Dose-dependent effects of SGLT2 inhibitors on circadian blood pressure in hypertensive patients with diabetes: A systematic review and Bayesian network meta-analysis

Abstract

Objective

The impact of dose variation of Sodium-glucose co-transporter 2 (SGLT2) inhibitors in reducing blood pressure (BP) in patients with diabetes and hypertension remains unclear.

Methods

A systematic search up to December 29, 2024 identified randomized trials reporting SGLT2 inhibitor effects on 24-h, daytime, nighttime, and office BP, along with hypoglycemia, urinary tract infections, and volume depletion. Continuous outcomes were synthesized as mean differences and safety outcomes as odds ratios within a Bayesian random-effects network meta-analysis. Dose–response patterns were examined using Bayesian meta-regression (ΔDIC), and treatment rankings were derived using SUCRA.

Results

Nine RCTs involving 9093 participants were included. SGLT2 inhibitors produced modest reductions across 24-h, daytime, nighttime, and office BP, with the largest numerical effect for empagliflozin 25 mg (SBP –5.93 mm Hg), but no credibly significant differences among agents in head-to-head comparisons. Safety outcomes showed no credibly significant excess risk for hypoglycemia, UTI, or volume depletion, with overall event rates low. Bayesian meta-regression yielded ΔDIC values near zero, indicating no detectable dose–response pattern for either blood pressure or safety outcomes.

Conclusion

SGLT2 inhibitors produced modest and consistent reductions in BP, with no evidence of dose-dependence or meaningful differences among agents. Safety events were not credibly increased. These findings support their use as adjunctive therapy in patients with diabetes and hypertension.

1. Introduction

Despite numerous advancements in medicine, diabetes remains a highly prevalent disease worldwide, affecting 10.5 % of adults [1]. Previous studies indicated that up to 82 % of diabetic patients exhibit elevated blood pressure (BP) levels, rendering hypertension a common comorbidity in diabetes [[2], [3], [4]]. Both diabetes mellitus and hypertension are significant contributors to microvascular complications and cardiovascular (CV) diseases, raising the risk of CV death 4 times higher in diabetic patients and 2–3 times higher in the hypertensive population [5,6]. Diabetic patients who also have hypertension face a 25 % chance of experiencing any CV incident and a 30 % risk for all-cause mortality [7]. Previous studies have shown that controlling BP can lead to lower death rates and fewer CV events in people with diabetes [[8], [9], [10]]. These findings emphasize the critical role of BP managements in the diabetic population. Furthermore, the most recent international guidelines, including the American Diabetes Association 2025, European Society of Cardiology (ESC) 2024, and European Society of Hypertension 2024 guidelines, recommend 130/80 mmHg as the target BP using drugs and lifestyle modifications in diabetic patients [[11], [12], [13]].

Sodium-glucose cotransporter 2 (SGLT2) inhibitors are a category of drugs initially employed for hypoglycemic effects by reducing glucose reabsorption in the kidney [14]. It has been suggested that SGLT2 inhibitors' diuretic effects can exert positive impacts on the CV system and BP regulation [15]. These drugs showed promising results in systolic blood pressure (SBP) and diastolic blood pressure (DBP) regulation, as well as CV outcomes in the patients with diabetes [[16], [17], [18], [19]]. SGLT2 inhibitors are also suggested for helping to lower BP moderately in patients with chronic kidney disease according to the ESC 2024 blood pressure management guideline [11]. Moreover, these drugs indicated significant impacts on circadian BP, which is defined as 24-h variation in BP through ambulatory monitoring. Researchers have recently investigated the potential of circadian BP as an indicator of cardiovascular health, as it significantly affects numerous physiological processes [20]. Mie K. Eickhoff and colleagues found that people taking dapagliflozin had a 2.9 mm Hg bigger drop in diastolic ambulatory BP than those not taking the medication [21]. Similarly, the study by Keith C. Ferdinand et al. demonstrated that empagliflozin effectively lowered ambulatory SBP and DBP in Black hypertensive patients with type 2 diabetes [22].

Considering various available doses of SGLT2 inhibitors, effects and side effects associated with different doses should be noted. Neeta B. Amin and others found that different doses of ertugliflozin (1, 5, 25 mg) lower BP by varying degrees in patients with diabetes and high BP. SGLT2 inhibitors’ dose-dependent impacts were subsequently confirmed by other studies [[23], [24], [25]]. However, there is still no consensus on the differences in outcomes between low and high doses of these medications regarding circadian BP regulation. Thereby, in this study, we aimed to systematically review the effects of SGLT2 inhibitors on circadian BP in the patients with diabetes and hypertension to evaluate their dose-dependent efficacy, and where applicable, provide the updated quantitative pooled analyzed results.

2. Methods and materials

This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and registered on PROSPERO (CRD420251023938).

2.1. Search strategy and study selection

A comprehensive literature search was conducted for articles published up to December 2024 from the following databases which were Scopus, PubMed, and Embase. The search strategy was conducted using the PICOS tool: Participants (P): diabetic patients with hypertension; Intervention (I): sodium-glucose cotransporter 2 inhibitors; Comparators (C): placebo or standard care; Outcomes (O): 24-h ambulatory systolic blood pressure (SBP), 24-h ambulatory diastolic blood pressure (DBP), daytime SBP, daytime DBP, nighttime SBP, nighttime DBP, urinary tract infection (UTI), events related to volume depletion, and hypoglycemia; and Study design (S): randomized controlled trials (RCTs). Detailed search strategies and keywords are provided in Supplementary file1, Table 1. In addition, we hand-searched the reference lists of included studies after full-text screening to identify any potentially eligible studies not captured by the database search. No language restrictions were applied during the initial database search.

Table 1.

Baseline characteristics of included studies.

Study, year	Country	Case/Control No.	Male No. (%)	Mean age	Intervention	Follow-up (weeks)	BP outcomes reported
Amin, 2015 [24]	Multi-center	A = 39, B = 38, C = 39 Control = 77	131 (67 %)	NA	A = Ertugliflozin 1 mg B= Ertugliflozin 5 mg C= Ertugliflozin 25 mg	4	24h-SBP and DBP, daytime and nighttime SBP and DBP.
Cheng, 2022 [26]	China	62/62	77 (62.1 %)	71.45	Empaglifozin 25 mg	12	24h-SBP and DBP, daytime and nighttime SBP and DBP, office-seated SBP and DBP.
Ferdinand, 2019 [22]	United States	78/72	79 (52.7 %)	56.8	Empagliflozin 10–25 mg	24	24h-SBP and DBP, daytime and nighttime SBP, office-seated SBP and DBP.
Heerspink, 2016 [27]	Multi-center	167/189	228 (64 %)	54.93	Dapagliflozin 10 mg	12	Office-seated SBP
Sjöström, 2015 [28]	Multi-center	2294/2222	2622 (58.1 %)	59.5	Dapagliflozin 10 mg	24	Office-seated SBP and DBP
Tikkanen, 2015 [25]	Multi-center	552/271	495 (60.1 %)	60.2	Empagliflozin 10 mg and 25 mg	12	24h-SBP and DBP, daytime and nighttime SBP and DBP, office-seated SBP and DBP.
Townsend, 2015 [29]	USA	113/56	98 (58.0 %)	58.6	Canagliflozin 100 mg and 300 mg	6	24h-SBP and DBP, daytime and nighttime SBP and DBP, office-seated SBP and DBP.
Weber, 2015 [30]	USA	225/224	247 (55 %)	56.5	Dapagliflozin 10 mg	12	24h-SBP and DBP, daytime and nighttime SBP, office-seated SBP and DBP.
Weir, 2014 [31]	Multi-center	1667/646	1146 (49.5 %)	55.9	Canagliflozin 100 mg and 300 mg	26	Office-seated SBP and DBP

Abbreviations: SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; 24h, 24-h; NA, not available; mg, milligrams; BP, blood pressure.

All references retrieved from database searches were imported into the Rayyan Intelligent Systematic Review web application (Rayyan Systems Inc. Software Development, Cambridge, MA) for duplicate removal. Subsequently, two reviewers (K.R. and E.E.) independently screened the titles and abstracts of the remaining records, and potentially eligible studies were then assessed in full text by the two reviewers (A.S. and D.S).

2.2. Data extraction and risk of bias assessment

Initially, a data extraction form was developed, and two independent authors (S.S., T.A.) reviewed and extracted data from each study, including baseline clinical and demographic characteristics, sample size, and changes in outcomes of interest. Continuous BP outcomes (24-h SBP/DBP, daytime SBP/DBP, nighttime SBP/DBP) were analyzed as mean differences from baseline for each treatment arm. When numerical values were not explicitly reported in the manuscript, data were extracted from published Fig.ures using WebPlotDigitizer (version 5.2). For safety outcomes, raw binary event data were extracted directly from the included trials to allow calculation of odds ratios within the network meta-analysis framework. The risk of bias of included studies was evaluated using version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2). Each trial was assessed across five domains: the randomization process, deviations from intended interventions, incomplete outcome data, measurement of the outcome, and selective reporting. Two reviewers (M.S. and F.C.) independently performed all risk-of-bias assessments, and disagreements were resolved in consultation with a senior reviewer (K.H.) until consensus was reached.

2.3. Statistical analysis

2.3.1. Bayesian network meta-analysis

A Bayesian network meta-analysis was performed using the gemtc and rjags frameworks to compare the effects of individual SGLT2 inhibitors. A random-effects hierarchical model was fitted using four Markov chains. The number of adaptation and sampling iterations was incrementally increased until satisfactory convergence was achieved, defined as a potential scale reduction factor (PSRF) ≤ 1.05 across all monitored parameters. The final model used 10,000 adaptation iterations and 100,000 sampling iterations. We initially planned to assess inconsistency by comparing direct and indirect evidence using node-splitting and network heat plots; however, because all active treatments were connected only through placebo, the evidence network formed a star-shaped structure with no closed loops, leaving no independent pathways for inconsistency assessment. Treatment ranking was derived from posterior rank probabilities and the Surface Under the Cumulative Ranking curve (SUCRA). For continuous outcomes, results are reported as mean differences (MDs) with 95 % credible intervals (CrIs), and for binary outcomes, as odds ratios (ORs) with 95 % CrIs. To evaluate potential small-study effects, comparison-adjusted funnel plots were generated and Egger's asymmetry test was performed, with statistical significance defined as a two-sided p < 0.05. The analysis used default non-informative priors implemented in gemtc, including vague priors for treatment effects and a broad prior for between-study heterogeneity.

2.3.2. Bayesian meta-regression

To evaluate whether SGLT2 inhibitor dose influenced BP responses, a Bayesian meta-regression model was developed by incorporating dose as a study-level covariate, coded ordinally (1 = low, 2 = moderate, 3 = high). The regression model was compared to the standard Bayesian model using the deviance information criterion (DIC). DIC differences were used to assess improvement in model fit, with lower DIC indicating better fit.

All statistical analyses were performed using R software (version 4.5.1; R Foundation for Statistical Computing). All p-values were two-sided, and values < 0.05 were considered statistically significant. The following R packages were used for the analyses: dmetar, gemtc, netmeta, dplyr, rjags, rgl, and metafor.

3. Results

The systematic search yielded a total of 4002 records. After the removal of duplicates and screening of titles and abstracts, 47 articles were selected for full-text review. Following the application of eligibility criteria, nine randomized controlled trials were included in the final analysis [22,[24], [25], [26], [27], [28], [29], [30], [31]]. The PRISMA flow diagram is presented in Fig. 1.

Fig. 1.

PRISMA flowchart.

3.1. Baseline characteristics

A total of 9093 participants were included across the 9 randomized controlled trials identified through the systematic search (initial n = 4002; full-text reviewed n = 47). Of the total population, 5161 (56.3 %) were male. Among the participants, 5274 were assigned to the intervention groups and 3819 to the control groups. The overall mean age of participants ranges from 54.93 to 71.45 years. Detailed baseline characteristics of the included studies are provided in Table 1.

3.2. Quality assessment

Out of the nine included studies, six were judged to have a low risk of bias across all five domains. Two studies were rated as having high risk of bias, and one study was concluded to have some concerns. A summary of the RoB 2 assessment results is presented in Supplementary file 1, Fig. 1.

3.3. Quantitative synthesis

3.3.1. Ambulatory blood pressure

The network graph for 24-h outcomes is shown in Fig. 2. SGLT2 inhibitors produced modest reductions in 24-h ambulatory BP compared with placebo. The largest decrease in 24-h SBP was observed with empagliflozin 25 mg (−5.93 mm Hg; 95 % CrI, −11.1 to −0.61) (Fig. 3a). The greatest reduction in 24-h DBP was also seen with empagliflozin 25 mg (−3.3 mm Hg; 95 % CrI, −6.92 to 0.48), although this was not credibly significant (Fig. 3b). Other agents showed similar but nonsignificant reductions, and no meaningful differences emerged in head-to-head comparisons, which indicated comparable class-wide efficacy (Supplementary File 2, Tables 1 and 2) (see Fig. 4).

Fig. 2.

Network Plots for 24-Hour Blood Pressure Comparisons. Network diagrams illustrate the evidence structure for 24-h SBP and DBP across SGLT2 inhibitors versus placebo. Line thickness corresponds to the number of trials contributing to each comparison.

Fig. 3.

Effect of SGLT2 Inhibitors on 24-h Blood Pressure. Mean differences in 24-h SBP (a) and DBP (b) versus placebo are shown, with points indicating posterior mean differences and lines showing 95 % CrI; negative values indicate BP reduction.

Fig. 4.

Bayesian Dose–Response Meta-Regression for Blood Pressure and Safety Outcomes. Panels a and b display ΔDIC values from Bayesian meta-regression models assessing whether adding dose as a covariate improves model fit for BP and safety outcomes. Values clustering around zero indicate no detectable dose-response pattern across efficacy or safety endpoints.

3.3.2. Daytime blood pressure

The network graph for daytime BP outcomes is presented in Supplementary File 1, Fig. 2. Across daytime assessments, SGLT2 inhibitors were associated with numerical decreases in both SBP and DBP. The greatest reductions were observed with empagliflozin 25 mg for both SBP (−6.63 mm Hg; 95 % CrI, −13.0 to 0.31) and DBP (−3.90 mm Hg; 95 % CrI, −8.13 to 0.44). Other agents demonstrated reductions of similar magnitude but without credible significance (Supplementary File 1, Fig. 3a and b), and no significant differences were detected among active treatments (Supplementary File 2, Tables 3 and 4).

3.3.3. Nighttime blood pressure

The corresponding nighttime network graph appears in Supplementary File 1, Fig. 4. A similar pattern was observed during night-time periods. The largest SBP reduction occurred with empagliflozin 25 mg (−5.31 mm Hg; 95 % CrI, −10.4 to −0.026), while the greatest DBP decrease was seen with ertugliflozin 5 mg (−3.45 mm Hg; 95 % CrI, −9.60 to 2.70) (Supplementary File 1, Fig. 5a and b). Other agents showed modest, but nonsignificant reductions, with no detectable differences among active regimens (Supplementary File 2, Tables 5 and 6).

3.3.4. Office blood pressure

The office BP network graph is shown in Supplementary File 1, Fig. 6. SGLT2 inhibitors also lowered office BP numerically relative to placebo. The greatest reduction in office SBP was observed with ertugliflozin 5 mg (−7.21 mm Hg; 95 % CrI, −14.1 to −0.269), and the largest DBP reduction with empagliflozin 25 mg (−4.14 mm Hg; 95 % CrI, −7.69 to −0.659) (Supplementary File 1, Fig. 7a and b). Reductions with other agents were similar in magnitude but not credibly significant, and no head-to-head comparison showed a meaningful difference, supporting broadly comparable efficacy across the class (Supplementary File 2, Tables 7 and 8).

3.3.5. Safety outcomes

Network graphs for all safety outcomes are provided in Supplementary File 1, Fig. 8. Across volume depletion, hypoglycemia, and UTI, SGLT2 inhibitors did not demonstrate any consistent or credibly significant excess risk relative to placebo. Point estimates varied substantially; however, the credible intervals were wide because event counts were sparse, which limited the strength of the conclusions (Supplementary File 1, Fig. 9a, b, c). No consistent or repeatable safety concerns were detected, and direct comparisons between individual agents did not show statistically significant differences (Supplementary File 2, Tables 9–11). Overall, adverse-event rates were uniformly low and comparable across all SGLT2 inhibitors.

3.3.6. Dose-response Bayesian meta-regression

Including dose as a covariate did not meaningfully improve model fit; ΔDIC values remained near zero, indicating no detectable dose-response relationship in blood-pressure or safety outcomes (Fig. 4, Supplementary file 1, Table 2).

3.3.7. SUCRA rankings for efficacy and safety

SUCRA rankings placed empagliflozin 25 mg highest for BP lowering, followed by canagliflozin 300 mg, though differences were modest and largely overlapping (Fig. 5a and b). No agent demonstrated a consistent safety disadvantage.

Fig. 5.

SUCRA Rankings for Blood Pressure and Safety Outcomes. (a): Cumulative ranking probabilities (SUCRA) for all blood-pressure outcomes across SGLT2 inhibitor regimens, illustrating modest separation with empagliflozin 25 mg generally scoring highest. (b): SUCRA values for safety endpoints (UTI, volume depletion, hypoglycemia), showing minimal differentiation between treatments and no consistent safety disadvantages.

3.3.8. Publication bias and small-study effects

Across most blood-pressure outcomes, funnel plots appeared symmetric, and Egger's tests were nonsignificant, including 24-h SBP (p = 0.76), 24-h DBP (p = 0.32), and daytime SBP (p = 0.42). Modest asymmetry was noted for nighttime SBP (p = 0.033) and office SBP (p = 0.007), likely reflecting heterogeneity rather than publication bias. For safety endpoints, a small-study effect was suggested only for hypoglycemia (p = 0.0286), whereas volume depletion (p = 0.12) and UTI (p = 0.61) demonstrated no evidence of bias. Overall, these findings support the robustness of the network. All funnel plots are shown in Supplementary File 1, Figs. 10–12.

4. Discussion

4.1. Overall findings

Across the Bayesian network meta-analysis, four consistent themes emerged: (i) SGLT2 inhibitors produced modest reductions in 24-h, daytime, nighttime, and office BP, with empagliflozin 25 mg showing the largest numerical effects, though most estimates were not credibly significant; (ii) head-to-head comparisons revealed no significant differences among active agents, indicating comparable efficacy across agents; (iii) Bayesian meta-regression demonstrated no evidence of a dose-dependent relationship as ΔDIC values clustered near zero, indicating that antihypertensive effects were consistent across standard doses; and (iv) SUCRA rankings showed only modest separation between treatments, with empagliflozin 25 mg ranking highest for BP lowering but with substantial overlap, and no agent displaying a consistent safety disadvantage. Together, these findings indicate that SGLT2 inhibitors exert a modest but uniform BP-lowering effect without discernible dose-response gradients or between-drug differences in efficacy or safety.

4.2. Mechanisms underlying blood pressure reduction

Our meta-analysis aligns with previous research, demonstrating that SGLT2 inhibitors, whether used as monotherapy or add-on therapy, exert modest but clinically relevant reductions in BP in patients with type 2 diabetes and hypertension [32,33]. This pattern of effectiveness corresponds closely with the pharmacological actions of SGLT2 inhibitors. By promoting urinary glucose excretion, these agents trigger osmotic diuresis and enhance sodium excretion, reducing plasma volume and arterial stiffness, contributing to lower BP [34]. Moreover, SGLT2 inhibitors’ cardiovascular benefits appear to involve multiple pathways, including reduced renal glucose reabsorption, inhibition of the renin-angiotensin-aldosterone system, and improvement in endothelial function [35,36]. Additionally, SGLT2 inhibition downregulates renal sodium-hydrogen exchanger activity, attenuates sympathetic nervous system activation, and suppresses the renin-angiotensin-aldosterone system (RAAS), collectively reducing vascular resistance. Emerging evidence also highlights improvement in endothelial function, reduction in oxidative stress, and anti-inflammatory effects as complementary pathways [37,38]. Together, these hemodynamic and vascular alterations provide a plausible explanation for the consistent reduction in BP observed across studies.

4.3. Mechanistic basis for dose-independent blood pressure effects

The mechanistic basis underlying the largely dose-independent BP–lowering effect of SGLT2 inhibitors appears to relate to the transient nature of natriuresis and osmotic diuresis, which are counterbalanced by compensatory mechanisms, including increased distal sodium reabsorption and restoration of plasma volume over time [[39], [40], [41]]. Consistent with this, SGLT2 inhibitors preferentially reduce interstitial rather than intravascular volume, thereby limiting sustained volume depletion compared with traditional diuretics [42]. Notably, BP lowering persists even as kidney function declines and glycosuric and natriuretic effects diminish, arguing against natriuresis alone as the dominant mechanism [43].

4.4. Comparison with previous meta-analyses and circadian blood pressure regulation

In a meta-analysis, Tian et al. showed that combination therapy with SGLT2 inhibitors and ACEIs/ARBs in type 2 diabetes mellitus (T2DM) reduces BP when compared to ACEIs/ARBs alone [33]. This may be due to the mechanism of action of this drug or the reduction of activation of the RAAS system by SGLT2 inhibitors through the down-regulating kidney expression of type 2 angiotensin II receptor as shown in an animal model [44]. Similarly, the meta-analysis by Georgianos et al., which included 2381 participants from 7 studies, demonstrated that SGLT2 inhibitors reduce 24-h, daytime, and nighttime BP, consistent with our findings. Additionally, our analysis did not identify any dose-dependent differences, consistent with the absence of a dose–response relationship in Bayesian meta-regression. Both studies also reported daytime reductions that numerically exceeded nighttime values, a pattern that was directionally similar to our findings [45]. Although reductions in nighttime BP were smaller in magnitude, even modest nocturnal decreases may contribute to a shift from non-dipping to a dipping BP pattern [46,47]. Experimental and limited clinical evidence indicates that SGLT2 inhibitors may influence circadian BP regulation through mechanisms beyond volume reduction, including improvements in insulin resistance and reductions in serum uric acid and triglyceride levels, all of which have been implicated in non-dipping BP profiles [[48], [49], [50], [51], [52]]. Nevertheless, the relative contribution of these pathways to nighttime BP regulation during SGLT2 inhibitor therapy remains incompletely defined.

4.5. Consistency of effects across agents and doses

Furthermore, another meta-analysis indicated that ertugliflozin at both 5 mg and 15 mg doses has similar efficacy and safety profiles, and both dosages contribute to a significant reduction in BP [53]. A meta-analysis (7 studies, 8499 patients) comparing tirzepatide, liraglutide, and SGLT2 inhibitors found that liraglutide (at doses ≥1.2 mg) has a more pronounced effect on lowering blood glucose compared to SGLT2 inhibitors. However, SGLT2 inhibitors demonstrated a greater ability to reduce BP, making them particularly suitable for diabetic patients with hypertension [54].

4.6. Evidence from cardiovascular outcome trials

Notably, large cardiovascular outcome trials such as EMPA-REG OUTCOME, CANVAS, and DAPA-HF have reported modest but statistically significant reductions in BP with SGLT2 inhibitors, often cited as secondary outcomes [16,17,55]. These trials primarily focused on cardiovascular mortality and heart failure hospitalization. However, the consistent observation of BP reductions across these studies further reinforces the validity of our findings.

4.7. Safety profile of SGLT2 inhibitors

In terms of safety outcomes, our analysis did not identify any credibly significant increase in hypoglycemia with SGLT2 inhibitors, although estimates were imprecise due to sparse events. In prior studies, this risk appears to be amplified when SGLT2 inhibitors are administered in combination with other antidiabetic agents, such as sulfonylureas and/or insulin [56]. Similarly, Rosenstock et al. stated that the incidence of hypoglycemia episodes is higher in patients using a combination therapy of methionine and SGLT2 inhibitors than in patients using methionine alone [57]. In addition, in another meta-analysis (17 studies, 11464 patients), SGLT2 inhibitors, except for canagliflozin 100 mg/d (OR: 3.12, 95 % CI: 1.18–8.36), were not significantly linked to an increased risk of hypoglycemia compared to placebo [58]. Roden et al. also demonstrated that the use of SGLT2 inhibitors did not significantly increase the risk of hypoglycemic events compared to placebo (RR = 0.92, 95 % CI: 0.83–1.02) [59]. The underlying reason is that SGLT2 inhibitors do not enhance insulin secretion or modulate glucose production [60].

The evidence surrounding adverse events associated with SGLT2 inhibitors, particularly UTI, remains inconsistent across studies. Xu L et al. demonstrated that none of the SGLT2 inhibitors evaluated in their study were associated with an increased risk of UTI, and there were no significant differences in UTI risk between the various combinations of SGLT2 inhibitors and metformin [61]. In another meta-analysis (21 studies, 12399 patients), only dapagliflozin at a dose of 10 mg per day was significantly linked to an increased risk of UTI compared to placebo (OR 2.14, 95 % CI 1.03–4.44), while no significant associations were observed with other treatments [58].Although, in our study, the point estimates indicated a possible rise in UTI events; the effect was not statistically credible.

In the present study, SGLT2 inhibitors did not demonstrate any credibly significant excess risk relative to placebo, although the interpretation is limited by low event counts. This pattern is consistent with prior data showing that serious volume depletion was slightly more frequent in patients receiving empagliflozin than placebo (0.94 vs. 0.88 events per 100 patient-years; rate ratio 1.08 [95 % CI: 0.88–1.31]), yet without statistically significant differences in either comparison [62].

4.8. Clinical implications and future directions

The findings of this meta-analysis reinforce the modest but consistent BP-modifying effects of SGLT2 inhibitors in patients with type 2 diabetes and coexisting hypertension. Nevertheless, these agents should not be considered as first-line antihypertensive therapies. Instead, they should be regarded as adjunctive agents that confer additional cardiovascular and renal benefits beyond glycemic control and BP modulation [63]. Their integration into antihypertensive regimens, particularly in patients already receiving renin-angiotensin system blockers or other cardioprotective agents, may help target residual risk in this high-risk population. Notably, SGLT2 inhibitors demonstrate a unique therapeutic profile, offering multi-organ benefits such as reduced cardiovascular mortality, heart failure hospitalizations, and improved renal outcomes, which appear not to be solely mediated by BP reduction. Ultimately, the strategic use of SGLT2 inhibitors as add-on therapy in the comprehensive management of patients with T2DM and hypertension may optimize cardiometabolic outcomes without compromising safety. Future RCTs are warranted to evaluate the additive BP-modifying effects of SGLT2 inhibitors in combination with other drugs.

5. Conclusion

SGLT2 inhibitors produced modest and consistent reductions across ambulatory, daytime, nighttime, and office BP, with no meaningful differences among individual agents and no evidence of dose-dependency. Safety outcomes showed no credibly significant excess risk, although interpretation was limited by sparse events. These findings support the use of SGLT2 inhibitors as adjunctive agents in hypertensive patients with diabetes, offering stable BP-lowering effects within a favorable safety profile.

6. Limitations

Despite the comprehensive nature of our analysis, several limitations must be noted. First, although we included only randomized controlled trials, the heterogeneity in study designs, follow-up durations, baseline characteristics, and BP measurement techniques may introduce variability into pooled estimates. Second, the included trials did not use uniform dose stratification, and dose–response categorizations were not prespecified across studies; although we used Bayesian meta-regression to evaluate dose as a continuous covariate, this limitation may still influence interpretation. Third, while our Bayesian network meta-analysis enabled pairwise head-to-head comparisons across all active agents, the structure of the evidence network imposes constraints. Because the evidence network was predominantly connected through the placebo arm and did not contain closed loops, formal inconsistency testing using node-splitting was not feasible. This is a structural limitation of star-shaped networks rather than a methodological issue. In addition, we did not apply the GRADE framework. This decision was driven by the predominantly star-shaped structure of the Bayesian network, which lacked closed loops and therefore did not meet the methodological assumptions required for reliable assessment of inconsistency and indirectness using GRADE. Fourth, although we evaluated ambulatory and office BP measures, data beyond 24 weeks were limited, which restricted our ability to assess durability of BP effects and long-term safety. Fifth, adverse events such as hypoglycemia, urinary tract infections, and volume-depletion events may have been underreported or inconsistently defined, and several outcomes were affected by sparse event counts, which contributed to wide credible intervals. Finally, we were unable to assess differential effects of SGLT2 inhibitors across specific hypertension phenotypes, including renin-dependent or volume-dependent forms of hypertension.

CRediT authorship contribution statement

Davood Semirani-Nezhad: Formal analysis, Data curation, Conceptualization. Khatere Roozbehi: Writing – original draft, Data curation, Conceptualization. Arman Soltani Moghadam: Writing – review & editing, Data curation. Elham Ebrahimi: Project administration, Data curation. Keyvan Salehi: Writing – original draft. Anahita Hashempoor: Writing – original draft, Data curation. Fatemeh Chichagi: Writing – original draft, Data curation. Sima Shamshiri: Writing – original draft, Data curation. Tara Azardar: Writing – original draft, Data curation. Mahshad Sabri: Writing – original draft. Ali Moradi: Writing – review & editing. Mani Khorsand Askari: Writing – review & editing. Toshiki Kuno: Writing – review & editing. Kaveh Hosseini: Writing – review & editing, Supervision, Conceptualization.

Consent to participate

Not applicable.

Availability of data and material

No new data were generated or analyzed in this study. All data used in this meta-analysis were obtained from open access sources and prior studies.

Code availability

Not applicable.

Ethics approval

Not applicable, as this study was a systematic review and meta-analysis of previously published studies.

Consent for publication

Not applicable.

Disclosure of AI programs

We employed AI tools, specifically ChatGPT 5.1 and AI-assisted writing tools, to faciliate in rephrasing. The aim was to enhance clarity and fluency only. All AI-generated content was carefully reviewed and validated by the authors to ensure accuracy and integrity, and all authors take full responsibility for final publication.

Funding

This research received no grant.

Competing interests

The authors declare no competing interests.

Appendix A. Supplementary data

The following are the Supplementary data to this article.