This study was approved by the Medical Ethics Committee of the General Hospital of the Northern Theater of the Chinese People’s Liberation Army [No. Y (2022) 092] and registered with the China Clinical Trials Center (trial registration number: ChiCTR2300067317). The trial was officially conducted in the Obstetrics and Delivery Center in accordance with the Declaration of Helsinki, and a total of 45 parturients who met the criteria were eventually enrolled for clinical observation. All parturients and their families signed an informed consent form. The inclusion criteria were as follows: (1) singleton, fetal head position and full term (37–42 weeks of gestation), (2) age 20–35 years, height 150–170 cm, body mass index (BMI) < 35 kg/m2, American Society of Anesthesiologists (ASA) Class II, (3) uterine orifice dilation of 2–3 cm, regular uterine contractions assessed by the obstetrician and midwife, (4) eligible for vaginal delivery. The maternal with any of the following conditions during the trial would be excluded and discontinued from observation: (1) unable to cooperate with the epidural puncture and visual analog score (VAS) evaluation, (2) VAS < 6 before analgesia, (3) converted to cesarean delivery for various reasons during labor analgesia, (4) allergic to local anesthetics or opioids, (5) contraindications to intervertebral block, (6) pregnancy complications such as eclampsia and gestational diabetes mellitus or other serious organic diseases, (7) rapid cervical orifice change (cervical orifice dilation ≥ 5 cm) within 45 min after epidural analgesia, (8) withdrawal midway and those with suspicious results.

When the cervical orifice was enlarged to 2–3 cm, the parturient was placed in the delivery room, and connected to a maternal fetal monitor (SRF618K9, Guangzhou Sunray Medical Apparatus, Guangzhou, China). The electrocardiogram (ECG) was monitored and blood pressure (BP), heart rate (HR), pulse oxygen saturation (SPO2), fetal heart rate (FHR), uterine pressure and temperature were recorded. Upper limb venous access was opened, and sodium lactate Ringer’s solution was given for fluid replacement. The parturient was instructed to a left-lying position and an epidural puncture was performed in the L2-3 space. The catheter was placed into the epidural space with 4 cm towards the head, and 3 mL of 1% lidocaine was given as a test dose, then the catheter was properly secured after the test dose was safe. The maternal was excluded if the dura was punctured or the epidural tube was failed to insert. An epidural catheter was connected to an electronic analgesic pump and programmed intermittent epidural injection (PIEB) combined with patient-controlled epidural analgesia (PCEA) mode was selected. According to the guidelines [3] and clinical experience, when 10 mL of the study solution was given as the pulse volume in this experiment, the pulse interval was set to 50 min to minimize the effect of maternal motor block, and the other parameters were set to 3 mL of background dose, 8 mL of maternal self-control dose, and 20 min of lockout time.

The VAS evaluation (0 for no pain, 10 for intolerable pain) and leg lift test were performed. The parturients were asked to raise both legs at a 45° angle and kept straight for 4 s while closing the eyes. The latest Bromage score (1—complete motor block; 2—almost complete motor block, patient can only move the foot; 3—partial motor block, patient has knee motion; 4—detectable hip flexion and extension weakness, patient can lift the legs but cannot maintain them; 5—no hip flexion and extension weakness, patient can lift the legs for more than 10 s; 6—no weakness) was used to assess the degree of motor block and motor intensity in the lower extremity. The degree of upper sensory block was assessed by the movement of an alcohol-impregnated swab from the level of S2 toward the head. In the lower extremity, the plane of block was assessed by stimulation of the inguinal fold (L1), anterior thigh (L2), medial knee (L3), medial malleolus (L4), dorsoventral area between the big toe and second toe (L5), lateral heel (S1), and medial popliteal fossa (S2). All of the above indicators were recorded at 15, 30, and 45 min after epidural analgesia and fully dilation of cervical orifice.

Adverse events during labor analgesia were addressed promptly. In case of hypotension (systolic blood pressure dropping more than 20% of the basal value or less than 90 mmHg), the parturient was treated in lateral position or with intravenous fluids, and norepinephrine or ephedrine could be given if there was no significant relief. The maternal bradycardia (HR < 50 bpm) was treated with atropine. When the fetal bradycardia (FHR < 110 bpm for more than 10 min) occurred, the parturient was examined and evaluated by the obstetrician and midwife. If there was no significant improvement after changing the position or administering oxygen via high-flow nasal cannula, emergency cesarean delivery was performed to ensure fetal life. The VAS 30 min after epidural labor analgesia less than or equal to 3 was considered effective. In cases of incomplete analgesia, the possible causes were considered and appropriate measures were taken according to the manifestations of incomplete analgesia. For example, if the analgesic plane was adequate (T10-S4) but the analgesic intensity was insufficient, the concentration of analgesic could be increased. For bilateral blocks and insufficient analgesic planes, a large volume (5–15 mL) of low-concentration local anesthetic was used for planar diffusion. If maternal labor pain was relieved, the next patient would receive a graded increase in alfentanil concentration. Otherwise, the remedy was considered ineffective and marked as suspicious, the case was excluded from the study, and the next woman also adapted this alfentanil concentration.

The type of this study was single-center, prospective, and double blind. Epidural puncture and catheterization, preparation of analgesic solution and assessment of analgesic effect were performed by anesthesiologists who were unaware of the study protocol. The study used BCD-UDM, which was the optimal upper and lower design in the general category, allowed accurate estimation of target doses for low or high quantiles, and had the advantages of small sample size and reasonably simple study performance. Based on the results of the pretest, the epidural concentration of alfentanil was set at 3 μg/mL in the first parturient, and then the concentration in the next parturient was depended on the block effect in the previous parturient. If the block failed (VAS 30 min after epidural labor analgesia more than 3 was considered incomplete analgesia), the concentration of alfentanil was increased in a gradient of 0.5 μg/mL for the next patient. If the block was successful (VAS 30 min after epidural labor analgesia less than or equal to 3 was considered effective analgesia), there was an 11% (b = 0.11) probability of a one-unit decrease (0.5 μg/mL) and an 89% (1− b = 0.89) probability of no change for the next parturient according to the BCD-UDM method. The sequential assignment of the biased coins was achieved by a computer-generated list of random responses, which was prepared by a biostatistician with excel. This random list was used by a research assistant to develop analgesic protocol for the next parturient based on the observer-recorded response of the previous parturient. The protocol was delivered in a closed envelope to the researcher who prepared the analgesic solution.

The VAS 30 min after epidural analgesia was used as the primary observation, and the primary outcomes were EC90 and 95% confidence intervals (CIs) for alfentanil. Secondary outcomes were recoded, including the values of the mentioned indicator at five time points, duration of analgesia and labor, mode of delivery, number of PCEA, oxytocin dosage after analgesia and amount of epidural analgesic solution. Midwifes measured Apgar scores at 1 and 5 min after birth and withdraw cord artery blood for pH measurement. The incidence of lower limb numbness, skin pruritus, urinary retention, nausea, vomiting and other adverse effects were observed and recorded. Finally, maternal satisfaction was scored.

Based on previous studies, a minimum of 20–40 subjects was required to obtain a stable estimated sample size [8,9,10], so 45 subjects were ultimately included in this study. The inherent expected failure rate according to the biased coin design sequential method was 11%, so the study endpoint for this experiment was at least 41 cases with successful block. The equivariate regression estimator for EC90 was a linear interpolation of the dose between p*(r) and p*(r + 1): EC90 = (x(r + 1) − x(r)) + x(r).Where x (r) = maximum (x (1): p*(i) ≤ 0.9), p* (i) was the main influence dose for x(i) supervision, i = 1, 2, 3,…, k, which was estimated by the Pool Adjacent Offender Algorithm (PAVA). The observed rate of p = (p(1), p(2),. p(k)) might not increase with increasing dose level, which was an implicit assumption of dose seeking studies. Therefore, the PAVA algorithm was first used for equivalence regression to obtain increasing adjustment rates based on p' = [p*(1) ≤ p*(2) ≤ p(k)]. Based on these results, the EC90 of alfentanil was calculated. Considering the independent linear relationship between doses, the 95% CI of EC90 was obtained according to the prediction model [E(Y∣X) = β0 + β1X].