Scolaris Content Display Scolaris Content Display

Base de datos Cochrane de Revisiones Sistemáticas Revisión - Intervención

Beta‐blockers and inhibitors of the renin‐angiotensin aldosterone system for chronic heart failure with preserved ejection fraction

Declaraciones de intereses de los autores

Versión publicada: 22 mayo 2021 Historial de versiones

https://doi.org/10.1002/14651858.CD012721.pub3
Contraer todo Desplegar todo

Background

Beta‐blockers and inhibitors of the renin‐angiotensin‐aldosterone system improve survival and reduce morbidity in people with heart failure with reduced left ventricular ejection fraction (LVEF); a review of the evidence is required to determine whether these treatments are beneficial for people with heart failure with preserved ejection fraction (HFpEF).

Objectives

To assess the effects of beta‐blockers, angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers, angiotensin receptor neprilysin inhibitors, and mineralocorticoid receptor antagonists in people with HFpEF.

Search methods

We updated searches of CENTRAL, MEDLINE, Embase, and one clinical trial register on 14 May 2020 to identify eligible studies, with no language or date restrictions. We checked references from trial reports and review articles for additional studies. 

Selection criteria

We included randomised controlled trials with a parallel group design, enrolling adults with HFpEF, defined by LVEF greater than 40%.

Data collection and analysis

We used standard methodological procedures expected by Cochrane.

Main results

We included 41 randomised controlled trials (231 reports), totalling 23,492 participants across all comparisons. The risk of bias was frequently unclear and only five studies had a low risk of bias in all domains.

Beta‐blockers (BBs)

We included 10 studies (3087 participants) investigating BBs. Five studies used a placebo comparator and in five the comparator was usual care. The mean age of participants ranged from 30 years to 81 years.

A possible reduction in cardiovascular mortality was observed (risk ratio (RR) 0.78, 95% confidence interval (CI) 0.62 to 0.99; number needed to treat for an additional benefit (NNTB) 25; 1046 participants; three studies), however, the certainty of evidence was low. There may be little to no effect on all‐cause mortality (RR 0.82, 95% CI 0.67 to 1.00; 1105 participants; four studies; low‐certainty evidence). The effects on heart failure hospitalisation, hyperkalaemia, and quality of life remain uncertain.

Mineralocorticoid receptor antagonists (MRAs)

We included 13 studies (4459 participants) investigating MRA. Eight studies used a placebo comparator and in five the comparator was usual care. The mean age of participants ranged from 54.5 to 80 years.

Pooled analysis indicated that MRA treatment probably reduces heart failure hospitalisation (RR 0.82, 95% CI 0.69 to 0.98; NNTB = 41; 3714 participants; three studies; moderate‐certainty evidence). However, MRA treatment probably has little or no effect on all‐cause mortality (RR 0.91, 95% CI 0.78 to 1.06; 4207 participants; five studies; moderate‐certainty evidence) and cardiovascular mortality (RR 0.90, 95% CI 0.74 to 1.11; 4070 participants; three studies; moderate‐certainty evidence). MRA treatment may have little or no effect on quality of life measures (mean difference (MD) 0.84, 95% CI ‐2.30 to 3.98; 511 participants; three studies; low‐certainty evidence). MRA treatment was associated with a higher risk of hyperkalaemia (RR 2.11, 95% CI 1.77 to 2.51; number needed to treat for an additional harmful outcome (NNTH) = 11; 4291 participants; six studies; high‐certainty evidence).

Angiotensin‐converting enzyme inhibitors (ACEIs)

We included eight studies (2061 participants) investigating ACEIs. Three studies used a placebo comparator and in five the comparator was usual care. The mean age of participants ranged from 70 to 82 years.

Pooled analyses with moderate‐certainty evidence suggest that ACEI treatment likely has little or no effect on cardiovascular mortality (RR 0.93, 95% CI 0.61 to 1.42; 945 participants; two studies), all‐cause mortality (RR 1.04, 95% CI 0.75 to 1.45; 1187 participants; five studies) and heart failure hospitalisation (RR 0.86, 95% CI 0.64 to 1.15; 1019 participants; three studies), and may result in little or no effect on the quality of life (MD ‐0.09, 95% CI ‐3.66 to 3.48; 154 participants; two studies; low‐certainty evidence). The effects on hyperkalaemia remain uncertain.

Angiotensin receptor blockers (ARBs)

Eight studies (8755 participants) investigating ARBs were included. Five studies used a placebo comparator and in three the comparator was usual care. The mean age of participants ranged from 61 to 75 years.

Pooled analyses with high certainty of evidence suggest that ARB treatment has little or no effect on cardiovascular mortality (RR 1.02, 95% 0.90 to 1.14; 7254 participants; three studies), all‐cause mortality (RR 1.01, 95% CI 0.92 to 1.11; 7964 participants; four studies), heart failure hospitalisation (RR 0.92, 95% CI 0.83 to 1.02; 7254 participants; three studies), and quality of life (MD 0.41, 95% CI ‐0.86 to 1.67; 3117 participants; three studies). ARB was associated with a higher risk of hyperkalaemia (RR 1.88, 95% CI 1.07 to 3.33; 7148 participants; two studies; high‐certainty evidence).

Angiotensin receptor neprilysin inhibitors (ARNIs)

Three studies (7702 participants) investigating ARNIs were included. Two studies used ARBs as the comparator and one used standardised medical therapy, based on participants' established treatments at enrolment. The mean age of participants ranged from 71 to 73 years.

Results suggest that ARNIs may have little or no effect on cardiovascular mortality (RR 0.96, 95% CI 0.79 to 1.15; 4796 participants; one study; moderate‐certainty evidence), all‐cause mortality (RR 0.97, 95% CI 0.84 to 1.11; 7663 participants; three studies; high‐certainty evidence), or quality of life (high‐certainty evidence). However, ARNI treatment may result in a slight reduction in heart failure hospitalisation, compared to usual care (RR 0.89, 95% CI 0.80 to 1.00; 7362 participants; two studies; moderate‐certainty evidence). ARNI treatment was associated with a reduced risk of hyperkalaemia compared with valsartan (RR 0.88, 95% CI 0.77 to 1.01; 5054 participants; two studies; moderate‐certainty evidence).

Authors' conclusions

There is evidence that MRA and ARNI treatment in HFpEF probably reduces heart failure hospitalisation but probably has little or no effect on cardiovascular mortality and quality of life. BB treatment may reduce the risk of cardiovascular mortality, however, further trials are needed. The current evidence for BBs, ACEIs, and ARBs is limited and does not support their use in HFpEF in the absence of an alternative indication. Although MRAs and ARNIs are probably effective at reducing the risk of heart failure hospitalisation, the treatment effect sizes are modest. There is a need for improved approaches to patient stratification to identify the subgroup of patients who are most likely to benefit from MRAs and ARNIs, as well as for an improved understanding of disease biology, and for new therapeutic approaches.  

PICO

Population
Intervention
Comparison
Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Beta‐blockers and inhibitors of the renin‐angiotensin aldosterone system for chronic heart failure with preserved ejection fraction

Review question

We investigated the effects of beta‐blockers (BBs), mineralocorticoid receptor antagonists (MRAs), angiotensin converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs) and angiotensin receptor neprilysin inhibitors (ARNIs) on survival, hospital admissions for heart failure, quality of life and potassium levels in people with heart failure with preserved ejection fraction (HFpEF).

Background

Heart failure is a common condition that occurs when the function of the heart muscle is impaired, being associated with symptoms of breathlessness and fatigue, and a reduction in survival. In around half of cases of heart failure, where the left ventricular ejection fraction is reduced to less than 40% (reflecting significant impairment of contractile function), several drug treatments are known to be effective at improving survival and reducing hospitalisation. In the remaining cases, where the ejection fraction is normal or only mildly reduced (HFpEF), it is not clear whether the same drug treatments are effective at improving outcomes.

Selection criteria

We sought to investigate whether treatments for heart failure with reduced ejection fraction are also effective in HFpEF. We conducted a comprehensive search for all trials investigating BBs, MRAs, ACEIs, ARBs or ARNIs (evidence current to 14 May 2020).

Results and conclusions

We included 10 studies with 3087 randomised participants for BBs, 13 studies with 4459 randomised participants for MRAs, eight studies with 2061 randomised participants for ACEIs, eight studies with 8755 randomised participants for ARBs and three studies with 7702 randomised participants for ARNIs. We combined the evidence in a pooled analysis for each drug class and for each of the outcomes assessed. Not all included studies are part of each analysis.

We found that BBs may improve cardiovascular mortality. However, the certainty of evidence was low due to small trials and uncertainty about the methods used. For MRAs, the results suggest a reduction in heart failure hospitalisation but little or no effect on cardiovascular and all‐cause mortality; however, the certainty of evidence was only moderate. For ACEIs, treatment probably has little or no effect on the outcomes of cardiovascular mortality, all‐cause mortality and heart failure hospitalisation; however, the certainty of evidence was only moderate. We found high certainty of evidence for ARB treatment, with the results suggesting little or no effect. We found that ARNI treatment has little or no effect on cardiovascular mortality (moderate‐certainty evidence), all‐cause mortality (high‐certainty evidence), or quality of life (high‐certainty evidence). ARB treatment may reduce slightly heart failure hospitalisations (moderate‐certainty evidence). Treatment with MRAs and ARBs was found to increase the risk of high potassium in the blood.

In conclusion, treatment with MRAs, and possibly ARNIs, was found to result in a slight reduction in the risk of hospitalisation due to heart failure. There was some evidence of a possible beneficial effect of BB on mortality due to cardiovascular disease. Treatment with ACEI probably has no beneficial effect in people with HFpEF, however, this remains uncertain due to a lack of evidence from clinical trials. For ARBs, the evidence suggested that treatment is of little or no benefit in people with HFpEF.

Certainty of the evidence

The certainty of evidence ranged from high to low across the outcomes and drug classes studied. With the exception of ARBs and ARNIs, there was a lack of large‐scale trials in HFpEF for the interventions and outcomes tested.

Authors' conclusions

Implications for practice

Our findings suggest that MRA and ARNI treatments have a modest beneficial effect on the risk of heart failure hospitalisation and that BB may reduce the risk of cardiovascular mortality. We did not find evidence supporting an important beneficial effect of ARBs (high‐certainty evidence), or ACEIs (moderate‐certainty evidence) on mortality and hospitalisation outcomes. Although we were not able to stratify our analysis by LVEF, we note evidence from individual‐participant level meta‐analyses that suggest benefit from BB, MRA, and ARNI treatment in heart failure patients with LVEF 40 to 49%, in addition to those with LVEF < 40%. For all comparisons, no effect on the quality of life was observed, although the certainty of the evidence was low. 

Implications for research

This review highlights persistent uncertainty concerning the effects of BBs, MRAs, ACEIs, and ARNIs in HFpEF, defined as LVEF greater than 40%. Although we were unable to explore LVEF as an effect modifier in this study‐level review and meta‐analysis, reports of individual participant‐level pooled analyses of BBs and RAAS inhibition suggest that beneficial effects may extend beyond patients with reduced ejection fraction (LVEF less than 40%) heart failure, to include patients with heart failure with mid‐range ejection fraction (LVEF 40% to 49%; HFmrEF). It is possible that heart failure patients with normal or increased LVEF have distinct aetiopathology. For example, a recent study found that one in seven patients with HFpEF had evidence of wild‐type transthyretin amyloidosis (Gonzalez‐Lopez 2015). The application of precision phenotyping at a population‐scale may yield important information about the disease mechanisms underlying onset and progression in these patients, and provide the basis for the development of more effective therapeutic approaches.

Summary of findings

Open in table viewer
Summary of findings 1. Beta‐blockers compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Beta‐blockers compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: beta‐blockers
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo or no treatment

Risk with beta‐blockers

Cardiovascular mortality (RR)
follow‐up: range 21 months to 3.2 years

Study population

RR 0.78
(0.62 to 0.99)

1046
(3 RCTs)

⊕⊕⊝⊝
LOW 1 2

Three additional studies (ELANDD; SWEDIC; Takeda 2004) reported that no deaths occurred

173 per 1000

135 per 1000
(107 to 171)

Heart failure hospitalisation (RR)
follow‐up: range 6 months to 3.2 years

Study population

RR 0.73
(0.47 to 1.13)

449
(4 RCTs)

⊕⊝⊝⊝
VERY LOW 1 3

Follow‐up unclear for SWEDIC. ELANDD reported that no hospitalisation due to heart failure occurred

117 per 1000

86 per 1000
(55 to 133)

Hyperkalaemia

follow‐up: mean 3.2 years

245 (1 RCT)

⊕⊝⊝⊝
VERY LOW1 6

J‐DHF reported one participant in the intervention group (N = 120) experienced hyperkalaemia but did not report on this outcome for the control group. No further data were available from any of the other studies.

All‐cause mortality (RR)
follow‐up: range 21 months to 3.2 years

Study population

RR 0.82
(0.67 to 1.00)

1105
(4 RCTs)

⊕⊕⊝⊝
LOW 1 2

Follow‐up unclear for Adamyan 2010. ELANDD, SWEDIC and Takeda 2004 reported that no deaths occurred.

243 per 1000

199 per 1000
(163 to 243)

Quality of life (MLHFQ): from 0 to 105
follow‐up: mean 6 months

Mean quality of life (MLHFQ) was 24

MD 1 lower
(9.05 lower to 7.05 higher)

‐

93
(1 RCT)

⊕⊝⊝⊝
VERY LOW 4 5

Lower = better, 5 point difference considered to be clinically meaningful

*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; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; RCT: Randomised controlled trial; 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

1Downgraded by one level due to study limitations (unclear selection bias in most studies).

2Downgraded by one level due to imprecision (concerns about the smaller study being more precise than the larger study).

3Downgraded by two levels due to imprecision (few events and wide CI).

4Downgraded by two levels due to imprecision (very small sample size).

5Downgraded by one level due to suspected publication bias (this is a patient‐relevant outcome that is not reported in most studies).

6Downgraded by two levels due to suspected publication bias (incomplete reporting).

Open in table viewer
Summary of findings 2. Mineralocorticoid receptor antagonists (MRAs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

MRAs compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: MRAs
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo or no treatment

Risk with MRAs

Cardiovascular mortality (RR)
follow‐up: range 12 months to 3.3 years

Study population

RR 0.90
(0.74 to 1.11)

4070
(3 RCTs)

⊕⊕⊕⊝
MODERATE 1

Two additional trials (RAAM‐PEF; Kurrelmeyer 2014) reported that no deaths occurred

88 per 1000

79 per 1000
(65 to 97)

Heart failure hospitalisation (RR)
follow‐up: range 24 weeks to 3.3 years

Study population

RR 0.82
(0.69 to 0.98)

3714
(3 RCTs)

⊕⊕⊕⊝
MODERATE 1

Three additional trials (ALDO‐DHF; Kurrelmeyer 2014; Upadhya 2017) reported that no hospitalisation due to heart failure occurred

136 per 1000

112 per 1000
(94 to 134)

Hyperkalaemia
follow‐up: range 24 weeks to 3.3 years

Study population

RR 2.11
(1.77 to 2.52)

4291
(6 RCTs)

⊕⊕⊕⊕
HIGH

Two trials defined hyperkalaemia ≥ 5.5 mEg/L

83 per 1000

175 per 1000
(146 to 208)

All‐cause mortality
follow‐up: range 9 months to 3.3 years

Study population

RR 0.91
(0.78 to 1.06)

4207
(5 RCTs)

⊕⊕⊕⊝
MODERATE 1

Two additional trials (RAAM‐PEF; Kurrelmeyer 2014) reported that no deaths occurred

133 per 1000

121 per 1000
(104 to 141)

Quality of life (MLHFQ): from 0 to 105
follow‐up: range 9 months to 12 months

Mean quality of life (MLHFQ) ranged from 20 to 25

MD 0.84 higher
(2.30 lower to 3.98 higher)

‐

511
(3 RCTs)

⊕⊕⊝⊝
LOW 2 3

Lower = better, 5 points are considered a clinically significant difference

We did not pre‐specify which QoL scale was to be reported in the 'Summary of findings' table. To aid comparisons among 'Summary of findings' tables we chose to include the Minnesota Living with Heart Failure questionnaire and not the SMD across two scales

*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; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; MRAs: Mineralocorticoid receptor antagonists; RCT: Randomised controlled trial; 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

1Downgraded by one level due to imprecision.

2Downgraded by one level due to study limitations (one trial was open label).

3Downgraded by one level due to imprecision (small sample size).

Open in table viewer
Summary of findings 3. Angiotensin‐converting enzyme inhibitors (ACEIs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

ACEIs compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: ACEIs
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo/no treatment

Risk with ACEI

Cardiovascular mortality (RR)
follow‐up: range mean 12 months to mean 26.2 months

Study population

RR 0.93
(0.61 to 1.42)

945
(2 RCTs)

⊕⊕⊕⊝
MODERATE 1

One additional trial (Kitzman 2010) reported that no deaths occurred

86 per 1000

81 per 1000
(53 to 123)

Heart failure hospitalisation (RR)
follow‐up: range 6 months to 26.2 months

Study population

RR 0.86
(0.64 to 1.15)

1019
(3 RCTs)

⊕⊕⊕⊝
MODERATE 1

13 per 1000

11 per 1000
(8 to 15)

Hyperkalaemia

follow‐up: 6 months

74 (1 RCTs)

⊕⊝⊝⊝
VERY LOW 1 3 4

One trial (Zi 2003) reported 2 events in the intervention group (N = 36), 0 events in the control group (N = 38) (RR 5.27, 95% CI 0.26 to 106.16)

All‐cause mortality (RR)
follow‐up: range mean 6 months to mean 26.2 months

Study population

RR 1.04
(0.75 to 1.45)

1187
(5 RCTs)

⊕⊕⊕⊝
MODERATE 1

One additional trial (Kitzman 2010) reported that no deaths occurred

102 per 1000

106 per 1000
(77 to 148)

Quality of life (MLHFQ): from 0 to 105
follow‐up: mean 12 months

Mean quality of life (MLHFQ) ranged from 10.9 to 29

MD 0.09 lower
(3.66 lower to 3.48 higher)

‐

154
(2 RCTs)

⊕⊕⊝⊝
LOW 2 3

Scale: 0 to 105, lower = better, 5 point difference considered clinically relevant

One trial (SNEGOVIK) reported mean change from baseline of ‐19.8 for intervention and ‐10.7 for control

*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).
ACEIs: Angiotensin‐converting enzyme inhibitors; CI: Confidence interval; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; RCT: Randomised controlled trial; 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

1Downgraded by one level due to imprecision (wide CI).

2Downgraded by one level due to study limitations (risk of bias (open label)).

3Downgraded by one level due to imprecision (low sample size).

4Downgraded by one level due to study limitations (unclear selection bias).

Open in table viewer
Summary of findings 4. Angiotensin receptor blockers (ARBs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

ARBs compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: ARBs
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo or no treatment

Risk with ARBs

Cardiovascular mortality (RR)
follow‐up: range mean 12 months to mean 49.5 months

Study population

RR 1.02
(0.90 to 1.14)

7254
(3 RCTs)

⊕⊕⊕⊕
HIGH

One additional trial (Parthasarathy 2009) reported that no deaths occurred

131 per 1000

133 per 1000
(118 to 149)

Heart failure hospitalisation (RR)
follow‐up: range mean 12 months to mean 49.5 months

Study population

RR 0.92
(0.83 to 1.02)

7254
(3 RCTs)

⊕⊕⊕⊕
HIGH

171 per 1,‐000

157 per 1,‐000
(142 to 174)

Hyperkalaemia
follow‐up: range 36.6 months to 49.5 months

Study population

RR 1.88
(1.07 to 3.33)

7148
(2 RCTs)

⊕⊕⊕⊕
HIGH

3 per 1,000

5 per 1,000
(3 to 8)

All‐cause mortality (RR)
follow up: range 1 years to 4.4 years

Study population

RR 1.01
(0.92 to 1.11)

7964
(4 RCTs)

⊕⊕⊕⊕
HIGH

One additional trial (Parthasarathy 2009) reported that no deaths occurred

72 per 1000

73 per 1,‐000
(66 to 80)

Quality of life (MLHFQ): from 0 to 105
follow‐up: range mean 13.8 weeks to mean 49.5 months

Mean quality of life (MLHFQ) ranged from 10.9 to 31.6

MD 0.41 higher
(0.86 lower to 1.67 higher)

‐

3117
(3 RCTs)

⊕⊕⊕⊕
HIGH

Scale: 0 to 105, lower = better, 5 point difference considered clinically relevant

*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).
ARBs: Angiotensin receptor blockers; CI: Confidence interval; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; RCT: Randomised controlled trial; 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

Open in table viewer
Summary of findings 5. Angiotensin receptor neprilysin inhibitors (ARNIs) compared to usual care for chronic heart failure with preserved ejection fraction

ARNIs compared to usual care for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: ARNI (sacubitril‐valsartan)
Comparison: ARB (valsartan)

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with ARNI

Cardiovascular mortality (RR)

Median follow‐up: 35 months

Study population

RR 0.96
(0.79 to 1.15)

4796
(1 study)

⊕⊕⊕⊝
MODERATE1

89 per 1,000

85 per 1,000
(70 to 102)

Heart failure hospitalisation, first (RR)

Range of follow‐up: 24 weeks to 35 months

Study population

RR 0.89

(0.80 to 1.00)

7362

(2 studies)

⊕⊕⊕⊝
MODERATE1

142 per 1,000

126 per 1,000

(113 to 142)

Hyperkalaemia

Range of follow‐up: 36 weeks to 35 months

Study population

RR 0.88
(0.77 to 1.01)

5054
(2 studies)

⊕⊕⊕⊝
MODERATE1

147 per 1,000

129 per 1,000
(113 to 148)

All‐cause mortality (RR)

Range of follow‐up: 36 weeks to 35 months

Study population

(RR 0.97

CI 0.84 to 1.11)

7663
(3 studies)

⊕⊕⊕⊕
HIGH

138 per 1,000

134 per 1,000
(117 to 153)

Quality of life

Range of follow‐up: 36 weeks to 35 months

PARAMOUNT reported change from baseline for the KCCQ overall summary score for the intervention arm (n =118) as 11.25 (2.185) and the control arm (n = 116) as 11.31 (2.183) and summarised the findings as "no difference in KCCQ score between treatment groups".

PARAGON‐HF reported a difference of the clinical summary score KCCQ between treatment arms as 1.0 (0.0 to 2.1).

PARALLAX reported "KCCQ improved in both treatment groups, with an early benefit of S/V that was no longer significant after 24 week".

⊕⊕⊕⊕
HIGH

*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).

ARB: Angiotensin receptor blocker; ARNIs: Angiotensin receptor neprilysin inhibitors; CI: Confidence interval; KCCQ: Kansas City Cardiomyopathy Questionnaire; MD: Mean difference; RCT: Randomised controlled trial; 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

1Downgraded by one level due to imprecision.

Background

Description of the condition

Heart failure is a clinical syndrome characterised by breathlessness and fatigue that results when abnormalities of cardiac structure and function lead to inadequate cardiac output, elevated ventricular filling pressures, or both (Ponikowski 2016). Based on available data from the United States and Europe, the prevalence of heart failure is estimated to range from 1% to 12% of the adult population and is projected to increase with populations ageing and with improved survival from cardiovascular disease (Roger 2013). Heart failure represents a significant public health problem, accounting for 5% of emergency medical admissions to hospital in the United Kingdom (NICE 2018). It is associated with significant mortality risk, with a ten‐year survival of 27%, compared with 75% in the general population, matched for age and sex (Taylor 2012). Heart failure is classified according to the left ventricular ejection fraction (LVEF) into two categories: heart failure with reduced ejection fraction (HFrEF, typically considered as LVEF less than 40%), and heart failure with preserved ejection fraction (HFpEF, typically LVEF greater than 40%). Recently, an intermediate subgroup was defined by the European Society of Cardiology as heart failure with mid‐range ejection fraction (HFmrEF) defined as LVEF 40% to 49% (Ponikowski 2016). This was defined by the American College of Cardiology as borderline HFpEF, defined as LVEF 41% to 49% (Yancy 2013). In this review, we defined HFpEF as LVEF greater than 40% because completed and ongoing HFpEF trials have used a range of LVEF cut‐offs, between 40% and 50%. HFpEF accounts for approximately half of all cases of HF; mortality outcomes are similar to those for HFrEF (Gerber 2015).

Description of the intervention

Neurohumoral inhibition with beta‐blockers (BB), angiotensin converting enzyme inhibitors (ACEIs), and mineralocorticoid receptor antagonists (MRAs) leads to improved survival and a reduction in hospitalisations for heart failure in people with HFrEF (CIBIS Investigators 1999; Consensus Trial Study Group 1987; Flather 2005; Hjalmarson 2000; Kotecha 2014; MERIT‐HF Study Group 1999; Packer 1999; Packer 2002; Packer 2001; Pitt 1999; Ponikowski 2016; SOLVD Investigators 1991; SOLVD Investigators 1992; Zannad 2011). Angiotensin receptor antagonists (ARBs) are recommended as an alternative to ACEIs in patients with intolerance to these agents or, in combination with ACEIs, for patients treated with BBs who are unable to tolerate MRAs (Ponikowski 2016). Angiotensin receptor neprilysin inhibitors (ARNIs) are recommended as an alternative to ACEIs, with superior efficacy in people with HFrEF who remain symptomatic despite optimal therapy (McMurray 2014). Although neurohumoral activation is observed in HFpEF (Hogg 2005), comparatively fewer clinical trials of neurohumoral inhibitor therapies have been performed in this population. The existing evidence from individual trials of BBs, MRAs, ACEIs, ARBs or MRAs in people with HFpEF does not support a reduction in mortality with these treatments (Ponikowski 2016). However, limited evidence indicates that candesartan (an ARB) (Yusuf 2003) and spironolactone (an MRA) (Pitt 2014) may be effective in reducing numbers of people hospitalised with HF.

This review sought to determine whether neurohumoral inhibition with therapies that improve mortality and morbidity in those with HFrEF (BBs, MRAs, ACEIs, ARBs, and ARNIs) have similar benefit in people with HFpEF.

How the intervention might work

In people with HFpEF, inadequate cardiac function triggers compensatory neurohumoral responses similar to those observed in HFrEF (Hogg 2005). Activation of the renin‐angiotensin aldosterone system (RAAS) and increased tone of the sympathetic nervous system may be adaptive in the short term; however, chronic activation is likely to be detrimental. Pre‐clinical disease models of HFpEF suggest that RAAS activation leads to maladaptive hypertrophy and fibrosis (Sharma 2014). ACEIs, ARBs or MRAs inhibit components of the RAAS system to counter the over‐activation that occurs in people with HF. ARNIs combine inhibition of RAAS with an ARB (valsartan) with augmentation of the natriuretic peptide system by inhibition of neprilysin (sacubitril). Neprilysin is a neutral endopeptidase that degrades a number of endogenous vasoactive peptides serving to counteract some of the effects of RAAS activation (McMurray 2014). The beneficial effects of BB therapy in people with HFrEF are thought to be mediated reducing the downstream effects associated with increased tone of the sympathetic nervous system, including increased heart rate, adverse myocardial energetics, and stimulation of RAAS (Sackner‐Bernstein 1995). These mechanisms may also be important in HFpEF and the effect of BBs to increase diastolic filling time may be particularly important (Sharma 2014); however, it is also possible that BB treatment may have adverse effects in this patient population. The population of people with HFpEF is heterogeneous, both with respect to disease aetiology and comorbidity. However, it is possible that neurohumoral activation represents a common pathophysiological mechanism that could be successfully targeted to improve clinical outcomes in people with heart failure across the spectrum of LVEF.

Why it is important to do this review

There is uncertainty as to whether BBs or RAAS inhibitors are effective at reducing mortality and heart failure hospitalisation and improving quality of life in people with HFpEF. Guidelines offer no specific treatment recommendations regarding the use of these therapies beyond the management of comorbidities, aside from a weak recommendation that ARBs and MRA might be considered to reduce hospitalisations (Yancy 2017). The European Society of Cardiology guidelines for the diagnosis and treatment of acute and chronic heart failure highlight a gap in the evidence for the effects of ARNIs and BBs in the treatment of HFpEF (Ponikowski 2016).

A recent systematic review and meta‐analysis of pharmacotherapy in HFpEF included BBs and RAAS inhibitors (ACEIs, ARBs and MRAs) and suggested a reduction in cardiovascular and all‐cause mortality with BB therapy (Zheng 2018). An updated review with a more comprehensive search strategy is needed to inform new guideline recommendations and to inform the conduct of further clinical trials.

Objectives

To assess the effects of BBs, ACEIs, ARBs, ARNIs, and MRAs in people with HFpEF.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) with parallel group design. We excluded cross‐over trials because we considered these to be inappropriate for our review question due to the progressive nature of HF.

Studies published in full‐text or as abstracts, or available only as unpublished data, were eligible for inclusion.

Types of participants

We included studies with adult participants (aged ≥ 18 years) with HFpEF defined by a LVEF greater than 40%. We recognised that there was likely to be significant heterogeneity among study populations, relating to the disease definition, and we summarise this narratively in the Discussion. In relation to ejection fraction, we contacted study authors to obtain data on the subgroup of interest for studies with mixed populations.

Types of interventions

We performed separate meta‐analyses of studies that compared BBs, MRAs, ACEIs, ARBs or ARNIs, in addition to standard care. For BBs, MRAs, ACEIs and ARBs, the comparator was with placebo or no‐treatment control. For ARNIs, we meta‐analysed studies that compared sacubitril‐valsartan with valsartan, an ARB.

Types of outcome measures

Reporting one of more of the listed outcomes in the trial was not an inclusion criterion for the review. We assessed outcomes at the longest reported follow‐up.

Primary outcomes

  1. Cardiovascular mortality.

  2. Heart failure hospitalisation (number of participants with at least one hospitalisation, analysed as risk ratios (RR), and time to first event, analysed as hazard ratios (HR)).

  3. Hyperkalaemia.

Secondary outcomes

  1. All‐cause mortality.

  2. Quality of life (measured with either the 'Minnesota Living with Heart Failure Questionnaire' (MLHFQ) or the 'Kansas City Cardiomyopathy Questionnaire' (KCCQ)).

  3. Withdrawal due to adverse event (hypotension, hyperkalaemia or renal impairment).

Search methods for identification of studies

Electronic searches

We updated the systematic searches of the following bibliographic databases on 14 May 2020:

  1. Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library (Issue 5 of 12, 2020);

  2. Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily and MEDLINE (Ovid, 1946 to 12 May 2020); and

  3. Embase and Embase Classic (Ovid, 1947 to 13 May 2020) .

The search strategies used in 2017 are included in Appendix 1 and those used in 2020 are in Appendix 2. We applied the Cochrane sensitivity‐maximising RCT filter (Lefebvre 2019) to MEDLINE (Ovid). For Embase, the RCT filter with the best optimisation of sensitivity and specificity was applied, as in the previous search. (Wong 2006).

We did not impose any restriction on language of publication.

Searching other resources

We searched ClinicalTrials.gov (clinicaltrials.gov) for ongoing trials on 14 May 2020. We were unable to search the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch) as it was unavailable at the time of searching. Search terms for the trials registers are also listed in Appendix 1 (from 2017) and Appendix 2 (from 2020).

We checked all primary references of included studies and systematic reviews for additional references. For studies identified as eligible from clinical trial register records, we searched for associated publications on PubMed and in recent conference presentations.

We contacted study authors to clarify details or obtain additional information not included in the published reports.

Data collection and analysis

Selection of studies

Two review authors (KM and NM) independently screened titles and abstracts of all records identified in our search and coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. In the event of disagreement, a third review author was asked to arbitrate (TL). We then retrieved the full‐text study reports for records identified as eligible, potentially eligible or unclear. Two review authors (KM and NM) independently screened the full‐text articles and identified studies for inclusion. We recorded reasons for exclusion of ineligible studies. We resolved any disagreement by consensus or consulted a third review author (TL). We identified and excluded duplicates and collated multiple reports of the same study so that each study rather than each report is the unit of interest in the review. We recorded the selection process and completed a PRISMA flow diagram (Figure 1) and the Characteristics of excluded studies table.

Figure 1
Study flow diagram

Study flow diagram

Data extraction and management

We used a data collection form to record study characteristics and outcome data from included studies, which had been piloted on two studies in the review (PEP‐CHF; TOPCAT). Some modifications were made after the pilot phase. Two review authors (NM and TL for Martin 2018, and NM and KM for this update) extracted study characteristics from included studies as follows:

  1. Methods: study design, duration of follow‐up, details of any 'run in' period, number of study centres and location, study setting, withdrawals, and start/end date of enrolment.

  2. Participants: number randomised/withdrawn/lost to follow‐up/analysed, mean age/age range, percent male, inclusion criteria, exclusion criteria, systolic blood pressure, heart rate, body mass index, serum creatinine, B‐type natriuretic peptide, NT pro B‐type natriuretic peptide, LVEF, New York Heart Association (NYHA) class, comorbidity (hypertension, diabetes, atrial fibrillation, hospitalisation for HF, coronary heart disease, stroke, medications at baseline).

  3. Interventions: intervention, comparison, concomitant medications (diuretic, digoxin, BBs, ACEIs, ARBs, MRAs).

  4. Outcomes: planned and reported.

  5. Notes: sources of funding, and notable conflicts of interest of trial authors.

Two review authors (NM and TL for Martin 2018 and NM and KM for this update) independently extracted outcome data from included studies. Disagreements were resolved by consensus. One review author (NM) transferred data into the Review Manager 5 (RevMan 2014) file. One review author (TL) double‐checked that data were entered correctly by comparing the data presented in the systematic review with the data extraction sheet.

Assessment of risk of bias in included studies

Two review authors (NM and TL) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Disagreements were resolved by consensus. We assessed the risk of bias according to the following domains.

  1. Random sequence generation.

  2. Allocation concealment.

  3. Blinding of participants and personnel.

  4. Blinding of outcome assessment.

  5. Incomplete outcome data.

  6. Selective outcome reporting.

  7. Other bias.

We judged each potential source of bias as high, low or unclear and provided quotes from study reports together with justification for our judgment in the 'Risk of bias' table. We summarised the 'Risk of bias' judgements across different studies for each of the domains listed. Where information on risk of bias related to unpublished data or correspondence with a triallist, we noted this in the 'Risk of bias' table.

When considering treatment effects in the pooled analysis, we accounted for risk of bias for the studies that contributed to each outcome tested.

Assessment of bias in conducting the systematic review

We conducted the review according to the published protocol and reported protocol deviations in the Differences between protocol and review section.

Measures of treatment effect

We analysed dichotomous data as RR with 95% confidence intervals (CIs), and continuous data as mean difference (MD) or standardized mean difference (SMD), with 95% CIs. In addition, wherever possible, we performed a pooled analysis of outcomes reported as HRs. We used SMD for one analysis when combining quality of life data reported for two different scales and followed Cohen's effect sizes of 0.2 representing a small effect, 0.5 a moderate effect and 0.8 a large effect (Cohen 1988). We entered data presented as a scale with a consistent direction of effect.

Unit of analysis issues

We included one three‐arm trial (Hong Kong DHF). Because two intervention arms contributed to two separate comparisons, no unit of analysis issue arose.

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics and where possible, obtain missing numerical outcome data.

Assessment of heterogeneity

We used the I² statistic to measure heterogeneity among the trials in each analysis. We considered possible causes in cases of substantial heterogeneity (I² = 50% to 100%).

Assessment of reporting biases

We pooled fewer than 10 trials for each comparison. Therefore, we did not examine funnel plots to explore possible small‐study biases for the primary outcomes. We plan to do so, should a sufficient number of trials become available in future updates of this review.

Data synthesis

We undertook meta‐analyses only where this was meaningful, that is, if the treatments, participants and the underlying clinical question were similar enough for pooling to make sense. We used a fixed‐effect model in the absence of substantial heterogeneity (I² < 50%) and a random‐effects model when unexplained substantial heterogeneity was present (I² ≥ 50%). We applied a random‐effects model for quality of life analyses for MRAs to account the high heterogeneity observed for the KCCQ (Analysis 2.7; I² = 86%) and to permit a combined analysis with outcome data from the MLHFQ (Analysis 2.6; I² = 50%).

We considered two relevant quality of life scales: the MLHFQ or KCCQ. The MLHFQ score has a range from 0 to 105, lower scores indicate better quality of life. The KCCQ score has a range from 0 to 100, higher scores indicate better quality of life. To account for the difference in the direction of the scale of the KCCQ, the mean values were multiplied by ‐1 (Cochrane Handbookfor Systematic Reviews of Interventions, section 9.2.3.2, Deeks 2011). For the purpose of interpretation, we considered a five point difference in score as clinically significant for the MLHFQ (Rector 1995) and KCCQ (Spertus 2005).

Subgroup analysis and investigation of heterogeneity

We planned to carry out the following subgroup analyses.

  1. Age. The potential effect of age as an effect modifier of RAAS inhibition is not well described. We planned to explore outcomes for subgroups defined by age as less than 70 years and 70 years and over. This is based on the median age of participants in two major HFpEF trials: the TOPCAT median age was 68.7 years and the PARAGON‐HF mean age was 72 years.

  2. Sex.

  3. HFmrEF LVEF 40% to 49% and preserved LVEF equal to or greater than 50%.

  4. Length of follow‐up less than 12 months and equal to or greater than 12 months.

We were unable to perform subgroup analyses, due to insufficient data (Deeks 2019).

If we had been able to conduct subgroup analyses, we had planned to use the outcomes of cardiovascular mortality and hospitalisation for heart failure and the formal test for subgroup interactions in Review Manager 5 (RevMan 2014).

Sensitivity analysis

We performed a sensitivity analysis for risk of bias by performing a pooled analysis that included only studies with a low risk of bias (where at least four of the six domains for bias assessment were judged to be low risk, and no domain was at high risk of bias).

Summary of findings and assessment of the certainty of the evidence

We created 'Summary of findings' tables for each of our five interventions and included the following outcomes: cardiovascular mortality, heart failure hospitalisation, all‐cause mortality, quality of life (when available MLHFQ, alternatively KCCQ) and hyperkalaemia. We used the five GRADE considerations (study limitations, inconsistencies, imprecision, indirectness and publication bias) to assess the certainty of the body of evidence as it related to the studies which contributed data to the meta‐analyses for the prespecified outcomes. We used methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) using GRADEpro software (GRADEpro GDT). Four review authors assessed the certainty of evidence (TL, NM, KM, CD). We used footnotes to document our justification for decisions to downgrade the certainty of evidence.

Results

Description of studies

Results of the search

We previously included 37 studies, reported in 207 references (Martin 2018). For this update, the database searches identified 1223 new reports, and a search of ClinicalTrials.gov retrieved 251 additional records. After de‐duplication, we screened 1130 records based on the titles and abstracts. Of these, 1038 did not meet the inclusion criteria and were excluded. The remaining 92 records were assessed for eligibility in full‐text and 24 references were included. We identified eight additional references for four previously included studies (CHARM‐Preserved; STRUCTURE; TOPCAT; Yuksek 2012). We included four new studies in this updated review (McDiarmid 2020; PARAGON‐HF; PARALLAX; PARAMOUNT), reported in 17 references. PARALLAX had previously been identified as an ongoing study. We thus now include a total of 41 studies in this review

We identified 11 new ongoing studies, bringing the total of ongoing studies to 13. Three new studies were added to the category of 'Studies awaiting classification' (Botoni 2010; PER‐010‐15; Przewlocka‐Kosmala 2017) bringing the total to nine studies in this category.

Included studies

We included 41 studies (234 reports) that involved a total of 26,059 participants.

BBs

We included 10 studies (3087 participants) that investigated BBs for HFpEF. Of these, five studies compared BBs versus placebo (ELANDD; Mittal 2017; Sahoo 2016; SENIORS; SWEDIC) and five versus usual care (Adamyan 2010; Aronow 1997; J‐DHF; Shu 2005; Takeda 2004). Four studies investigated carvedilol: Adamyan 2010 (up to 50 mg daily), J‐DHF (up to 10 mg twice daily), SWEDIC (up to 25 mg twice daily or 50 mg twice daily in people weighing over 85 kg), Takeda 2004 (up to 20 mg daily). Two studies used nebivolol: ELANDD (up to 10 mg daily) and SENIORS (up to 10 mg daily). One study used propranolol: Aronow 1997 (30 mg, three times daily); and two studies investigated metoprolol succinate: Mittal 2017; Sahoo 2016 (up to 100 mg daily). Shu 2005 investigated bisoprolol (up to 10 mg daily).

Numbers of participants randomised ranged from 40 (Mittal 2017; Takeda 2004) to 643 (SENIORS).

Four were multicentre studies. ELANDD was conducted across 12 centres in eight countries in Europe; J‐DHF was assumed to have taken place in Japan; SENIORS took place in 11 countries (Czech Republic, France, Germany, Hungary, Italy, Netherlands, Romania, Spain, Switzerland, UK and Ukraine), and SWEDIC took place in 12 centres in Sweden. Mittal 2017 and Sahoo 2016 were each conducted in one centre in India. Adamyan 2010, Aronow 1997 and Shu 2005 did not report numbers of centres or countries, but we assumed that Adamyan 2010 likely took place in Armenia. Takeda 2004 was a single centre trial in Japan.

Three studies did not report LVEF of the included participants at baseline (Adamyan 2010; Shu 2005; SWEDIC). Six studies reported LVEF at baseline with a mean ranging from 56% to 63% (Aronow 1997; ELANDD; J‐DHF; Mittal 2017; Sahoo 2016; Takeda 2004). SENIORS included participants with a "clinical history of chronic HF with at least 1 of the following features: documented hospital admission within the previous 12 months with a discharge diagnosis of congestive HF or documented LVEF ≤ 35% within the previous 6 months". The SENIORS study reported a subgroup of participants with LVEF greater than 40% and these outcome data were used in our analysis (643 participants).

Most participants were NYHA class II (51% to 78%). Shu 2005 did not report participants' NYHA class at baseline. Participants' mean age ranged from 30 years to 81 years; six studies reported mean age less than 70 years (Adamyan 2010; ELANDD; Mittal 2017; Sahoo 2016; Shu 2005; SWEDIC) and four reported mean age above 70 years (Aronow 1997; J‐DHF; SENIORS; Takeda 2004).

Three studies were funded by industry (ELANDD; SENIORS; SWEDIC); two studies were funded by not‐for‐profit organisations (J‐DHF; Mittal 2017); and five did not report sources of funding (Adamyan 2010; Aronow 1997; Sahoo 2016; Shu 2005; Takeda 2004).

MRAs

We included 13 studies that investigated MRAs for HFpEF. Of these, eight compared MRAs versus placebo (ALDO‐DHF; AREA IN‐CHF; Kurrelmeyer 2014; Mottram 2004; RAAM‐PEF; STRUCTURE; TOPCAT; Upadhya 2017) and five versus usual care (Karapysh 2015; Mak 2009; McDiarmid 2020; Orea‐Tejeda 2007; Wang 2010). Ten studies investigated spironolactone (ALDO‐DHF; Kurrelmeyer 2014 (initiated at 25 mg daily and up‐titrated to a maximum of 50 mg daily); McDiarmid 2020 (25 mg daily); Mottram 2004; STRUCTURE; Upadhya 2017 (25 mg daily); Karapysh 2015; Orea‐Tejeda 2007 (25 mg daily, up‐titrated if tolerated to 50 mg daily); TOPCAT (15 mg daily, increased to a maximum of 45 mg daily); Wang 2010 (50 mg daily)). Two studies used eplerenone (Mak 2009; RAAM‐PEF (25 mg daily to a maximum of 50 mg daily)). AREA IN‐CHF investigated canrenone at a maximum dose of 50 mg daily.

Numbers of participants randomised ranged from 28 (Orea‐Tejeda 2007) to 3445 (TOPCAT). Four were multicentre trials; ALDO‐DHF included 10 centres in Germany and Austria; AREA IN‐CHF was conducted in 46 centres in Italy; STRUCTURE had centres in Poland (the number is unclear but publication states "of each centre"); and TOPCAT was conducted across 233 sites in six countries (Argentina, Brazil, Canada, Georgia, Russia, USA). Six studies were single‐centre trials; two in USA (Kurrelmeyer 2014; RAAM‐PEF), one in the UK (McDiarmid 2020), one in Australia (Mottram 2004) and one in Taiwan (Wang 2010); Mak 2009 was a single‐centre trial but the country was unspecified. Three trials did not report on numbers of centres or countries (Karapysh 2015; Orea‐Tejeda 2007; Upadhya 2017).

Two studies (Karapysh 2015; Mottram 2004) did not report participants' LVEF at baseline. AREA IN‐CHF had a mean LVEF at baseline of 39.9% (intervention) and 39.7% (control) for the overall included participants (N = 467). However, we obtained outcome data for the subgroup of participants with LVEF greater than 40% (N = 225). The LVEF in the remaining seven studies ranged from 54% to 72%.

In one study, most participants were NYHA class I (McDiarmid 2020, 61%). Most participants in five studies were NYHA class II (52% to 88%; ALDO‐DHF; Mak 2009; RAAM‐PEF; STRUCTURE; TOPCAT). Most participants in two studies were NYHA class III (58% to 64%; Kurrelmeyer 2014; Upadhya 2017). Three studies did not report NYHA class for participants eligible for inclusion in our review (AREA IN‐CHF; Karapysh 2015; Mottram 2004). Orea‐Tejeda 2007 reported that most participants in the intervention arm were NYHA class III (57.1%) and NYHA class I (75%) in the control arm.

Participants' mean age ranged from 54.5 years to 80 years; seven studies included participants whose mean age was less than 70 years (ALDO‐DHF; AREA IN‐CHF; Karapysh 2015; Mottram 2004; Orea‐Tejeda 2007; STRUCTURE; TOPCAT). In five studies, participants' mean age was over 70 years (Kurrelmeyer 2014; Mak 2009; McDiarmid 2020; RAAM‐PEF; Upadhya 2017).

AREA IN‐CHF was industry funded; seven studies were funded by not‐for‐profit organisations (ALDO‐DHF; Kurrelmeyer 2014; McDiarmid 2020; RAAM‐PEF; STRUCTURE; TOPCAT; Upadhya 2017). Five studies did not report sources of funding (Karapysh 2015; Mak 2009; Mottram 2004; Orea‐Tejeda 2007; Wang 2010).

ACEIs

We included eight studies that investigated ACEIs for HFpEF. Of these, three compared ACEIs with placebo (Kitzman 2010; PEP‐CHF; Zi 2003), and five versus usual care (Aronow 1993; Aronow 1998; Hong Kong DHF; SNEGOVIK; Yuksek 2012). Two studies investigated enalapril (Aronow 1993 up to 20 mg daily; Kitzman 2010 up to 10 mg daily). Aronow 1998 investigated benazepril (up to 40 mg daily). Two studies investigated perindopril (PEP‐CHF up to 4 mg daily; Yuksek 2012, up to 10 mg). Hong Kong DHF investigated ramipril in one of two active arms (maximum of 10 mg daily). Two studies investigated quinapril (SNEGOVIK, dose not reported; Zi 2003, up to 40 mg daily).

Numbers of participants randomised ranged from 21 (Aronow 1993) to 850 (PEP‐CHF). Two studies were reportedly multicentre trials (Hong Kong DHF; PEP‐CHF). Hong Kong DHF did not report details on the number of centres. PEP‐CHF was conducted at 53 centres in Bulgaria (3), Czech Republic (5), Hungary (10), Ireland (1), Poland (26), Russia (1), Slovakia (2), and the UK (5). Zi 2003 took place at one hospital in the UK and Yuksek 2012 was conducted in Turkey. The countries or number of centres were not reported in four studies (Aronow 1993; Aronow 1998; Kitzman 2010; SNEGOVIK).

The mean LVEF of the included participants at baseline was not reported by two studies (SNEGOVIK; Zi 2003). LVEF ranged from 61% to 69% in five studies (Aronow 1993; Aronow 1998; Hong Kong DHF; Kitzman 2010; PEP‐CHF). Most participants were classified in NYHA class II in four studies (Hong Kong DHF; Kitzman 2010; PEP‐CHF; Zi 2003) and in NYHA class III in one study (Aronow 1993). Two studies did not report participants' NYHA class at baseline (Aronow 1998; SNEGOVIK).

Participants' mean age ranged from 70 years to 82 years with all studies equal to or over a mean age of 70 years.

Four studies did not report funding sources (Aronow 1993; Aronow 1998; SNEGOVIK; Yuksek 2012). Three studies were industry‐funded (Hong Kong DHF; PEP‐CHF; Zi 2003) and one study was funded by a not‐for profit organisation (Kitzman 2010).

ARBs

We included eight studies that investigated ARBs for HFpEF. Of these, five compared ARBs versus placebo (CAN‐DHF; CHARM‐Preserved; I‐PRESERVE; Kasama 2005; Parthasarathy 2009) and three compared ARBs versus usual care (CandHeart; Hong Kong DHF; SUPPORT). Four studies investigated candesartan (CAN‐DHF; CandHeart; CHARM‐Preserved (up to 32 mg daily), Kasama 2005 (8 mg to 12 mg daily)). Two studies investigated irbesartan (one of the two active treatment arms in Hong Kong DHF (up to 75 mg daily), I‐PRESERVE (up to 300 mg)). Parthasarathy 2009 investigated valsartan (80 mg daily). SUPPORT investigated olmesartan (up to 40 mg daily).

Numbers of participants randomised ranged from 22 (CAN‐DHF) to 4128 (I‐PRESERVE). Seven were multicentre trials: CAN‐DHF was conducted at eight centres in Germany; CandHeart at 70 centres in Italy; CHARM‐Preserved was conducted at 618 centres in 26 countries; I‐PRESERVE involved 293 centres in 25 countries; Parthasarathy 2009 was conducted at five centres each in Germany and the UK; and SUPPORT was conducted at 17 centres in Japan. Hong Kong DHF was reported to be a multicentre trial but no details were provided on numbers of centres or countries. Kasama 2005 was reported to be a single‐centre trial in Japan.

The mean LVEF of the included participants at baseline was not reported by CAN‐DHF and ranged from 49% to 72% in seven studies (CandHeart; CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; Kasama 2005; Parthasarathy 2009; SUPPORT). Most participants were assessed as NYHA class II at baseline in five studies (CandHeart; CHARM‐Preserved; Hong Kong DHF; Kasama 2005; SUPPORT); NYHA class III in I‐PRESERVE; and was not reported by two studies (CAN‐DHF; Parthasarathy 2009).

Participants' mean age ranged from 61 years to 75 years. Mean age was below 70 years in six studies (CAN‐DHF; CandHeart; CHARM‐Preserved; Kasama 2005; Parthasarathy 2009; SUPPORT) and over 70 years in two studies (Hong Kong DHF; I‐PRESERVE).

Six studies were funded by industry (CAN‐DHF; CandHeart; CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; Parthasarathy 2009). SUPPORT was funded by a not‐for‐profit organisation. Kasama 2005 did not report the source of funding.

ARNIs

For this update, we elected to include three studies comparing an ARNI (sacubitril‐valsartan) with an ARB (valsartan) (PARAMOUNT and PARAGON‐HF), and one study comparing an ARNI (sacubitril‐valsartan) with individualised medical therapy whereby the comparator was specified according to the RAAS treatment status of patients at study enrolment; patients treated with ACEI at enrolment received LCZ696 or enalapril, those treated with ARB received LCZ696 or valsartan, and those without prior treatment with RAAS inhibition received LCZ696 or matching placebo) (PARALLAX).

Administration of ARBs or ACEIs in combination with sacubitril‐valsartan is contraindicated due to safety concerns. Therefore, the investigators specified the active comparator of valsartan, given that many patient with HFpEF receive ARB or ACEI treatment for hypertension (Solomon 2012). Since both treatment arms received valsartan, these studies isolate the effects of neprilysin. However, this can only be safely administered in combination with an ARB (i.e. the ARNI class of therapeutics).

Number of participants randomised ranged from 308 (PARAMOUNT) to 4822 participants (PARAGON‐HF). PARAGON‐HF was a multicentre trial across 848 centres in 43 countries and PARAMOUNT had 65 centres and 13 countries. The mean LVEF was 56 to 58% across the three trials. Most participants in these trials were assessed as NYHA class II at baseline. Participants' mean age ranged from 71 to 73 years of age. Novartis funded the three studies.

PERSPECTIVE (EUCTR2016‐001254‐17) is an ongoing RCT to examine the effect of LCZ696 compared to valsartan on cognitive outcomes in participants with HFpEF, defined as LVEF greater than 40%.

Excluded studies

We previously excluded 303 studies (324 references) based on full‐text assessment (Martin 2018). Details for the reasons for exclusion are provided in the Characteristics of excluded studies table. In this update, we have excluded 52 studies (54 references) and added these to the list for exclusions below. We referenced in the review only those that most closely missed the inclusion criteria (four references). In summary, we made exclusions based on the following considerations:

  • population does not meet protocol: n = 118;

  • wrong intervention: n = 8;

  • wrong comparator: n = 23;

  • wrong study design: n = 125;

  • subgroup of interest but no response to our enquiry for data: n = 8;

  • subgroup not of interest: n = 36;

  • unclear eligibility and no response to our enquiry for details: n = 12;

  • unclear eligibility and no current contact details: n = 15;

  • completed status in trial registry record but no published results and no response to our enquiry for data: n = 1;

  • missing data and response that no details can be provided: n = 6;

  • retraction: n = 1; and

  • did not take place as planned: n = 2.

Studies awaiting classification

We identified nine studies that are awaiting classification (Anonymous 2003d; Botoni 2010; Dielievska 2015; EUCTR2005‐001306‐87; EUCTR2005‐002109‐22Metra 1999; PER‐010‐15; Przewlocka‐Kosmala 2017; Rapezzi 1999. See Characteristics of studies awaiting classification. We are waiting to retrieve the full‐text (n = 4), await responses from translators (n = 3) and await response from the triallists to clarify eligibility (n = 2).

Ongoing studies

From the previously identified ongoing studies, two have now been included (McDiarmid 2020; PARALLAX). We identified 11 new ongoing studies, bringing the current total to 13 ongoing studies (Characteristics of ongoing studies).

Risk of bias in included studies

The 'Risk of bias' assessments are detailed in the Characteristics of included studies tables. We summarised them in the text below and in Figure 2 and Figure 3.

Figure 2
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Twelve studies reported random sequence methods and were rated as being at low risk of bias (ALDO‐DHF; CandHeart; ELANDD; Hong Kong DHF; I‐PRESERVE; McDiarmid 2020; PARAGON‐HF; PARAMOUNT; PEP‐CHF; Sahoo 2016; SENIORS; TOPCAT). We assessed 28 studies to be at unclear risk of bias for this domain because no information was provided in study reports. One study was assessed as high risk of bias in this domain as it randomised participants on the basis of admission sequence (Shu 2005).

Twelve studies used a method for allocation concealment that was judged to be at low risk of bias (ALDO‐DHF; CHARM‐Preserved; I‐PRESERVE; Kurrelmeyer 2014; McDiarmid 2020; Mittal 2017; PARAGON‐HF; PARAMOUNT; PEP‐CHF; SENIORS; STRUCTURE; TOPCAT). We assessed 29 studies to be at unclear risk of bias for this domain because no information was provided in study reports.

Blinding

We assessed 22 studies to be at low risk of bias regarding blinding of participants and personnel (ALDO‐DHF; AREA IN‐CHF; CAN‐DHF; CHARM‐Preserved; ELANDD; I‐PRESERVE; Kasama 2005; Kitzman 2010; Kurrelmeyer 2014; Mittal 2017; Mottram 2004; PARAGON‐HF; PARAMOUNT; Parthasarathy 2009; PEP‐CHF; RAAM‐PEF; SENIORS; STRUCTURE; SWEDIC; TOPCAT; Upadhya 2017; Zi 2003). Six studies were open‐label designs and therefore were judged to be at high risk of bias for this domain (CandHeart; Hong Kong DHF; J‐DHF; Mak 2009; McDiarmid 2020; SUPPORT). The remaining 13 studies were assessed at unclear risk of bias because no information was provided.

Detection bias was judged to be at low risk in 18 studies (ALDO‐DHF; CHARM‐Preserved; Hong Kong DHF; J‐DHF; Kasama 2005; Kitzman 2010; Mak 2009; Mittal 2017; Mottram 2004; Orea‐Tejeda 2007; PARAGON‐HF; PARAMOUNT; PEP‐CHF; RAAM‐PEF; Sahoo 2016; STRUCTURE; SUPPORT; TOPCAT). One study was judged to be at high risk of detection bias as outcome assessors were not blinded (McDiarmid 2020). The remaining 22 studies did not provide information and were judged to be at unclear risk of detection bias.

Incomplete outcome data

Attrition bias was judged to be at low risk in 14 studies (ALDO‐DHF; CHARM‐Preserved; ELANDD; I‐PRESERVE; J‐DHF; PARAGON‐HF; PARAMOUNT; Parthasarathy 2009; PEP‐CHF; RAAM‐PEF; SENIORS; STRUCTURE; TOPCAT; Zi 2003). We judged Kasama 2005 to be at high risk of bias for this domain because the study report did not indicate if losses to follow‐up or withdrawals occurred. Shu 2005 (unclear reporting of withdrawals) and McDiarmid 2020 (completed case analysis with uneven drop‐out in treatment arms) were also judged to be at high risk of bias. All 24 remaining studies were assessed to be at unclear risk of bias for attrition bias as no information was reported to allow judgement.

Selective reporting

We assessed 12 studies to be at low risk of reporting bias (ALDO‐DHF; CAN‐DHF; CHARM‐Preserved; ELANDD; I‐PRESERVE; J‐DHF; PARAGON‐HF; PARAMOUNT; PEP‐CHF; SENIORS; SUPPORT; TOPCAT). These 12 studies reported planned outcomes in either published protocols or clinical trial registers before enrolment started. We were unable to assess reporting bias in 29 studies either because no information was available in the form of protocols or clinical trial registry entries, or they were published/entered after enrolment was completed.

Other potential sources of bias

We judged 20 studies to be at low risk of other bias (mainly based on providing details on funding and declaring any conflict of interest by the authors) (ALDO‐DHF; AREA IN‐CHF; CHARM‐Preserved; ELANDD; I‐PRESERVE; J‐DHF; Kitzman 2010; Mak 2009; Mittal 2017; PARAGON‐HF; PARAMOUNT; Parthasarathy 2009; PEP‐CHF; RAAM‐PEF; SENIORS; STRUCTURE; SUPPORT; SWEDIC; TOPCAT; Zi 2003).

We judged six studies to be at high risk of other bias. Kurrelmeyer 2014 was originally registered as an observational study and this detail was changed after completion of the trial, but before the results were published. Five studies (Adamyan 2010; CAN‐DHF; Karapysh 2015; SNEGOVIK; Upadhya 2017) were published as conference abstracts only; withholding the full results from publication may present a form of bias. The remaining 15 studies were judged to be at unclear risk of bias.

Effects of interventions

See: Summary of findings 1 Beta‐blockers compared to placebo or no treatment for chronic heart failure with preserved ejection fraction; Summary of findings 2 Mineralocorticoid receptor antagonists (MRAs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction; Summary of findings 3 Angiotensin‐converting enzyme inhibitors (ACEIs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction; Summary of findings 4 Angiotensin receptor blockers (ARBs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction; Summary of findings 5 Angiotensin receptor neprilysin inhibitors (ARNIs) compared to usual care for chronic heart failure with preserved ejection fraction

BBs versus placebo or no treatment

We included 10 studies involving a total of 3087 participants that assessed BBs versus placebo or no treatment. The main outcomes for this comparison are included in summary of findings Table 1.

Cardiovascular mortality

Six studies reported cardiovascular mortality (Aronow 1997; ELANDD; J‐DHF; SENIORS; SWEDIC; Takeda 2004). Three studies reported that no deaths occurred (ELANDD; SWEDIC; Takeda 2004). We included three studies in the meta‐analysis (Aronow 1997; J‐DHF; SENIORS) (15% of participants in the intervention arm versus 19% in the control arm; RR 0.78, 95% CI 0.62 to 0.99; NNTB 25; 1046 participants; I² = 0%; low‐certainty evidence; Analysis 1.1).

J‐DHF reported cardiovascular mortality but with different numbers for events within the same table (Table 2 in the primary reference). We contacted the study authors to seek clarification but are yet to receive a response; we used the higher numbers in the analysis.

SENIORS reported a hazard ratio (HR 0.80, 95% CI 0.49 to 1.32; 643 participants).

Heart failure hospitalisation

We included five studies that reported heart failure hospitalisation (ELANDD; J‐DHF; Shu 2005; SWEDIC; Takeda 2004). ELANDD reported that no hospitalisation occurred due to HF. Data from four studies (J‐DHF; Shu 2005; SWEDIC; Takeda 2004) contributed to the meta‐analysis (RR 0.73, 95% CI 0.47 to 1.13; 449 participants; I² = 22%; very low‐certainty evidence; Analysis 1.2).

Hyperkalaemia

J‐DHF reported that one participant in the intervention group (N = 120) experienced hyperkalaemia but did not report on this outcome for the control group (very low‐certainty evidence). No further data were available from any other studies.

All‐cause mortality

We included seven studies that reported all‐cause mortality (Adamyan 2010; Aronow 1997; ELANDD; J‐DHF; SENIORS; SWEDIC; Takeda 2004). Of these, three studies reported that no deaths occurred (ELANDD; SWEDIC; Takeda 2004). We included data from four studies in the meta‐analysis (Adamyan 2010; Aronow 1997; J‐DHF; SENIORS) (RR 0.82, 95% CI 0.67 to 1.00; 1105 participants; I² = 0%; low‐certainty evidence; Analysis 1.3).

J‐DHF reported all‐cause mortality but with different numbers for events within the same table (Table 2 in the primary reference). We contacted the study authors to seek clarification but are yet to receive a response. We used the higher number of deaths in the analysis.

SENIORS reported a hazard ratio (HR 0.92, 95% CI 0.61 to 1.36; 643 participants).

Quality of life

We included two studies that reported quality of life (ELANDD; Mittal 2017). Mittal 2017 reported quality of life using SF‐36, which was not a scale we considered for our analysis. ELANDD reported end scores for the MLHFQ total score and showed MD ‐1.00 between the treatment arms, favouring the intervention (95% CI ‐9.05 to 7.05; 93 participants; very low‐certainty evidence).

Withdrawal due to adverse event

We included five studies that reported withdrawals due to adverse events (Aronow 1997; ELANDD; J‐DHF; Mittal 2017; Sahoo 2016). Mittal 2017 and Sahoo 2016 reported no withdrawals due to adverse events. Aronow 1997 reported 11 withdrawals due to "worsening CHF [chronic heart failure] in 7 patients and hypotension in 4 patients" but did not provide this information by intervention arm. Only two studies (ELANDD; J‐DHF) contributed data for meta‐analysis (9% of participants in the intervention arm versus 0% in the control arm, RR 18.07, 95% CI 2.45 to 133.04; 338 participants; I² = 0%; Analysis 1.5; number needed to harm (NNTH) = 11).

MRAs versus placebo or no treatment

We included 13 studies (4459 participants) that assessed MRAs versus placebo or no treatment. The main outcomes for this comparison are included in summary of findings Table 2. The findings for this comparison were driven by one trial (TOPCAT). Four trials (Karapysh 2015; Mottram 2004; Orea‐Tejeda 2007; Wang 2010) did not contribute any outcome data of interest for this review.

Cardiovascular mortality

We included five studies that reported cardiovascular mortality (ALDO‐DHF; AREA IN‐CHF; Kurrelmeyer 2014; RAAM‐PEF; TOPCAT). Of these, two studies reported that no deaths occurred (Kurrelmeyer 2014; RAAM‐PEF). We included data from three studies in the meta‐analysis (ALDO‐DHF; AREA IN‐CHF; TOPCAT) (RR 0.90, 95% CI 0.74 to 1.11; 4070 participants; I² = 0%; moderate‐certainty evidence; Analysis 2.1).

TOPCAT also reported a hazard ratio (HR 0.90, 95% CI 0.73 to 1.12; 3445 participants).

Heart failure hospitalisation

We included six studies that reported heart failure hospitalisation (ALDO‐DHF; AREA IN‐CHF; Kurrelmeyer 2014; RAAM‐PEF; TOPCAT; Upadhya 2017). Of these, three studies reported no hospitalisations due to heart failure(ALDO‐DHF; Kurrelmeyer 2014; Upadhya 2017). We included data from three studies in the meta‐analysis (AREA IN‐CHF; RAAM‐PEF; TOPCAT) (11% of participants in the intervention arm versus 14% in the control arm, RR 0.82, 95% CI 0.69 to 0.98; 3714 participants; NNTB 41; I² = 22%; moderate‐certainty evidence; Analysis 2.2).

Hazard ratios for time to first heart failure hospitalisation were reported for two studies (AREA IN‐CHF; TOPCAT) (HR 0.82, 95% CI 0.69 to 0.98; 3670 participants; I² = 59%; Analysis 2.3). The substantial heterogeneity was explained by differences in population characteristics (TOPCAT, LVEF ≥ 45%; AREA IN‐CHF subgroup, LVEF 40% to 45%).

Hyperkalaemia

We included six studies that reported hyperkalaemia (ALDO‐DHF; AREA IN‐CHF; Kurrelmeyer 2014; RAAM‐PEF; STRUCTURE; TOPCAT) (16% of participants in the intervention arm versus 8% in the control arm, RR 2.11, 95% CI 1.77 to 2.51; 4291 participants; I² = 0%; high‐certainty evidence; Analysis 2.4).

All‐cause mortality

We included eight studies that reported all‐cause mortality (ALDO‐DHF; AREA IN‐CHF; Kurrelmeyer 2014; Mak 2009; RAAM‐PEF; STRUCTURE; TOPCAT; Upadhya 2017). Of these, three studies reported that no deaths occurred (Kurrelmeyer 2014; RAAM‐PEF; STRUCTURE). The meta‐analysis included data from five studies (ALDO‐DHF; AREA IN‐CHF; Mak 2009; TOPCAT; Upadhya 2017) (RR 0.91, 95% CI 0.78 to 1.06; 4207 participants; I² = 0%; moderate‐certainty evidence; Analysis 2.5).

TOPCAT also reported a hazard ratio (HR 0.91, 95% CI 0.77 to 1.08; 3445 participants).

Quality of life

We included six studies that reported quality of life (ALDO‐DHF; Kurrelmeyer 2014; Mak 2009; RAAM‐PEF; TOPCAT; Upadhya 2017). TOPCAT reported quality of life in a report by Lewis 2016, but the end scores per treatment arm were not provided. We contacted the investigators and await details.

Three studies (ALDO‐DHF; Mak 2009; Upadhya 2017) reported total MLFHQ scores and were pooled for analysis (MD 0.84, 95% CI ‐2.30 to 3.98; 511 participants; I² = 0%; low‐certainty evidence; Analysis 2.8). Kurrelmeyer 2014 and RAAM‐PEF reported KCCQ results and were pooled (MD ‐0.78, 95% CI ‐28.02 to 26.46; 92 participants; I² = 86%; Analysis 2.7). The substantial heterogeneity could not be explained.

All five studies that used MLHFQ and KCCQ were pooled (SMD 0.05, 95% CI ‐0.23 to 0.34; 603 participants; I² = 50%; Analysis 2.6). The substantial heterogeneity could not be explained.

Withdrawal due to adverse event

Five studies reported this outcome (ALDO‐DHF; Kurrelmeyer 2014; McDiarmid 2020; TOPCAT; Upadhya 2017) and contributed to the meta‐analysis (RR 1.10, 95% CI 1.00 to 1.21; 4037 participants; five studies; I² = 0%; Analysis 2.9).

ACEIs versus placebo or no treatment

We included eight studies involving a total of 2061 participants that assessed ACEIs versus placebo or no treatment. The main outcomes for this comparison are presented in summary of findings Table 3. The findings for this comparison were driven by PEP‐CHF. Two studies (Aronow 1993; Yuksek 2012) did not contribute any outcome data of interest for this review.

Cardiovascular mortality

Three studies reported cardiovascular mortality (Hong Kong DHF; Kitzman 2010; PEP‐CHF). Kitzman 2010 reported that no deaths occurred. Hong Kong DHF and PEP‐CHF contributed data to the meta‐analysis (RR 0.93, 95% CI 0.61 to 1.42; 945 participants; I² = 0%; moderate‐certainty evidence; Analysis 3.1).

PEP‐CHF also reported a hazard ratio (HR 0.98, 95% CI 0.63 to 1.52; 850 participants).

Heart failure hospitalisation

Three studies (Hong Kong DHF; PEP‐CHF; Zi 2003) reported heart failure hospitalisation and were pooled for analysis (RR 0.86, 95% CI 0.64 to 1.15; 1019 participants; I² = 0%; moderate‐certainty evidence; Analysis 3.2).

PEP‐CHF also reported a hazard ratio (HR 0.86, 95% CI 0.61 to 1.20; 850 participants).

Hyperkalaemia

Zi 2003 reported hyperkalaemia (RR 5.27, 95% CI 0.26 to 106.16; 74 participants; very low‐certainty evidence; Analysis 3.3).

All‐cause mortality

We included six studies that reported all‐cause mortality (Aronow 1998; Hong Kong DHF; Kitzman 2010; PEP‐CHF; Yuksek 2012; Zi 2003). Kitzman 2010 reported that no deaths occurred. Five studies (Aronow 1998; Hong Kong DHF; PEP‐CHF; Yuksek 2012; Zi 2003) contributed to the meta‐analysis (RR 1.04, 95% CI 0.75 to 1.45; 1187 participants; five studies; I2 = 0%; moderate‐certainty evidence; Analysis 3.4).

PEP‐CHF also reported a hazard ratio (HR 1.09, 95% CI 0.75 to 1.58; 850 participants).

Quality of life

Three studies reported quality of life assessed using the MLHFQ scale (Hong Kong DHF; Kitzman 2010; SNEGOVIK). SNEGOVIK reported quality of life assessment based on the MLHFQ scale as change from baseline per treatment arm (‐18.9 for intervention, ‐10.7 for control). We were unsuccessful in our attempts to contact study authors to obtain scores at the end of follow‐up. Two studies (Hong Kong DHF; Kitzman 2010) contributed to the meta‐analysis (MD ‐0.09, 95% CI ‐3.66 to 3.48; 154 participants; I² = 4%; low‐certainty evidence; Analysis 3.5).

Zi 2003 assessed quality of life using the McMaster quality of life questionnaire and reported end scores at six months (12.9 ± 3.1 for the intervention and 13.1 ± 4.7 for the control arm).

Withdrawal due to adverse event

Four studies (Hong Kong DHF; PEP‐CHF; Yuksek 2012; Zi 2003) reported this outcome and were pooled for analysis (RR 2.52, 95% CI 0.49 to 12.87; 1127 participants; four studies; I² = 58%; Analysis 3.6).

ARBs versus placebo or no treatment

We included eight studies involving a total of 8755 participants that assessed ARBs versus placebo or no treatment. The main outcomes for this comparison are included in summary of findings Table 4. The findings for this comparison were driven by two studies (CHARM‐Preserved; I‐PRESERVE). Three trials (CAN‐DHF; CandHeart; Kasama 2005) did not contribute any outcome data of interest for this review.

Cardiovascular mortality

Four studies reported this outcome (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; Parthasarathy 2009). Parthasarathy 2009 reported that no deaths occurred. Three studies (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE) contributed to the meta‐analysis (RR 1.02, 95% CI 0.90 to 1.14; 7254 participants; I² = 0%; high‐certainty evidence; Analysis 4.1).

Two studies (CHARM‐Preserved; I‐PRESERVE) were also pooled for an analysis of hazard ratio (HR 1.00, 95% CI 0.89 to 1.13; 7148 participants; Analysis 4.2).

Heart failure hospitalisation

Three studies (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE) reported this outcome and were pooled for analysis (RR 0.92, 95% CI 0.83 to 1.02; 7254 participants; I² = 0%; high‐certainty evidence; Analysis 4.3).

Two studies (CHARM‐Preserved; I‐PRESERVE) were also pooled for analysis  of hazard ratio (HR 0.90, 95% CI 0.80 to 1.01; 7148 participants; Analysis 4.4).

Hyperkalaemia

Two studies reported this outcome and were pooled for analysis (CHARM‐Preserved; I‐PRESERVE) (0.9% of participants in the intervention group and 0.5% in the control group; RR 1.88, 95% CI 1.07 to 3.33; 7148 participants; high‐certainty evidence; Analysis 4.5).

All‐cause mortality

Five studies reported this outcome (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; Parthasarathy 2009; SUPPORT). Parthasarathy 2009 reported that no deaths occurred. Four studies (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; SUPPORT) contributed to the meta‐analysis (RR 1.01, 95% CI 0.92 to 1.11; 7964 participants; I² = 0%; high‐certainty evidence; Analysis 4.6). For the SUPPORT trial, data for participants with LVEF ≥ 50% were analysed according to the definition of HFpEF used in this trial.

Two studies (I‐PRESERVE; SUPPORT) were also pooled for analysis of hazard ratio (HR 0.99, 95% CI 0.88 to 1.12; 4838 participants; Analysis 4.7).

Quality of life

Four studies reported this outcome (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; Parthasarathy 2009). CHARM‐Preserved reported quality of life (MLHFQ) in a study report (Lewis 2007): however, end scores per treatment arm were not provided. Three studies (Hong Kong DHF; I‐PRESERVE; Parthasarathy 2009) contributed to the meta‐analysis for MLHFQ (MD 0.41, 95% CI ‐0.86 to 1.67; 3117 participants; I² = 19%; high‐certainty evidence; Analysis 4.8).

Withdrawal due to adverse event

Four studies (CHARM‐Preserved; Hong Kong DHF; I‐PRESERVE; Parthasarathy 2009) reported this outcome and contributed to the meta‐analysis (16% of participants in the intervention arm versus 13% in the control arm; RR 1.22, 95% CI 1.09 to 1.36; 7406 participants; I² = 0%; Analysis 4.9 NNTH = 33).

ARNIs versus usual care

We included three studies that involved a total of 7702 participants. The main outcomes for this comparison are included in summary of findings Table 5.

Cardiovascular mortality

One study, PARAGON‐HF, reported proportional hazards and total event counts for CV mortality; a hazard ratio (HR) of 0.95 (95% CI 0.79 to 1.16) was reported and we calculated the risk ratio (RR), based on the total event counts, as 0.96 (95% CI 0.79 to 1.15; 4796 participants; moderate‐certainty evidence; Analysis 5.1).

Heart failure hospitalisation

PARAGON‐HF and PARALLAX reported data for first hospitalisation due to HF, which led to a RR of 0.89 (95% CI 0.80 to 1.00; 7362 participants; I² = 56%; moderate‐certainty evidence; Analysis 5.2).

PARAGON‐HF and PARALLAX also reported HRs for time to first hospitalisation due to HF, which gives a pooled effect of HR 0.87 (95% CI 0.76 to 0.99; 7362 participants; two studies; I² = 82%; Analysis 5.3).

Hyperkalaemia

Two studies (PARAGON‐HF; PARAMOUNT) reported this outcome (RR 0.88, 95% CI 0.77 to 1.01; 5054 participants; two studies; I² = 8%; moderate‐certainty evidence; Analysis 5.4).

All‐cause mortality

The three studies reported data for all‐cause mortality (RR 0.97, 95% CI 0.84 to 1.11; 7663 participants; three studies; high‐certainty evidence; Analysis 5.5).

Quality of life

PARAMOUNT reported change from baseline for the KCCQ overall summary score for the intervention arm (n = 118) as 11.25 (2.185) and the control arm (n = 116) as 11.31 (2.183), and summarised the findings as "no difference in KCCQ score between treatment groups".

PARAGON‐HF reported a difference of the clinical summary score KCCQ between treatment arms as 1.0 (0.0 to 2.1).

PARALLAX reported "KCCQ improved in both treatment groups, with an early benefit of S/V that was no longer significant after 24 week".

Withdrawal due to adverse event

The three studies reported withdrawals due to adverse events (RR 1.02, 95% CI 0.91 to 1.14; 7663 participants; three studies; I2 = 62%; Analysis 5.6).

Subgroup analyses

We did not have sufficient data to allow for meaningful subgroup analyses.

Sensitivity analyses

We conducted a sensitivity analysis by only including studies assessed at low risk of bias. Across comparisons, the estimates were not significantly changed, with the exception of BBs, where no effect on cardiovascular mortality was observed (one low risk of bias study (SENIORS): RR 0.81, 95% CI 0.50 to 1.29; versus overall analysis of three studies: RR 0.78, 95% CI 0.62 to 0.99).

Discussion

Summary of main results

We examined the evidence for the effects of BBs and RAAS inhibitors for the treatment of HFpEF. We included 41 trials, reported in 234 publications that involved a total of 26,059 participants. We identified 13 ongoing trials with treatment arms that include interventions assessed in this review. A further nine studies await assessment.

We performed a pooled analysis for the outcomes of cardiovascular and all‐cause mortality, heart failure hospitalisation, quality of life and hyperkalaemia. For most studies, we use the MLHFQ instrument to evaluate quality of life outcomes since this was the most frequently reported measure, except for ARNI trials that used KCCQ.

Withdrawals due to adverse events were inconsistently reported; these data could not be included in 'Summary of findings' tables.

We conducted a sensitivity analysis by including only studies assessed with overall low risk of bias. The effect estimates were not significantly changed, except for BBs.

BBs

A total of 10 included studies (3087 participants) assessed BBs compared with placebo or no intervention. We performed meta‐analyses including up to four studies and 1105 participants. The results suggested that treatment may improve cardiovascular mortality, however, the quality of evidence was low due to imprecision and risk of bias. When we performed a sensitivity analysis by including only studies with a low overall risk of bias, the effects on cardiovascular mortality did not persist. The two largest studies (J‐DHF; SENIORS) reported high rates of study drug discontinuation due to intolerance rather than adverse events, which may have attenuated any true treatment effects. A recent individual participant meta‐analysis found evidence that beta‐blockers improve LVEF and prognosis in heart failure patients with LVEF  40 to 49%, but not in those with  LVEF over 50% (Cleland 2018).

MRAs

A total of 13 studies (4459 participants) assessed MRA compared with placebo or no intervention. We combined evidence from up to six trials and 4291 participants in meta‐analyses. We found that treatment with MRA reduces the risk of heart failure hospitalisation but found little or no effect on cardiovascular and all‐cause mortality; however the certainty of evidence was moderate and uncertainty remains over these treatment effects. As expected, MRA treatment was associated with an increased risk of hyperkalaemia; potassium monitoring is therefore required in people being treated using MRA. A large, registry‐randomised clinical outcomes trial of spironolactone for HFpEF is ongoing and due to complete in 2022 (NCT02901184: Spironolactone Initiation Registry Randomized Interventional Trial in Heart Failure with Preserved Ejection Fraction, SPIRRIT).

ACEIs

A total of eight included studies (2062 participants) assessed ACEIs versus placebo or no intervention. We conducted a meta‐analysis of data from up to five trials and 1187 participants. We found that there was probably little or no effect on cardiovascular mortality, all‐cause mortality, heart failure hospitalisation or quality of life. Data on hyperkalaemia were limited. No large clinical trials (more than 1000 participants) were available and the certainty of evidence was assessed as moderate due to imprecision. The effectiveness of ACEI therapy in the treatment of people with HFpEF remains unclear.

ARBs

A total of eight included studies (8755 participants) assessed ARB therapy for people with HFpEF, with the evidence certainty assessed as high. We combined evidence from up to four trials and 7964 participants for meta‐analysis and found little or no overall difference on the outcomes of cardiovascular mortality, all‐cause mortality, heart failure hospitalisation or quality of life. The CHARM study found a beneficial effect on HF hospitalisation based on a time‐to‐event analysis; the strength of this association was increased in a subsequent analysis, based on recurrent events (Rogers 2014). As expected, ARB treatment was associated with an increased risk of hyperkalaemia; potassium monitoring is therefore required.

ARNIs

Three studies (7696 participants) assessed ARNI therapy for people with HFpEF. Only one study (PARAGON‐HF) provided usable data for cardiovascular mortality and found little or no difference between the treatment groups (moderate‐certainty evidence). Meta‐analysis of risk ratio for first heart failure hospitalisation from the PARAGON‐HF and PARALLAX trials indicated a probable modest benefit, however there was little or no effect on cardiovascular or all‐cause mortality.

A pooled analysis suggested that a slightly reduced risk of hyperkalaemia in ARNI versus standardized usual care or placebo comparison overall (moderate‐certainty evidence).

Overall completeness and applicability of evidence

This review provides the most comprehensive appraisal of the evidence to date. We included 41 studies (234 reports) that involved 26,059 participants. The included trials assessed BBs (10 studies, 3087 participants), MRAs (13 studies, 4459 participants), ACEIs (eight studies, 2061 participants), ARBs (eight studies, 8755 participants) and ARNIs (three studies, 7702 participants).

We searched clinical trials registries and identified 13 ongoing clinical trials, several of which have potential to influence the review findings. We also identified nine studies that were classified as Studies awaiting classification. There was insufficient information to determine whether these studies met our inclusion criteria. These studies were mostly small and it is therefore unlikely that they would influence the results of this review. In total, we identified 234 reports of 41 trials, nine studies awaiting classification, and 13 ongoing trials, compared with a total of 22 identified by Zheng 2018 for the same comparisons.

The LVEF threshold for defining the HFpEF trial populations varied among the included studies. This may contribute to indirectness, with implications for the applicability of the evidence. Nine studies included participants with an ejection fraction cut‐off of 40%, 12 used 45%, 15 used 50%, and one used 55%. Adamyan 2010 included participants with preserved ejection fraction but did not specify the cut‐off. SENIORS reported a subgroup of participants with LVEF greater than 40%, and for AREA IN‐CHF we obtained outcomes for a subgroup with LVEF greater than 40%.

The included studies had enrolment start dates from 1997 to 2014. In more recent studies, B‐type natriuretic peptides have been used as a key inclusion criterion to improve the specificity of the HFpEF population and enrich the trial populations for people at higher risk for clinical outcomes (e.g. CAN‐DHF; Mak 2009; RAAM‐PEF). Similarly, new measures of diastolic function have been included in more recent studies to increase the specificity (e.g. ELANDD; J‐DHF; PEP‐CHF). We noted considerable clinical heterogeneity among study populations with respect to comorbidities and cardiovascular therapies at baseline, which may influence the applicability of the evidence.

Quality of the evidence

We used the GRADE method to assess evidence certainty for the outcomes of cardiovascular mortality, all‐cause mortality, heart failure hospitalisation, quality of life (assessed using the MLHFQ), and hyperkalaemia. For BBs, evidence certainty for clinical outcomes (cardiovascular mortality, all‐cause mortality and heart failure hospitalisation) ranged from low to very low. In the combined analysis, most participants were from a subgroup of a single large trial; the other included studies were small, with high or unclear risk of bias.

For MRAs, the TOPCAT study contributed the majority of participants to the meta‐analysis, for which the overall evidence certainty for clinical outcomes was assessed as moderate. We noted differences in participant populations among the included studies (TOPCAT, LVEF greater than 45%; RAAM‐PEF, LVEF equal to or greater than 50%; AREA IN‐CHF, LVEF 40% to 45%) and it is reported that ejection fraction is a modifier of treatment effect for MRAs (Solomon 2016). Notably, a post hoc analysis of the TOPCAT study reported important differences in the placebo event rates among participants enrolled from the Americas (Argentina, Brazil, Canada, USA) and participants enrolled from Russia and Georgia (Pfeffer 2015). Furthermore, a pharmacology substudy of participants at 12 months (206 participants from USA and Canada; 160 participants from Russia) found that drug metabolites were undetectable in a greater proportion of participants from Russia, compared with participants from the USA and Canada (30% versus 3%, P < 0.001) (de Denus 2017). A geographical subgroup analysis suggested possible clinical benefit from spironolactone in HFpEF in participants who were enrolled in the Americas (United States, Canada, Brazil, Argentina; cardiovascular mortality HR 0.74, 95% CI 0.57 to 0.97; all‐cause mortality HR 0.83, 95% CI 0.68 to 1.02; heart failure hospitalisation HR 0.82, 95% CI 0.67 to 0.99). These findings will be investigated further in the ongoing SPIRRIT study (NCT02901184).

For ARBs, several large trials contributed a large number of events to the meta‐analysis for the clinical outcomes of mortality and heart failure hospitalisation, and the evidence certainty was high. For ACEIs, fewer trials were included in the combined analysis, event numbers were low and the evidence certainty was assessed as moderate. For BBs, the certainty of evidence was low due to small study sizes and risk of bias.

For ARNIs, the evidence certainty was classified as moderate for all outcomes except for all cause mortality and quality of life which was of high certainty; the cardiovascular mortality and hyperkalaemia outcomes were downgraded due to imprecision.

Potential biases in the review process

Although we conducted a comprehensive search of major databases, it is possible we missed studies on clinical trials registers, studies that had not been reported, or both. Where information on relevant subgroups or outcomes were not reported, we attempted to contact the study authors; however, only a limited number of responses were received. Given the small number of included studies per analysis, we were unable to formally assess the presence of publication bias.

Agreements and disagreements with other studies or reviews

The results of this study were largely consistent with those from a comprehensive review of the evidence for BBs and RAAS inhibitors in people with HFpEF (Zheng 2018); however, important differences were noted. Our analysis included more studies for each comparison, but the additional studies were small and did not significantly alter the overall effect estimates. In contrast to Zheng 2018, we did not find clear evidence of an effect of BB treatment on all‐cause mortality, after the exclusion of studies with an unclear or high risk of bias. Our findings were consistent with meta‐analyses of ACE and ARB (Khan 2017) and MRA (Li 2018).

A recent individual‐patient level analysis of RCTs of BBs in HF, across the full spectrum of LVEF, reported evidence that BBs reduced cardiovascular mortality; the benefit was observed for heart failure patients in sinus rhythm at all levels of LVEF less than 50%, including patients with LVEF 40% to 49% (Cleland 2018). Similarly, an individual patient‐level pooled analysis of PARADIGM‐HF and PARAGON‐HF found evidence of effect modification of ARNI efficacy by LVEF, but with benefit on heart failure hospitalisation and cardiovascular mortality probably extending to patients with mid‐range LVEF (40% to 49%) (Solomon 2020). Consistent with these reports, a study‐level meta‐analysis of trials of RAAS interventions found beneficial effects on heart failure hospitalisation in patients with LVEF greater than 50%; however, increasing LVEF was associated with a decreasing trend for benefit (Emdin 2015). Data from the CHARM‐preserved study suggests a similar benefit of ARB on heart failure hospitalisation in patients with LVEF 40% to 49%, compared to those with LVEF less than 40% (CHARM‐Preserved).

Taken together, these findings suggest that the beneficial effects of BBs and RAAS inhibitors may extend beyond patients with LVEF less than 40% to include patients with LVEF 40% to 49%; however, evidence for this group of patients is currently limited.

Study flow diagram

Figuras y tablas -
Figure 1

Study flow diagram

Ir a la figura de la revisiónAbrir en una pestaña nueva

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figuras y tablas -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figuras y tablas -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Figuras y tablas -
Analysis 1.1

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 2: Heart failure hospitalisation (RR)

Figuras y tablas -
Analysis 1.2

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 2: Heart failure hospitalisation (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 3: All‐cause mortality (RR)

Figuras y tablas -
Analysis 1.3

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 3: All‐cause mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 4: Quality of life (Minnesota)

Figuras y tablas -
Analysis 1.4

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 4: Quality of life (Minnesota)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 5: Withdrawal due to adverse event

Figuras y tablas -
Analysis 1.5

Comparison 1: Beta‐blockers versus placebo or no treatment, Outcome 5: Withdrawal due to adverse event

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Figuras y tablas -
Analysis 2.1

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 2: Heart failure hospitalisation (RR)

Figuras y tablas -
Analysis 2.2

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 2: Heart failure hospitalisation (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 3: Heart failure hospitalisation (HR)

Figuras y tablas -
Analysis 2.3

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 3: Heart failure hospitalisation (HR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 4: Hyperkalaemia

Figuras y tablas -
Analysis 2.4

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 4: Hyperkalaemia

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 5: All‐cause mortality (RR)

Figuras y tablas -
Analysis 2.5

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 5: All‐cause mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 6: Quality of life

Figuras y tablas -
Analysis 2.6

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 6: Quality of life

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 7: Quality of life (KCCQ)

Figuras y tablas -
Analysis 2.7

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 7: Quality of life (KCCQ)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 8: Quality of life (Minnesota)

Figuras y tablas -
Analysis 2.8

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 8: Quality of life (Minnesota)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 9: Withdrawal due to adverse event

Figuras y tablas -
Analysis 2.9

Comparison 2: Mineralocorticoid receptor antagonists versus placebo or no treatment, Outcome 9: Withdrawal due to adverse event

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Figuras y tablas -
Analysis 3.1

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 2: Heart failure hospitalisation (RR)

Figuras y tablas -
Analysis 3.2

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 2: Heart failure hospitalisation (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 3: Hyperkalaemia

Figuras y tablas -
Analysis 3.3

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 3: Hyperkalaemia

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 4: All‐cause mortality (RR)

Figuras y tablas -
Analysis 3.4

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 4: All‐cause mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 5: Quality of life (Minnesota)

Figuras y tablas -
Analysis 3.5

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 5: Quality of life (Minnesota)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 6: Withdrawal due to adverse event

Figuras y tablas -
Analysis 3.6

Comparison 3: Angiotensin converting enzyme inhibitors versus placebo or no treatment, Outcome 6: Withdrawal due to adverse event

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Figuras y tablas -
Analysis 4.1

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 1: Cardiovascular mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 2: Cardiovascular mortality (HR)

Figuras y tablas -
Analysis 4.2

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 2: Cardiovascular mortality (HR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 3: Heart failure hospitalisation (RR)

Figuras y tablas -
Analysis 4.3

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 3: Heart failure hospitalisation (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 4: Heart failure hospitalisation (HR)

Figuras y tablas -
Analysis 4.4

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 4: Heart failure hospitalisation (HR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 5: Hyperkalaemia

Figuras y tablas -
Analysis 4.5

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 5: Hyperkalaemia

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 6: All‐cause mortality (RR)

Figuras y tablas -
Analysis 4.6

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 6: All‐cause mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 7: All‐cause mortality (HR)

Figuras y tablas -
Analysis 4.7

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 7: All‐cause mortality (HR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 8: Quality of life (Minnesota)

Figuras y tablas -
Analysis 4.8

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 8: Quality of life (Minnesota)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 9: Withdrawal due to adverse event

Figuras y tablas -
Analysis 4.9

Comparison 4: Angiotensin receptor blockers versus placebo or no treatment, Outcome 9: Withdrawal due to adverse event

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 5: ARNI versus usual care, Outcome 1: Cardiovascular mortality (RR)

Figuras y tablas -
Analysis 5.1

Comparison 5: ARNI versus usual care, Outcome 1: Cardiovascular mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 5: ARNI versus usual care, Outcome 2: Heart failure hospitalisation, first (RR)

Figuras y tablas -
Analysis 5.2

Comparison 5: ARNI versus usual care, Outcome 2: Heart failure hospitalisation, first (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 5: ARNI versus usual care, Outcome 3: Heart failure hospitalisation (HR)

Figuras y tablas -
Analysis 5.3

Comparison 5: ARNI versus usual care, Outcome 3: Heart failure hospitalisation (HR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 5: ARNI versus usual care, Outcome 4: Hyperkalaemia

Figuras y tablas -
Analysis 5.4

Comparison 5: ARNI versus usual care, Outcome 4: Hyperkalaemia

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 5: ARNI versus usual care, Outcome 5: All‐cause mortality (RR)

Figuras y tablas -
Analysis 5.5

Comparison 5: ARNI versus usual care, Outcome 5: All‐cause mortality (RR)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 5: ARNI versus usual care, Outcome 6: Withdrawal due to adverse event

Figuras y tablas -
Analysis 5.6

Comparison 5: ARNI versus usual care, Outcome 6: Withdrawal due to adverse event

Ir a la figura de la revisiónAbrir en una pestaña nueva
Summary of findings 1. Beta‐blockers compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Beta‐blockers compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: beta‐blockers
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo or no treatment

Risk with beta‐blockers

Cardiovascular mortality (RR)
follow‐up: range 21 months to 3.2 years

Study population

RR 0.78
(0.62 to 0.99)

1046
(3 RCTs)

⊕⊕⊝⊝
LOW 1 2

Three additional studies (ELANDD; SWEDIC; Takeda 2004) reported that no deaths occurred

173 per 1000

135 per 1000
(107 to 171)

Heart failure hospitalisation (RR)
follow‐up: range 6 months to 3.2 years

Study population

RR 0.73
(0.47 to 1.13)

449
(4 RCTs)

⊕⊝⊝⊝
VERY LOW 1 3

Follow‐up unclear for SWEDIC. ELANDD reported that no hospitalisation due to heart failure occurred

117 per 1000

86 per 1000
(55 to 133)

Hyperkalaemia

follow‐up: mean 3.2 years

245 (1 RCT)

⊕⊝⊝⊝
VERY LOW1 6

J‐DHF reported one participant in the intervention group (N = 120) experienced hyperkalaemia but did not report on this outcome for the control group. No further data were available from any of the other studies.

All‐cause mortality (RR)
follow‐up: range 21 months to 3.2 years

Study population

RR 0.82
(0.67 to 1.00)

1105
(4 RCTs)

⊕⊕⊝⊝
LOW 1 2

Follow‐up unclear for Adamyan 2010. ELANDD, SWEDIC and Takeda 2004 reported that no deaths occurred.

243 per 1000

199 per 1000
(163 to 243)

Quality of life (MLHFQ): from 0 to 105
follow‐up: mean 6 months

Mean quality of life (MLHFQ) was 24

MD 1 lower
(9.05 lower to 7.05 higher)

‐

93
(1 RCT)

⊕⊝⊝⊝
VERY LOW 4 5

Lower = better, 5 point difference considered to be clinically meaningful

*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; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; RCT: Randomised controlled trial; 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

1Downgraded by one level due to study limitations (unclear selection bias in most studies).

2Downgraded by one level due to imprecision (concerns about the smaller study being more precise than the larger study).

3Downgraded by two levels due to imprecision (few events and wide CI).

4Downgraded by two levels due to imprecision (very small sample size).

5Downgraded by one level due to suspected publication bias (this is a patient‐relevant outcome that is not reported in most studies).

6Downgraded by two levels due to suspected publication bias (incomplete reporting).

Figuras y tablas -
Summary of findings 1. Beta‐blockers compared to placebo or no treatment for chronic heart failure with preserved ejection fraction
Ir a la tabla de la revisión
Summary of findings 2. Mineralocorticoid receptor antagonists (MRAs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

MRAs compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: MRAs
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo or no treatment

Risk with MRAs

Cardiovascular mortality (RR)
follow‐up: range 12 months to 3.3 years

Study population

RR 0.90
(0.74 to 1.11)

4070
(3 RCTs)

⊕⊕⊕⊝
MODERATE 1

Two additional trials (RAAM‐PEF; Kurrelmeyer 2014) reported that no deaths occurred

88 per 1000

79 per 1000
(65 to 97)

Heart failure hospitalisation (RR)
follow‐up: range 24 weeks to 3.3 years

Study population

RR 0.82
(0.69 to 0.98)

3714
(3 RCTs)

⊕⊕⊕⊝
MODERATE 1

Three additional trials (ALDO‐DHF; Kurrelmeyer 2014; Upadhya 2017) reported that no hospitalisation due to heart failure occurred

136 per 1000

112 per 1000
(94 to 134)

Hyperkalaemia
follow‐up: range 24 weeks to 3.3 years

Study population

RR 2.11
(1.77 to 2.52)

4291
(6 RCTs)

⊕⊕⊕⊕
HIGH

Two trials defined hyperkalaemia ≥ 5.5 mEg/L

83 per 1000

175 per 1000
(146 to 208)

All‐cause mortality
follow‐up: range 9 months to 3.3 years

Study population

RR 0.91
(0.78 to 1.06)

4207
(5 RCTs)

⊕⊕⊕⊝
MODERATE 1

Two additional trials (RAAM‐PEF; Kurrelmeyer 2014) reported that no deaths occurred

133 per 1000

121 per 1000
(104 to 141)

Quality of life (MLHFQ): from 0 to 105
follow‐up: range 9 months to 12 months

Mean quality of life (MLHFQ) ranged from 20 to 25

MD 0.84 higher
(2.30 lower to 3.98 higher)

‐

511
(3 RCTs)

⊕⊕⊝⊝
LOW 2 3

Lower = better, 5 points are considered a clinically significant difference

We did not pre‐specify which QoL scale was to be reported in the 'Summary of findings' table. To aid comparisons among 'Summary of findings' tables we chose to include the Minnesota Living with Heart Failure questionnaire and not the SMD across two scales

*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; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; MRAs: Mineralocorticoid receptor antagonists; RCT: Randomised controlled trial; 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

1Downgraded by one level due to imprecision.

2Downgraded by one level due to study limitations (one trial was open label).

3Downgraded by one level due to imprecision (small sample size).

Figuras y tablas -
Summary of findings 2. Mineralocorticoid receptor antagonists (MRAs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction
Ir a la tabla de la revisión
Summary of findings 3. Angiotensin‐converting enzyme inhibitors (ACEIs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

ACEIs compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: ACEIs
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo/no treatment

Risk with ACEI

Cardiovascular mortality (RR)
follow‐up: range mean 12 months to mean 26.2 months

Study population

RR 0.93
(0.61 to 1.42)

945
(2 RCTs)

⊕⊕⊕⊝
MODERATE 1

One additional trial (Kitzman 2010) reported that no deaths occurred

86 per 1000

81 per 1000
(53 to 123)

Heart failure hospitalisation (RR)
follow‐up: range 6 months to 26.2 months

Study population

RR 0.86
(0.64 to 1.15)

1019
(3 RCTs)

⊕⊕⊕⊝
MODERATE 1

13 per 1000

11 per 1000
(8 to 15)

Hyperkalaemia

follow‐up: 6 months

74 (1 RCTs)

⊕⊝⊝⊝
VERY LOW 1 3 4

One trial (Zi 2003) reported 2 events in the intervention group (N = 36), 0 events in the control group (N = 38) (RR 5.27, 95% CI 0.26 to 106.16)

All‐cause mortality (RR)
follow‐up: range mean 6 months to mean 26.2 months

Study population

RR 1.04
(0.75 to 1.45)

1187
(5 RCTs)

⊕⊕⊕⊝
MODERATE 1

One additional trial (Kitzman 2010) reported that no deaths occurred

102 per 1000

106 per 1000
(77 to 148)

Quality of life (MLHFQ): from 0 to 105
follow‐up: mean 12 months

Mean quality of life (MLHFQ) ranged from 10.9 to 29

MD 0.09 lower
(3.66 lower to 3.48 higher)

‐

154
(2 RCTs)

⊕⊕⊝⊝
LOW 2 3

Scale: 0 to 105, lower = better, 5 point difference considered clinically relevant

One trial (SNEGOVIK) reported mean change from baseline of ‐19.8 for intervention and ‐10.7 for control

*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).
ACEIs: Angiotensin‐converting enzyme inhibitors; CI: Confidence interval; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; RCT: Randomised controlled trial; 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

1Downgraded by one level due to imprecision (wide CI).

2Downgraded by one level due to study limitations (risk of bias (open label)).

3Downgraded by one level due to imprecision (low sample size).

4Downgraded by one level due to study limitations (unclear selection bias).

Figuras y tablas -
Summary of findings 3. Angiotensin‐converting enzyme inhibitors (ACEIs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction
Ir a la tabla de la revisión
Summary of findings 4. Angiotensin receptor blockers (ARBs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

ARBs compared to placebo or no treatment for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: ARBs
Comparison: placebo or no treatment

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with placebo or no treatment

Risk with ARBs

Cardiovascular mortality (RR)
follow‐up: range mean 12 months to mean 49.5 months

Study population

RR 1.02
(0.90 to 1.14)

7254
(3 RCTs)

⊕⊕⊕⊕
HIGH

One additional trial (Parthasarathy 2009) reported that no deaths occurred

131 per 1000

133 per 1000
(118 to 149)

Heart failure hospitalisation (RR)
follow‐up: range mean 12 months to mean 49.5 months

Study population

RR 0.92
(0.83 to 1.02)

7254
(3 RCTs)

⊕⊕⊕⊕
HIGH

171 per 1,‐000

157 per 1,‐000
(142 to 174)

Hyperkalaemia
follow‐up: range 36.6 months to 49.5 months

Study population

RR 1.88
(1.07 to 3.33)

7148
(2 RCTs)

⊕⊕⊕⊕
HIGH

3 per 1,000

5 per 1,000
(3 to 8)

All‐cause mortality (RR)
follow up: range 1 years to 4.4 years

Study population

RR 1.01
(0.92 to 1.11)

7964
(4 RCTs)

⊕⊕⊕⊕
HIGH

One additional trial (Parthasarathy 2009) reported that no deaths occurred

72 per 1000

73 per 1,‐000
(66 to 80)

Quality of life (MLHFQ): from 0 to 105
follow‐up: range mean 13.8 weeks to mean 49.5 months

Mean quality of life (MLHFQ) ranged from 10.9 to 31.6

MD 0.41 higher
(0.86 lower to 1.67 higher)

‐

3117
(3 RCTs)

⊕⊕⊕⊕
HIGH

Scale: 0 to 105, lower = better, 5 point difference considered clinically relevant

*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).
ARBs: Angiotensin receptor blockers; CI: Confidence interval; MD: Mean difference; MLHFQ: Minnesota Living with Heart Failure Questionnaire; RCT: Randomised controlled trial; 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

Figuras y tablas -
Summary of findings 4. Angiotensin receptor blockers (ARBs) compared to placebo or no treatment for chronic heart failure with preserved ejection fraction
Ir a la tabla de la revisión
Summary of findings 5. Angiotensin receptor neprilysin inhibitors (ARNIs) compared to usual care for chronic heart failure with preserved ejection fraction

ARNIs compared to usual care for chronic heart failure with preserved ejection fraction

Patient or population: patients with chronic heart failure with preserved ejection fraction
Setting: secondary care
Intervention: ARNI (sacubitril‐valsartan)
Comparison: ARB (valsartan)

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with ARNI

Cardiovascular mortality (RR)

Median follow‐up: 35 months

Study population

RR 0.96
(0.79 to 1.15)

4796
(1 study)

⊕⊕⊕⊝
MODERATE1

89 per 1,000

85 per 1,000
(70 to 102)

Heart failure hospitalisation, first (RR)

Range of follow‐up: 24 weeks to 35 months

Study population

RR 0.89

(0.80 to 1.00)

7362

(2 studies)

⊕⊕⊕⊝
MODERATE1

142 per 1,000

126 per 1,000

(113 to 142)

Hyperkalaemia

Range of follow‐up: 36 weeks to 35 months

Study population

RR 0.88
(0.77 to 1.01)

5054
(2 studies)

⊕⊕⊕⊝
MODERATE1

147 per 1,000

129 per 1,000
(113 to 148)

All‐cause mortality (RR)

Range of follow‐up: 36 weeks to 35 months

Study population

(RR 0.97

CI 0.84 to 1.11)

7663
(3 studies)

⊕⊕⊕⊕
HIGH

138 per 1,000

134 per 1,000
(117 to 153)

Quality of life

Range of follow‐up: 36 weeks to 35 months

PARAMOUNT reported change from baseline for the KCCQ overall summary score for the intervention arm (n =118) as 11.25 (2.185) and the control arm (n = 116) as 11.31 (2.183) and summarised the findings as "no difference in KCCQ score between treatment groups".

PARAGON‐HF reported a difference of the clinical summary score KCCQ between treatment arms as 1.0 (0.0 to 2.1).

PARALLAX reported "KCCQ improved in both treatment groups, with an early benefit of S/V that was no longer significant after 24 week".

⊕⊕⊕⊕
HIGH

*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).

ARB: Angiotensin receptor blocker; ARNIs: Angiotensin receptor neprilysin inhibitors; CI: Confidence interval; KCCQ: Kansas City Cardiomyopathy Questionnaire; MD: Mean difference; RCT: Randomised controlled trial; 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

1Downgraded by one level due to imprecision.

Figuras y tablas -
Summary of findings 5. Angiotensin receptor neprilysin inhibitors (ARNIs) compared to usual care for chronic heart failure with preserved ejection fraction
Ir a la tabla de la revisión
Comparison 1. Beta‐blockers versus placebo or no treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Cardiovascular mortality (RR) Show forest plot

3

1046

Risk Ratio (M‐H, Fixed, 95% CI)

0.78 [0.62, 0.99]

1.2 Heart failure hospitalisation (RR) Show forest plot

4

449

Risk Ratio (M‐H, Fixed, 95% CI)

0.73 [0.47, 1.13]

1.3 All‐cause mortality (RR) Show forest plot

4

1105

Risk Ratio (M‐H, Fixed, 95% CI)

0.82 [0.67, 1.00]

1.4 Quality of life (Minnesota) Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

1.5 Withdrawal due to adverse event Show forest plot

2

338

Risk Ratio (M‐H, Fixed, 95% CI)

18.07 [2.45, 133.04]
Figuras y tablas -
Comparison 1. Beta‐blockers versus placebo or no treatment
Ir a la tabla de la revisión
Comparison 2. Mineralocorticoid receptor antagonists versus placebo or no treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

2.1 Cardiovascular mortality (RR) Show forest plot

3

4070

Risk Ratio (M‐H, Fixed, 95% CI)

0.90 [0.74, 1.11]

2.2 Heart failure hospitalisation (RR) Show forest plot

3

3714

Risk Ratio (M‐H, Fixed, 95% CI)

0.82 [0.69, 0.98]

2.3 Heart failure hospitalisation (HR) Show forest plot

2

3670

Hazard Ratio (IV, Fixed, 95% CI)

0.82 [0.69, 0.98]

2.4 Hyperkalaemia Show forest plot

6

4291

Risk Ratio (M‐H, Fixed, 95% CI)

2.11 [1.77, 2.51]

2.5 All‐cause mortality (RR) Show forest plot

5

4207

Risk Ratio (M‐H, Fixed, 95% CI)

0.91 [0.78, 1.06]

2.6 Quality of life Show forest plot

5

603

Std. Mean Difference (IV, Random, 95% CI)

0.05 [‐0.23, 0.34]

2.7 Quality of life (KCCQ) Show forest plot

2

92

Mean Difference (IV, Random, 95% CI)

‐0.78 [‐28.02, 26.46]

2.8 Quality of life (Minnesota) Show forest plot

3

511

Mean Difference (IV, Random, 95% CI)

0.84 [‐2.30, 3.98]

2.9 Withdrawal due to adverse event Show forest plot

5

4037

Risk Ratio (M‐H, Fixed, 95% CI)

1.10 [1.00, 1.21]
Figuras y tablas -
Comparison 2. Mineralocorticoid receptor antagonists versus placebo or no treatment
Ir a la tabla de la revisión
Comparison 3. Angiotensin converting enzyme inhibitors versus placebo or no treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

3.1 Cardiovascular mortality (RR) Show forest plot

2

945

Risk Ratio (M‐H, Fixed, 95% CI)

0.93 [0.61, 1.42]

3.2 Heart failure hospitalisation (RR) Show forest plot

3

1019

Risk Ratio (M‐H, Fixed, 95% CI)

0.86 [0.64, 1.15]

3.3 Hyperkalaemia Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

3.4 All‐cause mortality (RR) Show forest plot

5

1187

Risk Ratio (M‐H, Fixed, 95% CI)

1.04 [0.75, 1.45]

3.5 Quality of life (Minnesota) Show forest plot

2

154

Mean Difference (IV, Fixed, 95% CI)

‐0.09 [‐3.66, 3.48]

3.6 Withdrawal due to adverse event Show forest plot

4

1127

Risk Ratio (M‐H, Random, 95% CI)

2.52 [0.49, 12.87]
Figuras y tablas -
Comparison 3. Angiotensin converting enzyme inhibitors versus placebo or no treatment
Ir a la tabla de la revisión
Comparison 4. Angiotensin receptor blockers versus placebo or no treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

4.1 Cardiovascular mortality (RR) Show forest plot

3

7254

Risk Ratio (M‐H, Fixed, 95% CI)

1.02 [0.90, 1.14]

4.2 Cardiovascular mortality (HR) Show forest plot

2

7148

Hazard Ratio (IV, Fixed, 95% CI)

1.00 [0.89, 1.13]

4.3 Heart failure hospitalisation (RR) Show forest plot

3

7254

Risk Ratio (M‐H, Fixed, 95% CI)

0.92 [0.83, 1.02]

4.4 Heart failure hospitalisation (HR) Show forest plot

2

7148

Hazard Ratio (IV, Fixed, 95% CI)

0.90 [0.80, 1.01]

4.5 Hyperkalaemia Show forest plot

2

7148

Risk Ratio (M‐H, Fixed, 95% CI)

1.88 [1.07, 3.33]

4.6 All‐cause mortality (RR) Show forest plot

4

7964

Risk Ratio (M‐H, Fixed, 95% CI)

1.01 [0.92, 1.11]

4.7 All‐cause mortality (HR) Show forest plot

2

4838

Hazard Ratio (IV, Fixed, 95% CI)

0.99 [0.88, 1.12]

4.8 Quality of life (Minnesota) Show forest plot

3

3117

Mean Difference (IV, Fixed, 95% CI)

0.41 [‐0.86, 1.67]

4.9 Withdrawal due to adverse event Show forest plot

4

7406

Risk Ratio (M‐H, Fixed, 95% CI)

1.22 [1.09, 1.36]
Figuras y tablas -
Comparison 4. Angiotensin receptor blockers versus placebo or no treatment
Ir a la tabla de la revisión
Comparison 5. ARNI versus usual care

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

5.1 Cardiovascular mortality (RR) Show forest plot

1

Risk Ratio (M‐H, Fixed, 95% CI)

Totals not selected

5.2 Heart failure hospitalisation, first (RR) Show forest plot

2

7362

Risk Ratio (M‐H, Fixed, 95% CI)

0.89 [0.80, 1.00]

5.3 Heart failure hospitalisation (HR) Show forest plot

2

7362

Hazard Ratio (IV, Fixed, 95% CI)

0.87 [0.76, 0.99]

5.4 Hyperkalaemia Show forest plot

2

5054

Risk Ratio (M‐H, Fixed, 95% CI)

0.88 [0.77, 1.01]

5.5 All‐cause mortality (RR) Show forest plot

3

7663

Risk Ratio (M‐H, Fixed, 95% CI)

0.97 [0.84, 1.11]

5.6 Withdrawal due to adverse event Show forest plot

3

7663

Risk Ratio (M‐H, Fixed, 95% CI)

1.02 [0.91, 1.14]
Figuras y tablas -
Comparison 5. ARNI versus usual care
Ir a la tabla de la revisión