Effectiveness and safety of beta blockers in the management of hypertension in older adults: a systematic review to help reduce inappropriate prescribing

Background The benefit from a blood pressure lowering therapy with beta blockers may not outweigh its risks, especially in older populations. The aim of this study was to look for evidence on risks and benefits of beta blockers in older adults and to use this evidence to develop recommendations for the electronic decision support tool of the PRIMA-eDS project. Methods Systematic review of the literature using a stage approach with searches for systematic reviews and meta-analyses first, and individual studies only if the previous searches are inconclusive. The target population were older adults (≥65 years old) with hypertension. We included studies reporting on the effectiveness and/or safety of beta blockers on clinically relevant endpoints (e.g. mortality, cardiovascular events, and stroke) in the management of hypertension. The recommendations were developed according to the GRADE methodology. Results Fifteen studies were included, comprising one meta-analysis, four randomized controlled trials, six secondary analyses of randomized controlled trials and four observational studies. Seven studies involved only older adults and eight studies reported subgroup analyses by age. With regard to a composite endpoint (death, stroke or myocardial infarction) beta blockers were associated with a higher risk of events then were other antihypertensive agents. Further, beta blockers showed no benefit compared to other antihypertensive agents or placebo regarding mortality. They appear to be less effective than other antihypertensive agents in reducing cardiovascular events. Contradictory results were found regarding the effect of beta blockers on stroke. None of the studies explored the effect on quality of life, hospitalisation, functional impairment/status, safety endpoints or renal failure. Conclusion The quality of current evidence to interpret the benefits of beta blockers in hypertension is rather weak. It cannot be recommended to use beta blockers in older adults as first line agent for hypertension. Electronic supplementary material The online version of this article (doi:10.1186/s12877-017-0575-4) contains supplementary material, which is available to authorized users.


Background
Hypertension is an important public health challenge worldwide as nearly 1 billion people or~26% of the adult population of the world suffers from hypertension [1,2]. In Europe, the overall prevalence of hypertension appears to be around 30-45% of the general population [3], with a steep increase with ageing [4][5][6]. Hypertension is a major risk factor for cardiovascular and renal disease, and numerous clinical trials including studies in older adults have documented that effective treatment improves survival and confers cardiovascular benefits [4,5]. The Framingham Heart Study and a meta-analysis done by Lewington et al. have shown a continuous and positive relationship between cardiovascular events and usual blood pressure above a baseline level of approximately 115/75 mmHg at all ages and in both sexes [6,7]. Although the strength of the association is weakened with age [7], the absolute difference in cardiovascular risk between the highest and lowest usual blood pressure levels is much greater in older subjects [8]. Hence, it could be expected that especially older adults benefit from a blood pressure lowering therapy. However, there is a phenomenon called "reverse epidemiology" in older adults suffering from hypertension [9] which describes that higher blood pressure can be associated with increased survival [10]. The phenomenon likely reflects the confounding effects of other comorbid conditions that are quite common in this age group. This means that a lower blood pressure may be a proxy for poor health and a warning sign that further blood pressure lowering may actually be harmful. Thus, lower -for older adults -may not always be better [8]. Moreover, some studies suggest that in older adults tight blood pressure control (systolic blood pressure < 140 mmHg) may not reduce stroke more efficiently than a goal blood pressure < 150/90. In adults aged ≥75 years tight blood pressure may increase the incidence of cardiac and vascular disease, and renal failure [11,12]. Tight blood pressure control is not recommended for patients with dementia or cognitive impairment or functional impairment and in patients with limited life expectancy (<2 years) [13].
Drugs used in treating hypertension are among the most commonly used drug classes that are associated with preventable drug-related hospital admissions especially in older populations [14,15]. Among antihypertensive drugs, particularly beta blockers appear to be a problematical drug class for the treatment of hypertension [16]. A recently published Cochrane review on beta blockers for hypertension concludes that beta blockers are not recommended as first line treatment for hypertension as compared to placebo due to their modest effect on stroke and no significant reduction in mortality or coronary heart disease. Compared to other drug classes (diuretics, calcium channel blockers, and renin angiotensin system inhibitors) beta blockers were not efficient in decreasing mortality and morbidity [17]. It is argued that the use of beta blockers in older adults may not be justified because of physiologic changes in people over 60 years of age. These include a low cardiac output, bradycardia, high total peripheral resistance, reduced renal blood flow and glomerular filtration rate, and low plasma renin activity [18].
Nevertheless, to the best of our knowledge, so far no systematic review has analysed the specific evidence on the use of beta blockers for the management of hypertension in aged populations. This systematic review aims to explore the effectiveness and safety of beta blockers in the treatment of hypertension in older adults (≥65 years). A further aim was to develop recommendations on when to discontinue or to adjust the dose when using beta blockers in the treatment of hypertension in older adults. These recommendations were implemented in the electronic decision support tool of the PRIMA-eDS project (Polypharmacy in chronic diseases: Reduction of Inappropriate Medication and Adverse drug events in elderly populations by electronic Decision Support, www.prima-eds.eu).

Methods
This systematic review was developed following an adaptation of the methods proposed by both the Cochrane Handbook for Systematic Reviews of Interventions [19] and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [20]. A full description of the methods has been published previously [21]. A protocol for this systematic review has been elaborated and it is available from the authors on request.

Study inclusion criteria Types of studies
We included systematic reviews, meta-analyses, controlled interventional studies and observational studies reporting on risks and benefits of the use of beta blockers in the treatment of hypertension in older adults.

Types of participants
Older adults, ≥65 years old, with arterial hypertension were included. We ensured, that there was a sufficient number of older adults using the following criteria by type of design.
For systematic reviews and meta-analyses (any of the following criteria): overall mean or median age ≥ 65 years, overall mean or median age < 65, but subgroup analysis of participants <65, or overall mean or median age not reported but more than 80% of studies reported a mean or median age ≥ 65 years.
For controlled interventional studies and observational studies (any of the following criteria): ≥80% of participants ≥65 years or <80% of participants ≥65 years but subgroup analysis of participants <65. We included any setting reporting on the management of hypertension using beta blockers.

Types of interventions
We included studies reporting on the effectiveness and/ or safety of beta blockers for the management of hypertension. Studies were included irrespective of beta blockers prescribed as monotherapy or in combination with any other drug for the treatment of hypertension. We included studies comparing beta blockers versus placebo, no treatment, other antihypertensive drugs or a non-pharmacological intervention.

Types of outcomes
We included any of the following clinically relevant endpoints as primary outcomes: Mortality Hospitalization Cardiovascular event including stroke Quality of life Adverse drug event Life expectancy Cognitive impairment or cognitive status Functional impairment or functional status Renal failure Composite end points including any of the above Any of the above evaluated as safety endpoints

Study exclusion criteria
We excluded conference abstracts, pooled analyses, editorials, opinion papers, case reports, case series, narrative reviews, letters, and qualitative studies. We excluded studies evaluating only blood pressure values.

Search method
Database searches were conducted by YVM and AW. We performed a stage approach with searches 1 and 2 in various databases for systematic reviews and metaanalyses, and search 3 for individual studies. During study selection for search 1 and 2, we identified eligible individual studies from excluded systematic reviews and meta-analyses and transferred those to search 3A for potential inclusion. The list of studies in search 3A was checked for inclusion following the procedures described in the section "Selection of studies". Search 3B was designated as a last step to identify controlled interventional studies and observational trials in various databases. The respectively following search was performed only if the prior searches were inconclusive. We did not apply any language restriction to the search. The following paragraphs contain detailed information about search stages, databases and search dates:  [17] which had covered all individual studies eligible for the time before 2012 (see 3A). This Cochrane review has recently been updated, and again, no additional eligible studies for our review were identified [22]. Our hand search comprised a review of the bibliographies of all included studies. Study protocols from all searches were also collected to consider for future updates of this systematic review.
The PICOS-framework was used to develop the search strings. A selection of keywords used were: Population (older adults, aged) AND Condition (high blood pressure, hypertension) AND Intervention (beta blocker) AND Outcome (quality of life, mortality, life expectancy, hospitalization, cognitive impairment, functional impairment, cardiovascular event including stroke, renal failure) AND Study designs (systematic reviews, meta-analyses, controlled interventional studies and observational studies). Additional file 1 shows the full search strings. Endnote X7 was used to retrieve search results and to deduplicate references.

Selection of studies and data management
First, BF, AV and TJ assessed titles and abstracts from search 1 and 2 (including the search update) and identified studies to include using Endnote X7. Search 3A was done by YVM, TJ and AV, and search 3B by CS, AR, ALT, LS and BF. Second, full manuscripts were obtained for all titles and abstracts that appeared to meet the inclusion criteria or for cases of uncertainty for inclusion. ARG, YVM or AS were consulted when the two researchers could not reach an agreement on whether to include a study.

Data extraction
Two reviewers (AV, YVM or TJ respectively CS, AR, ALT, LS, SW or BF) independently conducted data extraction of the included studies using a standardised and piloted data collection form which was published with the protocol [23]. The two reviewers checked each other's data extraction to look for completeness and accuracy. The data collection form included information related to the study design and aim, characteristics of the participants (i.e. age, sex, setting, comorbidity, use of concomitant medications, frailty, functional and cognitive status), the intervention (i.e. beta blockers), comparison, time to follow-up, and reported outcomes.

Quality appraisal
We used three validated assessment tools to evaluate the quality of each study design: for systematic reviews/ meta-analyses the measurement tool to assess systematic reviews (AMSTAR) [24,25] for intervention studies the Cochrane Collaboration's tool for assessing risk of bias [26] and for observational studies the Critical Appraisal Skills Programme (CASP) [27,28]. Risk of bias was assessed independently by at least two review authors according to the Cochrane Handbook [26].

Data synthesis
Methods utilised to synthesise the studies depended on their quality, design and heterogeneity. According to the protocol we wanted to pool results of studies if at least two were homogeneous regarding participants, interventions and outcomes [23]. If that was not possible due to differences in the reporting or substantive heterogeneity, we reported a narrative synthesis describing all included studies, participants and findings. No additional metaanalyses were performed.

Identification of additional "references of interest" and the development of recommendations
During the search process, BF, AV, TJ, YVM and ARG identified additional references for the development of recommendations according to the methodology described by Martinez-Renom Guiteras et al. [21]. We identified nine additional references which can be seen in Additional file 2. Two were found in search 2, one in the search update, one in the hand search of references of included studies, two were suggested by two clinical researchers (IK and MMV) and three were found by snowballing [29]. The studies did not meet our inclusion criteria (mainly due to age < 65 years), but provided supportive information for the formulation of the recommendations. Included studies and additional references were summarised in a document that was used in team meetings to discuss recommendations on when the use of beta blockers should be discontinued or new doses should be considered in older adults for the management of hypertension [21]. Each recommendation was given strength (weak or strong) and quality (low, moderate or high) following the GRADE methodology [30][31][32]. The recommendations are provided in Table 1.

Results of the search
We identified 1449 records through database searching and 434 through other sources (hand search of reference lists of included studies). After removing 116 duplicates, we screened 1767 records and excluded 1543 records checking titles and abstracts. We assessed 224 full texts for eligibility and excluded 209 records. Additional file 3 provides the comprehensive list of reasons for exclusion of studies after full text analysis. The PRISMA flow diagram is presented in Fig. 1.

Interventions and outcomes
Detailed information about interventions and outcomes can be found in Table 3: Summary of characteristics of the included studies.

Any beta blocker
The meta-analysis [33] included only randomized controlled trials evaluating the efficacy of beta blockers as first-line therapy for hypertension in preventing major cardiovascular events (i.e. stroke, myocardial infarction or death). In the analyses of the group of studies that we are including, the authors compared beta blockers to no treatment, placebo, diuretics, ACE-inhibitors, calciumchannel blockers and angiotensin-receptor blockers. The Matsuzaki trial [47] and the related secondary analysis of Ogihara et al. [41] analysed the optimal calciumchannel blocker benidipine-based combination therapy comparing three intervention arms, one with additional angiotensin receptor blocker, one with a beta blocker and one with a thiazide diuretic. The primary outcomes of these studies were the composite of cardiovascular morbidity and mortality. The observational study of Carlsson et al. [43] compared the following beta blocker treatments: selective vs. no treatment, non-selective no treatment and non-selective vs selective with regard to the outcome of mortality. Gelber et al. [45], another observational study, analysed beta blocker alone and beta blocker in combination with one other antihypertensive drug for the outcome of the development of cognitive impairment.

Atenolol
Atenolol (50 or 100 mg/d) was examined in three randomized controlled trials [34,35] and the related secondary analyses [36,38,39,42], a post hoc analyses of a randomized controlled trial [40] and in two observational It is suggested to discontinue the beta blocker or change it to another antihypertensive drug (unless another indication for beta blockers exists), because beta blockers may increase the risk of stroke and other composite cardiovascular outcomes compared to other antihypertensive agents while not revealing any benefit regarding cardiovascular outcomes or mortality compared to placebo for adults >60 years.

Strong
Low Downgraded for indirectness because only one meta-analysis and one RCT were focused on older people 2 Cochrane reviews [17,49], 2 meta-analyses [33,54], 2 recommendation papers from the Canadian Hypertension Education Program [51,52], and 1 RCT [35] It is suggested to discontinue atenolol for the management of hypertension because it appears to be less effective than other antihypertensives in reducing cardiovascular events, and to have a higher risk of adverse events.

Strong
Low Downgraded for indirectness because only one RCT was focused on older people 2 Cochrane reviews [17,50] and 2 RCTs [35] It is suggested to discontinue beta blockers as monotherapy for the management of hypertension because it may be inferior to other antihypertensives in preventing stroke, and not to have any benefits in decreasing the rates of cardiovascular events. This recommendation does not apply if the patient has other indications for beta blockers (heart failure, arrhythmia, previous myocardial infarction, angina pectoris).

Strong
Low Downgraded for indirectness because only the meta-analysis included a subgroup analysis in older people 1 Cochrane review [17], 1 meta-analysis [33] and 1 evidence based guideline [38] Legend: RCT randomized controlled trial studies [44,46]. In Pepine et al. [34] atenolol was compared to a treatment with verapamil, in the MRC trial [35] and its related secondary analyses it was compared with hydrochlorothiazide plus amiloride (Moduretic©) or placebo [36,38,39,42], and in Coope et al. [34] atenolol was compared to no treatment. In Ruwald et al. [40], atenolol was compared to losartan plus hydrochlorothiazide and in the observational study of Collier et al. [44], an atenolol based treatment was compared with an amlodipine based treatment. The observational study of Testa et al. [46] compared atenolol to no-atenolol. The primary outcomes of these trials can be seen in Table 3.

Main findings
The results of the included studies are displayed in Tables 5 and 6 and summarized below.
Composite outcome of death, nonfatal stroke and nonfatal myocardial infarction The meta-analysis [33] found that no reduced event rates for the composite outcome were found in beta blocker treatment groups compared to placebo in trials enrolling older adults (relative risk RR 0.89, 95% confidence interval CI 0.75-1.05). But in the comparison with other antihypertensive agents, a higher risk of events was found in the beta blocker-group (RR 1.06, 95% CI 1.01-1.10). In the subgroup analysis of one of the included randomized controlled trials [34], no significant difference in the first occurrence of the composite outcome was found between the verapamiltreatment-group and the atenolol-treatment-group (RR 0.93, 95% CI 0.84-1.03). Ruwald et al. [40] analysed the combined endpoint of cardiovascular death, nonfatal stroke and nonfatal myocardial infarction and found that in patients ≥67 years losartan significantly reduced the risk of this combined endpoint compared to atenolol (hazard ratio HR 0.79, 95% CI 0.69-0.91).

Conclusion
Beta blockers should not be considered first-line therapy for older hypertensive patients without another indication for these agents; however, in younger patients beta blockers are associated with a significant reduction in cardiovascular morbidity and mortality.

Quality appraisal
Quality criteria for systematic reviews and metaanalyses   RCT To test the hypothesis, that the risk for adverse outcomes in older people with hypertension and coronary artery disease treated with a calcium antagonist based strategy or a non-calcium antagonist (atenolol) based strategy is equivalent.
Step   [43] found that the prescription of beta 1 selective or non-selective beta blockers did not affect mortality other than no treatment (for numbers, see Table 6). Further, the prescription of non-selective beta blockers compared to beta 1 selective beta blockers was associated with lower mortality in sex-adjusted models HR 0.62 (95% CI 0.41-0.95).

Cardiovascular eventsincluding stroke
The analysis of the included group of studies of the meta-analysis [33] states that in older adults, beta blockers were associated with statistically significant reductions in nonfatal stroke ( [47] analysed the composite endpoint of cardiovascular mortality and morbidity (see Table 3) and found, that there were no statistical   Regarding the endpoint of total stroke, Coope et al. [37] analysed data comparing beta blocker treatment to no treatment in a subgroup analysis involving older adults >70 and found a similar reduction in both groups. They further state that the sample was not large enough for the differences to be significant (numbers were not available from the paper). Ogihara et al. [41] states that benidipine plus beta blocker reduced fatal and non-fatal stroke less than benidipine plus thiazide diuretics (HR 2.74; 95% CI 1.08-6.96).

Cognitive impairment/status
Bird [36], a secondary analysis of the the MRC trial [35] found that beta blocker treatment of moderate raised blood pressure in older adults did not impair cognitive function or produce symptoms or behaviour changes, which would cause concern in confidants of treated patients. Gelber et al. [45] found that beta blocker use as the sole

Risk of bias -meta-analysis
A summary of risk of bias in the meta-analysis [33] is presented in Table 2. The main limitations contributing to a risk of bias were related to methodology issues. The literature search was done without using appropriate search strings and only two data bases were used, namely MEDLINE and the Cochrane library. A quality appraisal of included studies was lacking. The review process was unclear and a list of excluded studies with reason for exclusion was missing. Risk of publication bias was not assessed.

Risk of bias -Clinical trials
A summary of risk of bias for intervention studies is presented in Fig. 2. The main limitations contributing to risk of bias were related to the design (e.g. inadequate randomisation, blinding, attrition bias and selective reporting).
Other potential sources of bias were related to intentionto-treat analysis, contamination bias, sample size and power calculation. In two studies a high crossover rate between groups was registered [34,37]. Detailed information of the risk of bias is available in Additional file 4.

Risk of bias -Observational studies
Quality appraisal of the four observational studies is shown in Table 7. Three studies can be judged as being of fairly good quality. One study reported on most of the CASP items insufficiently and cannot be considered of high quality.

Additional references of interest for the development of recommendations
We found nine additional references that were taken into consideration for the development of the recommendations: three Cochrane reviews [17,49,50], one update of a Cochrane review [22], one meta-analysis [33], two evidence based recommendation paper [51,52], one evidence based guideline [38] and one randomized controlled trial Table 5 Summary of study findings of the meta-analysis (subgroup analysis) Author year Objective Kahn 2006 [33] To explore the efficacy (stroke, myocardial infarction and death) of beta blockers in "younger"(< 60 years) and "older" (≥ 60 years) patients.

Results:
1. Beta blockers compared to placebo or no treatment 1 (5 trials and n = 8019 patients, range mean age 65 to 75.7 years)  Prescription of selective or non-selective beta blockers did not affect mortality other than no treatment. Prescription of non-selective beta blockers was associated with lower mortality in sex-adjusted models Full regression model of the whole study sample adjusted for sex and all other covariates: non-selective beta blockers vs. beta 1-selective beta blockers HR 0.62 (95% CI: 0.41-0.95). Collier 2011 [44] Compared with the atenolol-based regimen, the amlodipine-based regimen reduced: -the relative risk of cardiovascular events more effectively in both age groups: by 17% in patients ≥65 years (hazard ratio: 0.83; 95% CI 0.75, 0.91; P < 0.01) and 15% in patients <65 years (hazard ratio 0.85; 95% CI 0.78, 0.95; P < 0.01) -cardiovascular mortality by 23% in the older group (hazard ratio 0.77; 95% CI 0.63, 0.94; P < 0.01) and by 24% in the younger group (hazard ratio 0.76; 95% CI 0.58, 1.00; P = 0.05) -fatal and nonfatal stroke by 30% in the older group (hazard ratio 0.70; 95% CI 0.59, 0.84; P < 0.01) and by nonsignificant 9% in the younger group (hazard ratio 0.91; 95% CI 0.71, 1.15; P = 0.42) --significant in total of coronary endpoints, nonfatal MI (excluding silent MI) and fatal CHD in the younger, but not in the older group Coope 1986 [37] Overall no significant difference in the total mortality was found neither in treatment nor in the control group. The rate of all deaths in the intervention group was 0.97 of that in the control group (95% CI 0.70-1.42).The rate of fatal stroke in the intervention group was 0.30 of that in the control group (95% CI 0.11-0.84) p < 0.025. Rate of all stroke in the intervention group was 0.58 of that in the control group (95% CI 0.35-0.96) p < 0.03. The subgroup analysis of patients by age (70-79 years) showed a similar reduction in total stroke in both groups, but the study was not large enough for these differences to be significant.
Gelber 2013 [45] Beta blocker use as the sole antihypertensive medication was associated with a lower risk of developing cognitive impairment (defined as a CASI [Cognitive Abilities Screening Instrument]) score < 74) compared with untreated men (IRR 0,69; 95% CI 0,50-0,94). Non-beta blocker drug combinations were also associated with a reduced risk (IRR 0,78; 95% CI 0,62-0,98). Cognitive decline (defined as a ≥ 9 point decrease in CASI score) occurred in 1167 men (53.1%). Beta blocker use was also associated with a trend toward a decreased risk of cognitive decline: model 2 IRR 0.78 (95% CI 0.61-1.00) for beta blocker use alone; 0.81 (95% CI 0.64-1.03) for beta blocker in combination with other drugs. None of the other drug categories was significantly associated with cognitive decline.
Pepine 2003 [34] No significant differences in primary outcome (first occurrence of all-cause death, nonfatal MI or nonfatal stroke) were seen between the calcium antagonist group and the non-calcium antagonist group (RR 0.98 CI 0.90-1.06). No significant differences were seen in these outcomes analysed individually. Ruwald 2012 [40] In this post-hoc analysis of the LIFE study, patients were divided in subgroups of aged 66 or younger and aged 67 or older.
MRC 1992 [35] Main trial Overall, the beta blocker group had significantly more withdrawals than the diuretic group for both suspected major side effects (beta blocker 30.2% (n = 333); diuretic 14.8% (n = 160); placebo 3.7% (n = 82), and inadequate blood pressure control (beta blocker n = 12; diuretic n = 1; placebo n = 175). After adjusting for baseline characteristics the diuretic group had significantly reduced risk of stroke (31% 95 CI 3% to 51%, p = 0.04), coronary events (44% 95 CI 21% to 60%, p = 0.0009), and all cardiovascular events (35% 95 CI 17% to 49%, p = 0.0005) compared with the placebo groups. The beta blocker group showed no significant reductions in these end points.  [53]. Details of the studies can be seen in Additional file 2. These references did not meet the inclusion criteria of our systematic review because of the age of the included participants. They were counted as being relevant to recommendations principally because they provided information about the risks and benefits of beta blockers for the management of hypertension in younger populations. Three of the additional references evaluated atenolol [50,53,54]. Atenolol was reported to be associated with an increased risk of stroke compared with other antihypertensive agents [54] and to be less effective at preventing cardiovascular morbidity and death than losartan [53]. However, in a systematic review no statistical differences were found between atenolol and placebo regarding the risk of fatal and non-fatal stroke, major vascular events, death from all causes, and death from cardiovascular causes [50]. The other five additional references evaluated beta blockers [17,38,49,51,52] in general. Antihypertensive treatment with either 25 mg hydrochlorothiazide and 2.5 mg amiloride or atenolol 50 mg for 9 months in a population aged between 65 and 74 years old with moderate raised blood pressure did not impair cognitive function, depression, or behavioural changes which would cause concern in confidants. The study did not find a linear relationship between blood pressure levels in their moderately elevated range and psychometric test scores, nor showed that lowering of blood pressure levels is associated with any impairment of performance on these tests Carr 2012 [42] This secondary analysis of the MRC trial found evidence that atenolol does not perform as well as the hydrochlorothiazide plus amiloride (Moduretic ©) in terms of stroke prevention: Moduretic 41.6 number of events /1000 patient years, beta blocker 50.8% and placebo 60.5% (statistical tests missing). For stroke, we found that after adjusting for current systolic blood pressure variability in systolic blood pressure over time, as measured by the standard residual or root successive variance, contained significant prognostic capability: the rate ratio associated with an increase of 1 standard deviation in the standard residual was 1.25 95% CI 0.86-1.81 and 1.16 95% CI 0.85-1.59 for the root successive variance.
Lever 1992 [48] In this article reporting on the MRC trial, the beta blocker group showed no significant differences to placebo regarding the outcomes stroke, coronary events, all cardiovascular events and all-cause mortality.
Lever 1993 [39] In this article reporting on the MRC trial, a RCT with 4396 patients (mean age 70.3 years) randomized to beta blocker, diuretic or placebo, the beta blocker group showed no significant differences to placebo regarding the outcomes stroke, coronary events, all cardiovascular events and all-cause mortality.
Although the risk for stroke was lower for beta blockers compared to placebo, there was an increase in stroke in beta blockers compared to calcium channel blockers and renin-angiotensin system inhibitors [17]. This finding is in the same direction as the results reported by Khan et al. [33] for the group of studies enrolling older patients. Total mortality and cardiovascular disease were also higher for beta blockers compared to calcium channel blockers [17]. Furthermore, it was reported that second line beta blockers reduced blood pressure but with little or no effect on pulse pressure compared to thiazides [46]. The Canadian Hypertension Education Program papers were the only sources that recommended beta blockers as first line therapy for patients <60 year with coronary artery disease [37]. These additional references provide preliminary evidence that certain risks like fatal and non-fatal stroke, major vascular events, cardiovascular morbidity, death from all causes, and death from cardiovascular causes may be increased with the use of beta blockers compared to other antihypertensive treatments. The uncertainties regarding these risks imply that independently from the age group beta blockers should be prescribed with caution for the treatment of hypertension.

Recommendations
We developed recommendations following a standardised schema and reflecting the strength and the quality of the evidence. From the results of our systematic review and the additional "references of interest" we were able to develop three recommendations in two meetings by AV (researcher and clinician), YVM (researcher) and ARG (researcher and geriatrician).AS participated in one of these as a senior clinician and researcher. The three recommendations were later discussed and confirmed with IK and MMV for their inclusion in the Comprehensive Medication Review (CMR) tool, which was developed within the PRIMA-eDS project.
The three recommendations we developed are about the discontinuation of beta blocker use in older adults with hypertension ( Table 1). The recommendations were considered as strong recommendations. The quality was downgraded to low for indirectness.

Discussion
This systematic review aimed to explore the effectiveness and safety of beta blockers for the management of hypertension in older adults (≥ 65 years). A further aim was to develop recommendations on when the use of beta blockers could be discontinued or its dose could be reduced in the treatment of hypertension in older adults.

Summary of main results
One meta-analysis, four randomized controlled trials, six secondary analyses (post hoc analyses) of randomized controlled trials, and four observational studies were included in our systematic review. The included studies were heterogeneous in terms of study designs, interventions, settings, participants, reporting and definitions of  The included studies reported on mortality, cardiovascular events including stroke, cognitive impairment or cognitive status and adverse events. We did not find studies analysing the effect on quality of life, life expectancy, hospitalisation, functional impairment or functional status, renal failure, or safety endpoints. The results of the included studies suggest that beta blockers appear to increase the risk of death, stroke and myocardial infarction analysed together as a composite outcome if compared to other antihypertensive agents [31]. Further, beta blockers showed no benefit compared to other antihypertensive agents or placebo regarding the single outcome of mortality. They appear to be less effective than other antihypertensive agents in reducing cardiovascular events. Regarding the endpoint of cognitive impairment/decline, there seems to be a trend towards a decreased risk associated with beta blocker use versus no treatment. Regarding the single endpoint of stroke we found contradictory results. In addition to the results of the effectiveness of beta blockers identified in the present review, also some issues regarding medication safety have to be risen. First, high cross-over-rates occurred in two studies [32,34], where a high cross over from the beta blocker group to the comparison group occurred. Pepine et al. [32] gave the following possible reasons for this: physician bias against beta blockers, and suspected higher metabolic complications or previous intolerances against beta blockers.
Second, in another study [33], withdrawal of patients due to suspected major side effects and inadequate blood pressure control in the beta blocker group was significantly higher than in the comparison group.
Regarding the quality of the included studies and the interpretation of the results, a major limitation was that the results are partly based on subgroup analyses (3 out of 5 studies) [31,32,34] and secondary analyses of randomized controlled trials [36,[39][40][41][42]48]. Subgroup analyses and secondary analyses in general are used for generating hypotheses and not for testing hypotheses. The included meta-analysis has some major limitations and it can be argued that the systematic review on which it is based does not fulfil the criteria for a systematic review [19]. Only three of the 11 studies included in the subgroup analysis of older people in the meta-analysis were considered eligible for our review. The reporting of the participants' characteristics was very poor with regard to their comorbidities, use of concomitant drugs, frailty, and cognitive status. Thus, the generalizability of the results to the heterogeneous group of older people is limited as the evidence is indirect. However, it could be expected that concerns with regard to safety and benefits of beta blockers for the management of hypertension identified in the present study may be even higher for frailer older people.

Strengths and limitations
We conducted this systematic review following an adaption of the standard methodology recommended by the Cochrane collaboration [19] and the PRISMA statement [20]. The search strategy comprised a stepwise approach searching first for systematic reviews and meta-analysis (searches 1 and 2), and then for additional relevant controlled intervention and observational studies (searches 3A and 3B). We restricted search 3B to the period 2011-2016, as earlier literature had already been covered in the process of searches 1, 2 and 3A. Taking into account that the new update of the Cochrane Review of Wiysonge published in 2017 [22] did not identify any additional studies that would be eligible for our review (and the review does not apply any age restrictions), it is very unlikely that any eligible studies have been missed by our search strategy. We excluded a large number of studies because they were either not focussing on beta blockers or they had been carried out in younger populations (< 65 years). Our search strategy included specific terms for older adults. This might have limited our ability to capture all studies that include evidence on older adults carrying a risk of missing studies like in a rapid review. But we believe our stepwise approach overcomes some of the limitations of a rapid review, for which there is a lack of published guidelines or explicit methods [51,52]. Unlike our review, some rapid reviews do not have clear research questions including participants, interventions, comparisons, outcomes, and study design (PICOS) [51].
Three of the four individual randomized controlled trials we included [34,35,37] also formed part of the included subgroup analysis of the meta-analysis [33]. Although this does carry a risk of "double-counting", we decided to include both in our systematic review as the overlap was small (3 out of 11 trials in the meta-analysis) and the metaanalysis did not examine these studies to the same level of detail as we are doing in our systematic review.
Two of the secondary analyses [39,48] did not really provide any new data and they also could have been treated as duplicates but as they formally met our inclusion criteria we included them as secondary analyses of the MRC study.
In our systematic review we only found studies specifically addressing Atenolol. It therefore remains unclear whether the results of these studies are also applicable for newer beta blockers such as metoprolol or bisoprolol.

Implications for future research
Older adults comprise the great majority of consumers of drugs used in the treatment of chronic diseases such as hypertension. Yet, there has been little work to systematically assess the research evidence for usage in this population (≥65 years with polypharmacy and/or multimorbidity). Research groups should focus on clinically relevant outcomes i.e. mortality, cardiovascular diseases (including stroke) or hospital admissions (drug related) whenever applicable, as these are critical in the context of beta blocker treatment in hypertension. We also want to mention that there are a number of other important endpoints, e.g. quality of life, functional and cognitive status which should be considered especially in studies on older populations. A proper power calculation and definition of a primary endpoint (single or composite) is important to show an effect and to avoid multiple testing. In most studies important information is lacking in the methods section. Study authors should provide a clear and comprehensive description of the population (including comorbidities), study design and intervention. We highly recommend researchers to use the Consolidated Standards of Reporting Trials statement (CONSORT) [55] to improve the quality of reporting of randomized controlled trials.

Conclusions
This is one of the first systematic reviews exploring the risks and benefits of beta blockers in treating hypertension in aged populations (>65 years) on patient relevant outcomes. This study highlights the limited number of studies that have been carried out on this issue. Furthermore, the quality of current evidence to interpret the benefits of beta blockers in hypertension is generally weak. It cannot be recommended to use beta blockers in older adults as a first line agent for hypertension. Further studies of good quality should analyse the risks and benefits of beta blockers in older people with hypertension taking into account their comorbidities, use of concomitant medications, functional and cognitive status, and evaluating clinically relevant outcomes for this population.