The safety and efficacy of remimazolam, ciprofol, and propofol anesthesia in endoscopy: a systematic review and network meta-analysis

Abstract

Background

While propofol remains widely used for endoscopic sedation, its cardiovascular depression and injection pain limitations have prompted exploration of novel agents (remimazolam, ciprofol). This study aimed to compare their safety and efficacy profiles systematically.

Methods

We conducted a network meta-analysis to evaluate remimazolam, ciprofol, and propofol for gastrointestinal endoscopy. Bayesian random-effects models were used to estimate relative risks (RR) and mean differences (MD) with 95% credible intervals(CrI).

Results

Forty-two randomized controlled trials (N = 10,540 patients) were included. Remimazolam demonstrated superior cardiovascular safety (RR = 0.44, 95%CrI 0.35–0.54 vs propofol) and lowest respiratory depression risk (RR = 0.36, 0.28–0.46). Propofol showed faster recovery (MD -14.22 min, -2.35 to -30.83 vs remimazolam). Both remimazolam (RR = 0.045) and ciprofol (RR = 0.054) significantly reduced injection pain versus propofol.

Conclusion

Remimazolam should be prioritized for high-risk patients (cardiovascular/respiratory comorbidities) despite slightly longer recovery times. Propofol remains suitable for low-risk procedures requiring rapid turnover, while ciprofol offers balanced efficacy for endoscopy.

Trial Registration

The study was registered with the UK National Institute for Health Research's PROSPERO platform (CRD42024569405;https://www.crd.york.ac.uk/prospero/).

Introduction

Endoscopic procedures typically encompass non-invasive techniques such as gastrointestinal endoscopy, bronchoscopy, and hysteroscopy, among others [1]. The current standard practice involves performing endoscopic surgery with sedation [2]. Sedation is crucial for both patient comfort and the success of endoscopic procedures. This procedure is essential as patients undergoing endoscopy may experience anxiety, pain, fear, and gastrointestinal discomfort, which can reduce their cooperation and increase the risk of negative cardiovascular events. [3]. For procedural sedation and analgesia (PSA), sedative agents need to strike a balance between providing enough sedation and maintaining hemodynamic stability, while also minimizing negative effects [4].

While sedation improves endoscopy and overall patient satisfaction, it poses a risk to patient safety if sedative medications are used inappropriately [5]. Among the various agents available, propofol has emerged as a popular choice due to its unique pharmacological profiles and sedative properties for sedation during endoscopic procedures [6]. As a long-standing sedative, propofol is renowned for its rapid induction and emergence characteristics [7, 8]. While effective, its use is associated with some limitations, such as cardiovascular depression and pain injection [9, 10].

Remimazolam is an ultra-short-acting benzodiazepine that has been noted for its rapid onset and offset of action, making it a contender for procedures requiring short-term sedation [11,12,13]. Its unique metabolism, largely through tissue esterases, allows for a predictable recovery profile, which can enhance patient satisfaction and operational efficiency in clinical settings. Ciprofol, a novel sedative agent developed with a pharmacokinetic profile tailored for short outpatient procedures, has gained attention for its potential to provide effective sedation with a reduced incidence of respiratory depression compared to traditional sedatives [14, 15]. Preliminary studies suggest that ciprofol may share some advantages over established agents in terms of safety and recovery time, although comprehensive evaluations are still needed [16]. Despite an increasing number of randomized controlled trials (RCTs) comparing these anesthetics, there remains a lack of consensus on the optimal choice for endoscopic procedures. Previous systematic reviews have primarily focused on direct comparisons between two agents [17, 18], leaving a gap in our understanding of the relative safety and efficacy of remimazolam, ciprofol, and propofol anesthesia in endoscopy. Given their widespread application, it is imperative to continually evaluate the effectiveness of remimazolam and ciprofol in comparison to propofol.

This systematic review and network meta-analysis aims to synthesize the existing literature and provide a comprehensive comparison of remimazolam, ciprofol, and propofol regarding their safety and efficacy in endoscopic procedures.

Methods

Data sources and strategy

This network meta-analysis evaluated the effectiveness and safety of three drugs, including ciprofol, propofol, and remimazolam. The study was conducted by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. The study was registered with the UK National Institute for Health Research's PROSPERO platform (Date: 15 Jul 2024, CRD42024569405; https://www.crd.york.ac.uk/prospero/). No external funding was provided for this study.

A systematic search of the PubMed, Embase, and Web of Science databases was conducted up to December 11, 2024. To ensure the inclusion of as many relevant studies as possible, a combination of free and subject terms was used, including Ciprofol, Propofol, and Remimazolam. The search terms included Ciprofol, Propofol, Remimazolam, and endoscopy. The search strategy was adapted to the characteristics of each database (Supplementary Appendix 1).

Study selection

Titles and abstracts were independently screened by two researchers, with final inclusion determined at their discretion. Disagreements regarding the full text were resolved through discussion and consensus. If consensus could not be reached, advice was sought from independent experts. No language restrictions were applied for the inclusion of studies (Table S1).

Inclusion criteria

The study aimed to compare the effects of propofol, ciprofol, and remimazolam in pairs during endoscopic surgery. The RCTs included were required to have at least one outcome of interest. Only adults (age > 18 years) were included.

Exclusion criteria

RCTs were excluded according to the following criteria: 1) studies comparing combinations of any two of propofol, ciprofol, and remimazolam with other drugs; 2) projects that had not yet begun; 3) studies that did not involve endoscopic surgery; and 4) studies that did not assess the safety and efficacy of the three drugs.

Outcomes

Primary outcomes were respiratory/cardiovascular adverse events. Additional outcomes were as follows: induction success rate, drug-related adverse effects(injection pain, postoperative nausea and vomiting), and patient anesthesia satisfaction time to induction, time to full consciousness. For the definitions of the above indicators, please refer to Supplementary Files.

Data extraction

Data were extracted by two investigators using a pre-designed form, including the following: author names, type of procedure, study inclusion and exclusion criteria, number of patients enrolled and randomized, baseline demographic characteristics, ASA classification, induction dose of each drug, and outcome data. Only trials with extractable data were included. No additional information was requested from the study authors.

Quality and risk of bias assessment

The risk of bias (RoB) in randomized controlled trials was assessed using the Cochrane Collaboration tool to evaluate the quality of the included trials [20]. Disagreements in the assessment were resolved through discussion and consensus. The Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach was used to assess the certainty of evidence for each outcome [21].

Data synthesis and analysis

For continuous outcomes, including patient satisfaction, induction time, and time to full consciousness, the mean difference (MD) corresponding to 95% credible intervals (CrIs) were calculated for the outcomes. Since the patient satisfaction measurement scales were not identical, the linear transformation was applied to convert the measurements to outcomes within the range of 0–10, and MD was calculated to estimate the overall study outcomes. All eligible RCTs were primarily analyzed. For dichotomous outcomes, including hypotension, bradycardia, and injection pain, relative risks (RR) and corresponding 95% CrIs were calculated. Statistical heterogeneity between trials was assessed using visual inspection of forest plots and the I² statistic. For continuous outcomes reported as median and interquartile range, the method of Shi et al. [22] was used to assess whether they conformed to a normal distribution. If they did, they were transformed into means and standard deviations according to this method [23, 24]. Data that did not meet the transformation requirements were excluded. No interpolation was performed on the data.

The feasibility of performing network meta-analyses for each outcome was assessed. When appropriate, Bayesian random-effects models were used to calculate direct effect estimates for different comparisons using the Markov Chain Monte Carlo (MCMC) method. The heterogeneity of the random-effects model was evaluated using the I² statistic. Convergence diagnostic plots, trajectory plots, and density plots were used to assess the stability and reliability of the MCMC simulation results. Cumulative probability ranking plots were generated, followed by the use of the Surface Under the Cumulative Ranking Curve (SUCRA) to explain the relative effectiveness of different interventions. Consistency assumptions were verified using node splitting [25].

A network diagram with nodes and lines was constructed to represent different interventions, where the size of the nodes represented the population size, and the thickness of the lines between the nodes indicated the number of studies.

The network structure was visualized using STATA software version 17.0. All analyses were performed using RStudio version 2024.04.2 + 764, with the gemtc package version 1.0.2, which interfaces with the rjags package version 4.16 to execute MCMC simulations.

Results

Search results and study characteristics

A total of 83 records were identified, of which 56 were reviewed in full text. A total of 14 studies were excluded: six were excluded due to incompatibility with the study objectives [26,27,28,29,30,31], three were research protocols [32,33,34], four were registrations only, and one was unrelated to endoscopic surgery [35]. Finally, 42 RCTs were finally included, with a total sample size of n = 10,540. Among these, 31 studies compared propofol with remimazolam [36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66], 10 compared ciprofol with propofol [67,68,69,70,71,72,73,74,75,76], and one was a comparison of the three drugs (Fig. 1) [77]. The characteristics of the included trials were detailed in the table. Of these 42 studies, 31 were assessed as having a low RoB [36, 37, 39,40,41, 44,45,46,47,48,49,50, 69, 78], two as having a high RoB [51, 62], and the remainder as unclear RoB [38, 42, 43, 63, 65, 67, 68, 70, 73] (Fig. 2).

Fig. 1

Study Flow Chart

Fig. 2

Risk of Bias Summary Chart

Primary outcomes

Respiratory adverse events

Respiratory adverse events, such as apnea, hypoxemia, and respiratory depression, were defined by the included studies. Figure 3A presents a forest plot of these findings. The forest plot showed that the RRs for respiratory adverse events for ciprofol and remimazolam, using propofol as the baseline, were 0.4800 (95% CrI [0.3200, 0.7000], moderate certainty) and 0.3600 (95% CrI [0.2800, 0.4600], high certainty), respectively. The p-value for the indirect comparison of remimazolam to ciprofol was greater than 0.05, indicating a high level of credibility in the network analysis results. The RR of remimazolam relative to ciprofol was 0.7600 (95% CrI [0.4800, 1.2000], moderate certainty), indicating no significant difference in respiratory adverse events. The probability ranking plot showed that remimazolam ranked first, with a probability of 0.8817 and an SUCRA value of 0.9408, whereas propofol ranked third, with a probability of 0.9995 and an SUCRA value of 0.0003 (Fig. 4).

Fig. 3

A Forest Plot of Respiratory Adverse Events. B Forest Plot of Cardiac Adverse Events

Fig. 4

A Cumulative Probability Distribution Diagram of Respiratory Adverse Events. B Cumulative Probability Distribution Diagram of Cardiac Adverse Events

Cardiac adverse events

Adverse reactions affecting the cardiovascular system included hypotension and bradycardia. The network diagram for this outcome is available in Fig. 5. The risk ratio (RR) for ciprofol relative to propofol was 0.7600 (95% CrI [0.5300, 1.1000], moderate certainty), whereas that for remimazolam was 0.4400 (95% CrI [0.3500, 0.5500], moderate certainty). The network analysis indicated an RR of 0.5800 (95% CrI [0.3800, 0.8700], moderate certainty) for remimazolam compared with ciprofol, supported by higher-quality evidence (Fig. 3B). The probability that remimazolam would result in the fewest adverse events affecting the cardiovascular system was 0.9951. Propofol demonstrated a third-ranked probability of 0.9379 for inducing adverse events affecting the cardiovascular system, exceeding that for ciprofol and remimazolam (Fig. 4). The SUCRA values for remimazolam and ciprofol were 0.9976 and 0.4714, respectively.

Fig. 5

Network map for Cardiac adverse events for x-node analysis. The size of the node corresponds to the number of patients randomised to that intervention. The thickness of the line and the associated numbers correspond to the number of studies comparing the two linked interventions

Secondery outcomes

Induction time and sedation recovery time

Ciprofol demonstrated a relatively longer induction time than propofol (mean difference [MD] 0.2039 min longer, 95% CrI [−0.2177, 0.6284], low certainty). Similarly, remimazolam had a longer induction time compared to propofol (MD 0.1596 min longer, 95% CrI [−0.1428, 0.4576], high certainty). The mean difference (MD) for remimazolam was −0.0450 (95% CrI [−0.5700, 0.4700], low certainty) when compared to ciprofol, indicating a lower effect size in the indirect comparison of remimazolam with ciprofol (Fig. 3A). The probability that propofol had the fastest induction time was 0.7251, which was greater than 0.1509 for ciprofol and 0.1240 for remimazolam. The probability of ciprofol having the longest induction time among the three drugs was 0.5606, and the cumulative probability ranking plot is shown in Fig. 4. The SUCRA values for ciprofol, propofol, and remimazolam were 0.2951, 0.8504, and 0.3545, respectively, further supporting the cumulative probability results.

When comparing sedation recovery time, propofol had a significantly shorter time to full consciousness than ciprofol and remimazolam (MD reduction of 1.0271 min, 95% CrI [–0.5486, 2.6190], low certainty; MD reduction of 0.2382 min, 95% CrI [−0.8651, 1.3340], very low certainty). The indirect mean difference (MD) between remimazolam and ciprofol was 0.7900, with a 95% CrI of [−1.1000, 2.7000], with very low certainty (Fig. 3B). Propofol had the highest probability of the shortest awakening time among the three drugs, at 0.6088. The probability of ciprofol having the shortest awakening time was 0.0795, and remimazolam ranked first with a probability of 0.3117. Additionally, ciprofol had a higher probability of ranking third compared to both ciprofol and propofol, with a probability of 0.7834 (Fig. 4). The SUCRA values for the three drugs were 0.7885, 0.5634, and 0.1481, in descending order, indicating that propofol ranked first for the shortest time to full consciousness, while ciprofol ranked last.

Injection pain

Injection pain is a common adverse event associated with propofol. The probabilities of the three drugs, ciprofol, propofol, and remimazolam, causing the least injection pain were ranked as follows: remimazolam first at 0.5046, ciprofol second at 0.4955, and propofol third at 0. (Supplementary file). The SUCRA values ​​of ciprofol, propofol, and remimazolam were 0.747725, 0.000000, and 0.752275, respectively. The RR for ciprofol versus propofol was 0.0550 (95% CrI [0.0130, 0.1600], low certainty). The RR for remimazolam versus propofol was 0.0540 (95% CrI [0.0190, 0.1200], high certainty). The RR for remimazolam versus ciprofol was 0.9900 (95% CrI [0.2400, 4.300], low certainty) (Supplementary file).

Gastrointestinal adverse events

The included studies defined gastrointestinal adverse events as nausea and vomiting. Ciprofol ranked first with the highest probability of causing the fewest nausea and vomiting events (0.8281) and a SUCRA value of 0.8668. Remimazolam ranked third with the greatest probability of 0.4940 and a SUCRA value of 0.3058 (Supplementary file). Using propofol as the reference, the RRs and 95% CrI for ciprofol and remimazolam were 0.7300 (95% CrI [0.4000, 1.3000], moderate certainty) and 1.0000 (95% CrI [0.7800, 1.5000], moderate certainty), respectively. The indirect RR for remimazolam relative to ciprofol was 1.400 (95% CrI [0.7700, 3.0000], moderate certainty) (Supplementary file).

Patient satisfaction

In a comparison of patient satisfaction, the MD for ciprofol versus propofol was 0.3610 (95% CrI [−0.4363, 1.1810], low certainty). The MD for remimazolam versus propofol was 0.3984 (95% CrI [−0.2070, 1.0240], very low certainty), whereas the indirect MD for remimazolam versus ciprofol was 0.0370 (95% CrI [−0.9800, 1.1000]), very low certainty (Supplementary file). The rank order of cumulative probability for patient satisfaction is shown in Supplementary file, with remimazolam having the greatest probability of ranking first at 0.5230. Propofol ranked third with a probability of 0.7648, which was higher than that for ciprofol (0.1609) and remimazolam (0.0743). Remimazolam ranked second with a probability of 0.4027, exceeding the second-ranking probabilities of both ciprofol and propofol (Supplementary file). The SUCRA values for propofol, ciprofol, and remimazolam were 0.1258, 0.6498, and 0.7243, respectively.

Other adverse events

Other common adverse reactions include body movements, dizziness, and cough. The three drugs, ciprofol, propofol, and remimazolam, had the highest safety-ranked probabilities of 0.7541, 0.1343, and 0.1116, respectively. The third-ranked probability for remimazolam was the highest at 0.6115, indicating a higher probability of body movements compared to the other two (Fig. 4). Compared with propofol, ciprofol and remimazolam had relative risks (RRs) of 0.7000 (95% CrI: 0.3000–1.600) and 1.100 (95% CrI: 0.7100–1.9000), respectively, with moderate certainty for ciprofol and very low certainty for remimazolam. The indirect relative risk (RR) of body movement for remimazolam compared with ciprofol was 1.6000, with a 95% CrI of 0.5700–4.7000, and the evidence was rated as very low certainty (Supplementary file).

According to the probability plot, the probability of remimazolam being safer than the other two drugs in causing dizziness is the highest, at 0.7900. The probability of propofol causing postoperative dizziness is 0.6677, indicating that compared to the other two drugs, propofol is more likely to cause dizziness (Supplementary file). The indirect comparison p-value for remimazolam versus ciprofol is 0.9031, suggesting that the result obtained through network analysis is reliable. The specific value was RR 0.6800 with a 95% CrI of [0.2400, 1.800], which is rated as low certainty (Supplementary file).

When comparing the adverse effects of cough, ciprofol demonstrated the highest safety probability of 0.6091, which was greater than those of propofol at 0.2621 and remimazolam at 0.1289. Remimazolam was ranked third, indicating that it was more likely to induce cough compared to both ciprofol and propofol, with a probability value of 0.6339 (Supplementary file). Compared with propofol, the relative risks (RRs) of ciprofol and remimazolam were 0.7000 (95% CrI: 0.0910–5.4000) and 1.6000 (95% CrI: 0.4500–7.6000), respectively, with the evidence graded as high certainty for ciprofol and very low certainty for remimazolam. The figure showed that remimazolam had an RR of 2.2000 for the occurrence of cough compared to ciprofol, with a 95% CrI ranging from 0.2200 to 31.00. The certainty of the evidence was rated as very low (Supplementary file).

Discussion

This systematic review and network meta-analysis summarized the strengths and weaknesses of the three sedative drugs. Specifically, propofol was found to have the shortest induction and awakening times, although it was associated with a higher incidence of cardiovascular and respiratory adverse effects and was linked to postoperative dizziness, gastrointestinal adverse effects, and lower patient satisfaction. Both remimazolam and ciprofol exhibited a higher safety profile than propofol. They significantly reduced the incidence of propofol injection pain and were associated with a lower incidence of cardiovascular and respiratory adverse effects. Remimazolam was associated with fewer cardiovascular and respiratory events than ciprofol and significantly reduced the incidence of bradycardia. However, remimazolam was linked to an increased incidence of cough and body movements.

The duration of patient recovery from PSA is significant from a resource utilization perspective, as close monitoring is required until full recovery is achieved. According to the results of the network meta-analysis, propofol demonstrated superior performance in induction and postoperative awakening times, having the shortest induction time and enabling a faster awakening process. However, these differences were not statistically significant, with minimal differences observed between groups. Therefore, selecting propofol solely based on time and resource considerations may not be advisable. The results of this study indicate that remimazolam has the longest induction time. A longer induction time may require extra time for patient preparation and monitoring. In clinical practice, it is essential to continuously assess the level of consciousness in patients undergoing PSA. This ongoing evaluation is important for monitoring the patient's physiological status and ensuring that anesthesia is adequate [79].

Based on safety considerations during patient examinations, remimazolam may be the preferred choice for patients with compromised airway conditions, those undergoing respiratory endoscopy, and patients requiring sedation with cardiovascular disorders (e.g., critically ill patients). The current meta-analysis also concluded that propofol has a higher risk of respiratory depression compared to remimazolam [18]. Propofol, being strongly linked to respiratory and cardiovascular events, should be avoided in patients with respiratory or cardiovascular complications [80, 81]. These observations may be due to cardiorespiratory inhibition caused by the effect of propofol on central chemoreceptor sensitivity [82]. Although benzodiazepines may cause hypotension, combining sedative drugs has been shown to reduce single-drug doses, thereby potentially mitigating adverse effects [1].

In terms of gastrointestinal adverse reactions, ciprofol demonstrated the lowest incidence, followed by remimazolam, while propofol exhibited the highest rate. Based on this analysis, propofol may be considered more suitable for PSA in patients undergoing gastrointestinal endoscopy or those with pre-existing gastrointestinal conditions. However, it is important to note that this comparison did not reach statistical significance, underscoring the necessity for further research to identify the optimal sedative for gastrointestinal endoscopic procedures.

The findings also revealed that both remimazolam and ciprofol significantly reduced the incidence of injection pain. In particular, our results indicated that remimazolam had a significantly lower incidence of injection pain compared to propofol with high certainty. This may be due to the short metabolic half-life of remimazolam, reduced local and in vivo accumulation, and decreased release of active metabolites [83]. Propofol injection pain (PIP) is a long-standing issue [84, 85]. It can cause discomfort, heightened tension, anxiety, and other unpleasant experiences, as well as induce body movements and symptoms that hinder the successful completion of endoscopy [86]. No significant differences were found between remimazolam and propofol in indirect comparisons. Therefore, both drugs may be suitable options for minimizing injection pain. It is important to acknowledge that the decrease in injection pain can significantly enhance the patient experience during anesthesia induction. However, it should be noted that this advantage was not seen at higher doses of ciprofol, suggesting that the effect is dose-dependent [79].

However, remimazolam may lead to higher rates of cough and body movements. Our research indicated that when comparing the adverse effects of cough, the RR of ciprofol was 0.70 (95% CrI: 0.091–5.4) compared with propofol, with the evidence graded as high certainty. Our study was the first meta-analysis to observe that ciprofol had a lower incidence of cough compared to propofol. Due to the gastroscopy or bronchoscopy will be performed in the throat area, sudden unexpected cough may cause damage to unwanted tissues or nerves, which may be difficult to predict[87], it is crucial to constantly monitor any reactions of the patient during the PSA. However, propofol exhibited the highest incidence of neurological adverse effects, such as post-examination dizziness. These differences were not statistically significant, and no studies have reported serious neurological complications associated with any of the three drugs. As a benzodiazepine, greater consideration should be given to the possibility that remimazolam may induce paradoxical reactions [88]. A study on dream induction reported that two patients experienced unpleasant dreams with remimazolam use[77]. While this result does not significantly affect the overall comparison, attention should still be given to patient experience and neurological symptoms, particularly in patients with pre-existing conditions. Reducing benzodiazepine use may improve sedation comfort in such cases.

In terms of patient satisfaction, propofol achieved the lowest scores. Although these differences did not reach statistical significance, they may be related to the higher incidence of propofol-induced adverse reactions, such as injection pain, gastrointestinal discomfort, and dizziness, potentially contributing to lower patient satisfaction and a less favorable sedation experience.

All in all, based on our findings, we propose the following clinical recommendations: For high-risk respiratory patients (e.g., COPD, OSA), remimazolam is preferred due to its significantly lower risk of respiratory depression (RR = 0.36 vs propofol; Fig. 6B), This result aligns with current European guidelines[89], which emphasize the need for careful evaluation of cardiovascular and respiratory risk factors before elective procedures. Similarly, for cardiovascular-compromised patients (e.g., heart failure, hypotension), remimazolam demonstrates superior safety with a 56% reduction in bradycardia risk (RR = 0.44). Therefore, perhaps in staffing models where an anesthesiologist is not available, remimazolam might be the safest option. In gastrointestinal endoscopic procedures, ciprofol emerges as the optimal choice given its favorable gastrointestinal tolerance profile (SUCRA = 0.8475). For injection pain-sensitive patients, either remimazolam or ciprofol is recommended, both showing dramatically reduced incidence compared to propofol (RR = 0.045). These evidence-based selections should be tailored to individual patient characteristics and procedural requirements. Based on patient-specific characteristics, we recommend the following medication protocols: For elderly patients (> 65 years), remimazolam (0.1–0.2 mg/kg) is recommended with titrated administration to avoid drug accumulation[56]. Obese patients (BMI ≥ 30) should receive ciprofol (0.4 mg/kg) with halved initial doses and close monitoring for respiratory depression[69]. For patients with hepatic impairment, remimazolam requires no dose adjustment due to its metabolism by tissue esterases.

However, the interpretation of our findings must consider several key factors: 1) Protocol heterogeneity: Variability in sedation protocols (e.g., adjuvant opioids, monitoring standards) may explain the wide CrIs for recovery times (Fig. 3B). 2) Dose–response ambiguity: While remimazolam reduced injection pain (RR = 0.045), dose-dependent effects were rarely analyzed—higher doses may attenuate this benefit. 3) Population generalizability: Exclusion of high-risk subgroups (e.g., ASA III-IV) limits applicability to real-world endoscopic settings. 4) Indirect comparison limitations: The single direct trial comparing remimazolam and ciprofol necessitates cautious interpretation of their relative rankings. Future research directions should include: RCTs directly comparing remimazolam and ciprofol (currently only 1 study [78]); Dose optimization studies for special populations (particularly pediatric, obese patients); Comprehensive evaluation of long-term cognitive impacts.

Limitations of this study include the low incidence of adverse reactions and the difficulty in including special groups, such as underage and obese patients, due to an insufficient sample size. This limits generalizability to these populations. Future studies should prioritize inclusion of diverse cohorts to validate our findings. Secondly, although sedation protocols standardized MOAA/S ≤ 3 as the target depth, variability in drug dosing and monitoring practices across studies may have influenced complication rates. While stratified analyses showed no significant interaction between sedation depth and adverse events (p > 0.05), residual confounding cannot be entirely excluded. The use of different criteria for defining adverse reactions may result in biased statistical outcomes. Therefore, more case reports and scoping reviews are needed to explore and standardize the criteria for defining adverse reactions. Thirdly, while indirect comparisons of remimazolam and ciprofol could reduce heterogeneity, further studies are necessary to directly compare the sedative effects of both drugs in endoscopy. Forthly, our protocol excluded studies comparing combination regimens (e.g., remimazolam + opioids). Consequently, we could not evaluate synergistic effects or safety profiles of polypharmacy, which is common in clinical practice. This gap underscores the need for future RCTs to assess combination strategies. Fively, variations in dosing regimens (e.g., remimazolam 0.1–0.4 mg/kg) across trials may have introduced heterogeneity. We attempted to mitigate this by pooling standardized mean doses, but residual variability in pharmacokinetic responses remains a concern. Last but not least, the safety and efficacy of sedatives are sometimes dose-dependent, and future research requires larger sample sizes and more dose subgroup analyses.

Conclusion

In summary, this network meta-analysis demonstrates that remimazolam significantly reduces cardiovascular risk by 56% compared to propofol while maintaining comparable recovery times, making it a safer alternative for high-risk patients. Furthermore, ciprofol exhibits superior gastrointestinal safety, positioning it as the optimal choice for prolonged endoscopic procedures requiring extended sedation. Importantly, these findings underscore that drug selection should be guided by individualized patient risk profiles—prioritizing remimazolam for cardiovascular/respiratory compromise, ciprofol for gastrointestinal-focused interventions, and propofol for rapid-turnover settings in low-risk populations.