Use of Machine Learning to Predict Improvement After Hip Arthroscopy
The management of expectations is crucial when counseling patients undergoing treatment for a musculoskeletal injury or condition. In hip arthroscopy, this is especially critical when discussing with patients—including athletes seeking to return to play—their anticipated outcomes following surgical treatment for femoroacetabular impingement syndrome (FAIS).
In the latest issue of JBJS, Kunze et al. report on their investigative efforts to develop and internally validate machine learning algorithms that can yield patient-specific predictions of which athletes will reach clinically relevant improvement in function after arthroscopy for FAIS.
A total of 1,118 athletes, identified through a retrospective review of clinical registry data, met the inclusion criteria. The primary outcome was attaining the minimal clinically important difference (MCID) in the Hip Outcome Score-Sports Subscale (HOS-SS) at a minimum of 2 years postoperatively. Six machine learning algorithm models were tested.
The authors found that 23.1% of the athletes did not achieve the MCID for the HOS-SS. Six variables optimized algorithm performance, with the following cutoffs found to decrease the likelihood of achieving the MCID:
- Preoperative HOS-SS score of ≥58.3
- Tönnis grade of 1 (early osteoarthritis)
- Alpha angle of ≥67.1° on anteroposterior radiograph
- Body mass index (BMI) of >26.6 kg/m2
- Tönnis angle of >9.7° (indicating subtle instability or dysplasia)
- Patient age of >40 years
The elastic-net penalized logistic regression (ENPLR) model was the most accurate model in this study.
The findings suggest that patient selection is paramount to the ability to achieve clinically relevant improvements in outcomes for patients treated with arthroscopy for FAIS. Multiple studies have demonstrated that increasing arthritis level and age, along with BMI, are associated with inferior patient-reported outcomes. In addition, hip instability and increased Tönnis angle have been shown to be associated with worse outcomes following hip arthroscopy. A greater alpha angle indicates a larger “deformity” and thus the potential for more damage at the time of surgery that cannot be completely addressed with today’s surgical techniques. “Higher” preoperative HOS-SS (although on a scale of 0 to 100, 58 is not that high) may make it more difficult for a patient to achieve enough of an improvement in their outcome score to be considered as having attained the MCID.
The ENPLR model was converted into an open-source application, although as Kunze et al. point out, external validation is necessary before wider adoption of the application. Nonetheless, the model demonstrates the potential to help hip surgeons better educate our patients on expected outcomes and to assist with proper patient selection for the ever-evolving treatment of FAIS.
Matthew R. Schmitz, MD
JBJS Deputy Editor for Social Media
Co-author Kyle N. Kunze, MD discusses this study in an “Author Insights” video, found here.
Risk Factors for Failure after FAI Treatment
Orthopaedic surgeons continually seek to refine techniques to improve their patients’ surgical outcomes. Surgical treatments for femoroacetabular impingement (FAI) syndrome are no exception, and careful patient selection is also critical to the success of these interventions. In the June 17, 2020 issue of The Journal, Ceylan et al. analyzed a single-surgeon prospective database to identify risk factors for treatment failure after a particular hip-preservation surgery known as mini-open femoroacetabular osteoplasty (FAO). In this study, the authors defined “failure” as the eventual need for a total hip arthroplasty (THA) over a minimum 2-year follow-up.
The 749 procedures studied were performed between 2004 and 2016 and involved treatment of the femur, acetabular rim, labrum, and chondral surfaces if necessary. Labral repair was performed on all hips that had adequate healthy tissue, while those that did not were treated with partial or total excision of the labrum.
Sixty-eight hips (9%) underwent THA. The patients who did not need a hip replacement were significantly younger (mean age of 33 years vs nearly 42) and were operated on after the surgeon had more experience. Other significant differences among the failure group included the duration of symptoms (twice as long, at 3.6 years), higher preop alpha angles, and a higher percentage of total labral resections performed.
Radiographic evidence of hip dysplasia was also a significant risk factor for failure, along with labral hypertrophy and acetabular retroversion (both of which may be considered proxies for volume-deficient acetabuli). After adjusting for covariates, Ceylan et al. found that less surgeon experience, older patient age, prolonged preoperative symptoms, increased medial joint space narrowing and Tonnis grade, and developmental hip dysplasia were all associated with a higher risk of failure after FAO surgery.
Although these findings do not represent results using the most up-to-date arthroscopic techniques for FAI treatment, they do highlight characteristics that can and should be discussed with patients with FAI when the subject of expected surgical outcomes arises during shared decision making.
Matthew R. Schmitz, MD
JBJS Deputy Editor for Social Media
Where Does the Blood Flow in the Femoral Head?
Osseous vascular anatomy has always been clinically relevant to orthopaedists, but its importance is sometimes overlooked. In the July 19, 2017 issue of The Journal, Rego et al. provide a precise topographic map of arterial anatomy in and around the femoral head.
Ever since Trueta’s classic work published in the British volume of JBJS in 1953, we’ve known that the terminal branches of the medial femoral circumflex system (also known as the lateral epiphyseal artery complex) supply blood to the majority of the femoral head. This information has proved critical in supporting treatment decisions for the management of femoral head and neck fractures. In those cases, surgeons typically perform ORIF through an anterior approach because it is remote from this posterior vascular supply.
The details in the Rego et al. study will help today’s and tomorrow’s arthroscopists more safely manage acetabular labral tears associated with cam deformities. In those settings, when increasing the “offset” across the femoral neck to decrease impingement, surgeons should limit the depth of bone removal to avoid injury to this important vascular network. Thanks to this study, operating surgeons now have precise anatomic information (albeit derived from non-deformed cadaver hips) with which to limit the risks of increasing the femoral head offset.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Does Hip Arthroscopy Really Help?
Over the past 15 to 20 years, the use of arthroscopic procedures for hip pathologies has rapidly increased. Leaders in sports medicine have standardized many arthroscopic techniques, including methods of joint distraction, portal location, approaches to labral repair or debridement, and management of cartilage lesions.
Many in the orthopaedic community have wondered whether this expansive use of hip arthroscopy is justified by significant improvement in patient function or is simply a first (and perhaps overused) step toward inevitable hip arthroplasty. To help answer that question, in the June 21, 2017 issue of The Journal, Menge et al. document the 10-year outcomes of arthroscopic labral repair or debridement in 145 patients who originally presented with femoroacetabular impingement (FAI).
Whether these patients were treated with debridement or repair, their functional outcomes and improvement in symptoms were excellent over the 10-year time frame, and the median satisfaction score (10) indicates that these patients were very satisfied overall. However, as seen in other similar studies in the peer-reviewed literature, the results in older patients with significant cartilage injury or radiographic joint space narrowing were inferior, and most of the patients with these characteristics ended up with a hip replacement.
The Menge et al. study helps confirm that arthroscopic repair or debridement in well-selected FAI patients yields excellent longer-term outcomes, and it provides concrete criteria for patient selection.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Plusses and Minuses of Capital Femoral Growth Plate Extension
Femoroacetabular impingement (FAI), especially in adolescent athletes, has received a lot of attention from orthopaedists in the last 15 years. In the May 18, 2016 edition of The Journal of Bone & Joint Surgery, a longitudinal radiographic study by Morris et al. sheds light on how a measurement called the epiphyseal extension ratio (EER) delivers excellent diagnostic accuracy for predicting cam morphology of the femoral head, one of the main causes of FAI.
The authors carefully analyzed at least five consecutive annual hip radiographs from 96 healthy adolescents. Specifically, they measured changes in the anteroposterior alpha angle and the superior EER (the superior epiphyseal extension divided by the femoral head diameter). They found a mean increase in alpha angle and EER between Oxford bone age (OBA) stages 5 and 7/8. The mean EER increased significantly at each stage, with the greatest increase occurring between OBA stages 6 and 7/8.
In this study, the EER showed excellent diagnostic accuracy for predicting a final alpha angle of ≥78, which prior research has suggested is a threshold that predicts an increased risk for developing end-stage hip osteoarthritis. However, as commentator John H. Wedge, MD emphasizes, Morris et al. “do not recommend radiographic screening for this marker.”
Dr. Wedge adds that this study lends credence to the hypothesis that cam deformity develops from chronic impingement before rather than after proximal femoral physeal closure. But perhaps the most interesting messages are in the discussion section, where Morris et al. state that “epiphyseal extension may be a physiologic, protective response to increased physeal shear forces that decreases the risk of progression to SCFE [slipped capital femoral epiphysis].” The authors describe the cam-morphology downside of epiphyseal extension as “the unfortunate long-term consequence of a short-term adaptive response.”
Naproxen Effective for Preventing HO after Hip Arthroscopy
Heterotopic ossification (HO) is a known complication of hip arthroplasty. A double-bl
ind, randomized, placebo-controlled trial by Beckmann et al. in the December 16, 2015 Journal of Bone & Joint Surgery showed that prophylaxis with naproxen dramatically reduced the prevalence of HO after hip arthroscopy, without serious medication-related side effects. These findings bolster findings from previous retrospective investigations that showed large reductions in HO prevalence among those taking nonsteroidal anti-inflammatory drugs (NSAIDs).
The patients in the study took naproxen (500 mg) or a placebo twice a day for three weeks following arthroscopic surgery for femoroacetabular impingement. After one year, the prevalence of radiographically determined HO in patients randomized to the naproxen group was 4% versus 46% in the patients randomized to the placebo group, an 11-fold difference. While the potential for serious GI and renal side effects with NSAIDs is well-documented, in this study only minor adverse reactions to study medication were reported in 42% of those taking naproxen and in 35% of those taking placebo.
Noting that the clinical consequences of HO following hip arthroscopy are “largely undetermined,” the authors still suggest a role for HO prophylaxis “because it could reduce the risk of developing symptomatic HO or requiring revision surgery for HO excision.”
In an accompanying commentary, Sverre Loken praises the authors for the well-designed study, but he cautions that “clinically relevant HO is uncommon, and this has to be weighed against the risk of serious side effects caused by NSAIDs.” He also emphasizes the observation Beckmann et al. make in the last paragraph of their study: that “the lowest dose and shortest duration of NSAID prophylaxis that still prevent HO remain to be determined.”
JBJS Editor’s Choice—Patient Pain: A Hurdle for All Orthopaedists
Ev
ery clinician treating musculoskeletal injury or disease knows that pain perception among patients is highly subjective and variable. Given the same objective magnitude of a pain stimulus, one person will grade it a 2 on the visual analog scale (VAS), while another will rate it an 8. I am sure that every dentist experiences similar patient variability! What is behind this, and what can we do with our decision making related to pain management to ensure compassionate and effective orthopaedic care?
We know that cultural and social factors play a role in pain perception, as do smoking and opiate-abuse history. Now, in a prognostic study in the August 5, 2015 edition of The Journal of Bone & Joint Surgery, Ernat et al. identify an association between pharmacologic treatment for anxiety and depression and poor outcomes, including higher postoperative pain scores, following primary surgery for femoroacetabular impingement (FAI) among members of the US military. The between-group difference in pain scores was significant only for antidepressant use, but 33 of the 37 patients in the study who took mental-health medications were on antidepressants.
I wonder whether the anxiety and depressive response to situational or relational stimuli that prompt an individual to seek mental-health treatment may be closely related to the same person’s response to painful musculoskeletal stimuli. Alternatively, incompletely treated anxiety or depression may influence a patient’s pain response to surgical treatment of FAI.
Either way, we need more research in this area so we can better manage our patients. An interesting study by Kane et al. that tested various approaches to standardizing patient pain reports showed how difficult normalizing pain scores is, but we still need to encourage further research into responses to painful stimuli, whether they be psychological or physical.
Marc Swiontkowski, MD
JBJS Editor-in-Chief
Try Clinical Classroom
Recent Comments