Natural language processing and entrustable professional activity text feedback in surgery: A machine learning model of resident autonomy

BackgroundEntrustable Professional Activities (EPAs) contain narrative ‘entrustment roadmaps’ designed to describe specific behaviors associated with different entrustment levels. However, these roadmaps were created using expert committee consensus, with little data available for guidance. Analysis of actual EPA assessment narrative comments using natural language processing may enhance our understanding of resident entrustment in actual practice.MethodsAll text comments associated with EPA microassessments at a single institution were combined. EPA—entrustment level pairs (e.g. Gallbladder Disease—Level 1) were identified as documents. Latent Dirichlet Allocation (LDA), a common machine learning algorithm, was used to identify latent topics in the documents associated with a single EPA. These topics were then reviewed for interpretability by human raters.ResultsOver 18 months, 1015 faculty EPA microassessments were collected from 64 faculty for 80 residents. LDA analysis identified topics that mapped 1:1 to EPA entrustment levels (Gammas >0.99). These LDA topics appeared to trend coherently with entrustment levels (words demonstrating high entrustment were consistently found in high entrustment topics, word demonstrating low entrustment were found in low entrustment topics).ConclusionsLDA is capable of identifying topics relevant to progressive surgical entrustment and autonomy in EPA comments. These topics provide insight into key behaviors that drive different level of resident autonomy and may allow for data-driven revision of EPA entrustment maps.     

Machine learning-based prediction of upgrading on magnetic resonance imaging targeted biopsy in patients eligible for active surveillance

ObjectiveTo examine the ability of machine learning methods to predict upgrading of Gleason score on confirmatory magnetic resonance imaging-guided targeted biopsy (MRI-TB) of the prostate in candidates for active surveillance.Subjects and methodsOur database included 592 patients who received prostate multiparametric magnetic resonance imaging in the evaluation for active surveillance. Upgrading to significant prostate cancer on MRI-TB was defined as upgrading to G 3+4 (definition 1 - DF1) and 4+3 (DF2). Machine learning classifiers were applied on both classification problems DF1 and DF2.ResultsUnivariate analysis showed that older age and the number of positive cores on pre-MRI-TB were positively correlated with upgrading by DF1 (P-value ≤ 0.05). Upgrading by DF2 was positively correlated with age and the number of positive cores and negatively correlated with body mass index. For upgrading prediction, the AdaBoost model was highly predictive of upgrading by DF1 (AUC 0.952), while for prediction of upgrading by DF2, the Random Forest model had a lower but excellent prediction performance (AUC 0.947).ConclusionWe show that machine learning has the potential to be integrated in future diagnostic assessments for patients eligible for AS. Training our models on larger multi-institutional databases is needed to confirm our results and improve the accuracy of these models’ prediction.     

High vs. low entrustment behaviors in the operating room

BackgroundOperative experience with an appropriate degree of supervised autonomy is critical to resident training. Progressively greater intraoperative entrustment has been associated with gradually higher levels of resident autonomy. This study attempts to identify consistently observed intraoperative behaviors that are linked with higher resident entrustment.MethodsThis qualitative study analyzed observational notes recorded by trained raters who provided entrustment scores for 204 surgical cases at Michigan Medicine from 2015 to 2017. Notes were coded in NVivo12. Thematic analysis was used to identify themes and patterns within the data.ResultsThe analysis generated 144 codes. Codes were clustered into 10 themes. These themes manifested differently in intraoperative behaviors strongly associated with high entrustment versus low entrustment.ConclusionThis study demonstrates key differences in intraoperative behaviors exhibited by residents and faculty in high and low entrustment interactions. Awareness of behaviors that enhance entrustment can help faculty augment resident learning and enable higher resident operative autonomy.     

Greater faculty familiarity with residents improves intraoperative entrustment

BackgroundLongitudinal contact between faculty and residents facilitates greater faculty entrustment. The purpose of this study is to assess the relationship between faculty familiarity with residents and faculty entrustment.Materials and methodsResearchers observed and rated entrustment behaviors using OpTrust, September 2015–June 2017 at Michigan Medicine. Faculty familiarity with resident was measured on a 1–4 scale (1 = not familiar, 4 = extremely familiar). ANOVA and Sidak adjusted multiple comparisons were used to assess the relationship between faculty familiarity and faculty entrustment.Results56 faculty and 73 residents were observed across 225 surgical cases. Faculty entrustment scores increased to 2.48 when resident familiarity was reported as “slightly familiar”. Faculty entrustment scores for “moderately familiar” increased to 2.57. Faculty entrustment scores for “extremely familiar” increased to 2.84.ConclusionsWe found a positive relationship between faculty familiarity and entrustment. These findings support greater continuity in faculty/resident relationships. Longitudinal contact allows learners to be granted progressive entrustment.SummaryThis study demonstrates a positive relationship between faculty familiarity with residents and an increase in intraoperative entrustment. These findings support greater continuity in faculty/resident relationships.     

Struggles with autonomy: Exploring the dual identities of surgeons and learners in the operating room

BackgroundDeveloping autonomy is a critical component of becoming an attending surgeon. General surgery training has evolved in recent decades, however, leaving residents less time to work with attendings to establish entrustment. Limited entrustment can impact resident learning and engagement.MethodsA constructivist grounded theory approach was used to guide interviews of 12 general surgery residents and 10 attendings.ResultsEngagement in the OR is perceived by both residents and attendings as fundamental to achieving autonomy. Our study uncovered three key tensions: 1. Residents and attendings both occupy dual roles in the OR; 2. System demands put those roles in tension and opposition constantly; 3. Residents and attendings do deploy strategies to seek balance in those tensions.ConclusionsIn an academic OR setting, competing priorities can negatively impact resident engagement. Participants described some strategies for helping residents and attendings prioritize learning and teaching to better prepare residents for future practice.     

Identification of teaching excellence in operating room and clinic settings

BACKGROUND: A system for obtaining learner feedback on surgical faculty teaching is a program-specific resource for recognizing faculty accomplishments as well as being a requirement of the Accreditation Council for Graduate Medical Education (ACGME). This investigation uses 5 years of feedback from residents to identify surgical teaching behaviors that define teaching excellence. METHODS: Between 1995 and 1999 full-time surgeons in a division of general surgery were evaluated biannually by every resident on their services, using two 10-item Likert scales to assess frequency of performing selected teaching behaviors. Response categories ranged from 0 (does not demonstrate) to 4 (demonstrates the behavior to a very high degree). Mean scores > or =3.7 (1 SD above the mean) were categorized as evidence of superior teaching, whereas mean scores < or =2.4 (1 SD below the mean) were categorized as mediocre. Residents wrote statements identifying teaching strengths. RESULTS: There were 753 individual resident assessments of 16 faculty. The overall mean rating for operating room and clinic teaching was 3.1, with 24% of the ratings > or =3.7 and 14% of the ratings < or =2.4. For operating room, discriminant behaviors were: demonstrates sensitivity to resident learning needs (3.85 versus 1.62, P <0.01) and provides direct feedback (3.60 versus 1.27, P <0.01). Residents' statements yielded themes tied to superior teaching: demonstrates technical expertise, allows resident participation, and maintains a learning climate of respect. CONCLUSIONS: A resident-based teaching assessment system can offer a reasonable and valid form of feedback to academic surgeons. The use of mixed methods to identify teaching behaviors that characterize excellence informs faculty of how they are perceived as educators and provides examples of specific behaviors that merit commendation.     

Timeliness and quality of surgical discharge summaries after the implementation of an electronic format

BackgroundAs electronic discharge summaries (EDS) become more prevalent and health care systems increase their focus on transitions of care, analysis of EDS quality is important. The objective of this study was to assess the timeliness and quality of EDS compared with dictated summaries for surgical patients, which has not previously been evaluated.MethodsA retrospective study was conducted of a sample of discharge summaries from surgical patients at an urban university teaching hospital before and after the implementation of an EDS program. Summaries were evaluated on several dimensions, including time to summary completion, summary length, and summary quality, which was measured on a 13-item scoring tool.ResultsAfter the exclusion of 5 patients who died, 195 discharge summaries were evaluated. Discharge summaries before and after EDS implementation were similar in admission types and discharge destinations of the patients. Compared with dictated summaries, EDS had equivalent overall quality (P = .11), with higher or equivalent scores on all specific quality aspects except readability. There was a highly significant statistical and clinical improvement in timeliness for electronic summaries (P < .01). Obvious use of copying and pasting was identified in 8% of discharge summaries and was associated with decreased readability (P = .02).ConclusionsThe implementation of EDS can improve the timeliness of summary completion without sacrificing quality for surgical patients. Excessive copying and pasting can reduce the readability of discharge summaries, and strategies to discourage this practice without the use of appropriate editing should be used.     

Association between sucrase-isomaltase and p53 expression in colorectal cancer

Background: Sucrase-isomaltase (SI) is a tissue-based phenotypic marker that is an independent prognostic factor in colorectal cancer (CRC). DF3 and galectin 3 are two other tissue-based markers that are upregulated during neoplastic transformation. Because p53 mutations are acquired during neoplastic progression, we reasoned that alterations in SI and p53 may be associated despite an apparent lack of biological interaction. Methods: Paraffin sections from 183 patients who underwent surgery at New England Deaconess Hospital (NEDH) between 1965 and 1977 were analyzed first by immunohistochemistry (IHC) for the expression of the markers SI, DF3, and galectin 3, which were scored as absent or present. Paraffin sections from a second group of 59 patients who underwent surgery at NEDH between 1985 and 1992 were analyzed by IHC for the expression of p53 as well as SI, DF3, and galectin 3. p53 nuclear staining was scored as absent or present. Previous work has shown that p53 is mutated in all cells with nuclear staining and in 10% of tumors that are unstained. Results: SI expression was not associated with the expression of either DF3 or galectin 3, and neither DF3 nor galectin 3 were prognostic factors in CRC. None of the phenotypic markers were associated with any of the clinicopathologic variables. However, 21 of 24 p53-positive cases (88%) expressed SI, whereas 15 of 35 p53-negative cases (43%) were also SI negative (p=0.02, Fisher exact test). p53 expression was not associated with expression of DF3 or galectin 3. Conclusions: SI expression and p53 mutation are associated significantly in CRC. Although the mechanism underlying such an association is presently unknown, the association may define a subset of patients with a worse prognosis.Presented at the 49th Annual Cancer Symposium of The Society of Surgical Oncology, Atlanta, Georgia, March 21–24, 1996.     

Association of intraoperative entrustment with clinical competency amongst general surgery residents

BackgroundLack of transparency and meaningful assessment in surgical residency has led to inconsistent intraoperative entrustment and highly variable trainee competence at graduation. The relationship between faculty entrustment and resident entrustability on clinical competency remains unclear. We sought to evaluate the dynamic between entrustment/entrustability and clinical competency in general surgery residency.MethodsIntraoperative observations were conducted across a 22-month period at an academic tertiary center. Entrustment/entrustability were measured using OpTrust. Clinical competencies were appraised via ACGME Milestones and Objective Structured Assessment of Technical Skill (OSATS) scores. Mixed effects linear regression was used to investigate the relationship among overall ACGME Milestone scores, OSATS domain scores, and overall OpTrust scores.ResultsOverall OpTrust scores significantly correlated with overall Milestone scores and multiple OSATS score domains.ConclusionsOpTrust demonstrated a positive association between ACGME general surgery Milestones and OSATS scores. Overall, OpTrust may help optimize intraoperative faculty entrustment and resident entrustability, facilitating surgical trainee success during residency.     

Possible relation between partial small bowel obstruction and severe postprandial reactive hypoglycemia after Roux-en-Y gastric bypass

BackgroundAlthough dietary treatment ameliorates symptoms in most patients with postbariatric hypoglycemia (PBH), there is a subgroup with severe symptoms that do not respond sufficiently to either diet or drugs. A clinical observation showed that those patients additionally experienced postprandial abdominal discomfort or pain.ObjectivesThis report describes patients with severe PBH following laparoscopic Roux-en Y gastric bypass undergoing corrective surgery to alleviate partial small bowel obstruction (kink, adhesions, dysfunctional anastomosis) and the subsequent outcome regarding symptoms of PBH.SettingSahlgrenska University Hospital, Sweden.MethodsRetrospective analysis regarding hypoglycemic symptoms from medical records and a complementary telephone interview.ResultsOut of 80 patients treated for severe PBH at our tertiary academic surgical unit over the last 4 years, 38 underwent corrective surgery (adhesiolysis and/or a reconstructed jejuno-jejunostomy). Out of 21 patients using medications to control hypoglycemia before corrective surgery, 19 (90.5%) discontinued the medication, and 5 of 19 (26%) no longer needed to use a blood glucometer or a continuous glucose monitor. Although patients after surgical correction still could experience PBH, symptoms were less frequent and less severe. Postprandial abdominal symptoms decreased, and patients reported improved eating quality. Patient interviews revealed that 8% became entirely free of PBH symptoms after surgery and 71% experienced significant relief.ConclusionsWe propose a possible association between severe hypoglycemic symptoms after laparoscopic Roux-en Y gastric bypass and partial small bowel obstruction. Patients with PBH not responding adequately to diet and drug treatment may benefit from assessment and intervention for partial small bowel obstruction.     

Use of error management theory to quantify and characterize residents’ error recovery strategies

BackgroundTraditional checklist metrics for surgical performance can miss key intraoperative decisions that impact procedural outcomes. Error-based assessments may help identify important metrics for evaluating operative performance and resident readiness for independent practice.MethodsThis study utilized human factors error analysis and error management theory to investigate a previously collected video database of resident performance during a simulated laparoscopic ventral hernia (LVH) repair on a table-top simulator using standard laparoscopic tools and mesh. Errors were deconstructed and coded using a structured observation tool and video analysis software. Error detection events and error recovery events were categorized for each operative step of the ventral hernia repair.ResultsResidents made a total of 314 errors (M = 15.7, SD = 4.96). There were more technical errors (63%) than cognitive errors (37%) and more commission errors (69%) than omission errors (30%). Almost half (47%) of all errors went completely undetected by the residents for the entire LVH repair. Of the errors that residents attempted to recover (n = 136), 86.0% were successfully recovered. Technical errors were four times more likely to be successfully recovered than cognitive errors (p = .020).ConclusionsOur results revealed specific details regarding residents’ error management strategies and provides validity evidence for the use of human factors error frameworks in surgical performance assessments. Practice in simulation-based learning environments may improve resident decision-making and error management opportunities by providing a structured experience where errors are explicitly characterized and used for training and feedback. Error management training may play a major role in equipping residents and junior faculty with the skills required for independent, high-quality operative performance.     

Defining entrustable professional activities for first year anesthesiology residents: A Delphi study

IntroductionThe use of entrustable professional activities (EPAs) as a basis for assessment may bridge the gap between the theory of competency-based education and clinical practice. The purpose of this study was to develop and validate EPAs for United States (US) first-year clinical anesthesia (CA-1) residents for anesthesiology residency programs to use as the basis for curriculum development and workplace assessment.MethodsFrom a list of EPAs abstracted from the literature, an expert panel through a modified Delphi consensus process established EPAs for the CA1 curriculum.ResultsThe final list of EPAs after group consensus had 28 EPAs, with 14 (50%) considered to be applicable to the CA-1 year. An 80% consensus rate was used to accept or reject the final list.ConclusionThis study applied a construct validity lens to EPA development providing assurance that the EPAs adopted are appropriate for use in workplace-based assessment and entrustment decision-making.     

Development of machine learning and natural language processing algorithms for preoperative prediction and automated identification of intraoperative vascular injury in anterior lumbar spine surgery

BACKGROUNDIntraoperative vascular injury (VI) may be an unavoidable complication of anterior lumbar spine surgery; however, vascular injury has implications for quality and safety reporting as this intraoperative complication may result in serious bleeding, thrombosis, and postoperative stricture.PURPOSEThe purpose of this study was to (1) develop machine learning algorithms for preoperative prediction of VI and (2) develop natural language processing (NLP) algorithms for automated surveillance of intraoperative VI from free-text operative notes.PATIENT SAMPLEAdult patients, 18 years or age or older, undergoing anterior lumbar spine surgery at two academic and three community medical centers were included in this analysis.OUTCOME MEASURESThe primary outcome was unintended VI during anterior lumbar spine surgery.METHODSManual review of free-text operative notes was used to identify patients who had unintended VI. The available population was split into training and testing cohorts. Five machine learning algorithms were developed for preoperative prediction of VI. An NLP algorithm was trained for automated detection of intraoperative VI from free-text operative notes. Performance of the NLP algorithm was compared to current procedural terminology and international classification of diseases codes.RESULTSIn all, 1035 patients underwent anterior lumbar spine surgery and the rate of intraoperative VI was 7.2% (n=75). Variables used for preoperative prediction of VI were age, male sex, body mass index, diabetes, L4-L5 exposure, and surgery for infection (discitis, osteomyelitis). The best performing machine learning algorithm achieved c-statistic of 0.73 for preoperative prediction of VI (https://sorg-apps.shinyapps.io/lumbar_vascular_injury/). For automated detection of intraoperative VI from free-text notes, the NLP algorithm achieved c-statistic of 0.92. The NLP algorithm identified 18 of the 21 patients (sensitivity 0.86) who had a VI whereas current procedural terminologyand international classification of diseases codes identified 6 of the 21 (sensitivity 0.29) patients. At this threshold, the NLP algorithm had a specificity of 0.93, negative predictive value of 0.99, positive predictive value of 0.51, and F1-score of 0.64.CONCLUSIONRelying on administrative procedural and diagnosis codes may underestimate the rate of unintended intraoperative VI in anterior lumbar spine surgery. External and prospective validation of the algorithms presented here may improve quality and safety reporting.     

Benchmarking the strategies for assessing clinical reasoning in osteopathic curricula

BackgroundBenchmarking between educational institutions is a vital component of quality assurance and contributes to greater consistency and quality in teaching and learning practices. The investigation of clinical reasoning in osteopathy has only recently begun to be explored in depth. Benchmarking builds confidence in our education practices and processes and demonstrates a maturing of the osteopathic academic profession.ObjectiveThe aim of this project was to benchmark the assessment strategy used for clinical reasoning across the final two years of the clinical components of four osteopathic programs.MethodsLearning objectives and clinical assessments from the final two years in each of the four programs were analysed to identify the types and frequency of assessments and the degree of alignment between learning objectives and Bloom's taxonomy and Miller's heirarchy.ParticipantsRepresentatives from Southern Cross University, Australia, Victoria University, Australia, Unitec, New Zealand and the British School of Osteopathy, UK.ResultsAll institutions assess clinical reasoning in a variety of ways such as the assessment of student's actual performance during real-time, in-situ clinical consultations; the assessment of simulated performance; the clinical supervisors' report and; oral or written reports - on simulated case study. The results show that the osteopathy teaching institutions in the present study do not scaffold the expected learning objectives to reflect an increase in difficulty as the student's progress; the learning objectives tend to be clustered and relatively stable. However, this may be a reflection of only investigating the final years of an osteopathy teaching program. This opens the field for future research.ConclusionsIt would be worthwhile if future studies benchmarked the criteria used in clinical assessments and made explicit the key professional values related to assessing clinical competencies in line with the Core Competencies outlined in the World Health Organization's Benchmarks for Training in Osteopathy.     

Increased flexor hallucis longus tension decreases ankle dorsiflexion

BackgroundRestricted excursion of the flexor hallucis longus (FHL) is associated with several clinical problems. An FHL excursion measurement device (EMD) was used to objectively assess differences between patients with clinically normal or tight FHL tendons.Methods188 patients (356 feet) were enrolled. The EMD measured maximum ankle dorsiflexion with the great toe in 15°, 30°, and 45° of dorsiflexion. All had clinical assessment of FHL tightness by their provider independently of the EMD measurement.ResultsIncreased hallux DF always caused decreased ankle DF. Patients with clinically tight FHLs demonstrated decreased ankle DF compared to normal subjects at all hallux positions (p < 0.01). The EMD measurement was not sensitive enough for detection of FHL tightness in individuals. A clinically tight FHL was seen in almost 50% of feet.ConclusionsTension in the FHL can limit ankle DF. Clinical tightness of the FHL is likely more common than currently recognized.     

Educational feedback in the operating room: a gap between resident and faculty perceptions

BackgroundImmediate feedback regarding performance in the operating room remains a key component of resident education. The aim of this study was to assess resident and faculty perceptions regarding postoperative feedback.MethodsAnonymous surveys were distributed to residents and faculty members. Questions addressed the timing, amount, and specificity of feedback; satisfaction; and the definition and importance of feedback. Additional questions regarded the importance and frequency of feedback in 7 specific areas of surgical competency.ResultsResident satisfaction with timing, amount, and specificity of feedback was significantly lower than faculty satisfaction. Perceptions of the importance of feedback for each of the 7 specific areas did not differ. Faculty members' perceptions on the frequency of feedback were higher than residents' perception in all competencies of feedback (5-point scale, all P values = .001).ConclusionsThere are significant differences between resident and faculty perceptions regarding postoperative feedback. Although faculty members believed they delivered appropriate amounts of timely, quality feedback, this perception was not shared by residents.     

Five-year attrition,active enrollment,and predictors of level of participation in the Longitudinal Assessment of Bariatric Surgery (LABS-2) study

BackgroundReporting high-quality bariatric surgery outcomes depends on participant attrition and level of study participation among enrolled participants.ObjectivesOur aims are to report participant attrition, active enrollment, and level of participation, and to evaluate pre-surgery sociodemographic, physical health, and psychosocial factors as predictors of attrition and level of participation through 5 years.SettingThe Longitudinal Assessment of Bariatric Surgery-2 study which enrolled 2458 adults undergoing a first bariatric surgical procedure at 1 of 6 US cites from 2006 through 2009.MethodsIn-person research assessments were conducted pre-surgery and annually for five years. Extensive retention strategies including offering remote assessments (telephone, email, mail, or a combination) were fully implemented in 2009. Among living participants, including those inactivated, annual follow-up assessments were categorized as in-person, remote or missed through 5 years.ResultsBy year 5, 1.7% of participants had died and 3.2% had withdrawn or were inactivated by the study staff; thus, attrition was 4.9% (n = 121). Controlling for site and calendar year, missed assessments increased from 14.7%–21.8% between years 1 and 2 and then stayed relatively stable (20.8%–19.6%) for years 3–5. Younger age, male sex, White race, lower body mass index, smoking, illicit drug use, and higher weight loss expectations preoperatively were independently associated with a higher likelihood of a missed versus in-person assessment across follow-up.ConclusionThe LABS-2 participant attrition was low. The percentage of missed assessments did not increase after year 2, perhaps due to implementation of a comprehensive retention plan. Predictors of missed assessments highlight subgroups to target for focused retention efforts.     

A clinical and radiographic 13-year follow-up study of 138 Charnley hip arthroplasties in patients 50–70 years old: Comparison of university hospital data and registry data

Background and purpose?Arthroplasty registers provide rates of implant survival in large populations based on implant revision. In an unrevised prosthesis population, some patients may have implants with clinically poor outcome or radiographic failure. We therefore evaluated medium-term clinical and radiographic results in patients with Charnley hip arthroplasties and compared our results with data from the Norwegian Arthroplasty Register (NAR).

Patients and methods?From 1989 through 1991, 138 Charnley arthroplasties with plain Palacos cement were performed in 123 patients who were 50–70 years old. At follow-up after 13 (12–15) years, 26 patients had died (28 hips). The 84 unrevised patients (93 hips) were interviewed and underwent clinical and radiographic assessment. Prosthesis survival was estimated by the Kaplan- Meier method.

Results?At follow-up, 83% of the patients were completely satisfied with their hip replacement. Mean Harris hip score (HHS) was 83 (SD 15), mean EQ-5D index was 0.75 (SD 0.24) and mean EQ-VAS score was 69 (SD 21). Most clinical assessments had poorer scores for Charnley category C (n = 47) than for Charnley category A + B (n = 46). Function, according to Charnley's modified Merle d'Aubigné and Postel scoring system, was improved compared to preoperative values. The survival at 10 years was 89% (95% CI: 84–95) and at 13 years it was 85% (95% CI: 79–92) with revision for any reason as endpoint. In addition to 20 revised hips, 8 implants were radiographically loose and 13 hips had HHS < 70, giving a clinical success rate of 76%. Only 4 primary operations (0.8%) had not been reported to the NAR, but all revisions had been reported.

Interpretation?Clinical follow-up studies give essential information that is additional to that gained from revision-based outcome studies. To fully appreciate the clinical effectiveness of an implant, specific hip function, patient satisfaction, quality of life, and radiographic analysis must also be considered. The functional status of the patient has an important influence on the clinical outcome after hip replacement.     

Desmoplastic fibroma of the spine: a series of 12 cases and outcomes

Background contextDesmoplastic fibroma (DF) is a benign, yet locally aggressive, tumor of the connective tissue. Desmoplastic fibroma in the spine is extremely rare, and only a few cases have been reported. Although surgical resection of DF arising in the spine is commonly regarded as a recommended treatment, it is difficult to achieve satisfactory results.PurposeThis study reviews the clinical patterns and follow-up data of patients with DF in the spine who underwent surgical treatment. We attempted to correlate surgical treatment and outcomes over time.Study designA retrospective clinical study of the surgical managements, including subtotal resection, total spondylectomy, and en bloc resection, for DF in the spine. Desmoplastic fibroma of the spine treatment occurred from 2004 to 2009 at the Department of Bone Tumor Surgery, AA Hospital.Patient sampleTwelve consecutive cases of DF of the spine underwent surgical treatment at our center between 2004 and 2009.Outcome measuresNeurologic outcomes were evaluated using Frankel score system and recurrence and metastasis were evaluated by computed tomography or magnetic resonance imaging of the surgical segments involved. Imaging was performed 3, 6, and 12 months after surgery, every 6 months for the next 2 years, and then annually for life.MethodsOverall, two different surgery protocols were applied. One protocol involved subtotal resection followed by radiotherapy (n=4), whereas the other involved total tumor resection (n=8). Postoperative radiotherapy was administered in six cases. Clinical data and surgery efficacy were analyzed via chart review.ResultsEleven patients were disease-free during their follow-up period, whereas one patient experienced recurrence without metastasis. Radicular pain nearly disappeared, and patients suffering from spinal cord compression recovered well. Local recurrence was detected in one-fourth (25%) of the cases that underwent subtotal resection and was not detected in any of the cases involving total spondylectomy.ConclusionsLocal recurrence of DF is not uncommon after insufficient removal. Therefore, total excision, while also preserving neural function, is recommended. In our study, patients who underwent a total spondylectomy had significantly lower local recurrence rates for DF in the spine. Radiotherapy may be an acceptable alternative therapy, whereas en bloc resection has the potential to result in significant functional impairment.