RESUMO
OBJECTIVE: To identify distinct clinical or imaging subtypes of spontaneous intracranial hypotension (SIH) due to spinal cerebrospinal fluid (CSF) venous fistula (CVF). BACKGROUND: Spontaneous intracranial hypotension is classically understood to present clinically with an orthostatic headache and stereotyped brain magnetic resonance imaging (MRI) findings; however, most prior literature examining clinical and brain MRI features of SIH has focused on all types of spinal CSF leaks concurrently. This study aimed to evaluate whether data support the possibility of internally consistent subtypes based on brain imaging features and clinical symptoms analogous to those seen in primary headache syndromes. METHODS: This retrospective cross-sectional single-institution study included 48 consecutive patients meeting the International Classification of Headache Disorders, 3rd edition criteria for SIH due to CVF. Clinical symptoms, pre-treatment brain MRI, and symptom duration were analyzed. Clinical and MRI data were analyzed to identify patterns and associations between symptoms and imaging findings. RESULTS: A total of 20 males and 28 females were evaluated, with a mean (standard deviation) age of 61 (10) years. In all, 44/48 (92%) patients experienced headaches, though 18/48 (40%) did not endorse relief when flat, including six of the 48 (13%) with worsening symptoms when flat. In all, 19/48 (40%) patients reported at least one migraine symptom, and six of the 48 (13%) presented with at least one migraine symptom and had no relief when flat. Clinical symptoms clustered primarily into a "classic" presentation consisting of relief when flat, occipital head pain, comorbid neck pain, a pressure/throbbing headache quality, and an "atypical" presentation that was characterized by having several differences: less relief when flat (nine of 22 (41%) vs. 20/23 (87.0%), p = 0.002; odds ratio [OR] 0.110, 95% confidence interval [CI] 0.016-0.53), more frontal head pain (14/22 (64%) vs. one of 23 (4%), p < 0.001; OR 35.0, 95% CI 4.2-1681.0), less neck pain (two of 21 (4.5%) vs. nine of 13 (69.6%), p < 0.001; OR 0.023, 95% CI 0.0005-0.196), and more stabbing/sharp headache quality (nine of 22 (41%) vs. two of 23 (9%), p = 0.017; OR 7.0, 95% CI 1.18-75.9). Brain MRI findings clustered into three groups: those presenting with most imaging findings of SIH concurrently, those with brain sag but less pachymeningeal/venous engorgement, and those with pachymeningeal/venous engorgement but less brain sag. CONCLUSION: This study highlights the clinical and imaging diversity among patients with SIH due to CVF, challenging the reliance on classic orthostatic headache alone for diagnosis. The findings suggest the existence of distinct SIH subtypes based on clinical and imaging presentations, underscoring the need for comprehensive evaluation in patients with suspected CVF. Future research should further elucidate the relationship between clinical symptoms and imaging findings, aiming to refine diagnostic criteria and enhance understanding of SIH's pathophysiology.
RESUMO
Postsurgical falls have significant patient and societal implications but remain challenging to identify and track. Detecting postsurgical falls is crucial to improve patient care for older adults and reduce healthcare costs. Large language models (LLMs) offer a promising solution for reliable and automated fall detection using unstructured data in clinical notes. We tested several LLM prompting approaches to postsurgical fall detection in two different healthcare systems with three open-source LLMs. The Mixtral-8×7B zero-shot had the best performance at Stanford Health Care (PPV = 0.81, recall = 0.67) and the Veterans Health Administration (PPV = 0.93, recall = 0.94). These results demonstrate that LLMs can detect falls with little to no guidance and lay groundwork for applications of LLMs in fall prediction and prevention across many different settings.
RESUMO
BACKGROUND: Postdural puncture headache has been traditionally viewed as benign, self-limited, and highly responsive to epidural blood patching (EBP) when needed. A growing body of data from patients experiencing unintended dural puncture (UDP) in the setting of attempted labor epidural placement suggests a minority of patients will have more severe and persistent symptoms. However, the mechanisms accounting for the failure of EBP following dural puncture remain obscure. An understanding of these potential mechanisms is critical to guide management decisions in the face of severe and persistent cerebrospinal fluid (CSF) leak. CASE PRESENTATION: We report the case of a peripartum patient who developed a severe and persistent CSF leak unresponsive to multiple EBPs following a UDP during epidural catheter placement for labor analgesia. Lumbar MRI revealed a ventral rather than dorsal epidural fluid collection suggesting that the needle had crossed the thecal sac and punctured the ventral dura, creating a puncture site not readily accessible to blood injected in the dorsal epidural space. The location of this persistent ventral dural defect was confirmed with digital subtraction myelography, permitting a transdural surgical exploration and repair of the ventral dura with resolution of the severe intracranial hypotension. CONCLUSIONS: A ventral rather than dorsal dural puncture is one mechanism that may contribute to both severe and persistent spinal CSF leak with resulting intracranial hypotension following a UDP.