After receiving IRB approval, written informed consent was obtained from 24 volunteers for this prospective, controlled, blinded, crossover trial. Exclusion criteria included age younger than 18 or older than 70, pre-existing paresthesia or neuropathy of any kind, history of excessive alcohol consumption ( >3 drinks per night), regular use of analgesics, history of any drug abuse, baseline systolic blood pressure ≥150 mm Hg, baseline diastolic blood pressure ≥85 mm Hg, BMI ≥35, pregnancy, and history of allergic reaction to local anesthetics. Participants were given the option to withdraw from the study at any time.
Randomization tables were generated by the study statistician and forwarded to the Investigational Drug Service (IDS). The IDS staff prepared the 20 mL syringes with either saline or 0.5% bupivacaine with epinephrine 1:200,000. The participants were randomized into 2 sequence allocation groups which determined whether the participant would receive bupivacaine or saline first. Randomization was also stratified by gender (6 males and 6 females in each group). Within each of these groups, half of the participants received axillary ring blocks on their dominant arm and the other half on their non-dominant arm (dominance determined based on their 'handedness'). Thus, 3 males and 3 females received saline blocks on their dominant arm, 3 males and 3 females received saline blocks on their non-dominant arm, 3 males and 3 females received bupivacaine blocks on their dominant arm, and 3 males and 3 females received bupivacaine blocks on their non-dominant arm. Participants and investigators were blinded to the substance administered for analgesia (bupivacaine vs. saline). When the participants returned one month later for a second session, they all received the opposite solution from the one they received during the first session, but the same arm (dominant or non-dominant) that was used for the first session was also used for the second session.
The axillary ring blocks were performed in the sitting position with standard ASA monitors. A 1.5 inch, 25 gauge needle was used to inject the solution at the level of the proximal axillary fossa, and the entire width of the medial aspect of the arm was infiltrated with 0.25% bupivacaine with Epinephrine 1:200,000 or saline to raise a subcutaneous wheal. 1 mL of solution for every inch of arm circumference up to a maximum of 15 mL was injected. For each subject, the tourniquet was inflated 10 min after completion of the axillary ring block. Prior to application of the tourniquet, the extremity was wrapped at the tourniquet site with soft gauze to prevent post-tourniquet discomfort and skin bruising. The arm was then elevated to allow passive exsanguination and a 5 inch (12.7 cm) Esmarch wrap was applied systematically from the distal part of the extremity to the tourniquet. This exsanguination of the extremity is routinely combined with a tourniquet to create an almost bloodless surgical field. Once the Esmarch was applied, the tourniquet was inflated to a pressure 100 mm Hg higher than the participant’s baseline systolic blood pressure and the Esmarch was unwrapped. From this point forward, a dedicated observer assessed and recorded heart rate, blood pressure on the opposite arm, 0–10 numeric pain rating scores under the tourniquet and anywhere below the tourniquet, sensation to fine touch and pin prick on the dorsum of the hand, and grip strength every 5 min. The subject’s response to “pin prick” was described as sharp, touch but not sharp, or no sensation. The subject’s response to “fine touch” as measured with a fine piece of cotton was described as normal, decreased, or absent. Grip strength was recorded as full strength, decreased strength, or no motor function. Subjects were asked to tolerate some degree of discomfort, but not severe pain. The primary outcome measure for this study was the amount of time that the subject was able to tolerate the tourniquet, up to a maximum time of 1 h. The tourniquet was deflated prior to the 1 h maximum at the subject’s request, or if the subject’s heart rate or blood pressure exceeded the maximum parameters set by the American Heart Association:
Up to age 30 190 beats per min.
Age 31 to 35 185 beats per min.
Age 36 to 40 180 beats per min.
Age 41 to 45 175 beats per min.
Age 46 to 50 170 beats per min.
Age 51 to 55 165 beats per min.
Age 56 to 60 160 beats per min.
Age 61 to 65 155 beats per min.
Age 66 to 70 150 beats per min.
Maximum systolic blood pressure; 160 mm Hg.
Maximum diastolic blood pressure; 100 mm Hg.
At the time of the second session, subjects were asked to request release of the tourniquet at the same degree of discomfort that they experienced at the time of deflation during the first session. Important secondary outcomes included pain scores under the tourniquet and pain scores anywhere distal to the tourniquet.
Statistical analyses
The effect of the anesthetic on the primary outcome in this two-period two-treatment crossover study was evaluated in two stages. First, it was necessary to assess carryover effect, as such effects, if substantial, may invalidate the use of the crossover design. Carryover effect is defined as the effect of the treatment from the first period on the treatment effect at the second period, and is tested using the model by Grizzle. If the carryover effect is significant (p <0.05), only the data from the first period should be analyzed using the ordinary parallel group analysis. Secondary outcomes in this study included numeric rating scale pain scores under the tourniquet and below tourniquet, as well as motor and sensory function. Pain trajectories between the two treatment groups were compared using the mixed effects model with random intercepts and random slopes, taking into account the correlated measurement errors within each subject. All analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC). P-values of less than 0.05 were considered statistically significant.
Power and sample size
Sample size for this study was determined using the PROC POWER procedure for paired means as implemented in SAS 9.3. Assuming the standard deviation of tourniquet duration is 10 min, Type I error rate 0.05, correlation between repeated assessments within each subject 0.3, a total of 20 subjects (allocated to 2 equal groups with treatment sequences) will yield the power of 0.98 to detect the difference of tourniquet time of 10 min. The planned sample size was increased by 20% (to a total n= 24) to take into account the likely drop out of study participants.