Trial Casts Doubt on Antibiotics for Refractory Low Back Pain

Patients with lower back pain related to spinal disc herniation got no more relief from the antibiotic combination of amoxicillin and clavulanate than from placebo in a randomized trial, creating new confusion around this approach to the common and often stubborn condition.

With 170 patients randomized to amoxicillin-clavulanate or placebo, reductions in pain scores were virtually identical 12 months after starting a 3-month course of treatment, according to Donna M. Urquhart, PhD, of Monash University in Melbourne, Australia, and colleagues. This lack of difference was seen not only for the overall sample but also for a subset of patients with so-called Modic changes (i.e., vertebral bone degeneration).

Meanwhile, adverse events were considerably more frequent in the antibiotic-treated group (40.0% vs 23.5% with placebo), the group reported in JAMA Network Open.

That, along with the public health risks associated with antibiotic overuse, prompted the authors to recommend against further use of antibiotics to treat lower back pain.

But that probably isn’t the end of the story, insofar as previous studies did show benefits from amoxicillin-clavulanate in this population. Most notably, a similarly sized, randomized, placebo-controlled trial from Denmark reported in 2013 found significantly (clinically as well as statistically) greater relief from pain and disability with the antibiotic combo at the 1-year mark. Lumbar pain scores averaged 2.6 points lower than in the placebo group on the standard 0-10 scale.

While the designs differed in some respects — the amoxicillin-clavulanate dosage was 50% higher in the Danish trial and treatment lasted 100 days versus 90 in the new Australian study, for example — it remains unclear why the results differed so markedly.

Urquhart and colleagues suggested that lower treatment adherence in their study as well as the geodemographic differences in the two samples may have worked to diminish the treatment response. They also argued that “given evidence from meta-analyses and reviews that suggest patients may benefit from antibiotics where the risks outweigh the benefits,” including the promotion of antibiotic resistance that comes from overly liberal prescribing, this treatment approach should probably be abandoned.

Why consider antibiotics at all for treating lower back pain? This concept originated with findings that herniated discs may harbor low-virulence bacteria such as Propionibacterium acnes and Corynebacterium propinquum, and it’s been hypothesized that these subtle infections may underlie Modic changes, which in turn lead to pain. Amoxicillin, either alone or combined with clavulanate, has therefore been used to treat patients whose lower back pain hasn’t responded to more conventional therapy and with imaging suggestive of vertebral bone involvement.

For the new study, patients with persistent lower back pain and MRI confirmation of disc herniation were enrolled and assigned in equal numbers to amoxicillin-clavulanate at 500/125 mg or to placebo, given twice daily for 90 days. (In the previous Danish trial, the same dosage was given three times a day.) About half of the overall sample were found to have Modic changes (mostly type 2; the Danish study only enrolled patients with type 1).

Primary outcome measures included the Low Back Pain Rating Scale (LBPRS) and a 100-point visual analogue pain intensity rating. Other outcomes included scores on the Roland-Morris Disability Questionnaire, as well as measures of work impairment and absenteeism.

Mean patient age was about 44, and some 60% were men. They had suffered from lower back pain for an average of 10 years, and two-thirds also reported pain extending into the legs. LBPRS scores averaged about 5.7, and 44 was the mean pain score on the 100-point scale. Roland-Morris disability scores averaged 7.6 in the antibiotic group and 6.9 among those assigned to placebo. Some 11% of participants missed the 1-year follow-up.

Both arms saw improvements in pain and disability over the trial’s 12 months, such that between-group differences were minimal to nonexistent. Mean LBPRS scores, for example, declined by about 1 point with both treatments, for an adjusted difference of just 0.06 points at 1 year (95% CI -0.58 to 0.70). Similarly, Roland-Morris scores declined by 2.5 points with antibiotics versus 1.7 points with placebo (adjusted difference -0.54, 95% CI -1.79 to 0.72).

All other outcomes also showed no significant differences between groups. Moreover, the same pattern was seen for the roughly half of participants with Modic changes (although the investigators acknowledged that the study wasn’t powered to reach conclusions on efficacy in this subgroup).

Limitations included the particulars of the treatment regimen and inclusion/exclusion criteria used in the study, as well as the 11% lost-to-follow-up rate. Urquhart and colleagues pointed out, too, that they recruited patients primarily through advertising as opposed to referrals from hospitals or specialty clinics. Finally, the study had originally planned to check for antibiotic resistance by taking fecal samples and swabs at different points on the body, but the researchers quickly abandoned this effort when they found it was inhibiting recruitment.