How can the labelling and the packaging of drugs impact on drug safety? Prof. Pascal BONNABRY. Head of pharmacy. Swissmedic, Bern, June 19, 2007

Transcription

1 How can the labelling and the packaging of drugs impact on drug safety? Head of pharmacy Swissmedic, To err is human (USA) Serious adverse events in 3% [ %] of hospitalizations 10% [ %] of adverse events led to death Extrapolation: to deaths each year in the USA (medication errors: 7 000)! 8 th leading cause of death (motor vehicle accidents , breast cancer , AIDS ) Equivalent of a BOEING 747 crash every 2 days... To err is human, IOM,

2 Where are the errors? Avoidable adverse events in 6.5% of hospitalizations Bates DW, JAMA 1995;274:29 Administration error Dispensation error AND Controls The right strategy : Improve the prevention ( probability of selection errors) AND Improve the detection ( probability of control ) Ph.Garnerin, HUG,

3 Contributive factors Improvement? Process organisation Human reliability Ergonomy of drugs Re-engineering Education, but limited impact ergonomy, based on objective data Human reliability «On the 6th day, God created man» but God was tired, and his creation was not perfect 3

4 Dispensation errors (manual) Nurses 3,0 % dispensation errors Control was not tested 20% 6% Selection error Repartition error Counting error 74% Pharmacy 3,6 % dispensation errors 79% detected during control 21% 8% 35% Wrong drug Wrong dosage Wrong form Others 36% Meier B, HUG, 2001 Cina JL, Jt Comm J Qual Patient Saf 2006;32:73 Selection errors 2% Efficacy of controls Introduction of errors during unit dose dispensing Detection ability during human-performed control: Pharmacists: 87.7% Nurses: 82.1% Facchinetti NJ, Med Care 1999;37:39-43 Efficacy 85% (known value in the industry) Do not be too confident with the double-checks! 4

5 Calculation errors Resolution of usual calculations from the medical field Mean error rates Anaesthetists: 10.4% (n=28) Nurses: 26.7% (n=30) % volunteers % 9% 18% 27% 36% 45% 72% error rate Meier B, HUG, 2001 Ergonomy of drugs Look-alike Sound-alike Information on labels Tall Man Letters Colour code Information on unit doses Barcode Impact on selection error Impact on control performance 5

6 Look-alike Infovigilance Is the risk taken in consideration? 6

7 Sound-alike Prava - Pravastatine Selipran - Seropram Dalmadorm - Dormicum Antra - Arixtra Carbostésine - Carbostésine Adrénaline Taxol - Taxotère AZT: azathioprine, zidovudine Information on label: injectable drugs 3 levels of standardisation For each level, 24 on-screen tasks For each task: 2 pieces of information of a label are displayed (concentration, quantity or volume) Select the correct answer among 10 propositions 5 professional groups of 15 volunteers each 7

8 Level 1: No standardisation 2 pieces of information Variable position Calculation necessary Concentration in % or mg/ml Level 2: Complete information 3 pieces of information Variable position No calculation Concentration in mg/ml 8

9 Level 3: Fixed sequence 3 pieces of information Fixed position No calculation Concentration in mg/ml Standardisation level 29.7% [ ] 6.4% [ ] 4.1% [ ] Error rate [%] info / 3 3 info, variable position 3 info, fixed position OR 1vs3= 12.1 [ ], p<0.001 OR 2vs3= 1.7 [ ], p=0.001 n=4968 9

10 Concentration (level 1) Error rate [%] % [ ] 24.4% [ ] % mg/ml OR = 1.8 [ ] p<0.001 Professional groups Error rate [%] Ward nurses ICU nurses Physicians Nurses anaesthetists Anaesthetists OR: Ward Nur vs Anaesth 13.2 ICU Nur vs Anaesth 11.4 Phys vs Anaesth 8.3 Anaesth nur vs Anaesth Overall 2 info / 3 3 info, variable position 3 info, fixed position 10

11 Interindividual variability from 0 to 40% Label vs Human Error rate Not standardised Very standardised Less specialised 43% 6% Most specialised 8% 1.5% 6 à 7 4 à 5 11

12 Tall Man Letters Filik R, Human Factors 2006;48:39-47 Colour-code No convincing data for or against... Church A: The dollar story Church B: The parking story Study actually under progress at HUG 12

13 Colour code Method Selection of syringes in trays (76 tasks) 16 anaesthetists, 16 anaesthetists nurses Tsoncheva D, HUG 2007 Colour code Analysis Done Detection Detected Control Failed Wrong Selection Omitted ERROR Right NO ERROR NO ERROR ERROR 13

14 Colour code Results (n=2432) Colour No No Yes Yes Intrasimilarities Similarities No Yes No Yes No Selection errors (%) Control omission (%) Control (%) Administration errors (%) Yes Yes Yes Tsoncheva D, HUG 2007 Information on unit doses Prescription: Omed 20mg 1x/d? 14

15 From very bad to very good Administration error rate If 100% of controls are possible Selection error 2% First control 15% AND Administration error Final control 15% 0.05% 15

16 Administration error rate If 0% of controls are possible Selection error 2% First control 100% AND Administration error 2% Final control 100% UD identifiable Control X% 15% 100-X% 100% «Drug-mix»? AND? AND in real-life??? Selection error 2% OR First control Y% AND UD NOT identifiable OR Final control Y% Control Administration error Z% 16

17 UD identifiable Control UD NOT identifiable Control If 50% of controls are possible 50% Selection error 2% AND 15% OR First control 57% AND 50% OR Final control 57% AND 100% Administration error 0.7% Administration error rate 2% Administration error rate 1% 0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Unit dose identification rate Hypothesis 50% 0.7% error rate 0.05% 17

18 Extrapolation for a 2200 bed hospital? Yearly number of doses dispensed Extrapolation from HUG statistics Ratio errors : serious ADE 100 : 1 Bates D, J Gen Intern Med 1995;10:199 Cost / serious ADE $4685 Frs Leape L, JAMA 1999;281:267 Estimated HUG events/year UD identification Errors reaching patients Significant ADE 0% % %

19 Estimated HUG cost/year HUG cost / year [Mio Frs] Hypothesis 50% 8 Mio Frs /yr 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Unit dose identification rate 0.55 Barcode on unit doses Datamatrix Safety Traceability Product ID Batch number Expiry date Small size 19

20 The benefit of bedside scanning BPOC impact Wrong drug - 75% Wrong dose - 62% Wrong patient - 93% Wrong administration time - 87% Globally - 80% Johnson, J Healthcare Inf Manag 2002;16:1 The scanning in Geneva Pilot study with chemotherapies CYTOS-TRACE (database) Caregiver ID (RFID) Drug ID (RFID) Patient ID (RFID) 20

21 What can Swissmedic do? Bad labelling risk for patients! Objective data demonstrating the impact of some measures begins to be available The authorities must apply these results in their regulations : Sound-alike and look-alike Labelling of drugs (injectables in priority) Identification of unit-doses Barcoding of unit-doses Expert group with hospital pharmacist? AFSSAPS (France) Injectable drugs: labelling harmonization Longitudinal writing Concentration + total quantity + total volume Suppression of % (except for electrolytes) Standardized positions on the label Colour-code for some products Tall man letters for some products 10/morphine/reco_harmon_isation_etique_tage.pdf AFSSAPS,

22 Application in hospital pharmacy EAHP request for unit-doses Unit doses blisters, with each single dose containing the whole information Trade name Active substance Dosage Expiry date Batch number Barcode Including product ID, expiry date and batch number Use of a recognized international standard (i.e GS1) Datamatrix EAHP,

23 Council of Europe Unit-doses identification and barcode Expert group on safe medication practices, 2007 FDA Barcode on unit-doses Product identification mandatory (batch number and expiry date encouraged) Mandatory since April 26, 2006 Prevention of adverse events / year? FDA,