20110719 mcguinness deborah_ontologies_for_the_real_world_microsoft_faculty_summitfinal
Www.huntingdon.com Fiona McGuinness Investigating Out of Specification Results 1.
-
Upload
daisy-thompson -
Category
Documents
-
view
214 -
download
0
Transcript of Www.huntingdon.com Fiona McGuinness Investigating Out of Specification Results 1.
www.huntingdon.com
Fiona McGuinness
Investigating Out of Specification Results
1
www.huntingdon.com
Introduction
g What do we mean by OOS, OOT, atypical or unexpected results?
g Documented procedureg Investigationg What happens when you can’t find the cause?g Examplesg Questions
2
www.huntingdon.com
What do we mean…?
g Out of Specification (OOS)n Result that does not comply with stated accepted criteria
g Out of trend (OOT)n A stability result that does not follow the trend of other
stability results during a stability studyn A result that does not follow the trend based on data
already collected with the same compoundg Atypical
n May be within specification but unexpected or questionableg All unexpected results!
3
www.huntingdon.com
Documented Procedure
g Important that all laboratories have a fully documented OOS investigation proceduren Clearly state what the investigation will be based
on, what you will do in each case, actions and outcomes from the investigation
n Clients and regulatory agencies like to see a procedure and that it is being followed!
4
www.huntingdon.com
Initial Investigation
g First step is to QC Check the data generated to ensure that no identifiable calculation or preparation errors have been made
g Investigate the analysis….n Calibration standards, extraction volumes, dilution
steps etcn Instrument set-up, injection volumes, mobile
phases, diluents, instrument working…
5
www.huntingdon.com
Re-dilution
g An investigative runn Has a dilution error been made?
g What if dilution results are different?g Don’t cherry pick results!
n Confirm results by a 3rd dilution!
6
www.huntingdon.com
Preparation of the Formulation
g Check the request, method, weights of test substance and vehicle as appropriate.
g Check calculations
7
www.huntingdon.com
Contingency Analysis
g No obvious analytical error (including dilutions), no obvious preparation error…
n Perform analysis of contingency samples
8
www.huntingdon.com
What happens when we can’t find the cause of an unexpected result?
g Meetingg Root Cause Analysis investigationg Corrective and Preventative Actionsg Analysis of next formulated occasiong Discussion and agreement with senior members
of department and SD
9
www.huntingdon.com
Fishbone
10
www.huntingdon.com
5 Whys
11
www.huntingdon.com
Example 1
12
Week of Formulation
Group NominalAnalysed
concentration (mg/ml) RME% Difference from
Mean
inclusion Analysis 1 Analysis 2 Mean (%) Analysis 1 Analysis 2
(mg/ml)
13 1 0 ND ND - - -2 0.75 2.6904 2.734 2.71 261.3 -0.80 0.803 1.5 1.4978 1.5041 1.50 0.0 -0.21 0.214 2.75 0.61669 0.60652 0.61 -77.7 0.83 -0.83
• Group 2 samples are above the quantification range
• Group 4 samples are below the quantification range
• Obvious error?
www.huntingdon.com
Investigation
g Samples had been booked into the department in the wrong order
g Analyst followed the booking in rather than looking at the labels of the samples
g New samples were taken from the bulk sample provided by Pharmacy
13
www.huntingdon.com
Re-sampled Analysis
14
Week of Formulation
Group NominalAnalysed
concentration (mg/ml) RME% Difference from
Mean inclusion Analysis 1 Analysis 2 Mean (%) Analysis 1 Analysis 2 (mg/ml)
13 1 0 ND ND - - -2 0.75 0.72346 0.71577 0.720 -4.0 0.53 -0.533 1.5 1.4978 1.5041 1.50 0.0 -0.21 0.214 2.75 2.7186 2.5703 2.64 -4.0 2.80 -2.80
• All samples now within specification• Original results for Groups 2 and 4 not
reported due to an identified analyst error
www.huntingdon.com
Example 2
Week of Formulation
Group NominalAnalysed concentration
(mg/ml) RME% Difference from
Mean inclusion Analysis 1 Analysis 2 Mean (%) Analysis 1Analysis 2
(mg/ml)
1 1 0 ND ND - - -2 10 9.6428 8.1326 8.89 -11.1 8.50 -8.503 30 27.686 26.675 27.2 -9.3 1.86 -1.864 60 58.827 57.72 58.3 -2.8 0.95 -0.95
15
• Investigation shows no calculation errors• Re-dilution confirms the results• No preparation errors found with formulation
preparation
www.huntingdon.com
Contingency Analysis
Week of Formulation
Group NominalAnalysed concentration
(mg/ml) RME Precision% Difference from
Mean
inclusion Analysis 1 Analysis 2 Mean (%)from mean
Analysis 1
Analysis 2
(mg/ml) (%)
1 2 cont 10 9.9601 10.022 9.99 -0.1 0.44 -0.31 0.31
16
• Contingency analysis showed that results were acceptable
• What should we do about the original analysis?
• Outlier test?
www.huntingdon.com
Outlier Test – Dixon’s Q Test
n 3 4 5 6 7 8 9 10
Q90% 0.941 0.765 0.642 0.560 0.507 0.468 0.437 0.412
Q95% 0.970 0.829 0.710 0.625 0.568 0.526 0.493 0.466
Q99% 0.994 0.926 0.821 0.740 0.680 0.634 0.598 0.568
17
www.huntingdon.com
Reportable Results
g 8.1326 mg/mL was rejected from the mean as an outlier with 90% confidence.
18
Group Nominal Analysed concentration (mg/ml) RME Precision % Difference from Mean
inclusion 1 2 3 4 Mean (%)from mean 1 2 3 4
(mg/ml) (%)
2 10 9.6428 [8.1326] 9.9601 10.022 9.87 -1.3 2.06 -2.35 - 0.86 1.49
www.huntingdon.com
Example 3
g UV Method – controls usually reportable as <LOQg One sample (from 3) showed a response >LOQg Full spectrum scan did not show a test item responseg No dilutions available to re-readg Time sensitive as pre-dose analysis!
19
www.huntingdon.com
Contingency analysisg The next 3 samples were read <LOQg Full scans were done of all controls = clear of test item
responseg Rejected the initial sample result as anomalous as 5/6
controls were <LOQg Full investigation was not required as able to
demonstrate the control group was clear by full spectrum scans (of all controls) and results of 5/6 controls
20
www.huntingdon.com
g Any Questions?
21