5:00 PM
|
Evaluating the Efficacy of Ultrasound-Guided Carpal Tunnel Release: A Systematic Review and Meta-Analysis
Introduction: Ultrasound-guided carpal tunnel release (USCTR) has emerged as an increasingly popular minimally invasive technique for the treatment of carpal tunnel syndrome (CTS), boasting reduced scarring and morbidity compared to open carpal tunnel release (1). The purpose of this systematic review and meta-analysis is to evaluate the efficacy and safety of USCTR and compare it to traditional open techniques.
Methods: A search was conducted within PubMed, Embase, ScienceDirect, and CENTRAL to include papers from inception to January 2025 using the following search query: ("ultrasound") AND ("guided" OR "guidance" OR "guide") AND ("carpal tunnel release"). Studies were included if they described USCTR with complete transverse carpal ligament release. Review papers and studies with combined techniques, steroid injections, or cadaver subjects, were excluded. Included studies were assessed using Cochrane risk of bias, Newcastle Ottawa, and Joanna Briggs Institute critical appraisal tools for randomized controlled trials, cohort studies, and case series, respectively. Extracted data included patient demographics, functional outcomes, advanced imaging and nerve conduction metrics, patient-reported outcomes, and complications or reoperations. Meta-analysis was conducted on studies comparing USCTR with open techniques using a random effects model. Sensitivity analysis was performed for all meta-analyses with a heterogeneity greater than 50%, and significance was set to p<0.05.
Results: Literature search yielded 284 results, with 43 articles representing 39 unique patient populations. Risk of bias assessment showed all studies had low to moderate risk. Studies included 4,067 total patients (4,222 wrists): 3,825 USCTR, 130 mini-open carpal tunnel release, and 112 open carpal tunnel release. Follow-up times ranged from 6-332 weeks, with 75% of articles reporting follow-up times of at least 26 weeks. USCTR demonstrated improvements in functional outcomes relative to pre-operative baseline, with improvement in two-point discrimination and the Semmes-Weinstein monofilament test. Nerve conduction metrics were also improved from baseline for both distal motor latency and sensory conduction velocity, similar to changes seen after open release (OR: -0.11; p=0.56, I2=0%; OR: -1.01; p=0.46, I2=35%). Patient-reported outcomes, including the Boston Carpal Tunnel and Quick Disabilities of the Arm, Shoulder, and Hand Questionnaires, also showed improvements exceeding the established minimal clinically important difference in the literature (MCID). Time to return to work was faster by 10 days (p=0.02; I2=10%), and time to normal activities significantly was faster by 21 days (p<0.001; I2=0%) in patients receiving USCTR compared to open techniques. Overall complication rates were low after USCTR (1.57%), with the most common complications being pillar pain (0.24%), infection (0.24%), and post-operative paresthesia (0.18%). Overall complication rates were found to be comparable to open techniques (OR, 0.87; CI, 0.16 to 4.87; p=0.88; I2 = 46%).
Conclusions: USCTR is safe and effective, with similar overall complication rates to open release and the benefit of faster return to work and normal activities. Further research should include large-scale prospective randomized trials comparing USCTR, endoscopic release, and open release.
References:
1. Petrover D, Richette P. Treatment of carpal tunnel syndrome : from ultrasonography to ultrasound guided carpal tunnel release. Joint Bone Spine. 2018;85(5):545-552. doi:10.1016/j.jbspin.2017.11.003
|
5:05 PM
|
GAME OVERuse: Management of Upper Extremity Pain in Esports Players
Introduction
Electronic sports (Esports) players are prone to overuse injuries in their hands and wrists due to repetitive motions and extended gaming sessions. The high frequency of clicking on game controllers or keyboards in Esports players and fine, quick movements can lead to excess strain and injuries to the upper extremity, such as tendonitis and carpal tunnel syndrome. This study investigates the association of level of Esports, type of game, and time spent gaming with hand, wrist, and shoulder injuries in gamers.
Methods
A survey study was conducted to analyze patterns of upper extremity injuries in the gaming population. Data was collected through a questionnaire that was distributed to participants at gaming competitions, collegiate Esports organizations, and online gaming communities. The survey included demographics, gaming habits and experience, history of upper extremity injuries or pain, and medical treatments for these conditions. Univariate analysis was performed to determine patterns of pain.
Results
A total of 143 survey responses were received. For gaming occupation, there were 84 (73.0%) casual gamers, 3 (2.6%) professional streamers, and 25 (21.7%) collegiate esports athletes. The average age was 26.0 ± 7.5 with no major comorbidities. 111 (78.2%) identified as male, and 31 (21.8%) identified as female. For input, 77 (66.4%) used mouse and keyboard and 39 (33.6%) used a controller. All mouse and keyboard users used a right-handed mouse. On average, players spent 7.0 ± 5.4 hours of structured gaming time (tournaments, scrimmages, training, etc.) and 18.2 ± 15.6 hours of casual gaming every week. Over half of the respondents (74, 51.7%) reported experiencing hand or wrist pain. There was no significant difference between pain and age, gender, occupation, game type, or type of input. Ten of eleven (90.9%) left-handed respondents experienced pain while 58 of 124 (46.8%) right-handed respondents experienced pain (p = 0.008). Respondents who experienced hand pain spent an average of 7.0 ± 6.3 hours per day in front of a computer, compared to 5.6 ± 3.0 hours for participants who experienced no pain (p = 0.09). Of the participants who sought medical advice, the most common diagnoses were sprain/strain (n = 12), carpal tunnel syndrome (n = 3), and tendonitis (n = 5). Most common interventions include none (n = 67), rest and avoidance of activity (n = 25), ergonomic changes (n = 9), hand therapy/exercises (n = 18), bracing (n = 20), over-the-counter pain medications (n = 5), and surgery (n = 2).
Discussion
From this study, over half of Esports players report hand, wrist, or shoulder pain, indicating a common problem for this occupation. While no significant association was found between hours spent gaming weekly and pain, it is likely that increased gaming time can lead to overuse strain. These findings can raise awareness amongst gamers in this occupation to recognize upper extremity injuries and seek medical intervention with hand specialists when necessary.
|
5:10 PM
|
Shifting Trends in Thumb Base Arthritis Epidemiology: Evidence for Rising Trapeziometacarpal Arthrosis in Younger Patients
Purpose: Arthrosis of the trapeziometacarpal (TMC) joint is a common pathology that is associated with increased age. Research conducted by Sodha et. al (2005), prior to the invention of the smartphone, demonstrated that radiographic findings of TMC arthrosis are uncommon in patients under 50 years old (1). Given the increased use of smartphones, gaming systems, and computers, one may expect to see higher rates of TMC arthrosis in younger patients owing to thumb loading and repetitive use. This study aims to describe the prevalence of TMC osteoarthrosis across the lifespan in the digital era.
Methods: We conducted a retrospective chart review of patients who presented with a distal radius fracture (DRF) at a single academic hospital between August 2022 and December 2023. Radiographs were assessed for trapeziometacarpal (TMC) osteoarthrosis (OA) by two independent raters. Inter-rater agreement was moderate (k=0.513), while intra-rater reliability was substantial (k=0.713). OA severity was classified on a three-grade scale: Grade I (no arthrosis), Grade II (mild to moderate arthrosis), and Grade III (complete joint destruction). Continuous variables are presented as mean ± standard deviation. Ordinal logistic regression was used for univariate analysis to assess OA prevalence across age groups. A multivariate logistic regression model, adjusted for age, was used to evaluate sex differences in OA severity. A chi-squared analysis was also conducted to compare OA rates with Sodha et. al (2005).
Results: A total of 602 patients were included in the analysis and 60.2% of the cohort was female. Evidence of early-onset trapeziometacarpal (TMC) osteoarthritis (OA) was observed in younger age groups (patients < 50 years old). Among patients aged 11-20, 5.0% had Grade II arthrosis, a significantly higher prevalence than previously reported by Sodha (2005) (p = 0.033). Similarly, OA rates were significantly elevated in patients aged 21-30 (13.5% vs. 3.2%, p<0.001) and 31-40 (21.2% vs. 2.0%, p = 0.004). The radiographic prevalence of OA increased with age but stabilized above 90% when aged >70 years old. However, the proportion of patients with Grade III arthrosis continued to increase with age. Univariate ordinal regression demonstrated a 2.63-fold increased risk of developing OA with each successive age cohort (p<0.001). After adjusting for age, there was no significant difference in OA severity between sexes.
Conclusion: Radiographic prevalence of TMC OA may be increasing in younger patients, with significantly higher rates than previously reported by Sodha (2005). The findings are also consistent with past research that shows increases in the prevalence of TMC arthrosis with advancing age. The results of this study suggest a potential shift in the epidemiology of TMC OA, which may warrant earlier screening and preventive strategies. Further research is needed to evaluate the role of modern technology use in TMC OA development in younger cohorts.
References
1) Sodha S, Ring D, Zurakowski D, Jupiter JB. Prevalence of osteoarthrosis of the trapeziometacarpal joint. J Bone Joint Surg Am. 2005 Dec;87(12):2614-2618. doi: 10.2106/JBJS.E.00104. PMID: 16322609.
|
5:15 PM
|
Predictors of Digit Replantation and Revascularization Outcomes
Purpose: The purpose of this study was to determine how different factors affect digit replantation and revascularization outcomes. We hypothesized that positive outcomes would be associated with primary arterial/venous repairs, less comorbidities, and shorter ischemia times.
Methods: Patients that received digit replantation or revascularization procedures between 2019-2024 were queried through our institution's electronic medical record system using ICD-10 code. All patients maintained adequate follow-up to determine digit survival. Extracted data included demographics, injury characteristics, time to care, and surgical management. Fisher's exact tests were used to analyze categorical variables and Mann-Whitney U tests were used for non-parametric comparisons, with p<0.05 for statistical significance. Odds ratios and 95% confidence intervals were calculated, with post-hoc stratification used to control for confounding variables.
Results: Our analysis included 60 digits (43 patients), with 21 replantations (19 patients) and 39 revascularizations (25 patients). Success rates for replants and revascs were 33.3% and 74.4%, respectively, with an overall success rate of 60%. Time from injury to presentation for replants had a mean time of 2.7 hours for successful replants and 4.9 hours for failures (OR 0.6, P=.04). Revascs had a mean Elixhauser comorbidity index of 1.75 for successes and 4.7 for failures (OR 0.06, P<.001), with the most significant comorbidities being cardiac arrhythmia (OR 0.02, P<.001), history of significant unintentional weight loss (OR 0.1, P=.02), and psychosis (OR 0.08, P=.03). Sharp injuries had a revasc success rate of 95% while crush/avulsion injuries were 47% (OR 0.04, P=.001). Replants with one artery repaired by venous autograft and at least one repaired vein had a 77.8% higher success rate compared to other vessel repair combinations (OR 22.6, P=.03). In non-sharp revascs, increased success rates were associated with primary venous repairs (OR 13.7, P=.03), while primary arterial repairs were negatively associated (OR 0.07, P=.049). Repairing nerves increased revasc success rate by 66% (OR 34.7, P<.001). Surgeons with hand fellowship training had a 55% higher revasc success rate compared to plastic surgeons without hand fellowship training (OR 12, P=.04). Intra-op use of papaverine during vessels anastomosis decreased revasc success rates by 45% (OR 0.05, P=.002). Starting a continuous heparin drip at 500 U/h post-op and continuing until discharge increased replant success rate by 57% (OR 15, P=.02). Post-op aspirin for active smokers increased the success rates of revasc by 86% (OR 25, P=.03). Patients that drink alcohol had 33% lower revasc success rates (OR 0.13, P=.04).
Summary:
• Using venous autografts to repair arteries in combination with repairing at least one vein may have better replant outcomes than other vascular repair techniques.
• Post-op heparin drip at 500 U/h may increase replant success rate.
• Non-sharp injuries, such as crush/avulsion injuries, may have better survival rates when arteries are repaired with autografts instead of primary anastomosis.
• Intra-op papaverine use during vessel anastomosis may decrease revasc success rates.
• The Elixhauser comorbidity index may be beneficial in predicting outcomes, with cardiac arrythmia, weight loss, and psychosis being more influential.
|
5:20 PM
|
Impact Of Patient Frailty On Brachial Plexus Repair Outcomes
Purpose: Brachial plexus nerve procedures, including repair, decompression, reconstruction, and tumor excision are highly efficacious procedures. However, to date, there has been no reporting on the effect of patient frailty on postoperative outcomes. The current literature identifies a need for a comprehensive review of surgical outcomes beyond motor recovery. This study aims to provide an analysis of the effect of patient frailty on brachial plexus surgery outcomes.
Methods: The 2020 National Inpatient Sample was utilized to perform a retrospective analysis to compare brachial plexus repair postoperative outcomes based on frailty status. All patients over the age of 18 who underwent brachial plexus repair from 2016-2020 were included. The Modified Frailty Index (mFI-5, mFI-11), Charleston Comorbidity Index (CCI), and Risk Analysis Index (RAC-A, RAC-C) were used to calculate frailty scores, and the scores were used to categorize patients into one of the following frailty categories: Not Frail, Pre-Frail, Frail, or Severely Frail. Postoperative outcomes, including in-hospital mortality, complication rate, length of stay, and cost, were compared between groups using one-way ANOVA.
Results: 16,215 patients were included in the final analysis. CCI and mFI were most effective in attributing frailty to postoperative outcome across all assessed metrics based on area under ROC curve. Based on mFI-5 score, 6,810 patients (42%) were not frail, 6,695 (41.3%) were pre-frail, 1,970 (12.1%) were frail, and 740 (4.6%) were severely frail. Age, Medicare/Medicaid status, and rural/urban status were significantly associated with frailty categorization (p < 0.001). Severely frail patients were associated with significantly higher in-hospital mortality than the total cohort (1.4% vs 0.1%, p < 0.001), increased incidence of non-home discharge (29.7% vs 15.7%, p < 0.001), and longer length of stay (5.946±8.240 vs 4.183±6.910 days, p < 0.001). Frail patients were also significantly associated with increased overall complication rate (23.0% vs 9.7%, p < 0.001), specifically, cardiac (2.7% vs 0.6%, p < 0.001), pulmonary (8.8% vs 4.7%, p < 0.001), thromboembolic (2.0% vs 1.3%, p < 0.001), renal, and genitourinary (15.5% vs 4.4%, p < 0.001), DVT (2.0% vs 1.3%, p < 0.001), and sepsis (4.7% vs 0.6%, p < 0.001). Based on mFI-5, frailty was associated with a 72.5% increase in odds of complication (OR: 1.725, 95% CI: 1.457-2.043, p < 0.001), and a 28.3% increase in odds of extended length of stay (OR: 1.283, 95% CI: 1.130-1.457, p < 0.001).
Conclusions: Frailty is associated with worse postoperative outcomes for patients undergoing brachial plexus nerve procedures across nearly every outcome metric assessed in this study. The Charleston Comorbidity Index (CCI) and Modified Frailty Index (mFI-5) showed the greatest ability to attribute frailty to outcome measures. Further research is required to identify differential patient-reported outcomes, such as pain, based on frailty status.
|
5:25 PM
|
Rheumatoid arthritis increases odds of short-term and long-term adverse events following distal radius fracture open reduction and internal fixation
Purpose:
While open reduction and internal fixation (ORIF) is commonly performed for distal radius fractures, patients with rheumatoid arthritis (RA) may experience increased complications related to medical considerations, impaired healing, compromised bone quality, and medication effects. The current study evaluated outcomes of RA patients undergoing distal radius ORIF.
Methods:
Using the M165Ortho PearlDiver Mariner Database (2010-2022), patients undergoing distal radius fracture ORIF were identified, excluding those with neoplasm, infection, or polytrauma. Patients were categorized into two groups based on whether they did or did not have rheumatoid arthritis (RA). A 1:4 matching algorithm was applied based on age, sex, and Elixhauser Comorbidity Index (ECI) between the two cohorts. RA patients were further stratified by medication use within one year before surgery.
Short-term complications included 90-day serious adverse events (DVT, PE, SSI, sepsis, cardiac) and minor adverse events (pneumonia, UTI, AKI, wound dehiscence). Long-term complications included 2-year rates of tendon rupture, complex regional pain syndrome (CRPS) and wrist diagnoses within 3 to 6 months post-surgery. Short-term and long-term complications following surgery were compared with univariate and multivariate analyses.
Results
Of 344,115 patients who underwent DRF ORIF, RA was noted for 13,978. After matching, 54,439 patients were in the non-RA cohort and 13,651 in the RA cohort. Of those with RA, related medications were taken within one year prior to surgery by 10,663 and not taken by 2,988.
RA patients demonstrated significantly higher odds of 90-day serious adverse events (OR 1.61, p<0.001) and minor adverse events (OR 1.35, p<0.001). Specific complications included increased surgical site infections (OR 2.16), pulmonary embolism (OR 2.12), deep vein thrombosis (OR 1.68), UTI (OR 1.55), and acute kidney injury (OR 1.76) (all p<0.001).
Long-term complications were notably higher for those with RA, including tendon rupture (OR 2.33), CRPS (OR 1.24), and wrist stiffness (OR 1.15) (all p<0.001). RA patients without medication use showed worse outcomes compared to those on medication management with regard to tendon rupture (OR 2.55 vs 2.28) and surgical site infections (OR 2.07 vs 1.93).
Conclusions:
RA patients undergoing distal radius fracture ORIF face significantly higher odds of both short-term and long-term complications. These findings emphasize the importance of tailored perioperative management and careful monitoring for this high-risk population. The fact that medications use appeared to correlate with improved surgical outcomes suggests that control of the underlying disease helps mitigate its effects.
|
5:30 PM
|
Linking Radiographic Severity And Clinical Symptoms: A Simplified Three-grade Classification For Trapeziometacarpal Osteoarthritis
Purpose: The relationship between radiographic findings of trapeziometacarpal (TMC) osteoarthritis (OA) and clinical symptoms is often inconsistent. Many patients with severe radiographic OA experience minimal pain and functional impairment, while others with mild changes report significant symptoms. Prior studies primarily utilized the Eaton-Littler classification or specific radiographic features, limiting their clinical utility. This study aims to establish a practical framework for assessing the correlation between radiographic severity and symptoms by implementing a simplified three-grade classification system.
Methods: We conducted a retrospective chart review of patients who presented with a distal radius fracture at an academic hospital between August 2022 and December 2023. Radiographs were assessed for TMC OA by two independent raters. Inter-rater agreement was moderate (k=0.513), while intra-rater reliability was substantial (k=0.713). OA severity was classified on a three-grade scale: Grade I (no arthritis), Grade II (mild-to-moderate arthritis), and Grade III (complete joint destruction). Univariate analyses were conducted to determine the association between radiographic OA severity and prior thumb pain and treatment. A multivariate logistic regression model, adjusted for age and sex, was used to evaluate differences in symptoms and treatment rate across OA severity grades.
Results: A total of 602 patients were included in the analysis and 38.7% of patients being >50 years old. Among patients over 50 years old, univariate analysis indicated that those with Grade II OA were about 2.5 times more likely to report thumb pain history (OR=2.54, 95% CI: 1.15–6.28, p=0.029), whereas those with Grade III OA had 3.75 times greater odds of reporting pain (OR=3.75, 95% CI: 1.48–10.25, p<0.007) compared to individuals with Grade I OA. After adjusting for age and sex, Grade III OA remained significantly associated with increased odds of thumb pain (OR=2.77, 95% CI: 1.18–6.41, p=0.019), whereas the association for Grade II OA was not statistically significant (OR=1.95, 95% CI: 0.91–4.40, p=0.095).
Univariate analysis showed that patients with Grade III OA were 7.5 times more likely to have received treatment for thumb OA (p=0.003), while those with Grade II OA (OR=3.07, p=0.078) did not differ significantly from Grade I OA. After adjusting for age and sex, Grade II OA (OR=2.1, 95% CI: 0.64–9.8, p=0.259) remained non-significant, whereas Grade III OA was significantly associated with increased treatment likelihood (OR=5.0, 95% CI: 1.3–24.6, p=0.027). Each additional year of age increased the odds of receiving treatment for thumb OA by 4.6%, though this did not reach statistical significance (p=0.073).
Conclusion: Our findings demonstrate that a simplified three-grade radiographic classification system for TMC OA is associated with increasing odds of thumb pain and arthritis treatment in patients over 50 years old. Unlike previous studies, our approach provides a more accessible and clinically intuitive method for assessing OA severity and identifying patients at risk for symptomatic disease-one that can be easily utilized by primary care physicians and occupational therapists. These findings reinforce the importance of age as a predictor of treatment for thumb OA and suggest that future prospective studies to assess prospective risk of developing TMC OA.
|
5:35 PM
|
Implementing Privacy-Focused AI for Clinical Note Data Extraction in Neonatal Brachial Plexus Palsy Care
Purpose
Patient notes contain a wealth of information valuable for clinical decision-making, quality improvement, and tracking surgical recovery. However, manual extraction of this data is time-consuming and error-prone. Though large language models (LLMs) like ChatGPT could streamline the extraction of clinical data, the use of ChatGPT in healthcare is constrained by privacy concerns due to data being sent to external servers. LLMs run on a single computer could help overcome these barriers, facilitating real-time data collection to support clinical decision-making while ensuring patient privacy. We hypothesize that privacy-focused large language models can accurately extract critical clinical data from neonatal brachial plexus palsy (NBPP) patient notes, supporting clinical decision-making for surgery and postoperative outcome tracking while preserving patient privacy.
Methods
A cohort of 355 NBPP patients from Duke University was identified using ICD codes for brachial plexus birth injuries, with preoperative clinical notes available for 86 patients. To extract critical clinical data, we used Llama 3.2, a large language model deployed via Ollama, an open-source platform enabling local execution of LLMs. Variables extracted including the presence of Horner's syndrome (yes/no), laterality (right/left), pseudomeningocele (yes/no), and functional scores across 15 Active Movement Scale (AMS) categories (scored 1 to 7). Structured prompting was used to guide model output, using queries such as "Extract the laterality of the injury (right/left)". For AMS scores, we prompted the model to extract each of the 15 functional scores from clinical notes. The output is a structured CSV (one row per patient).
Results
For structured data, such as laterality and Horner's syndrome, the model demonstrated strong performance: 94% accuracy, 96% sensitivity, and 91% specificity for laterality, and 93% accuracy, 99% sensitivity, and 75% specificity for Horner's syndrome. For pseudomeningocele identification, the model achieved 76% accuracy, 75% sensitivity, and 83% specificity. AMS scores showed variability in performance, with accuracies ranging from 58% to 71% across categories. These results highlight the model's strength in extracting categorical data but also reveal limitations in its ability to consistently extract granular numerical information, such as AMS scores.
Conclusions
Privacy-focused, locally-run LLMs can accurately and securely extract key clinical data. The model performed well with structured categorical information but showed variability with granular numerical data. These findings highlight both the promise and limitations of AI-driven data extraction in clinical settings. Continued refinement, validation, and oversight will be essential to optimize AI tools for accurate, privacy-preserving data extraction in patient care and research.
|
5:35 PM
|
Implementing Privacy-Focused AI for Clinical Note Data Extraction in Neonatal Brachial Plexus Palsy Care
Purpose
Patient notes contain a wealth of information valuable for clinical decision-making, quality improvement, and tracking surgical recovery. However, manual extraction of this data is time-consuming and error-prone. Though large language models (LLMs) like ChatGPT could streamline the extraction of clinical data, the use of ChatGPT in healthcare is constrained by privacy concerns due to data being sent to external servers. LLMs run on a single computer could help overcome these barriers, facilitating real-time data collection to support clinical decision-making while ensuring patient privacy. We hypothesize that privacy-focused large language models can accurately extract critical clinical data from neonatal brachial plexus palsy (NBPP) patient notes, supporting clinical decision-making for surgery and postoperative outcome tracking while preserving patient privacy.
Methods
A cohort of 355 NBPP patients from Duke University was identified using ICD codes for brachial plexus birth injuries, with preoperative clinical notes available for 86 patients. To extract critical clinical data, we used Llama 3.2, a large language model deployed via Ollama, an open-source platform enabling local execution of LLMs. Variables extracted including the presence of Horner's syndrome (yes/no), laterality (right/left), pseudomeningocele (yes/no), and functional scores across 15 Active Movement Scale (AMS) categories (scored 1 to 7). Structured prompting was used to guide model output, using queries such as "Extract the laterality of the injury (right/left)". For AMS scores, we prompted the model to extract each of the 15 functional scores from clinical notes. The output is a structured CSV (one row per patient).
Results
For structured data, such as laterality and Horner's syndrome, the model demonstrated strong performance: 94% accuracy, 96% sensitivity, and 91% specificity for laterality, and 93% accuracy, 99% sensitivity, and 75% specificity for Horner's syndrome. For pseudomeningocele identification, the model achieved 76% accuracy, 75% sensitivity, and 83% specificity. AMS scores showed variability in performance, with accuracies ranging from 58% to 71% across categories. These results highlight the model's strength in extracting categorical data but also reveal limitations in its ability to consistently extract granular numerical information, such as AMS scores.
Conclusions
Privacy-focused, locally-run LLMs can accurately and securely extract key clinical data. The model performed well with structured categorical information but showed variability with granular numerical data. These findings highlight both the promise and limitations of AI-driven data extraction in clinical settings. Continued refinement, validation, and oversight will be essential to optimize AI tools for accurate, privacy-preserving data extraction in patient care and research.
|
5:40 PM
|
An Assessment of the Diagnostic Approach in 105 Cases of Neurogenic or Mixed Thoracic Outlet Syndrome
Purpose
The diagnosis and management of neurogenic or mixed thoracic outlet syndrome (TOS) is complex. Various diagnostic tools including several imaging modalities, electromyography (EMG) and anterior scalene muscle blocks (ASMB) can be useful in identifying this difficult pathology. We assessed the diagnostic approach used in 105 patients who underwent thoracic outlet decompression (TOD).
Methods
A single-institution retrospective chart review was performed between January 2017 and September 2024 to identify patients with isolated/combined TOS treated with surgical
decompression. Demographic information, pre and post-operative symptoms, surgical technique, and functional outcomes were collected. Diagnostic data was collected from radiology and procedural reports in addition to clinic notes. Imaging findings collected included the presence of anatomic abnormalities, provocative nerve or vessel compression, nerve swelling or edema, and any documented signal abnormalities (hyperintensity and hypoechogenicity). Descriptive statistics and a Fisher Exact test were used to describe study findings.
Results
The diagnostic approach of 105 patients (54.28% female) was assessed. MRI was the modality that identified a brachial plexus pathology most commonly. Of the 54 patients with diagnostic evidence of nTOS (either imaging, ASMB, or EMG), 29 (53.7%) had a peripheral nerve abnormality identified on imaging, in addition to nTOS. US was the most common modality used to identify a nerve pathology distal to the brachial plexus, and the ulnar nerve was most affected. The most common brachial plexus imaging finding on MRI or US was altered nerve morphology or thickening/edema. Of the patients who had relief after ASMB, 22 (68.8%) had a good response to surgical decompression. There was no significant association found between response to ASMB and imaging evidence of TOS (p = 1.0).
Conclusions
A thorough examination of the entire upper extremity is recommended when assessing a patient for TOS, as comorbid compression of more distal peripheral nerves may be present. Of importance, the ulnar nerve may be more commonly affected. Some patients who do not have imaging evidence of a nerve pathology may respond positively to an ASMB. As such, a multi-pronged diagnostic approach may be best suited for assessing a patient with TOS. Future prospective studies are needed to establish a standard diagnostic approach for this challenging disorder.
|
5:45 PM
|
Disparities in Breast Cancer-Related Lymphedema: A Systematic Review of Inequities and Barriers in Care
Purpose: This systematic review summarizes the evidence for disparities and barriers surrounding breast cancer-related lymphedema (BCRL) care, particularly in diagnosis, education, and accessibility to treatment. This study also addresses the lack of guideline recommendations specific for BCRL in education, referral patterns, and transfer of care, and provides a comprehensive analysis on patient experience and perception relevant to BCRL.
Methods and Materials: The study protocol was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO registration no. CRD42024597105). A search of PubMed/MEDLINE, Embase, Scopus, and Web of Science following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines was conducted from January 1, 1990, through October 3, 2024. Mixed-methods, qualitative, cross-sectional, multiple-case, longitudinal, and randomized controlled trials that reported disparities and barriers surrounding BCRL care were included. Review articles, editorials, commentaries, abstracts, poster papers, translation or validation of patient-reported outcome measures, and non-English articles were excluded. To understand our outcome questions related to disadvantaged patient populations, extracted study data were pooled and reported in a narrative format and tabulated into systematic review tables.
Results: The search yielded 1059 articles, and 39 met inclusion criteria. Themes identified included the following: racial and ethnic disparities; increased risk with younger age, low education level, low income, rural geographic location, and presence of medical comorbidities; inadequate provider and patient knowledge; low patient education; burden and challenges with lifelong self-management; and barriers in receipt of healthcare provider diagnosis or adequate BCRL management. Subthemes included cumulative cost burden, psychosocial barriers, and the role of patient self-efficacy.
Conclusions: Younger non-Caucasian females, residence in rural regions, and those with low income or education levels appeared to be at greatest risk for self-reported (rather than physician-diagnosed) BCRL. Patients of diverse racial and ethnic backgrounds and low socioeconomic status were at increased risk for inadequate self-care practice education and breast cancer survivorship support. Active prevention with multidisciplinary interventions is imperative to lower BCRL rates, empower breast cancer survivors, and strengthen self-efficacy.
|
5:50 PM
|
Scientific Abstract Presentations: Hand Session 2 - Discussion 1
|