Language Processing Algorithms Can Boost Orthopaedic Research
Manufacturing, farming, and shopping…These are just 3 diverse examples of how technology is advancing daily and automating tedious tasks, decreasing costs, and improving efficiencies. Orthopaedics and orthopaedic research are not being left behind in this progression. In the November 6, 2019 edition of JBJS, Wyles et al. evaluate the accuracy of natural language processing (NLP) tools in automating the extraction of orthopaedic data from electronic health records (EHRs) and registries. The findings suggest that NLP-generated algorithms can indeed reliably extract data without the labor-intensive and costly process of manual chart reviews.
First, using an open-source NLP “engine,” the researchers developed NLP algorithms focused on 3 elements of >1,500 total hip arthroplasty (THA) procedures captured in the Mayo Total Joint Registry: (1) operative approach, (2) fixation technique, and (3) bearing surface. They then applied the algorithm to operative notes from THAs performed at Mayo and to THA-specific EHR data from outside facilities to determine external validity.
Relative to the current “gold-standard” of manual chart reviews, the algorithm had an accuracy of 99.2% in identifying the operative approach, 90.7% in identifying the fixation technique, and 95.8% in identifying the bearing surface. The researchers found similar accuracy rates when they applied the algorithm to external operative notes.
The findings from this study strongly suggest that properly “trained” NLP algorithms may someday eliminate the need for manual data extraction. That, in turn, could substantially streamline future research, policy, and surveillance tasks within orthopaedics. As Gwo-Chin Lee, MD predicts in his Commentary on this study, “When perfected, NLP will become the gold standard in the initial data mining of patient records for research, billing, and quality-improvement initiatives.” Dr. Lee is quick to add, however, that “no machine learning can occur…without the integral and indispensable input of the human element.”
Orthopaedic surgeons are already using robots to assist them in performing total joint arthroplasties. Wyles et al. show how we can use technology to reliably expedite research on that same subject. I believe the future holds much promise for the use of ever-advancing technologies in orthopaedic surgery and research.
Matthew R. Schmitz, MD
JBJS Deputy Editor for Social Media
Negative Findings from Level-I Trial Still a Step Forward
Donor-site morbidity from harvesting autologous bone graft has driven the decades-long search for a substitute that performs at least as well as a patient’s own bone. Much of the clinical research on donor-site morbidity is flawed by detection bias, but other factors such as operating-room time and expense are still driving the search for the ideal substitute for autologous bone. Still, the discovery of an ideal bone-graft substitute continues to be elusive.
In The November 6, 2019 issue of The Journal, Myerson et al. report findings from a Level-I trial that investigated the use of adipose-derived cellular bone matrix (ACBM) as a graft substitute in patients undergoing subtalar arthrodesis. Among 57 patients who received autograft and 52 who received ACBM, the substitute delivered lower fusion rates as determined by both CT and plain radiographic/clinical evaluations at 6 months. In addition, patients treated with autologous bone graft had lower rates of serious adverse events.
I commend the authors and funders (AlloSource) of this well-designed clinical trial for reporting these negative results, because it is often just as important to know what doesn’t work as what does. (This manuscript was submitted even after AlloSource decided to halt further production of its ACBM product in 2017.) Such transparent reporting saves other investigators and graft substitute-focused companies from going down similar avenues of investigation. Perhaps even more importantly, publishing negative results such as this might save patients from undergoing procedures with similar formulations that would probably have minimal chance of helping and could do harm.
By contributing to the scientific “process of elimination,” this study brings us one step closer to the identification of a worthy substitute for autologous bone graft.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
“Appropriate” Management of Distal Radial Fractures Improves Outcomes, Lowers Cost
Many surgeons realize that to improve value, we must improve the quality of care while decreasing its cost. Clinical Practice Guidelines (CPGs) developed by the AAOS and other medical societies are designed to help improve the quality of care and safety for patients, while also reducing inappropriate care and decreasing cost. Unfortunately, the evidence used for the development of CPGs is often of mixed quality. It is therefore crucial that studies evaluate patient outcomes when clinicians do and do not adhere to CPGs, so we can ensure that the guidelines are achieving their objective of improving care.
In the October 16, 2019 issue of The Journal of Bone and Joint Surgery, Giladi et al. hypothesize that adhering to Recommendation 3 of the AAOS CPG regarding radiographic indications for operative management of distal radial fractures would yield improved patient outcomes and cost benefits. Recommendation 3 of the CPG suggests that fractures with post-reduction radial shortening of >3 mm, dorsal tilt of >10°, or intra-articular displacement or step-off of >2 mm should be operatively treated. The authors retrospectively reviewed 266 patients, 145 of whom were treated operatively and 121 of whom were treated nonoperatively. Based on the guideline recommendation, only 6 patients were determined to have undergone inappropriate operative fixation, but 68 patients in the nonoperative cohort received inappropriate treatment; many of those had higher-grade fractures that, per the guideline, should have been surgically treated.
Using QuickDASH outcome scores at 4 time points up to 1 year and 1-year direct cost data, the authors compared the appropriately treated operative cohort to both the appropriate and inappropriate nonoperative groups. They also compared the appropriate and inappropriate nonoperative groups to each other. QuickDASH outcomes for appropriate nonoperative treatment were better than those for inappropriate nonoperative treatment at 1 year. In addition, inappropriate nonoperative treatment cost 60% more than appropriate nonoperative treatment. Although this cost comparison did not reach statistical significance, (p=0.23), it does suggest a cost savings with adherence to the CPGs. Appropriately treated operative patients reported less disability than the inappropriately nonoperative group.
As we continue to work at increasing health-care value, it is critical that we review CPGs in action, as Giladi et al. have done in this study. A potential next step would be to investigate whether the modest improvements in QuickDASH scores noted between appropriate operative treatment and inappropriate nonoperative treatment justify the 6-fold higher cost of operative treatment.
Matthew R. Schmitz, MD
JBJS Deputy Editor for Social Media
ACL Grafts: Diameter Does Matter, Sort Of
Clinical failure of anterior cruciate ligament (ACL) reconstructions continues to be a too-common scenario. The increasing incidence of ACL revision is due to a variety of factors, including greater intensity of postsurgical physical activity, technical issues, and anatomical influences of the proximal tibia and distal femur. Registries are important sources of data for ACL-related investigations, but I think they are most useful in clarifying experimental designs for more sophisticated clinical research.
In a cohort study in the October 16, 2019 issue of The Journal, Snaebjornsson et al. examined the influence of ACL graft diameter on the risk of revision surgery over 2 years in >18,000 subjects whose data resided in the national knee ligament registries of Sweden and Norway. The vast majority of those patients (92.8%) received a hamstring autograft, with 7.2% receiving a patellar tendon autograft. Overall, the 2-year rate of ACL revision was 2.63% for patellar tendon autografts and 2.08% for hamstring autografts, a statistically nonsignificant difference in relative risk.
However, the authors found an important correlation between graft diameter in the hamstring tendon cohort, with autografts <8 mm in diameter being associated with a higher risk of revision, compared with larger-diameter hamstring autografts. Additionally, patients treated with hamstring graft diameters of ≥9 mm or ≥10 mm had a lower risk of ACL revision surgery than those treated with patellar tendon grafts of any size.
One key limitation that should influence our interpretation of this study is a lack of detail regarding how compliant surgeons were intraoperatively with the use of the measurement device that is depicted in the manuscript and shown above. In addition, the limitations of registry data did not permit the authors to adjust for postsurgical exposures, such as return to sport, the increasing intensity of which makes rerupture more likely. Additional relevant information that would have aided interpretation of the findings includes the relative size of the tibia and femur, lateral condyle size and shape, and proximal tibial slope.
Despite these limitations, this study should prompt further research that uses robust clinical designs to more fully investigate the impact of graft diameter on ACL rerupture rates.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Surgery for Rotator Cuff Tears: The Better of Two Goods
We’re all familiar with the phrase “lesser of two evils,” but I’m an optimist and prefer the phrase “better of two goods.” In the October 2, 2019 issue of JBJS, Ramme et al. compare surgical versus nonsurgical treatment of full-thickness rotator cuff tears. Both cohorts had improved outcomes relative to baseline, but surgical management was the better of two goods.
The authors retrospectively analyzed a prospective cohort of adult patients with full-thickness rotator cuff tears who had elected either surgical or nonsurgical treatment. Ramme et al. utilized propensity score matching to pair up patients in each group according to factors thought to influence outcome, such as age, sex, tear size, chronicity, muscle atrophy, and the Functional Comorbidity Index. This matched-pair analysis is a valiant attempt to eliminate bias that is inherent in retrospective analyses, and this study design also mimics the real-world scenario of shared decision making between physician and patient.
The 2-year follow-up analysis of 107 propensity score-matched patients revealed that both groups improved in 4 patient-reported functional outcomes and pain compared to their baseline measures before treatment. However, the final outcome measurements and magnitude of improvement were statistically greater in the surgical management group (p <0.001).
This study will help shoulder surgeons have more meaningful discussions with their patients about treatment options for full-thickness rotator cuff tears. We know that with proper treatment—either surgical or nonsurgical—patients can expect improvement in pain and function. However, patients who elect surgical management may have the potential for even greater outcomes, and that definitely sounds like the “better of two goods.”
Matthew R. Schmitz, MD
JBJS Deputy Editor for Social Media
Vancomycin Powder May Thwart Spinal Fusion
Prompted by relatively high infection rates associated with surgical treatment of pediatric spinal conditions such as scoliosis and spinal-deformity surgery in immunocompromised adults, spine surgeons have led “deep dive” clinical research into the possible benefits of local, intrawound antibiotic therapy. Consequently, the administration of antibiotic powder around the spine’s posterior elements and internal-fixation devices has become fairly widespread. But are there possible downsides to this approach that can impact patient outcomes?
This important question is addressed in the basic-science study by Ishida et al. in the October 2, 2019 issue of The Journal. The authors analyzed the fusion-specific impact of varying concentrations of intrawound vancomycin and tobramycin in a well-characterized rat model of posterolateral fusion performed with syngeneic iliac-crest allograft plus clinical bone-graft substitute. Ishida et al. found that a high dose of vancomycin (71.5 mg/kg, about 5 times higher than spine surgeons typically use) but not tobramycin had detrimental effects on fusion-mass formation in this model, as demonstrated by micro-computed tomography and histological analysis.
We now need further clinical research from the spine community to determine the optimal doses and types of intrawound antibiotics in this setting. Based on the currently available data, power calculations should be performed when designing future trials focused on this question. There seems to be little remaining doubt that locally delivered antibiotics help limit surgical-site and deep infections in spinal surgery. The impact of antibiotics on fusion rates must now be investigated further.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Who’s at Risk for Prolonged Opioid Use after THA?
Much has been written in recent years about the orthopaedist’s predilection for prescribing opioids, most of which has been aimed at helping us become better stewards of these medications. It is imperative that we continue learning how best to prescribe opioids to maximize their effectiveness in postoperative pain management, while minimizing their many harmful and potentially lethal effects. With some patients, finding that balance is much easier than with others. Learning to identify which patients may struggle with achieving that equilibrium is one way to address the current opioid epidemic.
In the September 18, 2019 issue of The Journal, Prentice et al. identify preoperative risk factors that are associated with prolonged opioid utilization after total hip arthroplasty (THA) by retrospectively evaluating the number of opioid prescriptions dispensed to >12,500 THA patients. Many of the findings are in line with those of previous studies looking at this question. Prentice et al. found that the following factors were associated with greater opioid use during the first postoperative year:
- Preoperative opioid use
- Female sex
- Black race
- Anxiety
- Higher BMI
- Substance abuse
- Back pain
- AIDS
- Chronic pulmonary disease
For me, the most noteworthy finding was that almost 25% of all patients in the study were still using opioids 271 to 360 days after their operation. That is a much higher percentage than I would have guessed prior to reading this study. Somewhat less surprising but also concerning was the finding that 63% of these patients filled at least 1 opioid prescription in the year prior to their THA, leading the authors to suggest that orthopaedic surgeons “refrain from prescribing opioids preoperatively” or “decrease current opioid users’ preoperative doses.”
Although some readers may be suffering from “opioid fatigue” in the orthopaedic literature, I encourage our community to continue addressing our role in the current opioid crisis. While I believe that we have changed our prescribing practices since the data for this study were collected (2008 through 2011), we cannot dismiss these findings. The opioid epidemic is multifactorial and has many deep-rooted tendrils in our healthcare system. We owe it to our patients and to the public at large to be as significant a part of the solution as possible.
Chad A. Krueger, MD
JBJS Deputy Editor for Social Media
Stop Adding Antibiotics to Irrigation Solutions
The rate of adoption of knowledge gleaned from multiple well-done randomized clinical trials into medical practice is disappointingly slow. This has been well-documented in cardiovascular medicine, and the examples in orthopaedic surgery are embarrassingly similar. A corollary phenomenon exists with the slow rate of transfer of information from basic science studies to orthopaedic clinical practice.
These “disconnects” occur largely because we tend to adopt the practices of our residency faculty, often without any rational inquiry. Having been an oral examiner for the Part II ABOS Oral Boards, I frequently asked, “Why did you decide on that approach to the patient’s problem?” And I often heard in response, “That’s the way it was done in my residency.”
In the September 18, 2019 issue of The Journal, Goswami et al,. report findings from a well-designed in vitro study demonstrating that the common practice of adding the antibiotics polymyxin and bacitracin to irrigation solution to lower the risk of infection is not based on sound evidence. While adding antibiotics might make intuitive sense, according to these authors, it is “a futile exercise.”
After testing 8 different irrigation solutions for efficacy against S. aureus and E. coli and for toxicity to musculoskeletal cells, Goswami et al. concluded that “our results provide further support for the use of dilute povidone-iodine because of its bactericidal properties, relatively limited toxicity,… and modest cost.” They go on to say that their findings bring into question the widespread usage of polymyxin-bacitracin.
Certainly, we need to assemble more evidence from additional research to identify the optimal irrigation solution for orthopaedic surgery, but in the interim, we should probably stop using polymyxin-bacitracin. Doing so would have the added benefits of lowering costs and not exacerbating the serious problem of antimicrobial resistance. There are many areas of clinical practice where we have no evidence either for against a particular approach. But when we do have solid evidence, even if it’s from an in vitro study, we should work together to improve the rates of adoption into clinical practice.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Plate–Tendon Contact: How Important Is It?
There are few things more discouraging for an orthopaedic surgeon than a late postoperative complication after what was an otherwise successful surgery. One such scenario occurs when patients who have undergone open reduction/internal fixation (ORIF) for a distal radial fracture subsequently experience a flexor pollicis longus (FPL) tendon rupture. While previous literature has suggested that plate positioning plays a role in that complication, no studies have evaluated whether newer plate designs decrease contact with the FPL tendon and therefore reduce the risk of rupture.
With that question in mind, Stepan et al. evaluated two cohorts of patients who had undergone ORIF for a distal radial fracture. In the September 4, 2019 issue of JBJS, they report on findings from 40 patients, 20 of whom received a standard distal radial volar locking plate, and 20 of whom received a plate designed with a distal cutout to afford the FPL more room to traverse.
Ultrasound analysis revealed that similar percentages of patients in each group had FPL–plate contact (65% in the FPL-plate group and 79% in the standard-plate group), and there were no differences between groups in terms of FPL tendon degeneration as seen on ultrasound. However, patients who received the FPL plate had significantly less of the tendon come in contact with the plate at 0° and 45° of wrist extension. The authors noted, however, that this difference may have been influenced by the fact that patients with the FPL-specific plate also had significantly lower volar tilt than patients with the standard locking plate. It is therefore not possible to determine whether it was the plate design or the bone position (or both) that led to these results.
It is also noteworthy that the two senior authors of this study work as consultants for the company that manufactures the plates that were evaluated. It is also important to note that because all the patients in this study were asymptomatic, further research is needed to determine the clinical importance of reduced tendon–plate contact area. We should temper our excitement about specially designed volar plates until we have more clinical data supporting their success in avoiding the problem for which they were designed.
Chad A. Krueger, MD
JBJS Deputy Editor for Social Media
Let’s Continue Improving Hip-Implant Longevity
In October 2017, JBJS published results from a 10-year randomized controlled trial by Devane et al. documenting the dramatic reduction in polyethylene wear in total hip arthroplasties (THAs) using highly cross-linked polyethylene (HXLPE). This followed decades of research documenting that wear debris was implicated in macrophage activity that was ultimately responsible for implant loosening. In the September 4, 2019 issue of The Journal, Hart and colleagues produce further evidence of the improved performance of HXLPE, this time showing revision rates among THA patients with osteonecrosis that rival the rates among patients with osteoarthritis.
In this matched cohort of 922 THAs performed from 1999 to 2007 that used an HXLPE bearing, the 15-year cumulative rate of revision was 6.6% among patients treated for osteonecrosis and 4.5% among patients treated for osteoarthritis (p = 0.09). There were no radiographic signs of component loosening in the entire cohort, and, despite a lower median preoperative Harris hip score (HHS) among patients with osteonecrosis, both groups had marked improvements in HHS score. These findings are especially noteworthy because patients with osteonecrosis typically undergo THA at an earlier age and have much higher functional demands than the typical 70- or 80-year-old osteoarthritis patient.
However, the 15-year revision rate—even with HXLPE—remains at 4.5% for osteoarthritis patients, which should provide impetus to continue our work identifying all possible factors and mechanisms that lead to THA revision. A partial list would include bearing-surface wear, reliability of implantation, biomechanics, biomaterials, and patient perception of postoperative pain. Also, in a subgroup analysis, Hart et al. found that the 15-year rate of any reoperation among osteonecrosis patients ranged from 0% for hips with radiation-induced osteonecrosis to 25% for hips with idiopathic osteonecrosis. These findings add to the list of factors for THA success that need further investigation.
The work list for improvements in THA will remain substantive for at least the next few decades, and we may never get to 0% revisions for all patients. But we have certainly demonstrated that our research can produce very worthwhile results.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Recent Comments