Predictive algorithm works for one hospital…but will it work for others?

Quickly identifying whether osteomyelitis is caused by methicillin-resistant Staphylococcus aureus (MRSA) or methicillin-sensitive S aureus (MSSA) is critical to delivering appropriate treatment. During the 2011 annual meeting of the Pediatric Orthopaedic Society of North America, Mininder S. Kocher, MD, MPH, director of the clinical effectiveness research unit at Children’s Hospital Boston, reported on a prediction algorithm developed at one hospital that could help quickly differentiate between MRSA and MSSA osteomyelitis.

The retrospective chart review, conducted at Children’s Hospital Boston, focused on 129 children (ages 1–18 years) treated for S aureus osteomyelitis, as determined by positive cultures from bone and surrounding tissue, from Jan. 1, 2000, to May 31, 2009. Of these, 118 were positive for MSSA and 11 were positive for MRSA.

Data obtained for each patient included age, gender, symptoms, duration, current antibiotic use, past medical history, hospitalizations, vital signs (temperature, heart rate, and blood pressure), laboratory results (serum white blood cell count [WBC] and differential, hematocrit, platelet count, erythrocyte sedimentation rate [ESR], C-reactive protein [CRP]), and the results of any imaging studies.

The search for predictors
Researchers then compared the two groups based on 16 different variables and found that MRSA patients had a greater use of antibiotics, were less likely to be weight bearing, had a higher temperature, and had a faster heart rate than MSSA patients.

“We then looked for independent predictors of MRSA using multivariate logistic regression,” said Dr. Kocher. They identified the following four independent predictors of MRSA:

• temperature greater than 38 degrees C
• hematocrit less than 34
• serum WBC greater than 12,000
• CRP greater than 13

“More than half of the MRSA patients had all of these predictors,” he continued, “and none of the MSSA patients had all four (Fig. 1). “Based on this algorithm, the probability that a patient has MRSA is high if the patient has all four predictors, and very low if the patient has zero or one predictor.”

He also cited previous research that generally supported the findings of this study, particularly with regard to the higher WBC counts and CRP levels in children with MRSA osteomyelitis compared to those with MSSA osteomyelitis.

A model for other conditions?
Dr. Kocher noted that a previous study following this same model had been conducted to create a clinical prediction rule for distinguishing between septic arthritis and transient synovitis of the hip in children. That rule was validated in a new population and formed the basis for a clinical practice guideline on the treatment of septic arthritis in children.

“Now, turning our attention toward the challenge of differentiating between MRSA and MSSA osteomyelitis, we have taken the initial step by developing this predictive probability algorithm to help facilitate the timely selection of appropriate antibiotics,” he said.

Because MRSA produces more virulent and invasive infections than MSSA, Dr. Kocher noted the importance of being able to quickly differentiate between the two. But, he noted, follow-up studies, including prospective validation studies, should be conducted to evaluate the accuracy and utility of this algorithm before it is put into clinical practice.

“It’s a tool that helps us stratify risk levels,” he said, “not a substitute for clinical treatment.”

Dr. Kocher’s coauthors for “Evidence-based Clinical Prediction Algorithm for Differentiating between MRSA and MSSA Osteomyelitis in Children” are Kevin L. Ju, MD, and David Zurakowski, PhD.

Interview with Mininder S. Kocher, MD, MPH, on a clinical prediction algorithm for differentiating between MRSA and MSSA (mp3)

Disclosure information: Dr. Kocher—Biomet; OrthoPediatrics; PediPed; Smith & Nephew; Fixes 4 Kids; Pivot Medical; WB Saunders; Journal of Bone and Joint Surgery–American; Journal of Shoulder and Elbow Surgery; ACL Study Group; American Orthopaedic Society for Sports Medicine; Herodicus Society. Drs. Ju and Zurakowski—no conflicts reported.

Mary Ann Porucznik is managing editor of AAOS Now. She can be reached at porucznik@aaos.org

Bottom Line

• A retrospective chart review at a single hospital was used to develop a clinical prediction algorithm to help distinguish between MRSA and MSSA osteomyelitis in children.
• Four independent predictors were found, providing a high probability of MRSA osteomyelitis if the child possesses all four and a very low probability if none are present.
• These results are based on 11 children with MRSA osteomyelitis, compared to 121 children with MSSA osteomyelitis.
• Results may not be reproducible at other hospitals or in other settings; prospective validation studies are needed before this algorithm is applied in clinical practice.