| Objectives:By mining the U.S.Food and Drug Administration Adverse Event Reporting System(FAERS)database and collecting real-world adverse event data on oxycodone extended-release tablets,we explored the factors influencing its central nervous system adverse events and thus constructed its adverse event prediction platform to assist clinical decision making.Methods:Firstly,a retrospective pharmacovigilance study was used to analyze the adverse events reported by FAERS with oxycodone as the primary suspected drug from Q1 2004 to Q2 2021,and the proportional imbalance method and Bayesian method were used to mine the risk signals of adverse events;secondly,a case-control study was used to collect the hospital information system from June 1,2021 to November 1,2022The data of patients’use of oxycodone extended-release tablets were used to explore their influencing factors associated with adverse events using traditional mathematical and statistical methods and LASSO analysis.Finally,logistic regression models and six machine learning algorithms of decision trees,support vector machines,random forests,gradient boosting decision trees,adaptive boosting and extreme gradient boosting were used to construct models,screen the optimal model and establish a platform for prediction of central nervous system adverse events of oxycodone extended-release tablets.Results:In this study:1)Based on the FAERS database,a total of 5793 reports of central nervous system adverse events were mined for oxycodone alone,and 366,622,and 740 reports for combined benzodiazepines,antidepressants,and anticonvulsants,respectively.Oxycodone alone was found to be associated with myoclonus[ROR=2.92,95%CI=2.28 to 3.76;PRR=2.92,χ~2=77.49;IC=1.52,IC025=0.65;EBGM=2.89,GBGM05=2.33],delirium[ROR=4.69,95%CI=4.24 to 5.21;PRR=4.66,χ~2=1052.64;IC=2.17,IC025=1.81;EBGM=4.50,GBGM05=4.13],psychotic abnormalities[ROR=2.95,95%CI=2.53～3.44,PRR=2.94,χ~2=206.93;IC=1.56,IC025=0.96;EBGM=2.95,GBGM05=2.58],acute central respiratory depression[ROR=2.87,95%CI=2.68to 3.08;PRR=2.82,χ~2=971.62;IC=1.52,IC025=1.33;EBGM=2.87,GBGM05=2.76]were strongly associated;co-administration of benzodiazepines was associated with psychiatric abnormalities[ROR=10.08,95%CI=9.38 to 10.78;PRR=9.90,χ~2=64.06;IC=3.33,IC025=1.65;EBGM=10.08,GBGM05=5.61],tremor[ROR=3.09,95%CI=2.76 to 3.42;PRR=3.08,χ~2=48.93;IC=1.63,IC025=1.17;EBGM=3.09,GBGM05=2.34]had the strongest association;co-administration of antidepressants was associated with delirium[ROR=13.23,95%CI=12.23 to 14.23;PRR=12.87,χ~2=43.86;IC=3.69,IC025=1.36;EBGM=12.23,GBGM05=5.32],drowsiness[ROR=6.74,95%CI=6.15～7.33;PRR=6.73,χ~2=53.42;IC=2.75,IC025=1.52;EBGM=6.73,GBGM05=4.10]had the strongest association;co-administration of anticonvulsants was associated with myoclonus[ROR=17.89,95%CI=17.46 to 18.32;PRR=17.72,χ~2=971.39;IC=4.16,IC025=2.70,EBGM=17.89,GBGM05=12.46],delirium[ROR=4.86,95%CI=4.45 to 5.27;PRR=4.82,χ~2=69.49;IC=2.28,IC025=1.51;EBGM=4.86,GBGM05=3.44]had the strongest association.2)Using traditional mathematical and statistical methods,the influencing factors associated with central nervous system adverse events of oxycodone extended-release tablets were explored,and the results revealed that the occurrence of adverse events may be associated with whether patients are opioid tolerant,whether they are concurrently use of cardiovascular system drugs and gene polymorphism OPRM1 rs1074287.3)Using R4.2.2 software,a prediction model was constructed,in which the extreme gradient boosting classifier performed the best with an accuracy of 0.813 and an AUROC value of 0.899.Therefore,the extreme gradient boosting model was selected to construct a prediction platform for oxycodone extended-release tablets adverse events.Conclusions:The adverse event prediction platform of oxycodone extended-release tablets constructed in this study can provide a reference for the safe and rational use of this drug in clinical practice and help to reduce the risk of adverse reactions of this drug. |
PDF Full Text Request