joint replacement

Participant characteristics

Comparison of the OAI cohort (Table 1) baseline characteristics between the structural progressors and no-progressors showed, for the former, a higher percentage of participants with a Kellgren-Lawrence (KL) score > 0–1, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores and JSN, and a lower JSW. Age and BMI were also slightly higher in the structural progressor group, but although they reached statistical differences, they were not clinically significant.

For the TASOAC cohort (Table 2), a comparison between the two groups showed that the structural progressors have a higher WOMAC score and JSN, a lower JSW, and fewer men.

OAI and TASOAC cohorts have the same proportion for structural progressors (31%) and no-progressors (69%).

Comparison of the machine learning methodologies

With the OAI cohort, seven methodologies were compared using the 12 independent variables (Eq. 1). Figure 2 indicates the accuracy of the different ML methodologies in PVBSP forecasting at both the training and testing stages.

Data from the whole population showed that in the training stage (Fig. 2a), both the SVM and RF methodologies had good performances in all the population for structural progressors and no-progressors (mean of about 93%). The other methodologies resulted in poorer performances, primarily related to the progressor population. To select the superior model in PVBSP forecasting, the performance of different methodologies was analyzed in individuals who had not played a role in the calibration process, the testing stage (Fig. 2b). Data showed that only SVM demonstrated excellent accuracy for both groups. SVM methodology was thus further used for the development of the prediction models.

Feature selection: sensitivity analysis

Using the whole OAI population, the effect of each 12 variables (Eq. 1) was evaluated in which all the models, except model 1, consisted of removing a variable (Fig. 3a). Data showed that model 1 had 94.8% accuracy in the training stage and 96.8% in the testing stage (Fig. 3b, c). Moreover, the removal of each variable demonstrated in the training stage (Fig. 3b) that not using GDF5 (model 5), DUS4L (model 6), TP63 (model 9), and age (model 13) as input variables slightly reduced the model’s accuracy in predicting PVBSP compared to model 1. At the testing stage (Fig. 3c), the variables BMI (model 12) and age (model 13) were also reduced compared to model 1. These data thus suggest that although the accuracy of these models is close, based on slight differences in training and testing stages, the important variables are age, BMI, TP63, DUS4L, and GDF5, and the following Eq. 2 can be considered as a model for PVBSP forecasting the progressor population:

When the population was divided into structural progressors and no-progressors, findings (data not shown) revealed that the difference found for the whole population was related to the progressor population. Hence, when only the progressor population was used, the model using all the 12 variables (model 1) showed an accuracy of 88.8% in the testing stage. For the no-progressors, there was no difference in the accuracy between the different models suggesting that the variables did not impact the outcome. Therefore, the no-progressor population was not detailed further.

Machine learning model development

To find if lesser input variables can provide an accurate model, we further evaluated, by using the structural progressor population and the five variables as in Eq. 2, the scenarios of using one variable at a time followed by combining two to five variables. As illustrated in Fig. 4, 31 different ML models in PVBSP forecasting were developed. Data revealed that for models with only one variable (Fig. 4a), the accuracies of TP63 (M3), DUS4L (M4), and GDF5 (M5) at both training and testing stages were nil. Although the accuracy improved for age (M1) and BMI (M2), the numerical values (≤ 21.3%) were still very low. However, this improvement substantiates the importance of these two variables (age and BMI) as found in the sensitivity analysis (Fig. 3).

For the models with two variables (Fig. 4b), the highest accuracy (testing stage 41.3%) considered age and BMI simultaneously (M6). A comparison of the combination of each of these variables with age (M7–M9), with one consisting of age and BMI together (M6), showed that replacing BMI with TP63 (M7), DUS4L (M8), and GDF5 (M9) reduced the modeling accuracy in the testing stage by 12.5%, 18.8%, and 11.3%, respectively. However, if one of the inputs was BMI and the other two were one of the SNP genes (TP63 [M10], DUS4L [M11], GDF5 [M12]), the prediction accuracy was further reduced. Moreover, by not using age and BMI as one of the inputs of the models with two variables, we could not predict any of the progressors; the accuracy value was zero. These findings indicate that using TP63, DUS4L, and GDF5 without the risk factors cannot yield an efficient model in PVBSP forecasting.

As shown in Fig. 4c, adding one of TP63, DUS4L, or GDF5 as a variable to both age and BMI showed an increased accuracy. Models with three features that employed only one of the risk factors (M19–M21 for age and M22–M24 for BMI) had lower accuracy (range 21.3–40.0%) than the model with these two variables (M6, 41.3%). It should be noted that the simultaneous use of three variables without those of the two risk factors did not predict the progressors with high accuracy. Hence, the significant effect of age and BMI on PVBSP forecasting was confirmed.

For models with four and five variables (Fig. 4d), M26–M28 had higher accuracy than the best model offered among those using only three variables, i.e., M16. M27 demonstrated the best performance (testing stage, 73.8%). This model uses, in addition to age and BMI, TP63 and GDF5. The combination of DUS4L and GDF5 (M28) and DUS4L and TP63 (M26) were in the second and third place, respectively. As for the models with fewer variables (Fig. 4b, c), the non-simultaneous use of BMI and age (M29 and M30) led to a model with lower accuracy.

These results (Fig. 4) show that increasing the number of inputs is effective when both age and BMI are considered. However, the best performance (testing stage 78.8%) was obtained with M31, which considers the five variables as in Eq. 2.

Synergy of variables

The above data showed that, for the structural progressor population, the accuracy of M31 (five variables; testing stage, 78.8%) was lower than model 1 (12 variables; testing stage, 88.8%). Therefore, we assumed that some variables that are not considered in M31, including mtDNA haplogroup, cluster, FTO, GNL3, SUPT3H, MCF2L, and TGFA, could exert a synergistic effect with one or more variables in M31. This led to examining the synergy between the variables, and 66 new and different ML models were developed. Three impact levels were defined by comparing the results of each of these models with model 1 (Eq. 2) as the base model and according to the accuracy.

Table 3 illustrates, from highest to lowest, the impact effect between a fixed variable (variable 1) and one of the variables as listed in variable 2. Data revealed that the highest synergy impact was found for age with (i) BMI, (ii) GNL3, (iii) MCF2L, and (iv) FTO. Those having a moderate impact were age with (i) mtDNA haplogroup, (ii) GDF5, (iii) SUPT3H, (iv) TGFA, and (v) TP63; BMI with (vi) TP63 and (vii) SUPT3H, and (viii) GDF5 and MC2FL.

According to the highest and moderate impacts, two different scenarios were defined to find the optimum model (Table 4). In scenario 1, in addition to the combination of age and BMI, the factors found to have a high impact on synergies, GNL3, MCF2L, and FTO, were added one at a time to M31. In scenario 2, as the mtDNA haplogroup showed a moderate impact with age, it was used in addition to age and BMI as a fixed input variable, and the other SNP genes, TP63, FTO, GNL3, DUS4L, GDF5, SUPT3H, and TGFA, were added one at the time or in combination. All of the SNP genes were tested to ensure the accuracy and reliability of the final results.

Scenario 1

Three different models (Table 4, scenario 1) named M32-1 to M31-3 were defined and included the five variables of the model M31 plus GNL3, MCF2L, or FTO, respectively. The performance in the testing stage of M32-1 (82.7%), M32-2 (81.3%), and M32-3 (85.0%) was slightly lower than model 1 (12 variables, 88.8%) but higher than that of M31 (five variables, 78.8%). M32-3 (M31 + FTO) outperformed M31 and M31-2. Therefore, and to have a model with a lower number of variables, M32-3 appeared to be a very good model and consisted of:

Scenario 2

To verify if we could obtain a better accuracy with fewer variables, we analyzed another scenario consisting of age, BMI, and mtDNA haplogroup as fixed variables with one to seven SNP genes. This resulted in 109 combinations with four to ten variables and was named MH1-109. The analyses of all input combinations showed that the best accuracy range at the testing stage was 80.0–88.8%, and only those are represented in Table 4, scenario 2. Data showed that for four models with six to nine variables, the accuracy was identical (MH46, MH80, MH101, and MH106) in the testing stage (88.8%) as the one for model 1 with 12 variables. The model MH2 with four variables was at 80.0%, and MH17 with five variables at 82.5%. Given that the optimum model should not only have an excellent accuracy but the least number of variables, MH17 was selected as the optimal model:

Effect of each variable on the optimum model, MH17

The effect of each variable on the model MH17 was done using sensitivity analysis. Figure 5 demonstrates the impact of each SNP mtDNA haplogroup (others, H, Uk, T, J) and gene genotype for FTO (CC, CT, and TT) and SUPT3H (AA, GA, and GG) in PVBSP forecasting. The high percentage of error indicates the high impact of the studied variable.

Data showed that the mtDNA haplogroup H has the highest impact with an error of 35.0%, followed by UK with 16.1%, and ≤ 10% for the mtDNA haplogroup others, T, and J. FTO and SUPT3H both showed an identical highest error (37.3%) for both the presence of CT and absence of AA, respectively. The lowest error of FTO and SUPT3H was attained for the absence of TT (17.4%) and the presence of GG (9.8%), respectively.

Validation of the developed models using cross-validation and reproducibility with an external cohort (TASOAC)

The performance of M32-3 (Eq. 3) and MH-17 (Eq. 4) models when using the ten repetitions of tenfold cross-validation showed an average accuracy of 95.1% ± 2.1 for M32-3 and 94.6% ± 2.1 for MH-17 (Additional file 7: Fig. S4).

Reproducibility experiment with the external cohort TASOAC also demonstrated an excellent accuracy for both M32-3 (90.5%) and MH17 (85.7%), confirming the reliability and performance of these two developed ML models in the early detection of at-risk knee OA structural progressors.

Intraosseous (IO) infusion of medication during surgery has been shown to be a new and effective way to manage pain in patients undergoing total knee arthroplasty (TKA), according to a recent study.

To determine the safety and efficacy of injecting pain medication directly into the tibia during surgery and the impact that this method may have on pain levels and time spent in the hospital, the researchers performed a double-blind, randomized controlled study examining patients undergoing TKA (n = 48). The patients were divided into 2 groups: the experimental group (n = 24) who received both an IO antibiotic injection and 10 mg of morphine, and the control group (n = 24) who received only a standard IO injection of antibiotics.

The researchers assessed pain, nausea, and opioid use up to 14 days post-surgery for all patients. Additionally, the researchers examined morphine and interleukin-6 serum levels in a subgroup of 20 patients 10 hours post-surgery.

The researchers used the Visual Analog Scale to determine the level of pain each patient had postoperatively. Patients in the experimental group had a lower pain score at 1-, 2-, 3-, and 5-hours post-surgery (P = .0032, P = .005, P = .020, P = .10) when compared with the control group. The decrease in pain continued for postoperative day 1 (40% reduction, P = .01), day 2 (49% reduction, P = .036), day 8 (38% reduction, P = .025), and day 9 (33% reduction, P = .041).

Furthermore, the researchers saw a lower opioid consumption within the first 48 hours and the 2nd-week post-surgery among the experimental group when compared with the control group (P < .05). Serum morphine levels in were significantly less in the experimental group than in the control group 10 hours after IO injection (P = .049). The experimental group also has significant improvement (P < .05) in the Knee Injury and Osteoarthritis Outcome Score for Joint Replacement scores at 2- and 8-weeks post-surgery.

Overall, the experimental group showed significant improvement and outcome post-surgery.

“IO morphine combined with a standard antibiotic solution demonstrates superior postoperative pain relief immediately and up to 2 weeks,” the researchers concluded. “IO morphine is a safe and effective method to lessen postoperative pain in TKA patients.”

 

—Jessica Ganga

Reference:

Brozovich AA, Incavo SJ, Lambert BS, et al. Intraosseous morphine decreases postoperative pain and pain medication use in total knee arthroplasty: a double-blind, randomized controlled trial. J Arthroplasty. 2022;37(6):139-146. doi:10.1016/j.arth.2021.10.009.

Newswise — April 18, 2022 – A “bundled care” Medicare program to improve care for patients undergoing hip or knee replacement surgery has led to reductions in some outcome disparities for Black compared with White patients, suggests a study in The Journal of Bone & Joint Surgery. The journal is published in the Lippincott portfolio in partnership with Wolters Kluwer.

The introduction of Medicare’s Comprehensive Care for Joint Replacement (CJR) Model coincided with a reduction of racial differences in hospital readmission rates after hip or knee replacement surgery, according to new research by Calin Moucha, MD, Jashvant Poeran, MD, PhD, and other colleagues at the Icahn School of Medicine at Mount Sinai, New York.

Despite gains, racial differences persist in patient characteristics and outcomes

With use of nationwide Medicare claims data, the researchers analyzed disparities between Black and White patients undergoing total hip or knee replacement surgery, before and after rollout of the CJR Model in 2016. Under the CJR Model, health-care organizations receive a single “bundled” payment for all services throughout an episode of care – from the initial hospitalization to 90 days postoperatively – providing incentives to reduce costs while improving quality of care.

The study included data on nearly 1.5 million hip or knee replacement surgeries performed from 2013 to 2018. About 5% of patients were Black.

The analysis showed substantial racial differences in patient characteristics, outcomes, and Medicare payments, both before and after implementation of the CJR Model. As a group, Black patients had higher rates of other health problems, received more blood transfusions, spent more days in the hospital, and were more likely to be discharged to an institution (such as a skilled nursing facility), rather than being sent directly home.

The CJR program led to improvements in several key outcomes, some of which differed by race. After adjustment for other factors, White patients who were managed under the CJR approach had reductions in length of hospital stay, complication rate, risk of hospital readmission within 90 and 180 days, discharge to institutional care, and Medicare payments to skilled nursing facilities.

Some of the improvements were greater among Black patients. In particular, Black patients had larger reductions in 90-day and 180-day hospital readmission rates, as well as in Medicare payments related to outpatient care.

The greater benefits among Black compared to White patients suggest that the CJR program has improved some pre-existing racial differences. “These observed racial differences may represent true ‘disparities’ as some may not be attributable to clinical factors and may be directly associated with poorer outcomes,” the researchers write.

Dr Moucha comments, “This is an important finding as it provides insights on how to effectively reduce these disparities that we know are widespread, not just on orthopaedics, but in medicine in general.”

Dr. Poeran adds, “These results indeed seem promising, but we do have to consider alternative perspectives and explanations of our results. For example, although the effects on readmission rates are promising, the difference in payments for outpatient care – where we saw lower Medicare payments for Black patients – may also indicate potential under-utilization of postdischarge care in certain subgroups.”

Together with some previous reports of outcomes after introduction of the CJR Model, the new findings “support the notion of adapting and leveraging the bundled payment program design to reduce disparities in [total hip and knee replacement] care and outcomes,” the researchers write. They note that their study could not demonstrate a causal relationship between the CJR Model and the observed improvements in patient outcomes.

“A first step toward reducing racial differences that represent disparities […] is to understand the sources of these disparities,” Dr. Poeran and colleagues conclude. They call for further studies to evaluate the potential of bundled payment models to reduce racial disparities, and the mechanisms by which they do so.

Click here to read “Racial Differences in Care and Outcomes After Total Hip and Knee Arthroplasties: Did the Comprehensive Care for Joint Replacement Program Make a Difference? “

DOI: 10.2106/JBJS.21.00465

###

About The Journal of Bone & Joint Surgery

The Journal of Bone & Joint Surgery (JBJS) has been the most valued source of information for orthopaedic surgeons and researchers for over 125 years and is the gold standard in peer-reviewed scientific information in the field. A core journal and essential reading for general as well as specialist orthopaedic surgeons worldwide, The Journal publishes evidence-based research to enhance the quality of care for orthopaedic patients. Standards of excellence and high quality are maintained in everything we do, from the science of the content published to the customer service we provide. JBJS is an independent, non-profit journal.

About Wolters Kluwer

Wolters Kluwer (WKL) is a global leader in professional information, software solutions, and services for the clinicians, nurses, accountants, lawyers, and tax, finance, audit, risk, compliance, and regulatory sectors. We help our customers make critical decisions every day by providing expert solutions that combine deep domain knowledge with advanced technology and services.

Wolters Kluwer reported 2020 annual revenues of €4.6 billion. The group serves customers in over 180 countries, maintains operations in over 40 countries, and employs approximately 19,200 people worldwide. The company is headquartered in Alphen aan den Rijn, the Netherlands.

Wolters Kluwer provides trusted clinical technology and evidence-based solutions that engage clinicians, patients, researchers and students in effective decision-making and outcomes across healthcare. We support clinical effectiveness, learning and research, clinical surveillance and compliance, as well as data solutions. For more information about our solutions, visit https://www.wolterskluwer.com/en/health and follow us on LinkedIn and Twitter @WKHealth.

For more information, visit www.wolterskluwer.com, follow us on Twitter, Facebook, LinkedIn, and YouTube.

As intelligent, connected technology becomes more prevalent in every aspect of our daily lives, it’s probably no surprise that apps and digital feedback are becoming a part of the high-tech world of orthopaedic surgery.

Vail-Summit Orthopaedics & Neurosurgery is currently offering a one-of-a-kind, interconnected pair of all-new tech tools which promise to significantly and proactively improve the outcome of knee replacement patients.

Dr. Nathan Cafferky, a well-respected total joint surgeon and partner in VSON, is one of the first surgeons in the country to use the Persona IQ® “smart knee,” an FDA-approved knee replacement prosthetic that has been very subtly implanted with a data-sharing unit.

---

---

When used in conjunction with the mymobility® app that many of VSON’s patients already use on their smartphones to prepare for and recover from knee surgery, Cafferky says the combination allows patients to more accurately gauge their success and set goals for themselves throughout their recovery.

“The smart knee appeared in the last six or months or so, and it’s a game-changer, especially for those who are super-into new technology,” Cafferky says. “It gives you real-time feedback on your recovery: your steps, your biometrics and range of motion.”

---

---

mymobility, he explains, still works very well as a tool for patients to understand the stages of their pre- and post-surgical progress, but it tends to simply provide passive information. The smart knee, equipped with a tiny RF-frequency transmitter, delivers data on strain, stride, strength and other biometric details to a HIPAA compliant cloud-based platform that can only be accessed by you and your healthcare team. It does not collect data on your location– Cafferky jokes that he cannot use the implant to see if you are at Wal-Mart. 

That data can be used to track recovery and help patients set goals for themselves. “But it can also point out other important issues,” he adds. “For example, it can tell us if the implant is not working because of infection, loosening of components or failure.”

While total joint surgery has become a very common procedure in recent years, Cafferky says VSON’s objective is better outcomes for everyone. Technology, such as the smart knee or the robotics systems he routinely uses in his surgical procedures, help him and VSON stay on the cutting edge of patient-focused care.

In the case of the new smart knee, Cafferky says the instantaneous nature of its digital feedback can be an invaluable tool to help patients monitor their progress and see how an active post-surgical regimen will contribute to their recovery.

“The implant software even uses predictive analysis and artificial intelligence to tell the patients that if they take 500 steps today, there’s a 90% chance of them being much improved in two months,” he notes. “It helps point out trends to help people stay on the right path, and it gives more patients the confidence in their care.”

Those affirmations are part of VSON’s overall objectives for its patients, and Cafferky says the interconnected tools represent another aspect of the organization’s mission  of helping patients comfortably and quickly return to their normal routines.

Cafferky admits the smart knee concept is not for everyone, but he says that patients who’ve embraced everything from self-driving cars to a range of home high-tech devices have been very excited by Persona IQ and other similar biotech looming on the horizon.

“This technology is sort of like a pacemaker for the knee, but we get to be connected to the patient every step of the way. This is how we get healthcare innovation.”

So far, Cafferky has had two patients implanted with the prosthetic, and plenty more signed up – of the 100 or so smart knee surgeries done in the US so far, most have been in Colorado, he says.

“New ideas like this generate more excitement, and as always, the goal is to help patients thrive, succeed and feel more engaged in their recovery.” 

FREMONT, Calif., Aug. 26, 2021 /PRNewswire/ — THINK Surgical, Inc., an innovator in the field of orthopedic active robot surgery, is pleased to announce that INOV8 Surgical is the first healthcare facility to utilize the second-generation TSolution One® Total Knee Application for total knee arthroplasty (TKA). Stefan Kreuzer, M.D., orthopedic surgeon, and the founder of INOV8 Orthopedics, performed the first procedure with the new system on August 24^th at INOV8 Surgical Ambulatory Surgery Center (ASC) in Houston, Texas. THINK’s TSolution One system is the only robot system for TKA that supports an open implant library, giving surgeons the largest choice of implant options from different manufactures facilitating broad patient customization.

“My team and I are pleased to offer our patients the accuracy and reproducibility made possible in joint replacement procedures by THINK Surgical’s next-generation active robot,” said Dr. Kreuzer*. “Adoption of this advanced technology demonstrates our commitment to deliver exceptional outcomes for our patients.”

Dr. Kreuzer incorporated the first-generation TSolution One system into the INOV8 Surgical orthopedics program in early 2020, shortly after initial FDA clearance. INOV8 Surgical is a leading healthcare facility specializing in outpatient total joint procedures.

“We are very excited to have our latest robotic technology available to patients through the INOV8 Surgical ASC,” said Jay Yang, acting CEO and COO of THINK Surgical, Inc. “Our innovative engineers continue to advance robotic orthopedic surgery as demonstrated by our next-generation system and by future product launches.”

The TSolution One system consists of TPLAN^®, a 3D pre-surgical planning workstation, and TCAT^®, an active robot. Pre-surgical planning allows the surgeon to design and prepare the patient’s personalized joint replacement surgical plan in a virtual environment. The active robot aids the surgeon in executing the preoperative surgical plan with precise, automated cutting and removal of the diseased bone and cartilage. The TSolution One system assists surgeons with optimizing joint implant placement based on each patient’s unique anatomy.

About THINK Surgical^®, Inc.
THINK Surgical, Inc., a privately held U.S.-based medical device and technology company, develops, manufactures, and markets active robotics for orthopedic surgery. The core technology of the TSolution One system has been used in tens of thousands of successful total joint replacements worldwide.

The TSolution One system is the only robot available for total joint replacement procedures that features an open implant library.

THINK Surgical actively collaborates with healthcare professionals around the globe to refine of our orthopedic products, improving the lives of those suffering from advanced joint disease with precise, accurate, and intelligent technology. Please refer to the instructions for use for the TSolution One system for a complete list of indications, contraindications, warnings, and precautions. For additional product information, please visit www.thinksurgical.com.

*Dr. Kreuzer is a paid consultant of THINK Surgical.  

THINK Surgical and TSolution One are registered trademarks of THINK Surgical, Inc. ©2021 THINK Surgical, Inc. All rights reserved.

Media Contact
Sheri Hensley
SHensley@thinksurgical.com
510-602-0951

View original content to download multimedia:https://www.prnewswire.com/news-releases/inov8-surgical-performs-first-surgery-using-think-surgicals-next-generation-robot-technology-for-knee-replacement-301363227.html

SOURCE THINK Surgical, Inc.