1. Mining for Opportunity in
BCC Board Meeting
September 29th
Mining for Opportunity in
the 21st century
Donor Vigilance: Using Hemovigilance data
Improving both Center Vision and Donor Experience
Kevin J Land MD
Chair, AABB Donor Hemovigilance Committee
Senior Medical Director, Blood Systems
ACT I

2. The Big Picture
“If you believe in continuous process improvement, then you must take a system-wide approach, comparing your self to others.”
“If your goal is to improve the donor and patient experience and outcomes, then you need to systematically collect and analyze and USE evidence to drive change.”
“local metrics drive local continuous process improvement and culture change, while regional and nationwide metrics help identify best practices. “
“Evidence based analytics is critical to your business patient care.”
DonorHart® is already being used by donor centers for a variety of purposes.

3. Without a way to organize information nuggets we often overlook their significance (especially with more clinical data)
‘I wonder…if we haven't become so numbed by all these numbers that we are no longer capable of truly assimilating any knowledge which might result from them.‘
Chapter 4: Field of Ignorance, p95
Michael Lewis (2004)
Data – Information – Knowledge – Wisdom
Information: understanding relationships between data. Formulation of strategic objectives
Knowledge: understanding patterns. Specific solution takes shape
Wisdom: understanding principles, solving problems

4. Mining for all that it’s worth Gold Mining in California
Gold rush began in 1848
primarily surface gold (lying in streams or spilling from cliffs)
New techniques had to be developed after 1850 as the easy to find surface gold disappeared.
peaked in 1852.
$2 billion (>750K lbs) extracted,
Some active mines still exist in the mother lode belt
Helped bring settlers to region and helped define state borders (especially with Nevada)

5. Where to get to the Gold? Small v Large Scale Manual v Automated
Easy v Hard to See at First
Limited v Large Yield Potential
Panning or Sluicing
Hydraulic mining
Sediment
Hard rock mining
Open Pit mining
Encased in Rock

6. Uses of Gold are similar to the uses of good data
Base units cause excitement but have little intrinsic value
Must be processed and
aggregated to visualize real value
Ultimately serves many strategic purposes

7. There are nuggets of information around us that should be used to improve donor experience & outcome

8. Donor Center Informatics Problem Statements
Collection centers have lots of data stored in many places that are underutilized to drive donor and business improvements
BECS systems are designed to collect data to demonstrate appropriate manufacturing steps for each donation, getting in the way of data analysis
We have our own internal and external barriers resisting change (maybe not!)
We have conflicting priorities (business, regulatory, competition, time/ resource limitations)
Until recently, there were no national/international naming & formatting data standards
Limited models available to take existing donor data and drive improvements
Short term needs can obscure long-term strategic goals: donor recruitment/ retention/safety, better understanding of own system, & business/process improvement.

9. In essence, we have a Big Data problem AND it requires new approaches to mine it
Data mining (aka Biovigilance) requires us to provide meaning and context to the various data elements (aka informatics) before we can put them together
Converting data to knowledge requires an understanding of the relationships between data
Converting knowledge to information requires the ability to see patterns
Converting Information into Wisdom requires an ability to distill (simplify) into unifying principles

10. Data Mining techniques and applications are not new
Behavioral/Attitudinal
Segmentation
Target Marketing
Data/Output Interpretation
Data Presentation
Cross Selling
Advertising
Campaign Management
Clustering
Summarization
& Visualization
Targeting Models
Fraud Detection Models
Association
Rules
Association
Rules
Association
Rules
Association
Rules
Association
Rules
Association
Rules
Association
Rules
Classification
& Regression
Data Mining
Applications
Change
Detection
Change
Detection
Change
Detection
Change
Detection
Change
Detection
Change
Detection
Change
Detection
Sequential
Patterns
Fault Detection Models
Intrusion Detection
Graph based
Analysis
Financial Forecasting
Demand Forecasting
Forecasting
Financial Forecasting
Social Network Analysis
Routing, Transportation, Link Analysis
© 2010 Madhav Erraguntla

11. Knowledge Discovery Steps
Define Goal, Business Interests
Create Target Data Set
Clean & Pre-Process Data
Perform Data Reduction & Transformations
Choose The Data Mining Task
Choose The Algorithms
Perform Data Mining
Interpret and Extract Actionable Insights
Refine and Repeat
© 2010 Madhav Erraguntla

12. * * * Added Salty snacks and muscle tensing 2010-2011 2009-2010
Add minimum blood volume criteria >3500
Courtesy Mary Townsend

13. US Donor Hemovigilance Working Group: Executive Summary of Key Pilot Successes
Developed US DHV common definition set (CDS) based on
Existing models, nationally/internationally.
Objective evidence-based criteria, signs and symptoms
Developed software with contract vendor (KBSI)
Presented and published results
Leading Donor Hemovigilance efforts internationally
VVR (inter)nationally decreased as interventions built on hv data have been implemented
Interventions which grew out of US HV data have been incorporated into a set of recommendations sent to ABO centers (North America, Western Europe, Australia and a little Japan)
Helped develop draft International CDS standard with defined minimal (MDS) & optional data elements
Experiencing growing national and international interest

14. AABB US Donor Hemovigilance Working Group Reports to AABB HV Committee (AuBuchon, Chair)
Anne Eder, MD, PhD
ARC National Headquarters
Madhav Erraguntla, PhD
KBSI
Mindy Goldman, MD
Canadian Blood Services
Michelle Greenland
LifeShare Community Blood Services
Linda Gruber
BloodCenter of Wisconsin
Mary Gustafson
PPTA
Barbara Hallenburg
Hany Kamel, MD
BSI
Kevin Land, MD
BSI, (Bonfils), Current Chair
Bruce Newman, MD
ARC, subject matter expert
Kadi Schroeder, RN
Bonfils Blood Center
James Stubbs, MD
Mayo Clinic
Peter Tomasulo, MD
BSI (past chair)
Mary Townsend, MD
BSI, (Coffee Memorial Hospital)
Barbee Whitaker, PhD
AABB
Johanna Wiersum-Osselton, PhD
TRIP, The Netherlands

15. Acknowledgements
James Berger, MS, MT (ASCP), SBB, US Department of Health and Human Services
CDR Richard Henry, ML, MPH, US Department of Health and Human Services
LTCOL David Lincoln†, Armed Services Blood Program Office, US Department of Defense
D. Michael Strong, PhD, University of Washington School of Medicine
Alan Williams, PhD, US Food and Drug Administration
†Deceased November 18, 2012

16. Original Purpose
The Donor Hemovigilance Working Group (DHVG) will implement a national monitoring program on donor safety issues as an important element of continuous improvement in the comprehensive biovigilance network.
This report is an update on activities that culminated in the publishing of the AABB Donor Hemovigilance Annual Report 2012.

17. Brief History: US Biovigilance Gap Report
Drafted in response to 2006 ACBSA recommendations (and concurrence by Assistant Secretary of Health):
DHHS coordinate Federal actions and programs to support and facilitate biovigilance in partnership with private sector initiatives
DHHS form a task group to perform a gap analysis of current systems and make recommendations for public-private partnerships in biovigilance (blood, cell, tissue, and organ therapies).
Adapted from Alan Williams

18. HHS Biovigilance Gap report: Key Deficiencies of Hemovigilance in the United States
Absence of……….
Long-term stability
National scope
Multicenter design
Common definitions
Broad data access and sharing
Real Time Data Availability
Active use to document practice improvement
Adapted from Alan Williams

19. Brief History: US Donor Hemovigilance
Funding
: Software - DHHS through SBIR with KBSI
to date: DH WG, marketing, salaries, etc - AABB
: DH WG - various blood centers
Focus on Donor Adverse Reactions
Key Participants – DHHS, AABB, ABC, ARC, DoD, Bonfils Blood Center, BSI, Coffee Memorial, Mayo, PPTA, Canadian Blood Services, KBSI, ISBT/IHN
AABB – 1 million
Blood centers – 1 million
Adapted from Alan Williams

20. HHS/AABB Donor Hemovigilance
National Standards for Donor Reaction Data Collection
Data Elements and Definitions
Reactions and Reaction Categorization
Systemic, Standard Mechanism to Calculate Donor Reaction Rates
Trends at Facility, Organization, Region and Nation Levels
Comparison With Peers, Region and Nation
Slide from Alan Williams

21. HHS/AABB Donor Hemovigilance
Predictive and Causality Analysis
Analyze Variables (Age, Sex, Weight, BP) Affecting Donor Reaction Rates
Device and Kit Analysis
Analyze Associations between Policies, Procedures of Organizations and Donor Reaction Rates
Intervention Analysis and Management
Slide from Alan Williams

22. Primary Charge of US Hemovigilance Focusing on Blood is to develop a system that will be
Electronic
Voluntary, confidential, non-punitive reporting service
Focused on improving donor safety
Managed by experts
Able to analyze data and understand implications for donors and those caring for them
Provide immediate data access for participants
Provide periodic access for external analysis
Avoid duplication of existing systems

23. Scope:
Standard Data Element Definitions, Reactions and Reaction Categorization
Each Organization Will Provide Mapping Between Organization Reaction Codes and National Reaction Codes
Analysis Methodology Should be Flexible to Support Incomplete Data

24. Number of Donations By Predefined Demographic Categories
Scope: Denominators Offer uni-variant, bi-variant, and multi-variant options
Number of Donations
Number of Donations By Predefined Demographic Categories
Age Ranges, Sex, First Time Vs Repeat,..
Independent Demographic Categories Vs Stratified
Demographic Data For All Donations
Comprehensive Causality Analysis
Future Ad-hoc Analysis

25. Q: What type of data elements are gathered
Q: What type of data elements are gathered? A: Reaction Type & Category + optional signs &symptoms
Table Reaction Example
Reaction Type
Reaction Category
Vasovagal
Prefaint, no LOC (uncomplicated or minor)
LOC, any duration (uncomplicated)
LOC, any duration (complicated)
Injury
Local Injury related to needle
Nerve Irritation
Hematoma/Bruise
Arterial Puncture
Apheresis
Citrate
Hemolysis
Air Embolus
Allergic
Local
Systemic
Anaphylaxis
Other

26. Summary of Pilot data study How much data is required?
Total Elements
Minimum required
Allows null
Min Elements
Donor Data 7 4 3
Organization name, Donor ID,
DOB, gender
Donation Data 36 5
(2 + 3) 31
Organization name, Donor ID, Collection Center, Donation ID,
Donation Date
Reaction Data 25
( ) 18
Reaction Type, Reaction Category, update-flag
142 potential unique data elements
Donor: 7
Donation: 30*+ 5**
Reaction:
(includes 2 multi-select lists)
signs and symptoms (60)
adverse reactions (23)
Denominator: (provided monthly only)
13 categories with 80 total elements
+ 80 elements in Denominator Data
Courtesy Pilot Facility Bonfils Blood Center, Denver Co

27. Initially Reported to DHV
Summary of Pilot feasibility and time study How much data can be readily imported (eg readily accessible)?
Total
Unique
Elements
(% avail)
Inform.
System*
(BECS)
Primary Forms
Not Collected
Not Relevant
Initially Reported to DHV
(%)
Donor
7 (100%) 7
Donation
34 (46%) 12
(height/ manufact/kit) 3
(total protein/ Hgb)
12 (35%)
Reaction
21 (83%)
21 +1 update flag
10 (48%)
62 (64%)
31%
44%
19% 6%
29 (47%)
Nothing is impossible, but some things are definitely easier to obtain than others
This represents data we were able to get only by remapping our terms and formatting forms to be consistent with Donor BV templates. No new elements were added.
Donation element examples we do not collect: height (none) and weight (only apheresis)
Component produced is available via DIS, but not on a screen that DR can easily access, but could be put into a standard query IT
Donor demographic
Donation information
Type I and II Vasovagal reactions and Hematomas
Denominator data – a standardized report needed to be created
Remaining data will require manually into Biovigilance system
Primary Forms: Included on DN Incident Form, Donation Record, or Apheresis Run Sheet
Not Relevant: Not intended for our business (eg total protein)
Courtesy Pilot Facility Bonfils Blood Center, Denver Co

28. Summary of Pilot feasibility and time study Monthly time commitment ~ 3 ¾ hrs/mo for 175K RBC center
IT Data download (<1h)
electronic reaction information
denominator data
DHV file manual data entry (2.5h)
Initially, additional documentation from forms added ~15m per record.
Reduced to <5 min each or ~ 15 reports/hr in <1mo use
Upload to DHV website (<15m)
Initially took 6h/mo longer
Now takes same time as before, with more data for analysis +
(3 hours) estimated for setting up initial query
Donor record, apheresis run record, and safetrace
Data entry – probably will cut in half once we get better at it.
TBD
How much time does it take for a “complete record
Vs.
How much time does it take to get the basic data entered
Shortly after the transition to the new forms on Aug 31, our staff were still preferentially writing prose for their most of the reaction information. As they move toward using the check lists it will be easier for the data entry person to just flow through the form.
15/report at first, now 9/report
Courtesy Pilot Facility Bonfils Blood Center, Denver Co

29. AABB First Donor Hemovigilance Annual Report: 2012 data
Adverse reactions from 1,171,906 individual donations
Denominator Data: 100% univariant
99% Allogenic donations (1% total autologous, directed, and therapeutic)
148 Potential data + 80 univariant denominator data elements
Adverse reactions from 1,171,906 individual donations

30. How many different attributes were reported? Who is reporting them?
Table X: Attribute Reporting
Donor Variable
Percent reporting
Age
100%
Donation History
Donation Type
Gender
Procedure Type
Ethnicity
80%
Collection Site
Pulse
60%
Sponsor Group Type
Weight
Blood Pressure
40%
Race
Device Manufacturer
20%
Device Model
Height
Device Software 0%
Container Manufacturer
Container Kit Type
(~22 facilities in the cue)
5 facilities reported data
2 other facilities with partial data
8 in contract talks
7+ adopting CDS

31. Donor Demographics (n=1,171,906 individual donations)
Attribute
Donation %
Reaction %
Reaction %/ Donation%
GENDER
Female
47.9 65
1.36
Male
52.1 35
0.67
Donation Status
First Time Donor
14.6
31.3
2.14
Repeat Donor
85.4
68.7
0.80
Donation Type
Whole Blood
75.5
83.6
1.11
Automated*
24.5
16.4
*Aph PLT 14.2%, dRBC 14.2%, PLT & RBC 1.2%, PLT & Plasma 1.6 %, other multi-comp 1.7%

32. Automated Donations (24.5% total donations, 16.4% Reactions)
DRBC 14.2% (58.0%)
Aph PLT 5.4% (22.0%)
PLT + PLSM 1.6% (6.5%)
PLR + RBC 1.2% (4.9%)
Whole Blood

33. Donations by Donor Age (n=1,171,837 donors)
61% >40yo
21% >60yo
11% HS aged
(25% reactions)
9% 23-29yo
(12% reactions)

34. Reactions By Type and Donor Age
Reaction rate for all reactions types per 1,000 Donations
(all p<0.001)*
years
29.7 (2.28)*
years
22.2 (1.69)
years
17.2 (1.3)
years
12.0 (0.91)
years
9.3 (0.7)
years
8.7 (0.65)
years
years
9.0 (0.67)
≥80 years
12.3 (0.93)
Reactions By Type and Donor Age
28.7% First time donor rate

35. Reaction Rate Summary Table
Type of Reaction
Reaction rate per 1,000 donations
Overall Reaction Rate (1,171,906 donations)
13.41
Vasovagal
9.65
Prefaint, no LOC (uncomplicated or minor)
7.33
LOC, any duration (uncomplicated)
1.87
LOC, any duration (complicated)
0.40
Injury
0.06
Local Injury related to Needle
2.48
Nerve Irritation
0.23
Hematoma / Bruise (symptomatic)
2.23
Arterial Puncture
0.03
Apheresis
0.83
Citrate
0.05
Hemolysis *
Air Embolus
Infiltration
0.77
Allergic
0.22
Local
0.18
Systemic
0.04
Anaphylexis
Other
Reaction Rate Summary Table

36. Seasonal Donation Patterns Among Donors

37. Reactions by Location
2% Pre-donation
55% While on bed
43% Post-donation
6% walking onsite
4% in bathroom
13% offsite

38. Reaction Rate by Collection Site
95% confidence intervals

39. Collection Facilities are Discovering Additional and Innovative Uses for DHV Data
Medical Affairs (Assessing post implementation effectiveness of risk reduction strategies)
Implementation of pre-hydration stations and salt replacement initiatives
Restriction of blood donations based on new total blood volume calculation
Impact of staff tension training on donor adverse events
Quality Assurance (As part of a quality essentials program)
Comparison of facility reaction rates to national aggregate results
Denominator data helps identify statistical sampling size needed for auditors
Operations
Impact on donor adverse events as a result of pure operational changes
Changes in bag size and manufacturer (450ml – 500ml)
New mobile double red cell collection program
New staff monitoring for high risk donor groups (eg high school blood drives)
Descriptive denominator data being used for donor marketing and recruitment
Business Review Made Easy (Standard DonorHART™ reports are used)

40. Implementation Analysis 450-500mL bags
This next slide is an example of our data after the implementation of increasing our collection from a 450 mL bag to a 500 mL bag. We saw a significant increase in our moderate vasovagal reactions, so we implemented a few things among our HS drives the following semester and brought the staff in for re-training on how to identify and prevent reaction from increasing their severity. Nip it in the bud, per se. We strongly encouraged (forced) our HS donors to drink a bottle of water prior to their donation (we gave it to them when they registered and encouraged them to drink along the way). We implemented a brochure for Muscle tension exercises and allowed the donors to read this information during their donation and practice the muscle tension. After that implementation you see out numbers drop to where they were prior to the 500 mL bag implementation.
FIX this slide?|!!!
Courtesy Bonfils Blood Center

41. Current and Future Directions of AABB Donor HV
Continued alignment with international DHV definitions (New Charge)
Help develop thoughts around both Minimum Data Set (MDS) and larger Common Definition Set (CDS)
Formally define DonorHART™ Lite
Implement bi-variant and multi-variant denominator data capabilities
Begin DHV research projects, such as…
impact of Iron and Calcium depletion in donors

42. International Efforts Summary
Presentations on DonorHART progress made at several ISBT/IHN meetings
Several countries have expressed interest potential to use DonorHART
ISBT/IHN in 2013 set up working group to update Donor HV definitions
Members of AABB DHV working group were asked to participate
Updated draft ISBT Donor HV definitions have been made
ISBT has agreed to adopt many of the US definitions and agrees to allow signs and symptoms (primarily from US system) to be collected optionally
Only 9 issues separate current US HV definitions and draft ISBT definitions
US DHV working group and ISBT have agreed to compromises
ISBT definitions anticipated to be approved before the end of 2014, thus leading to an international unified Donor HV definition set.
1. Drop tetany and LOC >60sec from signs & symptoms
2. Move injury to an OUTCOME.
3. Move to ISBT VVR Locations (see Table 3)
4. Resolve if category LOC complicated is useful? Should DH drop this and only look at it when centers supply symptoms since the category is relatively subjective?
5. add Re-bleeding/Delayed bleeding as Category
6. Adopt ISBT Painful arm as Reaction Category instead of nerve irritation/injury
Figure out how many have used Major Cardiovascular Event as AE Category. Is this capture elsewhere as a general OUTCOME?
Adopt ISBT proposed Citrate optional subcategories (see Table 5)

43. International Donor Vigilance Efforts: Evolving how collection centers view their data
Stage
Description
Benefits I
Internal data collection
Local data gathering and research II
Internal adoption of external standard vocabulary
Ability to compare your data to others based on internationally accepted vocabularies
III
Surveillance data (aka minimal data set)
Minimal common surveillance data that can be shared and compared internationally
Surveillance detail summarized in AABB annual report
Signs and symptoms are not reported
IVa
Basic Benchmarking
(aka DonorHART lite)
Data incorporated into AABB donor hemovigilance annual report
Donor hemovigilance dashboard
IVb
Enhanced Benchmarking with extended data mining capability (use of common definition set)
Shared process improvement expertise to improve donor outcomes
Increased BECS data liquidity (donor demographics, business analytics, etc)
NOTES:
Levels I and II can require sophisticated data warehouses
Adopting national and international vocabularies and standards are preferred

44. DonorHART frees BECs data and solves many of the problems currently plaguing donor centers
Independence from original data structure
BECS, excel, forms, testing devices, others (Hemasphere, Talisman, etc)
The ability to join multiple data sources at detail level
Limits impact to business system performance
Not slowing down BECS to run query
Improves speed of data mining and answering questions
If the data is already there, you do not have to wait for it (or requisition it)
Allows single database to be reused for a variety of reasons
Most projects require unique aggregations
However, your

45. Four Different Donor Vigilance Paths to take, which one will you take!
Real benefits to working as a team
Improved Donor outcomes
Set (inter)national standards
Greater than sum of individual effort
Donors and Donor groups need to know we are committed
Donors increasingly sophisticated
Groups (HS, businesses) asking about objective measures
?? Adequacy of blood supply using older techniques of recruiting / retaining donors
Adapt
Adopt
Go your own way
Hikinh, biking, nordic walking, camping, dogs/kids
Hiking Horsetooth Rock Trail Fort Collins, Colorado
Do nothing

46. Summary US Donor Hemovigilance Pilot Program was a success.
2012 Report released Report to begin after June with more centers.
Program likely will need bridge funding for 1-3 years.
AABB and Donor Centers are deriving value from DonorHART.
US Donor Centers are adopting US definitions.
There is some international interest in the system as well.
Most US donor centers, though, are waiting for new definitions and answers to questions about what access FDA/HSS wants and what will the future financing look like before jumping onboard.
Appropriate donation type, frequency, minimum Hgb, & volume
Male versus Female donors
Estimated total blood volume
Annual Blood and Plasma loss
Reducing events and reactions
Young versus older donors
First time versus repeat donors
Prompt Identification & Treatment
Falls and Injuries
Preventions & Interventions
Iron or Calcium depletion
Increasing donor satisfaction/return rate
Thank you!

47. Final thought… We already manage donor suitability using complex data
Final thought… We already manage donor suitability using complex data. Why not manage their donation outcomes in a similar fashion?
Management of donor issues used to be relatively easy.
Donors were managed by age, TBV
Age, gender, mini-physical, Hgb/Hct (ferritin ?) level, donation type, donation technology, donation frequency, adverse reactions.