Spinal anesthesia is recommended in total hip or knee arthroplasty, as it is associated with superior clinical outcomes and recovery profiles compared to general anesthesia [1]. Nevertheless, it can be technically challenging in many patients due to age-related degenerative spine disease or other spinal anomalies which restrict access to the vertebral canal at the commonly used L2-L5 intervertebral spaces. In such cases, an alternative strategy is to perform intrathecal injection at the L5-S1 level since it is the largest interlaminar space, is least affected by arthritic and degenerative changes, and has the lowest chance of spinal cord trauma [2,3]. However, a major drawback to spinal anesthesia at lower lumbar intervertebral levels, particularly L5-S1, is the increased odds of failure to achieve adequate surgical anesthesia [4]. This occurs because “plain” local anesthetic solutions such as 0.5% bupivacaine are clinically isobaric with respect to cerebrospinal fluid (CSF), resulting in unpredictable cranial distribution and sensory block height [5].

Although increasing local anesthetic dose may reduce therapeutic failure from inadequate block height, the 95% effective dose (ED95) of isobaric 0.5% bupivacaine injected at the L5-S1 interspace to achieve a sensory anesthesia level of T10 or higher is 25 mg [6]. This dose is much larger than the conventional doses of 12–15 mg used in total joint arthroplasty and would prolong sensory and motor block duration well beyond that required for surgical completion; with consequent delays in post-anesthetic care unit (PACU) discharge, mobilization, and recovery. This is at odds with the goals of modern fast-track arthroplasty protocols that emphasize same-day mobilization, physiotherapy, and even same-day hospital discharge. On the other hand, using lower intrathecal doses of isobaric bupivacaine increases the risk of inadequate surgical anesthesia that may necessitate conversion to general anesthesia [7].

One way to resolve this dilemma is to administer a local anesthetic solution that is either hyperbaric or hypobaric relative to CSF, in conjunction with appropriate positioning, to promote cranial distribution within the CSF and blockade of the higher spinal nerve roots. We have previously shown that low-dose (10 mg) hypobaric 0.33% bupivacaine provides successful surgical anesthesia of adequate, but not excessive, duration for total hip and knee arthroplasty in our institution [8]. The study cohort of 60 patients included two who received spinal anesthesia at the L5-S1 interspace. Both patients had adequate sensory block height and duration for commencement and completion of surgery without need for anesthetic supplementation. This suggests that the anesthetic success rate of hypobaric local anesthetic administered at the L5-S1 interspace may be comparable to higher lumbar interspaces, making it a useful strategy in patients with challenging spinal anatomy. To date though, the success rate of this specific technique has not been systematically quantified. We therefore conducted a prospective observational study to evaluate the success rate and clinical characteristics of spinal anesthesia with low-dose (10 mg) hypobaric bupivacaine injected at the L5-S1 interspace in patients undergoing total hip or knee arthroplasty surgery.

This was a prospective observational trial conducted at Toronto Western Hospital (TWH). Ethical approval for this study was provided by the University Health Network Research Ethics Board (24-5192) and written informed consent was obtained from all participants. This trial was prospectively registered on July 11, 2024, prior to patient recruitment (ClinicalTrials.gov NCT06526156, URL: https://clinicaltrials.gov/study/NCT06526156). Patient enrollment began on 22 July 2024, and the study was completed on 20 March 2025. The study was conducted in accordance with the Declaration of Helsinki (2024) and the guidelines of Good Clinical Practice. This manuscript adheres to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.

We enrolled patients 20 years of age or older, American Society of Anesthesiologists status I–III, who were scheduled for elective unilateral total hip or knee arthroplasty under spinal anesthesia. Exclusion criteria included patient refusal to participate in the study, inability to communicate due to language barriers or cognitive impairment, height > 200 cm, weight > 120 kg, body mass index > 45 kg/m², contraindication or allergy to amide-type local anesthetic, contraindications to spinal anesthesia at the L5-S1 space (including infection at the injection site and severe spinal stenosis at higher vertebral levels), previous lumbosacral spine surgery, planned administration of intrathecal opioids, pre-existing sensory or motor impairment in the lower extremities, and inability to identify the L5-S1 intervertebral space with ultrasonography.

Spinal anesthesia was performed in a dedicated block room approximately 30 minutes ahead of the anticipated start of surgery. Intravenous access and standard monitoring, including electrocardiogram, non-invasive blood pressure measurement, and pulse oximetry were established prior to the spinal anesthetic. Patients were then placed in a sitting or lateral decubitus position and conscious sedation was administered as needed at the discretion of the anesthesiologist performing the procedure. The L5-S1 intervertebral space was identified with ultrasonography and marked. Time to scan, locate and mark the L5-S1 interspace (defined as first skin-probe contact to completion of skin marking) was recorded. A hypobaric 0.33% bupivacaine solution was prepared by mixing 2 mL of plain (isobaric) 0.5% bupivacaine with 1 mL of sterile water [8]. Dural puncture was performed using a 25- or 27-gauge spinal needle and the total 3 mL volume was injected into the intrathecal space. Time to perform the spinal anesthetic (defined as skin infiltration to completion of intrathecal injection) and number of needle passes (defined as an insertion followed by a withdrawal of the spinal needle) was recorded. After completion of spinal injection, patients were immediately placed in a lateral decubitus position with the operative side uppermost, and a 10–15° head-up tilt to promote cranial distribution of local anesthetic. If hypotension occurred or the sensory block surpassed T6, the patient was returned to a level position. Sensory and motor blockade produced by the spinal anesthetic were assessed every 5 minutes for the first 30 minutes after completion of intrathecal injection or until transfer to the operating room, whichever occurred sooner. If an adequate level of surgical anesthesia was not achieved within this time, supplementation or conversion to general anesthesia was performed at the discretion of the attending anesthesiologist.

Patients undergoing total knee arthroplasty or total hip arthroplasty via an anterior approach were placed in a supine position upon arrival in the operating room. Patients undergoing total hip arthroplasty via a posterolateral approach remained in the lateral decubitus position with the operative side uppermost. All patients received intraoperative propofol sedation during the operation as per current institutional practice. The patients were transferred to the PACU following completion of surgery.

Successful spinal anesthesia was defined as completion of surgery without conversion to general anesthesia, administration of supplemental opioids, or surgical local anesthetic infiltration in the operating room. Adequate spinal anesthesia for initiation of surgery without these supplemental interventions was defined as a secondary outcome.

Sensory block was assessed according to standard dermatomal maps. Pinprick testing using an 18 G blunt-tipped needle (BD Blunt Fill needle) was performed on non-operative and operative sides every 5 minutes during the first 30 minutes after spinal injection to detect sensory loss over the lower limbs and torso. Sensation to pinprick testing in each dermatome was categorized as normal (“sharp” sensation), analgesia (“dull” sensation), or anesthesia (“no” sensation). The most cranial (upper) level of sensory analgesia and anesthesia was recorded based on the dermatomal map. Pinprick testing was resumed at the end of surgery and every 30 minutes thereafter in the PACU and inpatient ward until a return of normal sensation in the L2 dermatome was documented.

Motor block was assessed by active flexion and extension at the hip, ankle, knee, and toes. This was scored as either 1 (movement present and observed) or 0 (movement absent) at each joint on both lower extremities every 5 minutes during the first 30 minutes after spinal injection. Recovery of motor function was assessed at the end of surgery and every 30 minutes thereafter in the PACU and inpatient ward until full motor recovery had been achieved in both lower limbs. Motor recovery was also documented using the modified Bromage scale (0, able to move hip, knee, ankle and toes; 1, unable to move hip, able to move knee, ankle, and toes; 2, unable to move hip and knee, able to move ankle and toes; 3, unable to move hip, knee, ankle, able to move toes; 4, unable to move hip, knee, ankle, and toes).

Except for time to scan and mark the L5-S1 space, and time to perform the spinal anesthetic (as defined above), all other reported time intervals were calculated starting from completion of intrathecal injection. Preoperative sensory and motor block outcomes of interest included: most cranial level of sensory anesthesia documented prior to start of surgery, and time to reach this level; time to reach sensory anesthesia at level of T10 or higher on the operative limb (the minimum level we deemed adequate for hip arthroplasty); time to reach sensory anesthesia at level of L1 or higher on the operative limb (the minimum level we deemed adequate for knee arthroplasty); time to complete motor block at hip, knee, ankle, and toes in both lower limbs. Postoperative sensory and motor block outcomes of interest included: duration of analgesia at T10 and L1 or higher; time to recovery of normal sensation at L2 or lower in both lower limbs; time to recovery of normal motor function (ability to move hip, knee, ankle, toes and a score of 0 on the modified Bromage scale).

Hemodynamic stability after the spinal injection (i.e., the incidence of hypotension and bradycardia following the block) was evaluated by the need to administer vasopressor or vagolytic medications (e.g., phenylephrine, ephedrine, glycopyrrolate, atropine) after the spinal injection and prior to initiation of intraoperative sedation. These medications were administered at the discretion of the anesthesia care team. The total amount of vasopressors used in the block room, and in the operating room prior to initiation of intraoperative sedation was documented. The incidence of adverse effects including hypotension and/or bradycardia following spinal anesthesia and prior to intraoperative sedation (as defined above), postoperative nausea and vomiting (PONV) in the PACU (defined as need for administration of anti-emetic medication), and urinary retention (defined as need for urinary catheterization after surgery) were assessed by direct observation and chart review of the electronic medical record.

The surgeons’ perception of the quality of surgical anesthesia in terms of operating conditions provided by the spinal anesthetic was assessed at the conclusion of surgery using a 3-point Likert scale (1 = Poor; 2 = Acceptable; 3 = Good). In the event of a score of 1-2 on the Likert scale, surgeons were asked to clarify reasons for the suboptimal operating conditions, and how they thought this could be improved.

Patient satisfaction with their anesthetic experience was assessed at the final study assessment visit (when regression to L2 was documented) using a 5-point Likert scale (1 = very dissatisfied; 2 = somewhat dissatisfied; 3 = neutral (neither satisfied nor dissatisfied); 4 = somewhat satisfied; 5 = very satisfied) and by asking if they would choose the same anesthetic for similar surgeries in the future. Negative responses were followed by clarifying questions to determine specific reasons for dissatisfaction.

We enrolled fifty-five patients who met inclusion criteria. All 55 patients completed the study; however, one patient was later discovered to have significant spinal stenosis at the L4-5 level and was subsequently excluded from analysis as per protocol. Baseline characteristics and key metrics of the technical performance of spinal anesthesia in the 54 included patients are summarized in Tables 1 and 2, respectively. In four patients the interval between completion of spinal anesthesia and transfer to operating room was too short to permit any preoperative sensory and block assessment (Fig. 1). Sensory and motor block testing was curtailed in a proportion of patients by transfer to the operating room within less than 30 minutes after intrathecal injection that limited the assessment of block onset time and height (Fig. 1). Assessment of postoperative block regression was similarly limited by the fact that in a proportion of patients, sensory block had receded below the threshold levels of T10 and L1 at the time of the first postoperative assessment (Fig. 1). Patients with missing data from sensory and motor block testing were excluded in calculation and analysis of relevant outcomes.

Key metrics of the anesthetic and recovery profile of L5-S1 spinal anesthesia with 10 mg hypobaric bupivacaine are summarized in Tables 3 and 4. The dermatomal progression of sensory block over the first 30 minutes is illustrated in Fig. 2. The median highest dermatomal level of sensory anesthesia recorded in the operative limb at the final preoperative assessment before transfer to the operating room was L1 (T10, L3). Mean time from intrathecal injection to the final assessment before transfer to the operating room was 23 ± 8 minutes. Mean time from intrathecal injection to skin incision was 45 ± 13 minutes, and mean time from injection to surgical completion was 114 ± 18 minutes.

Out of the 54 patients included in the study, 50 completed surgery without conversion to general anesthesia, administration of supplemental opioids, or surgical local anesthetic infiltration in the operating room (92.6%, 95% CI [82.1–97.8]). Fifty-three pa­tients had adequate anesthesia for initiation of surgery without supplementation (98.2%, 95% CI [90.1–100]). One patient undergoing total hip arthroplasty had only achieved sensory anesthesia to the L2 dermatome in the operative limb 30 minutes after intrathecal injection. Sensory anesthesia at the surgical site remained incomplete based on pinch testing prior to incision (56 minutes after intrathecal injection). Surgery was completed successfully under propofol sedation with supplemental opioids (50 µg IV fentanyl at incision and intraoperative remifentanil infusion at 0.03 µg/kg/min). An additional three patients required opioid supplementation for completion of surgery due to inadequate anesthetic duration. All three patients underwent total hip arthroplasty with surgical durations of 67, 67, and 88 minutes respectively. However, there was a prolonged interval between intrathecal injection and start of surgery of 47, 55 and 59 minutes respectively; and supplemental opioids were administered 109, 118, and 123 minutes after intrathecal injection respectively. No patients required conversion to general anesthesia for completion of surgery.

Mean surgical duration of knee and hip arthroplasty was 66 ± 11 and 74 ± 10 minutes, respectively. Postoperative dermatomal regression of sensory block is illustrated in Fig. 3. Mean time to recovery of normal sensation at L2 or lower in both limbs was 195 ± 47 minutes. Mean time to full recovery of motor function was 198 ± 53 minutes in the operative limb and 163 ± 52 minutes in the non-operative limb.

Twenty-six patients (48%) had hypotension requiring vasopressor medication during the period between intrathecal injection and initiation of intraoperative sedation. In these patients, the mean phenylephrine and ephedrine doses administered were 168 ± 114 µg and 11 ± 8 mg respectively. Mean time from intrathecal injection to vasopressor administration was 11 ± 8 minutes. There were no instances of bradycardia requiring treatment.

Three patients (5.6%) had PONV requiring treatment, and no patients had urinary retention. Patient satisfaction with the anesthetic technique was high, with a median patient satisfaction score of 5 (5, 5) out of 5. All 54 patients stated that they would choose to have the same anesthetic technique for similar surgeries in the future. Surgeons rated the quality of spinal anesthesia as “good” in 51 cases (94%) and “acceptable” in 3 (6%) cases.

L5-S1 is the widest lumbar interlaminar space and is less prone to collapse with degenerative disease, making it a useful option for spinal anesthesia in elderly patients or those with challenging spinal anatomy [3,10–12]. However, concerns regarding an increased risk of inadequate block height with injection at lower lumbar intervertebral spaces have hitherto precluded more widespread use [4,13]. In this prospective observational study, spinal anesthesia with 10 mg of hypobaric 0.33% bupivacaine administered at the L5-S1 interspace was associated with successful completion of total hip and knee arthroplasty, without need for anesthetic supplementation, in 92.6% of patients. Adequate anesthesia for initiation of surgery was achieved in 98.2% of patients.

We defined our primary outcome as successful completion of surgery without need for anesthetic supplementation. One patient required analgesic supplementation for initiation of surgery with a modest dose of systemic opioid at skin incision due to inadequate block height. Another three patients required supplemental opioids for completion of surgery due to inadequate block duration. This last finding should be interpreted in the context of our specific institutional workflow and priorities. As an academic teaching hospital, our mandate is to balance the needs of learners with clinical goals of surgical readiness and early recovery. Spinal anesthesia in our institution is performed in a dedicated block room using a parallel processing model. An inherent risk and limitation of this model is that the interval from intrathecal injection to the start and end of surgery is highly influenced by the efficiency of operating room turnover. All three patients who required supplementation for completion of surgery had prolonged injection-to-incision intervals (47–59 minutes) due to unanticipated delays in PACU bed availability and operating room turnover. This may not apply in other environments where spinal anesthesia is performed in the operating room.

A recent retrospective study of 3542 spinal anesthetics in a similar population of patients undergoing elective total hip and knee arthroplasty provides a useful benchmark for evaluating our results [4]. Colish et al. looked specifically at therapeutic failure after obtaining CSF backflow, defined as need for conversion to general anesthesia or repeat spinal anesthesia within 2 hours of the first spinal anesthetic. They used exclusively hyperbaric or isobaric bupivacaine and observed an overall failure rate of 3.8% (success rate of 96.2%). However, subgroup analysis by level of injection showed that the odds of failure were 4.6 and 7.7 times higher when spinal anesthesia was performed at L4-5 and L5-S1 levels respectively, compared to the L2-3 level [4]. Although Colish et al. did not provide specific data on the nature of therapeutic failure, they postulated that larger local anesthetic doses may be required at lower interspaces to achieve adequate cranial spread that in turn implies that maldistribution was a major contributor. The 92.6% success rate of surgical completion observed in our study with the use of hypobaric bupivacaine compares favorably with the success rate of 87.7% reported by Colish et al. [4] for spinal anesthesia at the L5-S1 level, even after accounting for the workflow-related delays that contributed to failure in our study. This suggests that hypobaric bupivacaine may promote more consistent cranial distribution within CSF. This is further supported by the fact that our success rates were also achieved with a relatively low dose of 10 mg. In contrast, the mean bupivacaine dose in Colish et al. [4] was 12.4 ± 1.5 mg and was combined with a mean intrathecal fentanyl dose of 13.5 ± 8.4 µg.

Like hypobaric bupivacaine, hyperbaric bupivacaine also has a greater propensity for gravitational distribution within CSF with appropriate positioning, compared with isobaric bupivacaine. Nevertheless, Colish et al. [4] observed that hyperbaric bupivacaine was associated with a higher odds of failure (1.7 times) than isobaric bupivacaine. In another study of total hip and knee arthroplasty, 3.3% of patients had inadequate block height following injection of 3.5 mL of 0.5% hyperbaric bupivacaine at the L5-S1 level [14]. While the relative superiority of hypobaric versus hyperbaric versus isobaric bupivacaine can only be established by an adequately powered randomized controlled trial, these findings suggest caution when using hyperbaric bupivacaine for L5-S1 spinal anesthesia in this context. Possible explanations for failure include a greater risk of sacral pooling in the supine position if intrathecal injection occurs caudal to the apex of the lumbar lordosis (classically at L3), and preferential distribution to the dependent non-operative limb if patients are prematurely placed in lateral decubitus for total hip arthroplasty.

There are several points to note in interpreting the results of our study. First, the low dose of bupivacaine reflects our institutional practice for total hip and knee arthroplasty, where operating times are relatively short and predictable, and early mobilization is a priority. Faster block resolution also reduces the risk of urinary retention (a complication absent in our cohort) [15]. The mean sensory block duration above T10 and L1 was 135 ± 44 and 158 ± 33 minutes respectively, which aligns well with typical North American operative times for total hip and knee arthroplasty [16]. However, these pharmacodynamic estimates are based on a limited sample (15% and 46% of the cohort for T10 and L1 regression respectively) as sensation had regressed beyond these thresholds at first postoperative assessment in the rest of the cohort. Therefore, they should be regarded as broad approximations rather than definitive measures of block duration.

Second, this was a mixed cohort of total hip and knee arthroplasty patients. Compared to total knee arthroplasty, total hip arthroplasty requires a higher dermatomal level of sensory anesthesia for initiation and completion of surgery. The duration of surgery may also differ between the two. These factors may contribute to variation in the clinical success rate of spinal anesthesia despite standardization of technique and local anesthetic dose. In our cohort, the success rate was indeed lower in hip versus knee arthroplasty for both surgical incision (95% vs. 100%) and surgical completion (79% vs. 100%). However, our study was not designed or powered for a formal comparison between total hip and knee arthroplasty; and these subgroup findings should be regarded as hypothesis-generating rather than conclusive. Practitioners should tailor the local anesthetic dose to surgery type as well as their institutional practice norms. Larger bupivacaine doses can be used to extend duration if needed, with adjustments to volume of sterile water to maintain a dilution ratio of 2:1.

On a related note, although we did not formally measure the specific gravity of the bupivacaine solution used in this study, previous studies of hypobaric bupivacaine have used 0.5% bupivacaine:water volume dilution ratios of 2.3:1 and 1.7:1, and reported the specific gravity of these mixtures to be 0.997316 and 0.996922 g/mL, respectively [17,18]. Our dilution ratio of 2:1 falls into this range, and produced a clinical effect consistent with meaningful hypobaricity, evidenced by the pattern of preferential blockade of the non-dependent limb observed here as well as in our previous work [8].

Excessive cranial CSF distribution is a consideration with the use of non-isobaric intrathecal local anesthetic; however, there were no instances in our cohort of excessively high spinal anesthesia as judged by sensory testing or hemodynamic instability. There was also no marked bradycardia requiring treatment to suggest blockade of cardio-accelerator fibers. The incidence of hypotension was in line with the typical range for neuraxial anesthesia in older patients and was effectively managed with modest doses of vasopressors [19]. Nevertheless, practitioners should be mindful of the potential for excessive cranial spread and consequent sympathectomy with larger doses of hypobaric local anesthetic, or in patients of extremely short stature. If patients are placed in a reverse Trendelenburg or head-up position to promote cranial distribution of hypobaric local anesthetic, close monitoring is advised with a return to a level supine position once the desired block height is achieved. Furthermore, our findings may not apply in surgery requiring coverage at higher thoracic levels; guidance in these scenarios should be sought from other studies [20].

We acknowledge several limitations of the present study. First, this was a prospective observational study without a contemporaneous control group. An RCT was not feasible or ethically acceptable at this stage given previously reported failure rates of spinal anesthesia at the L5-S1 level. Second, the study was powered using a precision-based approach assuming a low anticipated failure rate. However, the relatively small sample size resulted in a wide CI for the primary outcome, and the estimate of success rate is sensitive to small numbers of additional failures. Our results, while encouraging, do not therefore provide definitive conclusions regarding the success rate of L5-S1 spinal anesthesia with hypobaric bupivacaine, nor its relative efficacy compared to hyperbaric or isobaric solutions. Larger studies and randomized controlled trials are required for this.

Third, the variation in time interval between intrathecal injection and transfer to the operating room limits the interpretation of onset and duration of effective surgical anesthesia. This was an unavoidable consequence of the block room parallel-processing model and clinical workflow at our institution, and limits generalization of our results to settings in which spinal anesthesia is performed in the operating room. This time interval was longer than expected for three patients resulting in inadequate duration of surgical anesthesia. In other patients, shorter-than-anticipated intervals limited the collection of data for analysis of sensory block onset and height. This also explains the apparent discrepancy between the recorded preoperative sensory level (median of only L1, with a lower bound of L3) and the high rate of successful anesthesia for surgical incision. In our workflow, there was an average time interval of 24 minutes between transfer to the operating room and skin incision, during which block height continued to evolve.

We wish to highlight two additional points related to the clinical application of the described technique. First, we are not advocating that the L5-S1 intervertebral space be targeted as a first-line approach, but rather as a fallback option where spinal anesthesia is technically challenging at other levels. In this scenario, we strongly recommend the use of pre-procedural ultrasound imaging to locate and ascertain the patency of the L5-S1 interlaminar space. Based on the results of this study, we also recommend the routine injection of hypobaric local anesthetic if the L5-S1 level is used. Second, injection at L5-S1 may slow onset time relative to higher levels: mean ± SD onset times for sensory block to L1 and T10 were 14 ±7 and 17 ± 8 minutes respectively, compared to 6 ± 3 and 8 ± 5 minutes in our previous study with the same local anesthetic solution [8]. Patients should therefore be placed in a reverse Trendelenburg position for at least 10–15 minutes after intrathecal injection (this can be done during surgical preparation) to promote and accelerate cranial distribution of local anesthetic within the CSF. As discussed above, the continual increase in block height during surgical preparation accounts for the fact that although the documented median sensory level prior to transfer from block room to the operating room was only L1, anesthesia was sufficient for initiation of surgery in all but one patient.

In conclusion, the injection of low-dose (10 mg) of hypobaric 0.33% bupivacaine at the L5-S1 intervertebral space was associated with a high observed rate of successful spinal anesthesia for total hip and knee arthroplasty, with anesthetic and recovery profiles suited to modern fast-track surgery pathways. This technique is particularly valuable as a fallback option in patients with challenging spinal anatomy and in whom L5-S1 is the most accessible level for intrathecal injection.