This week’s 5-minute Journal Article discussion covered

“Validation of a Clinical Prediction Rule for the Differentiation Between Septic Arthritis and Transient Synovitis of the Hip in Children” by Kocher et al.


Why is it important to differentiate between septic arthritis and transient synovitis in pediatric patients? Because both diseases can present similarly with acute onset of pain, fever, limp or inability to bear weight and patients holding their hip in the flexed, abducted, externally rotated (FABER) position. The difference is transient synovitis is exactly that, transient, while septic arthritis can lead to permanent joint damage and disability if not treated aggressively with surgical intervention and IV antibiotics.

This study’s goal was to validate the four independent predictors of septic arthritis of the hip in children that Kocher described in his original study, published in 1999 in the Journal of Bone and Joint Surgery, when tested in a new population. The four independent predictors, known as the “Kocher criteria” are:

  1. Fever with temp >38.5C in the past week
  2. Inability to bear weight
  3. Serum WBC > 12,000 cells/mm3
  4. ESR >40mm/hr

According to Kocher’s orginal study, if all 4 of these variables were present, the predicted probability of septic arthritis was as high as 99.6%.

Screen Shot 2018-11-14 at 10.35.07 PM

This validation study ultimately enrolled 213 patients with acutely irritable hip at Boston Children’s hospital, a large tertiary care children’s hospital, between the years of 1997 and 2002. Of the 213 patients, 24 were diagnosed with true septic arthritis (joint fluid with >50,000 wbcs + positive cultures), 27 with presumed septic arthritis (joint fluid with >50,000 wbcs with negative cultures) and 103 with transient synovitis (< 50,000 wbcs in joint fluid, negative cultures, resolution of symptoms without antibiotics and no further disease progression), and 59 were excluded for various reasons.


Screen Shot 2018-11-14 at 10.21.07 PM

The diagnostic performance of the prediction rule was analyzed using a Receiver Operating Characteristic Curve and compared to the original study. The area under the curve for the original study was 0.96, whereas the area under the curve for the validation study was 0.86. The authors argued that this diminished performance was expected in this new population as most clinical prediction rules are optimized towards the original study population. However, an area under the curve of 0.86 still shows very good diagnostic performance for a predictive test and because of this, many EM physicians and orthopedic surgeons still use the “Kocher criteria” for septic arthritis today.


Interestingly, the Kocher criteria does not include CRP as at the time of the study, Boston Children’s Hospital could only test CRP weekly so only about 40% of the participants of the validation study had a documented CRP. Today, it is common to use CRP in addition to serum WBC and ESR when working up a patient for septic arthritis.

Leave a Reply