Quality Risk Management from Concept to Practical Strategies*

Transcription

1 Quality Risk Management from Concept to Practical Strategies* H. Gregg Claycamp, Ph.D. Director Div. of Compliance Risk Management and Surveillance and Associate Director, CDER Office of Compliance * Represents thoughts of the author and not policy of the FDA. 1

2 Overview ICH Q9: Quality Risk Management Contents, purpose Applications, approach Implementation Strategies Impediments to implementation What works can also raise caution flags Principles for robust strategies Don t forget risk science! 2

3 ICH Q9 Definition of Risk It is commonly understood that risk is defined as the combination of the probability of occurrence of harm and the severity of that harm. A conceptual, general purpose definition! Backed by a fit for purpose concept from quality management theory. Follows generally device industry concepts. Pharmaceutical Life Cycle coverage 3

4 And linking with risk to patient* Design Opportunities to impact risk using quality risk management Q9 Process Materials Manufacturing Facilities Distribution Q8 Q10 Patient 4

5 Plainly speaking What can go wrong? s i = scenario 2. What is the likelihood (probability) it will go wrong? p = probability i 3. What are the consequences (severity)? x i = consequence 5

6 Risk Management asks 1. What can be done to manage risk(s)? 2. What options are available? 3. What are risk trade-offs in terms of risks, benefits and costs? 4. What are the impacts of current risk management decisions on future options? 6

7 What s in ICHQ9? A Sample Process... Initiate Quality Risk Management Process Risk Assessment Risk Communication Risk Identification Risk Analysis Risk Evaluation Risk Control Risk Reduction Risk Acceptance unacceptable Risk Management Tools Output / Result of the Quality Risk Management Process Risk Review Review Events 7

8 Risk Management is Ubiquitous Company Strategic Risks Operational Risks Financial Risks Compliance Risks Competitor Advantage Company Viability Shareholder Harm Patient Harm ICH Q9 Impact 8

9 What s in ICH Q9: Tools for Quality Risk Management HHM, SFN General Risk Ranking and Filtering FME(C)A; HACCP; Root Cause Analysis; Variation Risk Management Formality Probabilistic Risk Analysis; Event Trees; Decision Trees;... Detailed 9

10 A Few Tools of the Risk Trades... Tool/Approach SFN HHM RRF CRA FEM(C)A HACCP VRM RCA DT PRA FTA ETA Name Brainstorming/Storming-Forming-Norming Hierarchical Holographic Modeling Risk Ranking and Filtering Comparative Risk Analysis Failure Mode and Effects (Criticality) Analysis Hazard Analysis and Critical Control Points Variation Risk Management Root Cause Analysis Decision Trees Probabilistic Risk Analysis Fault Tree Analysis Event Tree Analysis 10

11 A Key QRM Principle: The Right Tool for the Job ICH Q9 Includes an Annex of Tools 11

12 ICH Q9 s Catalysis of QRM?* Better Risk-Based Decisions Q9 Progress: GMPs for the 21 st Century Proposed as a launch platform for applied risk management tools, ideas and programs. Progress? Success stories? Remaining barriers to implementation? *QRM=Quality Risk Management 12

13 Is it Time for a Deeper Understanding of Risk Management(?) 13

14 Barriers to Science-Based QRM: Risk is a Concept Risk is intuitive and familiar to everyone, yet sophisticated and arcane when applied by complex organizations to complex problems. 14

15 Better Understanding of risk Individual: Risk is a cognitive and emotional response to expected loss. Societies: Risk is a societal expression of expected harm tempered by expected benefits. Organizations: Risk is a combination of the probability of occurrence and CMC severity Applicability of selected harms. Technical: Risk is usually based on the expected value of the conditional probability of the event occurring times the consequence of the event given that it has occurred. 15

16 Severity and Probability Risk Increasing Probability of Occurrence Low Risk Medium Risk High Risk A great communication tool; but difficult to implement from first principles! Increasing Severity of Harm/Consequence 16

17 What s Wrong [and Right] About Risk Matrices?* SAMPLE TABLE ONLY Health Probability of Occurrence Severity Scale Very Low Low Medium High Very High Death Medium Medium High High High A Hospitalization Low Medium B MediumHigh Risk High High Medium Acute Illness Low Medium Medium High High Risk Low Worry Low Risk Low Low C Medium Medium e.g., Cox, LA, What s Wrong with Risk Matrices? Risk Analysis 28: 497 (

18 Severity of Effect Risk Scores are not Risk! Example: FMEAs S Occurrence Probability O S x O Detectability* Criticality S x O x D SOD or Risk Number *Higher detection ability lowers risk score. 18

19 Understanding Risk Scores Ranking numbers from multiplication are come with problems from the obvious (zero) to the subtle (e.g.,expectations). Scores have questionable, direct translation to numerical risk estimates for risks to, e.g., Patients, Processes, or Compliance. What value are risk scores? 19 19

20 Risk Rankings Of what value? To focus experts and teams of experts on risk, as defined in a useful context. To focus experts and teams of experts on prioritizing work based on risk. Should not be represented as measuring or determining risk(s)! Not new but improved concepts methods having a systems perspective

21 Purpose of the Analysis... Risk scoring methods, including FMEA and others, are mostly about prioritization under a consistent process, not quantitative risk assessment. Mixer 1 Bottle supply Active ingredients container Sealer/capper Inert ingredients container Filter -- Chart for illustration purposes only. -- Sealer/capper Criticality 21

22 Consistent Decisions QMS Concepts! Recalls Programs Public Health Priority Products of Concern Difficult Processes Public Health Priority Products of Concern Recalls Facility History Difficult Processes Facility History ADE Scheduling Knowledge Base ADE Scheduling A defined, consistent process Ad Hoc Ranking Sample illustration, only! Ranking 22

23 Fundamental Objectives To minimize... Risks from Drug Quality Problems* p prod Risks from Products p fac Risks from Manufacturing Facilities p proc Risks from Manufacturing Processes Even fundamental objectives the purpose for the system represent probability space for scenarios. 23

24 Modeling at the Attribute Level Site Risk Score* Product Facility Process v j ( facility ) = Nj i= 1 w i x i Attribute Attribute Attribute v j ( process ) = Nj i= 1 w i x i Attribute Attribute Attribute v j ( product ) = Nj i = 1 w i x i 24

25 Mixtures of Qualitative & Quantitative? Product Features Characteristic 1 Characteristic 2 Risk Assessment Tools Attribute1 Attribute2 Attribute3 Simple Scoring System Characteristic 3 Attribute1 Characteristic 4 Attribute2 Attribute3 RISK MATRIX 25

26 Uncertainty and Probability 26

27 Better Understanding of Uncertainty Uncertainty is a capacious term Morgan and Henrion (1990), p. 47 Uncertainty From a lack of information From linguistics, e.g., reasonable About quantity About the mathematical model Extrapolation from complete information to the real world All models are wrong, some models are useful. * G. Box (1979). 27

28 Better Probability x Severity Risk For example, Increasing Probability of Occurrence A Low Risk B Medium Risk Uncertainty in the severity of harm High for Risk risk C Uncertainty in in C the the probability of of occurrence for for risk risk C Increasing Severity of Harm/Consequence 28

29 Expert Judgments Most Common Implementation Strategy Individuals to Groups How is Expert Judgment of Probability Used? Consensus v. Majority? Average values of individual judgments; or single groupbased judgment? Quantitative methods? Group dynamics important! (e.g. group think ) 29

30 Combinations of Judgments Common Each Expert s Mental Model of Uncertainty in Score D C B G E F A Prior distributions for scores in FMEAs FMEAs are an example where team-based judgments are used. This raises questions: Are the judgments consistent? Variability in judged scores? How should the judgments be combined? 30

31 Probability density* of x The Value of Including Uncertainty Three Understanding distributions uncertainty of of risk are in in shown risk estimation (a, (a, b, b, c) c) leads that have to to a the same better average understanding risk. Note of of the that the impact uncertainties of of alternative or or widths decision of of the thresholds distributions for for (e.g.) differ risk significantly. acceptance, quality changes, etc.) a b c (Average Risk) Risk (x ) *PDFs used for illustration only. 31

32 Conclusions Implementation of Quality Risk Management is happening in various programs. As we move to implementation from the principle-driven, launching platform of ICH Q9, now is the time to check in re: the risk-scientific foundations of implementation strategies. 32

33 And, ICH Q9 EWG Numerous colleagues in industry, government and academia 33

34 ICH Q9: Briefing Pack Slides Q9: Quality Risk Management Q9 Briefing Pack On website Table of Contents you may download chapters or the entire 480-slide set for review. 34