RiskInformed Interventions in Community Pharmacy: Implementation and Evaluation

1
Risk-Informed Interventions in Community
Pharmacy Implementation and Evaluation

• Michael R. Cohen (Principal Investigator)
• Judy L. Smetzer (Project Manager)
• Institute for Safe Medication Practices
• September 14, 2009

2
Current Research Project

• Risk-informed Interventions in Community
  Pharmacy Implementation and Evaluation
• Three interventions
• Scripted mandatory patient counseling for
  targeted high-alert medications
• Readiness assessment for bar-coding technology
• Risk assessment/intervention scorecard

3
Prior Study Aims

• Using Risk Models to Identify and Prioritize
  Outpatient
• High-Alert Medications
• Identify a list of high-alert medications
  dispensed from community pharmacies
• Available error data, ISMP surveys, review of
  literature, litigation data,
• Develop comprehensive risk models for four
  high-alert medications (ST-PRA) using model
  building teams with facilitators
• Warfarin, fentanyl transdermal, insulin analogs,
  methotrexate oral
• Identify error pathways that have the highest
  probability of causing harm (fault trees)
• Identify and determine the impact of approaches
  for eliminating or reducing the risk of harm

4
High-Alert Medications

• Individual Drugs
• carbamazepine
• chloral hydrate liquid
• sedation of children
• heparin
• unfractionated/low-molecular weight
• methotrexate
• non-oncologic use
• midazolam liquid
• sedation of children
• propylthiouracil
• warfarin

• Drug Class/Category
• Antiretroviral agents
• Chemotherapy, oral
• Hypoglycemic agents, oral
• Immunosuppressant agents
• Insulin
• Opioids, all formulations
• Pregnancy category X drugs
• Pediatric liquid medications that require
  measurement

5
Socio-Technical Probabilistic Risk Assessment
(ST-PRA)

• Models combinations of failures that lead to
  undesirable consequence (Relex software)
• Used in other industries
• Differs from FMEA, which analyzes each failure
  separately, never in combination (pharmacy
  dispensing process)
• Begins by defining the top-level event (PADE)
• Medication dispensed to wrong patient at point of
  sale
• Patient given wrong dose of warfarin
• Uses experienced modeling team to yield
  probability estimates of basic events

6
Socio-Technical Probabilistic Risk Assessment
(ST-PRA)

• Risk model includes effects of
• Human error
• Socio-technical aspects
• At-risk behaviors and procedural deviations
• Mechanical/technology failures
• Some data readily available in community
  pharmacies
• Rx volume, exposure rates, technologies and
  percent of use, computer alerts followed,
  presence of certain steps or processes like use
  of drive through window, availability of 24 hr
  pharmacy, opening bag at P.O.S.

7
Example of ST-PRA Fault Tree Risk Model
8
Human Error Probabilities
ST-PRA uses probability estimates to quantify
risk

• Unfamiliar task performed at speed/no idea of
  consequences 510
• Task involving high stress levels
  310
• Complex task requiring high comprehension and
  skill 15100
• Select ambiguously labeled control/package
  5100
• Failure to perform a check correctly
  5100
• Error in routine operation when care required
  1100
• Well designed, familiar task under ideal
  conditions 410,000
• Human performance limit
  110,000

9
Performance shaping factors that impact on
probability of error in community pharmacy

• Task complexity
• Information complexity
• Work environment
• Stress
• Time urgency
• Training/experience
• Familiarity with task
• Design of labels
• Clarity of handwritten prescriptions
• Look-alike drug names or packages

10
Insulin Analog Data Entry Error (wrong drug)

• Start
• 1 data entry error per 100 prescriptions
• Capture
• 96 errors captured
• Risk (PADEs that reach patients)
• 3 wrong drug errors per 10,000 prescriptions
• 2,200 errors annually (chains in study)
• 6,400 errors annually (national)

11
Insulin Analog Data Entry Error (wrong drug)

• Interventions
• Use of tall man letters to distinguish products
• 50 improvement
• Increase patient counseling from 30 to 80
• 67 improvement
• Conduct a second redundant data entry
  verification during product verification
• 50 improvement
• All three interventions
• 95 improvement
• 3/10,000 to 1/1 million errors that reach
  patients

12
Fentanyl Patches Prescribing Errors (wrong dose)

• Start
• 1 dose error per 1,000 prescriptions
• Capture
• 27 errors
• Lack of information about opioid tolerance,
  indication
• Risk (PADEs that reach patients)
• 7 dose errors per 10,000 prescriptions
• 1,000 errors annually (chains in study)
• 3,400 errors annually (national)

13
Fentanyl Patches Prescribing Errors (wrong dose)

• Interventions
• Increase in patient counseling from 10 to 80
  and increase ability to detect inappropriate
  doses during counseling session
• 64 improvement
• Conduct an intake history of opioid use at
  drop-off
• 40 improvement (tested with 20 implementation)
• Both interventions
• 78 improvement
• 7/10,000 to 1/10,000 errors that reach patients

14
Warfarin Filling Errors (drug/dose)

• Start
• 1 drug selection error per 1,000 prescriptions
• 1 dose selection error per 10 prescriptions
• Capture
• 99.9 errors
• Consistent use of bar-coding technology
• Risk (PADEs that reach patients)
• 9 wrong drug errors/1 billion prescriptions
• 1 error every 14 years (chains in study only)
• 9 wrong dose errors/10 million prescriptions
• 7 errors annually (chains in study only)

15
Warfarin Filling Errors (drug/dose)

• Interventions
• Increase patient counseling from 30 to 80
• 67 improvement
• 9/1 billion to 3/1 billion errors reach pt (drug)
• 9/10 million to 3/10 million errors reach pt
  (dose)
• Eliminate bar-coding technology
• (95,340) reduction in safety
• Eliminate pill image on the product verification
  screen
• (334) reduction in safety
• Eliminate bar-coding and pill image
• (445,000) reduction in safety
• 9/1 billion to 4/100,000 errors that reach pt
  (drug)
• 9/10 million to 4/1,000 errors that reach pt
  (dose)

16
All Medications Point of Sale Error (wrong
patient)

• Start
• Due to bagging error (4 per 10,000 prescriptions)
• Due to misidentification of bag or patient (3 per
  1,000 prescriptions)
• Captured
• 64 errors captured
• Risk (PADEs that reach patients)
• 1 error per 1,000 prescriptions
• 1.3 million errors annually (chains in study)
• 4 million errors annually (national)

17
All Medications Point of Sale Error (wrong
patient)

• Interventions
• Increase patient counseling from 30 to 50
• 27 improvement
• Open the bag at the POS
• 56 improvement
• Increase compliance with ID process from 50 to
  80
• 34 improvement
• All three interventions together
• 86 improvement
• 1/1,000 to 2/10,000 errors that reach patients

18
Current Research Project

• Risk-informed Interventions in Community
  Pharmacy Implementation and Evaluation
• Scripted mandatory patient counseling
• Warfarin
• Fentanyl patches
• Methotrexate
• Insulin analogs
• Low-molecular weight heparin
• Hydrocodone and oxycodone (with acetaminophen)
  top 200
• Readiness assessment for bar-coding technology
• Risk assessment/intervention scorecard using risk
  models from first study HAMERS tool
• Added to increase frequency of observation of
  counseling sessions

19
Intervention 3 HAMERS(High-Alert Medication
Error Risk Scorecard)

• ST-PRA models translated into practical
  assessment tool and scorecard
• Tool Kit will include
• Introduction
• Key learning from risk models (prior study)
• User instructions
• HAMERS tool
• Scorecard with qualitative (distribution of risk)
  and quantitative (PADE rates) information
• Tool calculations driven by reports from original
  risk models

20
Intervention 3 HAMERS

• Inputs
• Set-up questions
• Relevance Would the step provide capture
  opportunity?
• System attributes Require data entry
  verification for pharmacists?
• Availability Use bar-coding technology? Specific
  computer alerts?
• Prescription volumes?
• Exposure rates
• Frequency of pharmacists/technicians entering
  prescriptions?
• Capture opportunities
• What percent of errors will be caught during this
  step?
• At-risk behaviors
• Frequency of choosing not to ask patient for
  second identifier?
• Human errors
• Frequency of forgetting to read back an oral
  prescription?

21
Intervention 3 HAMERS

• Outputs
• Scorecard that quantifies the risk of specific
  PADEs
• Bar-graph that shows distribution of risk
• Which elements contribute most to the PADE?
• Menu of interventions to reduce risk
• Pharmacy chooses from the menu of interventions
• Pharmacy makes changes to inputs based on the
  planned interventions
• Pharmacy receives a revised scorecard that
  quantifies improvements based on planned
  interventions
• If (intervention) is implemented, then risk that
  the PADE will reach the patient is ___.
• If risk factor is (increased/decreased) by __,
  risk that the PADE will reach the customer is
  reduced to __.

22
Intervention 3 HAMERS

• Tool can be used to measure risk within
  dispensing system for any medication or most
  types of errors/ PADEs
• Focus on high-alert medications
• Can measure risk of not capturing prescribing
  errors
• Cannot measure risk of patient self-administration
  
• Limited menu of interventions
• General in nature
• Specific to high-alert medications
• Include all tested interventions from prior study
  and others

23
Intervention 1 Patient Counseling

• Pre-intervention observation in pharmacies
• 50 observations completed
• 4 states
• 2 states with mandatory counseling
• 2 states with mandatory offer to counsel
• Preliminary findings
• No counseling in states with offer to counsel
• Counseling for OTCs more common than for
  prescription drugs
• More frequent counseling in states with mandatory
  counseling
• Differences between state enforcement of
  counseling
• Not covering information linked to PADEs

24
Intervention 1 Patient Counseling

• Implementation Tool Kit
• Scripted counseling materials, checklists, health
  questions
• Consumer handouts about targeted drugs
• Specifically targets known causes of PADEs
• Consumer outreach materials to promote counseling
  
• Availability on http//www.consumermedsafety.org
• Model state regulations for requiring/limiting
  mandatory counseling for high-alert drugs

25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
Intervention 1 Patient Counseling

• Measures
• Self-administered surveys to patients
• Perception of counseling encounter/value of
  handouts
• Increase understanding?
• Result in new information?
• Result in changed behavior?
• Reduce risk of self-administration error?
• Treatment for PADE?
• Toll-free number to call research team
• Incentives to send back survey
• Self-administered surveys to pharmacists
• Perceived value and impact of counseling

30
Intervention 1 Patient Counseling

• Measures (contd)
• Post-implementation observation
• Detection of prescribing or dispensing errors
• Detection of potential self-administration errors
• Barriers to counseling
• Factors that facilitate counseling
• Quality of counseling sessions

31
Intervention 2 Bar-coding Readiness Assessment

• 46-50 of community pharmacies in the US do not
  use barcode technology for product verification
• 100 pharmacies participating in the study
• Survey to determine why non-users are still
  non-users
• Phase 1
• 100 pharmacies will complete the assessment and
  submit findings
• Pharmacies will complete survey to measure
  perceived value
• Phase 2
• Pharmacies from Phase 1 that have since
  implemented bar-coding will complete survey to
  measure actual value