McCutcheon, Brandon A.

OBJECTIVENonhome discharge and unplanned readmissions represent important cost drivers following spinal fusion. The authors sought to utilize different machine learning algorithms to predict discharge to rehabilitation and unplanned readmissions in patients receiving spinal fusion.METHODSThe authors queried the 2012–2013 American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) for patients undergoing cervical or lumbar spinal fusion. Outcomes assessed included discharge to nonhome facility and unplanned readmissions within 30 days after surgery. A total of 7 machine learning algorithms were evaluated. Predictive hierarchical clustering of procedure codes was used to increase model performance. Model performance was evaluated using overall accuracy and area under the receiver operating characteristic curve (AUC), as well as sensitivity, specificity, and positive and negative predictive values. These performance metrics were computed for both the imputed and unimputed (missing values dropped) datasets.RESULTSA total of 59,145 spinal fusion cases were analyzed. The incidence rates of discharge to nonhome facility and 30-day unplanned readmission were 12.6% and 4.5%, respectively. All classification algorithms showed excellent discrimination (AUC > 0.80, range 0.85–0.87) for predicting nonhome discharge. The generalized linear model showed comparable performance to other machine learning algorithms. By comparison, all models showed poorer predictive performance for unplanned readmission, with AUC ranging between 0.63 and 0.66. Better predictive performance was noted with models using imputed data.CONCLUSIONSIn an analysis of patients undergoing spinal fusion, multiple machine learning algorithms were found to reliably predict nonhome discharge with modest performance noted for unplanned readmissions. These results provide early evidence regarding the feasibility of modern machine learning classifiers in predicting these outcomes and serve as possible clinical decision support tools to facilitate shared decision making.

Study Design: Retrospective database analysis. Objective: To examine the impact of training pathway, either neurosurgical or orthopedic, on complications, readmissions, and revisions in spine surgery. Summary of Background Data: Training pathway has been shown to have an impact on outcomes in various surgical subspecialties. Although training pathway has not been shown to have a significant impact on spine surgery outcomes in the perioperative period, long-term results are unknown. Materials and Methods: A retrospective analysis of 197,682 patients receiving 1 of 3 common spine surgeries [lumbar laminectomy, lumbar fusion, and anterior cervical discectomy and fusion (ACDF)] between 2006 and 2010 was conducted. Patient data were obtained from a large claims database. Postoperative adverse effects, all-cause readmission, revision surgery rates, and intermediary payments in these cohorts of patients were compared between spine surgeons with either neurosurgical or orthopedic backgrounds. Results: Patient demographics, hospital-stay characteristics, and medical comorbidities were similar between neurosurgeons and orthopedic surgeons. The risks of surgical complications, all-cause readmission, and revision surgery were also similar between neurosurgeons and orthopedic surgeons across all procedure types assessed, with several minor exceptions: neurosurgeons had marginally higher odds of any complication for lumbar fusions [odds ratio (OR) 1.14; 95% confidence interval (CI), 1.09–1.20] and ACDFs (OR, 1.09; 95% CI, 1.04–1.15). Neurosurgeons also had slightly higher rates of revision surgery for concurrent lumbar laminectomy with fusion (OR, 1.14; 95% CI, 1.08–1.22), and ACDFs (OR, 1.20; 95% CI, 1.14–1.28). No associations between surgeon type and any particular complication were consistently observed for all procedure groups. There were also no associations between surgeon type and 30-day all-cause readmission. Median total intermediary payments were somewhat higher for neurosurgery patients for all procedure groups assessed. Conclusions: Few significant associations between surgeon type and patient outcomes exist in the context of spine surgery. Those which do are small and unlikely to be clinically meaningful. Level of Evidence: Level 3.

Abstract BACKGROUND Stereotactic radiosurgery (SRS) is a commonly performed procedure for patients with intracranial meningiomas. OBJECTIVE To describe the clinical features of patients with radiation-induced cavernous malformations (RICM) after single-fraction meningioma SRS. METHODS Retrospective study of patients having single-fraction SRS for intracranial meningioma at our center from 1990 through 2009, and 1 patient who had single-fraction SRS elsewhere. Patients were excluded if they refused research authorization (n = 7), had a World Health Organization Grade II or III meningioma (n = 65), had a genetic predisposition for tumor development (n = 52), had prior or concurrent radiation therapy (n = 49), or had less than 2 yr of magnetic resonance imaging follow-up after SRS (n = 77). The median follow-up of the remaining 426 patients was 7.9 yr (range, 2-24.9). RESULTS Three RICM (0.7%) were identified at 2, 10, and 21 yr after SRS. Two patients were asymptomatic, whereas 1 patient had a brainstem hemorrhage causing facial weakness and numbness. The risk of developing an RICM after SRS was 0.2% at 5 yr and 0.9% at 15 yr. All patients were observed and remained stable without additional bleeding in follow-up of 7, 12.8, and 2 yr, respectively. A fourth patient developed progressive neurological dysfunction starting 7 yr after SRS at another center and was treated for several years with bevacizumab without improvement. Surgical resection was performed 11.5 yr after SRS and histologic examination was consistent with an RICM. CONCLUSION The risk of RICM after single-fraction SRS for intracranial meningiomas is very low, but the latency period noted until their detection emphasizes the need for extended imaging follow-up after SRS of benign lesions.

OBJECTIVEIn this analysis the authors compare the characteristics of smokers to nonsmokers using demographic, socioeconomic, and comorbidity variables. They also investigate which of these characteristics are most strongly associated with smoking status. Finally, the authors investigate whether the association between known patient risk factors and disability outcome is differentially modified by patient smoking status for those who have undergone surgery for lumbar degeneration.METHODSA total of 7547 patients undergoing degenerative lumbar surgery were entered into a prospective multicenter registry (Quality Outcomes Database [QOD]). A retrospective analysis of the prospectively collected data was conducted. Patients were dichotomized as smokers (current smokers) and nonsmokers. Multivariable logistic regression analysis fitted for patient smoking status and subsequent measurement of variable importance was performed to identify the strongest patient characteristics associated with smoking status. Multivariable linear regression models fitted for 12-month Oswestry Disability Index (ODI) scores in subsets of smokers and nonsmokers was performed to investigate whether differential effects of risk factors by smoking status might be present.RESULTSIn total, 18% (n = 1365) of patients were smokers and 82% (n = 6182) were nonsmokers. In a multivariable logistic regression analysis, the factors significantly associated with patients’ smoking status were sex (p < 0.0001), age (p < 0.0001), body mass index (p < 0.0001), educational status (p < 0.0001), insurance status (p < 0.001), and employment/occupation (p = 0.0024). Patients with diabetes had lowers odds of being a smoker (p = 0.0008), while patients with coronary artery disease had greater odds of being a smoker (p = 0.044). Patients’ propensity for smoking was also significantly associated with higher American Society of Anesthesiologists (ASA) class (p < 0.0001), anterior-alone surgical approach (p = 0.018), greater number of levels (p = 0.0246), decompression only (p = 0.0001), and higher baseline ODI score (p < 0.0001). In a multivariable proportional odds logistic regression model, the adjusted odds ratio of risk factors and direction of improvement in 12-month ODI scores remained similar between the subsets of smokers and nonsmokers.CONCLUSIONSUsing a large, national, multiinstitutional registry, the authors described the profile of patients who undergo lumbar spine surgery and its association with their smoking status. Compared with nonsmokers, smokers were younger, male, nondiabetic, nonobese patients presenting with leg pain more so than back pain, with higher ASA classes, higher disability, less education, more likely to be unemployed, and with Medicaid/uninsured insurance status. Smoking status did not affect the association between these risk factors and 12-month ODI outcome, suggesting that interventions for modifiable risk factors are equally efficacious between smokers and nonsmokers.

OBJECTIVEAggressive sacral tumors often require en bloc resection and lumbopelvic reconstruction. Instrumentation failure and pseudarthrosis remain a clinical concern to be addressed. The objective in this study was to compare the biomechanical stability of 3 distinct techniques for sacral reconstruction in vitro.METHODSIn a human cadaveric model study, 8 intact human lumbopelvic specimens (L2–pelvis) were tested for flexion-extension range of motion (ROM), lateral bending, and axial rotation with a custom-designed 6-df spine simulator as well as axial compression stiffness with the MTS 858 Bionix Test System. Biomechanical testing followed this sequence: 1) intact spine; 2) sacrectomy (no testing); 3) Model 1 (L3–5 transpedicular instrumentation plus spinal rods anchored to iliac screws); 4) Model 2 (addition of transiliac rod); and 5) Model 3 (removal of transiliac rod; addition of 2 spinal rods and 2 S-2 screws). Range of motion was measured at L4–5, L5–S1/cross-link, L5–right ilium, and L5–left ilium.RESULTSFlexion-extension ROM of the intact specimen at L4–5 (6.34° ± 2.57°) was significantly greater than in Model 1 (1.54° ± 0.94°), Model 2 (1.51° ± 1.01°), and Model 3 (0.72° ± 0.62°) (p < 0.001). Flexion-extension at both the L5–right ilium (2.95° ± 1.27°) and the L5–left ilium (2.87° ± 1.40°) for Model 3 was significantly less than the other 3 cohorts at the same level (p = 0.005 and p = 0.012, respectively). Compared with the intact condition, all 3 reconstruction groups statistically significantly decreased lateral bending ROM at all measured points. Axial rotation ROM at L4–5 for Model 1 (2.01° ± 1.39°), Model 2 (2.00° ± 1.52°), and Model 3 (1.15° ± 0.80°) was significantly lower than the intact condition (5.02° ± 2.90°) (p < 0.001). Moreover, axial rotation for the intact condition and Model 3 at L5–right ilium (2.64° ± 1.36° and 2.93° ± 1.68°, respectively) and L5–left ilium (2.58° ± 1.43° and 2.93° ± 1.71°, respectively) was significantly lower than for Model 1 and Model 2 at L5–right ilium (5.14° ± 2.48° and 4.95° ± 2.45°, respectively) (p = 0.036) and L5–left ilium (5.19° ± 2.34° and 4.99° ± 2.31°) (p = 0.022). Last, results of the axial compression testing at all measured points were not statistically different among reconstructions.CONCLUSIONSThe addition of a transverse bar in Model 2 offered no biomechanical advantage. Although the implementation of 4 iliac screws and 4 rods conferred a definitive kinematic advantage in Model 3, that model was associated with significantly restricted lumbopelvic ROM.

OBJECTIVEThe subspecialization of neurosurgical practice is an ongoing trend in modern neurosurgery. However, it remains unclear whether the degree of surgeon specialization is associated with improved patient outcomes. The authors hypothesized that a trend toward increased neurosurgeon specialization was associated with improved patient morbidity and mortality rates.METHODSThe Nationwide Inpatient Sample (NIS) was used (1998–2009). Patients were included in a spinal analysis cohort for instrumented spine surgery involving the cervical spine (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 81.31–81.33, 81.01–81.03, 84.61–84.62, and 84.66) or lumbar spine (codes 81.04–81.08, 81.34–81.38, 84.64–84.65, and 84.68). A cranial analysis cohort consisted of patients receiving a parenchymal excision or lobectomy operation (codes 01.53 and 01.59). Surgeon specialization was measured using unique surgeon identifiers in the NIS and defined as the proportion of a surgeon’s total practice dedicated to cranial or spinal cases.RESULTSA total of 46,029 and 231,875 patients were identified in the cranial and spinal analysis cohorts, respectively. On multivariate analysis in the cranial analysis cohort (after controlling for overall surgeon volume, patient demographic data/comorbidities, hospital characteristics, and admitting source), each percentage-point increase in a surgeon’s cranial specialization (that is, the proportion of cranial cases) was associated with a 0.0060 reduction in the log odds of patient mortality (95% CI 0.0034–0.0086) and a 0.0042 reduction in the log odds of morbidity (95% CI 0.0032–0.0052). This resulted in a 15% difference in the predicted probability of mortality for neurosurgeons at the 75th versus the 25th percentile of cranial specialization. In the spinal analysis cohort, each percentage-point increase in a surgeon’s spinal specialization was associated with a 0.0122 reduction in the log odds of mortality (95% CI 0.0074–0.0170) and a 0.0058 reduction in the log odds of morbidity (95% CI 0.0049–0.0067). This resulted in a 26.8% difference in the predicted probability of mortality for neurosurgeons at the 75th versus the 25th percentile of spinal specialization.CONCLUSIONSFor both spinal and cranial surgery patient cohorts derived from the NIS database, increased surgeon specialization was significantly and independently associated with improved mortality and morbidity rates, even after controlling for overall case volume.

IN BRIEF Diabetes has been associated with the incidence of back pain. However, the relationship between markers of diabetes progression and back pain has not been studied. The objective of this study was to correlate clinical and laboratory measures of diabetes disease severity to the presence of back pain to provide insight into the relationship between these conditions. Findings showed that markers of diabetes disease progression were associated with the presence of back pain, suggesting that uncontrolled diabetes may contribute to the development of chronic back pain.

OBJECTIVETemporal lobectomy is a well-established treatment modality for the management of medically refractory epilepsy in appropriately selected patients. The aim of this study was to assess 30-day morbidity and mortality after temporal lobectomy in cases registered in a national database.METHODSA retrospective cohort analysis was conducted using a multiinstitutional surgical registry compiled between 2006 and 2014. The authors identified patients who underwent anterior temporal lobectomy and/or amygdalohippocampectomy for a primary diagnosis of intractable epilepsy. Univariate and multivariable analyses with regard to patient demographics, comorbidities, operative characteristics, and 30-day outcomes were applied.RESULTSA total of 216 patients were included in the study. The median age was 38 years and 46% of patients were male. The median length of stay was 3 days and the 30-day mortality rate was 1.4%. Fourteen patients (6.5%) developed at least one major complication. Return to the operating room was observed in 7 patients (3.2%). Readmission within 30 days and discharge to a location other than home were available for 2011–2014 (n = 155) and occurred in 11% and 10.3% of patients, respectively. Multivariable regression analysis revealed that increasing age was an independent predictor of discharge disposition other than home and that male sex was a significant risk factor for the development of a major complication. Interestingly, the presence of the attending neurosurgeon and a resident during the procedure was significantly associated with decreased odds of prolonged length of stay (i.e., > 75th percentile [5 days]) and discharge to a location other than home.CONCLUSIONSUsing a multiinstitutional surgical registry, 30-day outcome data after temporal lobectomy for medically intractable epilepsy demonstrates a mortality rate of 1.4%, a major complication rate of 6.5%, and a readmission rate of 11%. Temporal lobectomy is an extremely effective therapy for seizures originating there—however, surgical intervention must be weighed against its morbidity and mortality outcomes.

Abstract BACKGROUND Cerebrospinal fluid (CSF) leak is a well-recognized complication after surgical resection of vestibular schwannomas and is associated with a number of secondary complications, including readmission and meningitis. OBJECTIVE To identify risk factors for and timing of 30-d readmission with CSF leak. METHODS Patients who had undergone surgical resection of a vestibular schwannoma from 1995 to 2010 were identified in the California Office of Statewide Health Planning and Development database. The most common admission diagnoses were identified by International Classification of Disease, ninth Revision, diagnosis codes, and predictors of readmission with CSF leak were determined using logistic regression. RESULTS A total of 6820 patients were identified. CSF leak, though a relatively uncommon cause of admission after discharge (3.52% of all patients), was implicated in nearly half of 490 readmissions (48.98%). Significant independent predictors of readmission with CSF leak were male sex (odds ratio [OR] 1.72, 95% confidence interval [CI] 1.32-2.25), first admission at a teaching hospital (OR 3.32, 95% CI 1.06-10.39), CSF leak during first admission (OR 1.84, 95% CI 1.33-2.55), obesity during first admission (OR 2.10, 95% CI 1.20-3.66), and case volume of first admission hospital (OR of log case volume 0.82, 95% CI 0.70-0.95). Median time to readmission was 6 d from hospital discharge. CONCLUSION This study has quantified CSF leak as an important contributor to nearly half of all readmissions following vestibular schwannoma surgery. We propose that surgeons should focus on technical factors that may reduce CSF leakage and take advantage of potential screening strategies for the detection of CSF leakage prior to first admission discharge.

Abstract BACKGROUND Patients recovering from decompressive laminectomy without fusion may require assistance with activities of daily living and physical/occupational therapy upon hospital discharge. OBJECTIVE To examine comorbidities and perioperative characteristics of patients undergoing lumbar decompression for associations with discharge status using a multicenter database. METHODS A multicenter database was used for this retrospective cohort analysis. Patients admitted from home with degenerative spine disease for lumbar decompression without fusion were included. Thirty-day outcomes and operative characteristics were compared as a function of patient discharge using chi-square and Wilcoxon Rank Sum tests. Multivariable logistic regression was used to determine factors associated with discharge to a nonhome facility. RESULTS Of the 8627 patients included for analysis, 9.7% were discharged to a nonhome facility. On multivariable analysis, age (85+ vs <65, odds ratio [OR] 13.59), number of levels of decompression (3+ vs 1, OR 1.75), African American race vs Non-Hispanic or Hispanic White (OR 1.87), female vs male gender (OR 1.97), body mass index (BMI) (40+ vs 18.5-24.9, OR 1.74), American Society of Anesthesiologists physical classification status (4 vs 1 or 2, OR 2.35), hypertension (OR 1.29), dependent functional status (OR 3.92), diabetes (OR 1.47), smoking (OR 1.40), hematocrit (<35 vs 35+, OR 1.76), international normalized ratio (≥1.3 vs <1.3, OR 2.32), and operative time (3+ h vs <1 h, OR 5.34) were significantly associated with an increased odds of discharge to nonhome facilities. CONCLUSION Preoperative status and operative course variables can influence discharge disposition in lumbar decompression patients. Identifying specific factors that contribute to a greater likelihood of dismissal to skilled facility or rehabilitation unit can further inform both surgeons and patients during preoperative counseling and disposition planning.