Evaluation du contrôle libératoire en utilisant un programme de simulations d’erreurs
(1) Pharmaceutical Technology Department, Bordeaux university hospital (CHU de Bordeaux), France.
(2) ARNA Laboratory-ChemBioPharm U1212 INSERM - UMR5320, CNRS - University of Bordeaux, France
Control release of preparations is mandatory referred to French good manufacturing practices(1) even for taylor-made cytotoxic preparations. However, the release process is based mainly on human multiple checks. Our work aimed to assess the reliability of pharmacists involved in this control using simulation program including intentional non-conformity (NC).
Material and methods
Simulation was based on 55 fictive preparations including NC (x20) or not (x35) were made and control release was carried out by 7 pharmacists. This assessment was carried out by senior trained (n= 4 ) and junior untrained pharmacists (n= 3 ).
NC includes : errors related to drug (e.g. wrong dose, wrong final concentration), wrong diluent (e.g. Nacl vs glucose), wrong administration device (e.g lack of required filter), lack of checks controls (drug, diluent device), errors related to co-administrated therapy (e.g. lidocaine or MESNA forgotten), errors in labelling (e.g label exchanges, errors of beyond to use date), and final packaging (e.g. UV protect). Among errors, 15 were considered as major errors (potential impact on the patients) and 5 were considered as minor (no impact on the patient). In addition, mean time for control release per preparation was determined.
Results and Discussion
Among 20 fictive NC, 8 were detected by all staff members and 12 were not detected by the whole staff members. NC undetected were : wrong final packaging (n=4), wrong administration device (n=3), wrong diluent volume (n=6), NC in co-administrated therapy (n=6), and drug NC (n=8). NC detected by all pharmacists were labelling NC (n=8). NC about labelling seemed to be easier to detect. An uneglitable number of errors were undetected by the pharmacist.
Among the 12 errors inconstantly detected, 9 were considered as major errors with potential health impact on patients.
Mean NC undetected by trained pharmacists was 1.5 while mean NC undetected by untrained pharmacist was 5.33 confirming the need to develop specific training for pharmaceutical control release.
Mean time for control release was : 32,6 ± 15,6s per preparation (max 90s, min 11s). Junior’s mean time was 37,2 ± 17,9s (max 90s, min 11s) while senior’s mean time was 27,4 ± 11,4s (max 60s, min 12s).
Untrained pharmacists were less efficient in NC detection. However, whatever the status, uneglitable number of errors were not detected by all pharmacist justifying implementation of an education training program for pharmaceutical control release validated by the simulation program based on errors simulation.