Article Citation: Kocher MS, Mandiga R, Zurakowski D, Barnewolt C, Kasser JR. Validation of a clinical prediction rule for the differentiation between septic arthritis and transient synovitis of the hip in children. J Bone Joint Surg Am. 2004 Aug;86-A(8):1629-35. PMID: 15292409

What we already know about the topic: Differentiation between septic arthritis and transient synovitis can be difficult, because both conditions have similar presenting symptoms. Delayed diagnosis of septic arthritis can lead to poor outcomes.

Why this study is important: Differentiation between septic arthritis and transient synovitis can be difficult. Accurate and early diagnosis of septic joint is critical. Previously, a clinical prediction rule that used 4 independent factors (fever, non-weight-bearing, ESR> 40 mm/hr, and serum WBC >12,000 cells/mm) to differentiate septic arthritis and transient synovitis yield a ROC of 0.96. Given the clinical prediction rule was trained with the original dataset, it was optimized for the original population. Thus, validating it in a new patient population was important.

Brief overview of the study:

The study design was a prospective study at a major children hospital that evaluated patient that presents with acute irritable hip. 213 patients were evaluated. 24 patients were diagnosed with true septic arthritis (WBC≥50,000 cells/mm3 plus positive blood culture), 27 were diagnosed with presumed septic arthritis, (WBC≥50,000 cells/mm3 plus negative blood culture), 103 were diagnosed with transient synovitis (WBC< 50,000 cells/mm3), and 59 were excluded. Exclusion: immunocompromised, renal failure, neonatal sepsis, postop infection, rheumatological disease, Legg-Calve-Perthes, joint fluid insufficiency. The ROC for the current patient population was 0.86 compared to 0.96 in the original population. The paper concludes that although the ROC diminished for the current population compared to that of the original population, the clinical prediction rule had a good diagnostic performance.


The paper used 4 independent factors: fever, non-weight-bearing, ESR> 40 mm/hr, and serum WBC >12,000 cells/mm. Fever is sometimes associated with elevated ESR and WBC. It would be useful to see the correlation between these factors. If some factors are highly correlated, certain combinations may yield more information than others. The ROC of the current population may improve if the clinical prediction rule is weighted for each factor.

Although the sample size is similar to the original study, this clinical prediction rule may benefit from testing on a larger sample size. There were 16 combinations of the four predictors, 51 septic arthritis patients and 103 transient synovitis patients. There were only a few patients in each combination group.

This prediction rule may also benefit from comparing it to clinical gestalt before incorporating it into clinical practice.

Take home points:

Accurate and early differentiation between septic arthritis and transient synovitis is critical. This study validated a previously designed clinical prediction rule with a new patient population. However, validating the clinical prediction rule for different clinical settings may be beneficial.