Diagnostic accuracy of an artificial intelligence algorithm versus radiologists for fracture detection on cervical spine CT.

OBJECTIVES: To compare diagnostic accuracy of a deep learning artificial intelligence (AI) for cervical spine (C-spine) fracture detection on CT to attending radiologists and assess which undetected fractures were injuries in need of stabilising therapy (IST). METHODS: This single-centre, retrospective diagnostic accuracy study included consecutive patients (age ≥18 years; 2007-2014) screened for C-spine fractures with CT. To validate ground truth, one radiologist and three neurosurgeons independently examined scans positive for fracture. Negative scans were followed up until 2022 through patient files and two radiologists reviewed negative scans that were flagged positive by AI. The neurosurgeons determined which fractures were ISTs. Diagnostic accuracy of AI and attending radiologists (index tests) were compared using McNemar. RESULTS: Of the 2368 scans (median age, 48, interquartile range 30-65; 1441 men) analysed, 221 (9.3%) scans contained C-spine fractures with 133 IST. AI detected 158/221 scans with fractures (sensitivity 71.5%, 95% CI 65.5-77.4%) and 2118/2147 scans without fractures (specificity 98.6%, 95% CI 98.2-99.1). In comparison, attending radiologists detected 195/221 scans with fractures (sensitivity 88.2%, 95% CI 84.0-92.5%, p < 0.001) and 2130/2147 scans without fracture (specificity 99.2%, 95% CI 98.8-99.6, p = 0.07). Of the fractures undetected by AI 30/63 were ISTs versus 4/26 for radiologists. AI detected 22/26 fractures undetected by the radiologists, including 3/4 undetected ISTs. CONCLUSION: Compared to attending radiologists, the artificial intelligence has a lower sensitivity and a higher miss rate of fractures in need of stabilising therapy; however, it detected most fractures undetected by the radiologists, including fractures in need of stabilising therapy. Clinical relevance statement The artificial intelligence algorithm missed more cervical spine fractures on CT than attending radiologists, but detected 84.6% of fractures undetected by radiologists, including fractures in need of stabilising therapy. KEY POINTS: The impact of artificial intelligence for cervical spine fracture detection on CT on fracture management is unknown. The algorithm detected less fractures than attending radiologists, but detected most fractures undetected by the radiologists including almost all in need of stabilising therapy. The artificial intelligence algorithm shows potential as a concurrent reader.

Surgical versus conservative treatment for odontoid fractures in older people: an international prospective comparative study.

BACKGROUND: The optimal treatment for odontoid fractures in older people remains debated. Odontoid fractures are increasingly relevant to clinical practice due to ageing of the population. METHODS: An international prospective comparative study was conducted in fifteen European centres, involving patients aged ≥55 years with type II/III odontoid fractures. The surgeon and patient jointly decided on the applied treatment. Surgical and conservative treatments were compared. Primary outcomes were Neck Disability Index (NDI) improvement, fracture union and stability at 52 weeks. Secondary outcomes were Visual Analogue Scale neck pain, Likert patient-perceived recovery and EuroQol-5D-3L at 52 weeks. Subgroup analyses considered age, type II and displaced fractures. Multivariable regression analyses adjusted for age, gender and fracture characteristics. RESULTS: The study included 276 patients, of which 144 (52%) were treated surgically and 132 (48%) conservatively (mean (SD) age 77.3 (9.1) vs. 76.6 (9.7), P = 0.56). NDI improvement was largely similar between surgical and conservative treatments (mean (SE) -11 (2.4) vs. -14 (1.8), P = 0.08), as were union (86% vs. 78%, aOR 2.3, 95% CI 0.97-5.7) and stability (99% vs. 98%, aOR NA). NDI improvement did not differ between patients with union and persistent non-union (mean (SE) -13 (2.0) vs. -12 (2.8), P = 0.78). There was no difference for any of the secondary outcomes or subgroups. CONCLUSIONS: Clinical outcome and fracture healing at 52 weeks were similar between treatments. Clinical outcome and fracture union were not associated. Treatments should prioritize favourable clinical over radiological outcomes.

Medical Malpractice in Neurosurgery: An Analysis of Claims in the Netherlands.

BACKGROUND AND OBJECTIVES: Studying malpractice claims is important to improve quality of health care and patient safety and to educate the individual healthcare providers. The objective of this study was to describe characteristics of neurosurgical claims in the Netherlands. METHODS: A nationwide retrospective observational study of neurosurgery-related claims closed by Centramed and MediRisk, 2 major insurance companies in the Netherlands, was performed. Relevant data, including type of neurosurgical pathology, theme and category of the claim, type and severity of injury, outcome, and financial burden, were extracted from anonymized claim files. The estimated annual risk was used to determine the risk for claims by adjusting for the number of annually practicing neurosurgeons in the Netherlands. RESULTS: A total of 388 claims against neurosurgeons were closed between 2007 and 2021. Liability was denied in a slight majority of claims (n = 230; 59%). The total burden during this period was €6 165 000 (amount paid out to patients: €5 497 000). The estimated annual risk per Dutch neurosurgeon for a claim was 15.5%, meaning 1 claim per 6.5 years. The case-level analysis of 238 available anonymized claims revealed that most claims were related to spinal pathology (81.5%), followed by cranial pathology (10.9%) and peripheral nerve (7.6%). The motivations for filing claims were mostly related to alleged surgical (56.3%) or diagnostic errors (22.3%). Most of these claims were denied (151/238; 63.4%), and fewer were settled (42/238; 17.6%), sustained (31/238; 13.0%), or closed without final decision (14/238; 5.9%). CONCLUSION: Neurosurgery-related malpractice claims primarily involved spinal pathology and were mostly related to alleged treatment errors. Most claims did not result in compensation because there seemed to be no liability or culpable injury. However, the annual risk for a claim for Dutch neurosurgeons is considerable.

The effect of the introduction of the Amsterdam Trauma Workflow Concept on mortality and functional outcome of patients with severe traumatic brain injury.

The purpose of this study was to analyze the effect of the introduction of an all-in workflow concept that included direct computed tomography (CT) scanning in the trauma room on mortality and functional outcome of trauma patients with severe traumatic brain injury (TBI) admitted to a level-1 trauma center. To this end, a retrospective comparison was made of a 1-year cohort prior to the implementation of the all-in workflow concept (Pre-CT in trauma room cohort [Pre-TRCT]) and a 1-year cohort after the implementation (Post-TRCT). All severely injured TBI patients aged 16 years or older that were presented in our level-1 trauma center and that underwent a CT of the head were initially included. Severe TBI was defined as an Abbreviated Injury Scale (AIS) score of >2 of the head region following trauma. Primary outcome parameter was TBI-related mortality during primary hospital admission. Secondary outcome parameter was the functional outcome based on GOS-Extended. A total of 59 patients were included in the Pre-TRCT and 49 in the Post-TRCT. Median age was 49 years in the Post-TRCT and 44 years in the Pre-TRCT (not significant [NS]). Median ISS was similar (ISS = 25). Median Head-AIS was higher in the Post-TRCT (5 vs. 4, NS). Initial CT scanning was completed faster in the Post-TRCT. There was a significant difference of 23% mortality in favor of the Post-TRCT for TBI-related mortality during primary hospital admission (p < 0.05). For acute neurosurgical interventions, time until intervention tended to be faster in the Post-TRCT (NS). Functional outcomes for survivors were higher in the Post-TRCT (6 vs. 5, NS).

Skin Augmentation as a Last-Resort Operative Technique During Decompressive Craniectomy.

OBJECTIVE: Since 2009, we have performed skin augmentation using a Gore-Tex patch as a last-resort measure to reduce intracranial pressure (ICP) in uncontrollable brain swelling during decompressive craniectomy (DC). Here, we report our experience and outcome in a consecutive series of patients undergoing DC with skin augmentation (DC+S). METHODS: In 2009-2015, a prospective database was created registering all patients who underwent DC+S when ICP increased >25 mm Hg while approximating the skin edges after DC (or when closing the skin was impossible because of uncontrollable brain swelling in patients without an ICP monitoring catheter). Patients' baseline characteristics and 1-year outcome were compared with patients undergoing DC without the need of skin augmentation in the same time frame. Outcome according to the Glasgow Outcome Scale (GOS) was dichotomized into favorable (GOS score 4-5) and unfavorable (GOS 1-3). RESULTS: Of a total of 180 consecutive patients with DC, 20 (11%) underwent DC+S. Four (20%) survived favorably, 2 (10%) unfavorably, and 14 (70%) died (compared with 36%, 22%, and 42%, respectively, in patients with standard DC). Four of 7 patients in whom DC+S was performed ≥24 hours after injury or at second surgery survived favorably, versus none of the 13 patients in whom DC+S was performed <24 hours after injury and at first surgery. Two of 10 patients surviving the first week after DC+S had a skinplasty-related infection. CONCLUSIONS: Decompressive craniectomy with skin augmentation may be used as a last-resort measure in cases of severe brain swelling despite DC.

Continuous monitoring of cerebrospinal fluid oxygen tension in relation to motor evoked potentials during spinal cord ischemia in pigs.

BACKGROUND: Perioperative assessment of spinal cord oxygenation might guide measures to prevent neurologic deficits secondary to ischemic or traumatic damage of the spinal cord. Although cerebrospinal fluid (CSF) partial pressure of oxygen (Po2) measurement has been used to detect spinal cord ischemia (SCI), the diagnostic value and the temporal resolution of CSF Po2 measurement compared with functional assessment of the spinal cord is unknown. This study compared CSF Po2 with transcranial motor evoked potentials (tcMEPs) for detection of experimental SCI. METHODS: The aorta and segmental arteries were exposed in 10 sufentanil-ketamine-anesthetized pigs (weight, 40-50 kg). Myogenic tcMEPs were recorded from the upper and lower limbs, and continuous assessment of CSF Po2 was provided by two Clark-type microcatheters inserted in the lumbar and thoracic intrathecal space. Graded lumbar SCI was produced by sequential clamping of segmental arteries. The relation between CSF Po2 and tcMEP during graded SCI was determined using linear regression. Diagnostic characteristics of CSF Po2 values for clinical SCI were determined using different cutoff points of CSF Po2. RESULTS: Lumbar CSF Po2 (baseline, 44 [interquartile range, 38-54] mmHg) decreased below 50% in all animals and was linearly related to loss of tcMEP amplitude in all animals. The median lumbar CSF Po2 during reduction of tcMEP to less than 25% of baseline was 11 (4-29) mmHg, whereas thoracic CSF Po2 remained constant (40 [28-50] mmHg). During absence of the tcMEP signal, lumbar CSF Po2 was less than 20 mmHg in 80% of the animals. Optimal sensitivity and predictive values of CSF Po2 measurement for SCI were in the range of 40-60% of baseline. CONCLUSIONS: The data indicate that intrathecal Po2 measurement is a sensitive monitoring technique to track real-time changes in local spinal cord oxygenation. Continuous monitoring of CSF Po2 might be applied for evaluation of patients who are at risk for direct or secondary SCI.