Objectives
This retrospective observational cohort study aimed to characterise and compare the demographics, initial laboratory tests and outcomes between patients with large-joint bacterial septic arthritis (BSA) and tuberculous septic arthritis (TBSA).

Methods
All patients with a culture from a large, native joint growing either non-mycobacterial bacteria or Mycobacterium tuberculosis between 1 January 2012 and 1 October 2018 in our institution were included. Clinical details and admission laboratory values were obtained from patient records. Comparisons were made by Mann-Whitney U, chi-squared tests, and logistic regression analysis.

Results
We identified 64 BSA and 29 TBSA. On average, the BSA cases were older, had higher CRP levels and neutrophil counts and lower albumin levels. The odds ratio for having a BSA was 46 in cases with a CRP greater than 100 mg/L (95% confidence interval (CI) 8.5–850, p < 0.001) and 24 with a neutrophil count greater than 7.5x109 (95% CI 6.1–160, p < 0.001).

51% of BSA were asymptomatic at last follow up compared to 72% of TBSA. 14% of the BSA cases died during admission; there were no deaths in the TBSA group.

Conclusions
Significant differences exist between patients with BSA and TBSA. Whilst no test is sufficient to exclude BSA, a raised neutrophil count or a CRP greater than 100 mg/L significantly increases the odds of a bacterial aetiology. Patients with BSA had worse long-term outcomes and higher incidence of inpatient mortality.