A predictive algorithm for perioperative complications and readmission after ankle arthrodesis.

PURPOSE: Ankle arthrodesis is a mainstay of surgical management for ankle arthritis. Accurately risk-stratifying patients who undergo ankle arthrodesis would be of great utility. There is a paucity of accurate prediction models that can be used to pre-operatively risk-stratify patients for ankle arthrodesis. We aim to develop a predictive model for major perioperative complication or readmission after ankle arthrodesis. METHODS: This is a retrospective cohort study of adult patients who underwent ankle arthrodesis at any non-federal California hospital between 2015 and 2017. The primary outcome is readmission within 30 days or major perioperative complication. We build logistic regression and ML models spanning different classes of modeling approaches, assessing discrimination and calibration. We also rank the contribution of the included variables to model performance for prediction of adverse outcomes. RESULTS: A total of 1084 patients met inclusion criteria for this study. There were 131 patients with major complication or readmission (12.1%). The XGBoost algorithm demonstrates the highest discrimination with an area under the receiver operating characteristic curve of 0.707 and is well-calibrated. The features most important for prediction of adverse outcomes for the XGBoost model include: diabetes, peripheral vascular disease, teaching hospital status, morbid obesity, history of musculoskeletal infection, history of hip fracture, renal failure, implant complication, history of major fracture. CONCLUSION: We report a well-calibrated algorithm for prediction of major perioperative complications and 30-day readmission after ankle arthrodesis. This tool may help accurately risk-stratify patients and decrease likelihood of major complications.

Machine learning-driven identification of novel patient factors for prediction of major complications after posterior cervical spinal fusion.

PURPOSE: Posterior cervical fusion is associated with increased rates of complications and readmission when compared to anterior fusion. Machine learning (ML) models for risk stratification of patients undergoing posterior cervical fusion remain limited. We aim to develop a novel ensemble ML algorithm for prediction of major perioperative complications and readmission after posterior cervical fusion and identify factors important to model performance. METHODS: This is a retrospective cohort study of adults who underwent posterior cervical fusion at non-federal California hospitals between 2015 and 2017. The primary outcome was readmission or major complication. We developed an ensemble model predicting complication risk using an automated ML framework. We compared performance with standard ML models and logistic regression (LR), ranking contribution of included variables to model performance. RESULTS: Of the included 6822 patients, 18.8% suffered a major complication or readmission. The ensemble model demonstrated slightly superior predictive performance compared to LR and standard ML models. The most important features to performance include sex, malignancy, pneumonia, stroke, and teaching hospital status. Seven of the ten most important features for the ensemble model were markedly less important for LR. CONCLUSION: We report an ensemble ML model for prediction of major complications and readmission after posterior cervical fusion with a modest risk prediction advantage compared to LR and benchmark ML models. Notably, the features most important to the ensemble are markedly different from those for LR, suggesting that advanced ML methods may identify novel prognostic factors for adverse outcomes after posterior cervical fusion.

Development of a Novel, Potentially Universal Machine Learning Algorithm for Prediction of Complications After Total Hip Arthroplasty.

BACKGROUND: As the prevalence of hip osteoarthritis increases, the number of total hip arthroplasty (THA) procedures performed is also projected to increase. Accurately risk-stratifying patients who undergo THA would be of great utility, given the significant cost and morbidity associated with developing perioperative complications. We aim to develop a novel machine learning (ML)-based ensemble algorithm for the prediction of major complications after THA, as well as compare its performance against standard benchmark ML methods. METHODS: This is a retrospective cohort study of 89,986 adults who underwent primary THA at any California-licensed hospital between 2015 and 2017. The primary outcome was major complications (eg infection, venous thromboembolism, cardiac complication, pulmonary complication). We developed a model predicting complication risk using AutoPrognosis, an automated ML framework that configures the optimally performing ensemble of ML-based prognostic models. We compared our model with logistic regression and standard benchmark ML models, assessing discrimination and calibration. RESULTS: There were 545 patients who had major complications (0.61%). Our novel algorithm was well-calibrated and improved risk prediction compared to logistic regression, as well as outperformed the other four standard benchmark ML algorithms. The variables most important for AutoPrognosis (eg malnutrition, dementia, cancer) differ from those that are most important for logistic regression (eg chronic atherosclerosis, renal failure, chronic obstructive pulmonary disease). CONCLUSION: We report a novel ensemble ML algorithm for the prediction of major complications after THA. It demonstrates superior risk prediction compared to logistic regression and other standard ML benchmark algorithms. By providing accurate prognostic information, this algorithm may facilitate more informed preoperative shared decision-making.

Risk Adjustment in the California Joint Replacement Registry: Is Patient Complexity Accurately Assessed in Academic Versus Nonacademic Hospitals?

BACKGROUND: We sought to report on the differences in observed versus expected arthroplasty outcomes between academic and nonacademic hospitals in a large joint registry. We utilized the California Joint Replacement Registry's data and risk adjustment model. METHODS: Observed versus expected hip and knee arthroplasty complications were utilized to assess hospital and surgeon risk-adjusted complication rates (RACRs). Based on a hospital and surgeon RACR, each was assigned a performance rating ("worse," "expected," "better"). Associations between academic status and performance ratings, rates of individual complications, prevalence of risk factors associated with increased complication rates, and differences in complication rates were calculated. RESULTS: A higher percentage of academic providers had "worse" than expected ratings, whereas a higher percentage of nonacademic providers had "expected" and "better" than expected ratings (P = .011) based on the observed versus expected complication rates. There was a higher incidence of patients with congestive heart failure and an elevated American Society of Anesthesiologists classification in academic institutions (P = .0001). The complication rate was higher in academic institutions for all total knee arthroplasties (P < .0016). CONCLUSIONS: We identified disparities in RACRs between nonacademic and academic institutions. This may reflect the difficulty of fully adjusting for medical risk and surgical complexity in a large arthroplasty database.

The peroneal strength deficits in patients with chronic ankle instability compared to ankle sprain copers and normal individuals.

BACKGROUND: Despite a consensus regarding the correlation of peroneal strength deficit with chronic ankle instability (CAI), there are conflicting reports in regards to peroneal strength as assessed by isokinetic dynamometer in patients with CAI. The purpose of this study was to evaluate the changes of isokinetic strength in patients with CAI compared to ankle sprain copers and normal individuals. METHODS: Forty-two patients (CAI group) with chronic ankle instability who were scheduled for the modified Broström procedure met inclusion criteria. Thirty-one ankle sprain copers (ASC group) who were eligible at 6 months after acute injury and 30 controls were recruited. The muscle strength associated with four motions of the ankle were evaluated using isokinetic dynamometer. RESULTS: Peak torque for inversion and eversion at 60°/s angular velocity were significantly lower in the CAI group compared to the ASC and control group (P=.004, P<.001, respectively). Deficit ratio of peak torque for eversion at 60°/s and 120°/s in the CAI group were 33.8% and 19.8%, respectively, which indicated significant side to side differences (both P<.001). The evertor/invertor strength ratio (0.59) for eversion at 60°/s was significantly lower in the CAI group (P<.001). CONCLUSION: As compared to the ankle sprain copers and normal individuals, patients with chronic ankle instability who were scheduled for modified Broström procedure demonstrated a significant weakness of isokinetic peroneal strength. Isokinetic muscular assessment can provide the useful preoperative informations regarding functional ankle instability focusing on peroneal weakness.

A randomized comparison between lateral ligaments augmentation using suture-tape and modified Broström repair in young female patients with chronic ankle instability.

BACKGROUND: Although various minimally invasive procedures for chronic ankle instability are increasingly being used, a question regarding whether these procedures can be a viable alternative of the modified Broström procedure remains controversial. This study was conducted to compare the intermediate-term clinical outcomes between lateral ligaments augmentation using suture-tape and modified Broström repair in a selected cohort of patients. METHODS: Sixty female patients with chronic lateral ankle instability were randomly assigned and underwent surgical treatments by one surgeon. Twenty-eight patients with suture-tape augmentation and 27 modified Broström procedures were followed ≥2 years and analysed in this comparative study. The clinical evaluation included the Foot and Ankle Outcome Score (FAOS), Foot and Ankle Ability Measure (FAAM), and stress radiographs. Medical expense related with operation was analysed to evaluate the cost-effectiveness. RESULTS: There were no statistically significant differences in the clinical outcomes between two procedures based on FAOS, FAAM, recurrence rate of instability, and stress radiographs. Total medical expense was approximately 1.3 times more in the suture-tape group (P<0.001), despite shorter operation time. CONCLUSIONS: Lateral ankle ligaments augmentation using suture-tape showed the similar clinical outcomes but low cost-effectiveness, as compared to modified Broström repair for young female patients with chronic ankle instability.

Is generalized ligamentous laxity a prognostic factor for recurred hallux valgus deformity?

BACKGROUND: This study was performed to evaluate the intermediate-term clinical outcomes after proximal chevron osteotomy for hallux valgus in patients with generalized ligamentous laxity, and to determine the effect on postoperative recurrence of deformity. METHODS: There were 23 cases in laxity group (Beighton score ≥5 points) and 175 in non-laxity group with a mean followup of 46.3 months. Clinical evaluation consisted of the AOFAS score, Foot and Ankle Ability Measure (FAAM), and radiographic measurement of hallux alignment. Risk factors associated with postoperative recurrence were evaluated using univariate analysis. RESULTS: Recurrence rates were 21.7% in the laxity group and 17.1% in non-laxity group (P=.218). There were no significant differences in clinical and radiographic measurements at final followup between the 2 groups. Preoperative HVA and IMA were found to be predictive factors of recurrence (OR=6.3, 4.2; P=.001, .018, respectively). CONCLUSION: There were no statistical differences in the clinical and radiographic outcomes between hallux valgus with and without generalized ligamentous laxity. Generalized ligamentous laxity demonstrated no definitive effects on postoperative recurrence of hallux valgus deformity.

Incidence and predictors of early complications following primary and revision total ankle arthroplasty.

BACKGROUND: Total ankle arthroplasty (TAA) offers an effective option for end-stage osteoarthritis. The incidence and preoperative risk factors for early adverse events (AEs) following primary and revision TAA may be useful information for providers. METHODS: A large database was queried from 2010 to 2016 to identify 905 patients of whom 818 underwent primary TAA (90.4%) and 87 underwent revision TAA (9.6%). Data on patient demographics, comorbidities, and hospital length of stay were analyzed as risk factors for reported 30-day AEs. RESULTS: The overall AE rate was 5.5% (50/905) for the entire cohort. AEs occurred more frequently for revision TAA (9/87) than primary TAA (41/818) cases (OR 2.43, p=0.022). Age (OR 1.03, p=0.045), BMI (OR 1.04, p=0.046), and revision TAA (OR 2.56, p=0.002) were independent risk factors for 30-day AEs in multivariate analysis. CONCLUSIONS: Older age, higher BMI, and revision cases are associated with a higher risk of AEs.

Reoperation and complications after anterior cervical discectomy and fusion and cervical disc arthroplasty: a study of 52,395 cases.

PURPOSE: The aim of this study was to analyze rates of perioperative complications and subsequent cervical surgeries in patients treated for cervical degenerative disc disease with anterior cervical discectomy and fusion (ACDF) and those treated with artificial cervical disc arthroplasty (ACDA) for up to 5-year follow-up. METHODS: California's Office of Statewide Health Planning and Development discharge database was analyzed for patients aged 18-65 years undergoing single-level ACDF or ACDA between 2003 and 2010. Medical comorbidities were identified with CMS-Condition Categories. Readmissions for short-term complications of the procedure were identified and rates of subsequent cervical surgeries were calculated at 90-day and 1-, 3-, and 5-year follow-up. Multivariate regression modeling was used to identify associations with complications and subsequent cervical surgeries correcting for patient and provider characteristics. RESULTS: A total of 52,395 eligible cases were identified: 50,926 ACDF and 1469 ACDA. Readmission was less common in the ACDA group (OR: 0.69, 95% CI: 0.48-1.0, p = 0.048). Subsequent cervical spine surgery was more common in the ACDF group in the immediate perioperative period (within 90 days of surgery) (ACDF 3.35% vs. ACDA 2.04%, OR: 0.63, 95% CI: 0.44-0.92, p = 0.015). At 1-, 3-, and 5-year postoperatively, rates of subsequent cervical surgeries were similar between the two cohorts. CONCLUSIONS: We found no protective benefit for ACDA versus ACDF for single-level disease at up to 5-year follow-up in the largest cohort of patients examined to date. Early complications were rare in both cohorts stressing the value of large cohort studies to study risk factors for rare events. These slides can be retrieved under Electronic Supplementary Material.

Adipose-derived Human Perivascular Stem Cells May Improve Achilles Tendon Healing in Rats.

BACKGROUND: Achilles tendon rupture is a common injury and the best treatment option remains uncertain between surgical and nonoperative methods. Biologic approaches using multipotent stem cells such as perivascular stem cells pose a possible treatment option, although there is currently a paucity of evidence regarding their clinical therapeutic use. QUESTIONS/PURPOSES: The purpose of this study was to determine whether injected perivascular stem cells (PSCs) would (1) improve histologic signs of tendon healing (such as percent area of collagen); and (2) improve biomechanical properties (peak load or stiffness) in a rat model of Achilles tendon transection. METHODS: Two subtypes of PSCs were derived from human adipose tissue: pericytes (CD146CD34CD45CD31) and adventitial cells (CD146CD34CD45CD31). Thirty-two athymic rats underwent right Achilles transection and were randomized to receive injection with saline (eight tendons), hydrogel (four tendons), pericytes in hydrogel (four tendons), or adventitial cells in hydrogel (eight tendons) 3 days postoperatively with the left serving as an uninjured control. Additionally, a subset of pericytes was labeled with CM-diI to track cell viability and localization. At 3 weeks, the rats were euthanized, and investigators blinded to treatment group allocation evaluated tendon healing by peak load and stiffness using biomechanical testing and percent area of collagen using histologic analysis with picrosirius red staining. RESULTS: Histologic analysis showed a higher mean percent area collagen for pericytes (30%) and adventitial cells (28%) than hydrogel (21%) or saline (26%). However, a nonparametric statistical analysis yielded no statistical difference. Mechanical testing demonstrated that the pericyte group had a higher peak load than the saline group (41 ± 7 N versus 26 ± 9 N; mean difference 15 N; 95% confidence interval [CI], 4-27 N; p = 0.003) and a higher peak load than the hydrogel group (41 ± 7 N versus 25 ± 3 N; mean difference 16; 95% CI, 8-24 N; p = 0.001). The pericyte group demonstrated higher stiffness than the hydrogel group (36 ± 12 N/mm versus 17 ± 6 N/mm; mean difference 19 N/mm; 95% CI, 5-34 N/mm; p = 0.005). CONCLUSIONS: Our results suggest that injection of PSCs improves mechanical but not the histologic properties of early Achilles tendon healing. CLINICAL RELEVANCE: This is a preliminary study that provides more insight into the use of adipose-derived PSCs as a percutaneous therapy in the setting of Achilles tendon rupture. Further experiments to characterize the function of these cells may serve as a pathway to development of minimally invasive intervention aimed at improving nonoperative management while avoiding the complications associated with surgical treatment down the line.

Can Preoperative Patient-reported Outcome Measures Be Used to Predict Meaningful Improvement in Function After TKA?

BACKGROUND: Despite the overall effectiveness of total knee arthroplasty (TKA), a subset of patients do not experience expected improvements in pain, physical function, and quality of life as documented by patient-reported outcome measures (PROMs), which assess a patient's physical and emotional health and pain. It is therefore important to develop preoperative tools capable of identifying patients unlikely to improve by a clinically important margin after surgery. QUESTIONS/PURPOSES: The purpose of this study was to determine if an association exists between preoperative PROM scores and patients' likelihood of experiencing a clinically meaningful change in function 1 year after TKA. METHODS: A retrospective study design was used to evaluate preoperative and 1-year postoperative Knee injury and Osteoarthritis Outcome Score (KOOS) and SF-12 version 2 (SF12v2) scores from 562 patients who underwent primary unilateral TKA. This cohort represented 75% of the 750 patients who underwent surgery during that time period; a total of 188 others (25%) either did not complete PROM scores at the designated times or were lost to follow-up. Minimum clinically important differences (MCIDs) were calculated for each PROM using a distribution-based method and were used to define meaningful clinical improvement. MCID values for KOOS and SF12v2 physical component summary (PCS) scores were calculated to be 10 and 5, respectively. A receiver operating characteristic analysis was used to determine threshold values for preoperative KOOS and SF12v2 PCS scores and their respective predictive abilities. Threshold values defined the point after which the likelihood of clinically meaningful improvement began to diminish. Multivariate regression was used to control for the effect of preoperative mental and emotional health, patient attributes quantified by SF12v2 mental component summary (MCS) scores, on patients' likelihood of experiencing meaningful improvement in function after surgery. RESULTS: Threshold values for preoperative KOOS and SF12v2 PCS scores were a maximum of 58 (area under the curve [AUC], 0.76; p < 0.001) and 34 (AUC, 0.65; p < 0.001), respectively. Patients scoring above these thresholds, indicating better preoperative function, were less likely to experience a clinically meaningful improvement in function after TKA. When accounting for mental and emotional health with a multivariate analysis, the predictive ability of both KOOS and SF12v2 PCS threshold values improved (AUCs increased to 0.80 and 0.71, respectively). Better preoperative mental and emotional health, as reflected by a higher MCS score, resulted in higher threshold values for KOOS and SF12v2 PCS. CONCLUSIONS: We identified preoperative PROM threshold values that are associated with clinically meaningful improvements in functional outcome after TKA. Patients with preoperative KOOS or SF12v2 PCS scores above the defined threshold values have a diminishing probability of experiencing clinically meaningful improvement after TKA. Patients with worse baseline mental and emotional health (as defined by SF12v2 MCS score) have a lower probability of experiencing clinically important levels of functional improvement after surgery. The results of this study are directly applicable to patient-centered informed decision-making tools and may be used to facilitate discussions with patients regarding the expected benefit after TKA. LEVEL OF EVIDENCE: Level III, prognostic study.

Revision with suture-tape augmentation after failed collateral ligament reconstruction for chronic interphalangeal instability of the hallux.

Chronic varus instability or recurrent subluxation following isolated interphalangeal dislocation of the hallux is a rare injury. No consensus has been reached regarding the best joint-salvage procedure for patients with the failed collateral ligament reconstruction using tendon graft. We report a case who achieved satisfactory clinical outcome through a modified surgical procedure (revision collateral ligament reconstruction augmented with suture-tape).

John Charnley Award: Preoperative Patient-reported Outcome Measures Predict Clinically Meaningful Improvement in Function After THA.

BACKGROUND: Despite the overall effectiveness of total hip arthroplasty (THA), a subset of patients remain dissatisfied with their results because of persistent pain or functional limitations. It is therefore important to develop predictive tools capable of identifying patients at risk for poor outcomes before surgery. QUESTIONS/PURPOSES: The purpose of this study was to use preoperative patient-reported outcome measure (PROM) scores to predict which patients undergoing THA are most likely to experience a clinically meaningful change in functional outcome 1 year after surgery. METHODS: A retrospective cohort study design was used to evaluate preoperative and 1-year postoperative SF-12 version 2 (SF12v2) and Hip Disability and Osteoarthritis Outcome Score (HOOS) scores from 537 selected patients who underwent primary unilateral THA. Minimum clinically important differences (MCIDs) were calculated using a distribution-based method. A receiver operating characteristic analysis was used to calculate threshold values, defined as the levels at which substantial changes occurred, and their predictive ability. MCID values for HOOS and SF12v2 physical component summary (PCS) scores were calculated to be 9.1 and 4.6, respectively. We analyzed the effect of SF12v2 mental component summary (MCS) scores, which measure mental and emotional health, on SF12v2 PCS and HOOS threshold values. RESULTS: Threshold values for preoperative HOOS and PCS scores were a maximum of 51.0 (area under the curve [AUC], 0.74; p < 0.001) and 32.5 (AUC, 0.62; p < 0.001), respectively. As preoperative mental and emotional health improved, which was reflected by a higher MCS score, HOOS and PCS threshold values also increased. When preoperative mental and emotional health were taken into account, both HOOS and PCS threshold values' predictive ability improved (AUCs increased to 0.77 and 0.69, respectively). CONCLUSIONS: We identified PROM threshold values that predict clinically meaningful improvements in functional outcome after THA. Patients with a higher level of preoperative function, as suggested by HOOS or PCS scores above the defined threshold values, are less likely to obtain meaningful improvement after THA. Lower preoperative mental and emotional health decreases the likelihood of achieving a clinically meaningful improvement in function after THA. The results of this study may be used to facilitate discussion between physicians and patients regarding the expected benefit after THA and to support the development of patient-based informed decision-making tools. For example, despite significant disease, patients with high preoperative function, as measured by PROM scores, may choose to delay surgery given the low likelihood of experiencing a meaningful improvement postoperatively. Similarly, patients with notably low MCS scores might best be counseled to address mental health issues before embarking on surgery. LEVEL OF EVIDENCE: Level III, prognostic study.

The Importance of Risk Adjustment in Reporting Total Joint Arthroplasty Outcomes.

BACKGROUND: This study describes and tests a risk adjustment model developed for the California Joint Replacement Registry to report predictors of complication rates. METHODS: Complication rates were analyzed for 9960 patients enrolled in the California Joint Replacement Registry at 22 medical centers. Multivariable logistic risk models were created to analyze risks of postoperative complications. RESULTS: Age and American Society of Anesthesiologists class were the strongest predictors of complication rates (P < .0001). Congestive heart failure and peripheral vascular disease were also statistically significant predictors of complications. Three hospitals were found to have statistically significantly worse than expected complication rates, and one was found to have a better than expected complication rate after case mix risk adjustment. CONCLUSION: Adequate risk adjustment is a key element in objective comparison of surgeons, hospitals, and devices using total joint arthroplasty registry data.

Bilateral Total Hip Arthroplasty has Similar Complication Rates to Unilateral Total Hip Arthroplasty.

Simultaneous bilateral THA has several advantages over staged THA, however its safety has not been well examined. This study reports a statewide 15-year experience of simultaneous bilateral THA and compares its complications to those of unilateral THA. Logistic regression was used to determine the role of bilateral surgery as a predictor of complications while correcting for patient comorbidities. Of 202,986 patients, 1.1% underwent bilateral THA. Bilateral THA was more commonly performed in males, younger patients, those with private insurance, and at higher volume hospitals. There was no difference in 30-day readmissions or revision surgeries. There was a higher rate of sepsis in the bilateral group but no difference in other complications. This study demonstrates that THA is a safe option in appropriately selected patients.

Responsiveness of patient reported outcome measures in total joint arthroplasty patients.

UNLABELLED: This study reports the responsiveness to change and minimal clinically important difference (MCID) of three patient reported outcome measures following total hip arthroplasty (THA) and total knee arthroplasty (TKA). Patient-reported outcome measures were collected preoperatively and 3 months postoperatively for 391 patients enrolled in the California Joint Replacement Registry. Effect size, standardized response means, and MCID were calculated for each measure. The WOMAC and the SF12v2 physical component summary (PCS) score were the most responsive to perioperative changes. The MCID was 4.97 for the SF12v2 PCS and 10.21 for the WOMAC. THA patients were more likely to exhibit improvements above the MCID than TKA patients. The WOMAC and SF12v2 PCS are useful to measure health status changes in TJA patients. LEVEL OF EVIDENCE: Prognostic Level II.

Operative versus nonoperative treatment of acute Achilles tendon rupture: An analysis of 12,570 patients in a large healthcare database.

BACKGROUND: The purpose of this study was to compare the latest patient demographics and rerupture rates of operative versus nonoperative treatment of acute Achilles tendon rupture in the United States. METHODS: Patients undergoing treatment of an acute Achilles tendon rupture from 2007 to 2011 were identified by cross-referencing ICD-9-CM and CPT codes through the PearlDiver Patient Record Database. RESULTS: In total, 12,570 patients were treated for an acute Achilles tendon rupture. The ratio of operative to nonoperative treatment increased from 1.41 to 1.65. Males were more likely to undergo surgery than females. There were no significant differences in short-term rerupture rate for operative (2.1%) versus nonoperative (2.4%) treatment. CONCLUSIONS: The proportion of patients who received operative treatment for an acute Achilles tendon rupture increased slightly during the 5 year period, suggesting that surgeons in the United States have been slower to adopt nonoperative treatment than their European counterparts.

Compliant Intramedullary Stems for Joint Reconstruction.

The longevity of current joint replacements is limited by aseptic loosening, which is the primary cause of non-infectious failure for hip, knee, and ankle arthroplasty. Aseptic loosening is typically caused either by osteolysis from particulate wear, or by high shear stresses at the bone-implant interface from over-constraint. Our objective was to demonstrate feasibility of a compliant intramedullary stem that eliminates over-constraint without generating particulate wear. The compliant stem is built around a compliant mechanism that permits rotation about a single axis. We first established several models to understand the relationship between mechanism geometry and implant performance under a given angular displacement and compressive load. We then used a neural network to identify a design space of geometries that would support an expected 100-year fatigue life inside the body. We additively manufactured one representative mechanism for each of three anatomic locations, and evaluated these prototypes on a KR-210 robot. The neural network predicts maximum stress and torsional stiffness with 2.69% and 4.08% error respectively, relative to finite element analysis data. We identified feasible design spaces for all three of the anatomic locations. Simulated peak stresses for the three stem prototypes were below the fatigue limit. Benchtop performance of all three prototypes was within design specifications. Our results demonstrate the feasibility of designing patient- and joint-specific compliant stems that address the root causes of aseptic loosening. Guided by these results, we expect the use of compliant intramedullary stems in joint reconstruction technology to increase implant lifetime.