When I spoke recently with a Venezuelan anesthesiologist about CSF leaks, she explained that she had relatively little experience performing epidural blood patches. Several Colombian general practitioners told me something similar, describing the epidural blood patch as advanced technology. That framing is worth pausing on, because the procedure is not new by any reasonable clinical standard. Robert Danis first proposed sealing a dural puncture with an epidural injection in 1924. The modern technique was performed and reported by James Gormley, a general surgeon rather than a specialist, in 1960, using two to three milliliters of a patient's own blood after he observed that "bloody taps" during spinal puncture seemed to reduce the incidence of post-dural puncture headache. A procedure first described by a general surgeon over one hundred years ago, using nothing more exotic than a needle and the patient's own blood, is not advanced technology in any technical sense. What it has become, in parts of South America, is a procedure that simply never entered the standard curriculum, which is a very different problem than technological scarcity, and one that says more about training pathways than about the difficulty of the technique itself.
The training gap I have observed is not really a training problem on its own. It is an incentive-alignment problem. Training follows incentives, curriculum reflects what gets tested, and skills are maintained by being performed regularly. Most physicians who learn the epidural blood patch learn it in the narrow context of obstetric anesthesia, managing a known post-spinal headache after a labor epidural, a relatively standardized presentation with a clear onset. Spontaneous intracranial hypovolemia does not announce itself the same way, may require thoracic or cervical approaches depending on where the leak sits, and training confined to the obstetric context does not prepare physicians for that complexity. When competency in recognizing and treating these leaks is not an explicit requirement across neurology, anesthesiology, pain management, and family medicine residencies, the skill stays concentrated in a small number of centers, and most patients with a leak never reach a clinician equipped to recognize what is happening to them.
While researching Spanish-language patient discussions, I intentionally searched for symptom descriptions rather than diagnoses. Instead of searching for "spinal leak," I searched for phrases describing headaches that improved after lying down, or symptoms that required patients to stay horizontal for relief. Across multiple forums I found numerous women describing patterns remarkably similar to those eventually diagnosed as spinal CSF leaks. Forum discussions cannot establish a diagnosis on their own, but they can reveal symptom patterns worthy of further clinical study, and I think that kind of pattern-listening across language and geography is undervalued in how research questions typically get formed. One conversation especially stayed with me. After a podcast in which I discussed my own spinal leak experience, an Argentine friend contacted me, describing a spinal anesthetic after a soccer injury years earlier and a strange, lasting vertigo triggered by moving text on a screen. His experience does not establish causation, but it reinforces why it matters to ask open-ended questions rather than assuming every patient recovers completely from spinal anesthesia.
An epidural blood patch works because the patient's own blood clots and adheres over the dural defect, both sealing the leak mechanically and, over time, promoting a durable fibrous repair. That means the quality of the blood being used matters as much as the skill of the person delivering it, and this is a variable that gets almost no attention in the spinal CSF leak literature I have reviewed. Many chronic leak patients accumulate long histories of dietary restriction, whether from food sensitivities, gastrointestinal motility problems, or the kind of restrictive eating that develops when almost everything seems to worsen symptoms. That pattern of restriction has predictable hematologic consequences. Vitamin K deficiency impairs the synthesis of clotting factors and can present with visible signs like petechiae and slow-clotting blood. Iron-deficiency anemia is common in patients with restricted diets and chronic gastrointestinal symptoms. Both vitamin B12 and folate deficiency are established causes of thrombocytopenia, low platelet counts, because bone marrow needs both nutrients to complete normal cell division during platelet production, and clinical reviews describe folate and B12 repletion as capable of rapidly correcting the resulting thrombocytopenia. Every one of those deficiencies directly affects a patient's capacity to form a stable clot, which is the entire mechanism an epidural blood patch depends on.
A patient with a long-standing, undiagnosed spinal CSF leak has often spent months or years modifying their diet in response to symptoms that no one has correctly attributed to the leak itself, which means the blood being drawn for a patch may already be compromised by the time a patient finally reaches treatment. A poorly clotting patch is more likely to fail, more likely to require repeat procedures, and more likely to generate the kind of "the patch didn't work" outcome that then gets attributed entirely to technique or to the wrong leak site, when a contributing factor may never have been assessed at all. This is a place where a basic coagulation panel, a ferritin level, and a B12 and folate check before a patch, standard, inexpensive, and unglamorous tests, could meaningfully change outcomes, and yet in my experience these are rarely ordered as a matter of routine before a procedure whose entire success depends on blood behaving normally.
This is also where I think Colombian and Korean care each have something to learn from the other, and where both currently fall short. In Colombia and Ecuador, dietitians are woven into general medical care in a way that is unusual by US standards, and the Colombian and Ecuadorian dietitians I have worked with are comfortable reading and interpreting a complete blood count or a coagulation panel as a routine part of their scope, not as something reserved for hematology. That integration is a real strength, and it is precisely the kind of nutrition-literate, lab-literate clinical culture that spinal CSF leak care needs and does not currently have anywhere I have observed. But even in Colombia, spinal CSF leak recognition itself remains rare, which means that strength in nutrition is not currently being applied to this specific patient population, because the patients are not being identified in the first place. South Korea has the opposite gap. Its pain anesthesiology workforce has, as I will describe below, become unusually procedurally strong at the epidural blood patch itself, but nothing in the incentive structure I have researched suggests that Korean pain medicine practices are screening leak patients for the nutritional status of the blood they are about to inject. A system that has solved the procedural bottleneck has not solved the blood-quality bottleneck, and a system that understands the blood-quality bottleneck has not yet solved the procedural or diagnostic bottleneck. Neither failure is a reason to rank one system above the other. It is a reason to ask what each could import from the other.
South Korea does not have a free-price private medical market in the way the United States, the United Kingdom, or Australia does. Nearly all providers, including privately owned hospitals and clinics, operate inside the national health insurance system, and for covered services the government negotiates a fixed national fee schedule. A famous specialist cannot simply decide that an ordinary covered procedure is worth five or ten times more because of their reputation. Using currently regulated Korean fee figures, a lumbar epidural block reimburses at roughly the equivalent of $27, a cervical or thoracic epidural block at roughly $46, and a fluoroscopic transforaminal-type procedure at roughly $88. Korean physicians have complained for decades that these fees are low relative to cost, and research on Korean reimbursement has found that regulated fee-for-service payment tends to push providers toward higher procedural volume and, where possible, toward non-covered services to make up the difference. For a physician doing epidural procedures, that produces a very specific outcome: low price per procedure, and a strong incentive to become extremely fast, efficient, and technically reliable at doing a large number of them.
Institutionally, "Anesthesiology and Pain Medicine" functions in South Korea as a single, combined specialty identity, and academic pain centers doing epidural steroid injections, nerve blocks, and radiofrequency procedures are routinely housed inside anesthesiology departments rather than being split off into a separate interventional pain specialty. Korean literature on spontaneous intracranial hypotension and epidural blood patching consistently comes out of these same anesthesiology and pain medicine departments, including case reports of technically demanding cervical patches. That produces a natural technical progression: epidural anesthesia leads into pain medicine, pain medicine leads into epidural steroid injections and nerve blocks under fluoroscopic guidance, and fluoroscopic epidural access leads naturally into epidural blood patching, because the hardest part of a blood patch is not drawing the blood. It is obtaining controlled epidural access without creating a second dural hole, recognizing loss of resistance accurately, and understanding how a viscous fluid behaves once it is in the epidural space, all skills built through sheer procedural repetition rather than through diagnostic imaging expertise.
Crucially, Korea's low reimbursement per procedure does not push anesthesiologists away from spinal leak care the way American reimbursement does. In the United States, Medicare data show lumbar and sacral epidural steroid injections rising from about 119,000 procedures in 2000 to 1.97 million in 2020, an increase of over 1,500 percent, alongside a similarly explosive rise in cervical and thoracic injections. That volume created an enormous, well-reimbursed chronic pain market that American pain anesthesiologists could specialize around, and spontaneous CSF leak patients, being rare, diagnostically uncertain, and comparatively unprofitable by comparison, were a far less attractive use of exactly the same underlying epidural skill set. American anesthesiology did not abandon the blood patch outright; it retained it strongly in obstetrics, where epidural and spinal anesthesia are used in roughly three-quarters of vaginal deliveries and a known accidental dural puncture creates an unambiguous, well-defined indication for a patch. What appears to have never developed as strongly is the middle category: a pain anesthesiologist comfortable determining the appropriate level and volume for a spontaneous or iatrogenic leak outside the obstetric setting. Korea's comparatively modest reimbursement for chronic pain procedures never created that same gravitational pull away from leak care, so a blood patch remained, economically, just another epidural procedure within an existing pain practice rather than a detour away from a more lucrative specialty.
None of this means radiology has nothing to contribute. It means the contribution is best aimed at diagnosis rather than at owning the needle. The Bern score, formally the brain spontaneous intracranial hypotension score, was developed by a neuroradiology group at the University Hospital of Bern in Switzerland and published in 2019. It assigns points across six brain MRI findings, ranging from pachymeningeal enhancement to venous sinus engorgement, to estimate the probability that a patient with suspected spontaneous intracranial hypotension has an underlying spinal CSF leak, without requiring an invasive myelogram to get that estimate. It has since been validated, and revalidated, in other cohorts, including a 2025 Korean study out of Seoul National University Hospital, and subsequent research has shown a Bern score of five or higher combined with a positive non-invasive MR myelogram is a reasonable basis to proceed directly to an empiric epidural blood patch before resorting to invasive myelography at all. That is the division of labor I think makes the most clinical and economic sense: radiology refines non-invasive brain and spine imaging to identify likely leak patients and localize difficult cases when necessary, and a well-trained pain anesthesiology workforce delivers the actual epidural treatment, repeating it when appropriate, because the evidence base already supports offering a second empiric patch before jumping to invasive localization in many patients.
In the United States, United Kingdom, Australia, and Switzerland, chronic spinal CSF leak care has tended to concentrate inside a small number of tertiary neuroradiology-led programs, and the reasoning behind that concentration is not irrational. Finding a subtle or occult leak, particularly a CSF-venous fistula, genuinely is a difficult diagnostic-radiology problem, involving digital subtraction myelography, dynamic CT myelography, and the interpretation of very subtle imaging findings. Once a patient's pathway becomes one in which a neuroradiologist diagnoses the leak, performs the myelogram, and decides on the target, it becomes a short institutional step for the same specialist to perform the patch as well, and there is legitimate precedent in the literature for CT-guided targeted patching performed by neuroradiologists. But that precedent establishes that a neuroradiologist can perform a technically excellent, image-guided patch in a genuinely complex case. It does not establish that a neuroradiologist is the best-positioned specialist for an ordinary, non-targeted lumbar or thoracic patch, which the current multidisciplinary consensus guideline for spontaneous intracranial hypotension actually recommends as first-line treatment before invasive leak localization in many patients. There is a real difference between two skill questions here. "Where exactly should the needle go" is a question that favors advanced imaging expertise. "How do I manipulate an epidural needle atraumatically and reliably, thousands of times over" is a question that favors whoever has the highest neuraxial procedural volume, and in most Western systems that is not neuroradiology.
Neurosurgery occupies a further, distinct role in this pathway. Neurosurgeons become genuinely indispensable when treatment means repairing a structural dural defect, ligating a CSF-venous fistula, or removing an offending disc spur, but routinely performing epidural injections is not the core, repetitive motor skill of neurosurgical training the way neuraxial access is for anesthesiology. When a system routes an ordinary patch through neurosurgery or through a scarce, prestige-priced neuroradiology consultant, it can end up in a strange position: a patient pays for a scarce specialist's time to perform something that a high-volume interventional anesthesiologist may have technically executed thousands of times over the course of an ordinary pain-medicine career. Price, in other words, is a poor proxy for who has actually built the specific manual skill the procedure requires.
Local anesthetics and steroid solutions used in an ordinary epidural steroid injection are thin, low-viscosity fluids that flow easily through a 22-gauge or even smaller needle. Autologous blood is a much more viscous fluid, and forcing a meaningful volume of it through a narrow lumen under the pressure needed to inject in a reasonable timeframe changes the physics of the procedure entirely. This is why 18-gauge Tuohy needles, not the smaller gauges common in routine pain-medicine injections, are the standard instrument described throughout the epidural blood patch literature, from routine lumbar patches to caudal and cervical approaches. The physics behind that choice is well studied outside the blood patch literature specifically. Research on hemolysis in needles and catheters has found that red blood cell rupture scales with wall shear stress and the pressure differential required to force blood through a given bore, and clinical studies of rapid blood transfusion through small needles found that passage through 23-gauge or smaller needles caused measurable hemolysis while 18- and 20-gauge needles did not. Pushing viscous blood through a needle several sizes smaller than standard, under enough pressure to complete the injection in a reasonable time, is very plausibly forcing exactly the kind of dynamic shear stress that this literature associates with red cell rupture, even though a study measuring hemolysis specifically during epidural blood patch procedures does not appear to exist yet, which is itself a notable gap in the literature given how central intact, clot-competent blood is to the procedure's mechanism.
Clinically, this matters for two separate reasons. First, a patch's ability to seal a dural defect depends on the blood's capacity to clot and form a stable mass over the defect, and hemolyzed blood, with damaged or destroyed red cells, is compromised as clotting material in a way that could plausibly reduce how durable or effective the resulting seal is. Second, free hemoglobin released by hemolysis is a known irritant in central nervous system spaces; its presence and breakdown products are part of what makes conditions like subarachnoid hemorrhage clinically dangerous and symptomatic, and there is no clear reason to assume free hemoglobin introduced into the epidural space would be inert. Neither of these concerns has been rigorously studied in the specific context of blood patch technique, and that absence of research is worth naming plainly rather than treating as reassurance.
Several American, French, and British patients have described to me being severely and repeatedly punctured, or in at least one case having air accidentally introduced, during attempts to obtain a blood patch performed by anesthesiologists rather than image-guided specialists. Their accounts describe rushed procedures performed without fluoroscopic or CT guidance and, by their telling, without the procedural volume that would let an operator work confidently by feel alone. This is a useful counterpoint to the argument that anesthesiology is inherently the right specialty for every ordinary patch. The Korean model works, to the extent it does, specifically because Korean pain anesthesiologists are performing an enormous number of epidural procedures under a system that rewards volume and, often, uses imaging guidance as a matter of course. An anesthesiologist who does not have that volume, or who is working without any imaging guidance in a system that does not reward slowing down, is not automatically safer than a radiologist. Specialty label alone, as the Korean literature illustrates in the other direction with reports of anesthesiology-performed CT-guided cervical patches, does not tell you procedural competence. What tells you competence is volume, guidance, and an incentive structure that rewards getting it right rather than getting through the schedule.
In an informal conversation, a Spanish neuroradiologist described performing epidural blood patches using a 22-gauge needle, a gauge suited to thin fluids like contrast or local anesthetic but far narrower than the 18-gauge standard described throughout the blood patch literature. When I asked whether hemolysis from forcing viscous blood through that narrow a bore had ever been a consideration, he laughed the question off and said he had never thought about it. I have not had the same reaction from the American anesthesiologists I have spoken with about technique specifically, several of whom explained unprompted that they use an 18-gauge Tuohy needle precisely because of blood's viscosity. I want to be careful not to overstate what that difference means. My conversations with US anesthesiologists suggest that even among clinicians who correctly use the larger needle, understanding of the hemolysis mechanism itself is uneven, and younger US anesthesiologists in particular have sometimes described the correct needle gauge as something they were taught by rote rather than something they could explain the physics of. A clinician can get the equipment right without understanding why it is right, and a clinician who has never been trained in a discipline built around high-volume neuraxial access, which describes most neuroradiologists, may not have had the exposure that would prompt the question in the first place.
The same American market that produced an extraordinarily high volume of chronic-pain procedures has also produced a documented backlash. Pain management specialists themselves have criticized epidural steroid injections as overused, with one prominent pain physician describing patients who received two to three dozen injections in a single year and describing the cumulative effect as a kind of trauma to the outer layer of the spinal cord. A Medicare oversight audit examined over 300,000 epidural steroid injection sessions and found tens of thousands exceeding applicable utilization limits. Patients in the US, and in other countries with similarly reimbursement-driven procedural markets, including reports I have heard from patients in Spain and Australia, have described being told that a steroid injection would resolve their pain, only to be left with new or worsened symptoms afterward, in some cases consistent with a new iatrogenic CSF leak caused by the injection itself. This backlash matters for the leak-care conversation specifically, because it means the same procedural volume that built Korea's pain anesthesiology expertise can, under a different incentive structure, produce the opposite outcome: a market that pushes procedures on patients who did not need them, rather than a market that builds deep technical competence in a procedure patients do need.
Some US and other Western procedural programs describe or market a "360-degree" patch, implying that circumferential coverage around the spinal cord requires multiple needle placements at different angles. Imaging studies of actual epidural patch distribution do not support that premise. A 2025 study of CT-guided epidural patching found that a single, adequately sized injection, with a mean volume of just over seven milliliters, spread a mean of 4.6 spinal levels craniocaudally, and imaging in that study explicitly documented spread circumferentially around the thecal sac, including ventrally, from a single dorsal injection site. Earlier MRI-based research found blood spreading over five to ten spinal segments from an 18 to 20 milliliter injection, again from a single site, with distribution influenced primarily by volume and injection level rather than by injecting at multiple points around the circumference. Blood, injected into a potential space, moves along the path of least resistance the way any fluid does; it does not need to be manually walked around the spinal cord by a series of separate needle passes. A "360-degree" framing that implies otherwise is not supported by how the epidural space actually behaves once blood is introduced into it, and multiplying needle passes to achieve a spread that adequate single-site volume already produces adds risk without a clear physiological rationale.
A related pattern some patients describe is a radiologist attributing a case to a suspected CSF-venous fistula, and recommending repeated rounds of digital subtraction or CT myelography to find it, even in cases where the leak site is already visible on imaging, for example at a site of known calcification, or where there is a documented prior dural puncture that already explains the presentation. Not every CSF leak involves a fistula, and current multidisciplinary guidance explicitly supports offering non-targeted, empiric epidural blood patching before escalating to invasive myelography in appropriate patients, precisely because those invasive studies carry real costs in radiation, procedural risk, and patient morbidity. A diagnostic pathway that defaults to the most invasive, highest-technology explanation before ruling out the simplest one inverts the order of operations that current guidance actually recommends, and it is worth asking, plainly, whether that default serves the patient or serves the procedural volume of the program performing it.
There is also a real methodological pattern worth naming in how some interventional studies in this space are structured. Published research on CSF-venous fistula embolization commonly uses objective, quantitative imaging measures, most often the Bern score derived from brain MRI, to determine which patients are included in a study and to demonstrate radiological improvement, while measuring treatment success itself using subjective, patient-reported scales like the Headache Impact Test or the Patient Global Impression of Change. That is not, on its own, evidence of bad faith; patient-reported outcome measures are standard practice across medicine because they capture something imaging cannot. But it does mean that the headline claim in a study, complete or significant symptom resolution in the large majority of treated patients, ultimately rests on what patients say about how they feel, and the patient community around spinal CSF leaks is, by its own account, not a neutral reporter of that experience. Patients frequently describe pressure, whether self-imposed or perceived, to report improvement after a difficult, expensive, and often geographically inconvenient procedure, both to avoid being dropped from further care and to avoid the discomfort of concluding that a widely recommended specialist's procedure did not work. That dynamic is a known and well-documented phenomenon in patient-reported outcomes research generally, sometimes described as social desirability bias, and it deserves more explicit acknowledgment in a field where the objective and subjective halves of an outcome measure can drift apart without anyone involved intending to mislead.
None of this argues that radiology should stop performing invasive diagnostic testing altogether, or that every patch belongs exclusively to anesthesiology. It argues for matching the skill to the task. Radiology's most valuable contribution is on the diagnostic side: refining non-invasive tools like the Bern score, expanding the use of MR myelography to identify and roughly localize leaks without a dural puncture, and reserving genuinely invasive digital subtraction myelography for the subset of complex or occult cases that actually require it, rather than as a default first step. A well-trained, high-volume pain anesthesiology workforce, of the kind Korea's reimbursement structure has inadvertently cultivated, is best positioned to deliver the actual epidural treatment for the great majority of ordinary cases, informed by that imaging when it exists, and supported by routine attention to a patient's coagulation status and nutritional history before the procedure ever begins. Neurosurgery remains essential for the smaller subset of cases that require structural repair. None of these specialties needs to own the entire pathway to contribute meaningfully to it, and the countries that get the best outcomes are likely to be the ones that stop treating this as a turf question and start treating it as a systems-design question, which is, in the end, the same lesson that every part of this article keeps returning to.