Incidence of subsequent vertebral fracture after kyphoplasty

A retrospective review of charts and radiographs of all consecutive patients who underwent kyphoplasty at the authors' center from the initial procedure in September 2000 to the end of the collection period in July 2002.

The best available natural history data would suggest that after experiencing an osteoporotic vertebral compression fracture, patients have a 19% incidence of subsequent fracture in the following year when no surgical intervention is performed. When kyphoplasty is performed, there are conflicting data regarding the incidence of subsequent fracture, ranging anywhere from 3 to 29%. These fractures occur at adjacent levels between 30 and 90% of the time, with no clear explanation for the wide variation in the results of three different studies. There are biomechanical data to suggest that injection of cement does increase the stiffness of the treated vertebra and that this increases strain on adjacent vertebrae, especially in forward bending.

A database was created containing patient age, gender, height, weight, medication history, comorbidities, fracture levels, and pain level before and after surgery. Subsequent fractures were confirmed with radiographs and MRI. Statistical analysis was performed.

Thirty-eight patients (10 men and 28 women) were treated for 47 levels initially. L1 and L2 were the most common level of fracture managed initially. The gender, smoking and medication history, location of fracture, and number of fracture levels of the patients did not correlate with the risk of subsequent fracture. Over the follow-up period (average, 8 months), 10 patients sustained 17 subsequent fractures. Eight patients sustained fractures in the first 2 months after the index procedure, all with at least one fracture at an adjacent level. Of the 17 subsequent fractures, there were nine at the adjacent-above levels, four at adjacent-below levels, and four at remote levels. The remote fractures occurred at significantly greater time intervals after the index procedure (P < 0.001).

This study demonstrated a higher rate of subsequent fracture after kyphoplasty compared with natural history data for untreated fractures. Most of these occurred at an adjacent level within 2 months of the index procedure. After this 2-month period, there were only occasional subsequent fractures, which occurred at remote levels. This confirms biomechanical studies showing that cement augmentation places additional stress on adjacent levels. Patients with an increase in back pain after kyphoplasty should be evaluated carefully for subsequent adjacent fractures, especially during the first 2 months after the index procedure.