Correlation between shunt series and scoliosis radiographs in children with myelomeningoceles

J Neurosurg Spine. 2012 Nov;17(5):410-4. doi: 10.3171/2012.8.SPINE12766. Epub 2012 Sep 14.


Object: Over 85% of patients with myelomeningoceles require placement of a ventriculoperitoneal shunt for hydrocephalus, and between 25% and 85% of these patients develop scoliosis. Although most patients undergo repeated shunt series radiography to evaluate for device malfunction, scoliosis radiographs are less consistently obtained. The authors sought to determine if a correlation exists between these 2 radiographic techniques for a given patient, as shunt series are obtained with the patient supine, whereas scoliosis radiographs are acquired with the patient standing upright. The authors also endeavored to study if shunt series radiographs can reliably detect significant scoliosis.

Methods: The authors retrospectively reviewed a single institution's series of 593 patients with myelomeningoceles and identified all patients in whom a shunt series and scoliosis radiographs were obtained within a 6-month period. They reviewed the medical records and radiographs of these patients for demographic and radiographic parameters. They then applied a linear regression model and determined shunt series curve cutoffs to detect scoliotic curves greater than 20° and 50°.

Results: Of the 593 patients identified, 116 did not have radiographs available for interpretation. Of the remaining 477 patients, 201 had radiographic evidence of scoliosis (42%), and 66 had both a shunt series and a scoliosis radiographs acquired within a 6-month interval. In 4 patients, both end vertebrae of the scoliotic curve could not be visualized on a single radiograph. The mean age of the remaining cohort was 10.6 ± 5.2 years and the mean curve magnitude was 58° ± 37°. Using identical end vertebrae, the mean shunt series curve magnitude was 49° ± 35°. The mean interval between both radiographs was 2.3 ± 3.3 months. The regression model showed a strong linear association between shunt series and scoliosis series curves. A curve greater than 19° on shunt series radiographs would detect significant curves of greater than 20° on scoliosis series with 91% sensitivity and 78% specificity. A shunt series curve greater than 37° had 100% sensitivity and 93% specificity in identifying significant scoliotic curves greater than 50°.

Conclusions: Although shunt series radiographs may not precisely depict scoliotic curve magnitude because the impact of gravity is negated, they may be useful in helping to confirm clinical suspicion of scoliosis. The authors' results suggest a strong correlation between both types of radiographs.

MeSH terms

  • Cerebrospinal Fluid Shunts*
  • Child
  • Female
  • Humans
  • Linear Models
  • Male
  • Meningomyelocele / complications
  • Meningomyelocele / diagnostic imaging*
  • Meningomyelocele / surgery*
  • Patient Positioning
  • ROC Curve
  • Radiography
  • Retrospective Studies
  • Scoliosis / complications
  • Scoliosis / diagnostic imaging*
  • Scoliosis / surgery*
  • Sensitivity and Specificity