Design
We conducted a dose-response, prospective, double-blinded study to determine the ED95 values of intrathecal hyperbaric prilocaine used with 2,5 mcg of sufentanil and 100 mcg of morphine for cesarean delivery.
The study was approved by the institutional Medical Ethics Committee (President E. Stevens, Research Ethics Board number O.M.007; date of protocol approval 24 of March 2016; protocol number NB076201627436). It was retrospectively registered on January 30, 2017 and results posted at the public database clinicaltrials.gov (NCT03036384).
Study population and setting
The present report was established according to ROBUST criteria for Bayesian based studies28, SPIRIT statement for interventional trials29 and CONSORT guidelines30.
Healthy term parturients presenting to our hospital between 1st of April and 30th of November 2016 for elective cesarean delivery were enrolled in the study after signed written informed consent had been obtained.
Inclusion criteria were age between 18 and 40, American Society of Anesthesiologists physical status (ASA) class I-II, body weight less than 100 kg, height between 155 and 175 cm, singleton pregnancy, and gestational age of more than 37 completed weeks.
Exclusion criteria were active labor, ruptured membranes, three or more previous caesarean deliveries, diabetes or gestational diabetes, pregnancy induced hypertension or preeclampsia, intrauterine growth retardation, placenta praevia, congenital anomaly, standard contraindications to neuraxial block, neurological impairment, and known allergy to local anesthetics.
Study protocol
All patients were premedicated with intravenous metoclopramide 10 mg, sodium citrate 30 ml and ranitidine 150 mg orally, 30 minutes before spinal anaesthesia. They received slowly upon arrival in the operating theatre 1000 ml of Ringer’s lactate solution via peripheral intravenous access as regular fluid therapy, which is standard care in our institution.
Continuous electrocardiography, pulse oximetry (SpO2) and non-invasive arterial blood pressure monitor were applied throughout the whole study protocol.
A combined spinal-epidural (CSE) was performed at the L3/L4 or L4/L5 interspace with the parturient in sitting position, under uterine and foetal heart rate monitoring.
Applying the midline approach, an 18G Tuohy needle (Vygon, Ecouen, France) was inserted into the epidural space using a loss-of-resistance-to-saline technique.
The spinal component was performed under aseptic conditions with a needle-through-needle technique using a 27G Whitacre needle (Vygon, Ecouen, France), with the orifice oriented cephalad.
Following observation of spontaneously flowing cerebrospinal fluid, the study solution of hyperbaric 2% prilocaine (Tachipri® Hyperbar, Nordic Pharma) at room temperature was injected over 20s associated with sufentanil 2,5 mcg and morphine 100 mcg. A multiple orifice epidural catheter was then threaded 3 cm into the epidural space, an aspiration test was performed but no drug was injected. Immediately after the procedure, parturient laid supine with a left lateral tilt to cause uterine displacement. A bladder catheter and an O2 face mask with 6l/min O2 were applied.
Each parturient enrolled in the study received an intrathecal dose of HP determined by the CRM and the success or failure of the block was assessed as being the primary endpoint. Off noted, the assessing anaesthesiologist remained blind to the administered dose.
For the purpose of the study, a successful block was defined as a bilateral T4 sensory level31 obtained within 15 min after intrathecal HP dose administration with no pain experienced upon incision and until the end of surgery.
Otherwise, a failure was recorded and epidural supplementation of 5 ml bolus injections of 2% lidocaine with epinephrine 1/200.000 were administered every 5 min through the epidural catheter, in order to obtain a VAS score ≤ 3.
Hypotension was defined as a 20% decrease in systolic blood pressure (SAP) compared to baseline value, recorded before spinal anaesthesia. When occurred, titration of ephedrine 5 mg or phenylephrine 100 mcg was administered at the discretion of the attending anaesthesiologist in order to keep SAP over 90% of baseline.
The surgical technique was uniform for all patients, including uterine exteriorization.
Blinding
To ensure proper blinding throughout the study, the same anaesthesiologist prepared the study dose according to the CRM and performed the combined spinal-epidural. Another investigator, blinded to the dose, assessed the success or failure of each intrathecal block, ensured the subsequent management of the patient and collected the data throughout the study protocol. Similarly, parturient was not aware of the dose administered.
Measurements
Demographic variables recorded in the study were: age, weight, height, body mass index, gestational age, parity and number of previous caesarean deliveries.
Regarding the new-born, weight and Apgar scores at 1, 5 and 10min were recorded after delivery, as well as umbilical vessels pH and methemoglobinemia measured from percutaneous umbilical cord blood samples, using arterial blood analysis.
The following surgical data were also collected: time from spinal anaesthesia to baby extraction, time from baby delivery to the end of surgery, the duration of surgery and total blood loss.
Sensory level was assessed bilaterally by loss of cold and sensation at the midclavicular line and recorded every 2,5 min after intrathecal dose administration of HP (T0) during the first 15 min. Then, every 5 min until the end of the procedure, and every hour in the Post-anesthesia care unit (PACU) until the patient declared regaining full sensitivity, signifying complete resolution of the sensory block. The time to achieve Th4 bilateral level, the maximum level obtained and the total duration of sensory block were also registered.
Bromage scale (1 = no motor block; 2 = hip blocked; 3 = hip and knee blocked; and 4 = hip, knee, and ankle blocked) was used to evaluate the motor block every 15 min after spinal anaesthesia (T0) and until the end of surgery. Patients’ follow-up continued in the PACU every hour until complete recovery of motor block was observed (Bromage score = 1) and total duration of motor blockade was recorded.
Total recovery of both motor and sensory blocks allowed discharge to the care-unit.
Pain was assessed using a 10-cm horizontal visual analogue scale (VAS; 0-10 cm; 0: no pain and 10: worst imaginable pain) at skin incision, new-born delivery, uterine exteriorization, peritoneal and skin closure; in addition, at 5-min intervals throughout surgery and at 15-min intervals during the follow-up in the PACU. Thereafter, pain was evaluated every 4 hours during the first postoperative day in the care-unit.
Maternal arterial blood pressure was recorded by non-invasive measurements at baseline, at 1-min intervals after drug dose administration during the first 15 min then at 2.5-min intervals until the end of surgery and every 20 min in the post anaesthesia care unit. The necessity of using vasopressors (ephedrine or phenylephrine) when hypotension occurred as well as total administered doses were recorded. Heart rate and SpO2 were monitored continuously.
Regarding side-effects, the incidence (presence or absence) of nausea, vomiting and pruritus, were recorded at 15 min intervals from intrathecal dose administration until the end of surgery and at the same time-points as pain was assessed. During the postoperative period and until hospital discharge, all parturients were questioned and examined as well for transient neurologic symptoms (TNS), urinary retention and dizziness.
From a quality point of view, maternal satisfaction (yes or not) was assessed 1 h after surgery and in the care-unit ward.
All collected data were registered anonymously, according to institutional ethics committee policy.
Dose allocation
To provide a valid estimation of the ED95 of 2% HP with sufentanil 2,5 mcg and morphine 100 mcg for caesarean section, the study design was based on the modified CRM.32
It is an adaptive Bayesian method, designed to estimate the targeted percentile on the response curve among several dose levels, requiring small sample of patients of around 20-30 to reach valid conclusions. Originally designed for dose-toxicity finding in oncology trials, it was then extended to dose-failure in phase II trials, notably in anaesthesiology.24
We set out to recruit 40 parturients, 4 per cohort, to benefit from spinal anaesthesia with 2% HP different given doses with sufentanil. The starting dose of 45 mg was determined using a priori estimates of the ED95 based on our previous experience. Subsequent doses were allocated based on the CRM power model (fig. 1), with the operator remaining blind to the given doses. Indeed, the results of each cohort were analysed by the statistical advisor researcher (Mr J-F. Fils) in order to propose to the clinical investigator the next dose to allocate.
Dose-response statistical analysis
Assuming a dose-failure relationship, with higher doses being more toxic and lower doses less efficacious, we want to find the ED95; that is, the dose defined as the 5th percentile of the dose–failure relationship, which is modelled throughout a power model as follows:
P(Y=1/xi) = piθ,
where θ is the model parameter to be estimated, considered as a random variable with exponential unit prior, xi is the administrated dose to the ith patient and pi (i = 1, …k) is the initial guess of failure probabilities at the ith dose level.
Six dose levels (= k) were chosen, specifically 30, 35, 40, 45, 50 and 55mg, whose range was based on our previous experience. The guesstimates failure probabilities associated to the retained doses were given by clinicians as 0.5, 0.25, 0.10, 0.05, 0.02, and 0.01, a priori corresponding respectively to ED50, 75, 90, 95, 98 and 99 of HP with sufentanil.
The CRM is conducted as follows: the first cohort of four patients is administered the initial candidate of the ED95, the dose level 45 mg. Then, depending on the response observed for all patients in the cohort, Bayes theorem is applied in order to provide the actualized posterior distribution of the model parameter. Subsequently, the posterior mean estimate is computed – that is, the mean distribution after taking into account the patients recruited so far in the trial – E (θ/y), and then is used in the power model to give an updated probability of failure at each dose level. The dose allocated to the next cohort is the one with an actualized posterior response closest to the target 0.95 (95%).
CRM allows to previously incorporate stopping rules which is important for an ethically and statistically reliable approach of patients.27,33,34
Our trial continued until one of the following stopping criteria was met:
- the planned number of 40 patients was reached;
- the estimated posterior probability of response was either too low or too high for all dose levels;
- a reliable estimation of the ED95 was obtained, based on the predictive gains (mean and maximum) of further patients’ inclusions on the response probability and on the width of its credibility interval lower than 5%.
Collected demographic, surgical and clinical data were expressed as mean ± standard deviation or absolute number, as appropriate.
The dose-finding allocation and analysis of remaining data were performed using R software version 3.2.2 (R CRAN, Vienna, Austria).