This prospective, sequential allocation, dose-finding trial recruited 80 candidates aged >65 years between May 2022 and June 2022 in the First Affiliated Hospital of Nanchang University in Nanchang, China. The study was registered with the China Clinical Trials Center (ChiCTR2200055709) and approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University (AF-SG-03-2.0). Written informed consent was obtained from the patients or authorized individuals.

The inclusion criteria were as follows: 1) Patients must be > 65-years-old and candidates for elective surgery; 2) Physical status of ASA I–III (Mayhew et al., 2019); 3) Body mass index (BMI) 20–25 kg/m²; 4) Patients who intend to undergo single-lumen endotracheal tube surgery under general anesthesia; 5) Informed consent form signed by the patient or authorized person. The exclusion criteria were as follows: 1) First intubation failed or the patient withdrew from the trial; 2) Patient’s alertness/sedation score was more significant than 1 point post-remimazolam tosylate injection after 3 min; 3) Patients intermittently used benzodiazepines shortly before the surgery; 4) Patients who had a history of hypertension and baseline systolic blood pressure (SBP) > 180 mmHg; 5) Patients with severe obstructive (chronic obstructive pulmonary disease gold ≥3) or restrictive (forced vital capacity <80% of predicted value) pulmonary disease or ischemic heart disease; 6) Test drug allergy; 7) Nervous system disease or inability to communicate appropriately.

The patient’s preoperative fasting status was confirmed in compliance with ASA guidelines. No premedication was used. An intravenous (i.v.) cannula was inserted in the pre-anesthesia room, and 6–8 mL/kg of lactated Ringer’s solution was infused. Radial artery puncture and catheterization were performed under local anesthesia by an anesthetist. Standard monitoring with continuous electrocardiography, invasive arterial BP monitoring, and pulse oximetry (S_pO₂) was used for frail and non-frail groups during the trial.

Elderly patients >65-years-old were assessed for weakness on the clinical frailty scale (CFS) (McIsaac et al., 2020b) the day before the operation. The CFS is a 9-point global rating scale; the score is assigned based on the assessment of mobility, energy levels, physical activity, and function. The primary study physician conducted all frailty assessments. Subjects with CFS scores <4 were included in the non-frail group, and those with scores ≥4 comprised the frail group.

Before administration, 2 mg/mL remimazolam tosylate (Jiangsu Hengrui Pharmaceutical Co., Ltd, Jiangsu, China) was prepared by diluting 36 mg remimazolam tosylate in 0.9% saline (18 mL). The study dose (6.63–12.48 mL) was further diluted in saline to a total volume of 15 mL. A bolus of targeting remimazolam tosylate was administered via an infusion pump (Shenzhen Mindray Bio-Medical Electronics Co., Ltd, Shenzhen, China) for 1 min, followed immediately by a flush of saline (5 mL). Dose sequencing was based on the biased coin design (BCD) method, with an initial dose of 0.3 mg/kg, and the dose gradient was 0.02 mg/kg for frail and non-frail groups. Dai et al. (Dai et al., 2021) showed a 94% success rate of anesthesia induction with 0.3 mg/kg remimazolam, deeming it a safe dose in clinical practice; hence, 0.3 mg/kg remimazolam was selected as the starting dose for both groups of patients in the trial. If the cardiovascular response to endotracheal intubation was positive in the previous patient, the next patient received an increasing dose gradient of remimazolam, while if the response was negative, the next patient had a 5% random probability of one dose gradient of remimazolam reduced and a 95% probability of the same dosage as the previous patient.

Upon arrival in the operation room and after preoxygenation, anesthesia was induced by targeting remimazolam tosylate for sedation. When the observers’ assessment of alertness/sedation (OAA/S) score (Chernik et al., 1990) was ≤1 point (unresponsive to mild shaking or prodding) or the bispectral index (BIS VISTATM, GE Healthcare, United States) was ≤60 after the patient’s consciousness disappeared, sedation was considered successful. Then, 4 μg/kg fentanyl (Yichang Humanwell Pharmaceutical Co., Ltd, Yichang, China) and 0.15 mg cisatracurium (Hangzhou Hongyou Pharmaceutical Technology Co., Ltd, Hangzhou, China) were administered for complete anesthesia induction. An experienced anesthetist used a video laryngoscope to intubate the patient if there was no response to train-of-four (TOF) stimulation 3 min after the infusion of fentanyl and cisatracurium. SBP, diastolic arterial BP (DBP), and HR were recorded by another experienced anesthesiologist, blinded to the remimazolam dose, at the following time points: baseline vital signs before induction (T0); successful sedation (T1); immediately before intubation (T2); immediately after intubation (T3); 30 s, 1 min, and 2 min post-intubation (T4, T5, and T6, respectively), using sustainable monitoring. Subsequently, the expected operation was performed. At the end of the procedure, the subjects were admitted to the post-anesthesia care unit (PACU) for resuscitation, wherein all subjects were assessed for awareness by the anesthesia nurse. In this study, BIS monitored the depth of anesthesia, and TOF monitored the degree of muscle relaxation until the end of endotracheal intubation. The detailed sequence of this trial is shown in Figure 1.

If the absolute HR was <45 bpm or the baseline mean arterial pressure (MAP) was decreased by >30% (Wesselink et al., 2015), immediate treatment was given with 0.3 mg atropine or 4 µg norepinephrine. The symptomatic treatment can be reused if necessary. If the MAP or HR is higher than the baseline value by >30%, it should be carefully observed for 1 min. If there is no remission, anesthesia or urapidil 5 mg or esmolol 0.5 mg/kg is deepened. These phenomena were regarded as adverse complications and recorded in the current study.

The positive responses were defined as MAPmax or HRmax ≥20% of the fundamental value within 2 min after intubation. Negative responses were defined as either the variation of MAPmax and HRmax during intubation at < 20% of the baseline value. Failure to sedate with the target dose of remimazolam tosylate was also considered a positive reaction, and remimazolam tosylate was administered further to deepen anesthesia; subsequently, the patient was withdrawn from the trial.

The BCD method was used to determine the adequate dose level of remimazolam tosylate at Г = 0.95 quantiles for frail and non-frail senile patients. Г is the level of drug effect determined by the study purpose. Since ED95 is desired, Г = 0.95 is assigned to produce a response in 100×Г = 95% of the target population (Pace and Stylianou, 2007). The K-sequential dose level was selected to increase or decrease the dose by 0.02 mg/kg between classes, and consecutive patients were exposed to one of the successive K-dose levels with an initial dose of 0.3 mg/kg. The recommended dose for the current study (Dai et al., 2021) was implemented at investigator’s discretion. The research dose in the current study was 0.26, 0.28, 0.30, 0.32, and 0.34 mg/kg, respectively. If the cardiovascular response to endotracheal intubation was positive in the previous patient, the next patient received an increased dose gradient of remimazolam, while in the case of a negative response, the next patient had a 5% random probability of reduced remimazolam by one dose gradient and a 95% probability of maintaining the dose level same as the previous patient.

In the BCD study, a priori calculation of the sample size was not possible due to the non-independence (the dose administered to the next patient was dependent on the intubation response of the previous patient) and the unknown dose distribution. The parametric estimates of all target doses were balanced in approximately 20 subjects and stabilized in 40 subjects, as described previously (Stylianou and Flournoy, 2002). Therefore, 40 subjects were recruited for each group in this study.

Isotonic regression with R Foundation was used to calculate the ED₉₅ estimate of û3 (Pace and Stylianou, 2007), which is the linearly interpolated dose between P* κ and P* κ+1 at the probability of Г = 0.95. P* κand P* κ+1indicated adjusted response rates at doses χ _κ and χ _κ+1, respectively. The adjusted response rate is calculated monotonically using the pool-adjacent-violators algorithm (PAVA), following which the response rate was recalculated by pooling the adjacent increasing and decreasing pairs. Isotonic regression is an adjustment of regression that limits the probability of a sequential dose response to be monotonic. The 95% CI for ED₉₅ was calculated using a bias-corrected percentile derived by bootstrapping with a resampling size of 40, a repeat number of 2000, and a target Г of 0.95. The above data were analyzed using R Foundation.

SPSS 26.0 software (IBM SPSS Inc, Armonk, NY, United States of America) was used for data analysis. The mean [standard deviation (SD)] of measurement data of normal distribution was computed. The comparison between the two groups was performed using independent sample t-test. In the case of non-normal data distribution, median (interquartile range, IQRs) was used, and the comparison between the two groups was conducted using a non-parametric Mann–Whitney U test with two independent samples. The enumeration data were expressed as absolute numbers, and the comparisons between groups were performed using the χ² test. Fisher’s exact probability test was applied where appropriate. Baseline BP and HR were compared between the two groups using unpaired Student’s t-test, and two-factor analysis of variance was used for repeated measures data, followed by Sidak’s test (p < 0.05) for multiple comparisons. p < 0.05 indicated a statistically significant difference.

A total of 96 patients undergoing elective surgery were recruited in this study, of which 16 declined the trial as their independent choice. The remaining 80 subjects (40 frail and 40 non-frail) completed the study (Figure 2). The analysis of demographic characteristics of the frail and non-frail groups did not show any significant differences in age, gender composition, BMI, and ASA classification (Table 1).

The positive or negative reaction of each subject to the specified dose of remimazolam tosylate in the two groups is shown in the standard plot, with the x-axis as the sequence of subjects and the y-axis as the dose of each subject (Figure 3). The specific response doses (y-axis) of remimazolam tosylate were 0.30, 0.32, and 0.34 mg/kg in the non-frail subjects compared to 0.26, 0.28, and 0.30 mg/kg in frail subjects, respectively. One candidate reacted positively at 0.30 mg/kg, and three candidates were positive to 0.32 mg/kg in the non-frail patient group (Figure 3A), while in the frail patient group, one was positive to 0.26 mg/kg and the other to 0.28 mg/kg dose (Figure 3B).

The number of subjects tested, the number of negative subjects, the observed response rates, and the PAVA-adjusted response rates for both groups at different target doses are presented in Table 2. In the non-frail group, the observed response rate was 0.927, 0.727, and 1.000 at the assigned doses of 0.30, 0.32, and 0.34 mg/kg of remimazolam tosylate, while the PAVA-adjusted response rate was 0.870, 0.920, and 1.000, respectively. The observed response rate decreased with increasing doses, but the response rate after PAVA adjustment did not decrease with increments in the dose. The estimated ED₉₅ using isotonic regression was 0.331 mg/kg (95% CI: 0.272–0.472) in the non-frail group and 0.297 mg/kg (95% CI:0.231–0.451) in the frail group. Overlapping CIs did not show any significant difference in ED₉₅ between the two groups.

DBP was significantly higher in the non-frail group than in the frail group at 1 and 2 min post-intubation (p = 0.006 and 0.010, respectively). However, no significant differences were detected in the SBP and HR between the two groups (Figure 4).

The present study has several limitations. First, we did not draw blood to detect changes in the blood stress indicators during endotracheal intubation. Second, the fluctuation in BP and HR within 20% in the intubation reaction was defined as successful inhibition of intubation. Although there was only a slight fluctuation, it may eventually lead to a large ED₉₅. Finally, only BP and HR changes during endotracheal intubation were monitored, and no cardiac output changes were observed using cardiac function monitoring.