We present a comprehensive analysis of VO patients from a prospective cohort focused on identifying and comparing predictors of early and late mortality. Our main findings were as follows: (i) Within the first 30 days, 5% of patients died, and another 14% died between day 31 and day 365. (ii) Patients who died within the first 30 days of diagnosis had a higher ASA score, a higher rate of CKD, were more likely to have bacteremia, and MSSA was detected more frequently as the causative agent. Patients who died between days 31 to 365 were older, more likely to have heart failure as well as CKD, and more likely to have bacteremia, concomitant IE, and psoas abscess. (iii) CKD and MSSA were identified as independent risk factors for early mortality. Our data suggest that late mortality did not seem to be clearly driven by recurrent VO, but other independent predictors such as ASA score > 2, age > 70 years, CKD, and bacteremia.

The mortality rates of this study are high compared with most other studies which range from 2 to 24%[2, 6, 13,14,15]. In contrast to our cohort, aforementioned studies had a high proportion of patients treated conservatively. It is noticeable that studies with a high ratio of surgical therapy show high mortality rates [10, 16]. Regarding the high rate of surgical interventions in our study, the mortality rate can be explained by the severity of the course of the disease and underlines the role of a tertiary care hospital: specialization of a spine center may facilitate the treatment of particularly complex cases that may not be adequately cared for in peripheral hospitals. The present analysis with 355 VO patients confirms the high 1-year mortality of 20% found by Kehrer et al. [2].

As only patients for whom a complete 1-year follow-up was available were included in this analysis, 9% of patients without a 1-year follow-up were excluded. Therefore the reported 1-year mortality may be slightly underestimated, since patients who are lost to follow-up often have a higher mortality. Nevertheless, since the 1-year mortality is in the range reported in the literature, it was important to us to focus on patients with complete follow-up to be able to provide a valid statement about the immediate postoperative and 1-year course after VO.

Previous studies have analyzed pre-existing conditions in VO and their influence on the course of the disease. Typically, VO occurs more often in multimorbid patients. Accordingly, the prevalences of diabetes, malignant tumor, heart failure, CKD, COPD, rheumatic disease, alcohol addiction, and drug disorder were higher in our cohort than in the normal population as also reported by Kehrer et al. [2]. While other studies revealed similar prevalence of comorbidities [2, 7, 17], the risk factors influencing the early and late mortality rate of VO are still only poorly understood.

In our study, age older than 70 years was found to be a predictive factor for death between day 31 and day 365. The significant impact of age on survival after VO diagnosis was also described in the study by Kehrer et al. [2]. They demonstrated an increased 1-year mortality starting at an age of 65 years. Furthermore, Akiyama et al. [18] and Loibl et al. [19] described a significant impact of an age over 60 years on in-hospital mortality. Issa et al. [13] also demonstrated significantly increasing probabilities of general death in VO patients older than 60 years in their epidemiological analysis. The rationale for the influence of age lies in increasing multimorbidity. The assumption is further strengthened by the associated ASA risk classification. Two thirds of patients were assigned to class 3 or worse with severe general or life-threatening diseases. In patients with VO, a higher ASA score is associated with a worse outcome [20]. This was confirmed in our study. We were able to show that an ASA score of > 2 was already an independent predictor of 1-year mortality.

We were able to show that MSSA is a risk factor for early mortality and increases the probability of dying in the first month after diagnosis by fourfold. A reason for this may be the virulence factors of S. aureus, which lead to increased pathogenicity, confer the ability to form abscesses and are associated with more infection-related complications.

In 2014 Aagard et al. analyzed long-term mortality of VO patients surviving the first year after diagnosis and were able to demonstrate that VO caused by S. aureus leads to an increased mortality in bacteremic patients [9]. However, in our study we could not demonstrate an effect of MSSA on late mortality. This may also be due to the virulence of this pathogen, causing patients to die within the first four weeks, rather than in the long-term. This is in accordance to a cohort study of patients with S.aureus bloodstream infection showing a high mortality of 46% within the first year [21].

In a recent study Stangenberg et al. [16] showed a significant influence of bloodstream infection on mortality rates. We were also able to proof this effect as late mortality was found to be significantly influenced by the presence of bacteremia in our analysis. Detecting bacteremia as early as possible is crucial and could therefore reduce late mortality in VO.

Preoperative CKD was a predictive factor for both early and late mortality in VO patients. In a large-scale epidemiological analysis, Issa et al. were able to associate CKD with increased mortality [13]. Aagaard et al. also described a strong relationship between renal failure and mortality in their retrospective study [6]. They observed an increase in mortality risk of more than fivefold. Bains et al. explicitly investigated the influence of CKD on mortality after spinal surgery and attributed a special role to CKD [22]. Optimal treatment of CKD alongside with specific therapy for VO could thus have a significant impact on mortality rates. Factors leading to worsening renal function should be identified and, if possible, reduced or eliminated. These include optimal glycemic control in existing diabetes, blood pressure control and avoidance of nephrotoxic substances [23, 24]. Especially the latter plays an important role in the therapy of VO. Adequate pain therapy is one of the pillars in treatment. In the WHO staging scheme, non-opioid analgesics are provided as basic therapeutics. They can cause additional damage to the kidneys and should be used with caution. Substances used as part of anti-infective therapy should also be evaluated for their nephrotoxicity and, if necessary, administered in renal-adapted doses or alternative drugs should be used [23, 24].

The current study includes a large number of patients and data from a prospective study with uniform 1-year follow-up, providing more detailed and robust information with strict inclusion criteria compared with case series or retrospective cohort studies. By differentiating between early and late mortality, our results give a more detailed view and provide important new clinically relevant insights into influencing factors and the course of this disease.

One limitation of the study is, that it is a single center study. Due to the retrospective design, there is a lack of data in single variables. Since missing data occur comparatively frequently in both groups and in particular the outcome variables mortality and recurrence are present in all patients, we assume that the data are missing at random and have no relevant influence on the analysis. Since our institution is a tertiary care centre, our cohort may have been influenced by referral bias; therefore, the results of the present study may not be generalizable. Because patients were often transferred from our maximum care centre to peripheral hospitals for continuation of anti-infective therapy, the duration of anti-infective therapy could not always be determined. Some parameters that would have been useful for this analysis, such as blood culture clearance or CKD stages, were not recorded in this database, but will be included in future prospective studies. Since the causes of death were not always recorded during follow-up, we unfortunately cannot distinguish whether patients died from reinfection, surgical complications, or general causes of death.