Focal progressively severe back pain, fevers are rare; MRI is the best diagnostic test.
Even if the patient has a pacer you can get an MRI if you do it right (PubMed).
Seeding from bacteremia (especially in them what use needles), and direct inoculation (especially from back surgery), and occasionally no good reason what so ever (although a history of preceding trauma is very common).
You cannot tell the difference between gram positive or gram negative infections (PubMed) except, course, if you culture the disc space. So get a biopsy first.
De novo infected discs can have anaerobes, including P. acnes. If the MRI is the right type (disc herniation with Modic Type 1) the herniated disc may indeed be infected (PubMed) (Review)(PubMed), although not supported in all studies, as is so often the case (PubMed).
Baaaaaaaad idea. Get a biopsy first. The microbiology and resistance patterns are unpredictable; it is too easy to guess wrong. No bug, no drug. But work with alacrity in areas of the spinal cord: not only can it progress to an epidural abscess but on occasion the infection will cause a vertebral artery thrombosis and then cord infarction. So if the biopsy isn’t going to happen that day, start antibiotics, perhaps an anti-MRSA agent plus a third generation cephalosporin.
And then there is the study where patients with Modic 1 changes on MRI were randomized to under dosed amoxicillin./clavulanic and got more better than placebo (PubMed). My advice, if you have a patient who presents with the MRI and history suggestive of infection, re-read the preceding paragraph.
Surgery leads to better outcomes: "In short, early surgical treatment of pyogenic spondylodiscitis typically achieves a better prognosis, shorter hospitalization period, and subsequent significant improvement in kyphotic deformity and quality of life" (PubMed). Now to make the surgeons a believer.
Once you get an organism it is 6 to 8 weeks of IV therapy directed against the infecting organism. If patient is improving clinically (decreasing pain, ESR) there is no reason to repeat the MRI, it gives no useful information (PubMed).
"...early surgical treatment of pyogenic spondylodiscitis typically achieves a better prognosis, shorter hospitalization period, and subsequent significant improvement in kyphotic deformity and quality of life."(PubMed). Although for every study there is an equal and opposite study "No additional long-term beneficial effect of surgical treatment could be shown in the studies comparing surgical versus antibiotic only treatment." (PubMed). I still lean towards debridement, being a big fan of surgical source control.
Infection ALWAYS goes to the disk then spreads to contiguous bone, destroying end plates in the process; tumor ALWAYS goes to bone and spares the end plates and disk space. You can take that to the bank. To date I have seen one exception.
Relevant links to my Medscape blog