Streamlining the Journey of Research Into Clinical Practice: Making Your Patients and Practice Flourish Evaluation and Treatment of Pyogenic Spondylodiscitis of the Spine: AO Spine Knowledge Forum Trauma and Infection

Clinical Rationale

Several attempts have been made to produce a classification/scoring system that guides decision-making in infectious spondylodiscitis.^7,8 Most studies are retrospective case series, with a small sample size, and rarely a comparison is made between them using the same case sample. Previous scoring systems used either subjective domains or variable radiographic parameters, limiting its applicability in daily clinical practice. The lack of a validated system generates discrepancies in decision-making, with delays in definitive treatment and unfavorable outcomes.

Study Summary

The authors aim to compare three scoring systems for spondylodiscitis:

- The Spinal Infection Treatment Evaluation Score (SITE Score) ⁶ ;

The SITE score ⁶ includes 5 domains (neurology, location, radiological changes, pain, comorbidities) ranging from 0-15 points, in which a score from 0-8 indicates the need for surgery.

The SISS score ⁷ includes four domains (location, extent of bone lesions, spinal alignment, mechanical pain) ranging from 0-14 points, with lesions with 10-14 points being considered unstable.

Methodological Review

This study uses a retrospective cohort of 123 spondylodiscitis cases rated by three experienced spine surgeons. Most of the patients in this cohort were treated surgically (70.7%), with a good distribution of different lesion types. The authors used two different indications for surgery for the SISS and SINS, 1 with the potentially unstable and unstable injuries (SISS>5 and SINS>7) and another including only unstable injuries (SISS>10 and SINS>13). Interobserver analysis using intraclass correlation coefficients, and predictive validity was analyzed by cross-tabulation, followed by the Student t test for paired samples.

The mean SITE score was 6.52, with 69% of the patients having a surgical indication. The authors also analyzed the surgical indication after excluding epidural abscess cases (61% had surgical indication).

The SINS score yielded 55% of potentially unstable and 35% unstable lesions, and after exclusion of epidural abscess cases (45% potentially unstable and 46% unstable).

The mean SISS was 7.53, with 56% being potentially unstable lesions and 35% unstable.

The SITE score had better ratings in comparison to the SISS and SINS, and better agreement rate to the real treatment offered to the patients. It is of note that the panel agreed unanimously with the treatment option in 61% of the cases, leaving 39% of the cases without full agreement. Even though scoring better, the SITE score recommended surgery for six patients who underwent conservative treatment, and in 14 patients’ surgery was performed without the SITE recommendation.

The main limitation of this study is that it was performed by the developers of the SITE score, meaning they were used to this score, and possibly had similar tendencies regarding surgical threshold for spinal infections. Although including some clinical information on their score, the SITE score does not take in consideration the pathogen identification and virulence which is associated with improved outcomes for surgical and clinical treatment.

Having only three raters and not been able to evaluate intraobserver agreement also limits the study’s validity. The quality of evidence is very low.

Recommendation for Integrating Into Clinical Practice

This study highlights the evolution of current scoring systems for spondylodiscitis and the need for further improvement to achieve a more standardized decision-making tool for treatment decision. The use of the SITE and SISS score is yet to be validated by external studies, as well as each of its components. Although radiological data and neurology are relevant for urgent decision making, more information about the patient and the pathogen is needed for an adequate decision-making tool.

Another key-point is the risk of over and under treatment associated with personal preferences or individual beliefs in comparison with validated scoring systems and check lists for specific pathologies.

It is encouraging that work is evolving to validate a decision-making aid for infectious spondylodiscitis. Although we cannot yet recommend to incorporate the SITE Score or SISS into practice, 1 should be aware of the principles used and its clinical need.

Clinical Rationale

Pyogenic native vertebral osteomyelitis (NVO) and spondylodiscitis (SD) are increasingly prevalent infections due to advances in diagnostic techniques and growing risk factors. While classic symptoms like fever, back pain, and neurological deficits can guide suspicion, diagnostic certainty relies on positive cultures (ie, obtained blood cultures, CT guided or open biopsy). This makes complementary diagnostic tools essential.

Plain radiographs, though low-cost, lack sensitivity, particularly in early disease. Magnetic resonance imaging (MRI) has become the preferred modality due to its superior sensitivity and specificity. This systematic review and meta-analysis evaluates both traditional and advanced imaging techniques, highlighting their roles in improving diagnostic accuracy for NVO and SD.

Study Summary

This study analyzed data from retrospective and prospective studies examining at least two imaging modalities or comparing 1 modality to a reference standard in adults (age>17 years) with suspected NVO/SD which were published between 1970 and July 2021. Studies involving 18F-FDG PET/CT, MRI, Gallium-67 (⁶⁷ Ga), Technetium-99m (^99mTc), or Indium-111 (¹¹¹In) were included. The reference standard for the diagnosis of NVO/SD was either growth of a compatible organism from biopsy or blood cultures, confirmatory histology from biopsy, or a highly suggestive clinical picture. In total, 20 studies were included in this review with 1093 extracted subjects. All included studies underwent quality and applicability assessment. The main outcomes of interest were sensitivity, specificity and diagnostic odds ratios (DOR) using constructed 2 × 2 contingency tables from each study. They state that the sensitivity and specificity in the setting is more relevant than the DOR. While the DOR indicates the overall performance, the authors state that missing a diagnosis is worse, therefore the sensitivity is stated to be the most valuable measure in this study.

Results

• MRI showed a sensitivity of 90%, specificity of 72% with a DOR of 24.

Methodological Review

The present systematic review and meta-analysis employed strong eligibility criteria for the selected studies and the review was conducted following PRISMA-guidelines. Further, literature review was conducted by a medical reference librarian in Ovid MEDLINE, Ovid EMBASE, Scopus, the Cochrane library, and ISI Web of Science, which enhanced the search in order to retrieve more available studies. The authors assessed quality and applicability of all selected studies using the quality assessment of diagnostic accuracy studies-2 tool to evaluate risk of bias of the selected studies. Another strength of the present analysis is the usage of a bivariate mixed-effects regression model that accounts for the correlation between sensitivity and specificity in each study and heterogeneity between studies. Overall, the heterogeneity of all assessed studies was less than 40% which suggests that the results within the imaging techniques were similar and therefore the pooling of the studies becomes reliable and the meta-analysis results are more certain. Limitations of this study are the exclusion of not-routinely used imaging techniques (eg, CT-alone) and the possibility of inclusion of non-NVO or postoperative SD, especially in older studies which underreported demographics. Another limitation is the wide range (1980 to 2020) of the analyzed studies resulting in heterogeneity regarding imaging technology. Lastly, the highly suggestive clinical picture, as a reference standard for diagnosis might produce false positive values.

Overall, this systematic review and meta-analysis was performed with a robust search strategy and rigorous statistical methods which increase the generalizability of the findings. Future prospective, comparative imaging studies with standardized imaging protocols and rigorous clinical reporting could overcome these limitations.

Recommendation for Integrating into Clinical Practice

The findings of this systematic review and meta-analysis suggest that MRI (current gold-standard) is slightly inferior in terms of sensitivity and specificity compared to 18F-FDG-PET/CT. The most compelling reason to use MRI in PS is the evaluation of the spinal cord and the epidural space. Moreover, due to the relative wide-spread availability of MRI-scanners, relative cost-effectiveness and greater anatomical detail compared to 18F-FDG-PET/CT, the slight inferiority of MRI does not justify a paradigm shift in the imaging diagnosis of native vertebral osteomyelitis or spondylodiscitis. ⁶⁷ Ga/^99mTc-combination and ¹¹¹In maybe justified in cases where MRI and/or 18F-FDG-PET/CT failed in aiding imaging diagnosis.

In conclusion, MRI remains the imaging gold-standard, whereas 18F-FDG-PET/CT is a viable option for early diagnosis or for patients who are unsuitable for MR-imaging.

We conditionally recommend the use of contrast MRI as the gold standard for diagnosis of PS and 18F-FDG-PET/CT as a viable adjunct for complex cases.

Clinical Rationale

The mainstay of PS treatment consists of antibiotic therapy with vast variation in duration and specifics; which are highly debated. ¹¹ Surgery becomes indicated in the presence of neurological deficits, mechanical instability, and failure of medical management. With this pathology carrying up to a 25% risk of mortality and substantial morbidity, identifying best treatment is crucial to avoid significant burden to patients and society. ³ Advocates for early surgery argue that source control and deformity prevention is necessary. However, the role of early surgery remains controversial. Thavarajasingam et al ¹⁰ report some of the best current available evidence that may help settle this ongoing debate.

Study Summary

The systematic review included 31 studies. It presents a vital overview of the current literature on this important topic with 31 manuscripts included and reviewed. Most of the studies included were retrospective in nature (27/31) with few prospective studies (3/31) and 1 ambispective study. The GRADE scoring is employed to weigh the quality of include studies.

Conservative treatment mostly consisted of intravenous and/or oral antibiotics with specifics being absent from most studies. Most common operations were decompression and debridement. The most common surgical indication was presence or worsening of neurological deficit followed by failure of medical management.

This systematic review illustrated the lack of consensus regarding the definition of treatment failure. The most common definition used was any “intensification of treatment”, which was defined by either: (1) failure of initial medical management with antibiotics requiring surgery; or (2) failure of surgical management requiring repeat surgery. A consensus on the definition of treatment failure will be necessary for future investigative efforts.

This meta-analysis consisted of a large overall pooled sample size of 10 954 patients from 21 studies. It showed that the pooled mortality among the early surgery patient subgroup was 8% compared to 13% for patients initially treated conservatively. Furthermore, early surgery was associated with a 39% risk reduction in mortality rate.

Additionally, the mean proportion of relapse/failure among early surgery subgroup was 15% vs 21% for the conservative treatment subgroup. Moreover, early surgery was associated with a 40% risk reduction in relapse/failure.

Early surgery was also associated with 7.75 days per patient reduction in length in hospital stay (P < 0.01).

Methodological Review

This rigorously conducted systematic review and meta-analysis compared the mortality, relapse rate and length of hospital stay of conservative vs early surgical treatment of pyogenic spondylodiscitis. All major databases were searched to include original studies, which were then evaluated using a qualitative synthesis, meta-analyses, influence and regression analyses.

The quality of evidence is graded as moderate. The true effect is likely to be close to the estimate of the effect reported in this meta-analysis. The Oxford Centre of Evidence-Based Medicine (OCEBM) guidance was used to determine the level of evidence of each included study. Twenty-one studies were classified as 2b, three studies as level 3b, and seven studies as level 4. Additionally, most of the studies included are retrospective in nature and the few prospective studies included are of small to moderate sample size. Egger’s asymmetry test was significant and yielded publications bias.

The authors did attempt to address baseline characteristics and confounders like age, comorbidities, and surgical candidacy in their included studies. However, the level of control over these variables was limited by the nature and quality of the primary studies.

Recommendation for Integrating into Your Clinical Practice

We conditionally recommend early surgery for patients with pyogenic spondylodiscitis over conservative management. Future prospective large-scale studies are necessary to solve this clinical dilemma especially in the absence of neurological deficits and to better select patients who may benefit from early surgery.

Clinical Rationale

PS, a serious infection of the spine, often results in significant morbidity and mortality. ¹³ Older adults with multiple comorbidities are especially at risk. ¹⁴ Historically, conservative management with antibiotics has been the first-line approach for patients without neurological deficits or spinal instability. ¹⁵ However, growing evidence suggests that delaying surgical intervention may lead to worse outcomes, including higher mortality rates. ¹⁰

This study is especially relevant as it compares outcomes between two well-resourced health care systems with standardized protocols, ensuring a homogenous treatment approach across centers. By doing so, the study provides insights into the effectiveness of these treatment strategies in an almost equipoise situation, enhancing its relevance for clinical decision-making. The study aims to compare the management of pyogenic spondylodiscitis: whether early surgical treatment offers superior outcomes compared to conservative care. Given the rising incidence of spinal infections in aging populations, identifying the optimal treatment strategy is paramount for improving survival and reducing hospital stays.

Study Summary

This retrospective study analyzed data from 392 patients with primary pyogenic spondylodiskitis treated between 2017 and 2022 across centers in the UK and Germany. While UK centers usually treat their patients conservatively in this setting, the German centers usually opt for surgery. After applying propensity score matching to balance confounding factors like age, CRP levels, and pre-existing conditions, two comparable cohorts of 95 patients each (CoT vs SuT) were created. The primary outcome was mortality, while secondary outcomes included hospitalization duration, infection relapse, and neurological status at discharge.

• The mortality rate was significantly higher in the CoT (24.2%) compared to the SuT (4.2%), P < .001.

Methodological Review

The study employs a robust methodological approach by using propensity score matching (PSM) to create balanced cohorts of patients receiving either conservative treatment or early surgical intervention for pyogenic spondylodiskitis. PSM ensures that the cohorts are well-matched on important confounding variables, such as age, CRP levels, comorbidities, and infection characteristics, which enhances the reliability of the comparison between treatment groups. This statistical adjustment is particularly important in retrospective studies, helping to minimize biases that could arise from non-random treatment selection.

One of the study’s notable strengths is that the patients were treated in two well-developed health care systems—Germany and the UK—both of which adhere to high standards of care with clearly established local protocols. The homogeneity in treatment protocols across the centers, despite being in different countries, ensures that all patients received the best possible care according to established regional/local/national guidelines. This not only increases the internal validity of the study but also improves its generalizability to other well-resourced health care settings.

There are limitations associated with the study’s retrospective design. Even with PSM, there remains a risk of residual confounding due to unmeasured variables, such as frailty, surgeon expertise or availability or patient preferences, that may have influenced treatment decisions. The retrospective nature also means that the study relies on the accuracy and completeness of medical records, which can introduce variability in data quality.

Another limitation is the absence of long-term follow-up, which restricts the analysis to short-term outcomes such as mortality and hospital stay. This leaves important questions unanswered about functional recovery, quality of life, and long-term complications like recurrent infections or spinal deformities. Including these outcomes would provide a more comprehensive assessment of the true benefits of early surgical intervention.

In conclusion, the study’s strengths lie in its rigorous statistical methods and the high-quality, homogeneous care provided across two advanced health care systems, which enhances the generalizability of its findings. Nonetheless, the lack of long-term follow-up and the potential for unmeasured confounding factors should be considered when interpreting the results. Finally, it has to be acknowledged that some patients do very well in the conservative group.

Recommendation for Integrating into Clinical Practice

The findings from this study provide moderate evidence (due to the methodological rigor and large effect size, limited by the retrospective design) in favor of early surgical intervention for PS, particularly for reducing mortality and hospital stay. Given that the study was conducted in well-resourced health care systems with homogenous treatment protocols, the results are likely applicable to similar settings worldwide.

We support a conditional recommendation that for patients who are suitable surgical candidates, early surgery is the preferred treatment strategy. However, for selective patients conservative therapy remains a viable option.

Clinical Rationale

There is no consensus regarding the duration of antibiotic treatment and most guidelines recommend 6 to 12 weeks as part of the standard of care. As we know, prolonged antibiotic therapy can be associated with adverse events, resistance, and higher health care costs. This study aimed to compare treatment duration with effective antibiotics for 6 weeks and 12 weeks in patients with microbiologically confirmed pyogenic vertebral osteomyelitis.

Study Summary

This is a multicentric, open-label, non-inferiority, randomized controlled trial initially involving 359 patients with microbiologically confirmed pyogenic vertebral osteomyelitis from 2006 to 2011 across 71 medical centers in France. In the intention-to-treat analysis, 176 patients were included in the 6-week treatment (98%) and 175 (99%) in the 12-week treatment. After exclusions from protocol extension violation (plus or minus six days of the duration assigned), 146 (82%) included in the per-protocol analysis received six weeks of antibiotic therapy whereas 137 (78%) received 12 weeks of antibiotic treatment. Antibiotic therapy followed French guidelines and involved the use of a combination of oral fluoroquinolones and rifampicin as a first-line treatment whenever possible. Intravenous oxacillin or cloxacillin was used as parenteral penicillin M. The primary outcome was the comparison of the proportion of patients who were classified as cured after 1 year of treatment. Secondary outcomes were the persistence of clinical symptoms, major adverse events, percentage of patients who were not cured at six months, quality of life, microbiological resistance, and identification of factors associated with treatment failure and death. Confirmed cure was defined as a sustained absence of fever, pain, and inflammatory syndrome (C-reactive protein ≤10 mg/L) 12 months after the end of treatment.

For the intention-to-treat population, 160 of 176 (90.9%) in the 6-week treatment and 159 of 175 (90.8%) in the 12-week treatment met the criteria for clinical cure. Non-inferiority was also shown in the per-protocol analysis between 6 and 12 weeks of treatment regarding the rate of cure (93.8% vs 96.4%, respectively). There were no differences in terms of quality-of-life scores, adverse events, and treatment failure between both groups.

In multivariate analysis, patients aged 75 years or older and S. Aureus infections were associated with a higher risk of treatment failure. Subgroup analyses indicated that non-inferiority of the 6-week regimen was not conclusively demonstrated in patients with complicating factors such as advanced age, diabetes, or abscesses. These findings likely reflect limited statistical power rather than clear evidence of inferiority.

Methodological Review

This is a high-quality study according to the Grading of Recommendations, Assessment, Development and Evaluation (GRADE). ¹⁷ The authors planned a sample size of 200 patients per group by assuming a 1-year cure rate of 85% in those receiving an effective antibiotic treatment regimen irrespective of whether the duration was 6 weeks or 12 weeks, using a one-tailed type-I error of 0.025, 80% statistical power, and assuming a 20% dropout rate. Authors tested the hypothesis of non-inferiority in both ways, the intention-to-treat analysis (initial assignment) and the per-protocol analysis (patients who finally received the treatment according to the protocol), the authors also searched for factors associated with a lower rate of treatment success through a multivariate logistic regression. Despite its strengths, including a large multicenter design and robust follow-up, the study’s open-label design and exclusion of specific subgroups, such as those with non-microbiologically confirmed vertebral PS, necessitate caution in broadly applying its conclusions.

Another limitation of the trial is the open-label design, even though masking was not possible for physicians and patients due to logistic reasons, the members of the committee that classified patients as cured were blinded regarding treatment duration. Another limitation of the study is that the duration of intravenous treatment was not standardized and was left to the treating physicians, however, no differences were observed between groups in terms of distribution of antibiotics.

This study has several strengths such as the sample size, randomization, multicentric design, guideline-supported drugs and follow-up.

Recommendation for Integrating Into Clinical Practice

This study proved, with high-quality evidence, that in patients with PS with microbiological identification, 12 weeks of antibiotic treatment has no clinical advantage over 6 weeks of treatment.

Recommendations for Practice:

• Strong Recommendation: For patients matching the study’s criteria (microbiologically confirmed pyogenic vertebral osteomyelitis without complicating factors), 6 weeks can be the standard of care.