One of the most common procedures performed in the geriatric population is the total hip arthroplasty [1]. Depending upon the age range examined, it is either the second or the third most common procedure performed within the age range of 65–90+ years. In patients aged 18–64 years, it is not even in the top ten. In older females, the incidence of bone fractures is so common that it is higher than the aggregate incidence of stroke, breast cancer, and heart disease [2]. Further, 40% of those with hip fracture will require nursing care and 20% will be unable to return to normal ambulation.

With these statistics, it is little wonder that perioperative efforts are focused on either identifying preoperative risk modifiers or working to reduce known comorbidities [3]. While the definition of geriatric tends to focus on age alone, the last decade has seen an explosion in the understanding and need for further research to better quantify important modifiers of the aging process. Chief among these modifiers is the diagnosis of frailty [4–6].

While frailty is an easy concept to grasp, providing an exact definition is more tenuous and is beyond the scope of this chapter; however, we refer the reader to Chaps. 4 and 6 for a more thorough review of the concept of frailty and its application to perioperative care [4–8]. There are also several guidelines available to assist in the perioperative care of the geriatric patient for hip surgery both emergently and electively [7–11].

Finally, there are a few reviews that have examined the interaction or intersection of the Perioperative Surgical Home (PSH) as well as ERAS with the various Frailty scales and measures [12]. The application and expansion of the PSH as a concept has resulted in the development of several guidelines and protocols for the management of hip fracture in the elderly most notably in the UK [13]; however, the propagation of these guidelines was assisted by the development of an active surveillance database in use for over a decade [14, 15].

Interestingly, the initial guidelines [14] were designed following a Cochrane Review of outcomes following emergent hip fracture surgery [16] and include a recommendation for regional anesthesia (specifically subarachnoid anesthesia ) even though this same review noted, “The effect of the removal of the oldest trial (McLaren 1978), which has an excessive mortality in the general anaesthesia group, also shows the weakness of the evidence.” Despite this comment as well as others suggesting that there were issues in the review, it served as the basis for the guidelines evaluated in two recent Anaesthesia Sprint Audits of Practice (ASAP) [13, 17]. While the exact guidelines may or may not be ideal suggestions, the framework of those guidelines act as an excellent roadmap for examining important aspects of anesthetic care for hip fracture patients.

The ASAP practice standards outline twelve standards for anesthetic practice [13]. While the first standard is not relevant to this chapter, those from two onward are.

Standard 2 – Spinal or epidural anaesthesia should be considered for all patientsStandard 11 – Hypotension should be avoided

Standard 2 seems to be the most controversial of the standards suggested. The choice of either anesthetic category, general or regional, as a “safer” technique is not a universally accepted tenet. There are several papers and reviews regarding this subject that have been referred to previously, and almost all the reviews from the twenty-first century can find no difference in outcome regarding the selection of anesthetic approach. More importantly, whatever approach is chosen, it should be one that is familiar to the provider and provides for scrupulous attention to blood pressure management [17].

The management of blood pressure in the geriatric population is an important variable in determining outcome as has been suggested by a variety of studies throughout the last decade [18, 19]; however, there remains at least one major question. Regardless of the chosen level of hypotension (i.e., MAP < 55, MAP < 70, SBP < 20% below awake, etc.), the relationship between the chosen blood pressure and the chosen outcome (generally mortality, cardiac or neurological injury) has not been shown to be causal, only related. One possible hypothesis is that those patients with lesser hemodynamic reserves are the most likely to suffer hypotensive episodes and would also be more likely to suffer further insults over time. Developing hypotension in response to an anesthetic may simply be a biomarker for this poor reserve. Thus, while the avoidance of hypotension remains a paramount concern for anesthetic personnel, this may or may not reduce the likelihood of current or future events. This in no way should suggest therapeutic nihilism, but simply that we need to focus our attempts on examining the role of avoiding hypotension directly instead of looking for surrogate markers for poor outcomes.

Standard 3 – Spinal anesthetics should be administered using hyperbaric bupivacaine (< 10mg) with the patient positioned laterally (bad hip down)

Standard 4 – Co-administration of intrathecal opioids should be restricted to fentanyl

These standards suggest that if one wishes to use spinal anesthesia, reducing the dose of bupivacaine to less than 10 mg reduces hypotension [17, 20]. There is also a strong suggestion that hypobaric spinal techniques be avoided for the same reason (hypotension) [20, 21]. Adding fentanyl to the intrathecal mixture allows for improved postoperative analgesia with fewer issues of delirium, sedation, and respiratory depression. However, there is little direct evidence that fentanyl improves outcomes in hip fracture patients, so this recommendation represents a significant research opportunity. The Sprint Audit [13] demonstrated that fentanyl was used in only 32% of cases with the majority (~50%) adding diamorphine. Thus, it seems that many anesthetists do not follow this practice which suggests that there should be a room for further exploration.

Standard 5 – If sedation is required, this should be midazolam or propofol

The advantage of both propofol and midazolam lie primarily in their pharmacokinetic profiles and their wide safety margins when used in the geriatric population [22]. There is a general sense that geriatric patients tend to meet discharge criteria post sedation more quickly following propofol compared to midazolam; however, the data show small absolute differences (17.6 vs. 10.1 min for midazolam vs. propofol, respectively); thus, this may not be relevant clinically [22]. This finding is similar to that of their younger brethren (10.4 vs. 4.2 min for midazolam vs. propofol, respectively) [23]. Intraoperative amnesia is more complete following the use of midazolam [23], but whether this is a crucial outcome to the geriatric patient is not clear (patient satisfaction scores of 4.6 vs. 4.7 for midazolam vs. propofol, respectively) [22]. In the Sprint Audit [13], oversedation was common and may have contributed to hypotension; thus, tight control of sedation level is necessary to avoid this outcome. Further, the Audit also suggested that the use of propofol was associated with a reduced incidence of postoperative confusion compared to benzodiazepines and opiates [13].

Ketamine is frequently used for sedation during spinal anesthesia primarily for its salutary hemodynamic effects. Unfortunately, there is a fine line between the dosing for sedation and the avoidance of postoperative confusion [13]. It has been suggested that when combined with general anesthesia at a dose of 0.5 mg/kg, ketamine does not increase the incidence of postoperative cognitive dysfunction (POCD) at days 1 and 6 [24].

Standard 6 – Supplemental oxygen should always be provided

The use of supplemental oxygen is based on several observations. The first is that the implementation of spinal anesthesia is associated with sedation independent of anesthetic agents [25, 26]. In addition, regional oxygen saturation falls below baseline levels in patients receiving subarachnoid anesthetics with or without supplemental sedation [27]. Thus, the addition of supplemental oxygen seems both prudent and perspicacious. Further, because regional cerebral oxygen saturations are associated with that of peripheral oxygen saturations [28], the use of supplemental oxygen in concentrations higher than that obtained with nasal cannula is highly recommended.

Standard 7 – Inhalational agents should be considered for the induction of general anaesthesia.

This standard could be interpreted exactly as it is written, or with some license, it could also be interpreted as an admonition to avoid excessive administration of anesthetic agents and use a deliberate and watchful induction technique. These authors prefer the latter interpretation. Indeed, the outcome of the Audit suggests that most anesthetists also believe in the latter interpretation [13]. Fully 93% of those audited pursued an intravenous induction rather than an inhalational one. We are sure that if the question, “Did you consider the use of inhalational agents for induction?” was asked, most of that 93% would say, “Sure, I considered it for about 10 s and then reached for my trusted intravenous agent.” Slow gentle inhalational inductions with sevoflurane are hemodynamically more stable than rapid intravenous inductions by both the nature of the rapidity of the transition from awake to anesthetized as well as the maintenance of spontaneous ventilation (see Standard 8). The important take away for the readers is that one should use the method most familiar to them with the caveat that there are well-established nomograms and guidelines for the reduction in dosing of anesthetic agents in the geriatric population [8, 29, 30].

Standard 8 – Spontaneous ventilation should be used in preference to mechanical ventilation

This is also a controversial recommendation as there are multiple reasons to select endotracheal management (ET) in preference to either LMA or mask supplementation. ET management reduces the risk of aspiration and allows for rapid control of the airway should the patient require urgent intervention. While spontaneous ventilation is not impossible with ET management, it increases both the work of breathing and the risk of hypoventilation for this reason (unless supplemented with pressure support). Spontaneous ventilation does allow for enhanced matching of ventilation and perfusion and is generally associated with decreased degrees of hypotension.

In the recent Audit [13], this recommendation was not as controversial as the previous standard but was clearly not followed in all or even most cases. Among those patients who received general anesthesia with an ET tube (44.2% of cases), 81% were paralyzed and mechanically ventilated, 9% were non-paralyzed but mechanically ventilated, 9% were not recorded or other, and in NONE of the cases, spontaneous ventilation was used. In those patients, whose airway was managed with an LMA (51% of cases), spontaneous ventilation was used in 73% of those cases, non-paralyzed but mechanically ventilated in 13%, and paralyzed and mechanically ventilated in slightly less than 9%. This suggests that less than half of all patients were allowed to breathe spontaneously.

Standard 9 – Consider intraoperative nerve blocks for all patients undergoing surgery

The use of peripheral nerve blocks (PNB) for all types of surgery and all ages including the geriatric group is increasing worldwide [31]. The chief advantage of these approaches is the reduction in the need for parenteral and oral opiates for managing analgesia. However, when they are placed immediately prior to surgery they also reduce the dose of anesthetic needed and can accelerate the rate of discharge from the PACU or ambulatory surgery [31]. Further, they can also assist in positioning the patient for subarachnoid anesthesia if placed prior to administration.

In the Audit [13], PNBs were used in 56% of patients and most (54%) were administered without the need for either ultrasound guidance or nerve stimulation. This was due in part to the use of fascia iliaca block in 56% of patients, instead of the more traditional (in the US) 3-in-1 (lateral cutaneous, obturator, and femoral nerve) or psoas compartment block. The fascia iliaca block, while not providing comparable analgesia to the 3-in-1 block, is easier to perform using landmark techniques, and this may explain its more common appearance in the Audit. Ultrasound guidance was used in 26% of cases in the Audit. What is most interesting is the very wide variation in the use of PNBs in the hospitals audited [13], ranging from 8% to 92%.

Standard 10 – Neuraxial and general anaesthesia should not be combined

While this technique is frequently used in younger and healthier patients, it is not appropriate except under very select circumstances in the geriatric population. The incidence of hypotension is higher than with either technique alone [13]. The incidence of hypotension overall was very high depending upon the definition. The Audit analyzed hypotension using eight different definitions: fall in systolic blood pressure of greater than 20 or 30%, lowest systolic blood pressure less than 90 or 100 mmHg, fall in mean arterial pressure greater than 20 or 30%, and mean arterial pressure of less than 70 or 55 mmHg.

Using these definitions, the combination of general anesthesia and subarachnoid anesthesia resulted in a prevalence of hypotension of 47–93%. With subarachnoid anesthesia alone, hypotension ranged from 22 to 85% compared to the prevalence rate for all anesthetics that ranged from 32 to 89%. The incidence of hypotension for the general anesthesia group was similar in both magnitude and direction compared to the combined group but it was not quite as severe, ranging from 40 to 92%. These data again reiterate the reasoning behind the preference for subarachnoid over general anesthesia as regards the avoidance of hypotension.

Standard 12 – Patients should be routinely assessed for the occurrence of Bone Cement Implantation Syndrome (BCIS)

The incidence of symptomatology compatible with the diagnosis of BCIS varies across hospitals and across countries [32]. A generally accepted definition of BCIS did not exist prior to this publication by Donaldson et al. [32]. Their definition includes “hypoxia, hypotension or both and an unexpected loss of consciousness occurring around the time of cementation, prosthesis insertion, reduction of the joint or, occasionally, limb tourniquet deflation in a patient undergoing cemented bone surgery” [32]. Their group also proposed a grading system for the severity of the reaction: Grade 1 is characterized by a fall in SpO2 to less than 94% or a fall in systolic blood pressure of 20% or more. Grade 2 is characterized by fall in SpO2 to less than 88% or a fall in systolic blood pressure of 40% or more or an unexpected loss of consciousness. Grade 3 is characterized by cardiovascular collapse requiring CPR [32].

Using these criteria, a separate study from Sweden [33] performed a retrospective analysis in 1016 patients undergoing cemented hemiarthroplasty. The incidence rates of BCIS Grades 1, 2, and 3 were 21%, 5%, and 1.7%, respectively. More importantly, early mortality was related to the severity of the grade. Overall perioperative mortality was 2% which is similar to the range reported in other large studies (1.3–2.5%) [34, 35]. Although there was no difference between the absence of vs. Grade 1 symptoms (5.2% vs. 9.3%, respectively), early mortality with Grade 2 symptoms was 33% and with Grade 3, 88% [33].

However, the role or importance of the syndrome in the long-term outcome of patients is disputed [36, 37]. The primary reason for the dispute is that the functional outcomes for cemented prostheses are felt to be superior to that from the non-cemented version [36, 37]. Thus, many now focus on identifying those patients at highest risk for morbidity and mortality from BCIS as a critical step in improving the safety of hip surgery [38, 39]. Both articles have identified similar risk stratifications regarding BCIS: cardiopulmonary compromise, particularly focused on drugs that suggest compromised cardiac reserve (diuretics, beta-blockers, ACEi); age, frailty was not measured or assessed in these reports; male sex, possibly related to the size of the femoral medullary canal, ASA 3 or 4 status, which is likely a marker for comorbidities; and, finally, hypotension/hypovolemia immediately preceding the insertion of cement.

Providers (geriatricians, anesthesiologists, surgeons) should also discuss with each other plans for managing patients who present with these markers. Clearly discussing the influence each of these risk factors will have on the proposed surgical, anesthetic, and postoperative approach will insure the optimal outcome for each patient. Monitoring hemodynamic status more invasively, while not conclusively shown to change outcome, allows for faster diagnosis and a more tailored therapeutic approach. As the old saying goes, “forewarned is forearmed.”

For most geriatric patients, it seems prudent to place an arterial catheter prior to the initiation of surgery. This serves the purpose of providing beat-to-beat analysis of blood pressure and the ability to rapidly assess the status of arterial blood gases if necessary. Some form of monitoring of cardiac output is also essential to tailoring treatment as most investigators report a drop in cardiac output with the onset of BCIS . The type of cardiac output monitor can take the form of an esophageal Doppler, transesophageal echocardiography, pulmonary artery catheter, or pulse contour devices [40]. Each of these approaches has advantages and disadvantages, but the chief defining characteristic is whether the device can be used in non-intubated, sedated patients (PA catheter and pulse contour devices). Of course, it should go without saying that all standard ASA recommended monitoring is in place prior to initiating the anesthetic.

Treatment for BCIS is directed at the primary probable cause of the hemodynamic derangement. While the exact etiology is not clearly defined, a constellation of physiologic alterations result including: an increase in pulmonary vascular resistance, increase in pulmonary artery pressure, decrease in right ventricular function, decrease in cardiac output, decrease in stroke volume, decrease in SpO2, and an increase in V/Q mismatch [38, 39, 41]. While the putative cause of most problems is related to a combination of embolic phenomena of one sort or another (fat, cement, bone, air) and activation of a variety of vasoactive substances (histamine, complement, cytokines, etc.) acting primarily on the right side of the heart [38], treatment is directed at increasing systemic blood pressure, increasing stroke volume and cardiac output.

Preventive volume loading and augmentation of inspired oxygen concentration immediately prior to cementation in high-risk patients combined with monitoring with a CVP or PA catheter is essential to successful management [32]. Management of hypotension can be accomplished with a variety of vasoactive drugs including phenylephrine, norepinephrine, and vasopressin for increasing systemic vascular resistance; epinephrine and dobutamine for increasing cardiac output; and if a pulmonary artery catheter is in place, milrinone could be used for pure right ventricular overload and failure. The latter compound however is a significant vasodilator and should rarely be used in this scenario without evidence of isolated right ventricular overload (high CVP, tricuspid regurgitation or poor right ventricular function, and an under-filled left ventricle as imaged on echocardiography), and even then, it is best used in combination with a vasoconstrictive agent.

The use of blood and blood products has become more controversial over the last decade. Originally, a more liberal (definitions vary but generally means transfusion for hemoglobin concentrations of less than 10 gm/dl) policy was used in the elderly. The prevailing belief was that the higher incidence of comorbidities (primarily cardiovascular and pulmonary) and a desire to rapidly regain functional status required a higher oxygen-carrying capacity [42].

However, in 2011, a large multicenter study (FOCUS – Functional Outcomes in Cardiovascular Patients Undergoing Surgical Hip Fracture Repair) from the NIH strongly suggested that this was not the case [43]. The study was carried out in 2016 patients over the age of 50 with a history of or risk factors for cardiovascular disease and a hemoglobin level of less than 10 gm/dl. Patients were then assigned to either a liberal (threshold of 10 gm/dl) or restrictive (threshold of less than 8 gm/dl) transfusion strategy. The primary outcome was mortality or inability to walk across a room without human assistance on 60-day follow-up. The average age of their participants was approximately 82 years and approximately one-quarter of the participants were male; there were no differences between groups regarding the type or extent of cardiovascular risk factors, type of fracture, type of anesthetic, or primary residence (approximately 88% in both groups were in a retirement home or at home). Likewise, there were no differences in hemoglobin prior to surgery (average of 11.3 ± 1.5) or at entry into the study (9.0 ± 0.8); however, blood loss was slightly and statistically (though not clinically relevant) greater in the restrictive group (209 ± 179 vs. 232 ± 257, respectively).

Fifty-nine percent of patients in the restrictive group did not receive transfusions, while only 3.3% of patients in the liberal group were not transfused. Compared to 54.9% of patients in the liberal group, 16.6% of patients in the restrictive group received 2 or more units of red cells. There was no difference in the age of the units transfused or the use of leukoreduction. The primary reason for transfusion in the restrictive group was tachycardia or hypotension. At 30 days, 46.1% of patients in the liberal group and 48% of patients in the restrictive group met the criteria for the primary endpoint (death or inability to walk), a nonstatistically significant difference. At 60 days, the percentages had decreased (35.2% and 34.7%, respectively), but there remained no statistical difference between the two groups. Mortality rates for these same two time periods were 5.2% and 4.3% for the liberal vs. restrictive groups and 7.6% and 6.6%, respectively.

This same finding was confirmed by at least two further studies [44, 45]. The first trial examined functional outcomes in 305 patients, but did not prospectively group patients by a transfusion strategy, but rather measured their ability to walk in a predetermined amount of time (6 min), maximal hand strength, and two measures (SF36 and CR10) for QoL (Quality of Life) following either hip or knee arthroplasty [44]. Patients were assessed preoperatively and again on postoperative days 1–10 where they completed the SF36 form and were asked to walk as far as possible in 6 min. They were then asked to assess their level of effort during the walk on a CR10 scale [46], and finally their grip strength was measured in their dominant hand. Patients were grouped according to their hemoglobin value on the day of their postoperative visit into four groups: ≤ 8, 8–9, 9–10, and ≥10 gm/dL. There were no differences in the four outcome variables across the four groups except for grip strength as the percentage of males in the ≥10 group was significantly higher (47% vs. 29%, 19%, and 32%, respectively). Most patients were examined postoperatively between days 4 and 5 (4.6 ± 1, 4.5 ± 1.5, 4.8 ± 1.5, and 4.6 ± 1.7 respective to Hgb group). While there were significant decreases over time in each of the groups, they all performed equally well compared to their preoperative states. Further, there were no significant differences between the groups with respect to adverse events (cardiac and respiratory), symptoms of anemia, length of stay, or incidence of prolonged hospital stay.

The second trial involved 603 patients who were prospectively randomized to a restrictive or liberal transfusion strategy [45] and followed for 14 days following operation. Outcome measures included complications (infectious, respiratory, neuropsychiatric, cardiovascular, and hemorrhagic), mobilization delay, QoL (FSI or Functional Status Index), and mortality. Demographic criteria were balanced across groups apart from a history of COPD which was higher in the restrictive group (incidence of 10.7 vs 4.6%). As expected, the number of transfused patients was smaller in the restrictive group (26.4% vs. 39.1%). There was no difference in hospital stay or median blood loss between groups. Infectious and respiratory complications occurred more frequently in transfused patients regardless of categorical assignment. Of those patients who developed infections, 66% had been transfused, while 70% of patients with respiratory complications were transfused. QoL scores were not affected by transfusion strategy.

Although these studies would appear to conclude the debate about where transfusion triggers should be set, a Danish study was published in 2016 that has reignited the debate [47]. This paper is a composite of three papers published as part of a thesis for PhD [48–50]. The three papers sought to examine the role that frailty and not simply age plays in responding to transfusion strategy following surgery for hip fracture in 284 patients.

The patients were drawn from two populations: one in nursing homes and the other in sheltered living facilities. The two groups were matched across a wide variety of demographic factors including but not limited to ADLs, gender, residence, comorbidity, dementia, age, and pre- and intraoperative transfusion. The only statistically significant difference was in age which was not clinically significant (85.7 vs. 86.9 for restrictive vs. liberal, respectively). The restrictive group was transfused at a level of 9.7 gm/dL and the liberal group at 11.3 gm/dL. This is an important distinction from almost all the other studies we have discussed. The restrictive group is being transfused at a level that would generally be “liberal” in almost all the other studies.

Thus, the first significant question to ask is to what degree are the results reflective of a comparison of essentially two different liberal transfusion strategies? Essentially, they created “more” and “less” liberal transfusion groups with a relatively small number of patients. They did find that their frailest patients were from nursing homes (interestingly, however, the incidence of dementia was not different between the two residency groups) and that these patients had the higher survival rate in the more liberal group (36% vs. 20% at 90 days). Further, 30-day mortality was significantly lower in all patients in the more liberal group (7% vs. 16%). There is a caveat to these findings, however, as they evaluated their outcomes with respect to both an intention to treat and on a per protocol basis.

The per protocol group was smaller than the intention to treat group with only 260 patients in total. While the 90-day mortality was higher in the restrictive group in both analyses, the 30-day mortality for all patients was only significantly different in the per protocol analysis. Also, they did not find an increase in infections with the more liberal group which other investigators have noticed. Overall, the most important findings from this study is the outcomes as related to frailty rather than simply age. Unfortunately, the use of relatively high values for the “restrictive” group made the ability to relate this study to so many others in the literature very difficult.

Thus, one is left with the impression that unless the patient has pre-existing coronary artery or severe pulmonary disease, the restrictive strategy appears to be as safe as the more liberal strategy. Further, if one lives or works in an environment where blood and blood products are expensive or difficult to locate, then the restrictive strategy can conserve these previous resources at no physiologic expense to the patient.

As noted, many (but not all) of the ERAS pathways suggest use of PNBs to reduce the need for intraoperative analgesia and anesthesia (if general anesthesia is used) or to enhance the postoperative analgesic management and reduce the reliance on opioids. The innervation of the skin around the knee and surrounding tissue comes from the femoral nerve, obturator nerve, and sciatic nerve (the last as two branches – the tibial and common peroneal nerves). The joint space is innervated by the femoral nerve anteriorly and the obturator and sciatic nerves posteriorly.

A very recent paper [57] has examined the use of a variety of different approaches for providing postoperative analgesia including PNBs , periarticular infiltration, and epidural analgesia. The authors identified 170 trials published between 1987 and 2016, encompassing over 12,500 patients and utilizing 17 different treatment modalities. They evaluated these modalities for three primary outcomes: acute postoperative pain during rest and movement, postoperative opioid consumption, and quality of early postoperative rehabilitation (range of motion combined with degree of flexion). Secondary outcomes included postoperative incidence of nausea, vomiting, pruritus, urinary retention, and DVT, LOS, and blood loss.

Approximately 59% of the trials (121) used some version of neuraxial anesthesia, but the clear majority of these (87 of 121) used only subarachnoid anesthesia . Of the 170 trials, 57 used general anesthesia (7 TIVA and the remainder volatile with 16 of the latter including N₂O). Seventy-one of the trials used acetaminophen with or without NSAIDs, while 9.4% used some form of gabapentin, and 24 trials (~14%) did not specify.

All forms of combined PNBs were superior to any single nerve block for analgesia. The cumulative ranking curves were different based on the primary outcome examined. The top five methods of analgesia for each primary outcome were summarized over the first 72 h: pain at rest, femoral/obturator, femoral/sciatic/obturator, lumbar plexus/sciatic, femoral/sciatic, and the fascia iliaca compartment block; range of motion, femoral/sciatic, femoral/obturator, femoral, lumbar plexus, and periarticular infiltration; reduction in opioid consumption, femoral/sciatic/obturator, femoral/obturator, lumbar plexus/sciatic, lumbar plexus, and femoral/sciatic; and pain with movement, femoral/obturator, intrathecal morphine, femoral/sciatic, periarticular infiltration, and lumbar plexus/sciatic.

Secondary outcomes showed similarly disparate results depending upon the outcome examined. The incidence of nausea was lowest with auricular acupuncture followed by femoral/obturator, lumbar plexus/sciatic, femoral/sciatic, and adductor canal block. The incidence of vomiting on the other hand was lowest with liposomal bupivacaine followed by femoral/obturator, periarticular infiltration, femoral, and femoral/sciatic. Pruritus was lowest with the lumbar plexus/sciatic block followed by auricular acupuncture, femoral, femoral/sciatic, and periarticular infiltration. Finally, the incidence of urinary retention was lowest with auricular acupuncture followed by lumbar plexus, lumbar plexus/sciatic, femoral/sciatic, and femoral. Length of stay was shortest with the adductor canal block followed by lumbar plexus/sciatic, periarticular infiltration, liposomal bupivacaine, and placebo. Finally, the incidence of deep venous thrombosis was lowest with femoral/sciatic blocks followed by placebo, epidural anesthesia, adductor canal block, and periarticular infiltration.

Perhaps the most interesting finding in this meta-analysis is the fact that auricular acupuncture placed in the top two in three of the six secondary outcomes measures. In fact, it was the top performer in two categories, lowest incidence of nausea and urinary retention, and was second in pruritus. The only PNB that placed consistently in the top five was the femoral/sciatic, placing in five of the six secondary outcomes. The lumbar plexus/sciatic was a close second placing in the top five in four of six secondary outcomes as did periarticular infiltration.