Postoperative wound infection is a well recognized complication of the operative treatment of adolescent idiopathic scoliosis. The prevalence of this complication varies, but it has been reported to be as high as 7.5 per cent (thirty-seven of 493 patients)2. The signs of postoperative infection (persistent fever, pain, swelling of the wound, erythema, and drainage) usually appear relatively early. Recently, however, Richards reported on ten patients who had a delayed infection at an average of twenty-five months after an operation for the treatment of scoliosis. Low-virulence organisms frequently grew on culture of specimens taken from deep within the wound, and four of the infections were thought to have originated at the time of the operation rather than as a result of hematogenous seeding of the area around the implant.
I present the case of a patient who had recurrent episodes of meningitis secondary to a late postoperative wound infection that developed eleven months after spinal arthrodesis and instrumentation for the treatment of adolescent idiopathic scoliosis.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†The Nemours Children's Clinic, P.O. Box 5720, Jacksonville, Florida 32247.
A fourteen-year-old girl was seen because of adolescent idiopathic scoliosis with a right thoracic curve of 54 degrees (Fig. 1). A review of previous radiographs revealed that the curve had been 47 degrees four months earlier.
Radiographs of the pelvis demonstrated incomplete excursion (ossification) of the iliac apophysis. Because the patient had growth remaining and had had recent progression of the curve, operative treatment was recommended. The patient was managed with posterior arthrodesis and instrumentation from the fourth thoracic to the second lumbar vertebra. The postoperative course was benign, and the patient was discharged four days after the operation.
Eleven months postoperatively, the patient began to have episodes of fever, headache, photophobia, nausea, and vomiting. Initially, these episodes occurred once a month and lasted from twelve to forty-eight hours. By fourteen months postoperatively, the episodes were occurring twice a week. The patient was seen by her primary-care physician and was managed with prochlorperazine and diclofenac. The results of a urinalysis were consistent with an infection of the urinary tract, and treatment with cephalexin was initiated.
Fifteen months postoperatively, the patient was seen in our emergency room because of headache, photophobia, nausea, vomiting, and fever (38.4 degrees Celsius). Physical examination revealed mild nuchal rigidity. Examination of the back demonstrated a well healed scar with no swelling, redness, increased warmth, or pain. A lumbar puncture was performed, and analysis of the cerebrospinal fluid revealed a white blood-cell count of 1440 per cubic millimeter (1.44 x 109 per liter), with 82 per cent polymorphonuclear leukocytes, 3 per cent lymphocytes, and 15 per cent monocytes. The level of glucose in the cerebrospinal fluid was twenty-four milligrams per deciliter (1.3 millimoles per liter), and the level of protein was 210 milligrams per deciliter (2.1 grams per liter). Cultures of blood and of cerebrospinal fluid were negative. The patient was thought to have meningitis. The negative cultures were attributed to the cephalexin that had been administered for treatment of the urinary tract infection. Ceftriaxone was administered intravenously, and the symptoms resolved. The patient was discharged on the fourth day, and intravenous administration of ceftriaxone (two grams every twenty-four hours) was continued at home for one week.
One week after discharge, the patient was readmitted because the symptoms had recurred. A lumbar puncture was performed, and analysis of the cerebrospinal fluid revealed a white blood-cell count of 2121 per cubic millimeter (2.121 x 109 per liter), with 86 per cent polymorphonuclear leukocytes, 1 per cent lymphocytes, and 13 per cent monocytes. The level of glucose in the cerebrospinal fluid was twenty-two milligrams per deciliter (1.2 millimoles per liter), and the level of protein was 260 milligrams per deciliter (2.6 grams per liter). A cell block of the cerebrospinal fluid was negative for neoplastic cells. A repeat culture of the cerebrospinal fluid was positive for coagulase-negative Staphylococcus. Radiographs of the spine showed possible dislodgment of the most caudad hook (Figs. 2-A and 2-B). Because the meningitis had recurred, the primary-care physician asked me to investigate the possibility of an infection in the area around the spinal implant. Ultrasonography of the spine showed a possible accumulation of fluid around the rods. However, aspiration of the area that had been identified as the site most likely to contain fluid yielded only a few drops of blood; therefore, the neurologist, the primary-care physician, and I concluded that the ultrasonographic findings were an artifact. Culture and gram-staining of the aspirate were negative. Bone-scanning showed symmetrical uptake in the operative site, and magnetic resonance imaging of the brain demonstrated normal findings. The symptoms resolved with intravenous administration of vancomycin (one gram every eight hours), trimethoprim (160 milligrams every six hours), and sulfamethoxazole (800 milligrams every six hours); the antibiotic regimen was continued for three weeks on an outpatient basis. Because the ultrasonographic findings were thought to have been an artifact and the symptoms resolved rapidly in response to the antibiotic regimen, we concluded that exploration of the wound was not necessary.
Three months later (eighteen months postoperatively), the symptoms recurred. A lumbar puncture was performed, and analysis of the cerebrospinal fluid revealed a white blood-cell count of 148 per cubic millimeter (0.148 x 109 per liter), with 80 per cent polymorphonuclear leukocytes, 16 per cent lymphocytes, and 4 per cent monocytes. A culture of the cerebrospinal fluid was negative. A myelogram revealed no epidural abscess or evidence of a dural fistula communicating with the spinal instrumentation. Because the symptoms had recurred, the patient was taken to the operating room for exploration of the site of the spinal arthrodesis. At the time of the operation, purulent-appearing fluid was found surrounding the instrumentation. In addition, a pseudarthrosis was noted at the thoracolumbar junction. The most caudad hook had pulled through the lamina of the second lumbar vertebra. The instrumentation was removed, and the wound was debrided. The pseudarthrosis was repaired with autogenous bone from the iliac crest. As the amount of motion at the site of the pseudarthrosis was slight, reinstrumentation of the spine was not performed. The wound was closed over suction drains. Gram-staining of fluid that had been obtained at the operation revealed no organisms, and culture of the fluid demonstrated no growth. Intravenous administration of vancomycin (one gram every twelve hours) and ceftriaxone (two grams every twelve hours) was continued for six weeks on an outpatient basis. The patient wore a thoracolumbosacral orthosis for six months postoperatively. One year after removal of the hardware, débridement of the wound, and initiation of the six-week course of intravenous antibiotic therapy, the patient reported that she had had no additional episodes of meningitis and no back pain. Radiographs of the spine that were made at that time showed no loss of correction of the curve (Fig. 3).
The persistent fever, back pain, and wound drainage that frequently are associated with a postoperative infection of the spine were not observed in my patient. In addition, there was no evidence of wound infection either on imaging studies or on aspiration of the wound. Despite the lack of evidence of a wound infection, exploration revealed purulent material that was consistent with infection around the hardware. Moreover, after removal of the hardware, débridement of the wound, and initiation of a six-week course of intravenous antibiotic therapy, the patient had no additional evidence of infection or of recurrent meningitis.
A wound infection after spinal arthrodesis and instrumentation for the treatment of a deformity is not rare, but a thorough search of the literature revealed no other reports on the spread of such an infection to the meninges or on meningeal irritation secondary to such an infection. Recurrent bacterial meningitis in children is uncommon. Lieb et al., in a recent review of the records of patients who had been managed at a large children's hospital in Europe during a fifteen-year period, identified twenty-five children who had had a total of eighty-four episodes (range, two to thirteen episodes) of bacterial meningitis. A pathogen was identified in association with sixty-five (77 per cent) of the episodes. In all instances, an anatomical lesion eventually was found. Operative correction of the lesion was necessary for resolution of the meningitis in twenty-four of the twenty-five patients.
On the basis of the case of my patient, I believe that exploration of the wound is warranted when meningitis that develops after posterior spinal arthrodesis and instrumentation is not readily eradicated by the usual medical treatment. Exploration should be considered even when evidence of a wound infection cannot be found.