Development of a prediction model for antimicrobial stewardship pharmacy consultations to identify high-risk pediatric patients: a retrospective study across two centers.

BACKGROUND

Antimicrobial Stewardship Pharmacy Consultation (ASPC) in China has been shown to reduce patients' length of stay (LOS). However, prolonged LOS remains a challenge, resulting in unnecessary psychological and financial burden for patients.

OBJECTIVE

This study aimed to develop a prediction model using ASPC parameters to identify high-risk pediatric patients with infectious diseases. These patients received ASPC interventions but still experienced prolonged LOS, which defined their high-risk status.

METHODS

Predictors for the ASPC model were selected using lasso regression, a nomogram was developed using multivariate logistic regression, and internal validation was performed using tenfold cross-validation. The data set consisted of 474 electronic medical records of pediatric patients with infectious diseases from two hospitals. LOS was dichotomized at the median, and patients with LOS greater than the median were considered to have achieved the outcome.

RESULTS

The proportion of outcome events was set at 50% by design. Five independent predictors were identified in the ASPC model: (1) the suggestions from the crucial consultation (OR: 1.74; 95% CI: 1.10 to 2.74), (2) weight (OR: 0.98; 95% CI: 0.97 to 1.00), (3) whether the patient received first aid (OR: 0.54; 95% CI: 0.3 to 1.00), (4) the aim of the crucial consultation (OR: 0.15; 95% CI: 0.03 to 0.66), and (5) whether the patient was critically ill (OR: 0.22; 95% CI: 0.12 to 0.41). The ASPC model showed good discrimination with a C-statistic of 0.772 (95% CI: 0.748 to 0.797) and good calibration performance with intercept and slope values of 0.00 (95% CI: -0.12 to 0.12) and 0.93 (95% CI: 0.82 to 1.04), respectively, under tenfold cross-validation.

CONCLUSIONS

The antimicrobial stewardship pharmacy consultation model has good discrimination and calibration, and effectively identifies patients at risk for prolonged length of stay.