Trial design

Patients were enrolled from January 18, 2019 to February 19, 2021. This study was first registered and released September 6, 2018 at ClinicalTrials.gov identifier NCT04028284. There was a dialog with reviewers which resulted in a rereview and final approval of the entry in July 2019 without modifications to trial procedures or conduct (online supplemental file 1).

Participants

The study was approved under a verbal assent consent model for patients who had provided consent to a thoracic epidural for postoperative analgesia. We reasoned the intervention did not entail any additional significant risk to the patient, as procedures in both study arms were ultimately guided by the same tactile feedback such as loss of resistance. Only patients who met inclusion criteria were approached for participation in this study before their procedure. They were informed about the purpose, procedures, benefits/risks, and were given enough time to read a lay study summary and ask questions about the study. The study research team ensured that the participants knew their participation was voluntary and they had the right to withdraw from the study/collected data at any time. Informed consent was obtained from the anesthesiologists performing the epidural procedure as they were subjects and randomized. Patients undergoing abdominal surgery with a planned thoracic epidural catheter participated in this study.

Patient inclusion criteria were ≥18 years old and American Society of Anesthesiology Physical Status I to IV undergoing any thoracic or upper abdominal elective surgery with planned thoracic epidural analgesia. Exclusion criteria were contraindications to epidural catheter, pregnancy, preexisting coagulopathy, localized infection, and known allergy to local anesthetics.

Randomization, allocation, blinding

In this study four staff anesthesiologists experienced in regional anesthesia and in preprocedure US-guided epidural techniques participated. All anesthesiologists completed a training session where they practiced the MR technique on thoracic spine phantoms until they were comfortable with the procedure and deemed competent by study staff.16 A fifth anesthesiologist participated on a pilot basis during the training phase only as they departed our institution. At each epidural insertion event, the four subjects were randomized (1:1) to landmark or preprocedure US-guided thoracic epidural technique, as per operator preference (group C), or MR guidance (group H). The randomization sequence was computer generated by non-clinical study staff with allocations placed in sealed, sequentially numbered envelopes, opened only at the time of epidural placement to maintain allocation concealment. The epidural level targeted was determined by the procedural anesthesiologist based on the planned surgical incision.

Study procedure and data collection

In group C (control), the staff anesthesiologist followed the traditional technique for landmark or preprocedure US-guided thoracic epidural insertion (as per their preference). Briefly, the anesthesiologist used landmarks or the US to identify and mark the appropriate spot on the skin for insertion of the epidural needle. Local infiltration was with 1–5 mL of 2% lidocaine solution with a 25G needle, followed by 2 min of waiting. The US probe was then placed at rest and the anesthesiologist proceeded with thoracic epidural needle insertion following standard techniques.

In group H, the staff anesthesiologist used the HoloLens system in combination with the US to obtain a hologram image of the trajectory toward the epidural space, thereby mitigating the need to walk off the lamina. The operator performs the procedure while wearing the HoloLens headset, which minimally interferes with procedure ergonomics and visibility. A preprocedural, non-sterile US scan of the spine was performed with the prepared tracking marker attached to the transducer (figure 1). The posterior complex is identified and placed in the middle of the US screen. The angulation of the US transducer is then registered by the headset detecting the position of the US transducer marker, and a hologram is projected into the user’s workspace, which replicates the needle path through the middle of the transducer (figure 1 inset) in a clear path from skin to posterior complex. The central axis of the US transducer is represented by a holographic orange line (10 mm length, 2 mm diameter).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

The user performs a preprocedural ultrasound, allowing the headset to detect the position of the ultrasound transducer and subsequently generate a spatially stable hologram. The inset shows the operator’s mixed reality view displaying a holographic guidance graphic which recreates the angulation of the transducer (orange line along the central axis of the ultrasound transducer).

The transducer is removed from the field, and its center point is marked on the skin in the usual fashion. The spatially stable hologram representing the ideal needle path remains projected into the user’s visual field although the hologram does not provide any information on the desired needle depth nor does it register the needle’s position in the user’s visual frame. Typical sterile prepping and draping and local anesthetic injection do not disrupt the position of the needle path hologram. Operators may then use this hologram to precisely align the needle angulation with the holographic projection in three dimensions from the marked skin entry point. The MR system served as a guide to needle angulation only, with the skin insertion site being noted relative to the US by standard techniques. The Tuohy needle was then inserted by following the MR/holographic trajectory overlaid on the patient’s back.

We have previously published additional figures of this MR technique and video performed on a phantom spine model.16

Outcomes

For the primary outcome, an independent observer documented the time to procedure completion starting at skin puncture to withdrawal of the Tuohy needle after successful catheter placement. We chose to use procedure time as this is the time period where patient morbidity might occur, and where patient anxiety might be greatest. Continuous secondary outcomes, the number of needle passes, needle skin punctures, and needle to bone contacts were also recorded by the trained observer. A “needle pass” was defined as any advancement of the needle along an unchanged trajectory. A “needle skin puncture” was defined as complete needle withdrawal from the skin and reinsertion at a new location. The anesthesiologist performing the procedure would indicate verbally to the independent observer when a needle to bone contact occurred. The observer also documented any signs of complications during the procedure such as paresthesia, dural punctures, as well as procedure failure rates.

A questionnaire regarding the patients’ level of discomfort during the procedure (on a scale of 0–10, with 0 being no pain/no discomfort and 10 being worst possible pain/extreme discomfort) was administered to the patient on completion of the epidural procedure, as well as in the recovery room after the surgical procedure. The ease of use and limitations of each technique were also evaluated using a questionnaire given to the regional anesthesiologist following each procedure.

Epidurals were managed intraoperatively by anesthesiologists not involved in the study with local anesthetic concentration and timing as per their standard of care. Epidural failure was defined as unexpected addition of intravenous patient controlled analgesia to the patient’s pain control regimen prior to discharge from the recovery room.

Sample size calculation

Based on historical data in the literature and from our own institution, mean thoracic epidural analgesia placement requires 10 min with an SD of 3 min. We considered a reduction in procedure time of 2 min (20% relative reduction) to be meaningful in the context of patient comfort and operating room efficiency. With a two-sided type I error rate of 5% and power 80%, 36 participants per group (72 total) were required; recruitment of 84 patients was planned to compensate for potential withdrawals.

Statistical analysis

The primary outcome, time to epidural placement, was analyzed as follows. Times were assessed for normality with the Shapiro-Wilk test. If normally distributed, data was analyzed with the independent samples t-test. If the assumption of normality was violated, the Mann-Whitney U test for non-parametric data was used. Secondary outcomes that were continuous in nature (number of skin punctures, number of needle passes, procedural pain) were analyzed in the same manner. For binary outcomes (parasthesias, dural puncture, success rate) the χ² test (or Fisher’s exact when expected cell counts are less than 5) was used. Categorical variables are presented as frequency and percentages. Normally and nonnormally distributed variables were presented as the mean (SD) and the median (1st–3rd quartiles, (range)), respectively. All inferential testing will be two sided and a p<0.05 was considered significant for the primary outcome. To account for multiple comparisons, inferential testing for secondary outcomes was adjusted with the Holm-Bonferroni method.