基于主成分分析算法的医院抗真菌药物使用监测模式

Monitoring mode of antifungal drugs in hospital based on principal component analysis algorithm

目的采用主成分分析(PCA)算法分析我院抗真菌药物的用量,探讨医院抗真菌药物使用监测的新模式。方法提取我院住院患者2010-2015年共24个季度21种抗真菌药物的用量数据,建立PCA模型。通过构建主成分得分图开展抗真菌药物使用的整体评价;通过构建主成分载荷图来筛选变化趋势较显著的品种;当某一季度的统计值超出95%控制限时,通过构建变量贡献图来分析用量异常的品种。结果 2010-2015年我院伏立康唑片(每片50 mg)、注射用伏立康唑(每支200 mg)、注射用卡泊芬净(每支50 mg)、注射用卡泊芬净(每支70 mg)、伏立康唑片(每片200 mg)的用量增加较显著;特比萘芬乳膏(每支100 mg)、注射用两性霉素B(每支25 mg)、咪康唑乳膏(每支25 g)、注射用两性霉素B脂质体(每支50 mg)、特比萘芬片(每片250 mg)的用量减少较显著。结论 PCA算法在抗真菌药物使用监测中具有有效性,可为医院抗真菌药物监测提供新的方法。

AIM To explore new mode for antifungal drugs monitoring, and to investigate the utilization of antifungal drugs in our hospital with the application of principal component analysis （PCA） algorithm. METH- ODS The utilization of 21 antifungal drugs in 24 quarters from 2010 to 2015 was extracted and the PCA model was established. The principal component scores plot was applied for global assessment of the antifungal drugs use. The principal component loading plot was applied for screening the antifungal drugs, which showed the most significant upward and downward trend. When the statistic value of a quarter was beyond 95% control limit, the antifungal drugs with abnormal usage were screened according to the contribution plot. RESULTS The use of voriconazole tablet 50 mg, voriconazole for injection 200 mg, caspofungin for injection 50 mg, caspofungin for injection 70 mg and voriconazole tablet 200 mg showed the most significant upward trend, while the use of terbinafine cream 100 mg, amphotericin B for injection 25 mg, miconazole cream 25 g, amphotericin B liposomes for injection 50 mg and terbinafine table 250 mg showed the most significant downward trend from 2010 to 2015. CONCLUSION This case study demonstrates that it is effective to employ PCA algorithm to monitor the clinical utilization of antifungal drugs. This research provides a new scientific tool for the monitoring of antifungal drugs in hospital.