Effects of General Anesthesia Plus Multimodal Analgesia on Immediate Perioperative Outcomes of Hamstring Tendon Autograft ACL Reconstruction

A Randomized, Double-Blinded, Placebo-Controlled Trial

Abstract

Background: 

Anterior cruciate ligament reconstruction with hamstring tendon autograft (H-ACLR) is a standard ambulatory procedure with the potential for considerable postoperative pain. We hypothesized that general anesthesia combined with a multimodal analgesia regimen would reduce postoperative opioid use associated with H-ACLR.

Methods: 

This study was a single-center, surgeon-stratified, double-blinded, placebo-controlled, randomized clinical trial. The primary end point was the total postoperative opioid use during the immediate postoperative period, and secondary outcomes included postoperative knee pain, adverse events, and ambulatory discharge efficiency.

Results: 

One hundred and twelve subjects, 18 to 52 years of age, were randomized to placebo (57 subjects) or combination multimodal analgesia (MA) (55 subjects). The MA group required fewer opioids postoperatively (mean ± standard deviation, 9.81 ± 7.58 versus 13.88 ± 8.49 morphine milligram equivalents; p = 0.010; effect size = −0.51). Similarly, the MA group required fewer opioids within the first 24 hours postoperatively (mean ± standard deviation, 16.56 ± 10.77 versus 22.13 ± 10.66 morphine milligram equivalents; p = 0.008; effect size = −0.52). The subjects in the MA group reported lower posteromedial knee pain (median [interquartile range, IQR]: 3.0 [0.0 to 5.0] versus 4.0 [2.0 to 5.0]; p = 0.027) at 1 hour postoperatively. Nausea medication was required for 10.5% of the subjects receiving the placebo versus 14.5% of the subjects receiving MA (p = 0.577). Pruritis was reported for 17.5% of subjects receiving the placebo versus 14.5% receiving MA (p = 0.798). The median time to discharge was 177 minutes (IQR, 150.5 to 201.0 minutes) for subjects receiving placebo versus 188 minutes (IQR, 160.0 to 222.0 minutes) for those receiving MA (p = 0.271).

Conclusions: 

A combination of general anesthesia and local, regional, oral, and intravenous multimodal analgesia appears to reduce postoperative opioid requirements after H-ACLR compared with placebo. Adding preoperative patient education and focusing on donor-site analgesia may maximize perioperative outcomes.

Level of Evidence: 

Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence.

Opioid use in patients undergoing orthopaedic surgery is a notable contributor to the current global opioid epidemic¹. In fact, orthopaedic surgeons are among the most frequent prescribers of opioid medications². However, adequate analgesia and postoperative pain control are essential to a successful surgical outcome, requiring the collaboration of the patient, orthopaedic surgeon, and anesthesiologist³. Anterior cruciate ligament reconstruction (ACLR) is a common orthopaedic surgery performed in an ambulatory setting, with the potential for considerable postoperative pain^4–6. Poorly controlled pain following ACLR is associated with increased utilization of health-care resources^5,7. One study found that acute pain was the most frequent complication after ACLR in the ambulatory setting⁸. Because of these issues, opioids are traditionally prescribed postoperatively⁹, thus providing a valuable setting for interventions aiming to reduce opioid medication use.

While regional blockade and local anesthetic injection into the knee joint have been utilized to control postoperative pain after ACLR using patellar tendon autograft¹⁰, studies have demonstrated that they are not particularly effective for managing the pain associated with hamstring tendon autograft ACLR (H-ACLR)^4,5,11. One study found that local infiltration analgesia is comparable with femoral nerve block (FNB) after H-ACLR and concluded that further randomized clinical trials evaluating FNB combined with local infiltration at the donor site are needed⁶. Therefore, an analgesia regimen for patients undergoing H-ACLR that reduces postoperative opioid use remains to be established.

In alignment with a multidisciplinary opioid-reduction program initiated at our medical campus to reduce opioid use after orthopaedic surgery, we aimed to determine the effects of general anesthesia combined with a multimodal analgesia regimen (combination multimodal analgesia) for H-ACLR on the postoperative outcomes associated with pain and opioid use. We hypothesized that a combination of general anesthesia and local, regional, oral, and intravenous (IV) analgesia would reduce postoperative opioid use associated with H-ACLR.

Materials and Methods

This research was conducted at the University of Wisconsin Hospitals & Clinics in Madison, Wisconsin. This study was approved by the university’s institutional review board (2017-0712-CR004), and written informed consent was obtained from all subjects participating in the trial. The trial was registered prior to patient enrollment at ClinicalTrials.gov (NCT01868425).

This study was a single-center, surgeon-stratified, double-blinded, placebo-controlled, randomized study conducted from April 2013 to December 2017. Patients undergoing H-ACLR were randomly assigned to receive standard-of-care general anesthesia plus either placebo or the multimodal analgesia regimen (MA). Eligible participants were adults aged 18 to 55 years who consented to H-ACLR. Patients were excluded if they had a contraindication to an FNB, allergy to the protocol medications, nervous system disease, renal or hepatic impairment, history of opioid dependence or current opioid use, or diagnosed psychiatric disease, were pregnant or lactating, or had a seizure disorder, history of postoperative nausea and vomiting, latex allergy, diagnosed cardiac or pulmonary disease, and/or body mass index of ≥40 kg/m². A total of 3 board-certified, fellowship-trained sports medicine orthopaedic surgeons performed all of the H-ACLR surgeries using the technique described by Walczak et al.¹². This manuscript adheres to the applicable CONSORT (Consolidated Standards of Reporting Trials) guidelines.

Interventions

After the anesthesiologist discussed the anesthetic care plan prior to surgery, the patients were approached by the study staff. Patients who met the inclusion criteria, agreed to participate, and provided consent were then randomized to 1 of 2 study arms: placebo or MA. Patients assigned to the MA group received an oral multimodal medication regimen, including 1,000 mg of acetaminophen and 300 mg of gabapentin preoperatively. Patients assigned to the placebo group were given placebo pills prepared by the hospital pharmacy that were similar in shape and size to the pills received by the MA group. All patients then received an ultrasound-guided FNB using 20 to 30 mL of 0.5% bupivacaine.

Monitors recommended by the American Society of Anesthesiologists (ASA)¹³ were placed, and the patients were pre-oxygenated with 100% oxygen via a facemask. Induction of general anesthesia included up to 2 mg of midazolam, 0.5 to 1 mcg of fentanyl, 40 to 100 mg of lidocaine, and 200 mg of propofol. After placing a laryngeal mask airway or endotracheal tube, 4 mg of dexamethasone was administered. IV antibiotics were administered within 30 minutes of incision. Once the patient was anesthetized, the lower extremity was aseptically prepared and draped, and the skin was incised.

Participants in the MA treatment arm received an IV bolus of 0.5 mg/kg of ketamine (up to 50 mg) prior to skin incision. Participants in the control arm received an IV bolus of normal saline solution, with a volume equal to that in the ketamine dosing parameters described above. Following the initial bolus, subjects in the treatment arm received a standardized, low-dose, 15-mg/hr ketamine infusion at a rate of 15 mL/hr. The participants in the placebo arm received normal saline solution at 15 mL/hr.

Immediately following the hamstring harvest, patients in the treatment arm received a local anesthetic infiltration of 20 mL of 0.125% bupivacaine administered by the surgeon using a flexible catheter inserted into the hamstring sheath, as previously described by Faunø et al.¹⁴ (see Appendix Supplemental Figure 1). Patients in the control arm received 20 mL of preservative-free normal saline solution.

During post-anesthesia care unit (PACU) phase-1 recovery, pain medication administration was standardized using a numeric rating scale (NRS) ranging from 0 to 10, where 0 = no pain and 10 = the worst pain possible. For an NRS pain score of 3 to 5, 0.1 to 0.3 mg of IV hydromorphone was given every 6 minutes as needed for pain control. For an NRS pain score of >5 to 7, 0.2 to 0.5 mg of IV hydromorphone was given every 6 minutes. Finally, for an NRS pain score of >7, 25 to 50 mcg of IV fentanyl was given every 5 minutes. The PACU phase-2 protocol for opioid administration was similarly standardized: for an NRS pain score of 4 to 7, 5 mg of oral oxycodone was given every 20 minutes, and for an NRS pain score of >7, either 5 to 10 mg of oral oxycodone or 0.25 to 0.5 mg of IV hydromorphone was given.

Outpatient pain management was standardized, with oral oxycodone in 5-mg tablets to be taken every 3 hours as needed, 10 mg of oral ketorolac taken 3 times daily for 5 days, and 1,000 mg of acetaminophen to be taken 3 times daily for 5 days. Patients kept a standardized pain diary after discharge until 24 hours after surgery.

Outcomes

The primary end point was the total postoperative opioid requirement, calculated as morphine milligram equivalents, during the immediate postoperative period. The “immediate postoperative period” was defined as the time from when the patient entered the PACU until discharge from the outpatient surgery center. In addition to evaluating the primary end point of total opioid consumption in the immediate postoperative period, we assessed total opioid use during the first 24 hours after surgery.

Secondary outcome measures included patient-reported pain scores during the initial 24 hours after surgery; the number of participants who received medication for nausea; the postoperative incidence and severity of nausea; the postoperative incidence and severity of pruritus; the severity of postoperative sedation; the number of patients who experienced a complication from the procedure; the impact of block characteristics on pain control; the intraoperative use of ketorolac, lidocaine, and fentanyl; and the time to discharge from phase-1 PACU and from phase-2 PACU.

Sample-Size Estimation

On the basis of PACU opioid use after outpatient surgery, we estimated that the mean opioid requirement for standard care would be 6.6 morphine milligram equivalents, with a standard deviation (SD) of 6.2 morphine milligram equivalents. A minimal sample size of 46 subjects per group was needed to achieve 80% power and a significance level of <0.05 for testing the ability of multimodal analgesia to reduce opioid use by >50%. In order to account for a potential randomization block failure of up to 20%, a total of 112 patients were recruited.

Blinding and Randomization

Randomization lists for each surgeon were created by the pharmaceutical research center (PRC) using a random number generator and forced blocks of 4. Patients were randomized to the 2 treatment arms in a 1:1 fashion, and subjects were stratified by surgeon prior to randomization. The PRC pharmacist was responsible for assigning randomization numbers and for the preparation, dispensing, and masking of medications for the operative and perioperative teams.

Statistical Analysis

Continuous data are presented as the mean and SD (normally distributed data) or as the median and interquartile range (IQR; non-normally distributed data), and categorical data are presented as the number and percentage. Normality of distributions was determined qualitatively by evaluating the residual quantile-quantile (Q-Q) plots and quantitatively by the Anderson-Darling (A2) statistical test. Normally distributed continuous variables were analyzed using the Student t test. Non-normally distributed continuous variables were analyzed using the Mann-Whitney U test. The Fisher exact test was used to analyze categorical data. A p value of <0.05 was considered significant. Calculations were performed using Prism 9.0.1 for Mac (GraphPad).

Source of Funding

No external funding was received for this study.

Results

Baseline Participant Characteristics

A total of 112 subjects, 18 to 52 years of age, met the inclusion criteria, consented to participate, and were randomized. Fifty-seven subjects received the placebo analgesia, and 55 subjects received MA (Fig. 1). Participants’ baseline characteristics are summarized by group in Table I.

Fig. 1: CONSORT (Consolidated Standards of Reporting Trials) diagram of patient enrollment, allocation, follow-up, and analysis. n = total number of subjects, ASA = American Society of Anesthesiologists, BMI = body mass index, FNB = femoral nerve block, SOC = standard of care.

TABLE I - Demographic Characteristics of Randomized Subjects, 2013 to 2017^*

Baseline Characteristic	Placebo (N = 57)	MA (N = 55)
Age† (yr)	32.16 ± 8.8	30.46 ± 8.5
BMI† (kg/m 2 )	26.46 ± 4.32	26.35 ± 4.13
Preop. NRS knee pain†	0.32 ± 0.81	0.20 ± 0.78
ASA class‡
1	39 (68.4)	40 (72.7)
2	18 (31.6)	15 (27.3)
3	0 (0)	0 (0)
Sex‡
Female	21 (36.8)	16 (29.1)
Male	36 (63.2)	39 (70.9)
Smoking status‡
Never	49 (86.0)	41 (74.5)
Former	5 (8.8)	5 (9.1)
Current	3 (5.3)	9 (16.4)

^*MA = multimodal analgesia, BMI = body mass index, NRS = numeric rating scale (from 0 to 10, where 0 = no pain and 10 = worse pain possible), and ASA class = American Society of Anesthesiologists physical status classification system.

^†Data are presented as the mean and standard deviation.

^‡Data presented as the number, with the percentage in parentheses.

Primary End Point and Postoperative Opioid Use

A total of 56 subjects (98.2%) in the placebo group and 52 subjects (94.5%) in the MA group completed the study’s opioid protocol and were analyzed. The MA group required, on average, 4.07 (95% confidence interval [CI], 0.99 to 7.15) fewer morphine milligram equivalents compared with the placebo group (p = 0.01) in the immediate postoperative period. This difference corresponded to a medium effect size (Cohen d = −0.51; 95% CI, −0.89 to −0.12) (Table II). Moreover, the MA group required a mean of 5.57 (95% CI, 1.48 to 9.66) fewer morphine milligram equivalents compared with the placebo group during the first 24 hours postoperatively (p = 0.008). This difference corresponded to a medium effect size (Cohen d = −0.52; 95% CI, −0.91 to −0.13). Subjects in the MA group were administered less IV hydromorphone in the PACU compared with subjects in the placebo group (Table III). The median oral oxycodone use following hospital discharge home was lower in subjects who received MA compared with placebo (see Appendix Supplemental Figure 2).

TABLE II - Comparison of Opioid Use in Patients After H-ACLR by Analgesia Type, 2013 to 2017^*

Parameter	Placebo (N = 56)†			MA (N = 52)‡			P Value
	Mean	SD	95% CI	Mean	SD	95% CI
PACU opioid requirement	13.88	8.49	11.61, 16.15	9.81	7.58	7.70, 11.92	0.010
Total 24-hour opioid use	22.13	10.66	19.27, 24.98	16.56	10.77	13.56, 19.56	0.008

^*The values are given as morphine milligram equivalents. MA = multimodal analgesia, SD = standard deviation, CI = confidence interval, and PACU = post-anesthesia care unit.

^†One patient did not complete the study protocol.

^‡Three patients did not complete the study protocol.

TABLE III - Comparison of Opioid Medications Administered in the PACU After H-ACLR by Analgesia Type, 2013 to 2017^*

Medication	Placebo			MA			P Value
	Median	IQR	Range	Median	IQR	Range
IV fentanyl (mcg)	0	0.0-0.0	0.0-100.0	0	0.0-0.0	0.0-50.0	0.285
IV hydromorphone (mg)	1.0	0.4-1.3	0.0-3.0	0.5	0.0-1.0	0.0-2.3	0.021
Oral oxycodone (mg)	5.0	1.3-10.0	0.0-15.0	5.0	0.0-10.0	0.0-15.0	0.102

^*PACU = post-anesthesia care unit, MA = multimodal analgesia, IQR = interquartile range, and IV = intravenous.

Secondary Outcomes

The predominant location of maximal pain for both groups was the posteromedial part of the knee (Table IV). However, the frequency of subjects reporting the posteromedial knee as the location of maximal pain was lower in the MA group than in the placebo group. Furthermore, the median pain intensity in the posteromedial knee was significantly less for subjects receiving MA compared with placebo (p = 0.027) (Table IV). There were no significant between-group differences in the frequency of pain scores of >3 and >5, which corresponded to the threshold levels requiring increased opioid administration. Similarly, the pain intensity in the anterolateral part of the knee did not differ significantly between the groups. NRS scores were similar between groups on postoperative day 1. There were no significant between-group differences in the amount of lidocaine, fentanyl, and ketorolac administered intraoperatively (see Appendix Supplemental Table 1).

TABLE IV - Comparison of Postoperative Pain in Patients After H-ACLR, 2013 to 2017^*

Parameter	Placebo (N = 57)	MA (N = 55)	P Value
Location of maximal pain (no. [%])
Posteromedial	46 (80.70)	32 (58.18)	0.013
Anterolateral	11 (19.30)	23 (41.82)
NRS pain score >3 (no. [%])
Yes	35 (61.40)	27 (49.10)	0.254
No	22 (38.60)	28 (50.90)
NRS pain score >5 (no. [%])
Yes	8 (14.04)	8 (14.55)	0.999
No	49 (85.96)	47 (85.45)
PACU at 1 hr postop.†
Posteromedial pain	4.0 (2.0-5.0) [0-7]	3.0 (0.0-5.0) [0-7]	0.027
Anterolateral pain	1.0 (0.0-2.5) [0-7]	1.0 (0.0-2.5) [0-6]	0.736
Home at day 1 postop.†
Current overall pain‡	3.0 (1.0-4.0) [0-7]	2.0 (1.0-3.0) [0-10]	0.482
Current posteromedial pain§	2.0 (0.5-4.0) [0-10]	2.0 (1.0-3.0) [0-5]	0.608
Current anterolateral pain§	1.0 (0.0-2.5) [0-6]	1.0 (0.0-3.0) [0-5]	0.923
Worst combined pain‡	4.0 (3.0-6.0) [1-10]	4.0 (3.0-5.0) [0-10]	0.585
Least combined pain‡	1.0 (0.0-2.0) [0-7]	1.0 (0.0-1.0) [0-4]	0.186

^*MA = multimodal analgesia, NRS = numeric rating scale (from 0 to 10; 0 = no pain, and 10 = worse pain possible), and PACU = post-anesthesia care unit.

^†The values are given as the median NRS score, with the interquartile range (IQR) in parentheses and the range in square brackets.

^‡Placebo: n = 57, and MA: n = 50.

^§Placebo: n = 49, and MA: n = 44.

Adverse Events and Ambulatory Discharge Efficiency

In the MA group, 8 (14.5%) of the subjects required treatment for nausea compared with 6 (10.5%) of the subjects in the placebo group (p = 0.577). One subject (1.75%) in the placebo group vomited postoperatively compared with 2 (3.64%) of the subjects who received MA (p = 0.615). Pruritis was reported by 10 (17.5%) of the subjects who received placebo compared with 8 (14.5%) of the subjects in the MA group (p = 0.798). Fifty-three (93.0%) of the subjects in the placebo group reported some level of sedation compared with 54 (98.2%) of the subjects in the MA group (p = 0.364). There were no significant differences in the severity of the side effects between the groups (Table V). Three subjects in the placebo group had complications possibly related to the anesthesia, including a nightmare (1), lightheadedness while walking (1), and headache (1). All resolved with observation in the PACU. The median time to discharge from phase-1 to phase-2 PACU was 60 minutes (IQR, 49.0 to 71.5 minutes) for the placebo group compared with 62 minutes (IQR, 52.0 to 74.0 minutes) for the MA group (p = 0.279). The median time to discharge from phase-2 PACU to home was 114 minutes (IQR, 93.5 to 141.5 minutes) for the placebo group compared with 121 minutes (IQR, 92.0 to 148.0 minutes) for the MA group (p = 0.671). The median total time to discharge (PACU 1 + PACU 2) was 177 minutes (IQR, 150.5 to 201.0 minutes) for subjects receiving placebo versus 188 minutes (IQR, 160.0 to 222.0 minutes) for those receiving MA (p = 0.271).

TABLE V - Comparison of the Severity of Perioperative Side Effects After H-ACLR by Analgesia Type, 2013 to 2017^*

Parameter	Placebo (N = 57)			MA (N = 55)			P Value
	Median	IQR	Range	Median	IQR	Range
PACU
Nausea	3.0	1.3-4.8	1-8	5.0	1.0-6.0	1-8	0.549
Sleepiness/sedation	5.0	4.0-6.0	1-10	5.0	3.0-7.0	1-9	0.513
Pruritis	3.5	2.0-5.5	1-7	2.5	1.3-3.8	1-4	0.158
Home at day 1 postop.
Nausea	4.0	1.8-7.0	1-10	3.5	2.0-7.0	1-9	0.711
Sleepiness/sedation	5.0	4.0-6.0	1-10	5.0	4.0-7.0	0-10	0.441
Pruritis	2.0	2.0-6.0	1-8	2.0	2.0-4.0	1-6	0.707

^*Data represent patient-reported intensity on a numeric rating scale (from 1 to 10; 1 = least possible, and 10 = worst possible). MA = multimodal analgesia, IQR = interquartile range, and PACU = post-anesthesia care unit.

Discussion

Postoperative pain management in patients undergoing outpatient ACLR remains challenging¹⁵. Multimodal analgesia employs drugs that act by different mechanisms on separate pathways to reduce nociception¹⁶. Commonly, multimodal analgesia includes a local anesthetic injection^17–19 or a regional anesthetic blockade^20–23. Collectively, these strategies aim to improve postoperative pain control and facilitate patient mobility compared with traditional opiate-based methods^6,7. Here, we compared general anesthesia and a multimodal regimen that included local, regional, oral, and IV analgesia to general anesthesia and placebo in patients undergoing H-ACLR.

With respect to our primary outcome measure, this study demonstrated that, compared with placebo, combined general anesthesia and MA reduced opioid requirements postoperatively. These findings are consistent with previous research in which multimodal analgesia regimens were used^24,25. For example, in a recent systematic review and meta-analysis of Level-I randomized clinical trials, Maheshwer et al. found that multimodal analgesia regimens were associated with significantly improved pain control after ACLR in the first 24 hours after surgery²⁵. It should be noted that this research primarily focused on studies that compared different types of regional anesthesia rather than other specific multimodal analgesia regimens and did not include any study that used a combination of regional, local, and oral approaches²⁵. Similarly, a retrospective comparative study demonstrated that a local anesthetic infiltration into the hamstring sheath in H-ACLR was associated with lower postoperative pain scores than were found for patients who did not receive the anesthetic infiltration¹⁸. However, that study did not examine postoperative opioid use and only assessed patient-reported pain scores during the first 2 hours after surgery¹⁸. This is consistent with our previous preliminary data analysis that aimed to examine the specific effects related to the intraoperative infiltrative hamstring anesthetic²⁶. That separate analysis found that the hamstring sheath anesthetic was associated with lower postoperative pain but did not significantly reduce opioid requirements, suggesting that the MA is primarily modulating the need for opioid medication²⁶. The inability of local anesthetic alone to modulate postoperative opioid requirement is consistent with a recent nonrandomized prospective study that evaluated a multimodal analgesia protocol to reduce opioid use after different ambulatory orthopaedic procedures²⁴. However, a randomized controlled study examining the analgesic effect of a hamstring tendon blockade found that adding an intraoperative hamstring tendon block reduced postoperative opioid requirements at 6 hours after surgery¹⁴. As in the previous retrospective comparative study¹⁸, the results were limited to only the immediate postoperative phase-1 PACU period¹⁴.

Regarding our secondary end points, we found that the subjects in the MA group had less posteromedial knee pain postoperatively than the placebo group, corroborating the findings of a retrospective comparative study examining postoperative pain control by a donor-site block for H-ACLR¹⁸. Conversely, a recent randomized clinical trial comparing an ultrasound-guided peri-hamstring injection or anterior obturator nerve block to no regional analgesia (control) after H-ACLR found no difference in postoperative knee pain or opioid requirements within the first 24 to 48 hours after surgery⁵. While the authors found better analgesia coverage in the nerve-block group, the failure to detect a difference in postoperative pain reduction in their study may be due to the preponderance of relatively low numeric pain scores, making the identification of important group differences difficult⁵. These authors also found worse postoperative knee pain, regardless of group, at the donor site, which is consistent with our data⁵. In the current study we found that MA was not associated with an increased incidence or severity of adverse events or a reduced discharge efficiency.

These results, taken together, suggest that a combination regional, local, oral, and IV analgesia regimen can effectively reduce postoperative opioid requirements for at least the initial 24 hours and decrease posterior knee pain during the immediate postoperative period.

Limitations

While this study was well powered a priori to identify the effects of MA on total postoperative opioid requirements, we could not determine whether the results were primarily the responsibility of the hamstring tendon sheath injection, regional FNB, or perioperative medication regimen. Additionally, 1 subject in the placebo group and 3 in the MA group were randomized but did not receive opioids per the study protocol, reducing the number of subjects available for analysis. Furthermore, the mean morphine milligram equivalents used by subjects were substantially higher than those predicted in the power analysis. The combination of a greater-than-expected mean opioid requirement and a threshold minimum number of subjects could lead to an overestimated effect, which amounted to less than one 5-mg oxycodone tablet or one 25-mcg dose of fentanyl over the first 24 hours postoperatively.

Conclusions

A combination of general anesthesia and local, regional, oral, and intravenous multimodal analgesia appears to reduce postoperative opioid requirements after H-ACLR compared with placebo. Adding preoperative patient education and focusing on donor-site analgesia may maximize perioperative outcomes.

Appendix

Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (https://links.lww.com/JBJSOA/A491).