Independent Analysis of a Clinical Predictive Algorithm to Identify Methicillin-resistant Staphylococcus aureus Osteomyelitis in Children

Abstract

The number of serious, life-threatening musculoskeletal infections in children due to methicillin-resistant Staphylococcus aureus (MRSA) infections is increasing. The early identification of the bacteria causing osteomyelitis is critical to determine the appropriate antibiotic treatment. A recent study proposed a clinical algorithm to predict which infections were caused by MRSA by stratifying basic clinical values at the time of admission for children with osteomyelitis. The purpose of this study is to apply that predictive algorithm on an independent patient population to determine its wider applicability. This was a retrospective chart review at a tertiary care children's hospital. All children who were treated for a culture-positive osteomyelitis were identified over a 3-year period. The previously reported predictors, determined by multivariate regression analysis, of MRSA infection (temperature >38°C, hematocrit <34%, white blood cell count >12,000/µL, and C-reactive protein >13 mg/L) were determined for each patient. The number of positive predictors was then correlated with the percentage of cases that were MRSA positive. A total of 58 patients with culture-positive osteomyelitis were identified from 2008 to 2010. Sixteen of the infections were caused by MRSA (overall 26%). The percentage of patients with MRSA osteomyelitis according to the number of risk factors were as follows: all 4 risk factors, 50% (1 out of 2 patients); 3 risk factors, 42% (5 out of 12 patients); 2 risk factors, 21% (4 out of 19 patients); 1 risk factor, 50% (6 out of 12 patients); and 0 risk factor, 0% (0 out of 13 patients). The previously reported clinical predictive algorithm had a relatively poor diagnostic performance in this independent patient population. Specifically, the percentages of MRSA were the same for 1 risk factor compared with 4 (50%). Differences in bacteria strain, host responses, and a variety of other confounding variables could be responsible for these differences. Specific genetic markers may be the best early test to identify MRSA infections in the future. Level III-case-control series.