Risk Factors for Infectious Spondylodiscitis in Patients Receiving Hemodialysis
A retrospective case-control and cohort analysis of hemodialysis patients was done to identify risk factors for spondylodiscitis. These risk factors included bacteremia, receipt of blood products, invasive procedures, and establishment of vascular access. The death rate was greater for case subjects than for control subjects (odds ratio, 2.7).
Received November 28, 2007; accepted March 12, 2008; electronically published May 30, 2008
In patients receiving chronic hemodialysis for end-stage renal disease, spondylodiscitis (ie, infection of the vertebral body and intervertebral disc) is most likely caused by bacteremia resulting from contaminated vascular access devices.1,2 Patients may be immunocompromised because of advanced age, diabetes mellitus, uremia, blood-dialyzer membrane incompatibilities, malnutrition, and/or anemia.3,4
Previous studies reporting risk factors for spondylodiscitis have been observational, without case-control analysis.5 The purpose of this case-control study was to identify risk factors for and outcomes of spondylodiscitis in patients receiving hemodialysis.
Methods
Subjects. This retrospective case-control study reviewed the medical records of hemodialysis patients treated from July 1999 to August 2005 at a tertiary healthcare center and its affiliated regional centers. The study was approved by the University of Manitoba Human Research Ethics Committee. Twenty-two of the 2,647 patients had spondylodiscitis, and they were identified by reviewing the databases of the dialysis units. For each patient with spondylodiscitis, 2 control subjects were selected and matched on the basis of their receipt of hemodialysis for 1 year before and 1 year after the date on which the case patient received a diagnosis of spondylodiscitis; no other factors were considered in the matching process. For case subject and matched 2 control subjects, the zero time point was defined as the date of diagnosis of spondylodiscitis in the case subject.
Medical record review and definitions. A retrospective review of the medical records of the case and control subjects provided information about their demographic characteristics and medical history. A cohort analysis was done for the 22 patients with spondylodiscitis; we collected data on method of diagnosis, treatment, and outcome. Spondylodiscitis was defined as a pyogenic infection of the intervertebral disc and/or the vertebral body, as determined from imaging studies or recovery of microorganisms from a biopsy specimen of the disc and/or vertebral body. Invasive procedures were defined as those diagnostic and therapeutic interventions that could possibly predispose the patient to bacteremia.
Dialysis access was defined as any means of establishing vascular access for hemodialysis, except that peripherally inserted central venous catheters were excluded because of inconsistent documentation in the medical records about placement and use of these catheters. Vascular manipulation was defined as any intervention that involved a catheter, fistula, or graft (eg, contrast study, fistulogram, thrombectomy, angioplasty, or surgical revision or removal), excluding the placement of a new catheter or the exchange of a catheter over a guide wire.
A culture was deemed positive if it grew any microorganism obtained from a swab or tissue sample taken from a wound site, a vascular access site, a catheter tip, urine, sputum, diarrhea, or blood. Bacteremia was defined by a positive blood culture result, regardless of the site of the culture sample.
Patient outcomes were classified as resolved (resumption of daily living activities without major impairment), resolving (incomplete clinical improvement), unresolved (absence of clinical improvement), and death (either related or unrelated to spondylodiscitis).
Data analysis. We performed data analysis using SAS version 9.1 (SAS Institute), which included a univariate analysis with paired Student t tests and paired Wilcoxon rank sum tests (for continuous variables) and a Mantel-Haenszel stratified χ2 test (for categorical variables). Multivariate analysis was not done because of an insufficient sample size. The level of significance was set at P less than or equal to .05.
Results
On average, case subjects received hemodialysis for 2–3 times longer than did control subjects (mean duration, 41 ± 33 vs 18 ± 13 months; P < .0008). The demographic characteristics were similar between case and control groups (ie, mean age, sex, ethnicity, alcohol and/or tobacco use, mean body mass index, place of residence, mean duration of diabetes, mean level of hemoglobin A1c, and/or prevalence of hypertension, diabetes, peripheral neuropathy, nephropathy, or foot ulcers). The reason for hemodialysis, the number of sessions per week, and the average time per dialysis session were also similar between case and control groups.
Of the 22 patients with a diagnosis of spondylodiscitis, 15 (68%) were receiving hemodialysis by use of a central venous catheter, and 16 (73%) had a permanent vascular access site for hemodialysis.
The most common symptoms of spondylodiscitis were back pain (20 patients [91%]; median duration, 2 months), fever (11 patients [50%]), and neck pain (3 patients [14%]), and symptoms were present for a mean of 59 days (range, 18–435 days) before diagnosis. The majority of patients (15 [68%]) had imaging- and/or biopsy-proven involvement of the lumbosacral region, 6 [27%] had involvement of the thoracic region, and 1 [5%] had involvement of the cervical region. Almost all patients (21 [95%]) had involvement of both the vertebral and disc space; 1 patient (5%) had isolated discitis. Twelve patients (55%) had a total of 16 extraspinal manifestations of spondylodiscitis, including a paraspinal, psoas, or epidural abscess.
Of the 22 patients with a diagnosis of spondylodiscitis, blood culture results were positive 3–6 months before diagnosis for 6 (27%) of the patients and within 3 months before diagnosis for 18 (82%) of the patients, with a variety of microorganisms isolated (Table 1). Only 5 patients (23%) had microorganisms recovered from culture of biopsy specimens (biopsy was performed for 12 [55%] of the 22 case subjects); for all these patients, a similar microorganism had been recovered from a blood culture within the 6 months before diagnosis (Table 1).
In 15 patients (68%), a dialysis access device was implicated as the source of infection—including central venous catheters in 13 patients (59%)—on the basis of medical records and culture results. A significantly greater proportion of case subjects than control subjects received blood products, had an invasive procedure, had the establishment of a vascular access site, or had placement of a temporary or permanent internal jugular or temporary femoral catheter during the months immediately before the diagnosis of spondylodiscitis (Table 2). There was no difference between case and control subjects in the proportion who underwent surgical procedures, vascular access manipulations, arteriovenous graft placements, or placement of a permanent femoral, temporary subclavian, or permanent subclavian catheter during the year before diagnosis of spondylodiscitis (data not shown). The proportion of patients requiring an arteriovenous fistula 6–12 months before diagnosis was significantly smaller in the group of case subjects than in the group of control subjects (Table 2). A significantly greater proportion of case subjects than control subjects used any temporary central venous catheter or a permanent subclavian catheter, and a smaller proportion of case subjects than control subjects used an arteriovenous fistula for vascular access at the time of diagnosis of spondylodiscitis (Table 2).
During the 3 months before the diagnosis of spondylodiscitis, a significantly greater proportion of case subjects than control subjects had a positive culture result from any sample, which included blood, urine, stool, catheter tips, and specimens from wound sites, vascular access sites, debrided tissue, amputated tissue, and the upper respiratory tract (19 [86%] of 22 case subjects vs 16 [36%] of 44 control subjects; P ⩽ .0007). Likewise, a significantly greater proportion had positive blood culture results (18 [82%] case subjects vs 6 [14%] control subjects; P ⩽ .0001) and received antimicrobial therapy (ie, any antimicrobial therapy received, regardless of course) (21 [95%] case subjects vs 21 [48%] control subjects; P ⩽ .0007).
After the diagnosis of spondylodiscitis, antimicrobial therapy was given to all 22 (100%) patients for a mean duration (±SD) of 21 ± 14 weeks (Table 1). Vancomycin, the most frequently used antimicrobial agent, was given to 17 patients (77%) (Table 1). Nine patients (41%) were treated with an orthosis (thoracolumbosacral orthosis or cervical collar), and 7 (32%) patients had surgery.
A follow-up evaluation as of July 2006, 1 year after the last included patient with a diagnosis of spondylodiscitis, showed that only 9 (41%) of the 22 patients were still alive; spinal infection was unresolved in 6 and resolved in 3 of these patients. For the 13 patients (59%) who had died, the cause of death was not noted; at the time of death, spinal infection was unresolved in 10 and resolved in 3 of these patients. One year after inclusion in the study, the proportion of deaths from all causes after diagnosis of spondylodiscitis was significantly greater among case subjects than among control subjects (13 [59%] vs 12 [27%]; odds ratio, 2.7 [95% confidence interval, 1.0–7.0]; P ⩽ .03).
Discussion
The longer duration of hemodialysis associated with spondylodiscitis may have resulted in an increased susceptibility to infection because of episodes of occult or overt vascular access–related sepsis and longer exposure to more vascular access devices, new device placements, blood products, and other invasive procedures that may increase cumulative risk of bacteremia.6
Potential sources of bacteremia and spinal infection included receipt of blood products, an invasive procedure, and the establishment of a vascular access site (Table 2). Spondylodiscitis was more frequently associated with central venous catheters than with arteriovenous fistulas or grafts, a finding consistent with previous studies showing that dialysis by use of a temporary central venous catheter had the highest risk of bloodstream infection and dialysis by use of an arteriovenous fistula had the lowest.1,3,7-9 Vascular access using an arteriovenous fistula is preferred to access using a central venous catheter,10-12 and the infection rate associated with use of a central venous catheter is double that for all other forms of access combined.7 However, exhaustion of available access sites may necessitate use of a temporary or permanent dialysis catheter.
The concordance between the microorganisms recovered from cultures of blood and of biopsy samples supports the hypothesis that spondylodiscitis in this patient population is hematogenous in origin and likely arises from use of vascular access devices.1,12 On the basis of their phenotypic characteristics, the microorganisms recovered from the biopsy samples most likely were the same microorganisms previously identified in the blood culture samples.
Limitations of this study included those inherent in a retrospective review; the incidence of spondylodiscitis may have been underestimated: subclinical cases of spondylodiscitis may have been undetected, particularly in patients treated with antibiotics for bacteremia who may have had concurrent back pain but were not evaluated for it. Microbiological diagnosis of spondylodiscitis may have been confounded by receipt of antimicrobial treatment prior to diagnosis in most patients. Furthermore, exclusion of peripherally inserted central venous catheters from the analysis may have resulted in an incomplete description of the source of bacteremia.
Acknowledgments
We thank Dr. John Conly for reviewing our manuscript and for helpful comments.
Financial support. This work was generously supported by the Dr. John Adamson and Dr. Sanford T. Fleming Studentship, University of Manitoba (to R.M.H), and by the Manitoba Renal Program.
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
References
- 1. Taylor G, Gravel D, Johnston L, Embil J, Holton D, Paton S. Prospective surveillance for primary bloodstream infections occurring in Canadian hemodialysis units. Infect Control Hosp Epidemiol2002;23:716–720.
- 2. Taylor G, Gravel D, Johnston L, Embil J, Holton D, Paton S. Incidence of bloodstream infection in multicenter inception cohorts of hemodialysis patients. Am J Infect Control2004;32:155–160.
- 3. Hoen B, Kessler M, Hestin D, Mayeux D. Risk factors for bacterial infections in chronic haemodialysis adult patients: a multicentre prospective survey. Nephrol Dial Transplant1995;10:377–381.
- 4. Ryan SV, Calligaro KD, Dougherty MJ. Management of hemodialysis access infections. Semin Vasc Surg2004;17:40–44.
- 5. Hadjipavlou AG, Mader JT, Necessary JT, Muffoletto AJ. Hematogenous pyogenic spinal infections and their surgical management. Spine2000;25:1668–1679.
- 6. Tsuchiya K, Yamaoka K, Tanaka K, Sasaki T. Bacterial spondylodiscitis in the patients with hemodialysis. Spine2004;29:2533–2537.
- 7. Abdulrahman IS, Al-Mueilo SH, Bokhary HA, Ladipo GOA, Al-Rubaish A. A prospective study of hemodialysis access-related bacterial infections. J Infect Chemother2002;8:242–246.
- 8. Cheesbrough JS, Finch RG, Burden RP. A prospective study of the mechanisms of infection associated with hemodialysis catheters. J Infect Dis1986;154:579–589.
- 9. George A, Tokars JI, Clutterbuck EJ, Bamford KB, Pusey C, Holmes AH. Reducing dialysis associated bacteraemia, and recommendations for surveillance in the United Kingdom: prospective study. BMJ2006;332:1435.
- 10. Jindal K, Chan CT, Deziel C, et al. Hemodialysis clinical practice guidelines for the Canadian Society of Nephrology. J Am Soc Nephrol2006;17(3 Suppl 1):S1–S27.
- 11. O'Grady NP, Alexander M, Dellinger EP, et al. Guidelines for the prevention of intravascular catheter-related infections. MMWR Recomm Rep2002;51(RR-10):1–29.
- 12. Philipneri M, Aly ZA, Amin K, Gellens ME, Bastani B. Routine replacement of tunneled, cuffed, hemodialysis catheters eliminates paraspinal/vertebral infections in patients with catheter-associated bacteremia. Am J Nephrol2003;23:202–207.
-
Address reprint requests to John Embil, MD, FRCPC, Infection Prevention and Control Unit, Health Sciences Centre, MS 673-820 Sherbrook Street, Winnipeg, MB R3A 1R9, Canada (jembil@hsc.
mb. ca).