Article Text


Headache outcomes in children undergoing foramen magnum decompression for Chiari I malformation
  1. Saba Raza-Knight1,
  2. Kshitij Mankad2,
  3. Prab Prabhakar3,
  4. Dominic Thompson4
  1. 1Neurosciences Division, Addenbrooke's Hospital, Cambridge, UK
  2. 2Department of Radiology, Great Ormond Street Hospital, London, UK
  3. 3Department of Neurology, Great Ormond Street Hospital, London, UK
  4. 4Department of Neurosurgery, Great Ormond Street Hospital, London, UK
  1. Correspondence to Dominic Thompson, Department of Neurosurgery, Great Ormond Street Hospital, Great Ormond Street, London WC1N 3JH, UK; dominic.thompson{at}


Objective A common symptom of Chiari I malformation (CIM) is headache, which is diagnosed using non-validated criteria from the International Headache Society (IHS). CIM-associated headaches should resolve following neurosurgical treatment of the malformation by foramen magnum decompression (FMD). We aimed to validate the IHS criteria and determine (1) the efficacy of FMD in treating headache and (2) whether duraplasty confers an advantage over simple bony decompression in the treatment of this symptom.

Methods A retrospective review of CIM cases treated with FMD at Great Ormond Street Hospital from 1989 to 2014 was carried out. Clinical headache characteristics were compared against IHS criteria and correlated with outcome following FMD.

Results Headache was a presenting symptom in 57/102 (55.9%) of patients. Of these, 42/57 (73.7%) could be classified as Chiari I-type headache. Following FMD, 42/57 (73.7%) of all presenting headaches showed a sustained improvement (>3 months) compared with 32/39 (82.1%) of Chiari I-type headaches. Duraplasty led to a sustained improvement in headache in 32/38 (84.2%) cases compared with 9/16 (56.3%) cases treated with bone-only decompression.

Conclusions Our data suggest that 80% of headaches classified as Chiari I-type will show a sustained improvement following FMD, and, as such, the IHS criteria are clinically useful in evaluating symptoms attributable to CIM. For all headaches associated with CIM, duraplasty may confer a benefit in terms of long-term improvement, compared with bone-only decompression.

  • paediatric neurosurgery
  • posterior fossa
  • headache

Statistics from

What is already known on this topic?

  • Chiari I malformation is a relatively common radiological finding on sagittal brain MRI and may have associated neurological symptoms.

  • Headache is a common symptom of Chiari I malformation and can be diagnosed using criteria from the International Classification of Headache Disorders.

  • The symptoms of Chiari I malformation can be treated neurosurgically by foramen magnum decompression.

What this study adds?

  • In children, 80% of Chiari I-type headaches show a long-term improvement following foramen magnum decompression.

  • Most headaches associated with Chiari I malformation that do not fulfil the diagnosis of Chiari I type will still improve following foramen magnum decompression.

  • Duraplasty may confer a benefit in headache outcome following foramen magnum decompression in children.


The Austrian pathologist Hans Chiari described a series of hindbrain malformations in postmortem specimens from 1891 to 1896.1 Four variants of Chiari malformation (I–IV) are currently recognised, with additional types 0 and 1.5 proposed on the basis of case reports2 (see online supplementary figure S1). Chiari I malformation (CIM) is the most common type3 and can be considered a deformation of normal hindbrain in response to local mechanical or hydrodynamic factors rather than a true malformation; indeed, the terms cerebellar tonsillar ectopia or hindbrain hernia may be more appropriate.

The pathognomic feature of CIM is caudal displacement of the cerebellar tonsils relative to the basion–opisthion line on sagittal MRI (see online supplementary figure S2). However, defining CIM according to a strict physical measurement is misleading since the foramen magnum is not planar, but a saddle-shaped ellipsoid.

The literature is inconsistent in defining the magnitude of descent that constitutes a diagnosis of CIM, although 5 mm tonsillar descent below the foramen magnum is the most commonly quoted definition. The position of the cerebellar tonsils varies with age: cerebellar tonsillar ectopia (displacement of the cerebellar tonsils ≤5 mm) appears to be physiological in infants.4

Many instances of cerebellar tonsillar ectopia are asymptomatic5 and do not require treatment. Even in the presence of symptoms that might be reasonably attributed to a CIM, the degree of tonsillar descent correlates poorly with symptom severity. CIM may be an incidental finding on brain imaging and, as such, is an entity that invariably will be encountered by doctors and surgeons working in the paediatric setting.


The prevalence of CIM is difficult to estimate. A large, retrospective study of brain MRI scans in a US tertiary care centre has suggested that 0.8% demonstrated CIM.6 The majority (86%) of these cases were symptomatic, although it is unclear whether the indications for brain MRI were due to CIM-attributable symptoms.

Clinical presentation

The symptoms of CIM arise due to brainstem and spinal cord compression or interruption of cerebrospinal fluid (CSF) flow. Symptoms may also be attributable to syringomyelia, which is found in 50–80% of individuals with CIM.7 ,8 Other manifestations of CIM include ocular and vestibular disturbances, bulbar symptoms and sleep apnoea, all of which may fluctuate.3 ,8–10

It has been suggested that headache is the most common symptom of CIM, particularly in the paediatric population.3 ,8 ,11 ,12 Furthermore, CIM may be an organic cause of headache in children with medically refractory headaches. The International Headache Society (IHS) has issued detailed diagnostic criteria (table 1).13

Table 1

Summarised International Headache Society criteria for headache attributed to Chiari I malformation (CIM)


In the absence of formal guidelines, the management of CIM is based on clinical acumen. MRI studies of the brain and spinal cord are essential to confirm the degree of CIM and any associated intracranial or spinal cord abnormalities. Asymptomatic CIM is extremely common,14 and, since the natural history in the majority of such cases is favourable, most authorities would not recommend treatment on a prophylactic basis. At Great Ormond Street Hospital (GOSH), treatment is offered to patients with progressive syringomyelia or scoliosis or neurological symptoms that are attributable to the CIM (eg, Chiari-type headache). More recent cases have had input from a paediatric neurologist.

Foramen magnum decompression (FMD) surgery is the mainstay of treatment for symptomatic CIM. Controversy exists over whether a simple bony decompression (removing the foramen magnum and posterior arch of the atlas) is sufficient treatment compared with dural opening (usually with placement of a dural patch). Those who advocate duraplasty believe it to decrease crowding within the foramen magnum to a greater extent and improve CSF flow in the craniocervical region.10 The potential complications of surgery include infection, CSF leak, hydrocephalus, neurological damage and pseudomeningocoele formation at the operation site. These complications tend to be higher after procedures where the dura has been opened compared with bone-only decompressions. The degree of improvement following surgery largely depends on the nature of the presenting symptoms, the degree of CIM, and any coexistent pathologies, particularly in syndromic cases.

At GOSH, duraplasty is generally performed in children undergoing FMD for CIM. However, dura-sparing bone decompression is considered in children under 5 years of age or in patients with no preoperative evidence of syringomyelia who display an improvement in intraoperative ultrasound or neuro-monitoring following bony decompression alone. The frequency of performing FMD for CIM has increased in the past 15 years in our unit: only 15 FMD cases were performed before 2000, whereas, in the past decade, approximately 10 cases per year have been performed, excluding revision surgery. Reasons for this increase in operative frequency include: increasing access to MRI, enabling identification of CIM and syringomyelia; improvements in understanding the symptomatology of CIM; and increasing operative experience with FMD, facilitating its use.

In the paediatric population, the onset of symptoms associated with CIM occurs in children of primary school age, potentially disrupting their development and education. With increasing identification of CIM, sometimes on an incidental basis, it is becoming essential to define which clinical symptoms are likely to benefit from surgery. As such, CIM and its management are highly relevant to the general paediatrician.

Our aims in this study were twofold: (1) to correlate the latest IHS criteria for Chiari I-type headache with outcomes following FMD at a large paediatric neurosurgery centre; (2) to investigate whether FMD with duraplasty was more effective than bone-only decompression in improving headache outcome.


A retrospective review of all cases of first FMD for CIM was carried out with ethics approval from GOSH, London, UK (audit registration number 1129). A database of paediatric patients (≤16 years) from the neurosurgical team was used to identify cases from 1989 to 2014. This database is maintained by consultant neurosurgeons for the purposes of audit and quality control, with data collected prospectively on a weekly basis, ensuring it is a comprehensive and reliable source of cases. Patients with an associated congenital syndrome involving skeletal dysplasia were excluded.

Case notes were analysed to confirm the diagnosis of CIM and symptomatic indications for FMD. In patients presenting with headache, the headache characteristics were correlated with IHS criteria. All included cases had a minimum follow-up of 1 year and involved an operation before July 2014.

Short-term (≤3 months) and long-term (>3 months) outcomes were evaluated. Headache improvement was defined as a reduction in the frequency or intensity of the headaches after recovery from the surgical procedure, as ascertained from the patient's notes and follow-up clinic letters. Only two paediatric neurosurgeons were involved in the evaluation and surgery of this cohort, and both share the same views regarding patient selection and operative intervention. Each of the experienced neurosurgeons was cognisant of the need to differentiate the rarer Chiari I-type headache from more common types such as tension headache and migraine.

Surgical procedure

Surgery was carried out in the prone position. From 2000 onwards, intraoperative neurophysiological monitoring (somatosensory evoked potentials (SSEPs) and brainstem auditory evoked potentials (BSAEPs)) has been used during FMD in an attempt to identify whether electrophysiological variables could be used to inform the need for dural opening in addition to bony decompression. Since that time, intraoperative ultrasound has also been used. Bone-only decompression was reserved for young patients (under 5 years) or those patients (without syringomyelia) showing improvement in electrophysiological monitoring (SSEPs and BSAEPs) or ultrasound appearance after bone removal.

The posterior arch of the atlas was routinely removed, but C2 and its muscle attachments were preserved. In cases where the dura was opened, the arachnoid was also opened, but tonsil resection or coagulation was only performed when the tonsil extended behind the posterior arch of the axis. During the course of the study, two different dural grafts have been used, DuraGen (Integra Life Sciences Corporation) and Dura-Guard (Synovis). The graft was cut to the size of the dural defect and secured with a monofilament suture. Dural sealants were not used.

Statistics were analysed on SigmaStat for Windows V.3.5 (Systat Software). Diagrams were drawn on Illustrator CS4 (Adobe). Where relevant, data are summarised as mean±1SD.


A total of 127 consecutive patients with a first FMD for CIM were identified from the paediatric neurosurgery database. In the sample group, 15% of patients (19/127) had a congenital disorder in addition to CIM, including achondroplasia, craniosynostosis and inborn errors of metabolism (see online supplementary figure S3). After exclusion of these cases, and a further six with incomplete follow-up data, 102 patient records were analysed, encompassing 25 years (1989–2014). To our knowledge, this is the largest single-centre cohort of paediatric CIM cases specifically addressing the symptom of headache described in the literature to date.

For each patient, the presence of a CIM was confirmed using sagittal brain MRI and the case notes. Only a single case was asymptomatic: a multidisciplinary team agreed that this patient had slowly progressive syringomyelia with a high probability of developing symptoms, and recommended proceeding to FMD. After surgery, this patient made an uncomplicated recovery, with a reduction in syrinx size. In the remaining cases, the mean age of symptom onset was 8.7±4.1 years. Girls presented with a later symptom onset than boys (9.5±3.8 years vs 7.5±4.3 years, respectively; p=0.032, Mann-Whitney Rank Sum Test). Consistent with previous analyses of CIM in symptomatic cohorts, the female/male ratio was increased (1.7:1).

Indications for FMD

All symptomatic cases had more than one indication for surgery. The most common indications for FMD were syringomyelia, headache and scoliosis; these are the main indications cited in the paediatric literature.3 ,15 ,16 Additional indications were present in 60/102 (58.8%) cases, and included non-headache sensory symptoms, such as limb pain and paraesthesia (29.4%), bulbar symptoms (16.7%) and ataxia (12.7%) (table 2).

Table 2

Frequency of different clinical indications for treatment of Chiari I malformation in the Great Ormond Street Hospital cohort

Headaches at presentation

Using the latest IHS criteria,17 42/57 (73.7%) presenting headaches could be classified as Chiari I-type (figure 1; see online supplementary figure S4). Additional headache sub-types included migraine, hydrocephalus-associated, tension-type and the neck–tongue syndrome (unilateral neck pain, provoked by movement and accompanied by ipsilateral tongue paraesthesia). A number of headaches could not be clearly classified as Chiari I type (5/57, 8.8%), as they lacked classic features (precipitated by cough or other Valsalva-like manoeuvre; occipital or suboccipital location; lasting <5 min) and did not develop with a clear temporal relationship to the CIM.

Figure 1

Headache characteristics and outcomes in the Great Ormond Street Hospital Chiari I malformation cohort. NTS, neck–tongue syndrome. Long-term improvement was defined as >3 months.

Headache outcomes following FMD

Of the 42 Chiari I-type headaches, there was insufficient long-term follow-up for three cases. Nevertheless, the majority of these headaches showed a sustained (>3 months) improvement after FMD (32/39; 82.1%). Of the seven Chiari I-type headaches that did not improve following FMD, refractory headache had been the predominant symptom at presentation in three cases. Additional indications were present in the remaining four cases: two involved drop attacks, which improved following FMD, and separate cases presented with progressive syringomyelia and visual symptoms, which did not improve after surgery (figure 2).

Figure 2

Headache outcomes following foramen magnum decompression with duraplasty compared with a dura-sparing procedure (bone-only decompression). Long-term improvement was defined as >3 months.

A sustained improvement was observed after FMD in cases of migraine, hydrocephalus-associated headache and tension-type headache (4/9 cases; 44.4%). However, a similar number of cases showed no long-term improvement after surgery (5/9 cases; 55.6%). All of the unclassified presenting headaches showed a long-term improvement following FMD, as did a case of neck–tongue syndrome, which has been associated with CIM in the past.18


In six duraplasty cases, there were insufficient data to assess FMD outcomes. Of the remaining 96 cases, 70 (73.0%) were carried out with duraplasty and 26 (27.0%) were carried out using bone-only decompression.

FMD carried out with duraplasty was associated with a lower rate of reoperation than bone-only decompression (4/70 (5.7%) vs 14/26 (53.8%), respectively). Of the bone-only decompression cases that required further surgery, 13/14 (92.9%) required duraplasty; the remaining case required occipitocervical instrumented fixation to manage a retroflexed odontoid process.

Complications following FMD in our cohort were limited to hydrocephalus, CSF leak, superficial infection and deep infection. Overall complication rates for duraplasty cases (8/70, 11.4%) and bone-only decompression cases (3/26, 11.5%) were similar (table 3). Post-surgical infections occurred in 7/96 (7.3%) of this series, with the majority being superficial wound infections treated in the outpatient department with antibiotics. Non-infection complications (hydrocephalus and CSF leak) occurred in 4/96 (4.2%).

Table 3

Comparison of foramen magnum complications based on the use of duraplasty or bone-only decompression

Duraplasty and headache outcome

There were insufficient long-term follow-up data for three headache cases. Most patients presenting with headache (38/54, 70.4%) underwent FMD with duraplasty, of which 32/38 (84.2%) demonstrated a long-term improvement in their headache. Patients who underwent bone-only decompression demonstrated a long-term improvement in their headache in 9/16 cases (56.3%).

Of the cases that did not show any headache improvement after bone-only decompression, 5/7 (71.4%) underwent further surgery with duraplasty; the headache subsequently improved in two of these cases (figure 2).


This retrospective study evaluated the headache outcomes for a cohort of 102 children undergoing FMD for CIM at GOSH, London, UK, from 1989 to 2014.

Several important limitations exist in our study. The complexity of CIM—with its heterogeneous presentation, evolving clinical course and uncertain aetiology—complicates data analysis. Retrospective analysis over a relatively long time period restricted the amount of data that could be collected and the ability to directly compare outcomes between different groups. Furthermore, headache is a subjective symptom that is difficult to assess in a paediatric population. Fifteen children in our cohort were below the age of five, an age group in which it is particularly difficult to elicit an accurate history. Nevertheless, our cohort of patients spanned a 25-year period, such that each case had a minimum of 1 year of follow-up data; to our knowledge, this represents the largest single-centre cohort of paediatric CIM patients analysed in the literature specifically addressing headache. Data were analysed by a single reviewer (SR-K), and FMD operations were carried out by two different paediatric neurosurgeons (DT and Richard Hayward). As described in our methods, there was a standard preference for duraplasty in our series. There is thus some selection bias in this series that inevitably constrains the conclusions that can be drawn concerning the efficacy of bone-only decompression compared with duraplasty in the treatment of headache.

Over half of the patients reviewed presented with headache as a component of their symptom complex. Using the latest ICHD criteria,17 the majority of these headaches (42/57, 73.7%) could be classified as Chiari I type.

Most Chiari I-type headaches (32/39; 82.1%) showed a sustained improvement following FMD. It is possible that the Chiari I-type headaches that did not improve following FMD represented a more severe underlying malformation with a more fluctuant or aggressive natural history. However, it is difficult to radiologically assess the severity of CIM given the physiological variation of cerebellar tonsillar position with age. Three-quarters of all headaches showed a sustained improvement (>3 months) following FMD, regardless of subtype (42/54; 77.8%). We conclude that some headaches that do not have ‘classic’ CIM-associated features, such as an occipital location and cough exacerbation, may respond well to surgery provided they are not better accounted for by headache types such as migraine or tension-type. One notable clinical syndrome that responded to FMD is the neck–tongue syndrome,18 in which occipital and upper cervical pain are precipitated by neck movement, with associated ipsilateral tongue paraesthesia.

A proportion of headaches associated with CIM could be functional and may not improve following FMD. The mean age of symptom onset for children whose Chiari I-type headaches did not improve was 11.4±4.5, compared with an overall age of 8.7±4.1. Increasing age of headache onset may thus be associated with poorer surgical outcome, possibly because of a functional component. Of the seven Chiari I-type headaches that did not improve following FMD, the headache was the predominant symptom and main indication for surgery in three cases (figure 1); this represents 5% of all headaches found in association with CIM. It is rare to carry out FMD in patients whose only presenting symptom is headache, and, at GOSH, the three cases in question were refractory headaches. Patients presenting in this way are evaluated by the paediatric neurology service at our centre before consideration of surgery.

There is no universally accepted radiological definition of CIM and there is poor correlation between symptoms and radiological severity (in the extent of tonsillar descent); deciding whether headache might be attributable to the radiological finding of CIM, and might benefit from surgical decompression, can therefore be difficult. In this surgical series, headache was a part of the clinical presentation in 55.9% (57/102) of cases, which is only slightly less than the incidence quoted in contemporary published series, where headache is present in ∼70% of children having surgery for CIM. There is little doubt that, in appropriately selected children, FMD can provide effective and long-lasting resolution of headache: in our series, 77.8% (42/54) of headaches improved after surgery, increasing to 82.1% (32/39) if only headaches satisfying the diagnostic criteria for CIM-type headache are considered.

This series has an inherent selection bias in that surgical cases were selected from a larger cohort of children with CIM and headaches and who were evaluated in the paediatric neurosurgical setting but not deemed appropriate for FMD. What clinical factors then should govern the selection of cases for surgery? Patients should be considered surgical candidates if the headache syndrome is compatible with current IHS criteria for Chiari-type headache, the headache has been refractory to simple remedies, and the headache impacts significantly on lifestyle even if it is the only symptom. In all patients with CIM, particularly those with atypical headaches, additional symptoms and signs should be specifically sought that might indicate coexisting bulbar dysfunction, as these would strengthen the case for surgery. Such symptoms include ocular movement disorders, swallowing or speech disturbance or sleep disordered breathing. Additional investigations such as swallowing assessment or respiratory sleep studies should be considered, particularly in younger children who are more prone to having a bulbar component to their CIM symptoms.

Patients with severe and refractory non-Chiari-type headaches may benefit from multidisciplinary evaluation. From the results of our survey, patients and their families can be counselled that up to 20% of Chiari I-type headaches do not improve following FMD. What of these children in whom surgical treatment fails? It is important to recognise that cerebellar tonsillar ectopia (CIM) is rarely, if ever, a primary malformation and underlying predisposing factors should always be considered. For example, raised intracranial pressure associated with idiopathic intracranial hypertension or clinically occult craniosynostosis can result in hindbrain herniation. Bony anomalies or instability at the craniovertebral junction can also be associated with hindbrain herniation. These disorders may have been overlooked or were perhaps not even readily detectable at the time of the initial evaluation.

The majority of FMD operations in our cohort were carried out with duraplasty, which carried a fourfold lower rate of reoperation than bone-only decompression. However, we found that overall complication rates for hydrocephalus, CSF leak, and superficial and deep infection were almost identical in both groups. Our data suggest a benefit for long-term headache outcome with duraplasty, although headache is only a relative indication for operative intervention in CIM. FMD practice at GOSH continues to default to duraplasty, apart from in children under 5 and where intraoperative neuromonitoring has demonstrated adequate bone-only decompression. A recent retrospective study of 65 patients has suggested that there is no difference in long-term clinical outcome or in the improvement or deterioration of syringomyelia following FMD using either technique.19 While bone-only decompression remains of use in certain subgroups, such as young children, our data suggest that duraplasty confers a more predictable effect on symptoms. The benefits of avoiding reoperation must be weighed against the risks of a more invasive procedure when considering whether duraplasty is indicated.

CIM can manifest with a range of neurological symptoms, including refractory headache, which may be difficult to assign to conservative management or neurosurgical intervention. Our survey has found that 80% of headaches classified as Chiari I-type using the latest IHS criteria will improve following FMD, showing that they provide an effective diagnostic tool. Regardless of subtype, the majority of patients presenting with headache should observe a long-term improvement following surgery, particularly when duraplasty is used. However, prospective studies of headache and CIM will be necessary to inform clinical decision-making and management of this complex, heterogeneous disorder.


View Abstract


  • Contributors SR-K collected and analysed data, prepared figures and tables, and wrote the manuscript. KM reviewed MRI images. PP contributed to the study idea and design. DT contributed to the study idea and design, and revised the manuscript. SR-K and DT are guarantors for the overall content of the work.

  • Competing interests None declared.

  • Ethics approval Great Ormond Street Hospital, London, UK.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles