Joint Bone Spine 89 (2022) 105333

Contents lists available at ScienceDirect

Joint Bone Spine

journal homepage: www.elsevier.com

Original article

Spine immobilization and neurological outcome in vertebral

osteomyelitis SPONDIMMO, a prospective multicentric cohort夽
Adrien Le Pluart a,∗ , Guillaume Coiffier b,m , Christelle Darrieutort-Lafitte a , Sophie Godot c ,
Sebastien Ottaviani d , Julien Henry e , Julia Brochard f , Grégoire Cormier g ,
Marion Couderc h , Emmanuel Hoppe i , Denis Mulleman j , Lydie Khatchatourian k ,
Aurélie Le Thuaut l , Benoit Le Goff a , Géraldine Bart b
a
Department of Rheumatology, CHU Nantes, Nantes, France
b
Department of Rheumatology, CHU Rennes, Rennes, France
c
Department of Rheumatology, AP-HP DCSS, Paris, France
d
Department of Rheumatology, AP-HP Bichat, Paris, France
e
Department of Rheumatology, AP-HP Kremlin-Bicêtre, Paris, France
f
Department of Infectious Diseases, CH Saint-Nazaire, Saint-Nazaire, France
g
Department of Rheumatology, CHD Vendée, La Roche-sur-Yon, France
h
Department of Rheumatology, CHU Clermont-Ferrand, Clermont-Ferrand, France
i
Department of Rheumatology, CHU Angers, Angers, France
j
Department of Rheumatology, CHU Tours, Tours, France
k
Department of internal medicine and infectious disease, CH Cornouaille, Quimper, France
l
Direction of research, Methodology and Biostatistics platform, CHU Nantes, Nantes, France
m
Department of Rheumatology, GHT Rance-Emeraude, CH Dinan/Saint-Malo, France

a r t i c l e i n f o a b s t r a c t

Article history: Objective: The aim of our study was to describe spine immobilization in a multicentric cohort of vertebral
Accepted 13 December 2021 osteomyelitis (VO), and evaluate its association with neurological complications during follow-up.
Available online 22 December 2021 Methods: We prospectively included patients from 2016 to 2019 in 11 centers. Immobilization, imaging,
and neurological findings were specifically analyzed during a 6-month follow-up period.
Keywords: Results: 250 patients were included, mostly men (67.2%, n = 168). Mean age was 66.7 ± 15 years. Diagno-
Vertebral osteomyelitis sis delay was 25 days. The lumbo-sacral spine was most frequently involved (56.4%). At diagnosis, 25.6%
Spondylodiscitis
patients (n = 64) had minor neurological signs and 9.2% (n = 23) had major ones. Rigid bracing was pre-
Spinal infection
Spine immobilization
scribed for 63.5% (n = 162) of patients, for a median of 6 weeks, with variability between centers (P < 0.001).
The presence of epidural inflammation and abscess on imaging was associated with higher rates of rigid
bracing prescription (OR 2.33, P = 0.01). Frailness and endocarditis were negatively associated with rigid
bracing prescription (OR 0.65, P < 0.01, and OR 0.42, P < 0.05, respectively). During follow up, new minor
or major neurological complications occurred in respectively 9.2% (n = 23) and 6.8% (n = 17) of patients,
with similar distribution between immobilized and non-immobilized patients.
Conclusion: Spine immobilization prescription during VO remains heterogeneous and seems associated
inflammatory lesions on imaging but negatively associated with frailness and presence of endocarditis.
Neurological complications can occur despite rigid bracing. Our data suggest that in absence of any
factor associated with neurological complication spine bracing might not be systematically indicated.
We suggest that spine immobilization should be discussed for each patient after carefully evaluating
their clinical signs and imaging findings.
© 2022 Société française de rhumatologie. Published by Elsevier Masson SAS. All rights reserved.

1. Introduction

夽 Investigation performed at Department of Rheumatology, CHU Nantes, Nantes,

Vertebral osteomyelitis (VO) is a spine infection involving
France.
intervertebral discs and adjacent vertebral endplates. It affects
∗ Corresponding author. 2.4/100,000 inhabitants per year in France (0.5 to 10/100,000 in
E-mail address: adrien.LEPLUART@chu-nantes.fr (A. Le Pluart). Europe) [1] and 5.4/100,000 in the United States in 2013, with

https://doi.org/10.1016/j.jbspin.2021.105333
1297-319X/© 2022 Société française de rhumatologie. Published by Elsevier Masson SAS. All rights reserved.
A. Le Pluart, G. Coiffier, C. Darrieutort-Lafitte et al. Joint Bone Spine 89 (2022) 105333

a mortality rate of 2.2% [2]. The incidence of VO has increased (e.g. isolated spondylitis, facet joint infection, or isolated epidu-
in recent decades. VO can be associated with neurological com- ral infection). Patients were excluded if an alternative diagnosis of
plications. In a previous study, we showed that they occurred in spondylodiscitis was found (e.g. crystal-induced spondylodiscitis).
40% of cases [3]. In severe cases, surgical intervention is needed:
e.g. progressive neurological deficits, progressive deformity, spinal 2.3. Data collection
instability despite adequate antimicrobial therapy, patients with
persistent or recurrent bloodstream infections, or worsening pain Clinical information (pain scale score, neurological examina-
despite appropriate medical therapy [4]. tion, Oswestry score for spinal function), imaging data, as well as
In addition to antimicrobial therapy, spine immobilization is the type (rigid or soft) and duration of spine immobilization were
usually prescribed to decrease pain, reduce inflammation and pre- recorded at baseline and during the two follow-up appointments
vent neurological complications [5]. However, the most recent (between 6 weeks and 3 months and at 6 months). We also recorded
guidelines published by the Infectious Diseases Society of Amer- microbiological results and modality of antibiotic therapy.
ica, did not mention the use of immobilization in management
of VO [4]. The French Infectious Diseases Society (SPILF) recom- 2.4. Definitions
mends spine immobilization for 2 months, with 1 to 3 weeks of
strict bed rest before rigid bracing [6]. Nevertheless, those rec- Immobilization was defined as prescription of rigid bracing. We
ommendations are increasingly questioned because of the lack of compared immobilized patients with the others who were pre-
evidence in the literature regarding the efficacy of immobilization scribed soft bracing or no bracing at all.
[7,8] and because patients frequently complain of poor tolerance Neurological complications were defined as minor (radicular
of their bracing. Moreover, from a pathophysiological perspective, pain, reflex abolition, sensitive loss) or major (motor weakness
motor deficit is most often due to spinal cord ischemia, a septic or sphincter dysfunction). New neurological complications were
embolism of the vertebral arteria, or a compressive abscess caus- defined as appearance of a new sign or worsening of neurological
ing spinal stenosis [9]. The utility of spine bracing to prevent these examination.
complications still therefore needs to be investigated. Imaging variables were grouped into 3 categories: structural
The objective of our study was to describe spine immobilization lesion (vertebral destruction > 50%, destruction of posterior arch
practices in a large cohort of VO patients recruited in rheumatology, and sagittal angulation), inflammatory lesion (epidural inflamma-
infectious disease, and neurosurgery units from several hospitals. tion and epidural abscess) and compressive lesion (spinal cord
Secondary objectives were to identify factors associated with the hypersignal, subarachnoid space effacement, and dural sac com-
prescription of immobilization, and to evaluate the association pression) (Fig. 1).
between immobilization and neurological complications.
2.5. Statistical analyses

2. Methods
Baseline characteristics of the overall population were
expressed as frequencies (percentages) for categorical variables, as
2.1. Study design and setting
mean ± standard deviation (SD) for continuous data and in the case
of non-normal distributions as median with interquartile range
We conducted a prospective national observational study in
(IQR; 25th–75th percentile). Risk factors for immobilization were
11 French Hospital centers, from February 2016 to December
assessed using mixed models to take center effect into account
2019. We prospectively included adult patients hospitalized for
(random effect). Multivariate analyses were performed. All vari-
VO in rheumatology, infectiology, or spinal surgery units. On
ables entered into the model were associated with a P value of
inclusion, patients were informed of the study, received an infor-
0.2 or lower in univariate analyses. The final set of predictors was
mation letter, and gave their oral consent. The study protocol was
selected via stepwise variable selection. Multiple imputations with
approved by the local ethic committee and recorded in Clinical
the use of chained equations were performed to address missing
Trials (NCT04655950). The study was conducted using the format
data under a missing-at-random assumption. All tests were two
recommended by the Strengthening the Reporting of Observational
tailed, and P values of less than 0.05 were considered significant.
Studies in Epidemiology (STROBE) guidelines.
Statistical analyses were performed with SAS software, version 9.4
(SAS Institute).
2.2. Inclusion and exclusion criteria
3. Results
Patients were over the age of 18 years and hospitalized in med-
ical or surgery units for VO occurring on a non-instrumented spine 3.1. Patient characteristics
(without foreign devices from a previous spinal procedure). VO
was defined on typical radiological features (Magnetic Resonance We included 250 patients with a mean age of 66.7 ± 15 years,
Imaging or Computed-Tomography plus radionuclear imaging) and mostly men (67.2%, n = 168). Median duration of pain before
identification of a microbiological agent (on blood cultures or verte- diagnosis was 25 days (interquartile range [IQR], 11-51 days).
bral biopsy). If microbiological samples were sterile, patients could Demographic characteristics and comorbidities are described in
be included if they had a good response to antibiotic therapy with Table 1.
signs of inflammation on the vertebral biopsy (Polynuclear cell infil- On clinical examination at diagnosis, 25.6% patients (n = 64)
trate or micro-abscesses), and thus, if the clinicians in charge of had minor neurological signs and 9.2% (n = 23) major neurologi-
the patient concluded that there was a septic cause to the spondy- cal signs. The imaging technique used for diagnosis was mainly
lodiscitis. magnetic resonance imaging (MRI), performed in 87.2% of patients,
Exclusion criteria were spinal bone device infections, VO occur- other imaging modalities were computed tomography, positron
ring within one month of a spinal surgical procedure, or if there emission tomography (PET) and bone scan. In our cohort, the
was no imaging available for review. Patients under the age of 18 lumbo-sacral level was the most frequently involved (56.4%), fol-
years, under legal protection, or pregnant women were excluded. lowed by the thoracic level (23.2%), and the cervical level (13.3%).
Patients with other types of spine infection were not included Epidural inflammation was the most frequent imaging anomaly,

2
A. Le Pluart, G. Coiffier, C. Darrieutort-Lafitte et al. Joint Bone Spine 89 (2022) 105333

Fig. 1. Illustration of structural, inflammatory and compressive damage in vertebral osteomyelitis. A. Inflammatory lesion on MRI: epidural abscess. B. Structural lesion on
MRI: destruction > 50% of the C6 vertebrae. C. Compressive lesion on MRI: dural sac compression and spinal chord hypersignal.

Table 1 found in 61.1% of patients. Epidural abscesses were found in 19.6%

General characteristics of patients with vertebral osteomyelitis.
of cases. A structural lesion (destruction > 50% of the vertebral body,
Total % (n = 250) destruction of the posterior arch or sagittal angulation) was found
Male, % (n) 67.2 (168/250) in 27.6% of patients, and a compressive lesion (spinal cord hyper-
Age, years, mean (SD) 66.7 (15.0) signal, subarachnoid space effacement, or dural sac compression)
Diagnosis delay, median [IQR] 25 [11–51] in 26.5%.
History of spine surgery, % (n) 5.6 (14/250) Blood cultures were performed in 97.6% of cases, 69.2% of
Frailty (0 “very fit” to 5 “severely frail”), median [IQR] 1.0 [1–2]
them were positive. Vertebral biopsies were performed in 33.2%
Diabetes, % (n) 20 (50/250)
Cancer, % (n) 22 (55/250) of patients, with 74.4% of positivity and surgical biopsies in 8.4%
Immunosuppressive treatment, % (n) 8.4 (21/250) of patients with 81% of positivity. Staphylococcus aureus was the
Renal insufficiency, % (n) 4.8 (12/249) most frequent pathogen found (33.6% of the patients). Microbio-
Obesity, % (n) 22.9 (57/249)
logical samples remained sterile in 5.6% of cases. Endocarditis was
Blood culture, % (n) 97.2 (243/249)
Positive blood culture, % (n) 69.2 (173/243) diagnosed in 22.4% (n = 56) of patients. Patients with endocarditis
Percutaneous biopsy, % (n) 33.2 (83/249) had significantly fewer minor neurological symptoms at diagno-
Positive percutaneous biopsy, % (n) 74.4 (61/82) sis (supplementary data). Median duration of antibiotic therapy
Surgical biopsy, % (n) 8.4 (21/249) was 45 days (IQR 42–62 days), with a median administration of 21
Positive surgical biopsy, % (n) 81.0 (17/21)
days intravenously (IQR 10–42 days) and 35 days orally (IQR 14–52
Microbiological findings
Staphylococcus aureus and Coagulase-negative 40.8 (102/250) days). In our cohort, 11.2% of patients (n = 28) underwent spine
staphylococci, % (n) surgery, those patients had more often major neurological compli-
Streptococcus sp., % (n) 18.0 (45/250) cations or compressive and inflammation lesions, and the cervical
Escherichia coli and other Gram-negative bacilli, % (n) 13.2 (33/250)
level was more often involved (Table S1, See the supplementary
Enterococcus sp., % (n) 7.2 (18/250)
Mycobacterium tuberculosis, % (n) 5.6 (14/250)
material associated with this article online).
Other or multiple bacterial findings, % (n) 9.6 (24/250)
No bacterial identification or missing data, % (n) 5.6 (14/250)
3.2. Immobilization findings
Endocarditis, % (n) 22.4 (56/249)
Level involved
Cervical (C2-C3 to C6-C7), % (n) 13.3 (32/241) In 76.7% (n = 191) of cases, strict bed rest was prescribed, with
Cervico-thoracic (C7-T1), % (n) 1.7 (4/241) median duration of 8 days (IQR 5-14 days). A verticalization table
Thoracic (T1-T2 to T11-T12), % (n) 23.2 (56/241) was used in 3.6% of patients. Rigid bracing was prescribed in 65.3%
Thoraco-lumbar (T12-L1), % (n) 5.4 (13/241)
(n = 162) of patients, with median duration of 6 weeks (IQR 6–12),
Lumbar (L1-L2 to L5-S1), % (n) 56.4 (136/241)
Multifocal, % (n) 25.5 (61/239) the duration was 6 weeks for 61 (42.7%) patients and 12 weeks for
Imaging anomaly 37 (25.9%). Initial use of soft bracing was prescribed in 4.0% (n = 11)
Structural lesion, % (n) 27.6 (61/223) of patients.
Vertebral destruction > 50%, % (n) 9.8 (23/234)
In order to explore factors associated with the prescription
Destruction of posterior arc, % (n) 8.4 (19/226)
Sagittal angulation, % (n) 20.1 (47/224) of immobilization, we compared patients immobilized with rigid
Inflammatory lesion, % (n) 62.6 (139/222) bracing with patients not immobilized (Table 2). In univariate anal-
Epidural inflammation, % (n) 61.1 (135/221) yses, the clinical factors significantly associated with prescription
Epidural abscess, % (n) 19.6 (44/224) of rigid bracing were: younger age (65.1 years old in the immo-
Compressive lesion, % (n) 26.5 (58/219)
bilization group and 69.3 in the non-immobilized group, P < 0.05),
Spinal cord hypersignal, % (n) 8.7 (19/218)
Subarachnoid space effacement, % (n) 19.9 (44/221) less frail patients on the autonomy scale (P < 0.05) and male gender
Dural sac compression, % (n) 22.5 (50/222) (71.6% and 58.1% respectively, P < 0.05). Prescription of immobiliza-
Duration of antibiotherapy: median [IQR] 45 [42–62] tion was significantly different between centers (P < 0.001) with
Surgery, % (n) 11.2 (28/249)
a variation of the prescription rate of rigid bracing from 13.3%
Pain scale evaluation (/100), mean (SD) 53.6 (28,65)
Minor neurological symptom, % (n) 25.6 (64/250)
(n = 2/15) and even 0% (n = 0/4) to 100% (n = 13/13). Imaging factors
Major neurological symptom, % (n) 9.2 (23/250) associated with rigid bracing were epidural inflammation (P < 0.01),
SD: standard deviation; Obesity: Body Mass Index > 30 kg/m2 ; IQR: interquartile
spinal cord hypersignal (P < 0.05), subarachnoid space effacement
range, Q1 = 25%, Q3 = 75%. (P < 0.05), and dural sac compression (P < 0.05). Patients with endo-
carditis were less likely to be immobilized than others (17.9%

3
A. Le Pluart, G. Coiffier, C. Darrieutort-Lafitte et al. Joint Bone Spine 89 (2022) 105333

Table 2
Univariate and multivariate analysis of factors associated with the prescription of rigid spine immobilization in vertebral osteomyelitis.

Univariable analysis Multivariable analysis

Population characteristics Rigid No rigid p Odds Ratio 95% Confidence p

Immobilization immobilization Interval
63.5% (n = 162) 34.7% (n = 86)

Centers <0.001 Take into account as random effect in

multivariate analysis
Male, % (n) 71.6 (116) 58.1 (50) <0.05
Age, mean (SD) 65.1 (15.2) 69.3 (14.2) <0.05
Diagnosis delay, median [IQR] 24 [11–52] 25 [11–44] 0.98
History of spine surgery, % (n) 6.8 (11/162) 3.5 (3/86) 0.39
Frailty (0 “very fit” to 5 “severely frail”), median [IQR] 1.0 [1–2] 1.0 [0–2] 0.05 0.65 [0.48; 0.88] <0.01
Diabetes, % (n) 19.8 (32/162) 20.9 (18/86) 0.83
Cancer, % (n) 25.3 (41/162) 16.3 (14/86) 0.10
Immunosuppressant treatment, % (n) 10.5 (17/162) 4.7 (4/86) 0.12
Renal insufficiency, % (n) 3.1 (5/161) 8.1 (7/86) 0.12
Obesity, % (n) 22.4 (36/161) 24.4 (21/86) 0.72
Endocarditis, % (n) 17.9 (29/162) 31.7 (27/85) <0.05 0.42 [0.21; 0.87] <0.05
Level involved
Cervical, % (n) 16.2 (25/154) 8.1 (7/86) 0.16
Cervico-thoracic, % (n) 2.0 (3/154) 1.2 (1/86) 1.00
Thoracic, % (n) 21.4 (33/154) 26.7 (23/86) 0.54
Thoraco-lumbar, % (n) 5.2 (8/154) 5.8 (5/86) 1.00
Lumbar, % (n) 55.2 (85/154) 58.1 (50/86) 0.82
Multifocal, % (n) 24.0 (37/154) 26.5 (22/83) 0.67
Imaging anomaly
Structural lesion, % (n) 30.1 (44/146) 22.7 (17/75) 0.24
Inflammatory lesion, % (n) 69.7 (101/145) 48.0 (46/75) <0.01 2.33 [1.21; 4.52] 0.01
Compressive lesion, % (n) 31.7 (45/142) 17.3 (13/75) <0.05
Duration of antibiotherapy, median [IQR] 45 [42–62] 46 [42–61] 0.96
Surgery, % (n) 14.2 (23/162) 5.9 (5/85) 0.05
Pain scale evaluation (/100): mean (SD) 54.7 (27.01) 51.2 (31.46) 0.57
Minor neurological symptom, % (n) 25.9 (42/162) 24.4 (21/86) 0.80
Major neurological symptom, % (n) 11.1 (18/162) 5.8 (5/86) 0.17

SD: standard deviation; IQR: interquartile range, Q1 = 25%, Q3 = 75.

(n = 29) and 31.7% (n = 27) respectively, P < 0.05). We found no sig- at baseline, 15.7% (8/51) still had a minor sign at 6 months. A major
nificant association between the level involved (lumbar, thoracic, sign persisted at 6 months in 30.8% (3/13) of the patient with a
or cervical), pain score, neurological complications at diagnosis, major neurological complication at baseline. The mortality rate was
recorded comorbidities and the prescription of a rigid immobiliza- 6.4% (n = 16).
tion. We analyzed factors associated with major neurological compli-
In multivariate analysis (Table 2), the only independent fac- cations during-follow-up (Table 3, Table S3). Factors significantly
tor positively associated with rigid spine immobilization was associated were mostly imaging features such as destruction of
inflammatory lesion on imagery (OR = 2.33 and IC [1.21–4.52], the posterior arc (P < 0.05), or sagittal angulation (P < 0.05). Signs
P = 0.01). Endocarditis was negatively associated with rigid bracing of compression were also associated with major neurological com-
(OR = 0.42 and IC [0.21–0.87], P < 0.05), as was frailness (OR = 0.65 plications: spinal cord hypersignal (P < 0.05), anterior effacement
and IC [0.48–0.88]), which was highly correlated to older age. of the subarachnoid space (P < 0.01), and dural sac compression
Some patients in the immobilized group (35.1%, n = 57/162) (P < 0.01). Patients who underwent surgery had significantly more
received a soft bracing prescription to follow on from the rigid one major complications during follow-up (P < 0.001). These patients
at the end of the initial hospitalization, or at 3 months. This relay were more severe at initial presentation (Table S1). We found no
bracing was more frequently prescribed in patients with a higher impact of prescribing rigid bracing on major neurological compli-
pain scale score on activity (P = 0.001). Of the patients with an initial cations during follow-up (P = 0.56).
prescription for rigid bracing, 23.3% (n = 30/129) were still wearing
a soft or rigid brace at 6 months. Observance was considered good 4. Discussion
for 83.0% (n = 112/135) of patients, who declared that they complied
with the prescription fully. To our knowledge, SPONDIMMO is the first study to assess
spine immobilization during VO. The lack of literature on this sub-
3.3. Immobilization and clinical evolution ject is surprising as immobilization remains a frequent question
in everyday practice and aims to prevent neurological complica-
During the 6-month follow-up, new minor neurological compli- tions, and to accelerate improvement of pain and inflammation
cations occurred in 9.2% (n = 23) of cases, at a median of 22 days (IQR, associated with this serious condition. With this large, prospective,
6–181) after diagnosis, 12.6% (n = 20) in the immobilized group and multicenter cohort we were able to describe both the prescrip-
3.5% (n = 3) in the other group (Fig. 2). Major neurological compli- tion of spine immobilization, and the neurological evolution of the
cations occurred in 6.8% of patients (n = 17), with a median onset patients during a 6-month follow-up period, in both medical and
of 11 days (IQR, 3–23), 6.2% (n = 10) in the immobilized group and surgical departments.
8.1% (n = 7) in the other group (Table S2). At the end of the follow There is a lack of clear guidelines on spine immobilization. Most
up, minor neurological complications persisted in 8.1% (n = 15/186) studies mention rigid spine immobilization for at least 6 weeks
patients and major neurological complications in 5.4% (n = 10/186) [7,10–12] for all patients, and even sometimes for 6 months or until
of patients. Among patients with minor neurological complication spine ankylosis occurs [13]. In our study, only two-thirds of our

4
A. Le Pluart, G. Coiffier, C. Darrieutort-Lafitte et al. Joint Bone Spine 89 (2022) 105333

Fig. 2. Survival curve of minor and major neurological complications during the follow-up of patients with vertebral osteomyelitis.

patients received a prescription for rigid bracing and the duration of We next sought the factors associated with the prescription of
this prescription varied considerably, with only half of the patients a rigid bracing. We first found that the presence of inflammatory
(48%) wearing their rigid bracing for 6 weeks or more. In their ret- lesions (epidural inflammation or epidural abscess) but not com-
rospective cohorts, Bettini et al. used a plaster brace for all patients pressive ones was a major determinant for the prescription of rigid
for 8 weeks with a canvas corset for 4 more weeks [7]. Legrand bracing in multivariate analysis. This is in opposition with the fact
et al. described bracing for 89.1% of patients for 81.5 days [14]. that compressive and structural lesions are clearly identified in
Rutges et al. noted that bed rest and orthosis have not been investi- the literature as risk factors for neurological impairment [3,18] as
gated in detail, and suggested a pragmatic approach: bed rest until confirmed in this cohort. These results highlight that physicians
pain and infection decrease, then start mobilization with a tho- give more credit to inflammatory lesions although the association
racolumbosacral orthosis for all patients, without mentioning any between inflammatory lesions and neurological complications
duration [15]. Finally, the French Infectious Diseases Society (SPILF) remains debated. Frailer and older patients received fewer pre-
recommends using rigid bracing for 1 to 3 months depending on scriptions for rigid spine immobilization, probably in order to
the localization [6]. Our data shows that these recommendations avoid immobilization complications in this vulnerable population
are rarely followed in clinical practice. known to have poorer outcomes, but without an increased risk of
Strict bed rest is also part of the standard treatment in VO, for neurological complication [19]. Endocarditis was also negatively
a duration of 1 or 2 weeks, or for some authors until improve- associated with rigid bracing, probably because VO was an inci-
ment in the pain [16] and to reduce inflammation. The aim of using dental diagnosis on body-scan imaging performed in the extension
spine bracing is to avoid prolonged bed rest [5] and its well-known check-up of patients with endocarditis. Those patients most often
morbidities, such as bedsores, deep vein thrombosis, pulmonary did not have VO symptoms (nor backpain neither stiffness), which
embolism, muscle loss, and cardiopulmonary disadaptation, espe- is consistent with the 41% of asymptomatic cases of VO in endo-
cially in the elderly [17]. carditis found by Carbone et al. [20]. Another determinant for the

5
A. Le Pluart, G. Coiffier, C. Darrieutort-Lafitte et al. Joint Bone Spine 89 (2022) 105333

Table 3
Factors associated with the apparition of neurological complications during follow-up of patients with vertebral osteomyelitis in univariable analysis.

Apparition of major neurological complications No major neurological complications

7.2% (18/250) 92.8% (232/250)

Male, % (n) 47.1 (8/17) 68.7 (160/233)

Age, years, mean (SD) 65.6 (11.86) 66.7 (15.25)
Diagnosis delay, median [IQR] 21.8 (20.51) 42.2 (54.1)
History of spine surgery, % (n) 5.9 (1/17) 5.6 (13/233)
Frailty (0 “very fit” to 5 “severely frail”): mean (SD) 1.35 (1.17) 1.26 (1.02)
Diabetes, % (n) 17.6 (3/17) 20.2 (47/233)
Cancer, % (n) 35.3 (6/17) 21.0 (49/233)
Immunosuppressive treatment, % (n) 0 (0/17) 9 (21/233)
Renal insufficiency, % (n) 0 (0/17) 5.2 (12/233)
Obesity, % (n) 41.2 (7/17) 21.6 (50/232)
Endocarditis, % (n) 23.5 (4/17) 22.4 (52/232)
Level involved
Cervical, % (n) 18.8 (3/16) 12.9 (29/225)
Cervico-thoracic, % (n) 6.3 (1/16) 1.3 (3/225)
Thoracic, % (n) 31.3 (5/16) 22.7 (51/225)
Thoraco-lumbar, % (n) 0 (0/16) 5.8 (13/225)
Lumbar, % (n) 43.8 (7/16) 57.3 (129/225)
Multifocal, % (n) 43.8 (7/16) 24.2 (54/223)
Imaging anomaly
Vertebral destruction > 50%, % (n) 12.5 (2/16) 9.6 (21/218)
Destruction of posterior arc, % (n) 25.0 (4/16) 7.1 (15/210)*
Sagittal angulation, % (n) 43.8 (7/16) 19.2 (40/208)*
Epidural inflammation, % (n) 81.3 (13/16) 59.5 (122/205)
Epidural abscess, % (n) 37.5 (6/16) 18.3 (38/208)
Spinal cord hypersignal, % (n) 25.0 (4/16) 7.4 (15/202)*
Subarachnoid space effacement, % (n) 50.0 (8/16) 17.6 (36/205)**
Dural sac compression, % (n) 56.3 (9/16) 19.9 (41/206)**
Duration of antibiotherapy, median [IQR] 43 (7–102) 45 (5–417)
Surgery, % (n) 47.1 (8/17) 8.6 (20/232)***
Pain scale evaluation (/100), mean (SD) 53.6 (25.3) 53.6 (28.9)
Minor neurological symptom, % (n) 11.8 (2/17) 26.6 (62/233)
Major neurological symptom, % (n) 0 (0/17) 9.9 (23/233)

SD: standard deviation; IQR: interquartile range, Q1 = 25%, Q3 = 75%.

*
P < 0.05.
**
P < 0.01.
***
P < 0.001.

decision of the type and duration of spine immobilization was the outcome was favorable for most patients, with only 5.4% of major
hospital center, highlighting the importance of local custom and neurological signs persistent at 6 months.
experience. This shows the need for consensual recommendations In conclusion, we showed that spine immobilization pre-
based on evidence-based data to harmonize practices. Of note, scriptions are highly heterogeneous between centers and seems
we found no association with cervical level or major neurological associated with inflammatory lesions on imaging but negatively
complications at diagnosis and rigid immobilization. However, associated with frailness and simultaneous endocarditis. Our data
those patients were more often managed surgically. suggest that, in absence of any factor associated with neurological
The aim of spine immobilization is to reduce the risk of neu- complication, spine bracing might not be systematically indicated.
rological complications. We thus studied the occurrence of new Therefore, bracing should be discussed for each patient after sys-
neurological symptoms during follow-up in relation to the pre- tematic analysis of clinical and imaging information. While Bernard
scription of rigid bracing. We found a similar rate of new major et al. investigated modified antibiotic practices in a randomized
neurological complications in patients immobilized with (6.2%) and trial [23], spine immobilization deserves proper investigation in
those without (8.1%) rigid bracing. Likewise, occurrence of minor randomized controlled studies to better define its indications and
neurological complications was similar with (12.6%) or without benefits for patients.
(3.5%) rigid bracing. The observational design of our study and
the absence of any randomization preclude drawing any definite Disclosure of interest
conclusions regarding the effect of immobilization on neurological
complications. However, this shows the low rate of such complica- The authors declare that they have no competing interest.
tions in the non-immobilized group and also the possibility of the
occurrence of complications, even in cases of rigid immobilization. Acknowledgments
As structural and compressive lesions were significantly associated
with neurological complications, patients with this type of anomaly This research did not receive any specific grant from funding
should therefore being carefully monitored. An important finding agencies in the public, commercial or not-to-profit sectors.
was the short delay after diagnosis before the occurrence of new
major neurological complications, with a median of 11 days in our Appendix A. Online Material
cohort. This enhances the requirement for strict clinical monitoring
during the first two weeks, in addition to controlling normalization Supplementary data (Tables S1-S3) associated with this arti-
of inflammatory markers such as C-reactive protein dosage (CRP) cle can be found, in the online version, at https://doi.org.10.
[21]. Consistently with other cohorts of VO [22], the neurological 1016/j.jbspin.2021.105333.

6
A. Le Pluart, G. Coiffier, C. Darrieutort-Lafitte et al. Joint Bone Spine 89 (2022) 105333

References [12] Butler JS, Shelly MJ, Timlin M, et al. Nontuberculous pyogenic spinal infec-
tion in adults: a 12-year experience from a tertiary referral center. Spine
[1] Kehrer M, Pedersen C, Jensen TG, et al. Increasing incidence of pyogenic spondy- 2006;31:2695–700, http://dx.doi.org/10.1097/01.brs.0000244662.78725.37.
lodiscitis: a 14-year population-based study. J Infect 2014;68:313–20. [13] Pola E, Taccari F, Autore G, et al. Multidisciplinary management of pyo-
[2] Issa K, Diebo BG, Faloon M, et al. The epidemiology of vertebral osteomyeli- genic spondylodiscitis: epidemiological and clinical features, prognostic factors
tis in the United States From 1998 to 2013. Clin Spine Surg 2018;31:E102, and long-term outcomes in 207 patients. Eur Spine J 2018;27:229–36,
http://dx.doi.org/10.1097/BSD.0000000000000597. http://dx.doi.org/10.1007/s00586-018-5598-9.
[3] Bart G, Redon H, Boutoille D, et al. Is there an association between mag- [14] Legrand E, Flipo RM, Guggenbuhl P, et al. Management of nontuber-
netic resonance imaging and neurological signs in patients with vertebral culous infectious discitis. Treatments used in 110 patients admitted
osteomyelitis? A retrospective observational study on 121 patients. Medicine to 12 teaching hospitals in France. Joint Bone Spine 2001;68:504–9,
2016;95:e2373, http://dx.doi.org/10.1097/MD.0000000000002373. http://dx.doi.org/10.1016/s1297-319x(01)00315-3.
[4] Berbari EF, Kanj SS, Kowalski TJ, et al. 2015 Infectious Diseases Society of [15] Rutges JPHJ, Kempen DH, van Dijk M, et al. Outcome of conservative and surgi-
America (IDSA) Clinical Practice Guidelines for the diagnosis and treatment cal treatment of pyogenic spondylodiscitis: a systematic literature review. Eur
of native vertebral osteomyelitis in adults. Clin Infect Dis 2015;61:e26–46, Spine J 2016;25:983–99, http://dx.doi.org/10.1007/s00586-015-4318-y.
http://dx.doi.org/10.1093/cid/civ482. [16] Tsantes A, Papadopoulos D, Vrioni G, et al. Spinal infections: an update. Microor-
[5] Gregori F, Grasso G, Iaiani G, et al. Treatment algorithm for spontaneous ganisms 2020;8:476, http://dx.doi.org/10.3390/microorganisms8040476.
spinal infections: a review of the literature. J Craniovertebr Junction Spine [17] Zarghooni K, Röllinghoff M, Sobottke R, et al. Treatment of spondylodiscitis. Int
2019;10:3–9, http://dx.doi.org/10.4103/jcvjs.JCVJS 115 18. Orthop 2012;36:405–11, http://dx.doi.org/10.1007/s00264-011-1425-1.
[6] Spondylodiscites infectieuses primitives, et secondaires à un geste intra-discal, [18] Widdrington J, Emmerson I, Cullinan M, et al. Pyogenic spondylodiscitis:
sans mise en place de matériel. Texte court. Med Mal Infect 2007;37:554–72, risk factors for adverse clinical outcome in routine clinical practice. Med Sci
http://dx.doi.org/10.1016/j.medmal.2007.03.008. 2018;6:96, http://dx.doi.org/10.3390/medsci6040096.
[7] Bettini N, Girardo M, Dema E, et al. Evaluation of conservative treat- [19] Courjon J, Lemaignen A, Ghout I, et al. Pyogenic vertebral osteomyelitis
ment of non specific spondylodiscitis. Eur Spine J 2009;18:143–50, of the elderly: characteristics and outcomes. PLoS ONE 2017;12:e0188470,
http://dx.doi.org/10.1007/s00586-009-0979-8. http://dx.doi.org/10.1371/journal.pone.0188470.
[8] Nene A, Bhojraj S. Results of nonsurgical treatment of tho- [20] Carbone A, Lieu A, Mouhat B, et al. Spondylodiscitis com-
racic spinal tuberculosis in adults. Spine J 2005;5:79–84, plicating infective endocarditis. Heart 2020;106:1914–8,
http://dx.doi.org/10.1016/j.spinee.2004.05.255. http://dx.doi.org/10.1136/heartjnl-2019-316492.
[9] Lemaignen A, Ghout I, Dinh A, et al. Characteristics of and risk [21] Duarte RM, Vaccaro AR. Spinal infection: state of the art
factors for severe neurological deficit in patients with pyogenic ver- and management algorithm. Eur Spine J 2013;22:2787–99,
tebral osteomyelitis: a case–control study. Medicine 2017;96:e6387, http://dx.doi.org/10.1007/s00586-013-2850-1.
http://dx.doi.org/10.1097/MD.0000000000006387. [22] McHenry MC, Easley KA, Locker GA. Vertebral osteomyelitis: long-term out-
[10] Lener S. Management of spinal infection: a review of the literature. Acta Neu- come for 253 patients from 7 Cleveland-Area. Hospitals 2002:1342–50.
rochir 2018;160:487–96. [23] Bernard L, Dinh A, Ghout I, et al. Antibiotic treatment for 6 weeks versus
[11] Karadimas EJ, Bunger C, Lindblad BE, et al. Spondylodiscitis. A 12 weeks in patients with pyogenic vertebral osteomyelitis: an open-
retrospective study of 163 patients. Acta Orthop 2008;79:650–9, label, non-inferiority, randomised, controlled trial. Lancet 2015;385:875–82,
http://dx.doi.org/10.1080/17453670810016678. http://dx.doi.org/10.1016/S0140-6736(14)61233-2.