Effectiveness and safety of chronic diuretic use in older adults: an umbrella review of recently published systematic reviews and meta-analyses of randomized-controlled trials

Key summary points

Aim

This umbrella review aimed to summarize the literature on the efficacy and safety of chronic diuretic treatment in adults.

Findings

Certain diuretics or diuretic subclasses offer significant benefits for key clinical outcomes, such as reducing cardiovascular mortality and heart failure-related hospitalizations. However, chronic diuretic use carries potential risks, including an increased risk of hyperkalemia, with age-related differences in adverse event risks [older adults (≥ 65 years) facing higher risks, while younger populations do not show similar concerns].

Message

Further research is needed to establish diuretic efficacy and safety in populations commonly seen in clinical practice, especially older adults living with multimorbidity and frailty.

Supplementary Information

The online version contains supplementary material available at 10.1007/s41999-025-01229-5.

Abstract

Background

Healthcare providers should balance the potential risks and benefits of chronic diuretic use, particularly in older adults, as with age, diuretic benefits may decline and risks increase. A comprehensive synthesis and critical evaluation of the available evidence on chronic diuretic treatment effects is currently lacking.

Methods

We conducted an umbrella review of systematic reviews and meta-analyses published since 2018 on health outcomes associated with diuretic use in randomized-controlled trials (RCTs). We conducted random-effects meta-analysis for pooled effect estimates and narratively summarized data that could not be pooled.

Results

We included 741 effect estimations from 117 systematic reviews (SRs) on 1566 RCTs in individuals aged 62 ± 6 years. Of our 33 meta-analyses, 11 provided convincing, high-quality evidence: finerenone reduced the risk of cardiovascular (CV) mortality and end-stage kidney disease in individuals with chronic kidney disease (CKD) and/or type 2 diabetes (T2D). Torasemide reduced the risk of heart failure-related hospitalization (HFH) more than furosemide in individuals with HF. Thiazides reduced CV events in individuals with hypertension. Mineralocorticoid receptor antagonists (MRAs) reduced HFH, but also increased hyperkalemia risk in individuals with HF. MRAs also reduced the risk of atrial fibrillation in those with HF or CVD, and reduced HFH, major adverse cardiovascular events (MACEs), > 40% eGFR decrease, and composite kidney outcomes in individuals with CKD and/or T2D. Lower quality evidence suggests that in older (≥ 65 years), but not in younger adults, diuretics may reduce CV mortality, but also increase adverse event (AE) risk.

Conclusions

Our umbrella review offers a comprehensive and up-to-date evaluation of the benefits and harms of diuretics. However, further research is needed to establish their efficacy and safety in populations commonly seen in clinical practice, especially older adults living with multimorbidity and frailty.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s41999-025-01229-5.

Introduction

Cardiovascular disease (CVD) is a leading cause of morbidity and mortality globally, disproportionately affecting older individuals [1, 2]. It is a primary cause of both death and disability in aging populations, with a particularly high burden in those aged ≥ 75 due to age-related physiological changes [3]. Diuretics have long been a cornerstone in the management of heart failure (HF) [4] and hypertension (HT) [5] and are widely used worldwide to prevent CVD and mitigate end-organ damage [5, 6].

Diuretic use can cause significant harm, particularly in older adults or those with limited life expectancy [7–10]. These individuals often derive less benefit from medications compared to younger, healthier individuals as they face competing risks such as premature death before any therapeutic benefits can be realized. Prolonged diuretic use in the absence of fluid retention lacks substantial evidence for benefit and may be associated with potentially life-threatening adverse events (AEs). These include interference with the guideline-directed up-titration of evidence-based HF medications [11, 12], the development of cardiorenal syndrome, diuretic resistance [13], or neurohumoral hyperactivation, especially in frail older adults [14–16]. Additional instances of inappropriate diuretic prescribing arise when diuretics are prescribed as part of an inappropriate prescribing cascade [17–19], are continued beyond the necessary duration, or prescribed at excessive doses [20]. Observational studies have shown that inappropriate loop diuretic use is prevalent among older adults [21, 22]. For example, a study using the STOPP (Screening Tool of Older People's Prescriptions) criteria [23] found that 26% of potentially inappropriate medications identified in older adults receiving home-based medical services were related to diuretic prescriptions [22]. Inappropriate diuretic prescribing is linked to higher risks of hospitalization and mortality [14, 24].

When continued chronic diuretic use is deemed potentially inappropriate, treatment options, such as dose reduction, withdrawal (deprescribing), or switching to a safer alternative, should be considered. Deprescribing involves the careful, supervised withdrawal of medications when the risks outweigh the benefits, taking into account individual factors such as patient preferences, goals of care, and remaining life expectancy [25–28]. Recent reviews suggest that diuretic deprescribing is a safe and feasible option for carefully selected individuals, potentially preventing AEs, but evidence on the long-term outcomes of deprescribing diuretics is limited [15, 29]. Undertreatment should be avoided as congestion is a key contributor to symptoms, poor quality of life (QoL), and adverse outcomes, particularly in persons with HF [30]. Furthermore, diuretics were among the most frequently mentioned medications that older adults would like to have deprescribed, according to a recent survey of 1,340 individuals across 14 European countries [31]. One of the main reasons older adults are eager to deprescribe diuretics is that these medications, by promoting urinary incontinence, can be burdensome, limiting social interactions and outdoor activities [32]. Thus, clinicians must routinely evaluate the ongoing use of diuretics in older adults, and guide discussions with their patients regarding continued prescribing, not prescribing or deprescribing based on thorough knowledge of the available evidence regarding the risks and benefits of chronic diuretic use.

Numerous systematic reviews (SRs) and meta-analyses (MAs) have been published, addressing the effects of chronic diuretic use on health outcomes across various populations. Although necessary to ensure patient-centered, appropriate diuretic prescribing [30], a comprehensive synthesis and critical evaluation of this published evidence is currently lacking. To address this gap, we conducted an umbrella review of SRs and MAs published from January 1, 2018, assessing the breadth, credibility, and certainty of associations between diuretic use and health outcomes through RCTs. A deeper understanding of the benefits and harms of chronic diuretic therapy, especially in older adults, is essential for optimal diuretic prescribing.

Methods

We conducted an umbrella review following guidance from the Cochrane Handbook, Chapter V: Overviews of Reviews [33] and followed the reporting guideline for overviews of reviews of healthcare interventions (PRIOR statement) [34]. The protocol for this study was prospectively registered on PROSPERO, the international prospective register of SRs (CRD42023423486).

Search strategy

The search strategy was developed in consultation with an experienced medical librarian. We systematically searched MEDLINE, Embase, and the Cochrane Library (CDSR and CENTRAL) for publications from January 1, 2018 to November 11, 2024 to identify SRs and MAs of RCTs examining associations between chronic diuretic use and health outcomes in adults. The search strategy is presented in Supplementary Table 1. Database searches were supplemented by reference list checking of the included references.

Inclusion and exclusion criteria

Two independent reviewers screened titles/abstracts and selected relevant records for eligibility. We considered SRs and MAs for inclusion, comparing the effects of chronic diuretic use with placebo or control, and the effects of diuretics (or subclasses) compared to the effects other diuretics (or subclasses) in adults (≥ 18 years). We considered all diuretics listed in the WHO (World Health Organization) ATC (Anatomical Therapeutic Chemical) classification code list, except vaptans. Language was restricted to English. We excluded MAs that investigated cost-effectiveness. Our PICOS (participants, intervention, control, outcomes, and study design) is presented in Supplementary Table 2.

Data collection and data synthesis

Relevant aggregated data from the included SRs were extracted into a predefined data collection form by two independent reviewers. The two independent reviewers discussed their conflicts until a final consensus was reached. When consensus was not reached, a third senior independent reviewer made the final decision. We extracted quantitative meta-analytic data together with the corresponding 95% confidence intervals (95% CI). If SRs examined more than one health outcome, each outcome was recorded separately.

We categorized the collected data based on population or indication for diuretic use, based on diuretic subclass and health outcome. If, for a certain specific outcome, there were ≥ 3 effect estimates originating from ≥ 2 SRs, reporting on a similar diuretic subclass within a comparable population and comparison group, we performed MAs (see paragraph on statistical analysis). In summary-of-findings (SoF) tables, we summarized the extracted data (quantitatively for MAs, narratively for data for which pooling was not appropriate). Per outcome (sub)category, we analyzed the data. We included data on age-related subgroup and meta-regression analyses.

Statistical analysis

The statistical analyses were performed with Comprehensive Meta-Analysis Version 4, (Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H., Biostat, Englewood, New Jersey, United States of America). For each MA, we estimated summary effect sizes and 95% CIs using random-effects models [35, 36] and estimated between-study heterogeneity was assessed using the I² metric. I² ranges between 0 and 100% and is the ratio of between-study variance over the sum of the within-study and between-study variances. Values exceeding 50 or 75% are generally considered large or very large heterogeneity, respectively [37]. In addition, we estimated the 95% prediction interval, which accounts for the between-study heterogeneity and evaluates the uncertainty for the effect expected in a new study addressing the same association [38, 39].

To ensure high quality and rigor of our work, we accounted for publication overlap in SRs and MAs (reviews or analyses including the same primary publications). We tabulated the RCTs from the included MAs that were originally involved in each review and calculated the pairwise overlap; the corrected covered area (CCA) [40, 41]. We visualized and quantified overlap for each MA using the GROOVE (Graphical Representation of Overlap for OVErviews) tool [42]. We classified overlap as “slight” (< 5%), “moderate” (5 to < 10%), “high” (10 to < 15%), or “very high” (≥ 15%).

Risk of bias and quality assessment

Two independent reviewers assessed the risk of bias (RoB) of the included SRs according to the standardized JBI (Joanna Briggs Institute) critical appraisal checklist for SRs and research syntheses (available from: https://joannabriggs.org/critical-appraisaltools). In this checklist, each of the 11 questions about the study’s methodology must be answered with one in four options: yes (Y), no (N), unclear (U), or not applicable (N/A). Disagreements in scoring were resolved by discussion between the two independent reviewers, and further discussed with a third independent reviewer until we reached consensus.

The overall RoB for each SR included in this umbrella review was calculated via merging the number of questions answered as “yes” in high, moderate or low RoB. If any question was answered with “not applicable”, it was not considered in the calculation of bias risk. Studies were classified as: low ≥ 70%, moderate between 50 and 70%, and high RoB with a score ≤ 49%. We performed sensitivity analyses, excluding data from SRs with high RoB.

Grading the evidence

We evaluated the evidence from our MAs applying the GRADE (Grading of Recommendations, Assessment, Development and Evaluation) assessment criteria [43] to classify the certainty of the evidence as i: high (we are confident that the true effect is similar to the estimated effect); ii. moderate (we expect the true effect to be close to the estimated effect); iii. low (we expect the true effect to be fairly different from the estimated effect) or vi: very low (we expect the true effect to be markedly different from the estimated effect (GRADEpro version 3.6.1; McMaster University, ON, Canada). Findings began as high certainty evidence, we downgraded by one point to moderate certainty evidence if i: overall RoB was “moderate”, and two points to low certainty if “high” in at least 50% of MAs, or ii: if there was inconsistency in the net effect (e.g., neutral in some, but lower in others). For statistically significant differences in continuous effect size estimations [mean differences (MDs), standardized MDs (SMDs), or weighted MDs (WMDs)], we additionally rated clinical relevance based on reports from the international guidelines or reports from the literature [e.g., minimally clinically important differences (MCIDs)].

Results

Search and descriptive results

Our search yielded 8076 unique results of which we deemed 278 suitable to cross-check for inclusion. After excluding a total of 161 records that did not match our inclusion criteria, we included 117 SR articles in our umbrella review (Fig. 1), reporting on 1566 RCTs among over 1.5 million participants treated with diuretics with a mean age of 62 ± 6 years. Characteristics of the included SRs are presented in Table 1. Overall RoB of the included SRs was low in 88%, moderate in 9%, and high in 3% (Supplementary Table 3).

Fig. 1.

PRISMA flow diagram

Table 1.

Included systematic review and meta-analysis articles

Review first author, publication year	Population, setting	Comparison	Number of individuals studied	Follow-up duration (mean/median or range)	Age of individuals (in years, mean/median or range)	Percentage females	Number of included RCTs on diuretics (K)
Abdelazeem et al., 2022 [77]	CKD and T2D	Finerenone vs placebo	13,847	1.6 years	64.7 ± 8.7	30	3
Abraham et al., 2020 [78]	CHF	Torasemide vs furosemide	–	15 months	–	–	9
Ahmed et al., 2023 [79]	HT (resistant)	MRA vs placebo	1414	16 weeks	58.89 ± 5.05	45	8
Albasri et al., 2021 [80]	HT	Thiazide vs placebo	24,311	2 to 4 years	–	–	5
Alexandre et al., 2019 [48]	not specified	MRA vs placebo	2843	12 months	64.2	–	11
Alexandrou et al., 2019 [81]	CVD	MRA vs placebo/active control	2767	–	–	–	31
Al-Sadawi et al., 2024 [82]	HFpEF	MRA vs placebo	1426	31 months	71	50	4
Asiimwe et al., 2021 [83]	not specified	Diuretic vs no diuretic	85,555	–	–	–	19
Bao et al., 2022 [84]	DKD	Finerenone vs placebo	13,510	–	–	–	4
Bazoukis et al., 2018a [85]	HFpEF and HFrEF	MRA vs placebo	13,354	9.4 months	66,3	34	7
Bazoukis et al., 2018b [86]	HT (resistant)	MRA vs placebo/active control	2736	4.83 months	58.5	36	21
Bidel et al., 2023 [49]	not specified	Thiazide vs placebo	–	4.2 years	66.2 ± 9.8 (women); 64.2 ± 9.4 (men)	42	–
Bonsu et al., 2018 [87]	HFpEF	MRA vs placebo	5238	3 to 40 months	67.7	59	12
Boulmpou et al., 2022 [88]	HFpEF	Spironolactone vs placebo	815	7 days to 2 years	–	–	4
Chen et al., 2024 [89]	CKD	Non-steroidal MRA or eplerenone vs placebo	15,817	–	–	–	11
Clark et al., 2024 [90]	HFpEF and HFrEF	MRA vs placebo	–	–	–	–	5
Desbiens et al., 2022 [50]	HT	Thiazide vs placebo/no treatment	68,107	4 weeks to 5.6 years	34 to 84 years	46	30
Dineva et al., 2019 [91]	HT	Chlorthalidone vs HCTZ	51,789	4 to 364 weeks	–	–	7
Dineva et al., 2020 [92]	HT	HCTZ vs placebo; chlorthalidone vs placebo; HCTZ vs chlorthalidone	–	4 to 364 weeks	–	–	33
Ding et al., 2023 [93]	CKD	MRA (network meta-analysis)	15,531	–	–	–	26
Du et al., 2024 [94]	CKD and T2D	Finerenone vs placebo	14,875	3 to 40.8 months	66.3	30	4
Dutta et al., 2022 [95]	DKD	Finerenone vs placebo/active control	–	42 days to 3.4 years	–	–	7
Eid et al., 2021 [96]	CHF	Furosemide vs other loop diuretics	2709	3 to 728 days	–	–	34
Elshahat et al., 2024 [97]	CHF	Eplerenone vs spironolactone	21,930	20 months	58.1	–	4
Faisal et al., 2022 [98]	HFpEF	Spironolactone vs placebo	–	12 or 40 months	–	–	2
Farmakis et al., 2022 [99]	HT	Thiazide vs placebo	7784	8 weeks to 5.8 years	56.2 ± 9.6	0	16
Fatima et al., 2023 [100]	not specified	MRA vs control	11,356	8 days to 38 months	66	–	10
Fernandes et al., 2018 [101]	HFpEF	MRA vs placebo	725	6 to 26 months	65	–	5
Ferre et al., 2021 [102]	Idiopathic hypercalciuria	Diuretic vs placebo/control	446	–	–	–	4
Frankenstein et al., 2020 [103]	HFrEF (NYHA class II to IV)	MRA vs placebo/standard medical care	12,213	4 to 60 months	46 to 70	–	14
Fu et al., 2021 [104]	CKD	Finerenone vs placebo	7048	1 to 30 months	–	–	4
Fukuta et al., 2019 [105]	HFpEF	MRA vs placebo	755	6 to 12 months	–	–	6
Geng et al., 2023 [51]	CHF	MRA vs placebo	9056	2.1 years	68	–	7
Ghosal and Sinha 2022b [106]	T2D	Finerenone vs placebo	13,026	–	–	–	2
Ghosal and Sinha 2023a [107]	CKD and T2D	Finerenone vs placebo	13,943	3 to 40.8 months	–	–	4
Gu et al., 2024 [108]	T2D	Finerenone vs placebo	13,847	–	59 to 66	65	3
Hall et al., 2020 [109]	HT	Thiazide vs placebo	76,608	35 weeks	58	45	95
Hansen et al., 2020 [110]	HFrEF	MRA vs placebo	–	1.6 to 4.1 years	–	–	3
Harrington et al., 2023 [111]	HFrEF	MRA vs placebo	–	–	–	–	3
Hasegawa et al., 2021 [112]	CKD	MRA vs placebo/standard care	1446	–	–	–	16
Ho et al., 2024 [113]	primary hyperaldosteronism	Eplerenone vs spironolactone	392	–	43.5 to 59.5	32 to 60	5
Hu et al., 2022 [114]	DKD	Eplerenone vs placebo/active control	838	6 to 48 weeks	–	–	8
Jiang et al., 2022 [115]	CKD and T2D	MRA vs placebo	14,450	–	–	–	8
Jyotsna et al., 2023 [116]	CKD (stages 1 to 4) and T2D	Finerenone vs placebo	39,995	–	–	–	7
Kapelios et al., 2019 [117]	HFpEF	MRA vs placebo/control	1164	6 to 18 months	–	–	9
Karakasis et al., 2024 [118]	CVD	MRA vs placebo	13,358	–	48.7 to 73	59	23
Kido et al., 2019 [119]	HFpEF and HFrEF	Torasemide vs furosemide	–	–	–	–	2
Kohjimoto et al., 2024 [120]	kidney stones (calcium oxalate)	Thiazide vs no thiazide	571	–	–	–	8
Li et al., 2018 [52]	HFpEF	Spironolactone vs placebo	4147	6 to 39.6 months	62 to 71	52 to 100	7
Li et al., 2020 [121]	kidney stones	Thiazide vs placebo/no treatment	–	–	–	–	8
Li et al., 2022 [122]	CKD and T2D	Finerenone vs placebo	–	–	–	–	2
Li et al., 2024 [123]	CHF	Spironolactone vs placebo	–	–	–	–	2
Liu et al., 2022 [124]	ESKD requiring dialysis	Spironolactone vs placebo/standard medical therapy	1258	2 weeks to 3 years	52.92 ± 6.90 to 70.45 ± 9.70	–	15
Lunney et al., 2020 [125]	CHF and CKD	MRA vs placebo	826	6 to 24 months	–	–	3
Ma et al., 2021 [126]	HT	ARB/HCTZ vs ARB alone	–	4 to 12 weeks	–	–	16
Macfarlane et al., 2019 [127]	HT (primary)	Indapamide vs bendroflumethiazide	21,315	18 months to 5 years	–	–	3
Martin et al., 2021 [128]	HFpEF	MRA vs placebo/no treatment	4459	24 weeks to 3.3 years	54.5 to 80	–	13
Martins et al., 2023 [129]	HT (primary)	Thiazide vs placebo	58,807	10 weeks	55	45	276
Miles et al., 2019 [130]	CHF	Torasemide vs furosemide	8127	5 to 12 months	63 to 75	22 to 58	10
Morita et al., 2022 [131]	HT and DKD	MRA vs placebo; MRA/SGLT2i vs SGLT2i alone	–	–	–	–	8
Musini et al., 2019 [44]	HT	Diuretic vs placebo/no treatment	–	3.8 years	–	–	8
Nguyen et al., 2023 [132]	CKD and DKD	Non-steroidal MRA vs placebo	6519	24 to 278 weeks	–	–	2
Noone et al., 2020 [133]	HT	Thiazide-like vs placebo	–	> 4 weeks	–	–	93
Oraii et al., 2024 [134]	CHF	MRA vs placebo	21,791	33.3 months	65.2	31	7
Pamporis et al., 2024 [135]	HFrEF	MRA vs control	15,685	4 to 540 weeks	–	–	32
Patoulias et al., 2021 [136]	CVD	MRA vs placebo	9866	–	–	–	6
Patoulias et al., 2022 [137]	CKD and T2D	Finerenone vs placebo	13,036	–	–	–	2
Peters et al., 2020 [138]	HT	Diuretic vs placebo/no treatment	–	> 1 year and > 5 years	–	–	4
Sakima et al., 2021 [53]	HT	MRA vs placebo/active control	877	4 weeks to 2 years	34 to 77	–	17
Sampaio Rodrigues et al., 2024 [139]	not specified	MRA vs placebo/standard medical therapy	11,419	6 months to 3.3 years	–	–	12
Seeley et al., 2020 [54]	HT	Diuretic vs placebo	–	–	51.2 ± 5.72	58	17
Shah et al., 2018 [60]	CHF	Torasemide vs furosemide	664	11 months	67,2	57	3
Shaman et al., 2020 [55]	HT	MRA vs placebo	4283	6 months	57	39	40
Sherif et al., 2019 [70]	CHF	Torasemide vs furosemide	–	–	55 to 82.3	–	14
Shi et al., 2023 [140]	T2D	Non-steroidal MRA vs standard treatment	471,038	6.0 months	57,7	43	–
Siddiqi et al., 2023 [141]	CHF	Furosemide vs torasemide	1996	240 days	58.2 to 75.1	–	13
Singh et al., 2023 [142]	CHF	Furosemide vs torasemide	4127	3 to 18 months	68	44	10
Sreenivasan et al., 2024 [143]	HFpEF	MRA vs placebo	–	15 months	71.7 ± 4.2	49	6
Sreenivasan et al., 2022 [144]	HFpEF	MRA vs placebo	–	15 months	62.4 to 80	49	6
Täger et al., 2019 [145]	CHF (systolic)	Diuretic vs placebo	2647	–	56 to 76	–	34
Teixeira et al., 2024 [146]	CHF	Torasemide vs furosemide	4115	–	–	–	12
Teles et al., 2023 [147]	HT	Thiazide(like) vs loop diuretic	–	24 days to 20 weeks	53.7 ± 7.8 to 67.5 ± 10.2	–	4
Thomopoulos et al., 2018 [47]	HT	Diuretic vs placebo	39,751	–	–	–	10
Tian et al., 2023 [148]	HT (resistant)	Spironolactone vs placebo	–	14.5 weeks	59 ± 3.68	39	7
Tsukamoto et al., 2022 [149]	CKD and DKD	MRA vs placebo	36,186	9.2 months to 4.2 years	–	–	3
Wang et al., 2023 [150]	CHF	MRA vs placebo	14,698	–	–	27	5
Wei et al., 2020 [151]	CVD	Thiazide vs placebo/standard medical treatment	–	> 6 months	–	–	13
Wright et al., 2018 [152]	HT (primary)	Thiazide vs placebo/no treatment	39,713	4.1 years	61	–	19
Wu et al., 2022 [153]	DKD	MRA vs placebo	–	12 to 54 weeks	45.6 ± 13.1 to 65.4 ± 8.9	–	12
Xiang et al., 2019 [154]	HFmrEF and HFpEF	Spironolactone vs placebo/standard medical therapy	4539	4 to 39 months	58 to 71.55	–	11
Xiang et al., 2022 [155]	HFmrEF and HFpEF	MRA vs placebo	–	6 to 39.6 months	66.3 to 76.4	–	5
Xie et al., 2018 [156]	HT	Diuretic vs placebo	–	–	–	–	2
Xu et al., 2018 [157]	not specified	MRA vs placebo/standard medical therapy	11,365	1 to 24 months	52 ± 10 to 64 ± 11	–	13
Xu et al., 2024 [158]	CKD and T2D	Non-steroidal MRA vs placebo	14,997	–	–	–	7
Yanai et al., 2021 [159]	CKD	MRA vs placebo	–	> 13 weeks	–	–	4
Yang et al., 2019 [160]	CHF (EF ≤ 45%)	MRA vs placebo; finerenone vs spironolactone; finerenone vs eplerenone; spironolactone vs eplerenone	13,597	1 to 24 months	–	–	13
Yang et al., 2022 [161]	CKD and T2D	Non-steroidal MRA vs placebo; eplerenone vs placebo; spironolactone vs placebo; non-steroidal MRA vs Nonselective MRA	15,025	–	–	–	12
Yang et al., 2023 [162]	CKD and T2D	Finerenone vs placebo	13,679	90 days to 3.4 years	62.4 to 66.8	20 to 30	4
Yang et al., 2024 [163]	DKD	Non-steroidal MRA vs placebo	–	–	–	–	5
Yang et al., 2023 [164]	CKD and DKD	MRA vs placebo	–	2 years	65.8	39	5
Yasmin et al., 2023 [165]	CKD	Finerenone vs placebo	15,462	–	–	–	7
Yi et al., 2024 [57]	HT	MRA or thiazide vs placebo	7758	–	–	–	15
Yuan et al., 2023 [56]	CKD	MRA vs control	22,792	2 to 42 months	34 to 74	0 to 81	53
Zafeiropoulos et al., 2024 [166]	HFmrEF and HFpEF	MRA vs placebo	–	33 months	71.4 ± 9.0	48	–
Zeng et al., 2019 [167]	ESRD	Spironolactone vs control	765	2 weeks to 3 years	–	–	7
Zhang et al., 2019 [45]	HT	Diuretic vs placebo	–	1.8 to 5.8 years	–	–	2
Zhang et al., 2022a [168]	CKD and T2D	Finerenone vs placebo	14,847	–	–	–	3
Zhang et al., 2022b [169]	CKD and T2D	Finerenone vs placebo	14,847	–	–	–	3
Zhang et al., 2022c [170]	CKD	Finerenone vs placebo	–	14 to 33 days	–	–	5
Zhao et al., 2019 [171]	CHF	Furosemide vs other loop diuretics	1956	–	–	–	12
Zheng et al., 2018 [172]	HFpEF	MRA vs placebo	4003	6 to 49.5 months	–	–	5
Zheng et al., 2022 [173]	DKD	Finerenone vs placebo	–	3.4 years	58.08 ± 13.08 to 66.75 ± 9.02	–	4
Zhong et al., 2021 [46]	CVD	Diuretic vs placebo	35,543	–	–	–	8
Zhu et al., 2020 [174]	DKD	Diuretic vs placebo	360	1 to 12 months	45 ± 5 to 60	–	5
Zhu et al., 2021 [175]	not specified	Spironolactone or eplerenone vs placebo	1128	3 to 36 months	–	–	9
Zhu et al., 2022 [176]	CKD and/or diabetes	Finerenone vs eplerenone/placebo	15,618	18 days to 53 months	–	–	7
Zonneveld et al., 2018 [177]	HT	Diuretic vs no diuretic	6216	–	–	–	3

ARB angiotensin receptor blocker, CHF chronic heart failure, CKD chronic kidney disease, CVD cardiovascular disease, DKD diabetic kidney disease, EF ejection fraction, HCTZ hydrochlorothiazide, HFmrEF heart failure with mid-range ejection fraction, HFpEF heart failure with preserved ejection fraction, HFrEF heart failure with reduced ejection fraction, HT hypertension, MRA mineralocorticoid antagonist, NYHA New York Heart Association, RCT randomized-controlled trial, SGLT2i sodium-glucose cotransporter-2 inhibitor, T2D type 2 diabetes mellitus

From the 117 SR papers, we extracted 741 effect estimations, 558 (75%) comparing diuretics and no diuretics (majority: placebo; Table 2 and Supplementary Table 4), and 183 (25%) comparing different diuretics (or subclasses; Table 3 and Supplementary Table 5). Of the 741 effect estimations, 192 (26%) were on cardiovascular (CV) effects, 125 (17%) on mortality, 123 (17%) on kidney effects, 89 (12%) on various AEs, 86 (12%) on biochemical effects, 54 (7%) on hospitalization, 35 (5%) on heart ultrasonography measures, 15 (2%) on heart biomarkers, 9 (1%) on QoL, 8 (1%) on functional performance, and 4 (0.5%) on cognitive performance. Of the 319 included placebo-controlled effect estimations, 285 (89%) were on different mineralocorticoid receptor antagonists (MRAs) or subclasses, 19 (6%) on thiazide(like) diuretics, 12 (4%) on diuretics (unspecified), and 3 (0.9%) on loop diuretics. Two SRs exclusively focused on diuretic effects in older (≥ 60 years) individuals [44, 45]; 12 SRs reported on potential age-related differences (either as main objective or based on predefined subgroup/meta-regression analyses) [45–56].

Table 2.

Summary of findings table, meta-analyses comparing diuretic therapy, and placebo

Review first author, year of publication	Outcome category	Specific outcome	Diuretic indication, population	Diuretic intervention	Comparison	Effect (metric, 95% CI)	Meta-analysis (number of pooled effect estimates, metric, 95% CI, I2)	GRADE (reasons for downgrading)	Overall overlap
Frankenstein et al., 2020 [103]	Biochemistry	Hyperkalemia	CHF	Canrenone	Placebo or standard medical care	HR 2.35 (1.13 to 4.87)	Hyperkalemia Patients with HF, MRA vs placebo: n = 4, RR 2.09 (1.87 to 2.33; p < 0.001, I2 = 0%)	High	Moderate
Geng et al., 2023 [51]				MRA	Placebo	RR 2.06 (1.78 to 2.39)
Frankenstein et al., 2020 [103]				Eplerenone	Placebo or standard medical care	HR 2.15 (0.91 to 5.07)
Martin et al., 2021 [128]			HFpEF	MRA	Placebo	RR 2.11 (1.77 to 2.52)
Tsukamoto et al., 2022 [149]	Biochemistry	Hyperkalemia	CKD and T2D	MRA	Placebo	RR 2.06 (1.79 to 2.37)	Hyperkalemia Patients with kidney disease, MRA vs placebo: n = 16, RR 2.31 (2.07 to 2.58; p < 0.001, I2 = 69%)	High	Low
Lunney et al., 2020 [125]			CKD			RR 2.91 (2.03 to 4.17)
Jiang et al., 2022 [115]			CKD and T2D			RR 2.07 (1.86 to 2.30)
Hu et al., 2022 [114]			DKD	Eplerenone		RR 1.55 (0.88 to 2.72)
Zheng et al., 2022 [173]				Finerenone		RR 2.03 (1.83 to 2.26)
Yang et al., 2024 [163]				MRA, non-steroidal		RR 1.01 (0.40 to 3.02)
Hasegawa et al., 2021 [112]			ESKD requiring dialysis	MRA		RR 1.41 (0.72 to 2.78)
Alexandrou et al., 2019 [81]			proteinuric kidney disease			RR 4.44 (1.99 to 9.93)
Wu et al., 2022 [153]			proteinuric kidney disease in diabetes	Eplerenone		RR 1.6 (0.57 to 6)
				Esaxerenone		RR 4.1 (1.8 to 11)
				Spironolactone		RR 8.4 (3.2 to 36)
Ghosal et al., 2022 [106]			T2D	Finerenone		RR 2.22 (1.93 to 2.54)
Chen et al., 2024 [89]			CKD	MRA, non-steroidal		RR 2.05 (1.85 to 2.28)
Jyotsna et al., 2023 [116]			CKD and T2D	Finerenone		RR 2.20 (1.90 to 2.55)
		Hyperkalemia, serious				RR 4.25 (3.11 to 5.83)
Zhang et al., 2022 [169]						RR 4.08 (2.39 to 6.96)
Al-Sadawi et al., 2024 [82]	Biochemistry	Hyperkalemia	HFpEF	MRA	Placebo	OR 3.90 (1.38 to 11.01)	Hyperkalemia Patients with HF, MRA vs placebo: n = 3, OR 1.82 (1.30 to 2.55; p = 0.001, I2 = 62%)	Moderate (inconsistency)	Low
Harrington et al., 2023 [111]		Hyperkalemia, serious	HF			OR 1.46 (1.16 to 1.82)
		Hyperkalemia				OR 1.97 (1.51 to 2.59)
Yang et al., 2019 [160]	Biochemistry	Hyperkalemia	CKD and T2D	Canrenone	Placebo	OR 3.3 (1.2 to 10)	Hyperkalemia Patients with CKD with and without T2D, MRA vs placebo: n = 7, OR 2.26 (1.96 to 2.61; p < 0.001, I2 = 17%)	Moderate (inconsistency)	Moderate
Yasmin et al., 2023 [165]		Hyperkalemia, mild	CKD	Finerenone		OR 1.76 (0.68 to 4.52)
		Hyperkalemia, moderate				OR 2.11 (1.77 to 2.52)
Zhu et al., 2020 [174]		Hyperkalemia, new-onset	DKD	Diuretics (nonspecified)		OR 1.8 (0.69 to 4.5)
				MRA		OR 3.34 (1.1 to 7.13)
Xu et al., 2024 [158]		Hyperkalemia	CKD and T2D	MRA, non-steroidal		OR 2.27 (1.90 to 2.71)
Yang et al., 2019 [160]				Spironolactone		OR 3.6 (2.3 to 7.4)
				Eplerenone		OR 1.8 (1.2 to 3)
Bazoukis et al., 2018 [86]	Cardiovascular	DBP	HF	Eplerenone	Placebo	MD − 1 (− 4.7 to − 2.0)	DBP Patients with HF, MRA vs placebo: n = 4, SMD − 0.20 (− 0.39 to 0.003 mmHg; p = 0.054, I2 = 51%)	Moderate (inconsistency)	Moderate
Fukuta et al., 2019 [105]			HFpEF	MRA		WMD − 3.58 (− 5.65 to − 1.52)
Bazoukis et al., 2018 [86]			HF			MD − 0.34 (− 3.37 to 2.68)
				Spironolactone		MD − 2.99 (− 7.3 to − 2.5)
Bazoukis et al., 2018 [86]	Cardiovascular	SBP	HF	Eplerenone	Placebo	MD − 0.04 (− 4.4 to 4.3)	SBP Patients with HF, MRA vs placebo: n = 4, SMD − 0.33 (− 0.68 to 0.024 mmHg; p < 0.001, I2 = 84%)	Low (inconsistency and imprecision)	Moderate
Fukuta et al., 2019 [105]			HFpEF	MRA		WMD − 8.39 (− 11.38 to − 5.41)
Li et al., 2024 [123]			HF	Spironolactone		WMD − 2.37 (− 3.81 to − 0.94)
Bazoukis et al., 2018 [86]						MD − 4.77 (− 22.3 to − 9.9)
Wright et al., 2018 [152]	Cardiovascular	SBP	HT, first-line therapy	Thiazide, high-dose	Placebo or no anti-hypertensive treatment	MD − 13.66 (− 14.4 to − 12.91)	SBP Patients with HT, thiazide vs placebo: n = 3, SMD − 4.23 (− 6.40 to − 2.10 mmHg; p < 0.001, I2 = 98%)	High	Moderate
				Thiazide, low-dose		MD − 12.56 (− 13.22 to − 11.91)
Martins et al., 2023 [129]			HT, primary	Thiazide	Placebo	MD − 10.37 (− 11.64 to − 9.10)
Jiang et al., 2022 [115]	Cardiovascular	SBP	CKD and T2D	MRA	Placebo	WMD − 4.48 (− 5.95 to − 3.72)	SBP Patients with CKD, MRA vs placebo: n = 6, SMD − 0.51 (− 0.87 to − 0.14 mmHg; p = 0.006, I2 = 90%)	Moderate (imprecision)	Low
Wu et al., 2022 [153]			Proteinuric kidney disease in Diabetes	Spironolactone		MD − 18 (− 42 to 6.4)
Xu et al., 2024 [158]			CKD and T2D	MRA, non-steroidal		WMD − 4.34 (− 5.28 to − 3.41)
Hu et al., 2022 [114]			DKD	Eplerenone		MD − 2.49 (− 4.48 to − 0.5)
Shaman et al., 2020 [55]			ESKD requiring dialysis	MRA		MD − 8.5 (− 16.4 to − 0.5)
Wu et al., 2022 [153]			Proteinuric kidney disease in diabetes	Finerenone		MD − 5.3 (− 35 to 25)
Karakasis et al., 2024 [118]	Cardiovascular	Atrial fibrillation	CVD	Spironolactone	Placebo	RR 0.76 (0.65 to 0.89)	Atrial fibrillation Patients with HF or CVD, MRA vs placebo: n = 6, RR 0.76 (0.71 to 0.82; p < 0.001, I2 = 0%)	High	Moderate
		Atrial fibrillation, recurrent		MRA		RR 0.75 (0.63 to 0.89)
Oraii et al., 2024 [134]			HF			RR 0.74 (0.63 to 0.87)
Karakasis et al., 2024 [118]		Atrial fibrillation, new-onset	CVD			RR 0.79 (0.64 to 0.97)
Oraii et al., 2024 [134]			HF			RR 0.81 (0.64 to 1.03)
		Atrial fibrillation				RR 0.76 (0.67 to 0.87)
Sampaio Rodrigues et al., 2024 [139]	Cardiovascular	Atrial fibrillation, new-onset	not specified	MRA	Placebo or standard medical care	OR 0.77 (0.55 to 1.06)	Atrial fibrillation Patients regardless of diuretic indication, MRA vs placebo: n = 8, OR 0.58 (0.47 to 0.72; p < 0.001, I2 = 48%)	Moderate (inconsistency)	Moderate
Alexandre et al., 2019 [48]				Canrenone	Placebo	OR 0.23 (0.12 to 0.44)
				Eplerenone		OR 0.58 (0.35 to 0.96)
				MRA		OR 0.52 (0.37 to 0.73)
Sampaio Rodrigues et al., 2024 [139]		Atrial fibrillation, new-onset or recurrent		Eplerenone	Placebo or standard medical care	OR 1.08 (0.34 to 3.51)
				MRA		OR 0.68 (0.51 to 0.92)
				Spironolactone		OR 0.63 (0.40 to 0.98)
		Atrial fibrillation, recurrent		MRA		OR 0.50 (0.30 to 0.83)
Clark et al., 2024 [90]	Cardio vascular	Composite	HFpEF	MRA	Placebo	RR 1.16 (0.86 to 1.57)	Composite CV outcome Patients with HF, MRA vs placebo: n = 4, RR 0.90 (0.76 to 1.07; p = 0.231, I2 = 58%)	Moderate (inconsistency)	Moderate
			HFrEF			RR 0.96 (0.81 to 1.14)
Wang et al., 2022 [150]		Composite (men and women)	HF			RR 0.96 (0.77 to 1.19)
		Composite (men)				RR 0.79 (0.69 to 0.91)
		Composite (women)				RR 0.77 (0.52 to 1.16)
Musini et al., 2019 [44]	Cardio vascular	CV events	HT, age > 60 years	Thiazide	Placebo	RR 0.67 (0.61 to 0.74)	CV events Patients with HT, thiazides vs placebo: n = 3, RR 0.85 (0.80 to 0.90; p < 0.001, I2 = 0%)	High	Low
Wright et al., 2018 [152]			HT, first-line therapy	Thiazide, high-dose	Placebo or no anti-hypertensive treatment	RR 0.72 (0.63 to 0.82)
				Thiazide, low-dose		RR 0.7 (0.64 to 0.76)
Wei et al., 2020 [151]			Not specified	Diuretic (nonspecified)	placebo	RR 0.73 (0.62 to 0.85)
Zhang et al., 2022 [169]	Cardiovascular	MACE	CKD	Finerenone	Placebo	RR 0.89 (0.75 to 1.05)	MACE Patients with CKD and/or T2D MRA vs placebo: n = 6, RR 0.88 (0.83 to 0.93; p < 0.001, I2 = 0%)	High	Low
Zhang et al., 2022 [168]			CKD and T2D	Finerenone		RR 0.88 (0.8 to 0.97)
Nguyen et al., 2023 [132]				MRA, non-steroidal		RR 0.92 (0.76 to 1.11)
Li et al., 2022 [122]			T2D	Finerenone		HR 0.87 (0.72 to 1.04)
Gu et al., 2024 [108]			T2D, established ASCVD			HR 0.92 (0.79 to 1.06)
			T2D, without established ASCVD			HR 0.85 (0.77 to 0.95)
Li et al., 2022 [122]	Cardio vascular	MI	T2D	Finerenone	Placebo	HR 0.9 (0.62 to 1.29)	MI Patients with CKD and/or T2D, finerenone vs placebo: n = 3, RR 0.90 (0.81 to 0.99; p = 0.044, I2 = 0%)	High	Low
Abdelazeem et al., 2022 [77]		MI, non-fatal	CKD and T2D			RR 0.91 (0.74 to 1.12)
Jyotsna et al., 2023 [116]						RR 0.89 (0.78 to 1.02)
Zhang et al., 2022 [168]	Cardiovascular	Stroke, non-fatal	CKD and T2D	Finerenone	Placebo	RR 1.00 (0.82 to 1.21)	Stroke Patients with CKD and/or T2D, MRA vs placebo: n = 5, RR 1.00 (0.92 to 1.10; p = 0.944, I2 = 0%)	High	Low
Abdelazeem et al., 2022 [77]		Stroke				RR 0.99 (0.80 to 1.22)
Nguyen et al., 2023 [132]				MRA, non-steroidal		RR 1.00 (0.71 to 1.4)
Li et al., 2022 [122]		Stroke, non-fatal	T2D	Finerenone		HR 1.00 (0.7 to 1.44)
Jyotsna et al., 2023 [116]			CKD and T2D			RR 1.01 (0.89 to 1.14)
Xiang et al., 2022 [155]	Hospitalization	HF-related	HFmrEF and HFpEF	MRA	Placebo	HR 0.83 (0.69 to 0.99)	HF-related hospitalization Patients with HF, MRA vs placebo: n = 6, RR 0.79 (0.75 to 0.83; p < 0.001, I2 = 0%)	High	High
Zafeiropoulos et al., 2024 [166]						HR 0.87 (0.68 to 1.12)
Martin et al., 2021 [128]			HFpEF			HR 0.82 (0.69 to 0.98)
Oraii et al., 2024 [134]			HF			HR 0.78 (0.73 to 0.84)
Xiang et al., 2022 [155]			HFpEF and HFmrEF			HR 0.83 (0.69 to 0.99)
Frankenstein et al., 2020 [103]			CHF	Canrenone	Placebo or standard medical care	HR 0.35 (0.12 to 0.99)
				Eplerenone		HR 0.75 (0.66 to 0.84)
Chen et al., 2024 [89]	Hospitalization	HF-related	CKD	MRA, non-steroidal	Placebo	RR 0.79 (0.67 to 0.92)	HF-related hospitalization Patients with CKD and/or T2D, MRA vs placebo: n = 8, RR 0.78 (0.73 to 0.82; p < 0.001, I2 = 0%)	High	Low
Zhang et al., 2022 [168]			CKD and T2D	Finerenone		RR 0.79 (0.67 to 0.92)
Abdelazeem et al., 2022 [77]						RR 0.79 (0.66 to 0.94)
Li et al., 2022 [122]			T2D			HR 0.78 (0.59 to 1.03)
Gu et al., 2024 [108]						HR 0.78 (0.66 to 0.92)
Jyotsna et al., 2023 [116]			CKD and T2D			RR 0.77 (0.70 to 0.84)
Yang et al., 2023 [164]				MRA	Control	RR 0.79 (0.67 to 0.92)
Tsukamoto et al., 2022 [149]					Placebo	RR 0.71 (0.57 to 0.9)
Nguyen et al., 2023 [132]				MRA, non-steroidal		RR 0.78 (0.66 to 0.92)
Fu et al., 2021 [104]	Kidney	eGFR	CKD	finerenone	Placebo	MD − 0.9 (− 3.84 to 2.04)	eGFR Patients with CKD and/or T2D, MRA vs placebo: n = 11, SMD − 0.40 (− 0.69 to − 0.11 ml/min per 1.73 m2; p = 0.007, I2 = 91%)	High	Low
Chen et al., 2024 [89]				MRA, non-steroidal		WMD − 2.83 (− 3.95 to − 1.72)
Xu et al., 2024 [158]			CKD and T2D			WMD − 2.44 (− 4.06 to − 0.83)
Jiang et al., 2022 [115]						WMD − 2.69 (− 4.47 to − 0.91)
Hu et al., 2022 [114]			DKD	Eplerenone		MD 1.8 (− 0.83 to 4.43)
Hu et al., 2022 [114]					Placebo or active control	MD 1.74 (− 0.87 to 4.35)
Alexandrou et al., 2019 [81]			Proteinuric kidney disease	MRA	Placebo	MD − 2.82 (− 3.98 to − 1.66)
Wu et al., 2022 [153]			Proteinuric kidney disease in diabetes	Esaxerenone		MD − 3.3 (− 11 to 4.5)
				Finerenone		MD − 2.6 (− 9.7 to 4.7)
				Spironolactone		MD − 5.9 (− 12 to 1.3)
Ghosal et al., 2023			T2D	Finerenone		SMD − 0.32 (− 0.37 to − 0.27)
Yasmin et al., 2023 [165]			CKD			MD − 0.02 (− 0.33 to 0.28)
Fu et al., 2021 [104]	Kidney	UACR	CKD	Finerenone	Placebo	MD − 0.3 (− 0.5 to − 0.11)	UACR Patients with CKD and/or T2D, MRA vs placebo: n = 10, SMD − 1.31 (− 1.84 to − 0.77 mg/g; p = 0.007, I2 = 97%)	High	Low
Hu et al., 2022 [114]			DKD	Eplerenone		MD − 48.29 (− 64.45 to − 32.14)
Zheng et al., 2022 [173]				Finerenone		MD − 0.3 (− 0.33 to − 0.27)
Wu et al., 2022 [153]			Proteinuric kidney disease in diabetes	Apararenone		MD − 0.63 (− 0.9 to − 0.35)
				Esaxerenone		MD − 0.54 (− 0.72 to − 0.3)
				Finerenone		MD − 0.21 (− 0.5 to 0.071)
Ghosal et al., 2023			T2D	Finerenone		SMD − 0.49 (− 0.53 to − 0.46)
Chen et al., 2024 [89]			CKD	MRA, non-steroidal		WMD − 0.41 (− 0.49 to − 0.32)
Xu et al., 2024 [158]			CKD and T2D			WMD − 0.39 (− 0.48 to − 0.31)
Jiang et al., 2022 [115]				MRA		WMD − 0.40 (− 0.48 to − 0.32)
Yang et al., 2024 [163]	Kidney	AKI	DKD	MRA, non-steroidal	Placebo	RR 1.08 (0.90 to 1.28)	AKI Patients with T2D and/or (diabetic) kidney disease, MRA vs placebo: n = 3, RR 0.97 (0.89 to 1.05; p = 0.42, I2 = 0%)	Moderate (inconsistency)	Low
Jyotsna et al., 2023 [116]			CKD and T2D	Finerenone		RR 0.94 (0.84 to 1.05)
Ghosal et al., 2022 [106]		AKI or eGFR reduction ≥ 40%	T2D			RR 0.93 (0.78 to 1.11)
Ding et al., 2023 [93]	Kidney	Composite	CKD	Finerenone	Placebo	HR 0.84 (0.77 to 0.92)	Composite kidney outcome Patients with CKD and/or T2D, MRA vs placebo: n = 8, RR 0.85 (0.82 to 0.88; p < 0.001, I2 = 0%)	High	Low
Tsukamoto et al., 2022 [149]			CKD and T2D	MRA		RR 0.86 (0.77 to 0.95)
Li et al., 2022 [122]			T2D	Finerenone		HR 0.84 (0.62 to 1.17)
Ghosal et al., 2023						HR 0.84 (0.77 to 0.92)
Gu et al., 2024 [108]						HR 0.84 (0.77 to 0.92)
Nguyen et al., 2023 [132]			CKD and T2D	MRA, non-steroidal		RR 0.84 (0.77 to 0.92)
Zhang et al., 2022 [169]			CKD	finerenone		RR 0.86 (0.79 to 0.93)
Chen et al., 2024 [89]				MRA, non-steroidal		RR 0.86 (0.79 to 0.93)
Chen et al., 2024 [89]	Kidney	eGFR, > 40% reduction	CKD	MRA, non-steroidal	Placebo	RR 0.85 (0.78 to 0.92)	> 40% eGFR reduction Patients with T2D and/or (diabetic) kidney disease, MRA vs placebo: n = 4, RR 0.85 (0.82 to 0.88; p < 0.001, I2 = 0%)	High	Low
Bao et al., 2022 [84]			CKD and T2D	Finerenone		RR 0.85 (0.78 to 0.93)
Zheng et al., 2022 [173]			DKD			RR 0.85 (0.78 to 0.93)
Jyotsna et al., 2023 [116]			CKD and T2D			RR 0.85 (0.80 to 0.90)
Bao et al., 2022 [84]	Kidney	ESKD	CKD and T2D	Finerenone	Placebo	RR 0.8 (0.65 to 0.99)	ESKD Patients with chronic (diabetic) kidney disease, finerenone vs placebo: n = 3, RR 0.80 (0.72 to 0.89; p < 0.001, I2 = 0%)	High	Low
Dutta et al., 2022 [95]			DKD			RR 0.79 (0.62 to 1.01)
Jyotsna et al., 2023 [116]			CKD and T2D			RR 0.80 (0.69 to 0.93)
Xu et al., 2024 [158]	Kidney	eGFR, > 30% reduction	CKD and T2D	MRA, non-steroidal	Placebo	OR 1.34 (0.83 to 2.18)	eGFR reduction of kidney failure Patients with CKD with and without T2D, MRA vs placebo: n = 9, OR 0.84 (0.74 to 0.96; p = 0.008, I2 = 68%)	Moderate (inconsistency)	Low
Yasmin et al., 2023 [165]		eGFR, > 40% reduction	CKD	Finerenone		OR 0.82 (0.74 to 0.91)
Dutta et al., 2022 [95]			DKD			OR 0.83 (0.75 to 0.92)
Yasmin et al., 2023 [165]		eGFR, > 57% reduction	CKD			OR 0.70 (0.59 to 0.82)
Yang et al., 2022 [161]			CKD and T2D	MRA, nonselective		OR 1.6 (0.54 to 4.92)
Dutta et al., 2022 [95]			DKD	Finerenone		OR 0.7 (0.6 to 0.82)
Yang et al., 2019 [160]		eGFR, reduction	CKD and T2D	Spironolactone		OR 3.3 (1.5 to 9.4)
Yang et al., 2022 [161]		Kidney failure		MRA, nonselective		OR 1.22 (0.05 to 32.9)
				MRA, non-steroidal		OR 0.85 (0.71 to 1)
Xiang et al., 2022 [155]	Mortality	All cause	HFmrEF and HFpEF	MRA	Placebo	OR 0.92 (0.77 to 1.08)	All-cause mortality Patients with HF, MRA vs placebo: n = 4, OR 0.88 (0.79 to 0.98; p = 0.021, I2 = 9%)	Moderate (inconsistency)	High
Al-Sadawi et al., 2024 [82]			HFpEF			OR 0.63 (0.43 to 0.92)
Sreenivasan et al., 2022 [144]						OR 0.9 (0.75 to 1.08)
Sreenivasan et al., 2024 [143]						OR 0.90 (0.75 to 1.08)
Zhang et al., 2022 [169]	Mortality	All cause	CKD	Finerenone	Placebo	RR 0.9 (0.8 to 1)	All-cause mortality Patients with CKD (with or without T2D), MRA vs placebo: n = 8, RR 0.89 (0.82 to 0.96; p = 0.004, I2 = 49%)	Moderate (inconsistency)	Low
Lunney et al., 2020 [125]				MRA		RR 0.61 (0.06 to 6.59)
Bao et al., 2022 [84]			CKD and T2D	Finerenone		RR 0.9 (0.8 to 1.00)
Tsukamoto et al., 2022 [149]				MRA		RR 0.9 (0.77 to 1.05)
Nguyen et al., 2023 [132]				MRA, non-steroidal		RR 0.89 (0.8 to 1)
Dutta et al., 2022 [95]			DKD	Finerenone		RR 0.89 (0.79 to 1)
Yang et al., 2024 [163]				MRA, non-steroidal		RR 1.12 (0.85 to 1.45)
Hasegawa et al., 2021 [112]			ESKD requiring dialysis	MRA		RR 0.45 (0.3 to 0.67)
Hansen et al., 2020 [110]	Mortality	All cause	HF	MRA	Placebo	HR 0.77 (0.68 to 0.88)	All-cause mortality Patients with HF, MRA vs placebo: n = 9, RR 0.86 (0.81 to 0.90; p < 0.001, I2 = 0%)	Moderate (inconsistency)	Moderate
Zafeiropoulos et al., 2024 [166]			HFmrEF and HFpEF			HR 0.83 (0.68 to 1.02)
Bonsu et al., 2018 [87]			HFpEF	Eplerenone		RR 1.01 (0.26 to 3.67)
Martin et al., 2021 [128]				MRA		RR 0.91 (0.78 to 1.06)
Geng et al., 2023 [51]			CHF			RR 0.82 (0.74 to 0.90)
Täger et al., 2019 [145]			HFrEF	Spironolactone		RR 1.00 (0.01 to 73.7)
Bonsu et al., 2018 [87]			HFpEF			RR 0.92 (0.79 to 1.08)
Faisal et al., 2022 [98]						RR 0.92 (0.79 to 1.08)
Bonsu et al., 2018 [87]						RR 0.97 (0.69 to 1.35)
Xiang et al., 2022 [155]			HFpEF and HFmrEF	MRA		HR 0.92 (0.77 to 1.08)
Zhang et al., 2022 [169]	Mortality	CV	CKD	Finerenone	Placebo	RR 0.88 (0.76 to 1.02)	CV mortality Patients with CKD (with or without T2D), finerenone vs placebo: n = 5, RR 0.86 (0.82 to 0.91; p < 0.001, I2 = 0%)	High	Low
Li et al., 2022 [122]			T2D			HR 0.88 (0.69 to 1.13)
Jyotsna et al., 2023 [116]			CKD and T2D			RR 0.86 (0.80 to 0.93)
Yasmin et al., 2023 [165]			CKD			HR 0.84 (0.74 to 0.95)
Abdelazeem et al., 2022 [77]			CKD and T2D			RR 0.88 (0.76 to 1.02)
Geng et al., 2023 [51]	Mortality	CV	CHF	MRA	Placebo	RR 0.80 (0.71 to 0.90)	CV mortality Patients with HF, MRA vs placebo: n = 7, RR 0.83 (0.79 to 0.88; p < 0.001, I2 = 0%)	Moderate (inconsistency)	Moderate
Oraii et al., 2024 [134]			HF			HR 0.82 (0.76 to 0.88)
Täger et al., 2019 [145]			HFrEF	Spironolactone		RR 1.06 (0.01 to 90.95)
Zafeiropoulos et al., 2024 [166]			HFmrEF and HFpEF	MRA		HR 0.74 (0.57 to 0.97)
Xiang et al., [155]						HR 0.91 (0.73 to 1.12)
Martin et al., 2021 [128]			HFpEF			RR 0.9 (0.74 to 1.11)
Faisal et al., [98]				Spironolactone		RR 0.91 (0.74 to 1.11)
Xiang et al., [155]			HFpEF and HFmrEF	MRA		HR 0.91 (0.73 to 1.12)
Ghosal et al., 2022 [106]	Various AEs	Any AE	T2D	Finerenone	Placebo	RR 1.00 (0.98 to 1.01)	Adverse events Patients with CKD (with or without T2D), MRA vs placebo: n = 6, RR 1.00 (0.99 to 1.01; p = 1.00, I2 = 0%)	High	Low
Jyotsna et al., 2023 [116]			CKD and T2D			RR 1.00 (0.99 to 1.01)
Fu et al., 2021 [104]			CKD			RR 1.00 (0.98 to 1.02)
Zheng et al., 2022 [173]			DKD			RR 1.00 (0.98 to 1.01)
Chen et al., 2024 [89]			CKD	MRA, non-steroidal		RR 1.00 (0.99, 1.01)
Bao et al., 2022 [84]			CKD and T2D	Finerenone		RR 1.00 (0.98 to 1.01)
Wright et al., 2018 [152]	VARIOUS AES	DISCONTINUATION	HT, first-line therapy	Thiazide	Placebo	RR 3.22 (2.9 to 3.57)	Discontinuation Patients with HT, thiazides vs placebo: n = 3, RR 3.25 (2.36 to 4.46; p < 0.001, I2 = 94%)	High	Low
Liu et al., 2022 [124]				Thiazide, high-dose	Placebo or no anti-hypertensive treatment	RR 4.48 (3.83 to 5.24)
Wright et al., 2018 [152]				Thiazide, low-dose		RR 2.38 (2.06 to 2.75)
Geng et al., 2023 [51]	Various AEs	Gynecomastia	CHF	Eplerenone	Placebo	RR 0.72 (0.32 to 1.61)	Gynecomastia Patients with HF, eplerenone or spironolactone vs placebo: n = 4, RR 2.50 (0.67 to 9.29; p = 0.17, I2 = 92%)	Very low (inconsistency and imprecision)	Low
Frankenstein et al., 2020 [103]					Placebo or standard medical care	HR 0.77 (0.31 to 1.88)
Geng et al., 2023 [51]				Spironolactone	placebo	RR 7.48 (4.42 to 12.68)
Frankenstein et al., 2020 [103]					Placebo or standard medical care	HR 8.44 (3.9 to 18.28)
Yang et al., 2019 [160]	Various AEs	Any AE	CKD and T2D	Canrenone	Placebo	OR 3.7 (1.1 to 13)	Adverse events Patients with CKD and T2D, MRA vs placebo: n = 3, OR 1.55 (0.81 to 2.96; p = 0.19, I2 = 75%)	Low (imprecision and inconsistency)	Low
Xu et al., 2024 [158]				MRA, non-steroidal		OR 1.00 (0.92 to 1.10)
Yang et al., 2019 [160]				Spironolactone		OR 1.8 (1.00 to 3.6)

AE adverse event, AKI acute kidney injury, BP blood pressure, CI confidence interval, CKD chronic kidney disease, CV cardiovascular, CVD cardiovascular disease, DBP diastolic blood pressure, DKD diabetic kidney disease, ESKD end-stage kidney disease, GRADE Grading of Recommendations, Assessment, Development and Evaluation, HF heart failure, HFmrEF heart failure with mid-range ejection fraction, HFpEF heart failure with preserved ejection fraction, HFrEF heart failure with reduced ejection fraction, HR hazard ratio, HT hypertension, MACE major adverse cardiovascular event, MD mean difference, MI myocardial infarction, MRA mineralocorticoid antagonist, OR odds ratio, RR risk ratio, SBP systolic blood pressure, SMD standardized mean difference, T2D type 2 diabetes mellitus, UACR urinary albumin-to-creatinine ratio, WMD weighted mean difference

Table 3.

Summary of findings table, meta-analyses comparing diuretics with no diuretics

Review first author, year of publication	Outcome category	Specific outcome	Diuretic indication, population	Diuretic A	Diuretic B	Effect (metric, 95% CI)	Meta-analysis pooled data (number of effect estimates, metric, 95% CI, I2)	GRADE (reasons for downgrading)	Overall overlap
Singh et al., 2023 [142]	Hospitalization	HF-related	HF	Torasemide	Furosemide	RR 0.61 (0.45 to 0.83)	RR 0.53 (0.41 to 0.69; p < 0.001; I2 = 46%)	High	High
Täger et al., 2019 [145]			HFrEF			RR 0.4 (0.28 to 0.58)
Teixeira et al., 2024 [146]			HF			RR 0.6 (0.43 to 0.83)
Shah et al., 2018 [60]	Hospitalization	HF-related	HF	Torasemide	Furosemide	OR 3.03 (2.00 to 4.55)	OR 1.18 (0.68 to 2.04; p = 0.552; I2 = 91%)	Low (imprecision and inconsistency)	Moderate
Abraham et al., 2020 [78]						OR 0.72 (0.51 to 1.03)
Kido et al., 2019 [119]						OR 0.79 (0.57 to 1.09)
Miles et al., 2019 [130]						OR 2.04 (1.16 to 3.6)
Siddiqi et al., 2023 [141]						OR 0.73 (0.54 to 0.99)
Zhao et al., 2020	Mortality	All cause	HF	Torasemide	Furosemide	OR 0.91 (0.56 to 1.47)	OR 0.96 (0.82 to 1.13; p = 0.651; I2 = 0%)	Moderate (imprecision)	Moderate
Sherif et al., 2019 [70]						OR 0.88 (0.57 to 1.38)
Abraham et al., 2020 [78]						OR 0.9 (0.58 to 1.41)
Kido et al., 2019 [119]						OR 1.00 (0.58 to 1.72)
Miles et al., 2019 [130]						OR 1.12 (0.7 to 1.8)
Siddiqi et al., 2023 [141]						OR 0.98 (0.75 to 1.29)

CI confidence interval, GRADE Grading of Recommendations, Assessment, Development and Evaluation, HF heart failure, HFrEF heart failure with reduced ejection fraction, OR odds ratio, RR risk ratio

We performed 33 MAs [30 comparing diuretics (or diuretic (sub) classes] to placebo (Table 2) and 3 comparing torasemide to furosemide (Table 3)). Overall overlap of original RCTs in our MAs was low in 64, moderate in 30, and high in 6% (Tables 2, 3). Ten results for MAs comparing diuretics with placebo, and for one MA comparing torasemide and furosemide had the combination of large statistical significance (p < 0.001), high certainty evidence (GRADE classification “high”) and low heterogeneity (I² ≤ 50%). For two MAs, we performed sensitivity analyses based on overall RoB classification.

Mortality

Our meta-analyses (Table 2) show that in persons with HF, MRAs compared to placebo reduced the risk of all-cause mortality [risk ratio (RR) 0.86; 0.81 to 0.90 and odds ratio (OR) 0.88; 0.79 to 0.98, respectively; both moderate certainty]. Similarly, CV mortality risk was lower with MRAs compared to placebo (RR 0.83; 0.79 to 0.88; moderate certainty). The odds of all-cause mortality were comparable between torasemide and furosemide (OR 0.96; 0.82 to 1.13; moderate certainty). In individuals with chronic kidney disease (CKD) and/or type 2 diabetes (T2D), all-cause mortality risk was lower with MRA compared to placebo (RR 0.89; 0.82 to 0.96; moderate certainty); CV mortality risk was lower with finerenone compared to placebo (RR 0.86; 0.82 to 0.91; high certainty).

Our non-pooled data (Supplementary Tables 4 and 5) suggest that diuretics may reduce CV mortality. One review [47] reported CV mortality in older (≥ 65 years old), but not in younger adults (< 65 years old). A Cochrane review showed that thiazides as first-line primary prevention in healthy adults ≥ 60 years old with moderate-to-severe HT may reduce all-cause, CV, cerebrovascular and coronary mortality [44]. A subgroup analysis of one SR showed that in adults with HT, MRA use was associated with a reduced risk of all-cause mortality in older (≥ 65 years), but not in younger individuals [57]. Diuretics or MRAs may lack beneficial effects on all-cause mortality in individuals with CVD. Diuretics or MRAs may have beneficial effect on vascular or HF-related mortality in individuals with a history of CVD. MRAs may reduce CV mortality in adults with end-stage kidney disease (ESKD) or a history of myocardial infarction (MI), but not with HFpEF [HF with preserved ejection fraction (EF)] or CKD and/or T2D. In individuals with HT, MRAs may have beneficial effects on all-cause mortality. MRAs may postpone death in individuals with HF. Spironolactone may have a favorable effect on all-cause mortality in individuals with ESKD requiring dialysis, but in persons with HFpEF, effects were inconsistent. Compared to eplerenone and spironolactone, finerenone may have a beneficial effect on CV mortality.

Cardiovascular effects

Our meta-analyses (Table 2) show that MRAs (irrespective of indication/population) compared to placebo reduce the odds of new-onset or recurrent atrial fibrillation (OR 0.58; 0.47 to 0.72; moderate certainty). In adults with HF, the risk of developing a composite CV end-point was comparable between MRAs and placebo (moderate certainty). In individuals with HF, the risk of new-onset or recurrent atrial fibrillation was lower with MRAs compared to placebo (RR 0.76; 0.71 to 0.82; high certainty). The effects on systolic and diastolic blood pressure (SBP and DBP, respectively) were comparable for MRAs and placebo (low and moderate certainty, respectively). In adults with HT, CV event risk and SBP were lower with thiazides compared to placebo (RR 0.85; 0.80 to 0.90; high certainty, and SMD − 4.23; − 6.40 to − 2.10; high certainty, respectively). The difference in SBP was clinically relevant [58, 59]. In persons with CKD with and without T2D, major adverse CV event (MACE) risk and SBP were lower with MRAs compared to placebo (RR 0.88; 0.83 to 0.93; high certainty, and SMD − 0.51; − 0.87 to − 0.14; moderate certainty, respectively), but stroke risk was comparable between MRAs and placebo (high certainty). The difference in SBP was not clinically relevant [58, 59]. The risk of MI was lower with finerenone compared to placebo (RR 0.90; 0.81 to 0.99; high certainty).

Our non-pooled data (Supplementary Tables 4 and 5) suggest that diuretics may reduce the risk of composite CV outcomes, CV events, CVD, and stroke. Results for potential age-related differences in stroke risk reduction showed inconsistent results: in one review, the risk was reduced in older, but not in younger (cut-off 65 years) adults [47], whereas in another review, the risk was reduced both in individuals older and younger than 60 years [46]. In persons with ESKD requiring dialysis, there were no age-related differences in BP effects of MRAs [55]. MRAs may have favorable effects on BP regulation, and may reduce the risk of composite CV outcomes, and CVD and CV events. MRAs may have favorable effects on atrial fibrillation, irrespective of age [48]). MRAs may reduce edema in individuals with CKD, but not with HT. Finerenone may reduce the risk of new-onset HT. Thiazide(like) diuretics may have favorable effects on BP regulation in persons with CKD and with uncontrolled HT, and may reduce the risk of MACE, MI, and revascularization and stroke risk. Meta-regression analysis in an SR of RCTs in individuals with HF showed no age-related differences in effects of MRAs on MACE [51]. An individual participant-level data MA showed that there may not be age-dependent effects of thiazides for the prevention of CVD [49]. A subgroup analysis of an SR among individuals with HT living in Sub-Saharan Africa showed that the BP effects of diuretics were independent of age [54]. A Cochrane review showed that thiazides as first-line primary prevention in healthy adults ≥ 60 years old with moderate-to-severe HT may reduce all-cause, CV, and cerebrovascular and coronary morbidity [44]. Data regarding the effects of finerenone on HF-related outcomes in persons with CKD were inconsistent. Low-dose but not high-dose thiazides may reduce the risk of coronary events in persons with HT. Compared to furosemide, torasemide may have favorable effects on HF.

Hospitalization

Our meta-analyses (Tables 2, 3) show that in persons with HF and in persons with CKD and/or T2D, HF-related hospitalization (HFH) risk was lower with MRAs compared to placebo (RR 0.79; 0.75 to 0.83; high certainty, and RR 0.78; 0.73 to 0.82; high certainty, respectively). The risk, but not the odds of HFH were lower with torasemide compared to furosemide (RR 0.53; 0.41 to 0.69; high certainty, and OR 1.18; 0.68 to 2.04; low certainty). Excluding the data from a review with high RoB [60] did not impact the results (data not shown).

Our non-pooled data (Supplementary Tables 4 and 5) suggest that MRAs may have a beneficial effect on HFH. Results for all-cause and for CVD-related hospitalization were inconsistent. In individuals with T2D or CKD, but not with HFpEF, MRAs may increase the risk of hyperkalemia-related hospitalization. Compared to furosemide, torasemide may have beneficial effects on all-cause and CV-related hospitalization, and on hospital admission duration.

Kidney effects

Our meta-analyses (Table 2) show that in individuals with CKD and/or T2D and compared to placebo, MRAs reduce the risk of developing a composite kidney outcome (RR 0.85; 0.82 to 0.88; high certainty), reduce the odds of estimated glomerular filtration rate (eGFR) worsening or kidney failure (OR 0.84; 0.74 to 0.96; moderate certainty), reduce the risk of a > 40% eGFR worsening (RR 0.85; 0.82 to 0.88; high certainty), reduce urinary albumin-to-creatinine ratio (UACR) (SMD − 1.31; − 1.84 to − 0.77; high certainty), and reduce eGFR (SMD − 0.40; − 0.69 to − 0.11; high certainty). The reductions in eGFR and UACR are not clinically relevant [61]. Acute kidney injury (AKI) risk was comparable with MRAs and placebo (moderate certainty). In individuals with chronic (diabetic) kidney disease, the risk of developing ESKD was lower with finerenone compared to placebo (RR 0.80; 0.72 to 0.89; high certainty).

Our non-pooled data (Supplementary Table 4) suggest that in persons with kidney disease, MRAs may lower UACR, 24 h albuminuria, and proteinuria. MRAs may have a beneficial effect on kidney function (eGFR, CKD progression, and composite kidney outcome) in individuals with CKD and/or T2D, but effects on ESKD and kidney failure are inconsistent. Thiazides may have a beneficial effect on kidney stone-related morbidity in individuals with idiopathic hypercalciuria or a history of kidney stones. Thiazide(like) diuretics may reduce GFR in persons with advanced CKD. In patients with diabetic kidney disease (DKD), eplerenone may be more effective in reducing microalbuminuria.

Effects on biochemistry

Our meta-analyses (Table 2) show that in persons with HF, MRAs compared to placebo increase the risk of hyperkalemia (RR 2.09; 1.87 to 2.33; high certainty, and OR 1.82; 1.30 to 2.55; moderate certainty). Sensitivity analysis showed that 2/3 of the estimated ORs were from a review with high RoB. In persons with CKD (with and without T2D), MRAs compared to placebo increase the risk of hyperkalemia (RR 2.31; 2.07 to 2.58; high certainty, and OR 2.26; 1.96 to 2.61; moderate certainty, respectively).

Our non-pooled data (Supplementary Table 4) suggest that thiazides may lower serum potassium in individuals with HT. MRAs may increase serum potassium in adults with CV disease [CKD, T2D, DKD, and HFmrEF (HF with mid-range EF) or HFpEF], but not with ESKD, and lower the risk of hypokalemia in persons with HF. The risk of hyperkalemia may be increased by MRAs irrespective of indication. Spironolactone may increase serum calcium. Nonsteroidal MRAs may increase hyponatremia risk in persons with CKD and/or T2D. Thiazides may increase (fasting) serum glucose. New-onset diabetes risk was comparable with diuretics and placebo, irrespective of age (younger vs older than 65 years) [45].

Effects on blood heart biomarkers

Our non-pooled data (Supplementary Tables 4 and 5) suggest that in patients with HFmrEF or HFpEF, spironolactone may reduce serum brain natriuretic peptide (BNP) and fibrosis marker PICP (procollagen type I C-terminal propeptide), and in patients with HFpEF, may reduce fibrosis marker PIIINR (amino-terminal peptide of procollagen type-III). Compared to furosemide, torasemide may have beneficial effects on BNP and BNP/NT-proBNP (N-terminal pro b-type natriuretic peptide) ratio.

Effects on heart ultrasonography variables

Our non-pooled data (Supplementary Tables 4 and 5) suggest that MRAs may have favorable effects on augmentation index, flow-mediated dilation (FMD), and pulse wave velocity (PWV), probably irrespective of age [53]. In persons with ESKD requiring dialysis, MRAs may reduce left ventricle mass (LVM). Spironolactone may reduce left-ventricular mass index (LVMI), but effects on left-ventricular ejection fraction (LVEF) are inconsistent. MRAs may have a favorable effect on LVEF in persons after MI, and a favorable effect on E', E/e', left atrial volume index (LAVI), and left-ventricular end-diastolic diameter (LVEDD) in individuals with HFpEF. Spironolactone may have a favorable effect on LVEDD in individuals with HFmrEF or HFpEF, and on E/e’ in persons with HFpEF, irrespective of age (70 year cut-off) [52]. Effects on left-ventricular end-diastolic volume (LVEDV) may be favorable with torasemide compared to furosemide.

Effects on functional performance

Our non-pooled data (Supplementary Table 4) suggest that the 6 min walking distance (6MWD) may be increased by spironolactone in individuals with HFmrEF or HFpEF; in persons with HFpEF, 6MWD may be comparable or reduced by spironolactone/MRAs.

Effects on cognitive performance

Our non-pooled data (Supplementary Table 4) suggest that diuretics may decrease the risk of a new dementia diagnosis in studies with a follow-up duration of ≥ 1, but not ≥ 5 years.

Adverse events and discontinuation

Our meta-analyses (Table 2) show that in persons with CKD with or without T2D, the risk and odds of AEs are comparable with MRAs and placebo (high and low certainty, respectively). In individuals with HF, the risk of developing gynecomastia was comparable for eplerenone or spironolactone compared to placebo (very low certainty). In adults with HT, the risk of discontinuation was higher with thiazides compared to placebo (RR 3.25; 2.36 to 4.46; high certainty).

Our non-pooled data (Supplementary Tables 4 and 5) suggest that diuretics may increase the risk of AEs. In older (≥ 65 years), but not in younger adults, the risk of AEs may be increased [47]. In persons with uncontrolled HT, addition of hydrochlorothiazide to ARB (angiotensin receptor blocker) therapy may increase the risk of drug-related AEs; the risk of any AEs was increased with high-dose, but not with low-dose hydrochlorothiazide. MRAs may increase the risk of hyperkalemia-related AEs in persons with HT. Finerenone may increase the risk of drug-related AEs, but reduce the risk of HF-related AEs and serious AEs in persons with CKD and/or T2D. Steroidal MRAs (especially spironolactone), but not non-steroidal MRAs (finerenone) may increase the risk of gynecomastia and breast pain. In patients with calcium oxalate stones, AE risk may be increased by thiazides. MRAs or spironolactone may increase the risk of discontinuation in individuals with CKD and/or T2D, ESKD or HFrEF. Nonsteroidal MRAs may reduce the risk of severe hypoglycemia in individuals with T2D. Finerenone may increase the risk of urinary tract malignancy, but not other neoplasms in adults with CKD and/or T2D. MRAs may increase the risk of hypotension in individuals with CKD, but not with HF, HT, or ESKD requiring dialysis. Thiazide(like) diuretics may have a neutral effect on fracture risk, irrespective of age [50]. Finerenone may have a more favorable AE profile than spironolactone (lower risk of AEs and hyperkalemia). The risk of gynecomastia and of hyperkalemia may be larger with spironolactone compared to eplerenone.

Quality of life

Our non-pooled data (Supplementary Table 4) suggest that MRAs or spironolactone may have a neutral effect on QoL.

Discussion

The findings of this umbrella review suggest both potential benefits and harms of chronic diuretic use in adults. We found beneficial effects for finerenone, which likely reduces the risk of CV mortality and ESKD in individuals with CKD and/or T2D. Thiazides probably reduce CV events in individuals with HT, and MRAs likely reduce HF-related hospitalization in individuals with HF and atrial fibrillation risk in individuals with CVD. Additionally, MRAs may reduce HFH, MACE, > 40% eGFR decrease, and composite kidney outcomes in individuals with CKD and/or T2D. Regarding potential harms of diuretic use, we found that MRAs increase the risk of hyperkalemia in individuals with HF.

Mean age in the SRs we analyzed was only 62 years, which is relatively young compared to most adults with CVD (> 70% of adults has developed CVD by age 70 [3], and 80% of HF patients are older than 65) [62, 63]. This younger mean age aligns with typical demographics in RCTs in the CV field, where participant ages generally range from 61 to 64 years. Yet, this is younger than the average age of patients in real-world clinical settings as indicated by epidemiological studies (e.g., the mean age for patients with acute coronary syndrome is generally between 66 and 70 years [64]). The potential clinical relevance of these age-related differences is underscored by our findings, which suggest that diuretics and MRAs may have differential effects across age groups. For instance, we observed potential benefits (e.g., reduced CV mortality) in older, but not in younger adults. Furthermore, in adults with HT, MRA use was associated with a reduced risk of all-cause mortality in older, but not younger individuals. This may be attributed to age-related decreases in plasma aldosterone levels [56, 65]. Additionally, we found that the risk of AEs associated with diuretic use was higher in older adults (≥ 65 years), but not in younger adults. These findings suggest that age-related differences in benefits and risks of diuretics must be considered. Although the population we analyzed is relatively young compared to the real-world patient population, our results remain highly relevant for older patients. It is important to note that individuals with CVD, as studied in our umbrella review, are biologically older than their peers of the same chronological age without CVD [66, 67]. Not only is the biological age of individuals with CVD (as studied in our umbrella review) typically higher, but their clinical characteristics also align with those of older individuals: patients with CVD generally experience a high prevalence of multiple geriatric conditions such as multimorbidity (≥ 2 chronic conditions), polypharmacy (≥ 5 medications), cognitive impairment, vision and hearing impairments, urinary incontinence, functional decline, frailty, and sarcopenia [68, 69]. Thus, despite the younger chronological age of our study population, our findings are applicable to older, more biologically frail patients with CVD. Another notable mismatch is the paucity of data on loop diuretics despite their widespread use in HF [> 80% HF patients are on loop diuretics [70], yet only 3 of 319 (0.9%) of the placebo-controlled effect estimations we included focused on loop diuretics]. Furthermore, the vast majority (97%) of the diuretic efficacy outcome estimates in our analyses were on objective, quantifiable outcomes such as mortality, hospitalization, or MACE. Although typical priorities for older adults include physical and cognitive function, symptom control, reduced therapy burden, HR-QoL, maintenance of independence, and overall well-being [6], less than 3% of the efficacy outcomes we included addressed these key factors that are highly relevant to older individuals. These discrepancies reflect broader issues in CV research, where older adults, particularly those with multimorbidity, are often underrepresented in clinical trials [71, 72]. Unfortunately, we were not able to perform subgroup or meta-regression analyses based on age and multimorbidity/frailty status.

A key strength of our review lies in its broad scope regarding settings, health outcomes, and diuretic therapy we deliberately adopted to fit the needs encountered in clinical practice. Additional strengths of our review include its rigorous methodology, including the use of the GRADE framework, RoB assessment, and performing sensitivity and overlap analyses. Furthermore, we evaluated the effects of diuretics on a range of biomarkers, such as NT-proBNP. Although biomarkers cannot replace clinical outcomes, some may serve as useful indicators of safety [46, 47]. Also, alongside statistical significance, we consider clinical relevance by incorporating the concept of minimal clinically important difference into our conclusions [48]. Our results may serve as a source of reference for clinicians, and will be used by our group to inform an evidence-based clinical practice guideline on diuretic deprescribing, aiming to support healthcare providers in discussions about the long-term use of diuretics and in making individualized treatment decisions in routine clinical practice.

However, our methodology does have limitations. First, the broad PICOS approach may reduce the generalizability of our findings. By focusing solely on published MA data, we excluded evidence from individual cohort studies or RCTs that have not been previously aggregated in MAs. Additionally, while we assessed the quality of the MAs, we did not evaluate the quality of all the primary studies included in these MAs, as this was beyond the scope of this review and would have been impractical given the large volume of studies involved. Finally, unlike traditional meta-analyses, umbrella reviews do not permit subgroup or sensitivity analyses due to several methodological and reporting limitations [73, 74]. As a result, we were unable to quantify potential differential effects of chronic diuretic use in specific subgroups such as the oldest old compared to younger counterparts or those residing in long-term care facilities (e.g., nursing homes) versus non-institutionalized individuals.

Further research is needed to establish causal relationships between chronic diuretic use and health outcomes, especially in older adults. Future studies should focus on real-world populations with multimorbidity and frailty [75], considering patient-centered outcomes like QoL, physical function, and independence [1, 63, 71]. Innovative methodologies on trials with large, diverse populations could help refine treatment decisions and quantify benefits and harms, while considering patient preferences [76]. This additional evidence will be critical for guiding clinical decisions on the initiation, continuation, or cessation of diuretic therapy, weighing the benefits, harms, and burdens in light of individual health priorities.

Conclusions

This umbrella review provides a comprehensive synthesis of the associations between diuretics and various health outcomes. Our findings suggest that certain diuretics, or diuretic (sub)classes, confer significant benefits for key clinical outcomes, including CV mortality, HFH, and CKD in specific populations with CVD. However, our results also underscore the potential risks associated with chronic diuretic use, such as an elevated risk of hyperkalemia in individuals with HF.

Age-related differences in AE risks were present, with older adults (≥ 65 years) exhibiting higher risks, while younger populations did not demonstrate similar concerns. Additionally, our analysis highlights a critical gap between the extensive body of existing data and the clinical relevance of this evidence for older individuals with CVD as encountered in clinical practice. This gap reflects broader challenges in CV research, where older adults—especially those with multimorbidity—are frequently underrepresented in clinical trials. Although older adults typically prioritize outcomes related to physical and cognitive function, symptom management, reduced treatment burden, health-related QoL, independence, and overall well-being, these outcomes represented less than 3% of all efficacy outcomes we reviewed.

To draw more definitive conclusions regarding the effects of chronic diuretic use and to refine clinical diuretic prescribing decision-making, further research is crucial, particularly for older adults with multimorbidity.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Eveline van Poelgeest was involved in the conception and design of the study, acquisition of data, data analysis and interpretation, GRADEing the evidence and drafting and revising the article. Luca Paoletti, Min Ji Kwak, Tuğba Erdogan, Serdar Özkök, Gulistan Bahat, Birkan Ilhan, Alessia Beccacece, Fatma Özge Kayhan Kocak, Karolina Piotrowicz, George Soulis, and Annette Eidam were involved in acquisition of data, quality assessment of included papers, and revising the article. Eva Topinková, Antonio Cherubini, Jerzy Gąsowski, Parag Goyal, Louis Handoko, and Wade Thompson were involved in revising the article. Joost Daams developed the literature search. Konstantinos Prokopidis, Giuseppe Dario Testa, and Nicola Veronese were involved in data analysis, assessment of the certainty in the body of evidence, and revising the article. Nathalie van der Velde was involved in the conception and design of the study, data interpretation, and revising the article. All authors approved the final version of the manuscript.

Funding

This project was not funded.

Data availability

Data are available upon request.

Declarations

Conflict of interest

Min Ji Kwak received a research grant from the US Deprescribing Research Network, receives consult fees from Novo Nordisk, and owns stocks in Eli Lilly and Novo Nordisk. Louis Handoko is supported by a research grant of the Dutch Heart Foundation (NHS; 2020T058), the Amsterdam Cardiovascular Sciences (ACS) Institute, The Netherlands Organization for Scientific Research (NWO), and funding for an investigator-initiated study of Vifor Pharma. He received an educational grant and/or speaker/consultancy fees from Novartis, Boehringer Ingelheim, Daiichi Sankyo, Vifor Pharma, AstraZeneca, Bayer, MSD, and Abbott; all not related to this work. Parag Goyal is supported by National Institute on Aging grants K76AG064428 and R21AG077092. Dr. Goyal was a member of the Junior Investigator Intensive Program of the US Deprescribing Research Network and is supported by a US Deprescribing Research Pilot Grant, which are funded by the National Institute on Aging (R24AG064025). He has received consulting fees from Sensorum Health, Agepha Pharma, and Axon therapies; and has received personal fees for medicolegal consulting and expert testimony related to HF. Jerzy Gąsowski has given talks on the topic of anti-hypertensive therapy in older persons and deprescribing, some of which were associated with lecture fees. Wade Thompson has received grants for deprescribing research from Health Canada and the National Institute on Aging (US), is supported by a salary award from Michael Smith Health Research British Columbia; and receives an honorarium for writing an article on deprescribing in Pharmacy Practice Plus magazine. Eva Topinkova work was partly supported by the Project New Technologies in Translational Research in Pharmaceutical Sciences/NETPHARM reg. No. CZ.02.01.01/00/22_008/0004607, co-financed by the European Union. George Soulis has received consulting fees by Reckitt Benckiser for Advisory Board. All other authors have no conflict of interest to declare.

Ethics approval

N/A.

Informed consent

For this type of study, formal consent is not required.