Background. Discitis and vertebral osteomyelitis are uncommon entities, and diagnosis often is confounded by their similar clinical presentation, because characteristic radiographic findings are not evident until late in the course of illness.
Objective. To compare the age distribution, clinical manifestations, and radiographic findings, especially magnetic resonance imaging (MRI), in children with discitis or vertebral osteomyelitis.
Methods. A retrospective review of 57 children with a discharge diagnosis of discitis or vertebral osteomyelitis hospitalized from January 1980 through December 1998.
Results. Fifty patients met inclusion criteria: 36 with discitis and 14 with vertebral osteomyelitis. The mean age at presentation was younger for children with discitis than for those with vertebral osteomyelitis (2.8 vs 7.5 years of age) and the duration of symptoms longer for children with vertebral osteomyelitis than for those with discitis (33 vs 22 days). The initial symptom for both groups of children was refusal to walk, limp, or back pain, but children with osteomyelitis more often were febrile (79% vs 28%) and ill-appearing than those with discitis. Thirty-three patients with discitis had radiographs of the spine; 25 (76%) had abnormalities that were diagnostic. Ten discitis patients had MRI; 9 (90%) had abnormalities consistent with this diagnosis. Thirteen children with vertebral osteomyelitis had radiographs of the spine, but in only 7 (54%) were these abnormal. However, 11 had MRIs, and in each the diagnosis of vertebral osteomyelitis was established.
Conclusion. This comparative study suggests that age and clinical presentation distinguish most patients with discitis from those with vertebral osteomyelitis. Although radiographs of the spine usually are sufficient to establish the diagnosis of discitis, MRI is the diagnostic study of choice for pediatric patients with suspected vertebral osteomyelitis.
Intervertebral disk infection (discitis) and vertebral osteomyelitis are uncommon entities with distinct epidemiologic, clinical, and radiographic features. Discitis generally afflicts children <5 years of age who have no or low grade fever. It occurs nearly exclusively in the lumbar region, causing refusal to walk or progressive limp, and 2 to 3 weeks into the illness, radiographs demonstrate narrowing of the disk space with variable degrees of destruction of adjacent vertebral end plates.1,,2 Vertebral osteomyelitis typically occurs in older children who are febrile and complain of back pain in the lumbar, thoracic, or cervical regions.3 Radiographs first demonstrate localized rarefaction of 1 vertebral body, and later, destruction of bone, usually the anterior portion, and osteophytic bridging. Early in the course of these disease processes, the differentiation between discitis and vertebral osteomyelitis often is difficult. Nuclear bone scans, computed tomography (CT) scans, and laboratory studies typically provide information that is nonspecific, and appropriate patient management demands a high degree of discernment by the pediatrician. Magnetic resonance imaging (MRI) has become widely available. This technique can not only unveil the anatomic spectrum of these spinal infections but also provides sufficient detail to guide the need for invasive diagnostic procedures.
To our knowledge, a comparative review of children with discitis or vertebral osteomyelitis has not been reported previously. We conducted a retrospective medical record review of children hospitalized between 1980 and 1998 with 1 of these diagnoses. In describing the clinical presentation, diagnostic evaluation, and treatment of these patients, we hoped to provide data that would assist the pediatrician with prompt and accurate diagnosis. We were especially interested in the role of diagnostic imaging in establishing a specific diagnosis.
This retrospective study included children hospitalized between January 1980 and December 1998 at Texas Children's Hospital in Houston. Forty-four children with discitis and 13 with vertebral osteomyelitis were identified by discharge diagnoses through the hospital Medical Records Department and the Infectious Disease Service inpatient consultation database. These 57 children constitute the study population.
A patient was considered to have discitis if there were: 1) clinical findings compatible with this diagnosis (refusal to crawl, sit, or walk; limp or back pain) and 2) an abnormal radiographic or nuclear bone scan study. The latter included a radiograph demonstrating narrowing of intervertebral space, a technetium-99m bone scan with increase tracer uptake, or an abnormal CT or MRI study of the spine demonstrating vertebral disk space involvement with a normal appearance of the nonadjacent vertebrae. Nuclear bone scan was allowed as 1 of the diagnostic criterion because during the initial period of this study more sensitive imaging techniques (ie, CT or MRI) were not available.
A patient was considered to have vertebral osteomyelitis if 2 or more of the following criteria were present: 1) clinical features compatible with the diagnosis (refusal to crawl, sit, or walk; limp or back pain; fever); 2) microbiologic evidence of infection (isolation of a pathogen from blood or biopsy specimen) or serology diagnostic forBartonella henselae infection; 3) histopathologic evidence of osteomyelitis by biopsy; and 4) abnormal radiographic studies of the spine (bone destruction with or without disk narrowing by radiograph or typical vertebral bone involvement by CT or MRI).
The following data, when available, were collected from the medical record of each patient who met the diagnostic criteria: date of birth, age, race, dates of admission and discharge, history of fever, pain, limping, trauma, treatment before admission, and duration of symptoms. The presence of point tenderness and other pertinent physical findings also were recorded. Laboratory studies analyzed were peripheral white blood cell (WBC) count and differential, erythrocyte sedimentation rate (ESR), and microbiologic results. Histopathology of biopsy or needle aspiration specimens, when available, also were recorded. Reports of radiographs, technetium-99m bone scan, MRI, and CT studies were reviewed, and available MRI or CT scans were blindly reviewed by 1 of us (C.L.C.) to confirm or dismiss the diagnosis of discitis or vertebral osteomyelitis. Cases then were reclassified, independently from the hospital discharge diagnosis, using our inclusion criteria. The type and duration of therapy were noted.
Forty-four children had a discharge diagnosis of discitis. After review, 3 were reclassified as vertebral osteomyelitis and 5 did not meet our diagnostic criteria; thus, 36 children met criteria for the diagnosis of discitis. Of the 13 children with a discharge diagnosis of vertebral osteomyelitis, 2 were excluded because they did not meet diagnostic criteria, but 3 patients from the discitis group were added; thus, 14 children met criteria for vertebral osteomyelitis. Tables 1 and 2summarize the demographic, clinical features, and laboratory characteristics of the 36 children with discitis and the 14 with vertebral osteomyelitis. Three children with osteomyelitis have been reported previously.3
The mean age at presentation (Fig 1) was significantly younger for children with discitis (2.8 years of age; median: 1.8; range: 7 months to 16 years) than that of those with vertebral osteomyelitis (7.5 years of age; median: 6.0; range: 2–13 years; P = .002). Twenty-seven (75%) children at presentation with discitis were 2½ years of age or younger, whereas only 2 (14%) of those with vertebral osteomyelitis were this young. None of the children in the discitis group had an underlying illness. One child with vertebral osteomyelitis had Cushing's disease and another had juvenile rheumatoid arthritis. There was a history of preceding trauma in 7 patients with discitis, but 5 of these incidents were not related to the anatomic location, and all preceded the presenting symptoms for >1 month. Only 2 patients with vertebral osteomyelitis had a history of preceding trauma.
The predominant presenting symptom in children with discitis was dictated by the child's verbal skills (Table 1). Children <3 years of age presented with refusal to walk or limp, whereas older patients often complained of back pain. Eight (57%) children with vertebral osteomyelitis presented with back pain (Table 2), and 1 each complained of neck, shoulder, rib, or abdominal pain. One child with vertebral osteomyelitis had incontinence as a presenting complaint and another had a history of prolonged fever. The site of involvement frequently correlated with the clinical presentation. Seventy-eight percent of children with discitis had involvement of the lumbar or lumbosacral area. Among children with vertebral osteomyelitis, the lumbosacral area was affected in 9 (64%), the thoracolumbar in 4 (29%), and the cervical in 1. No patient had multifocal involvement of the spine or other bones.
A minority (28%) of children in the discitis group had a temperature of >100.3°F by history or at presentation, and in only 3 did this exceed 101°F. None of these febrile patients were ill-appearing. Children with vertebral osteomyelitis were significantly more likely to have a history of fever than those with discitis (79% vs 28%;P = .02), and in each of these febrile patients, the temperature exceeded 102°F. The mean duration of symptoms was significantly longer for children with vertebral osteomyelitis than for those with discitis (33 vs 22 days; P = .04).
The mean WBC count at admission in children with discitis was 10 900/mm3 (range: 7500–25 400) (results were not available for 1 patient), and the mean ESR was 42 mm/hour (range: 4–85; results were not recorded for 2 patients). Similarly, in patients with vertebral osteomyelitis, the mean WBC count was 12 600/mm3 (range: 4300–23 000) and the mean ESR was 45.6 mm/hour (range: 9–100). Although minimally elevated in nearly all patients, the WBC and ESR provided nonspecific information.
A single blood culture was sterile in 28 (88%) of the 32 patients with discitis from whom this specimen was obtained. Among patients who had positive blood culture results, coagulase-negativeStaphylococcus was isolated from 2, and α-hemolyticStreptococcus (not S pneumoniae) and Gram-positive rods (not further identified) from 1 each. These organisms were interpreted to be contaminants by the physicians caring for these patients. Among the children with vertebral osteomyelitis, each had a blood culture obtained, and 8 (57%) of these specimens yielded organisms. Staphylococcus aureus was isolated from 5 patients, and Staphylococcus epidermidis andSalmonella group C from 1 patient each. Two children with vertebral osteomyelitis had serologic findings that established a diagnosis B henselae infection. One of these children presented with neck pain, unilateral cervical lymph node enlargement, and limp. Both of these patients had a history of cat exposure.
A biopsy or needle aspiration of the disk space was obtained from 8 (22%) children with discitis. Aerobic, anaerobic, fungal, and mycobacterial cultures of these specimens failed to grow in each case. Eight (57%) children with vertebral osteomyelitis also underwent open biopsy or percutaneous needle aspiration. In half potential pathogens were isolated and included S aureus in 2 patients, andSalmonella group C and Propionibacterium acnes in 1 patient each. Two of these 4 patients had the same organism isolated from blood and from the biopsy specimen.
Thirty-three (92%) patients with discitis had spine radiographs during their initial evaluation; in 25 (76%), these were abnormal. The most frequent finding was decreased height of the disk space and erosion of adjacent vertebral endplates. Thirty (83%) children also underwent technetium-99m bone scanning. Scans were abnormal in 27 (90%), demonstrating increased marker uptake of the affected area. Ten patients with discitis had MRI, and in 9, abnormalities were compatible with this diagnosis. One child had normal MRI after 4 weeks of illness. Thirteen (93%) patients with vertebral osteomyelitis had radiographs of the spine and in 7 (54%) these were normal. Eleven patients had MRI, and in each the diagnosis of vertebral osteomyelitis was established.
The usefulness of MRI to demonstrate vertebral bone involvement as well as to define inflammatory lesions in adjacent structures is illustrated in Figs 2 and3. Figure 2 (patient 3, Table 2) demonstrates increased signal intensity within the fourth lumbar vertebral body in the sagittal T2-weighted image, with bulging of the anterior margin of bone. The adjacent disk spaces are preserved. In contrast, Fig 3 (patient 8, Table 2) reveals not only focal osteomyelitis of the thoracic spine with vertebral body destruction, but also discitis, with a collapsed disk space and a paraspinous phlegmon and/or small collection of hemorrhage. This MRI illustrates the progressive involvement that can occur in some children with vertebral osteomyelitis.
Therapy and Outcome
Eighteen (50%) children with discitis received antimicrobials, 9 were treated with nonsteroidal antiinflammatory agents, and the remainder received no drug therapy. Some of these patients also were treated with body cast placement. Of the 18 children treated with antimicrobials, in 13 (72%) these were administered initially by the intravenous and later by the oral route. In 2 and 3 patients, respectively, only intravenous or oral antimicrobials were prescribed. A semisynthetic penicillin was given to each patient requiring intravenous therapy and a dicloxacillin or a first generation cephalosporin for those given oral therapy. The duration ranged from 10 days to 3 to 4 weeks. In contrast, each of the 14 children with vertebral osteomyelitis was treated with antimicrobials. Seven (50%) received intravenous followed by oral therapy, 5 (36%) were treated exclusively with intravenous, and 2 exclusively with oral antimicrobials. All but 1 child received semisynthetic penicillin (usually nafcillin). In 1 patient who had a methicillin-resistantS aureus infection, intravenous vancomycin was administered.
Infants and children with a diagnosis of discitis were divided into 2 groups based on year of hospital admission. From 1980 through 1989, there were 25 patients, and from 1990 through 1998, there were 11. Children in the earlier period tended to be younger (mean age: 1.9 years) than the 1990 to 1998 group (mean age: 4.8 years). Also, children in the earlier group had symptoms for a shorter duration (mean: 28 days), compared with the 1990 through 1998 group (mean: 46 days). Patients hospitalized after 1990 were more likely to have had MRI or CT scans (73%) than those before 1990 (12%). Furthermore, children who had MRI or CT scans were more likely to undergo needle aspiration or biopsy than those without these diagnostic imaging studies. Finally, 10 (91%) of the 11 children diagnosed after 1990 were treated with antimicrobial agents. Only 9 (36%) of those diagnosed before 1990 were treated with antimicrobials; the remainder received antiinflammatory agents or no drug treatment at all.
Discitis and vertebral osteomyelitis remain uncommon entities among pediatric patients, although their recognition may be increasing.4 Cushing1 estimated that the incidence of discitis in children was 1 to 2 cases per year in a hospital that evaluated 32 500 patients per year (outpatient and inpatients visits). Vertebral osteomyelitis in pediatric patients accounts for only 1% to 2% of all children with osteomyelitis.5,,6 These entities seem to be ends of the same disease spectrum, but differentiation is necessary for initiation of appropriate therapy for discitis or for vertebral osteomyelitis. However, children with either diagnosis have a history of limp, back pain, or refusal to walk, and laboratory findings that are nonspecific.7,,8 Characteristically, children with discitis are younger than those with vertebral osteomyelitis. In contrast, children with vertebral osteomyelitis are more likely to be febrile and ill-appearing at the time of presentation than those with discitis. Our series confirms these distinctive age patterns and clinical features. Initial radiographs of the spine, if abnormal, may help to distinguish between the 2 entities, but because these findings often appear late in the course of illness, additional studies are required to establish a diagnosis.
Series describing discitis and vertebral osteomyelitis report a biphasic age distribution. Discitis has a higher incidence early in childhood and a second subtler peak during adolescence.1Vertebral osteomyelitis, in contrast, has a peak during adolescence and a more prominent one in patients older than 50 years of age.4 A distinct age distribution was observed among our patients. Vertebral osteomyelitis was infrequently encountered among young children (<3 years of age), while discitis was uncommon in children 8 or more years of age. Children with intermediate ages may be afflicted with either of these spinal infections, and they should be carefully evaluated. Fever, although present in both groups of patients, is much more frequent and usually higher and of longer duration in children with vertebral osteomyelitis, and in the case series of Correa et al,3 it was the most frequent presenting symptom. In contrast, children <3 years of age who are not febrile or ill-appearing and who have typical radiographic features of discitis need no further evaluation. Laboratory studies, such as CBC and ESR, provide nonspecific information. Blood cultures and more invasive procedures, including biopsy, should be strongly considered in patients with suspected vertebral osteomyelitis, where definition of a causal agent is particularly important in selection of the appropriate antimicrobial therapy and in defining the duration of therapy.9 Vertebral osteomyelitis, although infrequently encountered, can be an atypical manifestation of B henselaeinfection. Availability of specific serologic assays allowed the confirmation of this cat-scratch disease organism as the cause in 2 of our 14 patients. In patients with cat exposure, this cause should be considered,10 and serologic testing should be performed in a reference laboratory, such as the Centers for Disease Control and Prevention (Atlanta, GA).
The pathophysiology of discitis and vertebral osteomyelitis remains poorly understood, but differences in blood supply to the intervertebral disk by age has been implicated for both. Discitis is thought to occur due to the presence in childhood of vascular channels in the cartilaginous region of the disk space that disappear later in life and because of abundant intraosseous arterial anastomoses in childhood that promote clearance of microorganisms or entrapped emboli.1,,11 Most authors propose an infectious cause for discitis.4,11–13 However, no conclusive data are available, and the fact that most cultures are sterile and patients recover without antimicrobial therapy mitigates against this hypothesis. Some investigators propose that the duration of symptoms may affect culture results, especially those from the biopsy of a disk space.1,,11 Vertebral osteomyelitis, in contrast, is thought to occur when microorganisms lodge in the low-flow, end-organ vasculature adjacent to the subchondrial plate region. Recently, some cases of vertebral osteomyelitis in adults have been associated with intravenous drug use, and the association with urinary tract infections is well known.4 These associated risk factors have not been noted in pediatric patients. Antecedent trauma has been associated with vertebral osteomyelitis as well as discitis,3,,13,14but this association was not found among the majority of our patients.
Before the advent of more sensitive radiographic techniques, the radiograph was considered to be the definitive study in establishing a specific diagnosis of discitis.15 In our series, initial roentgenograms were abnormal in 76% of children with discitis, but in only 46% of those with vertebral osteomyelitis. Radiographic changes sometimes are evident only late in the disease process, making this diagnostic method less reliable early in the illness. Spine radiographs should be obtained in all children with suspected discitis; if these demonstrate characteristic findings of discitis, the diagnosis is established. Gallium and technetium nuclear scans have been used, but the pattern of uptake is nonspecific and neither study can reliably differentiate between discitis and vertebral osteomyelitis.16,,17 The CT scan was not useful in providing a specific diagnosis in our patients. With the availability of MRI, nuclear bone-scanning methods should be abandoned and should be obtained only in very young patients when localization of symptoms to the spine, based solely on physical examination, may not be possible.18 The availability, sensitivity, and specificity of the MRI make it the radiographic imaging study of choice to evaluate children with possible vertebral osteomyelitis or those in whom radiographs are normal.18,,19 According to Donovan et al20 the optimal MRI technique for evaluation of spinal infection is thin-section surface-coil imaging with T1-weighted images in sagital and axial views and sagital T2-weighted images. Edema and purulent material in the marrow or disk space will appear as a dark signal on T1-weighted images and as a bright signal on T2-weighted images.18 A contrast-enhanced MRI should be considered for equivocal results. MRI has been found to have a sensitivity of 96% and a specificity of 93% for the diagnosis of vertebral osteomyelitis, making it substantially more sensitive and specific for this diagnosis than nuclear bone scan or routine roentograms.21 MRI is a rapid, accurate, and noninvasive method that can distinguish disk inflammation from pyogenic bone involvement, and it also provides anatomic information regarding involvement of surrounding tissue (Fig 3).21,,22
Based on our findings and those of others, we propose that the child who presents with a possible diagnosis of discitis or vertebral osteomyelitis should have, as part of the evaluation, at least 1 blood culture, a complete blood count and differential, and radiographs of the spine. Nuclear bone scanning may be helpful in very young children in whom localization of the inflammatory process is uncertain by physical examination. If the roentgenograms are not characteristic of discitis, MRI of the spine should be performed. Once the diagnosis is established, antimicrobial therapy is always indicated for the child with vertebral osteomyelitis, and the drug and length of therapy should be guided by the causal agent. Use of antimicrobial therapy in children with discitis remains controversial because there are no controlled studies to guide specific therapy, route of administration, or duration of treatment.
- Received June 23, 1999.
- Accepted November 18, 1999.
Reprint requests to (C.J.B.) Section of Infectious Diseases, Department of Pediatrics, Baylor College of Medicine, 1 Baylor Plaza, Room 302A, Houston, TX 77030. E-mail:
- CT =
- computed tomography •
- MRI =
- magnetic resonance imaging •
- WBC =
- white blood cell •
- ESR =
- erythrocyte sedimentation rate
- Cushing AH
- Wenger DR,
- Bobechko WP,
- Gilday DL
- ↵Le CT. Salmonella vertebral osteomyelitis: a case report with literature review. Am J Dis Child. 1982;136:722–724
- Santos EM,
- Sapico FL
- Bolivar R,
- Kohl S,
- Pickering LK
- Robson JMB,
- Harte GJ,
- Osborne DRS,
- McCormarck JG
- Kern RZ,
- Houpt JB
- Fisher GW,
- Popich GA,
- Sullivan DE,
- Mayfield G,
- Mazat BA,
- Patterson PH
- Adatepe MH,
- Powell OM,
- Isaacs GH,
- Nichols K,
- Cefola R
- Copyright © 2000 American Academy of Pediatrics