Using machine learning and big data for the prediction of venous thromboembolic events after spine surgery: A single-center retrospective analysis of multiple models on a cohort of 6869 patients.

Objective: Venous thromboembolic event (VTE) after spine surgery is a rare but potentially devastating complication. With the advent of machine learning, an opportunity exists for more accurate prediction of such events to aid in prevention and treatment. Methods: Seven models were screened using 108 database variables and 62 preoperative variables. These models included deep neural network (DNN), DNN with synthetic minority oversampling technique (SMOTE), logistic regression, ridge regression, lasso regression, simple linear regression, and gradient boosting classifier. Relevant metrics were compared between each model. The top four models were selected based on area under the receiver operator curve; these models included DNN with SMOTE, linear regression, lasso regression, and ridge regression. Separate random sampling of each model was performed 1000 additional independent times using a randomly generated training/testing distribution. Variable weights and magnitudes were analyzed after sampling. Results: Using all patient-related variables, DNN using SMOTE was the top-performing model in predicting postoperative VTE after spinal surgery (area under the curve [AUC] =0.904), followed by lasso regression (AUC = 0.894), ridge regression (AUC = 0.873), and linear regression (AUC = 0.864). When analyzing a subset of only preoperative variables, the top-performing models were lasso regression (AUC = 0.865) and DNN with SMOTE (AUC = 0.864), both of which outperform any currently published models. Main model contributions relied heavily on variables associated with history of thromboembolic events, length of surgical/anesthetic time, and use of postoperative chemoprophylaxis. Conclusions: The current study provides promise toward machine learning methods geared toward predicting postoperative complications after spine surgery. Further study is needed in order to best quantify and model real-world risk for such events.

Perioperative Factors Associated With Chronic Opioid Use After Spine Surgery.

STUDY DESIGN: Retrospective case control. OBJECTIVES: The purpose of the current study is to determine risk factors associated with chronic opioid use after spine surgery. METHODS: In our single institution retrospective study, 1,299 patients undergoing elective spine surgery at a tertiary academic medical center between January 2010 and August 2017 were enrolled into a prospectively collected registry. Patients were dichotomized based on renewal of, or active opioid prescription at 3-mo and 12-mo postoperatively. The primary outcome measures were risk factors for opioid renewal 3-months and 12-months postoperatively. These primarily included demographic characteristics, operative variables, and in-hospital opioid consumption via morphine milligram equivalence (MME). At the 3-month and 12-month periods, we analyzed the aforementioned covariates with multivariate followed by bivariate regression analyses. RESULTS: Multivariate and bivariate analyses revealed that script renewal at 3 months was associated with black race (P = 0.001), preoperative narcotic (P < 0.001) or anxiety/depression medication use (P = 0.002), and intraoperative long lumbar (P < 0.001) or thoracic spine surgery (P < 0.001). Lower patient income was also a risk factor for script renewal (P = 0.01). Script renewal at 12 months was associated with younger age (P = 0.006), preoperative narcotics use (P = 0.001), and ≥4 levels of lumbar fusion (P < 0.001). Renewals at 3-mo and 12-mo had no association with MME given during the hospital stay or with the usage of PCA (P > 0.05). CONCLUSION: The current study describes multiple patient-level factors associated with chronic opioid use. Notably, no metric of perioperative opioid utilization was directly associated with chronic opioid use after multivariate analysis.

Atlas fractures with and without simultaneous dens fractures differ with respect to clinical, demographic, and management characteristics.

Background: Patients with simultaneous fractures of the atlas and dens have traditionally been managed according to the dens fracture's morphology, but data supporting this practice are limited. Methods: We retrospectively examined all patients with traumatic atlas fractures at our institution between 2008 and 2016. We used multivariable regression and propensity score matching to compare the presentation, management, and outcomes of patients with isolated atlas fractures to patients with simultaneous atlas-dens fractures. Results: Ninety-nine patients were identified. Patients with isolated atlas fractures were younger (61 ± 22 vs. 77 ± 14, P = 0.0003), had lower median Charlson Comorbidity Index (3 vs. 5, P = 0.0005), had better presenting Nurick myelopathy scores (0 vs. 3, P < 0.0001), and had different mechanisms of injury (P = 0.0011). Multivariable regression showed that having a simultaneous atlas-dens fracture was independently associated with older age (odds ratio [OR] =1.59 [1.22, 2.07], P = 0.001), worse presenting myelopathy (OR = 3.10 [2.04, 4.16], P < 0.001), and selection for surgery (OR = 4.91 [1.10, 21.97], P = 0.037). Propensity score matching yielded balanced populations (Rubin's B = 23.3, Rubin's R = 1.96) and showed that the risk of atlas fracture nonunion was no different among isolated atlas fractures compared to simultaneous atlas-dens fractures (P = 0.304). Age was the only variable independently associated with atlas fracture nonunion (OR = 2.39 [1.15, 5.00], P = 0.020), having a simultaneous atlas-dens fracture was not significant (P = 0.2829). Conclusions: Among patients with atlas fractures, simultaneous fractures of the dens occur in older patients and confer an increased risk of myelopathy and requiring surgical stabilization. Controlling for confounders, the risk of atlas fracture nonunion is equivalent for isolated atlas fractures versus simultaneous atlas-dens fractures.

Frailty predicts readmission, reoperation, and infection after posterior spinal fusion: An institutional series of 3965 patients.

OBJECTIVE: Frailty is a measure of physiologic vulnerability conceptualized as the accumulation of deficits with aging, and may be useful for predicting risk of adverse events following posterior spinal fusion. Our objective was to investigate the utility of the Canadian Study on Health and Aging (CHSA) Modified Frailty Index (mFI) in patients undergoing posterior spinal fusion (PSF) as a predictor of several surgical quality metrics including readmission, reoperation, and surgical site infection. METHODS: We examined 3965 consecutive PSF patients treated at our institution between 2000 and 2015, and collected demographic, clinical, and frailty and comorbid disease burden measures using the mFI and Charlson Comorbidity Index (CCI). We examined trends and changes in these clinical and demographic characteristics over the course of the study period. We performed multivariable regression to identify independent predictors of readmission, reoperation, and surgical site infection. RESULTS: Over the course of the study period, the mean patient age increased linearly year-over-year (ß=0.60 [0.48, 0.72], p < 0.0001, R=0.94), while the SSI rate decreased linearly (ß=-0.14 [-0.27, -0.02], p = 0.0249, R=0.56), and frailty scores did not change significantly (p = 0.8124, R=0.065). Among all patients undergoing PSF, postoperative wound infection was independently associated with number of levels fused (OR=1.104 p < 0.001), frailty as measured by mFI (OR=1.150 p = 0.006), and BMI (OR=1.041 p = 0.008). Frailty was also independently associated with postoperative ICU admission (OR=1.1080 p = 0.005), 30-day readmission (OR=1.181 p < 0.001), and 30-day reoperation (OR=1.128 p < 0.001). Among all patients, rate of postoperative wound infection increased with increasing frailty (p = 0.0002) and increasing comorbid disease burden (chi-square p = 0.0012). CONCLUSION: The mFI predicts adverse events among patients undergoing PSF, including readmission, reoperation, and surgical site infection. When controlling for frailty, age was not an independent predictor of adverse events.

Percutaneous screw fixation versus open fusion for the treatment of traumatic thoracolumbar fractures: A retrospective case series of 185 Patients with a single-level spinal column injury.

STUDY DESIGN: Retrospective Single-Center Review of Data at a Level 1 Trauma Center. OBJECTIVE: Compare deformity correction and surgical outcomes of percutaneous instrumentation and open fusion in traumatic thoracolumbar fractures. METHODS: In our retrospective study, all patients undergoing elective spine surgery for TL fractures at a Level 1 trauma center between 2000 and 2017 were reviewed. Patients who underwent percutaneous fixation were given the option of hardware removal after the fracture had healed. RESULTS: A total of 185 patients were included in the study, with 109 treated with an open fusion, and 76 with percutaneous fixation. Twenty-five patients in the latter group had the instrumentation removed after the fracture had healed. None of them required reoperation. In the open fusion group 54.1% of patients required a decompressive laminectomy. Percutaneous fixation patients had a shorter operative time (98.3 min vs 214 min, p < 0.0001), shorter length of stay (9.8 days vs 13.5 days, p = 0.04), and less blood loss (68.4 cc vs 691 cc, p < 0.001). They also had a better correction of their traumatic kyphosis after surgery (p = 0.005). CONCLUSION: Percutaneous fixation is a valuable option for the treatment of TL fractures in cases without evidence of neural compression. It is still unclear whether hardware removal helps prevent adjacent segment degeneration. Percutaneous fixation could allow for better reduction of the fracture with improvement of postoperative alignment.

Neurosurgeons Relate Heterogeneous Practices Regarding Activity and Return to Work After Spine Surgery.

OBJECTIVE: Given the paucity of relevant data, the Council of State Neurosurgical Societies Workforce Committee launched a survey of neurosurgeons to assess patterns in activity restriction recommendations following spine surgery; the ultimate goal was to optimize and potentially standardize these recommendations. The aim of this initial study was to determine current practices in activity restrictions and return to work guidelines following common spinal procedures. METHODS: The survey included questions regarding general demographics and practice data, postoperative bracing/orthosis utilization, and guidelines for postoperative return to different levels of activity/types of work following specific spine surgery interventions. A spectrum of typical spine surgeries was assessed, including microdiscectomy, anterior cervical discectomy and fusion (ACDF), and lumbar fusion, both open and minimal invasive surgery (MIS) approaches. RESULTS: There was significant interprocedure and intraprocedure variation in the neurosurgeons' recommendations for postoperative activity and return to work recommendations after various spinal surgery procedures. Comparisons of the different surgical procedures evaluated revealed significant differences in cervical collar use (more often used following ≥2-level ACDF than single-level ACDF; P < 0.001), return to both sedentary and light physical work (greater restriction with ≥2-level ACDF than with single-level ACDF; P < 0.001), and return to a light exercise regimen (sooner following MIS versus open lumbar fusion; P < 0.001). CONCLUSIONS: This survey demonstrated little consistency regarding return to work recommendations, general activity restrictions, and orthosis utilization following common spinal surgical procedures. Addressing this issue also has significant implications for the societal and personal costs of spine surgery.

Educational impact of early COVID-19 operating room restrictions on neurosurgery resident training in the United States: A multicenter study.

BACKGROUND: The coronavirus (COVID-19) pandemic has caused unprecedented suspensions of neurosurgical elective surgeries, a large proportion of which involve spine procedures. The goal of this study is to report granular data on the impact of early COVID-19 pandemic operating room restrictions upon neurosurgical case volume in academic institutions, with attention to its secondary impact upon neurosurgery resident training. This is the first multicenter quantitative study examining these early effects upon neurosurgery residents caseloads. METHODS: A retrospective review of neurosurgical caseloads among seven residency programs between March 2019 and April 2020 was conducted. Cases were grouped by ACGME Neurosurgery Case Categories, subspecialty, and urgency (elective vs. emergent). Residents caseloads were stratified into junior (PGY1-3) and senior (PGY4-7) levels. Descriptive statistics are reported for individual programs and pooled across institutions. RESULTS: When pooling across programs, the 2019 monthly mean (SD) case volume was 214 (123) cases compared to 217 (129) in January 2020, 210 (115) in February 2020, 157 (81), in March 2020 and 82 (39) cases April 2020. There was a 60% reduction in caseload between April 2019 (207 [101]) and April 2020 (82 [39]). Adult spine cases were impacted the most in the pooled analysis, with a 66% decrease in the mean number of cases between March 2020 and April 2020. Both junior and senior residents experienced a similar steady decrease in caseloads, with the largest decreases occurring between March and April 2020 (48% downtrend). CONCLUSIONS: Results from our multicenter study reveal considerable decreases in caseloads in the neurosurgical specialty with elective adult spine cases experiencing the most severe decline. Both junior and senior neurosurgical residents experienced dramatic decreases in case volumes during this period. With the steep decline in elective spine cases, it is possible that fellowship directors may see a disproportionate increase in spine fellowships in the coming years. In the face of the emerging Delta and Omicron variants, programs should pay attention toward identifying institution-specific deficiencies and developing plans to mitigate the negative educational effects secondary to such caseloads reduction.

Spontaneous closure of myelomeningocele.

Fetal repair of myelomeningocele has been increasingly offered to mothers of children with myelomeningocele after the seminal Management of Myelomeningocele (MOMs) trial, which demonstrated decreased reliance on ventriculoperitoneal shunt following fetal closure. We present the case of a fetus diagnosed with a lumbar myelomeningocele in utero whose mother refused in utero closure and who was subsequently born with a skin-covered defect. A fetal MRI was obtained on a mother with a male fetus diagnosed with open neural tube defect at 20 weeks of gestation. The child demonstrated spinal dysraphism extending from L2 to L5 and associated Chiari II malformation with lateral and third ventriculomegaly. Based on our institutional criteria and the criteria of the MOMs trial, the parents were offered fetal repair of the myelomeningocele; however, they declined because of concerns about risks to the mother. At birth, the patient was found to have a skin-covered meningocele. He underwent elective repair of his occult meningocele and detethering of his spinal cord. Intraoperative findings demonstrated spinal nerve roots attached to the arachnoid within the defect, and a closed, tubularized neural placode. This represents a unique case in which a fetus with a clinical picture consistent with open spinal defect was found to have a lesion more consistent with meningocele on postnatal operative interrogation. Knowledge that this can occur should be taken into consideration when discussing fetal closure, although the frequency of this occurrence is not known. Additionally, identification of this case sheds light on the mechanism by which occult myelomeningoceles form.

Trans-Sulcal Endoport-Assisted Evacuation of Supratentorial Intracerebral Hemorrhage: Initial Single-Institution Experience Compared to Matched Medically Managed Patients and Effect on 30-Day Mortality.

BACKGROUND: The surgical management of supratentorial intracerebral hemorrhages (ICH) remains controversial due to large trials failing to show clear benefits. Several minimally invasive techniques have emerged as an alternative to a conventional craniotomy with promising results. OBJECTIVE: To report our experience with endoport-assisted surgery in the evacuation of supratentorial ICH and its effects on outcome compared to matched medical controls. METHODS: Retrospective data were gathered of patients who underwent endoport-assisted evacuation between January 2014 and October 2016 by a single surgeon. Patients who were managed medically during the same period were matched to the surgical cohort. Previously published cohorts investigating the same technique were analyzed against the present cohort. RESULTS: Sixteen patients were identified and matched to 16 patients treated medically. Location, hemorrhage volume, and initial Glasgow Coma Scale (GCS) score did not differ significantly between the 2 cohorts. The mean volume reduction in the surgical cohort was 92.05% ± 7.05%. The improvement in GCS in the surgical cohort was statistically significant (7-13, P = .006). Compared to the medical cohort, endoport-assisted surgery resulted in a statistically significant difference in in-hospital mortality (6.25% vs 75.0%, P < .001) and 30-d mortality (6.25% vs 81.25%, P < .001). Compared to previously published cohorts, the present cohort had lower median preoperative GCS (7 vs 10, P = .02), but postoperative GCS did not differ significantly (13 vs 14, P = .28). CONCLUSION: Endoport-assisted surgery is associated with high clot evacuation and decreases 30-d mortality compared to a similar medical group.