levels for serious imprecision and one level for indirectness. ^cDowngraded one level for imprecision, one level for indirectness, and one level due to high risk of bias. ^dDowngraded one level for imprecision, one level for indirectness, and one level due to publication bias.

Background

Description of the condition

A medical glossary is provided in Appendix 1.

Hypertension is a common and treatable medical condition. However, discrepancies exist in the cutpoints used for diagnosis of hypertension. According to the International Society of Hypertension, the European Society of Cardiology, and the European Society of Hypertension, hypertension is defined as systolic blood pressure values greater than or equal to 140 mmHg or diastolic blood pressure values greater than or equal to 90 mmHg, or both (ESC/ESH 2018; ISH 2020). In contrast, the American College of Cardiology and the American Heart Association consider systolic blood pressures greater or equal to 130 mmHg and diastolic blood pressures greater than or equal to 80 mmHg as hypertension (ACC‐AHA 2017).

Estimates of the overall prevalence of hypertension are greatly influenced by the cutpoint established to its categorisation, the methods used to establish the diagnosis, and the population studied (ACC‐AHA 2017). According to a study carried out with 135 population‐based studies from 90 countries, the prevalence of hypertension (defined as average systolic blood pressure ≥ 140 mmHg, average diastolic blood pressure ≥ 90 mmHg, or use of antihypertensive medication) was 31.1% in 2010 (Mills 2016).

Several factors have been identified as underlying causes of hypertension, many of which are modifiable, such as overweight and obesity, unhealthy diet, physical inactivity, or excess alcohol intake (ACC‐AHA 2017). In addition, hypertension becomes progressively more common with advanced age, with a prevalence of more than 60% in people aged more than 60 years (ESC/ESH 2018).

Hypertension is the leading preventable risk factor for cardiovascular disease and premature death worldwide (Mills 2020). Globally in 2017 high blood pressure was the first cause of death and disability‐adjusted life years worldwide, accounting for 10.4 million deaths and 218 million disability‐adjusted life years (GBD 2017). The Non‐Communicable Risk Factor Collaboration estimated that the prevalence of adults with high blood pressure (defined as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg) increased from 594 million in 1975 to 1.13 billion in 2015, with a larger increase in low‐ and middle‐income countries (NCD‐RisC 2017).

Published evidence has shown associations between high systolic blood pressure and increased cardiovascular risk. A review including results from 61 prospective observational studies found that at 40 to 69 years of age, each difference of 20 mmHg in systolic blood pressure or 10 mmHg in diastolic blood pressure was associated with a more than two‐fold difference in stroke death rate, and with two‐fold differences in death rates from ischaemic heart disease and other vascular causes (Lewington 2002). Another study analysing the association between blood pressure and occurrence of cardiovascular diseases found that people with hypertension had a lifetime risk of overall cardiovascular disease at 30 years of age of 63.3% compared with 46.1% for those with normal blood pressure, and developed cardiovascular diseases five years earlier (Rapsomaniki 2014).

One of the clinical effects of hypertension is left ventricular hypertrophy (LVH), a process of cardiac remodelling. Cardiac remodelling may be defined as genome expression, molecular, cellular, and interstitial changes that are manifested clinically as changes in size, shape, and function of the heart after cardiac injury (Cohn 2000). Cardiac remodelling is linked to heart failure progression. Individuals with major remodelling demonstrate progressive worsening of cardiac function, and it may underlie a sizeable proportion of cardiovascular morbidity and mortality (Cohn 2000).

Hypertension‐induced remodelling of the left ventricle is often grouped into three different geometric patterns: concentric remodelling, concentric LVH, and eccentric LVH (Yildiz 2020). LVH can be diagnosed by electrocardiography criteria, echocardiography criteria, or cardiac magnetic resonance imaging criteria (see Appendix 2). Its prevalence in people with hypertension varies between 36% according to more restrictive diagnostic criteria and 41% according to less conservative ones (Cuspidi 2012).

A meta‐analysis that included six cross‐sectional studies with 60,949 participants also found that left ventricular mass index (LVMI) and global LVH prevalence rates increased in prehypertensive patients compared with normotensive patients, as well as in hypertensive patients compared with prehypertensive patients. The review also found that concentric remodelling was the most common abnormal pattern in normotensive and prehypertensive patients, followed by eccentric and concentric LVH, whereas in hypertensive patients the most frequent altered left ventricular pattern was eccentric LVH. The review found prevalence rates of LVH, as defined by left ventricular mass/h criteria, of 2.5% in normotensive, 6.1% in prehypertensive, and 14.9% in hypertensive individuals (Cuspidi 2019b).

A study included in the review that analysed the 10‐year incidence of LVH found that the risk of developing LVH was significantly greater in prehypertensive patients who progressed to sustained hypertension than in those with persistent prehypertension. This may suggest that the long‐term transition from normal cardiac morphology to LVH is mainly driven by the progression from prehypertension to sustained hypertension (Cuspidi 2019a).

Factors influencing left ventricular geometry in people with hypertension include, amongst others, severity, duration, and rapidity of onset of the increased pressure load; the volume load; age, ethnicity, and sex; comorbidities such as coronary artery disease, diabetes mellitus, obesity, and valvular heart disease; and genetic factors (Aronow 2017). Black individuals with hypertension are more likely than white individuals with hypertension to develop concentric LVH (Aronow 2017). Women with hypertension are more likely than men with hypertension to develop concentric LVH (Aronow 2017).

LVH is considered to be the most potent predictor of morbidity and overall mortality in the hypertensive population, and an independent risk factor for coronary heart disease, sudden death, heart failure, atrial fibrillation, and stroke (Bauml 2010; Llancaqueo 2012; Pérez de la Isla 2010). Severity of the LVH is in turn associated with a higher prevalence of cardiovascular disease (González‐Juanatey 2007). Mortality of individuals with LVH is three to four times higher than in individuals without LVH (Águila‐Marín 2013).

Description of the intervention

Prevention or regression of left ventricular geometric changes with blood pressure control may be an effective way of decreasing future adverse cardiovascular events in individuals with hypertension (Oktay 2016). Indeed, current guidelines recommend treating hypertensive patients with LVH with antihypertensives (Hypertension Canada 2020). In this regard, a stricter blood pressure control is advocated in people at higher risk, such as those with LVH (ESC/ESH 2018).

Drugs currently available for lowering blood pressure include (WHO 2019):

antihypertensives (Anatomical, Therapeutic, Chemical (ATC) classification code: C02);
diuretics (ATC code: C03);
beta‐blocking agents (ATC code: C07);
calcium channel blockers (ATC code: C08);
agents acting on the renin‐angiotensin system (ATC code: C09).

Pharmacotherapy should be selected on an individual basis, taking into account that people with certain associated pathologies will benefit more from particular classes of drugs.

How the intervention might work

Cardiac adaptation in response to pressure overload in conditions such as hypertension usually turns into an increase in left ventricular mass influenced by various physiological and pathological stimuli (Lorell 2000; Schmieder 2000), triggering an increase in force‐generating units (sarcomeres) in the myocyte. The implication is that mechanical input transduces into biochemical events that modify gene transcription in the nucleus. The parallel addition of sarcomeres causes an increase in myocyte width, which in turn increases wall thickness; thus an increase in pressure can be offset (Lorell 2000).

Cardiomyocyte hypertrophy is one of many structural alterations occurring in hypertensive heart disease. Fibroblasts undergo hyperplasia and conversion to myofibroblasts, along with hypertrophy of vascular smooth muscle cells. Noncellular elements related to myocardial remodelling include expansion of interstitial and perivascular collagen that make up the extracellular matrix. Changes in intramyocardial capillary density and arteriolar thickening compound cardiac ischaemia in people with hypertension. These remodelling events are orchestrated via effects of biomechanical stress on the extracellular matrix that, in turn, signal stretch‐activated ion channels leading to intracellular transmission of signals to the nucleus, upregulating hypertrophic gene expression. Similar transduction occurs from cytokine signalling via intracellular calcium handling to myocardial transformation. In the short term, increasing wall thickness in proportion to increased pressure helps to normalise myocardial stress. However, long‐term outcomes clearly worsen with progressive hypertrophy, with increasing LVMI translating to commensurate increases in adverse cardiovascular events and all‐cause mortality (Raman 2010).

Regression of left ventricular mass is based on a reduction of wall thickness by all of the antihypertensive drug classes (Fagard 2009). LVH regression may be due to a decrease in both the myocyte volume and the fibrosis in the interstitium by afterload reduction as the main mechanism (Verdecchia 2004). Different drug classes may have different effects on the magnitude of left ventricular mass reduction (Devereux 2004; Gradman 2006; Klingbeil 2003).

Why it is important to do this review

Reviews comparing different antihypertensive drug classes in LVH regression are inconsistent, and do not analyse the repercussion on cardiovascular events or mortality (Fagard 2009; Roush 2018; Soliman 2017). In this regard, uncertainty remains on the prognostic relevance and the impact of LVH regression in terms of major clinical endpoints in people with hypertension‐induced LVH. To date, no systematic review specifically comparing the effects of antihypertensive drug treatment with placebo or no treatment on the morbidity and mortality of people with LVH and hypertension has been published.

Objectives

To assess the effect of antihypertensive pharmacotherapy compared to placebo or no treatment on morbidity and mortality of adults with hypertension‐induced LVH.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) with at least 12 months' follow‐up that analysed at least one of our primary outcomes.

Types of participants

Adults (18 years of age or older) with high blood pressure with LVH caused by hypertension.

In the case of clinical trials that included patients in whom the cause of the LVH was not specified and with coexisting pathologies other than hypertension that can lead to LVH (e.g. aortic stenosis, aortic regurgitation, mitral regurgitation, dilated cardiomyopathy, hypertrophic cardiomyopathy, ventricular septal defect, or infiltrative cardiac processes such as Fabry disease and Danon disease), we planned to select and include subgroups of participants with both LVH and hypertension without other alternative causes of LVH.

We identified these specific subgroups of patients through individual patient data from the included studies. If it was not possible to obtain individual patient data, we planned to include the study if ≥ 80% of the participants had LVH and hypertension without other possible causes of LVH. In the case of studies that did not explicitly provide data regarding the cause of the LVH, or when this could not be inferred from the participants' baseline characteristics, the study was included. We planned to carry out a sensitivity analysis by excluding studies with no specific information regarding the cause of LVH.

Types of interventions

Intervention

Antihypertensive pharmacological therapy (either in monotherapy or in combination):

antihypertensives;
diuretics;
beta‐blocking agents;
calcium channel blockers;
agents acting on the renin‐angiotensin system.

Control

Placebo or no treatment.

We excluded trials with multidimensional interventions.

Types of outcome measures

Primary and secondary outcomes were established. We analysed the outcomes at longest follow‐up in clinical trials with at least 12 months' follow‐up time.

Primary outcomes

All‐cause mortality.
Cardiovascular events (myocardial infarction (fatal or non‐fatal), stroke (fatal or non‐fatal), or atrial fibrillation).
Total serious adverse events. Serious adverse events were defined according to the International Conference on Harmonisation (ICH) Guidelines as events that at any dose result in death, are life‐threatening, require inpatient hospitalisation or prolongation of existing hospitalisation, result in persistent or significant disability, or are a congenital anomaly/birth defect, and any important medical event that may have jeopardised the participant or requires intervention to prevent it (ICH‐GCP 1997). These events were not required to have a causal relationship with the antihypertensive treatment.

Secondary outcomes

Hospitalisation for heart failure.
Reduction of the left ventricular mass index (LVMI).
Total adverse events.
Withdrawal due to adverse events.

Search methods for identification of studies

Electronic searches

The Cochrane Hypertension Information Specialist searched the following databases with no language, publication year, or publication status restrictions:

Cochrane Hypertension Specialised Register via the Cochrane Register of Studies (to 26 September 2020);
Cochrane Central Register of Controlled Trials (CENTRAL; 2020, Issue 9) via the Cochrane Register of Studies (to 22 September 2020);
Ovid MEDLINE and Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, Daily and Versions (from 1946 to 22 September 2020);
Ovid Embase (from 1974 to 22 September 2020);
US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov) (to 22 September 2020);
World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch) (to 26 September 2020).

The Information Specialist modelled subject strategies for databases on the search strategy designed for MEDLINE. Where appropriate, we combined them with subject strategy adaptations of the sensitivity‐ and precision‐maximising search strategy designed by Cochrane for identifying randomised controlled trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions) (Higgins 2021). The search strategies for major databases are presented in Appendix 3.

Searching other resources

The Cochrane Hypertension Information Specialist searched the Hypertension Specialised Register segment (which included searches of MEDLINE and Embase for systematic reviews) to retrieve existing reviews relevant to this systematic review, in order to scan their reference lists for additional trials. The Specialised Register also included searches for controlled trials in AMED (Allied and Complementary Medicine Database), CAB Abstracts and Global Health, CINAHL (Cumulative Index to Nursing and Allied Health Literature), ProQuest Dissertations & Theses, and Web of Science.
In addition to the above electronic databases, we searched LILACS BIREME (Latin American and Caribbean Health Science Information database) (lilacs.bvsalud.org/es/; to 19 February 2021), Epistemonikos (www.epistemonikos.org/es; to 19 February 2021), and Clarivate Web Of Science (www.recursoscientificos.fecyt.es/; to 26 February 2021).
We checked bibliographies of the included studies and any relevant systematic reviews identified for further references to relevant trials.
Where necessary, we contacted the authors and funders of key papers and abstracts to request additional information.
We searched clinical study reports for additional information about relevant trials.

Data collection and analysis

Study screening was carried out by pairs of review authors. Any discrepancies were resolved by consensus amongst all the review authors. Two review authors independently carried out the data extraction using a specially designed data extraction form. The same two review authors assessed risk of bias of the trials and the overall quality of the evidence.

Selection of studies

We summarised data using standard Cochrane methodologies (Higgins 2021). We applied no language restrictions.

Two review authors screened the search results for potentially relevant trials and independently assessed these for inclusion or exclusion based on the inclusion criteria, using a predesigned eligibility form. Any disagreements were resolved through discussion until consensus was reached. One review author acted as referee.

We selected studies for inclusion following the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021). We:

merged search results using Covidence (Covidence), removing duplicate records of the same report;
examined titles and abstracts to remove obviously irrelevant reports;
retrieved the full texts of reports deemed potentially relevant;
linked multiple reports of the same study;
examined the full‐text reports to determine study eligibility;
contacted investigators to clarify study eligibility (or to request further information, such as missing results) where appropriate;
made final decisions on study inclusion and proceeded to data collection.

The study selection process was illustrated using a PRISMA flow diagram.

Data extraction and management

Two review authors extracted data from the included trials using a spreadsheet data extraction form, and another review author checked the data entry.

We extracted the following data.

Eligibility criteria of the trials.
Demographics (age, gender, ethnicity, country).
Diagnosis method, diagnosis criteria.
Diabetes, chronic renal failure.
Left ventricular ejection fraction (LVEF).
Outcome data (all‐cause mortality, cardiovascular events, total serious adverse events, hospitalisation for heart failure, reduction of the LVMI, total adverse events, withdrawal due to adverse events).
Intervention data: type and regimens.

In the case of a discrepancy, one review author acted as referee to achieve final consensus.

Assessment of risk of bias in included studies

Two review authors independently assessed the risk of bias for each included trial using the domain‐based risk of bias tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Any discrepancies were discussed until consensus was achieved. One review author acted as referee during discussions.

We assessed the following risk of bias domains according to the definitions for each classification provided below (Higgins 2011).

Generation of allocation sequence (checking for possible selection bias)

For each included trial, we planned to describe the method used to generate the allocation sequence in sufficient detail to allow an assessment of whether it should produce comparable groups.

We assessed the methods as follows.

Low risk (any truly random process, e.g. random number table; computer random number generator).
High risk (any non‐random process, e.g. odd or even date of birth; hospital or clinic record number).
Unclear risk, if the trial was reported as randomised, but the method used for the allocation sequence generation was not described.

Allocation concealment (checking for possible selection bias)

For each included trial, we planned to describe the method used to conceal the allocation sequence in sufficient detail to determine whether intervention allocation could have been foreseen, in advance of or during recruitment, or changed after assignment.

We assessed the methods as follows.

Low risk (e.g. telephone or central randomisation; consecutively numbered, sealed, opaque envelopes).
High risk (open random allocation; unsealed or non‐opaque envelopes; alternation; date of birth).
Unclear risk, if the trial was reported as randomised, but the method used to conceal the allocation was not described.

Blinding or masking (checking for possible performance bias)

For each included trial, we planned to describe the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant had received. We judged trials to be at low risk of bias if they were blinded, or if we determined that lack of blinding could not have affected the results. We assessed blinding separately for different outcomes or classes of outcomes.

We assessed the methods as follows.

Low, high, or unclear risk for participants.
Low, high, or unclear risk for personnel.
Low, high, or unclear risk for outcome assessors.

We performed a sensitivity analysis to analyse results from blinded studies separately.

Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)

Low risk (any one of the following): no missing outcome data; reasons for missing outcome data were unlikely to be related to the true outcome (for survival data, censoring unlikely to be introducing bias); missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk was not enough to have a clinically relevant impact on the intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardised difference in means) amongst missing outcomes was not enough to have a clinically relevant impact on observed effect size; missing data have been imputed using appropriate methods.
High risk (any one of the following): reason for missing outcome data was likely to be related to the true outcome, with an imbalance in either numbers or reasons for missing data across intervention groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk was enough to induce clinically relevant bias in the intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardised difference in means) amongst missing outcomes was enough to induce clinically relevant bias in observed effect size; ‘as‐treated' analysis done with substantial departure of the intervention received from that assigned at randomisation; potentially inappropriate application of simple imputation.
Unclear risk (any one of the following): insufficient reporting of attrition or exclusions to permit a judgement of low or high risk (e.g. number randomised not stated, reasons for missing data not provided); the study did not address this outcome.

Selective reporting (reporting bias due to selective outcome reporting)

For each included trial, we described how we investigated the possibility of selective outcome reporting and what we found.

We assessed the methods as follows.

Low risk (any one of the following): the trial protocol was available, and all of the trial’s prespecified (primary and secondary) outcomes that were of interest in the review were reported in the prespecified way; or the trial protocol was unavailable but it was clear that the published reports included all expected outcomes, including those that were prespecified (convincing text of this nature may be uncommon).
High risk (any one of the following): not all of the study’s prespecified primary outcomes were reported; one or more primary outcomes were reported using measurements, analysis methods, or subsets of the data (e.g. subscales) that were not prespecified; one or more reported primary outcomes were not prespecified (unless clear justification for their reporting was provided, such as an unexpected adverse effect); one or more outcomes of interest in the review were reported incompletely so that they could not be entered in a meta‐analysis; the trial report failed to include results for a key outcome that would be expected to have been reported for such a trial.
Unclear: insufficient information to permit a judgement of low or high risk.

Other bias (bias due to problems not covered elsewhere in the tool)

For each included trial, we described any important concerns regarding other possible sources of bias (baseline imbalance, sponsorship bias, confirmation bias, bias of the presentation data, etc.).

Low risk of bias: the trial appeared to be free of other factors that could put it at risk of bias.
High risk of bias: there were other factors in the trial that could put it at risk of bias, e.g. no sample size calculation made.
Unclear risk of bias: the trial may or may not be free of other factors that could put it at risk of bias.

Measures of treatment effect

For each binary outcome, such as all‐cause mortality, cardiovascular events, total serious adverse events, hospitalisation by heart failure, total adverse events, and withdrawals due to adverse events, we estimated the risk ratio (RR) with 95% confidence intervals (CI).

For continuous outcomes (reduction of the LVMI), we calculated the mean difference (MD) with 95% CI.

Unit of analysis issues

For all prespecified primary and secondary outcome variables, we performed the analyses based on the number of randomised participants. For reduction of the LVMI, we assessed the change in LVMI in addition to estimating the differences between study arms in the number of participants who had a reduction of LVMI.

Dealing with missing data

We contacted the corresponding author of the included trial(s) to obtain information on missing data. If we were unable to obtain further information, we would carry out the analysis based on available participant information as our main analysis, that is the denominator for each outcome in each trial was the number randomised minus the number of participants whose outcomes were known to be missing.

We assessed the percentage of dropouts for each included trial and for each study group. We considered an intention‐to‐treat analysis for all trials that either reported this analysis or provided sufficient information to perform such an analysis. Otherwise, we included the analysis that was used, and conducted a sensitivity analysis including only results derived from an intention‐to‐treat analysis.

Assessment of heterogeneity

We used the I² statistic and the Q test to measure statistical heterogeneity between the trials. The I² statistic describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error (Higgins 2021).

Assessment of reporting biases

We planned to assess publication bias and other bias using a funnel plot in the case of 10 or more included trials (Sterne 2011).

Data synthesis

We performed meta‐analyses using a fixed‐effect model. We explored the robustness of the results using the random‐effects model. We reported the fixed‐effect model results for all outcomes, except in cases where the 95% CIs obtained through random‐effects models were wider than those from the fixed‐effect model.

We performed the meta‐analyses according to the Cochrane guidance (Higgins 2021). We used Review Manager Web to analyse the data (RevMan Web 2020). We considered the results of the meta‐analyses only in the absence of substantial heterogeneity (I² < 60%).

Trial sequential analysis

Meta‐analysis of cumulative data runs the risk of random errors (‘play of chance') due to sparse data and repetitive analyses of the same data (Brok 2008; Brok 2009; Thorlund 2010; Thorlund 2017; Wetterslev 2008; Wetterslev 2009). To assess the risk of random errors in our cumulative meta‐analyses, we planned to conduct diversity‐adjusted trial sequential analyses based upon the proportion with the outcome in the control group; an a priori set relative risk reduction of 20%; an alpha of 5%, a beta of 20%; and the squared diversity in the meta‐analysis (CTU 2011; Thorlund 2009; Thorlund 2017). We planned to conduct sensitivity analyses of the trial sequential analysis to estimate the need for further trials.

Subgroup analysis and investigation of heterogeneity

Where possible, we planned to explore the potential causes of substantial heterogeneity (I² > 60%) for a primary outcome (all‐cause mortality, cardiovascular events, and total serious adverse events) by conducting subgroup analyses according to the following aspects.

Participant gender
Participant ethnicity
Left ventricular ejection fraction: < 40% compared to ≥ 40%
Drug group

Sensitivity analysis

We tested the robustness of the results using several sensitivity analyses, restricting the analyses to the following.

Blinded trials
Trials at low risk of bias (see definition below)
Non‐industry‐sponsored trials
Trials that provided information regarding the cause of the LVH
Trials with results derived from an intention‐to‐treat analysis

As we considered it unlikely that many trials would be at low risk of bias in all domains, we planned to choose three core domains instead of all domains as at low risk of bias: generation of random sequence and allocation concealment, incomplete outcome data, and selective reporting bias.

Summary of findings and assessment of the certainty of the evidence

We used GRADE approach to assess the certainty of the body of evidence associated with specific outcomes (all‐cause mortality, cardiovascular events, total serious adverse events, hospitalisation for heart failure, and withdrawal due to adverse events) (Guyatt 2011a). The GRADE approach appraises the certainty of a body of evidence based on the extent to which one can be confident that an estimate of effect or association reflects the item being assessed. The GRADE approach considers within‐study risk of bias (methodological quality), directness of the evidence, heterogeneity of the data, precision of effect estimates, and risk of publication bias (Balshem 2011; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2011g; Guyatt 2011h; Guyatt 2011i; Guyatt 2013). We created a summary of findings table summarising this information (summary of findings Table 1).

Results

Description of studies

For details on included, excluded, and ongoing studies and studies awaiting classification, see Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies; and Characteristics of studies awaiting classification.

Results of the search

The screening process is shown in Figure 1. We identified 1970 records through database searching, seven of which were excluded as duplicates. We identified two additional references through searching Clarivate Web of Science. We screened the remaining 1965 records based on title and abstract, excluding 1671 records. We obtained and reviewed the full texts of the remaining 294 articles and excluded 280 of them. Of the 14 potentially eligible articles, we classified four as ongoing studies and seven as awaiting classification (corresponding to eight studies). We included three studies in both in the qualitative and quantitative synthesis (meta‐analysis). We also searched and reviewed other publications corresponding to the trials identified through the database searches.

Figure 1

PRISMA flow diagram

Included studies

We included three studies in the review (EWPHE 1991; SUPPORT 2015; TOPCAT 2014).

The European Working Party on High Blood Pressure in the Elderly trial analysed the effect of hydrochlorotiazide plus triamterene on the morbidity and mortality of individuals 60 years of age or older diagnosed with hypertension (EWPHE 1991). Around 36% of participants in each study arm had cardiovascular complications at entry. The study did not intend to determine outcomes in individuals with hypertension and LVH, and the trial investigators did not describe a specific LVH diagnostic criteria. During the study, standard 12‐lead electrocardiograms (ECG) were obtained from trial participants, and the review authors established the criteria 'RV1 + SV5 ≥ 35 AND RaVL > 12' by consensus to identify the subgroup of participants with LVH. Information regarding the cause of the LVH was not available, nor data on possible coexisting pathologies other than hypertension that could lead to LVH.

The Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist Trial evaluated the influence on cardiovascular outcomes of adding spironolactone to baseline treatment in adult patients with symptomatic heart failure (New York Heart Association (NYHA) class II‐IV) and LVEF of at least 45% (TOPCAT 2014). Participants were to present with systolic blood pressure under 140 mmHg; those patients with systolic blood pressure > 140 mmHg and ≤ 160 mmHg were eligible for enrolment if on three or more medications to control blood pressure. Individuals with known infiltrative or hypertrophic obstructive cardiomyopathy or pericardial constriction were excluded. Trial investigators defined LVH using the American Society of Echocardiography (ASE) criteria, which refers to LVMI > 115 g/m² in men or > 95 g/m² in women (Lang 2005). Information on the cause of the LVH was not available, therefore we included participants with hypertension and LVH and excluded those with aortic stenosis, aortic regurgitation and/or mitral regurgitation, as these conditions represent possible causes of LVH. None of the trial participants had obstructive cardiomyopathy.

The supplemental benefit of angiotensin receptor blocker in hypertensive patients with stable heart failure (NYHA class II‐IV) using olmesartan trial aimed to evaluate whether additive treatment with olmesartan reduced mortality and morbidity in hypertensive adult patients with stable heart failure (SUPPORT 2015). Trial investigators used the ASE criteria to define LVH (Lang 2005), which was a LVMI greater than 115 g/m² in males and greater than 95 g/m² in females. Data regarding the cause of the LVH were not provided, therefore we selected trial participants with hypertension and LVH, excluding those participants with dilated or hypertrophic cardiomyopathy, or both, which may correspond to possible causes of LVH.

Study design

Study designs of the included trials are shown in Table 1. In all three studies allocation was made on an individual basis, and a parallel design was used. All three trials were multicentre. In EWPHE 1991, centres in Belgium, the UK, Finland, France, Italy, the Netherlands, Ireland, Portugal, Norway, and Germany were involved. The TOPCAT 2014 trial was carried out in various centres in the USA, Canada, Russia, Georgia, Argentina, and Brazil. In SUPPORT 2015, all participating centres were in Japan. The three trials had two treatment arms. In all of the included trials analyses were carried out using the intention‐to‐treat principle. Participant mean follow‐up was 4.1 years (range 3.4 to 4.6 years).

Table 1. Study designs of the included trials

	EWPHE 1991	TOPCAT 2014	SUPPORT 2015
Participants, no.	840	3445	1147
Population of interest, no.	15	692	223
Design	Parallel	Parallel	Parallel
Unit of allocation	Individual	Individual	Individual
Setting	Multicentre, international	Multicentre, international	Multicentre, Japan
Recruitment start date	NR	August 2006	October 2006
Recruitment end date	NR	January 2012	March 2010
Recruitment method	NR	NR	NR
Number of arms	2	2	2
Analysis	Intention‐to‐treat	Intention‐to‐treat	Intention‐to‐treat

NR: not reported

EWPHE 1991 was carried out in consultation with the World Health Organization and supported by the Belgian National Research Foundation and the Belgian Hypertension Committee through a grant from Merck Sharp & Dohme and Smith, Kline & French. TOPCAT 2014 was funded by the National Heart, Lung and Blood Institute, US National Institutes of Health. SUPPORT 2015 was supported in part by grants‐in‐aid from the Ministry of Health, Labour, and Welfare and the Ministry of Education, Culture, Sports, Science, and Technology, Japan.

Trial participants

The TOPCAT 2014 trial had the largest sample size (n = 3445), followed by SUPPORT 2015 (n = 1147), and EWPHE 1991 (n = 840). Baseline characteristics of all trial participants are shown in Table 2. There were no baseline imbalances between treatment arms in any of the trials. In EWPHE 1991, 17.7% of participants left the study before the nine‐month follow‐up visit (n = 85 control group, n = 64 active group); these participants were not included in the mortality analysis. Overall, 36.4% of participants stopped the trial prematurely, with no significant differences between study arms (n = 157 control group, n = 149 intervention group; P = 0.18). The main causes of the premature stop were lost to follow‐up (41.8%); discontinuing trial medication for more than three months (17.0%); and the occurrence of a non‐fatal intercurrent disease (12.4%). The mean age of trial participants was 68.7 years (range 65.7 to 71.8 years). In TOPCAT 2014, 9.0% of participants were withdrawn or lost to follow‐up (n = 151 control group, n = 160 intervention group), with no statistically significant differences between treatment arms (P = 0.589). In SUPPORT 2015, 0.1% of trial participants were withdrawn or lost to follow‐up (three in the control group and six in the intervention group), with no significant differences between study arms (P = 0.327). One participant was excluded. Whereas in the SUPPORT 2015 trial males were predominant (74.7%), the majority of participants in EWPHE 1991 and TOPCAT 2014 were female. In TOPCAT 2014 93% of participants were white. The other two trials did not provide information on race.

Table 2. Baseline characteristics of all trial participants

	EWPHE 1991	TOPCAT 2014	SUPPORT 2015
Population	Elderly hypertensive patients	Individuals ≥ 50 years with symptomatic heart failure and LVEF ≥ 45%	Hypertensive patients aged 20 to 79 years with stable symptomatic chronic heart failure
Number of participants randomised	840	3445	1147
Baseline imbalances	No	No	No
Excluded, no. (%)	149 (17.7%)	0	1 (0.1%)
Withdrawn or lost to follow‐up, no. (%)	306 (36.4%)	311 (9.0%)	9 (0.8%)
Follow‐up in years, mean (SD)	4.6 (2.9)	3.4 (1.7)	4.2 (1.2)
Age in years, mean (SD)	71.8 (8.0)	68.6 (9.6)	65.7 (10.2)
Sex, no. males (%)	254 (30.2%)	1670 (48.5%)	857 (74.7%)
Race/ethnicity, no. (%)	NR	Black: 302 (8.8%) White: 3062 (88.9%) Hispanic: 50 (1.5%) Asian: 19 (0.6%) Other: 12 (0.3%)	NR
History of diabetes mellitus, no. (%)	72 (8.6%)	1118 (32.5%)	548 (47.8%)
History of chronic renal failure, no. (%)	2 (0.2%)	1330 (38.6%)	0
Previous myocardial infarction, no. (%)	56 (6.7%)	893 (25.9%)	0
Previous stroke, no. (%)	48 (5.7%)	265 (7.7%)	NR
History of atrial fibrillation, no. (%)	22 (2.6%)	1214 (35.2%)	479 (41.8%)
History of hypertension, no. (%)	840 (100%)	3147 (91.3%)	1147 (100%)
Number of antihypertensives, mean (SD)	NR	3.0 (0.9)	2.7 (1.1)
LVEF (%), mean (SD)	NR	57.1 (7.4)	54.1 (14.7)
LVEF < 40%, no. (%)	NR	0	200 (17.5%)
LVEF ≥ 40%, no. (%)	NR	3445 (100%)	939 (81.9%)
Cause of the LVH	NR	NR	NR

LVEF: left ventricular ejection fraction; LVH: left ventricular hypertrophy; NR: not reported; SD: standard deviation

The population of interest in this review, which corresponds to participants with hypertension and LVH without other possible causes of LVH, constituted a subgroup of participants from all of the included trials. Baseline characteristics of included study participants are shown in Table 3. LVH diagnostic criteria were predefined by trial investigators in TOPCAT 2014 and SUPPORT 2015. In EWPHE 1991, a definition of LVH was not stated, therefore we established the criteria for identifying trial participants with LVH according to available ECG data. Information regarding the cause of the LVH was not available for any of the included trials. We obtained individual‐level participant data for all trials, through which we selected participants with hypertension and LVH without other possible causes of LVH.

See: Summary of findings 1 Antihypertensive therapy versus placebo or no treatment for hypertension‐induced left ventricular hypertrophy

Table 3. Baseline characteristics of the included study participants

	EWPHE 1991	TOPCAT 2014	SUPPORT 2015
Number of participants (%)	15 (1.8%)	692 (20.1%)	223 (19.4%)
Baseline imbalances	Yes	Yes	No
Characteristics with imbalances	RV5	Previous stroke	NA
Excluded, no.	0	0	0
Withdrawn or lost to follow‐up, no. (%)	NR	100 (14.5%)	3 (1.3%)
Follow‐up in years, mean (SD)	3.6 (3.1)	3.5 (1.7)	4.3 (1.2)
Age, mean (SD)	75.7 (6.0)	66.6 (9.4)	66.0 (10.4)
Sex, no. males (%)	6 (40.0%)	335 (48.4%)	162 (72.6%)
Race/ethnicity, no. (%)	NR	Black: 46 (6.6%) White: 642 (92.8%) Hispanic: 3 (0.4%) Asian: 1 (0.1%)	NR
History of diabetes mellitus, no. (%)	1 (6.7%)	117 (25.6%)	96 (43.0%)
History of chronic renal failure, no. (%)	0	229 (33.1%)	0
Previous myocardial infarction, no. (%)	3 (20%)	227 (32.8%)	0
Previous stroke, no. (%)	3 (20.0%)	66 (9.5%)	NR
History of atrial fibrillation, no. (%)	0	230 (33.2%)	92 (41.3%)
Number of antihypertensives, mean (SD)	NR	3.0 (0.9)	2.7 (1.1)
LVEF (%), mean (SD)	NR	56.9 (7.7)	60.4 (13.2)
LVEF < 40%, no. (%)	NR	0	15 (6.7%)
LVEF ≥ 40%, no. (%)	NR	692 (100%)	208 (93.3%)
Cause of the LVH	NR	NR	NR

LVEF: left ventricular ejection fraction; LVH; left ventricular hypertrophy; NR: not reported; SD: standard deviation

The population of interest was 20.1% of all trial participants in TOPCAT 2014, 19.4% in SUPPORT 2015, and 1.8% in EWPHE 1991. We found imbalances between treatment arms in RV₅ in EWPHE 1991 (26.6 mm (standard deviation (SD) 8.1) in the control group; 16.9 mm (SD 7.8) in the intervention group; P = 0.028) and in history of stroke in TOPCAT 2014 (12% in the control group; 6.6% in the intervention group; P = 0.011). The percentage of included participants who were withdrawn or lost to follow‐up was 14.5% in TOPCAT 2014 and 1.3% in SUPPORT 2015, with no significant differences between treatment arms in any of the cases. Mean follow‐up was 3.8 years (range 3.5 to 4.3 years), and mean age of participants was 69.4 years (range 66.0 to 75.7 years). The prevalence of diabetes mellitus at baseline in the included trials ranged from 6.7% to 43.0%. Individuals with history of chronic kidney failure, myocardial infarction and/or atrial fibrillation accounted for 20% to 40% of all participants included in the review. A total of 7.4% participants had a previous stroke. Participants included in the review extracted from TOPCAT 2014 trial received a mean of three antihypertensives at baseline; those from SUPPORT 2015 received a mean of 2.7 antihypertensives at baseline. Mean basal LVEF was approximately 60% in participants in TOPCAT 2014 and SUPPORT 2015. Almost all of the participants included from these trials had an LVEF of at least 40%. LVEF was not measured in EWPHE 1991.

Characteristics of interventions and comparisons

In EWPHE 1991, the intervention consisted of treatment with hydrochlorotiazide plus triamterene or matching placebo. In the first phase, participants received one capsule containing 25 mg of hydrochlorotiazide plus 50 mg of triamterene daily or placebo. The dosage could be increased after not less than two weeks to two capsules per day. If after no less than one month blood pressure remained high with this therapeutic regimen, the second phase was started with the addition of alpha methyldopa or matching placebo. This treatment was started at a daily dose of half a tablet of 500 mg in the evening and was increased when necessary by half a tablet at intervals of not less than two weeks, until: a blood pressure of less than 160/90 mmHg was reached; or a total daily dose of four 500 mg tablets was obtained; or intolerable adverse reactions occurred, precluding a further increase in dosage. During the trial, 57% of participants in the intervention arm received only hydrochlorotiazide plus triamterene, and 42.5% required addition of alpha methyldopa. In the control arm, 32.5% of participants received only hydrochlorotiazide plus triamterene matching placebo, and 67.5% received additional methyldopa matching placebo. Participants could receive diuretics, calcium channel blockers, beta‐adrenoreceptor blocking agents, or Rauwolfia alkaloids for periods of less than three months. Participants requiring these treatments for three months or more terminated the trial. During the trial, there were no significant differences between study arms in the incidence of concomitant treatment with short‐term (less than three months) vasodilators plus other antihypertensives (19.6 per 1000 person years in the control arm versus 11.8 per 1000 person years in the intervention arm; P > 0.05), but short‐term beta‐blockers were less frequently prescribed in the intervention group (13.0 per 1000 person years in the control arm versus 5.1 per 1000 person years in the intervention arm; P < 0.05).

In TOPCAT 2014, the active drug used as intervention was the diuretic spironolactone, which was initiated at a dose of 15 mg once daily. All participants tolerating this dose without adverse events were up‐titrated by protocol to the target dose of 30 mg once daily at week‐four visit. Additional up‐titration to a maximum dose of 45 mg once daily was permitted at the site investigator's discretion after the month‐four visit for participants with refractory heart failure symptoms and acceptable laboratory parameters (potassium, creatinine). Participants in the intervention and control arms continued to receive other treatment for heart failure and coexisting illnesses throughout the trial.

In SUPPORT 2015, the intervention consisted on adding olmesartan for participants in the active treatment group. The drug was initiated at a dose of 5 to 10 mg; physicians were encouraged to increase the dose up to 40 mg/day where possible. No angiotensin receptor blockers were allowed in the control group.

In the EWPHE 1991 and TOPCAT 2014 trials, participants in the control group received a placebo identical in appearance to the active treatment. Participants in the control group of SUPPORT 2015 trial did no receive any add‐on treatment.

Excluded studies

We excluded a total of 280 articles for the following reasons: 169 used a different comparator; 58 had a different follow‐up period; 22 had a different study design; 13 studied a different patient population; 11 studied a different intervention; 3 analysed different outcomes; and 4 provided no specific data for the population of interest and the data could not be obtained. Further details for six of these (Black 2001; Hernández 2000; HOPE 2003; RENAAL 2005; TCCGIH 1994; VALIDD 2007) are provided in the Characteristics of excluded studies section.

Studies awaiting classification

We assessed eight studies as awaiting classification (CHARM 1999; FEVER 2005; HYVET 2001; PROFESS 2007; RALES 1999; Syst‐Eur 1991; TRANSCEND 2004; VAL‐HeFT 2001). In all of these studies, participants with hypertension and LVH were a subgroup of the total trial participants, and specific data for this subgroup were not published. We contacted the authors and funders of these studies in order to obtain the necessary data, but had not received a response or trial data by publication of this review.

Ongoing studies

We found four ongoing studies (ChiCTR‐INR‐16008079; ChiCTR‐IPR‐16009507; NCT02893358; NCT03315832). We will follow up with these studies upon their completion to assess their possible inclusion in future updates of the review.

Risk of bias in included studies

The risk of bias assessment of the included trials is shown in Figure 2. We judged two trials as having a high risk of bias in at least one domain (EWPHE 1991; SUPPORT 2015). We assessed the remaining trial as having an unclear risk of bias in one domain and low risk of bias in the all other domains (TOPCAT 2014).

Figure 2

Risk of bias summary

Allocation

We judged EWPHE 1991 and SUPPORT 2015 trials as having an unclear risk of bias for random sequence generation, as neither the protocol nor the publication provided information regarding the method used for randomisation. We assessed this domain as low risk of bias in TOPCAT 2014, as details regarding the randomisation method used were provided both in the protocol and in the main publication. Randomisation was carried out using permuted blocks, and a randomisation software (NERI’s Verandi software package) was used to allocate participants to either spironolactone or placebo.

We assessed EWPHE 1991 and SUPPORT 2015 trials as having an unclear risk of bias for allocation concealment. Even though randomisation was stratified according to participating centre, age, and sex in both studies, and additionally by the presence or absence of cardiovascular complications in the case of EWPHE 1991, there was no specific information regarding allocation concealment either in the study protocols or in the publications. We judged TOPCAT 2014 as having a low risk of bias in this domain.

Blinding

We considered EWPHE 1991 trial to have a low risk of bias for blinding of participants and personnel, as the trial was double‐blinded, and the protocol specified that tablets of active treatment and matching placebos were identical in shape, taste, and colour. In TOPCAT 2014, both participants and treating physicians were blinded, and the placebo and active treatment were identical in appearance, therefore we assessed this trial as at low risk of bias for this domain. We judged SUPPORT 2015 as having a high risk of bias for blinding of participants and personnel, as it was an open‐label study, and participants in the control arm did not receive placebo.

We judged all three trials as having low risk of bias for blinding of outcome assessment. In EWPHE 1991, deaths and other terminating events were classified and coded by two investigators who were not aware of the treatment that participants received. In TOPCAT 2014, individual components of the primary outcome, myocardial infarctions and strokes, were adjudicated by a clinical events committee that was unaware of treatment assignments. In SUPPORT 2015, endpoints were assessed in a blinded fashion.

Incomplete outcome data

Considering published and unpublished information as well as obtained individual‐level participant data, we judged the three trials as having a low risk of attrition bias. In all three studies, analyses were conducted according to the intention‐to‐treat principle (Table 1). In EWPHE 1991, 17.7% of trial participants left the study before the nine‐month follow‐up visit and were not included in the mortality analysis. Overall, 36.4% of trial participants stopped the trial prematurely, suggesting no differences between study arms (n = 157 control group, n = 149 intervention group; P = 0.18). Measures of the outcome variables prespecified in the review protocol were available for all participants included in the review (Table 2; Table 3).

In TOPCAT 2014, there may be no differences between groups either in the percentage of participants who discontinued the study (9.3% spironolactone arm versus 8.8% placebo arm) or in those with unknown vital status at last expected visit (3.9% spironolactone arm versus 3.8% placebo arm). Of participants who discontinued the study, 100 were in the subgroup included in the review (47 in the intervention group and 53 in the control group; P = 0.784) (Table 3).

In SUPPORT 2015, there may be no differences between groups in the number of participants lost to follow‐up either in the trial total population (six participants in the intervention group and three participants in the control group; P = 0.327) or in the population of interest for the review (two participants in the intervention group and one participant in the control group; P = 0.698), and these participants were included in the analyses. One participant in the control group was excluded due to lack of information, and no participants in the intervention group were excluded (Table 2; Table 3).

Selective reporting

We judged the EWPHE 1991 trial as having a high risk of reporting bias. The double‐blind phase of the study was ended when participants were lost to follow‐up or when a study‐terminating event occurred (e.g. death; cerebral or subarachnoid haemorrhage; non‐controllable congestive heart failure; increase in LVH; serum creatinine increase; rise in diastolic blood pressure; non‐hypertensive conditions requiring continuous long‐term (three months or more) therapy with diuretics, calcium channel blockers, beta‐adrenoceptor blocking agents, or Rauwolfia alkaloids), amongst other reasons. Data were analysed by intention‐to‐treat; however, in the case of participants who discontinued the study, only the date of death was registered. These methodological considerations could have led to an underestimation of the total clinical events.

We considered both TOPCAT 2014 and SUPPORT 2015 trials to have a low risk of reporting bias. Although information regarding some of the outcomes prespecified in trial protocols was not provided in the study publications, we identified all data needed for analysing the outcomes of interest for the review through individual‐level participant data.

Other potential sources of bias

In all three studies, participants with hypertension and LVH constituted a subgroup of all trial participants. In addition, it was not possible to ensure that LVH was caused by hypertension, as data on the cause of the LVH were not available in any of the studies. However, it was possible through individual‐level participant data to identify participants with coexisting pathologies other than hypertension that can potentially lead to LVH, and we did not include those participants in the analyses.

In EWPHE 1991, participants with hypertension and LVH did not correspond to a predefined subgroup of participants, and this subgroup represented 1.8% of the total population of the trial. Furthermore, the trial was supported by grants from pharmaceutical companies. We therefore judged EWPHE 1991 as having a high risk of other bias.

We classified TOPCAT 2014 and SUPPORT 2015 trials as having an unclear risk of other bias, since participants with hypertension and LVH without other possible causes of LVH represented around 20% of total populations of the trials.

Effects of interventions

summary of findings Table 1 shows the results and the certainty of the evidence for the following predefined outcome variables: all‐cause mortality, cardiovascular events, total serious adverse events, hospitalisation for heart failure, and withdrawal due to adverse events.

We obtained published and unpublished information as well as individual‐level participant data for all three trials (EWPHE 1991; SUPPORT 2015; TOPCAT 2014), through which we extracted and analysed specific data for participants with hypertension and LVH without other potential causes of LVH. The EWPHE 1991 trial contributed 15 participants (10 in the intervention group and five in the control group); TOPCAT 2014 692 participants (n = 334 in the intervention group and n = 358 in the control group); and SUPPORT 2015 223 participants (n = 124 in the intervention group and n = 99 in the control group), for a total of 930 participants.

All‐cause mortality

Results are shown in Analysis 1.1 and summary of findings Table 1. We analysed all‐cause mortality using data from the three trials, totalling 930 participants. We identified 67 deaths (14.3%) in the intervention arm and 63 (13.6%) in the control arm. Differences between arms were not statistically significant (risk ratio (RR) 1.02, 95% confidence interval (CI) 0.74 to 1.40; I² = 0%). In absolute terms, antihypertensive therapy was associated with 3 more deaths per 1000 participants than in the control arm (139 versus 136 per 1000 participants).

Sensitivity analysis restricting to blinded trials, EWPHE 1991 and TOPCAT 2014, did not change the results (RR 1.04, 95% CI 0.72 to 1.50; I² = 0%; n = 707). Additionally, when excluding the EWPHE 1991 trial, which received grants from pharmaceutical companies, we obtained similar results (RR 1.01, 95% CI 0.72 to 1.40; I² = 0%).

Results from TOPCAT 2014, the only study with low risk of bias in core domains (random sequence generation, allocation concealment, incomplete outcome data, and selective reporting bias), showed that there may be no difference between groups (RR 1.03, 95% CI 0.70 to 1.51; n = 692).

Cardiovascular events

Results for participants who suffered at least one cardiovascular event are shown in Analysis 2.4 and summary of findings Table 1. We carried out an estimation of the effect with data from the three trials (n = 930 participants). The three trials provided data for fatal or non‐fatal myocardial infarction and fatal or non‐fatal stroke (EWPHE 1991; SUPPORT 2015; TOPCAT 2014). Additionally, the TOPCAT 2014 and SUPPORT 2015 trials provided information regarding atrial fibrillation. In EWPHE 1991, arrhythmias events were described globally; specific data for atrial fibrillation events were not available.

A total of 59 participants (12.6%) in the intervention arm and 53 participants (11.5%) in the control arm suffered at least one cardiovascular event. We found that there may be no differences between arms in relative terms (RR 1.09, 95% CI 0.77 to 1.55; I² = 0%). In absolute terms, there were 10 more participants with cardiovascular events in the intervention arm per 1000 participants (125 versus 115 per 1000 participants).

Sensitivity analysis restricting to blinded trials, EWPHE 1991 and TOPCAT 2014, did not change the results (RR 1.00, 95% CI 0.63 to 1.57; I² = 0%; n = 707). Excluding the only trial with grants from the pharmaceutical industry (EWPHE 1991), we obtained similar results (RR 1.09, 95% CI 0.77 to 1.55; I² = 0%). Results from the TOPCAT 2014 trial, the only study with low risk of bias in core domains, showed that there may be no difference between groups (RR 0.97, 95% CI 0.61 to 1.54; n = 692).

Results for each type of cardiovascular event individually (fatal and non‐fatal myocardial infarction, fatal and non‐fatal stroke, and atrial fibrillation) are shown in Analysis 2.1, Analysis 2.2, and Analysis 2.3. We also found that there may be no differences between groups for these outcome variables. In relative terms, the difference in the incidence of myocardial infarction was RR 1.22, 95% CI 0.57 to 2.62 (I² = 1%, three studies, n = 930); for stroke RR 0.67, 95% CI 0.35 to 1.28 (I² = 0%, three studies, n = 930); and for atrial fibrillation RR 1.60, 95% CI 0.93 to 2.75 (I² = 0%, two studies, n = 915).

Total serious adverse events

Results for participants who suffered at least one serious adverse event are shown in Analysis 3.1 and summary of findings Table 1. We estimated results using data from the three trials (n = 930 participants). The TOPCAT 2014 trial provided a standardised definition of serious adverse events and analysed them systematically. In EWPHE 1991 and SUPPORT 2015 trials, a definition of serious adverse events was not established, thus we constructed the outcome variable following the definition provided by the International Conference on Harmonisation (ICH) (ICH‐GCP 1997). For EWPHE 1991, we considered all‐cause mortality, myocardial infarction, stroke, moderate and severe congestive heart failure, and severe increase in serum creatinine as serious adverse events. In SUPPORT 2015, serious adverse events included all‐cause death, myocardial infarction, stroke, hospitalisation due to heart failure, hospitalisation for any cardiovascular reason, new‐onset atrial fibrillation, fatal arrhythmia, and development of renal failure.

According to above‐mentioned criteria, 229 participants (48.9%) had at least one serious adverse event in the active treatment arm, compared with 222 participants (48.1%) in the control arm, suggesting no difference between groups (RR 1.02, 95% CI 0.89 to 1.16; I² = 0%). In absolute terms, there were more participants with serious adverse events in the active‐treatment arm (491 versus 481 per 1000 participants).

Sensitivity analysis restricting to blinded trials, EWPHE 1991 and TOPCAT 2014 (RR 1.01, 95% CI 0.87 to 1.18; I² = 0%; n = 707), and to trials with no pharmaceutical industry funding, TOPCAT 2014 and SUPPORT 2015 (RR 1.01, 95% CI 0.88 to 1.16; I² = 0%), did not change the results. Results from the TOPCAT 2014 trial, the only study with low risk of bias in core domains, showed that there may be no difference between groups (RR 1.00, 95% CI 0.86 to 1.17; n = 692).

Hospitalisation for heart failure

Results for participants who were hospitalised for heart failure are shown in Analysis 4.1 and summary of findings Table 1, and are derived from data from the TOPCAT 2014 and SUPPORT 2015 trials (n = 915). Forty‐nine participants (10.7%) in the treatment arm and 57 participants (12.5%) in the control arm were hospitalised for heart failure. There was no evidence of a difference between groups in relative terms (RR 0.82, 95% CI 0.57 to 1.17; I² = 0%). In absolute terms, more participants in the control arm (125 versus 103 per 1000 participants) were hospitalised for heart failure.

Sensitivity analysis restricting to TOPCAT 2014, the only study with low risk of bias in core domains and the only blinded study that provided data for this outcome, showed no differences between groups (RR 0.68, 95% CI 0.40 to 1.16; n = 692).

Reduction of the left ventricular mass index

We estimated both the differences in the number of participants who had reduction of LVMI (Analysis 5.1) and the change in LVMI (Analysis 5.2). Data regarding these variables were only available for the TOPCAT 2014 trial (n = 54).

We found that there may be no differences between groups in participants with reduction of LVMI (60% in the intervention arm versus 69% in the control arm; RR 0.87, 95% CI 0.58 to 1.30). There was no evidence of a difference between groups in change in LVMI (mean difference −0.30, 95% CI −5.87 to 5.27).

Total adverse events

The TOPCAT 2014 and SUPPORT 2015 trials provided data for this outcome (n = 915). A total of 313 of 458 participants (68.3%) in the treatment group and 307 of 457 participants (67.2%) in the control group experienced at least one adverse event, suggesting no differences between groups (RR 1.07, 95% CI 0.86 to 1.34; I² = 64%). However, the estimate showed substantial heterogeneity.

Results were similar when restricting to TOPCAT 2014, the only study with low risk of bias in core domains and the only blinded study that provided data for this outcome (RR 0.99, 95% CI 0.90 to 1.09).

Withdrawal due to adverse events

Results for participants who withdrew from the trials due to adverse events are shown in Analysis 6.1 and summary of findings Table 1. Data were only available for the TOPCAT 2014 trial (n = 522). In this trial, 15.2% of participants receiving antihypertensive therapy withdrew (39 out of 257) compared with 4.9% (13 out of 265) receiving placebo. The difference in relative terms was statistically significant, favouring placebo (RR 3.09, 95% CI 1.69 to 5.66). In absolute terms, 102 more participants withdrew due to adverse events in the intervention arm compared to the placebo arm (151 versus 49 per 1000 participants).

Sensitivity analysis

Results of the sensitivity analyses of the principal outcomes are shown in the corresponding sections. We carried out sensitivity analyses limited to blinded trials (EWPHE 1991; TOPCAT 2014); to trials with no pharmaceutical industry funding (SUPPORT 2015; TOPCAT 2014); and to the only trial with low risk of bias in core domains (TOPCAT 2014). It was not possible to perform analyses limiting to trials providing information on the cause of LVH, as none of the studies provided this information. We did not carry out sensitivity analysis restricting to trials derived from an intention‐to‐treat analysis, as all the included studies were analysed on this basis.

We did not perform subgroup analyses and trial sequential analyses as we did not find substantial heterogeneity in any of the primary outcomes.

Discussion

We identified three RCTs that met the review inclusion criteria (EWPHE 1991; SUPPORT 2015; TOPCAT 2014). EWPHE 1991 analysed the effect of hydrochlorotiazide plus triamterene on morbi‐mortality of elderly hypertensive patients. TOPCAT 2014 evaluated the influence on cardiovascular outcomes of adding spironolactone to baseline treatment in adults with symptomatic heart failure (NYHA class II‐IV) and left ventricular ejection fraction of at least 45%. SUPPORT 2015 assessed the effect of additive treatment with olmesartan in hypertensive patients with stable heart failure (NYHA class II‐IV). Through the individual participant level data from the three trials, we analysed data from participants with hypertension and LVH without other potential causes of LVH, amounting to a total of 930 participants.

Summary of main results

Based on the available evidence from RCTs, we are very uncertain about the effects of adding additional antihypertensive drugs on the morbi‐mortality of individuals with hypertension and LVH.

The evidence for all‐cause mortality was uncertain (RR 1.02, 95% CI 0.74 to 1.40). The percentage of people who died during the trials was around 14% in both groups. Sensitivity analyses restricting to blinded trials (EWPHE 1991; TOPCAT 2014), and separately to the only trial with low risk of bias (TOPCAT 2014), did not change the results.

Adding additional antihypertensive treatment may also fail to achieve benefit with respect to cardiovascular events (RR 1.09, 95% CI 0.77 to 1.55). 12.6% of the participants in the intervention arm and 11.5% of the participants in the control arm suffered at least one cardiovascular event. Sensitivity analyses revealed similar results. We also found that there may be no differences in the incidence of fatal or non‐fatal myocardial infarction (RR 1.22, 95% CI 0.57 to 2.62); fatal or non‐fatal stroke (RR 0.67, 95% CI 0.35 to 1.28); and atrial fibrillation (RR 1.60, 95% CI 0.93 to 2.75) when analysed separately.

There may be no differences between study arms in participants who were hospitalised for heart failure (RR 0.82, 95% CI 0.57 to 1.17). A total of 10.7% of participants in the treatment arm and 12.5% in the control arm were hospitalised due to heart failure during the trials' follow‐up. When restricting the analysis to the only blinded trial with low risk of bias that provided information for this variable (TOPCAT 2014), results were similar.

The confidence intervals for these outcome variables were wide, thus we cannot discard important benefits or important harms with additional antihypertensive drugs.

With regard to safety, withdrawals due to adverse events may be three‐fold higher in participants with additive antihypertensive treatment (RR 3.09, 95% CI 1.69 to 5.66), based on very low‐certainty evidence. A total of 15.2% of participants receiving antihypertensive therapy withdrew from the trials due to adverse events, compared to a 4.9% receiving placebo. However, as data for this variable corresponded to a single trial (TOPCAT 2014), caution is warranted when drawing conclusions.

Secondarily, we analysed incidence of adverse events. The proportion of participants who suffered at least one serious adverse event was around 48% to 49% in each study arm, thereby showing that there may be no differences between groups (RR 1.02, 95% CI 0.89 to 1.16). The results did not change after carrying out the sensitivity analyses. The proportion of participants suffering at least one adverse event of any type was somewhat higher (between 67% and 68% in each arm), but there may be no differences between groups (RR 1.07, 95% CI 0.86 to 1.34).

The limited available data showed that there may be no difference between study arms when analysing the outcome reduction of the LVMI (RR 0.87, 95% CI 0.58 to 1.30) and change in the LVMI (mean difference −0.30, 95% CI −5.87 to 5.27).

We did not carry out subgroup analyses as none of the primary outcomes revealed substantial heterogeneity. With regard to both primary and secondary outcome variables, heterogeneity was 1% or less in all cases except for the variable participants suffering at least one adverse event of any type, which showed an I² of 64% (P = 0.52).

Overall completeness and applicability of evidence

LVH is a secondary manifestation of hypertension and independently predicts the future cardiovascular disease events (ACC‐AHA 2017). Current guidelines advocate that hypertensive patients with LVH should be treated with antihypertensive therapy in order to decrease the rate of subsequent cardiovascular events (ESC/ESH 2018; Hypertension Canada 2020). Nevertheless, limited data derived from trials were available. The population of interest for the review constituted a subgroup of participants from the three included trials (20.1% from all trial participants in TOPCAT 2014, 19.4% from SUPPORT 2015, and 1.8% from EWPHE 1991). Despite this limitation, individual participant data permitted the identification of participants with hypertension and LVH, and exclusion of those with other possible causes of LVH. This approach enabled us to carry out meta‐analyses to assess the effect of antihypertensive therapy on all of the prespecified outcome variables. In addition, we ruled out heterogeneity for the primary outcomes, and performed sensitivity analyses to assess the robustness of the obtained results when restricting to blinded trials, to trials with low risk of bias in core domains, and to non‐industry‐funded trials. Results of the sensitivity analyses did not change the main findings.

The obtained results may have been influenced by the characteristics of the population included in the review. The mean age of participants included in the meta‐analyses was around 66 to 76 years, and except in the EWPHE 1991 trial, for which LVEF data were not available, almost all of the participants had an LVEF of 40% or more. Moreover, we did not include participants with other coexisting pathologies apart from high blood pressure that can lead to LVH, such as dilated and/or hypertrophic cardiomyopathy, aortic stenosis, and aortic and/or mitral regurgitation. All of these issues must be considered when interpreting the findings and when extrapolating results to other settings.

As we did not identify any completed RCT specifically focusing on people with high blood pressure and LVH caused by hypertension and fulfilling our remaining inclusion criteria, there is a need for RCTs to draw firm conclusions on the effect of antihypertensive therapy on morbi‐mortality of this specific population. We identified some ongoing trials that may help to broaden the existing evidence base on this issue (ChiCTR‐INR‐16008079; ChiCTR‐IPR‐16009507; NCT02893358). ChiCTR‐INR‐16008079 is an RCT designed by researchers from Shanghai Jiaotong University School of Medicine to analyse if treatment with spironolactone delays the progression of heart failure in adults with essential hypertension, echocardiographic signs of LVH, and preserved ejection fraction. Amongst other outcomes, the incidence of hospitalisation due to heart failure and cardiac death will be evaluated. Another trial carried out by the same investigators, ChiCTR‐IPR‐16009507, will assess the effect of spironolactone on the progression of diastolic dysfunction in individuals with essential hypertension, echocardiographic signs of LVH, suspected left ventricular diastolic dysfunction, and preserved ejection fraction. As in the preceding trial, hospitalisation due to heart failure and cardiac death will be evaluated. Antihypertensive Treatment in Masked Hypertension for Target Organ Protection (ANTI‐MASK) is a double‐blinded phase IV RCT that compares allisartan versus placebo in individuals with masked hypertension and at least one kind of target organ damage (LVH, large arterial stiffness, and microalbuminuria) (NCT02893358). The primary outcome will be improvement in target organ damage at one year. Secondarily, the incidence rate of all‐cause death and cardiovascular events (stroke and myocardial infarction) at one year will be assessed. These ongoing trials specifically target people with hypertension and LVH, and consequently are expected to help consolidate the existing evidence.

Another aspect to consider is that most of the participants included in the review (those from TOPCAT 2014 and SUPPORT 2015) were receiving around three antihypertensive drugs at baseline. With the available studies, the possible benefits that could be obtained with the intervention were from adding another antihypertensive to the baseline treatment (intensifying antihypertensive treatment), not of initially receiving an antihypertensive treatment.

Although the results of this review were found to be inconclusive, they could assist physicians who treat patients with hypertension and LVH in making decisions regarding the optimal therapeutic management, balancing the potential benefits and risks of introducing antihypertensive therapy.

Quality of the evidence

We rated the certainty of the evidence for all outcomes as very low. In the case of all‐cause mortality, cardiovascular events, and hospitalisation for heart failure, we downgraded the certainty of the evidence two levels due to imprecision and one additional level due to indirectness. Regarding imprecision, assuming a minimal important difference threshold of 10% for all‐cause mortality, myocardial infarction, stroke, and hospitalisation for heart failure as established by the National Institute for Health and Care Excellence (NICE) for adults with primary hypertension (NICE 2019), the 95% CI around the pooled estimate of effect for these outcomes crossed this threshold to the left and to the right side. As for indirectness, the target population of the included trials did not exactly coincide with the population of interest for the review, as they included patients presenting specific conditions, such as class II‐IV heart failure in TOPCAT 2014 and SUPPORT 2015, or elderly patients in EWPHE 1991. Participants that apart from fulfilling the trial inclusion criteria had hypertension and LVH were only a minor percentage of the total populations of the trials, for which we extracted data from individual participant‐level data. Additionally, in the case of cardiovascular events, the lack of blinding of participants and personnel in the SUPPORT 2015 trial could have affected the obtained results.

For the outcome total serious adverse events, we downgraded the certainty of the evidence one level each for imprecision, indirectness, and high risk of bias. With regard to imprecision and assuming a minimal important difference threshold of 15% (NICE 2019), the 95% CI around the pooled estimate of effect crossed this threshold to the left right side. In addition, the high risk of bias related to lack of blinding in the SUPPORT 2015 trial could have influenced the adjudication of serious adverse events. Moreover, of the three included trials, only TOPCAT 2014 trial provided a standardised definition of serious adverse events and analysed them systematically. In TOPCAT 2014, a serious adverse event was defined as an adverse event that met one or more of the following criteria: fatal, life‐threatening, requires in‐patients hospitalisation or prolongation of existing hospitalisation, persistent or significant disability/incapacity, congenital anomaly/birth defect, resulted in permanent impairment damage of a body function/structure, or required intervention to prevent permanent impairment of a body function/structure. In EWPHE 1991 and SUPPORT 2015, a definition of serious adverse events was not established, therefore we adopted the definition provided by the International Conference on Harmonisation (ICH) (ICH‐GCP 1997). In both cases, we constructed the serious adverse event outcome following this definition and through individual participant‐level data from the trials.

We downgraded the certainty of the evidence one level for imprecision, one level for indirectness, and a further level for publication bias in the case of withdrawals due to adverse events. With respect to publication bias, only TOPCAT 2014 provided information regarding withdrawals due to adverse events, therefore data were only available for a reduced number of participants from the entire population considered in the review.

Despite the constraint that participants with hypertension and LVH were only a minor percentage of trial populations, the obtained results can be considered robust, as we carried out the analyses on the basis of both published and unpublished information from the trials, including the original participant‐level data. In addition, we performed sensitivity analyses in order to minimise the possible uncertainty. This approach guarantees the validity of the findings.

Potential biases in the review process

We carried out searches in multiple databases in order to identify all potentially eligible trials. Two review authors screened the search results, and disagreements were resolved with the inclusion of a third review author and by consensus from all review authors. We additionally contacted main investigators, corresponding authors, and promoters of many of the potentially eligible trials identified during the screening process to further investigate if the trials included at least some participants with hypertension and LVH, and to obtain unpublished information for the included trials. Two review authors carried out data extraction and management of individual‐level participant data, and discrepancies were discussed until consensus was reached. The two review authors who carried out data extraction and analyses independently assessed the risk of bias of the trials and the certainty of the evidence using the GRADE approach. This strategy enhances the quality of the process.

However, there are some matters that should be taken into account when interpreting the results of the review. First, we did not identify any study specifically addressing people with high blood pressure and LVH caused by hypertension. We built evidence by identifying through individual‐level participant data the specific participants who suffered hypertension and LVH from the included trials. In this regard, characteristics of the target population varied amongst the included trials. TOPCAT 2014 involved adults with symptomatic heart failure (NYHA class II‐IV) and LVEF of at least 45%, and SUPPORT 2015 included hypertensive adults with stable heart failure. In contrast, EWPHE 1991 specifically addressed elderly hypertensive patients and excluded those with congestive heart failure not corrected without diuretics or antihypertensive drugs, or both.

The TOPCAT 2014 trial was designed to evaluate the effect of spironolactone versus placebo on morbidity, mortality, and quality of life in individuals with heart failure with preserved ejection fraction, not specifically with hypertension. In this trial, spironolactone was initiated at a dose of 15 mg once daily, which could be up‐titrated during the trial to a maximum of 45 mg daily. These doses are lower than those usually used when the drug is intended to reduce blood pressure, in which case starting doses are generally 50 to 100 mg per day, with the possibility of titration up to 200 mg per day. This issue may have led to a lower magnitude of effect of the intervention than would have been expected if the drug had been used with the intention of lowering blood pressure. However, the aim of the review was to analyse whether there were differences in benefits and harms when comparing a group of participants who received antihypertensive drugs of any type to a group of participants who received placebo, independently of the doses used.

None of the included trials provided information regarding the cause of the LVH, therefore it was not possible to ensure that the cause of the LVH was hypertension. However, where possible, we identified and excluded from the analyses participants who had coexisting pathologies other than hypertension that could potentially have caused the LVH. Another aspect that should be considered is the inconsistency in the LVH diagnostic criteria. In both TOPCAT 2014 and SUPPORT 2015 trials participants underwent echocardiograms during the trial, and in both cases trial investigators established the American Society of Echocardiograhy (ASE) criteria to diagnose LVH. According to these criteria, people with LVH were those with an LVMI greater than 115 g/m² in males and greater than 95 g/m² in females. In contrast, the EWPHE 1991 trial investigators did not define a specific LVH criteria. Trial participants underwent standard 12‐lead ECG during the trial, but echocardiograms were not performed. Based on the available information, the review authors established the criteria for LVH to select the subgroup of participants of interest for the review by consensus. In this case, we considered participants with LVH as those with 'RV1 + SV5 ≥ 35 AND RaVL > 12'.

Agreements and disagreements with other studies or reviews

Although the role of antihypertensive treatment in the regression of left ventricular mass in people with LVH hypertension has been studied extensively, there is still debate regarding which drug class provides the greatest magnitude of change. Moreover, regression of left ventricular mass is a subrogate endpoint, and thus may not necessarily translate into significant benefits in clinically relevant endpoints. At present, uncertainty persists regarding the prognostic relevance and the impact in terms of major clinical variables of LVH regression in people with hypertension. In this regard, to date no reviews specifically comparing antihypertensive drug treatment with placebo or no treatment in individuals with LVH and hypertension has been published. To our knowledge this is the first systematic review analysing the potential benefit of adding additional antihypertensive therapy compared to placebo in the morbidity and mortality of individuals with hypertension and LVH, which is crucial to determine in order to optimise patient management. We obtained evidence by identifying and selecting subpopulations of interest from clinical trials.

The review by Pierdomenico 2010 carried out a meta‐analysis on the impact of echocardiographic LVH regression on the incidence of cardiovascular events in hypertensive patients. Hypertensive patients with LVH regression or persistent normal left ventricular mass were compared with those with LVH persistence or LVH development. Five studies were identified, including 3149 participants (mean age range 48 to 66 years, 58% men), with a follow‐up duration of three to nine years. LVH regression/persistent normal left ventricular mass was associated with a significant benefit in cardiovascular events compared with LVH persistence/LVH development (adjusted hazard ratio 0.54, 95% CI 0.35 to 0.84; I² = 59%).

Several reviews have compared the effect of some antihypertensive drug classes versus other antihypertensive drug classes in left ventricular mass regression; however, there is a marked inconsistency in the results of these reviews. Fagard 2009 compared the effect of diuretics, beta‐blockers, calcium channel blockers, angiotensin‐converting enzyme inhibitors, and angiotensin receptor blockers on left ventricular mass regression in individuals with hypertension. The review included evidence up to December 2008. A total of 75 publications were identified, involving 6001 participants (mean age 53.8 years). All participants had hypertension, and 43.6% of the studies also required the presence of LVH. Drug treatment consisted of monotherapy in 59% of the studies, whereas the remaining studies permitted add‐on therapy. Median study duration was six months (range two to 48 months). Regression of left ventricular mass was significantly lower with beta‐blockers than with angiotensin receptor blockers (9.8% versus 12.5%), with no significant differences found in any of the other comparisons between drug classes. Beta‐blockers showed significantly less mass regression than the other four drug classes combined (P < 0.01), and regression was more pronounced with angiotensin receptor blockers than with the other classes (P < 0.01). However, the review findings did not coincide with those from a later review published in 2018 (Xing 2018).

Xing 2018 compared the efficacy of fat‐soluble and selective beta1 receptor blockers with other antihypertensive drug classes (angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers, calcium channel blockers, and diuretics) on regression of LVH. A total of 41 RCTs involving 2566 participants with hypertension and LVH were included. Bayesian network meta‐analyses showed that beta‐blockers had greater efficacy in LVH regression when compared with diuretics (mean difference 13.04, 95% CI 3.38 to 22.59) or calcium channel blockers (mean difference 10.90, 95% CI 1.98 to 19.49), but differences were not found when beta‐blockers were compared to angiotensin‐converting enzyme inhibitors or angiotensin receptor blockers. The probabilities of being amongst the most efficacious treatments were: beta‐blockers (72%), angiotensin receptor blockers (27%), angiotensin‐converting enzyme inhibitors (0.01%), calcium channel blockers (0.00%), and diuretics (0.00%).

Roush 2018 identified 12 RCTs that compared diuretics (chlorthalidone, indapamide, and potassium‐sparing diuretic/hydrochlorothiazide) to renin–angiotensin system inhibitors in reducing left ventricular mass. Left ventricular mass reduction was 37% greater with diuretics than with renin–angiotensin system inhibitors. Compared with renin–angiotensin system inhibitors, diuretics significantly reduced end‐systolic left ventricular internal dimension. The strength of the evidence was rated as at least moderate.

The review by Yang 2013 assessed the effect of angiotensin receptor blockers versus placebo or non‐angiotensin receptor blockers, and the impact of combined treatment with angiotensin receptor blockers and angiotensin‐converting enzyme inhibitors on LVH and left ventricular function in individuals on maintenance dialysis, based on evidence published up to November 2010. Six RCTs (207 participants) were included. Angiotensin receptor blockers led to a significantly greater regression of LVMI when compared with placebo/non‐angiotensin receptor blockers, whilst no significant differences were found between treatments in change in LVEF. The addition of angiotensin‐converting enzyme inhibitors to angiotensin receptor blockers did not show an added benefit when compared to treatment with angiotensin receptor blockers alone.

None of these reviews analysed the effect of different antihypertensive drug classes on cardiovascular events or mortality.

Our review, which compared antihypertensive therapy to placebo or no treatment, did not find a benefit of adding antihypertensive drugs on incidence of cardiovascular events or mortality in participants with hypertension and LVH, although the certainty of the evidence was very low. Limited data on the regression of LVMI limited our ability to draw any conclusions.

Figure 1

PRISMA flow diagram

Figure 2

Risk of bias summary

Analysis 1.1

Comparison 1: All‐cause mortality, Outcome 1: All‐cause mortality

Analysis 2.1

Comparison 2: Cardiovascular events, Outcome 1: Myocardial infarction

Analysis 2.2

Comparison 2: Cardiovascular events, Outcome 2: Stroke

Analysis 2.3

Comparison 2: Cardiovascular events, Outcome 3: Atrial fibrillation

Analysis 2.4

Comparison 2: Cardiovascular events, Outcome 4: At least 1 cardiovascular event

Analysis 3.1

Comparison 3: Total serious adverse events, Outcome 1: Total serious adverse events

Analysis 4.1

Comparison 4: Hospitalisation for heart failure, Outcome 1: Hospitalisation for heart failure

Analysis 5.1

Comparison 5: Reduction of the left ventricular mass index, Outcome 1: Participants with reduction of the left ventricular mass index

Analysis 5.2

Comparison 5: Reduction of the left ventricular mass index, Outcome 2: Reduction of the left ventricular mass index

Analysis 6.1

Comparison 6: Withdrawal due to adverse events, Outcome 1: Withdrawal due to adverse events

Summary of findings 1. Antihypertensive therapy versus placebo or no treatment for hypertension‐induced left ventricular hypertrophy

Outcome	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	No. of participants (studies)	Certainty of the evidence (GRADE)
Antihypertensive therapy versus placebo or no treatment for hypertension‐induced left ventricular hypertrophy
Patients or population: adults (18 years of age or older) with hypertension‐induced left ventricular hypertrophy Setting: outpatients Intervention: antihypertensive pharmacological therapy (either monotherapy or in combination) Comparison: placebo or no treatment
Outcome	With placebo/no treatment^a	With antihypertensive therapy	Relative effect (95% CI)	No. of participants (studies)	Certainty of the evidence (GRADE)
All‐cause mortality Mean follow‐up: 3.5 to 4.3 years	136 per 1000	139 per 1000 (101 to 190)	RR 1.02 (0.74 to 1.40)	930 (3 studies)	Very low^b
Cardiovascular events Mean follow‐up: 3.5 to 4.3 years	115 per 1000	125 per 1000 (89 to 178)	RR 1.09 (0.77 to 1.55)	930 (3 studies)	Very low^b
Total serious adverse events Mean follow‐up: 3.5 to 4.3 years	481 per 1000	491 per 1000 (428 to 558)	RR 1.02 (0.89 to 1.16)	930 (3 studies)	Very low^c
Hospitalisation for heart failure Mean follow‐up: 3.5 to 4.3 years	125 per 1000	103 per 1000 (71 to 146)	RR 0.82 (0.57 to 1.17)	915 (2 studies)	Very low^b
Withdrawal due to adverse events Mean follow‐up: 3.5 years	49 per 1000	151 per 1000 (83 to 277)	RR 3.09 (1.69 to 5.66)	522 (1 study)	Very low^d
The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI:* confidence interval; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.
^aControl group estimates come from pooled estimates of control groups. ^bDowngraded two levels for serious imprecision and one level for indirectness. ^cDowngraded one level for imprecision, one level for indirectness, and one level due to high risk of bias. ^dDowngraded one level for imprecision, one level for indirectness, and one level due to publication bias.

Summary of findings 1. Antihypertensive therapy versus placebo or no treatment for hypertension‐induced left ventricular hypertrophy

Table 1. Study designs of the included trials

	EWPHE 1991	TOPCAT 2014	SUPPORT 2015
Participants, no.	840	3445	1147
Population of interest, no.	15	692	223
Design	Parallel	Parallel	Parallel
Unit of allocation	Individual	Individual	Individual
Setting	Multicentre, international	Multicentre, international	Multicentre, Japan
Recruitment start date	NR	August 2006	October 2006
Recruitment end date	NR	January 2012	March 2010
Recruitment method	NR	NR	NR
Number of arms	2	2	2
Analysis	Intention‐to‐treat	Intention‐to‐treat	Intention‐to‐treat
NR: not reported

Table 1. Study designs of the included trials

Table 2. Baseline characteristics of all trial participants

	EWPHE 1991	TOPCAT 2014	SUPPORT 2015
Population	Elderly hypertensive patients	Individuals ≥ 50 years with symptomatic heart failure and LVEF ≥ 45%	Hypertensive patients aged 20 to 79 years with stable symptomatic chronic heart failure
Number of participants randomised	840	3445	1147
Baseline imbalances	No	No	No
Excluded, no. (%)	149 (17.7%)	0	1 (0.1%)
Withdrawn or lost to follow‐up, no. (%)	306 (36.4%)	311 (9.0%)	9 (0.8%)
Follow‐up in years, mean (SD)	4.6 (2.9)	3.4 (1.7)	4.2 (1.2)
Age in years, mean (SD)	71.8 (8.0)	68.6 (9.6)	65.7 (10.2)
Sex, no. males (%)	254 (30.2%)	1670 (48.5%)	857 (74.7%)
Race/ethnicity, no. (%)	NR	Black: 302 (8.8%) White: 3062 (88.9%) Hispanic: 50 (1.5%) Asian: 19 (0.6%) Other: 12 (0.3%)	NR
History of diabetes mellitus, no. (%)	72 (8.6%)	1118 (32.5%)	548 (47.8%)
History of chronic renal failure, no. (%)	2 (0.2%)	1330 (38.6%)	0
Previous myocardial infarction, no. (%)	56 (6.7%)	893 (25.9%)	0
Previous stroke, no. (%)	48 (5.7%)	265 (7.7%)	NR
History of atrial fibrillation, no. (%)	22 (2.6%)	1214 (35.2%)	479 (41.8%)
History of hypertension, no. (%)	840 (100%)	3147 (91.3%)	1147 (100%)
Number of antihypertensives, mean (SD)	NR	3.0 (0.9)	2.7 (1.1)
LVEF (%), mean (SD)	NR	57.1 (7.4)	54.1 (14.7)
LVEF < 40%, no. (%)	NR	0	200 (17.5%)
LVEF ≥ 40%, no. (%)	NR	3445 (100%)	939 (81.9%)
Cause of the LVH	NR	NR	NR
LVEF: left ventricular ejection fraction; LVH: left ventricular hypertrophy; NR: not reported; SD: standard deviation

Table 2. Baseline characteristics of all trial participants

Table 3. Baseline characteristics of the included study participants

	EWPHE 1991	TOPCAT 2014	SUPPORT 2015
Number of participants (%)	15 (1.8%)	692 (20.1%)	223 (19.4%)
Baseline imbalances	Yes	Yes	No
Characteristics with imbalances	RV5	Previous stroke	NA
Excluded, no.	0	0	0
Withdrawn or lost to follow‐up, no. (%)	NR	100 (14.5%)	3 (1.3%)
Follow‐up in years, mean (SD)	3.6 (3.1)	3.5 (1.7)	4.3 (1.2)
Age, mean (SD)	75.7 (6.0)	66.6 (9.4)	66.0 (10.4)
Sex, no. males (%)	6 (40.0%)	335 (48.4%)	162 (72.6%)
Race/ethnicity, no. (%)	NR	Black: 46 (6.6%) White: 642 (92.8%) Hispanic: 3 (0.4%) Asian: 1 (0.1%)	NR
History of diabetes mellitus, no. (%)	1 (6.7%)	117 (25.6%)	96 (43.0%)
History of chronic renal failure, no. (%)	0	229 (33.1%)	0
Previous myocardial infarction, no. (%)	3 (20%)	227 (32.8%)	0
Previous stroke, no. (%)	3 (20.0%)	66 (9.5%)	NR
History of atrial fibrillation, no. (%)	0	230 (33.2%)	92 (41.3%)
Number of antihypertensives, mean (SD)	NR	3.0 (0.9)	2.7 (1.1)
LVEF (%), mean (SD)	NR	56.9 (7.7)	60.4 (13.2)
LVEF < 40%, no. (%)	NR	0	15 (6.7%)
LVEF ≥ 40%, no. (%)	NR	692 (100%)	208 (93.3%)
Cause of the LVH	NR	NR	NR
LVEF: left ventricular ejection fraction; LVH; left ventricular hypertrophy; NR: not reported; SD: standard deviation

Table 3. Baseline characteristics of the included study participants

Comparison 1. All‐cause mortality

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 All‐cause mortality Show forest plot	3	930	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.74, 1.40]

Comparison 1. All‐cause mortality

Comparison 2. Cardiovascular events

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Myocardial infarction Show forest plot	3	930	Risk Ratio (M‐H, Fixed, 95% CI)	1.22 [0.57, 2.62]

2.2 Stroke Show forest plot	3	930	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.35, 1.28]

2.3 Atrial fibrillation Show forest plot	2	915	Risk Ratio (M‐H, Fixed, 95% CI)	1.60 [0.93, 2.75]

2.4 At least 1 cardiovascular event Show forest plot	3	930	Risk Ratio (M‐H, Fixed, 95% CI)	1.09 [0.77, 1.55]

Comparison 2. Cardiovascular events

Comparison 3. Total serious adverse events

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Total serious adverse events Show forest plot	3	930	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.89, 1.16]

Comparison 3. Total serious adverse events

Comparison 4. Hospitalisation for heart failure

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Hospitalisation for heart failure Show forest plot	2	915	Risk Ratio (M‐H, Fixed, 95% CI)	0.82 [0.57, 1.17]

Comparison 4. Hospitalisation for heart failure

Comparison 5. Reduction of the left ventricular mass index

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Participants with reduction of the left ventricular mass index Show forest plot	1	54	Risk Ratio (M‐H, Fixed, 95% CI)	0.87 [0.58, 1.30]

5.2 Reduction of the left ventricular mass index Show forest plot	1	54	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐5.87, 5.27]

Comparison 5. Reduction of the left ventricular mass index

Comparison 6. Withdrawal due to adverse events

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Withdrawal due to adverse events Show forest plot	1	522	Risk Ratio (M‐H, Fixed, 95% CI)	3.09 [1.69, 5.66]

Comparison 6. Withdrawal due to adverse events