Introduction
Hypertension has been shown to affect nearly one-third of adults aged 30 and older worldwide and is one of the most common chronic conditions. Globally, the number of adults with hypertension doubled from 650 million in 1990 to 1.3 billion in 2019 [1]. Among surgical patients, hypertension ranks as one of the most common comorbidities, with prevalence rates ranging from 48.8% to 62.6% in those undergoing noncardiac procedures [2–4].
Chronic hypertension shifts the cerebral autoregulatory curve to the right, increasing the threshold required for adequate organ perfusion. This shift increases patients’ vulnerability to tissue hypoperfusion, even during brief or mild episodes of hypotension [5]. Chronic hypertension is also characterized by blunted baroreflex sensitivity and increased arterial stiffness, both of which impair compensatory vascular responses [6–8]. Intraoperative hypotension most often occurs immediately after anesthesia induction and is considered an independent risk factor for myocardial injuries, acute kidney injuries, and postoperative deaths [4,9–14]. Observational studies have also reported that postinduction hypotension occurs in nearly one-third of unselected surgical populations but affects more than half of patients with treated hypertension [15–17]. This disproportionate risk emphasizes the need to evaluate hypertensive patients as a distinct high-risk group when assessing the hemodynamic effects of anesthetic agents.
Given these unique physiological vulnerabilities in patients with hypertension, selecting anesthetic agents that minimize hemodynamic instability is crucial. Clinicians commonly administer propofol to induce anesthesia because of its rapid onset and short duration [18,19]. However, the use of propofol is commonly associated with dose-dependent hypotension and bradycardia through the reduction of systemic vascular resistance and myocardial contractility [20,21] as well as baroreflex gain [22–24]. In contrast, remimazolam—a novel, ultrashort-acting benzodiazepine—delivers similar onset and recovery times while potentially providing greater hemodynamic stability [25–27].
Although recent meta-analyses have emphasized remimazolam’s hemodynamic advantages over propofol across various clinical settings [28–31], none have specifically assessed the outcomes in hypertensive patients. To address this limitation, a meta-analysis with a trial sequential analysis (TSA) was conducted to compare the hemodynamic effects of remimazolam and propofol during general anesthesia induction, specifically in hypertensive patients, providing targeted evidence for optimizing anesthetic selection in this vulnerable population.
Materials and Methods
Study design
This meta-analysis, which incorporated TSA, was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines [32]. The study protocol was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD42024607557).
Eligible criteria
Studies that met all of the following criteria were included: (1) randomized controlled trials (RCTs); (2) adult patients with hypertension undergoing general anesthesia, restricted to trials that explicitly used the terms hypertension in the title, abstract, or MeSH/Emtree terms to ensure a clearly defined target population; (3) a comparison between remimazolam and propofol as induction agents; and (4) reporting of at least one hemodynamic outcome, such as blood pressure or heart rate (HR), measured during the induction from the administration of the induction agent to the post-intubation period. Conversely, studies that involved animal experiments, observational designs, retrospective analyses, case reports, meta-analyses, review articles, editorials, or letters to the editor were excluded.
Search strategy and study selection
A systematic literature search was performed across MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials databases covering all records; the search was finalized on November 13, 2024. Our search terms included “remimazolam,” “propofol,” “hypertension,” and “randomized controlled trial.” The Supplementary Material 1 contains the complete search strategy. No language restrictions were applied during the search. However, all included studies were published in English.
Two investigators (S.E.S. and H.J.K.) independently screened the titles and abstracts to determine relevance based on the predefined eligibility criteria. Subsequently, they reviewed the full texts of the studies that appeared potentially eligible to confirm inclusion. When disagreements arose, a third reviewer (H.K.) facilitated the resolution through discussion.
Data extraction
After finalizing the selection of the eligible RCTs, two investigators (S.E.S. and H.J.K.) independently extracted the key study characteristics and outcome data using a standardized data extraction form. The extracted variables were as follows: study title, first author, journal name, year of publication, study design, country, type of surgery, patient demographics, sample size, and reported outcomes.
We primarily investigated the incidence of hypotension during the induction of general anesthesia. The secondary outcomes included the incidence of bradycardia and the maximum and minimum values of mean arterial pressure (MAP) and HR.
Due to the heterogeneity in the outcome definitions across the included trials, we used each study’s predefined criteria for hemodynamic events. Hypotension was most commonly defined as a MAP of < 65 mmHg or a decrease of ≥ 20%–30% from baseline. Hypertension was defined as a previous clinical diagnosis, the use of antihypertensive medication, or a baseline blood pressure of ≥ 140/90 mmHg. Most studies excluded patients with significantly uncontrolled hypertension, which is typically defined as a systolic blood pressure exceeding 160–180 mmHg. Bradycardia, when reported, was typically defined as a HR of < 50 beats/min. Supplementary Table 1 summarizes the detailed definitions used in each study.
Risk of bias and quality assessment
S.E.S. and H.J.K. independently assessed the methodological quality of each included study using the Revised Cochrane risk of bias tool for randomized trials (RoB 2.0). Five domains were assessed: (1) bias from the randomization process, (2) bias due to deviations from intended interventions, (3) bias from missing outcome data, (4) bias in the outcome measurements, and (5) bias in selecting the reported results. They rated each domain as “low risk,” “some concerns,” or “high risk.” Based on these ratings, the overall risk of bias of each study was defined as follows: low risk, all domains rated low; high risk, one or more domains rated high or multiple domains raised concerns; and some concerns, the ratings did not meet the criteria for either low or high risk.
The quality of evidence for each outcome was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach. This framework aims to assess the quality of evidence based on study limitations, inconsistency, indirectness, imprecision, and publication bias. In the present study, the strength of evidence was classified into four categories: (1) high, further research is unlikely to affect confidence in the estimate; (2) moderate, further research may impact confidence and may change the estimate; (3) low, further research will likely affect confidence and alter the estimate; and (4) very low, the estimate of effect remains highly uncertain.
Data analysis
All statistical analyses were performed using the Comprehensive Meta-Analysis software (version 2.0). Two investigators independently entered the extracted data and measured the pooled risk ratio (RRs) or mean differences (MDs) along with 95% CI. To assess heterogeneity across studies, Cochran’s Q test, Higgins’ I2, the DerSimonian–Laird estimator for τ, and prediction intervals (PIs) were applied. The PIs were omitted when τ equaled zero. Substantial heterogeneity was defined as a P value of < 0.1 in the Q test or I2 of > 50%. A fixed-effects model was used when heterogeneity was low and a random-effects model when it was high. The sensitivity analyses were conducted by excluding individual studies to assess their influence on the overall estimates. When the original data appeared as medians with interquartile ranges or ranges or means with standard errors, they were converted into means and standard deviations (SDs) for meta-analysis. To assess clinical relevance, the number needed to treat (NNT) and its 95% CI were calculated using the following formula: NNT = 1/[assumed control risk × (1−RR)]. Publication bias was not assessed due to the inclusion of fewer than 10 studies.
Trial sequential analysis
To mitigate the risk of random errors from sparse data, a TSA was performed. The TSA estimates the required information size (RIS)—the meta-analytic equivalent of a sample size calculation—to determine whether the cumulative evidence is sufficient and reliable. A cumulative Z-curve was generated using either a fixed- or random-effects model, with type I error set at 5%.
If the cumulative Z-curve crossed the monitoring boundary or entered the futility zone, the evidence was considered sufficient to confirm or refute the anticipated effect, making further studies unnecessary. If the Z-curve failed to cross any thresholds and the RIS remained unmet, the evidence was deemed inconclusive.
For binary outcomes, the RIS was estimated based on the event proportion in the propofol group, acknowledging the absence of robust prior data for effect-size specification. Thus, we pragmatically assumed a RR reduction of 30% for remimazolam, with an alpha of 5%, beta of 20%, and the observed heterogeneity across studies. To improve methodological transparency, sensitivity TSAs using more conservative assumptions (RR reduction of 20% and 10%) were additionally performed. For continuous outcomes, the observed SD, a pragmatically defined expected MD of SD/3 in the absence of prior evidence, an alpha of 5%, a beta of 20%, and the same heterogeneity parameters were used.
Results
Study selection
A thorough literature search comprising 583 database queries and three manual searches yielded 586 articles. After eliminating 501 duplicate records, 70 articles were excluded based on the title and abstract screening. The full texts of the remaining 15 articles were reviewed, and nine were removed for the following reasons: two did not compare remimazolam with propofol [33,34], six were not conducted under general anesthesia [35–40], and one was a commentary rather than original research [41]. Finally, six RCTs were included in the analysis (Fig. 1) [17,42–46].
Study characteristics
Table 1 shows the details of the six RCTs. These studies collectively enrolled 484 hypertensive patients and were conducted in Korea (n = 3), China (n = 2), and Japan (n = 1). The sample sizes ranged from 33 to 120 participants. The patient ages ranged from approximately 52 to 82 years, with one study specifically focusing on geriatric patients aged 80 and above. All participants were classified as American Society of Anesthesiologists physical status I–III and underwent several elective surgery procedures. Remimazolam was administered either as a single bolus injection or continuous infusion. Propofol was administered either as a bolus dose or target-controlled infusion. In all studies, opioid analgesia was consistently used, administering either remifentanil or sufentanil.
Meta-analysis and TSA
Incidence of hypotension
All six studies (totaling 484 patients) reported the incidence of hypotension [17,42–46], with the remimazolam group showing a significantly lower incidence compared with the propofol group (RR = 0.711, 95% CI [0.545–0.929], I2 = 67.54; Pchi2 = 0.009; τ = 0.250; 95% PI [0.430–1.177]; NNT 4; 95% CI NNT 3 to 7) (Fig. 2A). Sensitivity analysis, which was performed by sequentially excluding each study, confirmed the robustness of the findings (Fig. 3A). The TSA revealed that only 8.6% of the RIS (484 of 5652 patients) had been accrued. The cumulative Z-curve crossed the conventional test boundary, but it did not cross the monitoring boundary, indicating that further research is needed to confirm the effect (Supplementary Fig. 1A, Table 2). Supplementary Table 2 shows the sensitivity TSAs using more conservative assumptions for binary outcomes.
Incidence of bradycardia
Two studies (156 patients) assessed the incidence of bradycardia [42,46], revealing a lower rate of bradycardia in the remimazolam group than in the propofol group (RR = 0.256, 95% CI [0.101–0.649], I2 = 0.0; Pchi2 = 0.865, τ = 0.0; NNT 6; 95% CI NNT 4 to 16) (Fig. 2B). The TSA showed that only 6.3% of the RIS were accrued (156 of 2469 patients). The cumulative Z-curve crossed the conventional test boundary, but it did not cross the monitoring boundary (Supplementary Fig. 1B, Table 2). Supplementary Table 2 shows the sensitivity TSAs using more conservative assumptions for binary outcomes.
Maximum MAP
Four studies (331 patients) evaluated the maximum MAP [17,42,43,46]. No significant difference was observed between the remimazolam and propofol groups (MD = 5.523 mmHg, 95% CI [−1.142 to 12.188], I2 = 92.00; Pchi2 < 0.001; τ = 6.458; 95% PI [−15.029 to 26.075]) (Fig. 2C). However, the sensitivity analysis excluding the study by Choi et al. (2024) [43] altered the significance of the results (MD = 8.664 mmHg; 95% CI [3.321–14.007]) and eliminated heterogeneity (Fig. 3B). The TSA indicated that only 6.0% (331 of 5496 patients) of the RIS were accrued. Furthermore, the cumulative Z-curve did not cross the conventional test boundary or the monitoring boundary (Supplementary Fig. 1C, Table 2).
Minimum MAP
Five studies (451 patients) assessed the minimum MAP [17,42–44,46], with the remimazolam group showing higher values than the propofol group (MD = 9.023 mmHg, 95% CI [0.243–17.802], I2 = 97.50; Pchi2 < 0.001; τ = 9.820; 95% PI [−18.242 to 36.288]) (Fig. 2D). Sensitivity analysis, performed by removing one study at a time, generally did not alter the significance of the results, except when Choi et al. [43] and Xu et al. [46] were excluded. In that case, the results remained significant (MD = 4.701 mmHg, 95% CI [3.430–5.973]), and heterogeneity was eliminated (Fig. 3C). The TSA indicated that only 2.9% (451 of 15 587 patients) of the RIS were accrued. The cumulative Z-curve crossed the conventional test boundary, but it did not cross the monitoring boundary (Supplementary Fig. 1D, Table 2).
Maximum HR
Four studies (331 patients) assessed the maximum HR [17,42,43,46], with the remimazolam group showing higher values than the propofol group (MD = 6.280 beats/min, 95% CI [2.240–10.319], I2 = 70.80; Pchi2 < 0.001; τ = 3.367; 95% PI [−3.068 to 15.628]) (Fig. 2E). However, a sensitivity analysis removing the study by Choi et al. (2023) [42] changed the statistical significance (MD = 6.186 beats/min, 95% CI [−0.937 to 13.309]) and eliminated heterogeneity (Fig. 3D). The TSA indicated that only 18.3% (331 of 1809 patients) of the RIS were accrued. The cumulative Z-curve crossed the conventional test boundary, but it did not cross the monitoring boundary (Supplementary Fig. 1E, Table 2).
Minimum HR
Four studies (331 patients) assessed the minimum HR [17,42,43,46], with the remimazolam group showing higher values than the propofol group (MD = 7.200 beats/min, 95% CI [1.960–12.441], I2 = 86.40; Pchi2 < 0.001; τ = 4.875; 95% PI [−8.314 to 22.714]) (Fig. 2F). A sensitivity analysis excluding the study of Choi et al. (2023) [42] changed the significance of the findings (MD = 7.179 beats/min, 95% CI [−2.297 to 16.656]) and resolved heterogeneity (Fig. 3E). The TSA indicated that only 7.2% (331 of 4570 patients) of the RIS were accrued. The cumulative Z-curve crossed the conventional test boundary, but it did not cross the monitoring boundary (Supplementary Fig. 1F, Table 2).
Risk of bias and quality of evidence
Table 3 summarizes the risk of bias assessment using the Cochrane tool. All included studies showed a low risk of bias related to deviations from intended interventions, missing outcome data, outcome measurement, and the selective reporting of the results. However, four studies raised concerns regarding a potential bias in the randomization process [17,42,43,46].
Seven outcomes were assessed using the GRADE system (Table 4). The pooled evidence for bradycardia was rated as moderate quality. In contrast, the quality of evidence for hypotension, maximum MAP, minimum MAP, maximum HR, and minimum HR was rated as low.
Discussion
This meta-analysis assessed six RCTs involving a total of 484 hypertensive patients. The results indicated that remimazolam was associated with a lower incidence of hypotension and bradycardia compared with propofol. Moreover, secondary outcomes, such as minimum MAP, maximum HR, and minimum HR, were also shown to favor remimazolam. However, the TSA revealed that these differences were inconclusive due to the insufficient sample size, limited statistical power, and substantial heterogeneity across studies. To our knowledge, this is the first meta-analysis to incorporate TSA in assessing the hemodynamic stability of remimazolam in hypertensive patients.
The observed hemodynamic differences between remimazolam and propofol can be attributed to their distinct pharmacological profiles. Propofol tends to bind to the β3 subunit of GABA-A receptors, resulting in the suppression of central sympathetic activities and impairment of baroreceptor reflexes [22,23,47]. Propofol also inhibits voltage-gated calcium channels in the vascular smooth muscles, resulting in peripheral vasodilation and reduced myocardial contractility [21,24]. These combined effects often contribute to hypotension and bradycardia.
In contrast, remimazolam selectively targets the benzodiazepine binding site at the α-γ subunit interface of the GABA-A receptors, helping to preserve the autonomic nervous system balance [25,26]. Rapid carboxylesterase-mediated metabolism, independent of hepatic or renal function, ensures predictable pharmacokinetics and recovery [25,26]. These mechanistic distinctions align with the lower incidence of hypotension and bradycardia as observed in our findings.
Although the maximum and minimum MAP values differed by approximately 5–9 mmHg between the remimazolam and propofol groups, the clinical significance of these findings still requires cautious interpretation. A reduction of this magnitude in the MAP during induction may be relevant in hypertensive or high-risk cardiovascular patients, in whom even modest decreases in perfusion pressure can precipitate hypotension-related adverse events. However, the CIs, particularly for maximum MAP, introduce substantial uncertainty, suggesting that the observed differences cannot be directly translated into improved clinical outcomes. Thus, while the trend toward more stable MAP with the administration of remimazolam may be clinically favorable, the current evidence does not allow definitive conclusions with regard to its impact on peri-induction hemodynamic risk.
The NNT derived from this meta-analysis suggests that the administration of remimazolam may effectively reduce the risk of hemodynamic complications during anesthesia induction in hypertensive patients. Specifically, the NNT was 4 for hypotension and 6 for bradycardia, suggesting that one additional adverse event could be prevented for every 4–6 patients treated with remimazolam. However, these NNT estimates should be interpreted with caution, as the underlying data are limited and the CIs remain relatively wide, which may influence their precision. These findings are clinically relevant, as hypertensives are particularly vulnerable to hypotension due to chronic alterations in cerebral autoregulation [5] and the use of cardiovascular medications, such as β-blockers that may impair baroreflex-mediated compensatory responses [12–14]. Given its relatively stable hemodynamic profile, remimazolam may provide advantages for patients at increased cardiovascular risk. However, further high-quality trials are needed to validate these findings.
Our findings are consistent with those of previous meta-analyses comparing remimazolam and propofol across various clinical contexts. Multiple studies on general anesthesia have shown that remimazolam is associated with a lower incidence of intraoperative hypotension and bradycardia [28,31]. Similar trends have been reported in elderly patients [29] and during procedural sedation [30]. Moreover, research in endoscopic procedures consistently indicates superior cardiovascular safety with remimazolam compared with propofol [48,49]. However, these studies primarily assessed the general population and did not specifically assess patients with hypertension. Our study contributes new evidence by showing that remimazolam preserves its hemodynamic advantages in hypertensive patients. These findings support its use in high-risk patients who require hemodynamic stability during anesthesia induction.
Although our meta-analysis showed statistically significant reductions in hypotension and bradycardia following the administration of remimazolam, the TSA revealed that the required sample size was not achieved. Specifically, only 8.6% and 6.3% of the necessary sample sizes were achieved for hypotension and bradycardia, respectively. While the cumulative Z-curves surpassed the conventional significance thresholds, they did not cross the monitoring boundaries, suggesting that the current evidence is insufficient for definitive conclusions. Thus, additional large-scale RCTs are needed.
In addition to this limited information size, substantial heterogeneity was observed across hemodynamic outcomes. As highlighted in recent methodological reports, such variability likely reflects true clinical differences rather than statistical noise. In the context of such heterogeneity, the pooled estimates should be interpreted primarily as directional signals, while PIs encompassing the null effect indicate potential variability in treatment effects across clinical settings [50,51]. The included trials enrolled patients with diverse cardiovascular risk profiles, ranging from very elderly patients to middle-aged adults, with or without routine angiotensin-converting enzyme inhibitor or angiotensin receptor blocker therapy. They also encompassed various types of surgery, such as neurosurgical, breast, and orthopedic procedures, and applied different remimazolam regimens, including boluses and continuous infusions until loss of consciousness.
Furthermore, the definitions of hypertension and hypotension were not standardized, with hypertension defined either by previous diagnosis and antihypertensive treatment or by baseline blood pressure thresholds, and hypotension by either absolute MAP cutoffs (60–65 mmHg) or relative reductions of 20%–30% from baseline. These clinical and methodological variations provide a plausible explanation for the inconsistent effect sizes among trials and suggest that the hemodynamic impact of remimazolam is context-dependent within hypertensive populations. Moreover, the interpretation of the hemodynamic findings requires caution because hypertensive patients inherently show greater physiologic variability [6,7]. Increased arterial stiffness and impaired autonomic buffering can amplify blood pressure fluctuations [8]. The baseline blood pressure levels and hypertension severity varied across trials, and these factors may confound the magnitude of MAP reduction independent of the anesthetic agent. Thus, the observed heterogeneity likely reflected the patient-level variability in addition to pharmacologic effects.
This meta-analysis has several limitations. First, the small number of included studies resulted in several interrelated statistical limitations and the TSA showed that none of the outcomes reached the RIS. Thus, any observed patterns suggesting potential effect modifiers should be interpreted as hypothesis-generating rather than definitive. Second, two of the included studies administered remimazolam as an intravenous bolus, which is not an approved method for the induction of general anesthesia. This off-label administration may limit the applicability and regulatory relevance of the findings. Third, all included studies were conducted exclusively in East Asian countries, specifically Korea, China, and Japan, which may compromise the applicability of our findings to other populations and healthcare systems. Fourth, the observed differences in the MAP should be interpreted cautiously, as the wide CIs limit significant conclusions about their clinical relevance. Further multicenter RCTs involving diverse populations are essential to strengthen the evidence base. Finally, this review has a limitation related to identifying hypertensive patients. We included only trials that clearly reported hypertension in the title, abstract, or indexing terms. This ensured a well-defined study population but may have excluded mixed trials that contained hypertensive patients without separate reporting. Since hemodynamic responses tend to differ between hypertensive and normotensive individuals, the inclusion of such studies without subgroup data could cause a misclassification bias. Future RCTs that predefine hypertensive subgroups or provide accessible subgroup data would help strengthen the evidence base.
In conclusion, remimazolam indicates a trend toward more stable hemodynamic responses than propofol during anesthesia induction in hypertensive patients, with reduced risks of hypotension and bradycardia. These findings suggest that remimazolam may be a promising option in clinical settings wherein cardiovascular stability is important. However, given the limitations, further well-designed, large-scale studies are necessary to confirm the clinical significance of these results.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Full text via PMC
Download Citation
Supplement1
Print
