Derivation and Validation of a Risk Score for Methicillin-Resistant Staphylococcus aureus Bloodstream Infection

Compared with methicillin sensitive Staphylococcus aureus (MSSA), the prognosis of patients with methicillin resistant Staphylococcus aureus (MRSA) bloodstream infection is poor. Therefore, the construction of MRSA and MSSA identification model has certain value for the selection of antibiotics and treatment outcome control. This study aimed to derive and validate a simple risk prediction model for MRSA bloodstream infection in Chinese patients.

Three hundred and thirty-five patients with Staphylococcus aureus (S. aureus) bloodstream infection were retrospectively analyzed and divided into three groups. The first group was used for the risk score derivation (n = 163), the second group was used for internal validation (n = 80), and the third group was used for external validation (n = 92). According to the odds ratio (OR) obtained from multivariate logistic regression, the risk prediction model for MRSA bloodstream infection was established, and the prediction efficiency of the model in three cohorts were evaluated.

Hospital stay before BSI ≥ 7 days, hospital acquired BSI, infection source ≥ 2 sites, indwelling gastric tube before BSI and carbapenems used before BSI and after admission were independent influencing factors of MRSA in the derivation group, the above influencing factors were scored 3, 5, 4, 3, and 3, respectively. The derivation, internal and external validation groups showed adequate discrimination (the AUCs were 0.788, 0.780, and 0.742, respectively) and good calibration (H-L tests were χ2 = 3.896, p = 0.306; χ2 = 4.221, p = 0.298; and χ2 = 3.974, p = 0.352, respectively). The risk scores were further divided into very low-risk (score 0 - 3), low-risk (score 4 - 7), high-risk (score 8 - 12), and very high-risk (score ≥ 13) layers.

The simple risk score model for predicting MRSA bloodstream infection has good predictive effect, high predictive accuracy, and good clinical applicability, which can help clinicians choose sensitive antibiotics and reduce the adverse prognosis of patients.